DAW GR tL gi ill et BN APATHY er eS cA A EEN ALTE ES OP Otay at arigl tae mfg om RUE TAR EY Cs NAL ONS TE AW ¥agy, wahsa sPayNy peor Bee ee ee ee Rate ty hoe a

Se Bo x z WM oid PANN EOS EQ La hal Weg rae ea, bt Beary

Sater alin Pade oN Pettey Oey e Un tate Se Gal Sere aS 8 el ie airs tied 6 eles en eee Sere rea Rrporgcrnn prsaerrwenit rBsieTAEET th riage nee ne ity 2 neg sbbesin era nll Ree

s iokeaenanntacuct ane lr atm ie agesincnte! centri ta ga gmhi mad nin galieeta i ' PNRM res ADM ed matte oy heb ALA aD OENT EM SE BTA" NYO Ale oy MPO SIMA EN LE Cal dyad read atent

ats neni tee fighc Babe Ie Sele ores YE hry Qube te he i UY fee brake PANTERA hs PE AYE A SAR PAS 7 474 tam he r

fasbeae la Pet sin condeat g Nahas tbat Canis (> Wout ts th Uespw at tenes ph et ANY : PUGS pour ene on ; DA ae : " penieitt vate any

Wah Ae hha Guatty Matte tate Dube ete wonivra sitar AEE hetimtinsrd Wad aad Ah llehic st SEAL an Geiser y One tte FACES A OLN ENS EON are’ AVANCE AN Fe raha aya EROS TE SOM RDNA HN "ate ay 686A TN rik hg DIVER, rhea! mid dates fet ote

rene ene exits ime ae setae Netty May Leet FOF ENCE Me Siete tended a) ey UN OME A Ted Ha Oh ie ener LIT EONS, REESE, i e > aay ay rea Sade . Pore eyaratyys eye nhs AEN VED AB SNM MN esh SA ate OH GY eaih oot aHh Maho albred ay ah sot abe oP pegs

a pial ties eins tes mt atece ca ras’ Htpeainhietla a> oil he vite De tS aad tata i RuguL eo urcngetios aS Ment ee Simy "OL ey yan dV eye Cd edge a te Sista Vehalpy ZW VS yy sehnin « hin ay ele awh did «

als «tied ta tage oe iin @ B04 Caters ewe fetls Sa wt ae BO A AMA ty bole a (2b cig tat t& IVS a Yas acy odd Pea ie i) y vy SV eg Py dite ans LNibh 0 Vy Ap babes Ve are lank CYTE att Bevta bakes e

ese vation Caner A Sais HE nt he oe ber ae wine ha Sig Fife er

CT ee er

NAS OTe Cd Me Case eI Tt wha et eet ae . . - eta Yoo ost a” bap saay't jt vee sy

Fa tag Ged eile 2% dct Wilae ha tWreed RENLS AS CHES VANES gy IS Ot Pope BANGED oe ADRES Bd mo Raksha redo hess hipaa ac erp toes a ced TLRs ae oT tae ner ee aha

aortas Be We aren Ge Meh eM Tray UT OEE NEY EAT a taiecd apt io wut e EE A NARESE UE N AE bare HE Mar TEN Aaah SERS ae ded SE et RON BRN bata 3 8 eam y AOE WI TAC MY: gonna’

oa) ategineterk ad Getni et ealNd Meso tae 2 ata edatetyead etna tate! eat eek ert Pe bab Na die BNW a ae DUES aa Cpa pha i ashe ee avin ASR Boreas A Dod HEY ash aeeaths

wee a Ba netted wees aet acti Be ete panei Ante BEES Nite L EL 5 Mg We FANTAIL Aeron eo ae ON in ey fen Fahad Loader

Johan AiG oAMLENDE Poe SOE PUN EI Cw aN Bates dis WLR YIAE Fava dy SURE SAO athe tas eWeek ee

te dd ) ‘ siete stp Sey aa SHV EMA PO OTN ney UA area

atta Wengt wtlve OOS Ae frursdme* cama RO De eae VR ee PA PO tt Si Retiree. Res aia ql . AME Daatiag ong He aie nym

Vy hedia dann koe Piel Rb NSD 2 Oia Mela ty Area Nn fae (pase Peavy REP “Wf se fete 2 g a r OVE AI ES Mg eR Rat ail a akg

doe,

Pane ta)

abv eyape

Vaveags

FIBRIN poy

LY MCs aad tek a eh as

mevw HE Yay Savadtaihy dls oth

SSeS tito VANS ENON SMURF Mae gle Ap oy

ENTER yy” get AVIALIAENS BG

Rerrecn tet WAM V Ay ry AE Nee ve

: ray aray

SM HN ot TNF Ses

MADR R elec Pein halla atau twee

Ve Met retary dy Peta neny

NM tanetiy oe

Sighs EY Lane eR ite SS pthg Bact? se ant ey

SoBe Valet po Ate pyr dle

Seg tlatee he Ca A Tenn aera lia gna te yg N,

zane fin settee

ROALD ARE nig Cnt EO Ty thas ths

Se ta tt oti wnedets ey neta gs Yoneda

Fa Le ot tak al

tatae es Las

Find od cee tin tlelled 2 19 eF OFL a?

Meg a mee cee et Hoop

INE ls ee re Med RN Fle

PP Hee pate et ee ee tier ee Hy ae

HYP NT drat ce eaee ef

VN te obs HACE tha ben veibs

Pe Gaye tgat

Ld een WEN

Parekh Gola t

Tetees alm

COI aed

Eee rs 4

Av ghd

Vie,

wre Sei

CE AY ty My

Sar ores eens

awe pe

LAr ea ern

PUY ENS evel

eR aN crept ve» fey

REMAN ATE I ae DAM dF Se ae

Te rece yore wt Deeg eae yn eh

Net ae gt

PACS eT aS es Aaah

aren Osaki ekg Se, < ~ ' « Fi , Sakae Tee ae gs ae § ? me WN ae yt

rab aigey i098 SHUT SDA Magis yd adam AE hI a AN Weer are ode it pe ied WEF Po iia Soh aes) ALON Veg ¥ UARGEL SM ds SNipctha HEAPS YN bees enw rae ache dee Oot Pi ripe seetied eneth bie

WEE LBD WD NEMEM G UZ E whe siake 1 a sett fy iy WaT SLPS EEE ANH EE RIND SPO PS MUR Ey Gere het a grain arene ee

' +. tg Pye By satis mdr Nyy Neate ibe dy oaeae

sere tig Pas eB nic bbc I) IRA EM et eos ieee ofth dod Sete ete Ltyty ay td Bevel fy Lae Us fs EwYS gto On gees a EY OMS Vee Sains oo co wraniey E: ai”? Recut de til pte

Bedale wget Nebr d olin temietls het eMenee mE EU tee SSRN I BY Uy VE inthe Me Kis NE Ate AT eM iosene 8 habasntbohas ies my Neem TMN gt dhs SIA mi soy BH ir asi

sep See Te vate Fat at a ip ate 24: Sas in Ma Meee ADRS Ast Paral nie HG IE SL A RIANA LEN IT aT ANN SH Oe ected shone t8 rete UNEASY noite NUL TEMA Feat Abia hha et ate

A wont ca me eRe Male hy eS ATM t este As ANE gE CNT MEN Ged alts That Nee IRA Ake Se oe BES LEN Fa gad oe batt hie peel

, a econ thndia She LEONEL CT CONT EM CMP Ian UPD Nae PLE aha GHEY at Spas eas» tay Nei Sey Py Vigil tera ee NC) ee

Wh rete Mia eae Ss wan Hilts es See NH a HEARS ID Dd etent dene ot arene on nfRagmafn regina

SOOT koa) rer hey

ad teal MEI ARES EO Day LE Ml ELD ye

SH teal nity SGN ae BFW Ms Oat hae nett eth

ep e MEA OV Bd Y Aer yity

nee alblntien

meine SP dhe

Dee lintin ye iy eT Dy

WOR ERNSD merreate ms

ne cin fe cl MBS

Deaediosete

Aptos UA TYNE

TVA eye le

Bey MEE Oph Siar

PEO M ATEN Ee tates an si DAN strat were ie

a wig

Leet alingel freer tite

Pie eh ret

PIES MY RG 5a Sry VION Te Se gD Ube alk

WN ETS mS

AGE ete rE Aa ®

Norv greed why

vee Fins

Tae

manny edt tl

Sale art Sch

Ve en et

Pe neti Sadi tt Nin Hate. Pet te See LY uzeyier Yar Ue rt aeriat ie yay hore

4 oe Beg FENG tate es fh Kegan cota “

o : ease AYN NS » poker e

ee MIME ets ee Ne wa TY.

MAHOU iciewlb oth oh ore

Sra at aay! ahh «

Vay Dia

INU ENT Sab Loy

WEL hin a in Ne tet a Ete

UT eae ty

SOD

he Ta td Wd tte aydun Spaeth Mikes) bia 28s a ie

taf So £9 TRIM ge aber pietecatey vga SOME AU VAY git Beg

evel fone agree SNP ING th i> ane ee ee a eee SOLE wt ot Stee te UNA

7 Sac where PNG ASET NL Beaver et

‘ae gai Fhe ere Patel Te athe

ye

eke bene

yet bbe fe

Oe ee ea

Sete ale ta a cet

LIN Ma PVG? eras

EET ayia ye

tesosy

We ASAE le 80

fea

eat ee

Protein heters aff ‘

are

NP Deg

4 Patt a aS

eh NG EAN

naa re tvac RES marofln ice

fing ek

RA A dt er as ei HS,

pares

SOE eaten FeO Ey Pgh

Seen tN

Biya eeF adc

had Va ae Pg

VOLE ponte

Ter cree.

Shs tahey Sah

ey gu hes engines

2 EN eS earn

Breit etingetiete eetcend rsfortiy ‘ an A Hage y

vacueinr ycdereneat cane banda baltn O36 42 br it srt ena wna xia nsfancaleniyilisderPoes viernes mee

2 A % PY SOR a ay Qpeanaroy. webairgrr ans HH" 2 ber ites rena

sie OR citi Vas eee Re Nake ee NED BMG Eee UY SSW ee MN heel a PUNE yg FD UH bM EBON GE » i Dsiytes ay ha Rare ew our ad

>

| (

mil 7

NH VOLUME 83

Number 1

Jour nal of the March, 1993

WASHINGTON

ACADEMY .. SCIENCES

ISSN 0043-0439

Issued Quarterly

at Washington, D.C.

CONTENTS

Articles:

DOUGLAS H. UBELAKER & AGNES ROUSSEAU, “Human remains from

hospital San Juan de Dios, Quito, Ecuador”

ed

VALERY F. VENDA, “Work efficiency vs. complexity: Introduction to

ergodynamics”’

i

ROBERT E. LLANERAS, ROBERT W. SWEZEY, JOHN F. BROCK &

WILLIAM C. ROGERS, “Human abilities and age-related changes in driving

performance”

a & S100 6s © (@,4 (Cale 2.0 @ 68 60 8 a) 6 6s a a = 010, 0's @ 0.0 © e006 0 6 © « 0 66 0 6 6 6 0 p06 00 6 0 20's

SMIT HSONIZ NV

NOV 0 2 1998

LIBRARIES

\

Washington Academy of Sciences

Founded in 1898

EXECUTIVE COMMITTEE

President

Stanley G. Leftwich

President-Elect

John H. Proctor

Secretary

Nina M. Roscher

Treasurer

Norman Doctor

Past President

Walter E. Bock

Vice President, Membership Affairs

Cyrus R. Creveling

Vice President, Administrative Affairs

Grover C. Sherlin

Vice President, Junior Academy Affairs

Marylin B. Krupsaw

Vice President, Affiliate Affairs

Thomas W. Doeppner

Board of Managers

James W. Harr

John H. Proctor

Herbert H. Fockler

T. Dale Stewart

William B. Taylor

James H. Donahue

REPRESENTATIVES FROM

AFFILIATED SOCIETIES

Delegates are listed on inside rear cover

of each Journal.

ACADEMY OFFICE

2100 Foxhall Road, N.W.

Washington, D.C. 20007

Phone: (202) 337-2077

EDITORIAL BOARD

Editor:

Bruce F. Hill, Mount Vernon College

Associate Editors:

Milton P. Eisner, Mount Vernon Col-

lege

Albert G. Gluckman, University of

Maryland

Marc Rothenberg, Smithsonian Insti-

tution

Marc M. Sebrechts, Catholic Univer-

sity of America

Edward J. Wegman, George Mason

University

The Journal

This journal, the official organ of the Washing-

ton Academy of Sciences, publishes original

scientific research, critical reviews, historical

articles, proceedings of scholarly meetings of

its affliated societies, reports of the Academy,

and other items of interest to Academy

members. The Journal appears four times a

year (March, June, September, and De-

cember). The December issue contains a di-

rectory of the current membership of the

Academy.

Subscription Rates

Members, fellows, and life members in good

standing receive the Journal without charge.

Subscriptions are available on a calendar year

basis, payable in advance. Payment must be

made in U.S. currency at the following rates:

WiSwand Canada)! s..2.0) ee $25.00

Other counties .25).45 2 (ee 30.00

Single copies, when available ....... 10.00

Claims for Missing Issues

Claims will not be allowed if received more

than 60 days after the day of mailing plus time

normally required for postal delivery and

claim. No claims will be allowed because of

failure to notify the Academy of a change of

address.

Notification of Change of Address

Address changes should be sent promptly to

the Academy Office. Such notification should

show both old and new addresses and zip

codes.

POSTMASTER: Send address changes to

Washington Academy of Sciences, 2100 Fox-

hall Road, N.W. Washington, DC 20007-

1199.

Journal of the Washington Academy of Sciences (ISSN 0043-0439)

Published quarterly in March, June, September, and December of each year by the Washing-

ton Academy of Sciences, 2100 Foxhall Road, N.W., Washington, DC, 20007-1199. Second

Class postage paid at Washington, DC and additional mailing offices.

Journal of the Washington Academy of Sciences,

Volume 83, Number 1, Pages 1-8, March 1993

Human Remains from Hospital San Juan

de Dios, Quito, Ecuador

Douglas H. Ubelaker

Department of Anthropology, National Museum of Natural History,

Smithsonian Institution, Washington, D.C.

and

Agnes Rousseau

Paris, France

ABSTRACT

Excavations conducted in 1988 and 1989 at the earliest formal hospital in Ecuador

(founded.in 1565) revealed two samples of human remains. This analysis focuses on second-

ary, disarticulated remains found within the church and skulls found within a related os-

suary. The church remains reveal all ages and both sexes, with little evidence of skeletal

disease and moderate levels of dental disease. The ossuary skulls also reveal minimal evi-

dence of disease, with males showing higher frequencies of dental caries than females.

Since 1973, the excavation and analysis of human remains from archeological

contexts in Ecuador has revealed a great deal about biological patterns of change

within human populations of that area. Initially this work focused on Indian

populations dating prior to the arrival of the Spanish (Ubelaker, 1980a; 1980b;

1981; 1983a; 1983b; 1988a; 1988b). More recently, research has expanded to

include colonial period remains from the city of Quito (Ubelaker, 1992; 1994).

Information gleaned from skeletal analysis supplements the historic record with

unique information on demography, disease and other variables that normally

would not be available.

Corresponding author: D. H. Ubelaker, Department of Anthropology, NMNH, MRC 112, Smithsonian

Institution, Washington D.C. 20560. telephone 202 786 2505, fax 202 357-2208.

1

2 UBELAKER AND ROUSSEAU

From November 1988 to June 1989, with support from the Instituto Ecuato-

riano de Obras Sanitarias (Section of the Health Ministry), excavations were

conducted at the “Hospital San Juan de Dios” in the central part of Quito,

Ecuador. The excavations were initiated in conjunction with architectural stud-

ies and restoration of the structure of the hospital.

The Hospital San Juan de Dios represents the first formal hospital established

in Ecuador. The hospital was founded on March 9, 1565, and was in use in the

old central part of Quito until about 1974. The facility included infirmaries and

storage rooms distributed around a northern and southern patio. From 1705 to

1736, the hospital was expanded by construction of a church, immediately east

of the northern patio. The church was built over the remains of a house which

belonged to a Spanish merchant named Pedro de Ruanes which itself was con-

structed on pre-colonial ruins. Stone walls with Inca characteristics were found

beneath the church floor in no stratigraphical relation to the burial pits. The size

of these walls suggests the existence of a large structure at this site. The ceramic

material associated with them was representative of the ““domestic-Inca” type

ware.

The church was built by the Bethlemitas, a religious order created in 1650 in

Guatemala by Father José de Betancourt, to administer hospitals and care for

the sick. Prior to their arrival, the hospital was managed by a charitable brother-

hood. Due to conflict among the brotherhood members, the hospital had many

problems. The Real Audencia President approached the Bethlemitas to become

involved to help solve these problems.

A cemetery is associated with the hospital and dates from about 1565 to about -

1800.

Excavations clearly identified two burial areas within the complex: an ossuary

located outside the church and a group of individual skeletons within the

church. The ossuary was located within one of the storage rooms (Bodega D)

immediately west of the church, between it and an area of the hospital complex

called the northern patio. The bones were concentrated within the general fill of

this area. Archeological evidence suggests that the ossuary was filled in with the

bones, bricks, soil, and other construction materials after the wall of the north-

ern patio was destroyed by the earthquake of 1755. It is likely that the bones

originated from skeletons that had been buried within the same space between

1705 (date of initial construction of the church) and 1755. Presumably, the areas

in and around the church must have represented privileged sites for burial.

Two radiocarbon dates are available for the ossuary. Both were collected at

the time of excavation and sent, with support from the Smithsonian Institution,

to Beta Analytic Inc. of Miami, Florida. A sample of human bone was dated at

60 years plus or minus 60 years (essentially modern) and thus was obviously

HUMAN REMAINS FROM SAN JUAN DE DIOS 3

contaminated. A more reliable charcoal sample was dated at 160 years, plus or

minus 50 years. While the charcoal date should be relatively reliable, it likely

represents a terminal date for use of the ossuary. All of the remains were disarti-

culated which indicates they had been removed from another location (probably

in the same area) and reburied. The evidence suggests that the remains represent

not pre-contact Indians, but early colonial period individuals, most likely dating

between 1705 and 1755. No diagnostic artifacts were found in association with

the bones.

The remains were generally poorly preserved and fragmented upon removal.

All bones were disarticulated, i.e., no bones were in anatomical order. The

individuals had been located in the primary repository long enough for all soft

tissue to decompose, allowing the bones to disarticulate during the transfer to

the ossuary. No cut marks or other indicators of defleshing were noted.

Excavation of the floor of the church also revealed six roughly rectangular pits

containing skeletons. The pits were located around the periphery of the church

floor and were arranged symmetrically in line with the axis of the church. All of

these pits contained at least one articulated skeleton (one contained two). Disar-

ticulated bones were found in the fill above the skeletons within four of the pits.

Many immature individuals were present. Since not all of the area along the

walls inside the church was excavated, possibly other pits and skeletons remain.

The skeletons likely date between 1710 (the supposed date of completion of

construction of the church) and 1810, the date of the War of Independence and

the conversion of the hospital to a public facility.

The similar positions of the articulated skeletons within the church (one hand

on the opposite shoulder and the other one on the stomach area) and their small

number suggest they may have been high-ranking Bethlemitas. Unfortunately,

historical information about their burial customs is lacking. No remains of

wooden coffins or other associated artifacts were recovered. The origin of the

disarticulated bones found within the fill of the pits remains unknown. Burial of

the general public associated with the hospital was in the main cemetery, now

located beneath recent construction outside the southern part of the church.

In contrast to the bones of the ossuary, those from the church were well

preserved. All were disarticulated and removed for analysis except one, which

was preserved intact and removed within a block of soil for exhibition purposes.

In August of 1990, many of the bones were washed and dried in preparation for

analysis. Due to limitations of time, only the secondary samples from the church

and the crania recovered from the ossuary were studied. Most of the crania were

not sufficiently intact to allow reliable assessment of population origins. How-

ever, one elderly male ossuary skull (number two) appeared to be of European

origin.

A UBELAKER AND ROUSSEAU

Table 1.—Measurements and observations, Church

Male Female

Auricular height 106

Porion-bregma 105

Cranial length 170

Cranial breadth 143 137

Basion-bregma 124

Minimum frontal breadth 96

Nasal height 53

Nasal breadth 21

Maximum alveolar length 50

Maximum alveolar breadth 60

Palatal length 45

Palatal breadth 39

Bigonial breadth 94

Height of ascending ramus 67

Minimum breadth of ascending ramus 26

Secondary Sample from the Church

The bones from the secondary sample in the church represented at least 18

individuals, four adults and 17 immature individuals. Three adults are indicated

by the presence of three right radii, left and right femora, left fibulae, nght

temporals, gladioli of the sternum, innominates, and calvaria. Two individuals

are represented by the humeri, left radius, both ulnae, right clavicle, left tem-

poral, both maxillae, mandibles, patellae, thoracic vertebrae, sacra, left greater

multangulars of the hand, right first and fifth metacarpals, the second, third, and

fourth metacarpals, proximal hand phalanges, right calcanea and right ribs. A

single individual is represented by the tibiae, right fibula, scapulae, manubrium,

second cervical, other cervicals, the hand naviculars, right lunate, right capitate,

left first and fifth metacarpal, middle hand phalanges, tali, and the first through

fourth metatarsals.

Morphology of the innominates indicates that at least one male and two

females are present. Since morphology of the three calvaria indicates the likely

presence of one female and two males, at least four adults are probably present,

two males and two females.

Morphological assessment of the symphyseal surfaces of the pubic bones

suggests ages at death of 40 to 50 years and 50 to 70 years for the males and 40 to

50 years and 38 to 45 years for the females.

At least 15 immature individuals are indicated by the left humerus, 14 by the

femora, 12 by the right humerus, 10 by the left side of the mandible, eight by the

right scapula, left temporal and right mandible, seven by the right ilium, six by

the night ulna, left scapula and the left ilium, five by the left radius, left ulna, left

tibia, both clavicles, and left maxilla, four by the right radius, left fibula and ribs,

HUMAN REMAINS FROM SAN JUAN DE DIOS 5

Fig. 1. The ossuary at San Juan de Dios.

three by the right fibula and vertebrae, two by the left ischium and right pubis,

and one by the right maxilla, carpal and tarsal bones.

Ages estimated from long bone lengths indicate that 17 immature individuals

are present. Their estimated ages are four to five years, three years, 1.5 years, two

at one year, two at about 10 months, one at six months, four at about three

months, and five at newborn.

Table 2.—Measurements and observations, Ossuary

Male Female

Mean _ s.d. Range n Mean sd. Range

5

Auricular height

3 119 2 111-124 1 113 113

Porion-bregma 3 123 16.1 110-141 1 110 110

Cranial length 5 183 11.9 164-197 2 172 3.5 169-174

Cranial breadth 4 137 0.5 136-138 2 130 2.8 128-132

Basion-bregma 2 133 14.1 123-143

Minimum frontal breadth 6 96 2.3 92-98

Nasal height 4 33 1.0 52-54 1 44 44

Nasal breadth 4 26 1.4 24-27

Maximum alveolar length 3 33 13 53-56 1 51 51

Maximum alveolar breadth 4 62 Paes 63-68 1 61 61

Palatal length 3 47 2.6 44-49 | 40 40

Palatal breadth 3 42 2.0 40-44 1 38 38

Height of ascending ramus 1 64 64

Minimum breadth of ascending ramus _ 1 32 32

6 UBELAKER AND ROUSSEAU

The only evidence of bone pathology in this sample is a well remodeled

periosteal lesion on the posterior surface of the distal end of an adult night femur.

The affected area is located about 64 mm from the distal end. The distal articu-

lar surface also appears to be involved since it displays irregular bone deposits

and pits. Infection is the likely cause of the bony changes.

Four cervical vertebrae show stage one osteophytosis. Of 11 thoracic verte-

brae, eight are stage one and three are stage two. All six lumbar vertebrae show

stage one. These changes likely indicate normal age changes rather than disease.

Living stature can be estimated only for three adult individuals from femoral

lengths. Left femoral lengths of 473 mm and 490 mm (estimated) suggest living

statures of 174 cm and 178 cm respectively. One right femoral length of a

different adult suggested a stature of 166 cm. All of the femora likely originated

from males and all statures were calculated using Trotter’s formulae for White

males (Trotter, 1970; Ubelaker, 1989).

Only 32 adult teeth from three individuals were recovered from the church

sample. Of 70 observations of the presence or absence of permanent teeth, 38

(54 percent) were absent antemortem. Only two (3 percent) of 30 teeth were

carious. The two carious lesions were located on the mesial surface of a first

maxillary molar and on the mesial surface of a second maxillary molar. One (3

percent) of 33 teeth showed associated alveolar abscesses. The large percentage

of teeth lost antemortem reflects the cumulative effect of untreated caries, peri-

odontal disease and trauma.

Dental calculus was concentrated more on the lingual surfaces of the teeth

than on the buccal surfaces. Of the buccal surfaces examined, five showed no

deposits, 24 slight, one moderate, and two large. Lingual surfaces showed four

absent, 20 slight, five moderate and three large.

Only one tooth, a maxillary right central incisor, displayed linear enamel

hypoplasia. This lesion was located about 7 mm from the crown root junction

and likely formed about the age of three years.

Only seven deciduous teeth are present, maxillary left and right canines, left

maxillary first and second molars, and the left mandibular canine, and first and

second molars. None of these teeth show evidence of caries, alveolar abscess,

calculus or enamel hypoplasia.

Cranial and mandibular measurements are summarized in Table 1. Since

measurements were possible for only one male and one female, summary statis-

tics are not needed. Evidence of cranial deformation was not detected.

Observations on discrete cranial and mandibular traits were possible primar-

ily on one male and one female. Frontal grooves were present on one male and

absent on one female. All other observations were negative.

“HUMAN REMAINS FROM SAN JUAN DE DIOS 7

The Ossuary Sample

Of the ossuary material, nine crania were available for analysis. Additional

remains had been recovered but had not been thoroughly cleaned and prepared

for data collection. All of the crania originated from adults: seven males and two

females, all undeformed. Ages estimated for the males are 23 to 28, 55 to 60, 40

to 50, 50 to 60, and three between 30 and 35. Ages for the females are 30 to 40

and 20 to 30. The mean age of males in the cranial sample is 40 years, but only

30 years for females. Ages were estimated from the extent of cranial suture

closure and dental observations. Table 2 summarizes measurements of the os-

suary crania.

Dental data are available for only one female cranium, No. 5. Eight maxillary

teeth reveal no carious lesions and no alveolar abscesses. All teeth show slight

— calculus on both the buccal and lingual surfaces. No hypoplasia was noted.

The seven male crania presented 59 teeth of which 10 (16.9 percent) were

carious. All carious teeth were molars. Of 98 observations on teeth lost antemor-

tem, five (5 percent) had been lost. Of 97 observations on alveolar abscess, only

three teeth (3 percent) had associated abscesses.

Calculus was minimally present. Scores for buccal tooth surfaces were 14

absent, 35 slight, 11 moderate, and no large. Lingual scores were 14 absent, 45

slight, one moderate and no large. No examples of hypoplasia were noted.

Summary

Although the human remains from the ossuary and the later secondary

church sample are small, they add new perspective to the historic information

already available about human biology of the colonial populations in Ecuador.

Comparisons between the samples reveal slightly higher adult life expectancy for

both males and females in the church sample. This greater life expectancy

1s accompanied by greater tooth loss in the church sample (54 percent vs 5

percent).

The church sample also reveals a lower caries rate (3 percent vs 10 percent)

and similar rates of dental abscess (3 percent). No hypoplastic teeth were found

in the small ossuary dental sample, but three percent of the church teeth were

hypoplastic. In general, the dental disease frequencies fall within the range previ-

ously reported for skeletal populations within Ecuador and perhaps suggest

greater diversity during the historic period than previously thought.

Comparison between these two samples is complicated by lack of detailed

information about the populations they represent. In general, the hospital and

8 UBELAKER AND ROUSSEAU

church serviced the underprivileged populations of urban Quito. Those buried

directly within the church as single interments may represent higher status indi-

viduals, perhaps those with administrative positions within the church. The

secondary deposits found within the church and reported here are of unknown

population origins.

Although the sizes of these two samples are small, they contribute to the

growing information on biological change within ancient Ecuador. It is hoped

that these data can be augmented through future analysis of the additional

remains already excavated, and from new excavations at San Juan de Dios and

other historic mortuary sites in the area.

References

Trotter, M. (1970). Estimation of stature from intact limb bones. In T. D. Stewart (Ed.), Personal Identifica-

tion in Mass Disasters, (pp. 71-83). Washington: Smithsonian Institution.

Ubelaker, D. H. (1980a). Human Skeletal Remains from Site OGSE-80, A Pre-ceramic Site on the Sta. Elena

Peninsula, Coastal Ecuador. J. Wash. Acad. Sci., 70, No. 1, 3-24.

Ubelaker, D. H. (1980b). Prehistoric Human Remains from the Cotocollao Site, Pichincha Province, Ecua-

dor. J. Wash. Acad. Sci., 70, No. 2, 59-74.

Ubelaker, D. H. (1981). The Ayalan Cemetery: A Late Integration Period Burial Site on the South Coast of

Ecuador. Smithsonian Contributions to Anthropology 29. Washington, D.C.

Ubelaker, D. H. (1983a). Human Skeletal Remains from OGSE-172, an Early Guangala Cemetery Site on the

Coast of Ecuador. J. Wash. Acad. Sci., 73, No. 1, 16-26.

Ubelaker, D. H. (1983b). Prehistoric Demography of Coastal Ecuador. National Geographic Society Research

Reports, 15, 695-703.

Ubelaker, D. H. (1988a). Human Remains from OGSE-46, La Libertad, Guayas Province, Ecuador. J. Wash.

Acad. Sci., 78, No. 1, 3-16.

Ubelaker, D. H. (1988b). Prehistoric Human Biology at La Tolita, Ecuador, A Preliminary Report. J. Wash.

Acad. Sci., 78, No. 1, 23-37.

Ubelaker, D. H. (1989). Human Skeletal Remains. Excavation, Analysis, Interpretation, second edition. Wash-

ington: Taraxacum.

Ubelaker, D. H. (1992). (abstract) Patterns of Biological Change in Ancient Ecuador. American Journal of

Physical Anthropology, Suppl. 14, 165.

Ubelaker, D. H. (1994). The Biological Impact of European Contact in Ecuador. In Larsen & Milner (Eds.) In

the Wake of Contact: Biological Responses to Conquest. New York: Wiley-Liss.

Journal of the Washington Academy of Sciences,

Volume 83, Number |, Pages 9-31, March 1993

Work Efficiency vs. Complexity:

Introduction to Ergodynamics'

Valery F. Venda

Department of Mechanical and Industrial Engineering, University of Manitoba,

Winnipeg, Canada, R3T 2N2

ABSTRACT

Ergodynamics is proposed as a theoretical foundation and practical method for studying

and improving work efficiency in dynamic environments. Ergodynamics is based upon three

laws termed: 1) mutual adaptation; 2) plurality of functional work structures; and 3) trans-

formations. Work efficiency and complexity are interpreted as opposite criteria measured

with similar units. Analysis of correlations between criteria and factors of efficiency and

complexity may help to model work functional structures and to predict dynamics of trans-

formations between the structures. Recommendations to increase both work efficiency, and

the practical usefulness of the laboratory testing of products and workstations are given.

Ways to avoid losses and increase profits while upgrading software and technology are

suggested.

The Britannica World Language Dictionary (1954, p. 419) defines efficiency

as “1. The character of being efficient, effectiveness; 2. The ratio of the work

done by an organism or machine to the amount of food or fuel consumed and to

the energy expended. “The same Dictionary (p. 277) defines a complexity as

“the state of being complex; something complex”; and complex as “1. Consist-

ing of various parts or elements; composite; 2. Complicated; involved; intricate,

something composite or complicated”. The book by Streufert and Swezey

(1986) became a classic in studies and teaching theory of complexity in organiza-

tional psychology and management. We have used the theory by these authors

as a basis for analysis of complexity dynamics in decision making processes

(Venda and Venda, 1994).

"This paper is dedicated to fond memory of Yuri V. Venda (1969-1991) who discovered the Law of

Transformations.

Author appreciates the fruitful discussions and editing of this paper by Dr. John J. O'Hare and Dr. Robert

W. Swezey.

These studies were supported by Northern Telecom Ltd., Bell-Northern Research, and Natural Sciences and

Engineering Research Council of Canada.

10 VENDA

The terms “efficiency” and “complexity” are often combined, as in the saying

“this task is very simple for you (me), thus you (I) can do it easy, quickly,

effectively.”’ Or in another case, “‘this task 1s too complicated, and that is why

you spent so much time and made so many mistakes.” Time spent and mistakes

made are used as criteria of complexity: if more time and mistakes, then higher

complexity.

Productivity, work tempo, and the number of tasks solved in a certain

amount of time are used as criteria of efficiency. An inverse correlation occurs

between efficiency (what one has obtained) and complexity (what one has

spent). For example, efficiency may be measured as the number of a student’s

correct answers on an exam and as the number of wrong answers. Thus the

complexity criterion value (wrong answers) will supplement the efficiency crite-

rion value (correct answers). Higher complexity leads to lower efficiency; higher

efficiency leads to lower complexity. This concept of efficiency and complexity

has been applied successfully in many ergonomic and psychological studies and

projects (Lomov and Venda, 1977; Savelyev and Venda, 1989; Venda, 1975,

1980, 1990).

The concept offers an operational method for measuring and comparing

efficiency and complexity using the same units. Although it may be convenient

in design and ergonomic practice to increase efficiency, but there are many

situations when it is easier to find what causes complexity and influences on

complexity, for instance by optimizing the use of information displays, hard-

ware, and software (Venda, 1975, 1982). In industry, higher efficiency is gener-

ally considered better (i.e. more products and higher quality). In science and

education, higher efficiency may be of special interest as a way to success in

research, understanding of a new field, and/or acquisition of skills and knowl-

edge. Karwowski and Ayoub (1984) have suggested application of fuzzy set

theory to assess the stress and complexity of manual lifting tasks. Karwowski

and Mital (1986) have expanded application of fuzzy set theory to the main

areas of ergonomics; and Karwowski, Marek, and Noworol (1988), and Kar-

wowski (1991) have worked out a general approach to the theory of ergonomics

and complexity based upon fuzzy set theory and categories of entropy and

ergonomic incompatibility.

In our own work, we have tried to work out an operational theory of efficiency

and complexity of human work as well as human-machine-environment mu-

tual adaptation (Venda, 1975, Venda and Venda, 1991). In this view, complex-

ity depends on internal functional structures (work skills) and their mutual

adaptation with the external environment. Loss of efficiency, when a work task

and environment are constant, may mean there have been changes in the hu-

man internal state. Grandjean (1988) stated that fatigue is invariably associated

STUDIES OF EFFICIENCY AND COMPLEXITY 11

with ‘“‘a loss of efficiency and a disinclination for any kind of effort’ (p. 156).

This loss of efficiency is considered to be the result of mutual dysadaptation of

human internal components (subsystems, organs), and thus increases work

complexity, while efhciency and complexity are characteristics of human-en-

vironment mutual adaptation (Venda, 1975). Before the issue of efficiency and

complexity is addressed, users, goals, and functional structures need to be deter-

mined. If there is no goal, then complexity may not exist. Further, the same goal

(task) may be associated with different complexity criterion values for different

users using different functional work structures. Work is regarded generally, as

any kind of goal oriented human performance. For example, if students or

conference attendees are shown a control board for a power plant, and asked

whether it is complex, the question is improper. When persons do not have

some concrete task for reference, they cannot determine complexity. Ifa student

is not interested in watching the control board and no human-board contact

happens, there is no complexity. However, the same control board may present

the power-plant operator an emergency task of very high complexity; and task

complexity in that situation could be assessed as an obstacle to reaching maxi-

mal efficiency of the human-machine system. The higher the complexity (the

greater the obstacle on the way to a goal) the lower the work efficiency. Higher

complexity leads to efficiency losses, and thus to lower real efficiency. The need

to define human goals in all studies of the human-environment interaction has

been analyzed and demonstrated for many situations by Leontiev (1971), and

by the many Russian psychologists who were followers of the psychological

theory of human goal-oriented activity.

Work Efficiency and Complexity Criteria

Any work-output level to be increased may be used as a criterion of efficiency.

The criteria of efficiency may not only be engineering result, but work satisfac-

tion, health protection, and even happiness can also be considered as efficiency

criteria. The ergodynamics approach applies both to qualitative and quantita-

tive analyses of work processes and to the functioning of complex systems.

Efficiency is a positive measure of work and living processes. In an efficient

system, obstacles, difficulties, errors, and deviations from optimal work pro-

cesses and algorithms are minimal. These negative aspects (preferably measured

with quantitative measures) are criteria of functional complexity.

Functional efficiency and complexity criteria are opposites. If higher work

efficiency is attained, it automatically means that complexity has been lowered.

If some part of potential efficiency is lost, it means that functional complexity

12 VENDA

has become higher. Typical criteria used to assess complexity are extra time

spent for a given amount of work, number of errors committed, frequency of

defective products, and probability of a wrong decision (failure). If one measures

the time spent on the creation of a product, and determines the productive part

of that time, the remaining time defines a criterion of complexity. Functional

complexity may be measured by any criterion reflecting losses to be minimized,

and may be converted into efficiency, because less complexity is more eff-

ciency, and vice versa.

For example, if the probability of correct decisions is p,,, (aS a criterion of

efficiency), then the probability of wrong decisions p,,,, 1S Pyro = 1 — Door. Thus,

Q.e + Cre = 1, where Q,,; is a relative criterion of efficiency and C,,, is a relative

criterion of complexity. In absolute values, total productivity is Q,,, (number of

items produced), the effective productivity is Q (a number of quality items), and

the complexity of production is C (the number of defective items). Obviously, Q

fae Sa OD Q/ Qtot ai C/ Qtot 7" Qrot/ One Ox + Cre 7 i because Q/ Qrot is a relative

efficiency, Q,., and C/Q,,, is a relative complexity C,,,. Instead of addition of

efficiency and complexity criteria values, multiplication is used in many cases.

For example, productivity (a number of items produced) in a time unit (effi-

ciency, Q) could be multiplied on time spent on production of one item (com-

plexity, C), thus Q k C = 1, Q = 1/C, C= 1/Q.

Qualitative and Quantitative Analysis of Efficiency and Complexity

Some authors limit ergonomic analysis of human-machine-environment in-

teraction to qualitative methods, for example as skill-based, rule-based, and

knowledge-based behaviors (Rasmussen, 1986). However, a combined qualita-

tive-quantitative approach is supported by many leading scientists (Hendrick,

1992; Sheridan, 1992). Streufert and Swezey (1985) have shown the advantages

of a combined descriptive and predictive methodology, and mathematical mod-

els based on complexity theory for analysis of dynamic management decisions.

They stressed that full dedication of any scientist to only quantifiable predic-

tions limits the possibility of success to a relatively narrow area in real manage-

ment problem-solving processes and cases. If descriptive methodology is used

without a strong theory of development dynamics, quantification is like looking

through rear-view mirrors. One cannot drive safely using this kind of informa-

tion when the road makes sharp zigzags and becomes crowded. Thus, attempts

to utilize previous dynamics for the prediction of future events, based exclu-

sively on simple linear, or monotonic exponential models not only do not help

in organizational and technological control decisions, but may lead to serious

STUDIES OF EFFICIENCY AND COMPLEXITY 13

mistakes. In spite of the partial successes of mathematical decision-making

theory applications, described for example by Dickson (1983), there are many

negative results in practical use of quantitative predictive models in organiza-

tional spheres.

There is a very important difference between problem-solving processes in

organizational, management activities, and in human operator performance.

Even for the most complex technological control-system, in either a case of

emergency or in a normal situation, a human operator can make a successful

decision if the operator’s psychological model is adequate to the state and dy-

namics of the real object. If the operators at the Three Mile Island Nuclear

Power Plant during that famous accident were able to synthesize an adequate

model of the current events, in the short time allowed by quick dynamics of the

control processes, elimination of the emergency would have been a trivial task of

manipulating several control buttons and handles. All information needed, for

successful decision making for any technological control object, virtually exists

in the object. The problem is to find that information, extract its most important

features, display 1t to the human operators in a volume and structure appro-

priate for their knowledge, skills, psychophysiological state (i.e. possible high

stress), and cognitive strategies.

Decision makers in organizational systems, typically have more time than do

decision makers in emergency situations, but they, in principle, do not typically

have access to large portions of the information needed for a decision, because it

is often located outside the organization, among competitors, world market,

political institutions, etc. Despite those difficulties, Streufert and Swezey (1985),

have created both a strong theory and concrete practical methods for observing,

quantifying, and measuring structural characteristics of successful organiza-

tional decisions. They noticed that rational decisions need not always be based

on mathematical models or arithmetic calculations. Streufert and Swezey

(1985) proved that rationality and irrationality of decisions can be understood

more widely than in a dictionary definition. In Webster’s Dictionary (1976), the

term rational, (p. 1885), means (1) having reason or understanding, (2) of,

relating to, or based upon reason, (3) involving only multiplication, division,

addition and subtraction and only a finite number of times, (4) agreeable to

reason: intelligent, sensible, (5) capable of being measured in terms of mora in

Greek and Latin prosody: having the normal ration between argis and thesis.

Some authors may have considered this meaning when they suggested that

many corporations have, on occasion, reached great success via irrational deci-

sions (Streufert and Swezey, 1985). This is an important difference between

technological control-decisions during emergencies and organizational, manage-

ment, economic decisions when relating to competition. More detailed surveys

14 VENDA

on decision making complexity are found in these reports (Bodrov and Venda,

1992), (Streufert and Swezey, 1985), (Strickland, 1991), and (Venda and Venda,

1994) ). Even when a technological control-decision is not possible and accept-

able, there is some family of appropriate decisions, and every decision should be

compared with the real state and the dynamics of the control system, in order to

be adequate, and understandable for all participants (and, where appropriate, to

members of the commission that will assess the decisions if damages occur).

Of course every manager should be able to explain a decision to the Board of

Directors, Chairman, and shareholders. But the explanation occurs later in

time. Irrationality of an organizational decision means it should be sudden and

therefore unpredictable to competitors. If competitors can predict decisions and

responses on their actions (e.g. implementation of a new product and model, or

decreasing prices at 15, 25, and 40%), and take strong preventive steps, a deci-

sion will be wrong and may lead to bankruptcy. This scientific problem is

named conflict between systems or conflicting structures (Lefebr, 1971). Our

discussions herein, is limited to formalizing, quantitatively assessing efhiciency

and complexity, and examining the processes of problem solving and decision

making in technological control-systems and in industrial companies.

With the creation of a practical theory and methods of modeling organiza-

tional decisions, successful and experienced managers could use it to predict and

parry competitors’ decisions. Decisions that cannot be directly predicted and

explained on the basis of well-known theory may be qualified as “‘irrational”’

and may be successful with a higher probability than “‘rational” decisions.

To attain higher success, managers may waive traditional methods of organi-

zational decisions, but operators, on the contrary, typically need to follow stan-

dard technological control-decisions based on the current state and dynamics of

the control system. This difference could serve to stimulate quantitative predic-

tive decision-making theories in the human-machine-environment systems but,

in certain aspects, hamper similar studies in the organizational sphere.

There are not many studies on this problem, either in organization and man-

agement, or in ergonomics, human factors and engineering psychology. Studies

by Streufert and Swezey (1985), are among the few oriented to complex qualita-

tive and quantitative analyses of organizational decisions. Although one can

find statements in the literature about various problems associated with mathe-

matical decision theories (see very fundamental analysis and survey by Streufert

and Swezey, 1985), only a few scientists have studied control and technological

decisions using the complexity-based approach. Unfortunately, some of them

have met with difficulties, and therefore have concentrated their attention on a

single method. For example, after many years of quantitative studies of manual

control and decision making, J. Rasmussen (1986, 1989) has limited his studies

STUDIES OF EFFICIENCY AND COMPLEXITY 15

to strict qualitative analysis of skill-based, rule-based and knowledge-based

(SRK-models) human-operator behaviors. Qualitative models are not sufficient

even for “‘irrational’’ organizational decisions. But to improve efficiency, and

especially the safety of technological objects like nuclear power plants, decisions

concerning control, design and training, and on the processes of mutual adapta-

tion in human-machine-environment systems, should be based upon both de-

tailed and quantitative models, and mandatory qualitative analyses in order to

avoid principal mistakes.

Another difference between organizational and technological control-deci-

sions is the problem of repetitiveness. When an emergency situation recurs then

the same decision that was made earlier can sometimes be used. However, an

organizational decision that earlier led to success could lead to undesirable

outcomes even though the situation is the same. Hendrick (1986a,b) has studied

relations among cognitive complexity and optimal organizational and work

system design. Harvey, Hunt, and Schroder (1961), and Harvey (1963), have

also found that cognitive complexity levels underlie differences in how persons

conceptualize reality and, hence, strategies for reacting to changing or novel

situations. Barrif and Lusk (1977) have related cognitive complexity levels of

operators to the (ergonomic) design of management information systems.

Stamp (1981) has related human-complexity level to job-complexity level, and

suggested guidelines for joint optimization. These studies were continued by

Hendrick (1986a,b) with hotel managers.

Hendrick (1992) has suggested that success of any complex human-machine

system is contingent on its ability to adapt to its external environment. In

open-system terms, organizations require monitoring and feedback mecha-

nisms to follow and sense changes in their relevant task environments, as well as

the capacity to make responsive adjustments. For many organizations, telecom-

munications systems, and the adequacy of their ergonomic design are critically

important components of this adaptability. Of particular importance is the fact

that specific task environments vary along two dimensions that strongly influ-

ence the effectiveness of an organization’s macroergonomic design, i.e., their

degrees of environmental change and the environmental factors which affect

their human performance-complexity. Degree of change refers to the extent to

which a specific task environment is dynamic or remains stable over time;

degree of complexity refers to the number of relevant task environments. In

combination, these two dimensions determine the environmental uncertainty

of the system. In general, the greater the environmental uncertainty, the greater

is the need for work system design and related human-machine interfaces to

allow for, and support, operator flexibility within different functional structures.

Many interesting studies on the interdependence of work factors and eff-

16 VENDA

ciency criteria have been conducted in both former Soviet, and current Russian,

psychology and ergonomics fields (see surveys by Bodrov and Venda, 1992,

Strickland, 1991, Venda, 1990, Zarakovski et al., 1977, and Zinchenko and

Munipov, 1989).

A Paradox of Complexity

Many authors have emphasized the complexity of environment, object, and

information display systems, independent of human goals, functions, and work

processes. This perspective, however, is in error. Every real object has an endless

number of parts and particles, and therefore complexities of real objects cannot

be compared with real environments. From the point of view of a scientific

paradigm, if something cannot be observed, measured, or compared, it does not

exist as a subject of scientific interest. Therefore, a paradox exists 1.e. since real

environment complexity is endless, an environment itself does not possess com-

plexity. 2

Complexity is defined as a characteristic of human activities and attitudes,

toward achieving concrete goals. An environment could be friendly and support-

ive or hostile and complex, depending on the human goal, task, functional

structure, or generally speaking, the process of human-environment mutual

adaptation. Efficiency and complexity thus are not viewed as characteristics of

the environment, but of human performance, interaction, and adaptation with

the environment. Therefore, the proper scientific focus is on environmental

factors of eficiency and complexity, instead of environmental complexity, per

se. Also, if one cannot establish a criterion of efficiency in any system, it is

useless to try to find a criterion of complexity, and vice versa. If it is imprecise to

say “environmental efficiency” then the term “environmental complexity”

should also be avoided. It is, however, very important to study the environmen-

tal factors of eficiency and complexity in human performance. These factors

are critical parameters of the mutual adaptation human-environment processes

relevant to human-performance goals.

The complexity and efficiency of software, hardware, or graphic information

displays cannot be assessed if human functional-structures and tasks are un-

known. The same information structure may be effective (less complex) in a

normal control situation, but ineffective (more complex) in an emergency-con-

trol situation. The satisfying sight of a rain forest, full of trees and bushes, can be

reduced if one worries about the overall destiny of rain forests. When the com-

plexity of a task is low, an environment requires a small investment of time. The

same forest could increase in complexity should one attempt to cross it at night.

STUDIES OF EFFICIENCY AND COMPLEXITY 17

This example is of special interest, because it demonstrates the ease of assessing

complexity criteria such as expenditures of time, money, and human and tech-

nical resources; and the difficulty of agreeing on criteria of efficiency. If criteria

of complexity exist, however, there also exist criteria of efficiency (e.g. productiv-

ity, savings of money and human resources, probability of successful solutions

at limited time intervals, etc.).

In theory, complexity can be measured as the inverse of efficiency, which may

include: (1) a total productivity; (2) percentage of a quality product; (3) number

of correct actions; (4) probability of successful decision in a limited time; or (5)

profit. Respective measures of complexity would include: (1) loss of productiv-

ity; (2) percentage of a non-quality product; (3) number of errors (incorrect

actions); (4) probability of failure; or (5) monetary losses.

Quantitative Measures of Efficiency and Complexity

Measuring efficiency and complexity with the same units is convenient for

ergonomic and psychological practice, and for the improvement of work envi-

ronments, tools, and skills. This also means that the same factors of human-en-

vironment mutual adaptation may be considered as factors of efficiency and

complexity. Indeed, a bell-shaped curve has two sides. When factor values are

less than optimal for a certain functional structure, increasing those factor val-

ues leads to increased efficiency. In other words, the correlation between such

factors and efficiency criteria is positive. It is therefore logical to consider such

factors as measures of efficiency. If factor values increase above an optimal value

for this functional structure, then efficiency will decrease and complexity will

increase. Thus, factor values yield a negative correlation with efficiency, but a

positive correlation with complexity. Such factors could thus be considered, in

this case, as factors of complexity. So we may view these factors as indicators of

efficiency-complexity or of mutual adaptation.

Since decreasing complexity translates to increasing efficiency, in certain

practical situations, it may be convenient to maximize one or the other. If work

productivity is simply defined and measured, then an ergonomists’ attention

may be concentrated on increasing efficiency. However, if losses of a product

occurred, due to human error, then decreasing complexity (number of errors)

may be more easily accomplished rather than increasing the total number of

products, decisions, or actions.

Correlation coefficients between practical (industrial) criteria of work efh-

ciency-complexity and ergonomic factors of efficiency-complexity are asso-

ciated with the quality of factors (that is, which factors were chosen, the magni-

18 VENDA

tude of their correlation coefficients with criteria, etc), quantity of factors (how

many factors were analyzed), and methods for the measurement of the factors.

The range of factors of eficiency-complexity could thus be extended to find the

desired level of statistically-valid activity descriptions (Venda, 1990).

Fundamentals of Ergodynamics

In a paper dedicated to fundamental theoretical problems of ergonomics W.

Karwowski (1991) recalled that Wojciech Yastrzebowski, established in 1857

this name combining two Greek words (for work and natural laws), hoping that

future generations would discover the laws of the prospective science. The infa-

mous Russian psychologist and psychiatrist Vladimir Bekhterev organized the

first conference on ergonomics in 1921, (he named it “‘ergologia’’) and stressed

the necessity to study the laws of work and ergonomics (Zinchenko and Muni-

pov, 1989). However, to date we have been unable to find in the ergonomic

literature, any attempts to specifically state laws of ergonomics.

After publication of our paper on transformation dynamics theory and et

of transformations (Venda and Venda, 1991), and presenting an address at the

36th Annual Meeting of Human Factors and Ergonomics Society (Atlanta, GA,

October 1992), many valuable suggestions on this issue were received from the

colleagues. The main suggestions were: (1) the theory should be based on three

principles (fundamental laws) of transformation dynamics; (2) use simple ex-

periments to test and demonstrate the laws; (3) apply transformation dynamics

specifically to work-dynamics (ergodynamics) analysis and optimization, to en-

able more practitioners to benefit from it; and (4) carefully analyze basic ergo-

dynamics categories such as efficiency and complexity of work.

We began our efforts by reviewing the scientific work analysis history. In

1908, Taylor (1971), began to study work efficiency as a function of the dy-

namics of the work environment. In one effort, he manipulated shovel weights

systematically in order to achieve maximal work-productivity. Taylor’s discov-

ery was that work productivity, Q., is a bell-shaped function of the work environ-

ment, F (in his case, shovel weight). The function Q.(F) is modeling a work

functional structure, S, = Q,(F). He found that the work functional structure,

Q,(F), was different for small, Q.(F), middle Q,,(F), and big, Q,(F) men (Figure

1). Taylor organized a process for mutually adapting workers and shovels. Big-

ger workers were supplied with heavier shovels and smaller workers were moti-

vated to train themselves to use larger shovels, and thereby to achieve higher

productivity and pay. The weight of the shovel (with its material) was thus, a

factor of mutual adaptation between the worker and work environment (task,

STUDIES OF EFFICIENCY AND COMPLEXITY 19

F, F,opt=9 FmPt=10 Fpopt=12 F

Fig. 1. Work functional structures Q,(F) for shoveling by three groups of men: small, Q,, middle, Q,,, and

big, Q,, (after Taylor, 1911). Q—relative productivity of shoveling, F—weight of the shovel with material (kg),

Fp: represents optimal weight of shovel.

tool, machine). Productivity of shoveling was used as the criterion of work

efficiency.

This suggested that maximal work efficiency could be reached by mutually

adapting between the human-work functional structure (by professional selec-

tion, training, motivation), and the work environment (by ergonomic design of

machine, workstation, interior, or software). Yerkes and Dodson (1908), and

later, other researchers, also confirmed the bell-shaped function of work effh-

ciency on work factors (Freivalds, 1987; Konz, 1990; Tinker, 1963; Venda,

1975, 1986; Woodworth, 1938).

In another investigation on this issue, Warren (1984) studied the influence of

stair riser height on human climbing efficiency. He found that the “inverse

efficiency” (complexity (C), in our terms) of climbing (that is, energy expended

per step cycle/work done per step cycle) was a U-shaped curve function of riser

height (Figure 2). In Figure 2, we have added a curve displaying efficiency, Q, of

climbing (work done per step cycle/energy expended per step cycle). This is a

function of the factor, F, of mutual adaptation between the climber and the

stairs. Warren found a unique factor as a ratio of riser height/leg length (F =

RH/LL). He discussed this factor as intrinsic, internal, and evolutionary. The

efficiency, Q, is a bell-shaped curve of F, with constant optimum F,,, = 0.25 for

the groups of short climbers (the solid line at Figure 2), and tall climbers (the

broken line). Prior to actual climbing, Warren asked participants to visually

assess which riser height would be best for them. Arrows show visual-riser prefer-

ence prior to climbing. Dots on the curves show the location of F,,, for Qa, and

C,nin» Obtained in the experiments.

20 VENDA

0.1 0.2 Fopto.3 0.4 F

Fig. 2. Functional structure of climbing efficiency-complexity during visual evaluation. Efficiency, Q, mea-

sured as the stair height climbed per one calorie spent by the climber (m/cal). Complexity, C, measured as a

number of calories spent per one meter of stairs. Factor, F, is a riser height (m) (adapted from (Warren, 1984) ).

Subscripts for tall (T) and short (S) individuals. F,,, represents optimal stair height.

Secondly, an experiment similar to Taylor’s, but using a ““modern shovel’’, a

notebook computer, was conducted. In the Ergonomics laboratory in the Uni-

versity of Manitoba, students typed text in a sitting position. Chair height was

kept constant at 39 cm. Desk height was systematically changed. Typing produc-

tivity was found to generate a skewed bell-shaped function versus desk height

(Figure 3). The equation Q,., = Qua + Ca (where Q,,., 18 maximal possible

efficiency found as a peak point of the bell-shaped curve and Q,,, and C,,, are

actual efficiency and complexity values), represents the way in which complex-

ity, and each individual’s efficiency level, interact to comprise the maximal

possible efficiency.

These results can be summarized as Ergodynamics Law 1, (The Law of Mu-

tual Adaptation): ““Work efficiency is a bell-shaped function of the factor of

mutual adaptation between human work structure and its environment.” Fig-

STUDIES OF EFFICIENCY AND COMPLEXITY 21

75 100 125ee = 150 F

F OPt Fact = 120

Fig. 3. Work efficiency Q (typing productivity of characters per 3 minutes) as a function of the factor of

human-environment mutual adaptation F (desk height, cm). F°"' represents optimal desk height. Q*“—an

actual efficiency value Q** = 505 char./3 min. is shown for one particular F value: F = 120 cm. C** = Q™ —

Q** = 550 — 505 = 45 char./3 min.

ures 1-3 demonstrate that Law. There are two coordinates: efficiency, Q; and

the factor of human-environment (machine, person) mutual adaptation, F.

In the practice of work design and optimization, it is necessary to find an

equation or curve, QF), for each actual work-structure, S,. This will determine

Foor, When Q; = Qimax; and F,,;, and F,.,, when Q;O.

The goal of the notebook computer study was to find different work func-

tional structures S(F) that could be used for the same work task. The experi-

ment was extended so the students typed in a sitting (S,), as well as in a standing

(S,) position, with different desk-heights. Two different functional structures for

every student were presented.

Prior studies on the processes of reading and the perception of control board

information have also resulted in functional structures (cognitive strategies) that

can be modeled using bell-shaped, Q.(F), curves (Stishkovskaya, Venda et al.,

1993; Venda, 1980, 1986, 1990).

Results of these studies can be used to suggest Ergodynamics Law 2 (The Law

of Work Structures Plurality): “Every work task can be done with different work

structures modeled as a family of respective bell-shaped functions’.

The third law of ergodynamics was worded and analyzed in a previous paper

(Venda and Venda, 1992). It is known as the Law of Transformations: ““Trans-

formations between different structures of the system and interaction between

different systems’ structures are maximally effective if they go through a state

common and equal for the structures.”» The common state 1s modeled as the

intersect point of the respective Q(F) curves of the two structures (Figure 4).

22 VENDA

Fig. 4. Functional structures of typing in sitting (S,) and standing (S,) positions. Q—typing productivity

(characters per minute); F—desk height as a factor of human-environment mutual adaptation. F°™ represents

optimal desk height when typing person and work environment are mutually adapted and thus Q = Q™.

Common and equal states for the structures S, and S, occur when both struc-

tures have equal efficiency, that is, when F = F,, and Q,(F, 5) = Q)(F,2). The

third law of ergodynamics may be illustrated using results from the previously

discussed typing experiment. When the desk height is changed from minimal (F

= 50 cm) to maximal (F = 140 cm), a variable total productivity is obtained at all

heights, and it is important to find the optimal height for the transformation

from a sitting (S,) posture (work structure), to a standing one (S,). Three trans-

formation-heights are compared in the nght hand side of Figure 5: F = 100 cm,

F = 107 cm (intersection point for sitting and standing), and F = 120 cm. It is

seen in Figure 5, that a maximal integral productivity for the two consecutive

functional structures S, and S, with their transformation S, S, was obtained

when the transformation was made at F = 107 cm (solid line at the right side).

Transformations at F = 100 cm and F = 120 cm (see broken lines at Figure 5) led

to lower integral efficiency values.

In addition to graphs Q(F) and Q(T), graph F(T) is shown at Figure 5. F(T)

shows dynamics of the factor of mutual human-environment adaptation, F. If

one knows functional structures S, = Q,(F), and S, = Q,(F), as well as dynamics

F(T), then the dynamics of work efficiency, Q(T), for different trajectories of the

transformations S, ~ S, and S, ~ S, may be predicted and optimized.

Various work structures may belong to a single individual, or may be used

sequentially. Work structures may also belong to different individuals (or hu-

man-machine linkages) interacting in work processes.

Science (ergonomics, economics and psychology included) theoretically de-

scribes only the following types of development (i.e. progress, learning): 1) step

functions, where efficiency increases in stages; 2) linear increases in efficiency; 3)

STUDIES OF EFFICIENCY AND COMPLEXITY 23

Fig. 5. Visual image of the third law of ergodynamics shown on right side of graph. If F changes gradually

from Fmin = 50 cm to Fmax = 140 cm, and vice versa, forward and backward transformations between two

functional structures S, and S,, (S, S,), which go through a common and equal state with the coordinates Q, >,

= 522 char./3 min. F,,. = 107 cm at the trial #12, lead to the bigger integral productivity (as a sum of Q; at all

heights, F) than transformations at any other state, e.g. at F = 100 (trial #10) and F = 120 cm (trial #13) as

shown with broken lines. T = time (number of trials, consecutively from F = 50 cm to F = 140 cm with

difference between next to each other trials DF = 5 cm).

monotonic, or exponential increases in efficiency (Ebbinghaus, 1885), and 4)

increases in efficiency with intermediate plateaus (Bryan and Harter, 1899).

Using the Law of Transformations, we have previously explained a new type of

development dynamics, as a transformation having a wavy image (Venda and

Venda, 1991). For this purpose, human operator behavior was studied in emer-

gency situations, and information signal-flow characteristics were noted and

used to create simulations for a training experiment. Twelve engineering stu-

dents were asked to perform a compensatory tracking task, where dynamic

signals were presented simultaneously on several (from one to six) measurement

instruments, with the student controlling an equal number of switches (from

one to six). The signal flow, programmed as a Poisson process with its intensity

parameter in the range 0.04 to 1.50 signals per sec., initiated deviations of a

24 VENDA

dynamic model of simple control objects, and thus deflections in the needle on

displays. A computer-generated Poisson process, with intensity \ = 0.04-0.09

signals per sec., were fed into control objects with time constants from 10 to 20

sec. The outputs were fed to the measurement instruments (displays). The stu-

dents’ task was to adjust the control corresponding to the display in order to

counteract the induced deflection. Due to the inertia of the control system, the

student was required to make a sequence of regulating responses in both direc-

tions, much like controlling a vehicle. The signals were given four levels of

priority. The first had absolute priority; the others were relatively less important.

When the student did not have time to regulate all signals generated by the

computer, some were held in computer memory.

Most of the learning curves that resulted were non-monotonic in shape. Dif-

ferent cognitive and sensory-motor strategies and transformations could be

found between them (Venda, 1986). The criterion of efficiency (Q, number of

signals adjusted in a minute) as a function of a number of measurement instru-

ments presented simultaneously (n = 1-6), and training time (T—number of

training sessions), is shown in Figure 6a. The criterion of complexity (C, time

spent to adjust one signal) as a function of n and T, is shown at Figure 6b. The

efficiency and complexity values are as opposites to one another. Increasing

efhiciency and decreasing complexity is a general tendency of the training pro-

cess, but the result is not always monotonic. Sometimes, complexity of perfor-

mance increases when the functional structures of performance are trans-

formed. In these situations, temporary disadaptation of the functional structure

occurs to free components of the previous structure, so that they mutually adapt

each to other in new order in accordance with a new functional structure (Venda

and Venda, 1991). This oscillation between efficiency and complexity produces

wave-like dynamics of development, learning, and progress.

Ergostatics and Ergodynamics

If the independent factor has a constant value, F = Const, and Q = Const then

the result will be functional statics. But if the factor is changed in time and

efficiency, and analyzed as a function of both the factor and of time Q(F,T), then

the outcome will be functional dynamics. The main advantage of the mutual-

adaptation approach is an interrelated analysis and visual representation of the

functional statics (particularly ergostatics) and functional dynamics (ergodyna-

mics). Particularly ergostatics and ergodynamics emphasizes that this approach

is useful in the functional analysis of all human-machine-environment system

components. In reality, several factors may be essential for studying

STUDIES OF EFFICIENCY AND COMPLEXITY 25

sigznals/

3.0

ei), jit ye Soniye Typ WRT i atays he CPR Ara 'y Sy at ATS livys

Fig. 6. (a). Efficiency, Q, as a speed of signal tracking (signals/minute) (b). Complexity, C, as time spent to

track one signal (sec). Compensatory tracking as the function of number of measurement instruments pre-

sented (n = 1-6) and number (days) of training (T).

efficiency statics Q(F,,...,F,) and dynamics Q(F,,...,F,, T), thus, instead of

plain graphs, spatial and hyper-spatial models are frequently needed.

_Ergodynamics and Laboratory Testing of Workstations and Products

The second law of ergodynamics describes how ergonomic projects may fail

when they are tested in a laboratory by subjects using only one functional

structure (S,,,), but are implemented in the workplace where operators use

different strategies (S,,).

Several intervals of work factor F may be used to test the system (Figure 7).

Testing data obtained by subjects in the interval F,,,""" — F,,™", are useless (U)

because operators do not work in that interval. Usually, such tasks are of very

low complexity and are irrelevant in practice. The interval F,,™" — F,,°, is

slightly relevant (SR); whereas, the interval F,,°' — F,,,™™, 1s very relevant (VR)

and helpful because, in this interval, the correlation between subjects’ and opera-

tors’ work efficiency is positive. Thus, changes in system design which increase

subjects’ efficiency in that interval will also increase operators’ work efficiency.

The interval between F,,,°"' — F,,°P', seems to be most helpful because the

maxima of efficiency for the laboratory subjects and real operators are very

close. But an unexpected paradox exists. In reality, all data obtained in the

26 VENDA

Feupmin FeqpoPt Fs,q FopSPt Fopmax

Fig. 7. Work functional structures of real operators (S,,) and of laboratory individuals (S,,,,). Q—criterion

of efficiency (productivity); F—work factor; intervals of F for operators: U—useless, SR—slightly relevant,

H—harmful, VR—very relevant, I—inaccessible for the laboratory participants.

interval F,,°°' — F,,°", are harmful (H). Q,,, and Q,, have a negative correla-

tion. This means that when designers apply laboratory data to improve a system,

and the subjects’ efficiency increases, then the same changes will decrease the

operators’ work efficiency in the real system. The interval F,,,"" — F,,™*, 1s

inaccessible (I) for the laboratory subjects. The tasks in that interval may be too

complicated or the conditions were not modeled in the laboratory. It is likely

that all other intervals outside F,,,"" — F,,™™, are irrelevant for the operational

system and laboratory experiments.

The second law of ergodynamics explains why many practitioners do not trust

ergonomic laboratory testing to real systems and prefer to rely on their experi-

ence and common sense. The paradox is that improvement for the people with

one functional structure (professional skills, individual abilities) may cause

losses for the people with other functional structures. In other words, better

performance may occur in the laboratory, but worse in reality, when functional

structures are not identical.

Practical use of the first and second laws of ergodynamics can be illustrated in

the following example. A company is not satisfied with work productivity and

invites an ergonomist as a consultant. The consultant finds a main factor of

human-environment mutual adaptation (or work efficiency factor) F. The con-

sultant determines the direction of change for the work factor that will increase

STUDIES OF EFFICIENCY AND COMPLEXITY hal

efficiency (or decrease complexity). If the current factor values are at the left side

of the bell-shaped curve of the work functional structure, the factor value would

be increased to improve efficiency. If they are on the right side of the curve then

the factor value would be decreased to improve work efficiency.

If the consultant could model the functional structure, or conduct experi-

ments with the most successful employees, then the data would reveal the maxi-

mal efficiency, and the optimal factor value directly. Optimizing the work factor

would then become a relatively simple task. On the other hand, an experiment,

changing the factor to either a larger or smaller value, would demonstrate the

direction required for maximal work-efficiency to occur. If both factor changes

led to declines in efficiency, it would suggest that the factor value was optimal. If

optimal, the factor should be fixed; it would not be a source of additional work

improvement if the functional structure was retained.

If changes in the factor to both directions were to lead to higher efficiency-cri-

terion values, then the consultant should consider other possible work-struc-

tures and the transformations among them.

Ergodynamics: Practical Applications

Ergodynamics can help to answer many questions important to ergonomic

practice. The following are some examples:

1. How may human work efficiency be increased when a single work structure or

several different ones are used? The first law and Figure 1 provide an answer.

2. How can the quality and productivity of work be improved? Human-environment

mutual adaptation may be an effective way.

3. Can the usability level (user’s work efficiency) of a new technology or workstation

be predicted? When should the technology be upgraded to profit from such upgrad-

ing? The third law of ergodynamics can help to predict a level of efficiency during

transformation period.

4. Can the efficiency data of any new product, software, or information display in the

laboratory, based on naive subjects, be transferred to industry, for example to a

power plant control room? Figure 7 provides an answer. One should notice that

similarity of functional structures, identity of Fopt for both subjects and workers, is

much more important than identity of their efficiency levels.

5. What do the performance-efiiciency dynamics look like when the performance

functional-structure is constant and the environmental factor F reverses in both

directions? A wavy-like dynamics of efficiency may ensue.

6. Will human performance and efficiency stay constant when its structure is constant

and F = F,,, = Const? No, they will change as a result of such events as fatigue,

boredom, overexertion, and repetitive strain injuries (cumulative trauma de-

sorders).

7. What should an ergonomist recommend when the same work site characteristic (for

example, speed or volume of information input) affects one worker’s efficiency

28 VENDA

Fy=F,opt12 Fa Fax F4opt PF fom Tylyacr) Tale ep

Fig. 8. Predicting dynamics of firm’s efficiency (profit) Q as a result of upgrading software. S,;—work

functional structure when initial software S, was used by the firm; S,, S;, S,—software packages successively

available. F = efficiency-complexity factor. Shadowed areas on the right side of the graph depict the losses of

work efficiency during transformations relative to efficiency when first software S, was optimally used. T—

transformational complexity.

positively but another’s negatively? Provisions for individual adaptation of the

work site characteristic for different people should be introduced.

8. Can efficiency dynamics in the workplace be predicted and observed when new

technologies, management, and products are introduced? Again, the third law will

help. Ergodynamics suggests that similar structures can be transformed without

essential transformation dips; and that monotonic exponential dynamics will per-

mit slow, continuous improvements.

9. Can the usability level (user’s work efficiency) of new software be predicted theoreti-

cally when the user’s functional structure is S,, but the new software was designed

for structure S,? Figure 5 provides an answer. The user, having work structure S,,

will use the new software requiring S,, with maximal initial efficiency being Q, >.

10. How often should software be upgraded to obtain a maximal profit from upgrad-

ing? Software companies’ sales persons may promise clients immediate profit from

upgrading software, but typically they do not allow for inevitable losses during the

transformation periods. Analysis of work-functional structures and transforma-

tions can help to find a reasonable basis for frequency of upgrading software. IfS,,

S;, and S, represent different software packages that become available, then upgrad-

ing the firm’s existing software S, to S,, S, and S,, should be carefully analyzed.

Work functional-structures, S,, are shown on the left side of Figure 8 and their

transformations are shown on the right side.

Many different processes and consequences are possible in this situation: a)

When upgrading is made directly to S;, or especially to S,, a significant decrease

in efficiency could occur. b) When upgrading occurs too often, then losses of

efficiency following the transformation of S, into S,, S, into S,, S; into S,, are

bigger than the increases in efficiency. There may not be enough time between

STUDIES OF EFFICIENCY AND COMPLEXITY 29

upgradings to reach maximal efficiency for each new product generation. Thus

upgrading could not only cost money, but also lead to large losses in productiv-

ity. Shadowed areas (Figure 8) depict losses at the periods between T, and T,. c)

When maximal complexity (measured as a factor F value) of the initial tasks is

F nax> then the relative efficiency of using the most advanced (and usually the

most expensive) software S,, is low. The optimal software at that time would be

S;, although S, would be acceptable, considering the predicted losses for trans-

formation of S, into S;.

Conclusion

Ergodynamics is an application of the theory of Transformation Dynamics to

the analysis and improvement of work in the dynamic environment. It is being

based on the three laws of: 1) mutual adaptation, 2) plurality of functional work

structures, and 3) transformations. Along with methodological analysis of en-

tropy, fuzziness and incompatibility in human-machine systems, and with stud-

ies ON Macroergonomics, ergodynamics may help to build a general theory and

methodology of ergonomics. Work functional efficiency and complexity are

interpreted as an opposing criteria. This concept produces an operational

method for measuring and comparing efficiency and complexity using the same

or similar units. Efficiency and complexity depend on both internal functional

structure (work skills) and mutual adaptation-disadaptation with the external

environment.

Functional efficiency and complexity are always addressed to a specific hu-

man function, operator, and goal. Functional complexity can be assessed using

any criterion reflecting the decrements that are to be minimized and converted

into efficiency, because less complexity results in more efficiency, and vice

versa.

Goals and functions may be used to determine criteria for efficiency, com-

plexity, and work strategy. For instance, in order to attain the highest success

levels, managers may wish to avoid traditional methods for organizational deci-

sions and instead, follow technological control decision methods when the

current state and dynamics of the control system are known. Such an approach

can, in turn, stimulate quantitative analyses of transformational dynamics in

decision making.

Every real object has an infinite number of parts and particles. This mitigates

against direct comparisons among the complexities of real objects and environ-

ments. From the point of view of a scientific paradigm, if something cannot be

observed, measured, compared, it does not exist. Therefore, a paradox arises:

30 VENDA

complexity of real environments are not directly measurable, and thus, an envi-

ronment itself does not have complexity. Instead, complexity is a characteristic

of human activities, such as attitudes towards achieving concrete goals by inter-

acting with and adaptating to the environment. The environment can be

friendly and supportive or more hostile and complex, depending upon the hu-

man goal, task, and/or functional structure under investigation. Therefore, it is

necessary to focus on the environmental factors of efficiency and complexity

rather than on environmental complexity per se. To measure efficiency and

complexity with the same units, is convenient for ergonomic and psychological

practice in improving the work environment, tools, and skills. This would sug-

gest that the factors involved in human-environment mutual adaptation may be

considered as factors of eficiency and complexity. Maximal work-efficiency can

be achieved using mutual adaptation between human work-functional struc-

tures (by professional selection, training, motivation), and work environments

(by ergonomic design of machines, workstations, interiors, and software).

The laws proposed to describe these conditions are useful not only for work

dynamics specifically, but they are general in nature. The dynamics of any

complex system of laws may be summarized as follows: The first law of transfor-

mation dynamics: Functioning and development of any system is a process of

mutual adaptation between inner components of the system, and between the

system and its environment. The second law: Every complex system may have

several different functional structures. The third law: Transformations between

functional structures of the complex system move, with minimal losses, through

common and equal states for prior and new structures.

References

Bariff, M. L., & Lusk, E. J. (1977). Cognitive and personality tests for the design of management information

systems. Management Science, 23:820-837.

Bodrov, V. A., & Venda, V. F. (eds.) (1992). System approach in engineering and work psychology. Moscow:

Russian Academy of Science.

Britannica World Language Dictionary (1954). Funk and Wagnals Co.: New York, NY.

Bryan, W. L., & Harter, N. (1899). Studies on the telegraphic language: The acquisition of a hierarchy of

habits. Psychological Review, 57:94-107.

Dickson, D. N. (1983). Using logical techniques for making better decisions. Harvard Business Review, John

Wiley and Sons: New York, NY.

Ebbinghaus, H. (1885). Uber das Gedachtnis. Berlin: Leip-Dunker.

Freivalds, A. (1987). The ergonomics of tools. In: International Review in Ergonomics, 1:12-48.

Grandjean, E. (1988). Fitting the task to man: An ergonomic approach. Taylor and Francis: London.

Harvey, O. J. (1963). System structure, flexibility and creativity. In: O. J. Harvey (ed.), Experience structure

and adaptability. Springer: New York, NY. 13-37.

Harvey, O. J., Hunt, D. E., & Schorder, H. M. (1961). Conceptual systems and personality organization. John

Wiley and Sons: New York, NY.

Hendrick, H. W. (1986a). Organizational design. In: G. Salvendy (ed.), Handbook of human factors. John

Wiley and Sons: New York, NY. 470-494.

Hendrick, H. W. (1986b). Matching individual and job complexity: validation of stratufied systems theory. In:

Proceedings of the 30th Annual Meeting of HFS. Santa Monica: Human Factors Society. 999-1001.

STUDIES OF EFFICIENCY AND COMPLEXITY 31

Hendrick, H. W. (1990). Perceptual accuracy of self and others and leadership status as functions of cognitive

complexity. In: Measures of leadership. Leadership Library of America: West Orange, NJ. 511-520.

Hendrick, H. W. (1992). A macroergonomic approach to work organization for improved safety and produc-

tivity. In: S. Kumar (ed.), Advances in industrial ergonomics and safety. Taylor and Francis: London. 3-10.

Karwowski, W., & Ayoub, M. M. (1984). Fuzzy modelling of stresses in manual lifting tasks. Ergonomics,

27:641-649.

Karwowski, W., & Mital, A. (1986). Applications of fuzzy set theory. Human Factors. Elsevier: Amsterdam.

Karwowski, W., Marek, T., & Noworol, C. (1988). Theoretical basis of the science of ergonomics. In: Proceed-

ings of the 10th Congress of the International Ergonomics Association. Taylor and Francis: London. 756-

758.

Karwowski, W. (1991). Complexity, fuzziness and ergonomic incompatibility issues in the control of dynamic

work environments. Ergonomics, 34:671-686.

Konz, S. (1990). Work design: Industrial ergonomics. 3rd ed., Publishing Horizons: Scottsdale, AZ.

Lefebr, V. (1971). Conflicting structures. Sovetskoe Radio: Moscow.

Leontyev, A. N. (1971). Activity, conscious, personality. Politizdat: Moscow. (In Russian).

Lomoy, B. F., & Venda, V. F. (1977). Human factors; problems of adapting systems for the interaction to the

individual: The theory of hybrid intelligence. Proceedings of the Human Factors Society 21st Annual

Meeting. San Francisco, CA. 1-9.

Rasmussen, J. (1986). Information processing and human-machine interaction: An approach to cognitive

engineering. North-Holland: New York, NY.

Rasmussen, J., & Vicente, K. J. (1989). Coping with human errors through system design: Implications for

ecological interface design. International Journal of Man-Machine Studies, 31:517-534.

Savelyev, A. Y., & Venda, V. F. (1989). Higher education and computerization. Progress Publishers: Moscow.

Sheridan, T. B. (1992). Telerobotics, automation and human supervisory control. MIT Press: Cambridge,

MA.

Stamp, G. (1981). Levels and types of managerial capability. Journal of Management Studies, 18:277-297.

Stishkovskaya, N. N., Venda, V. F., Laliberte, T. G., Fu, W., & Longfield, K. (1993). Strategies of perception

and decision making based on graphic information. In: Advances in industrial ergonomics and safety.

Taylor and Francis: London. 570-576.

Streufert, S., & Swezey, R. W. (1985). Simulation and related research methods in environmental psychology.

In: J. Singer & A. Baum (eds.), Advances in Environmental Psychology, 5:99-117.

Streufert, S., & Swezey, R. (1986). Complexity, managers, and organizations. Academic Press: New York,

NY.

Strickland, L. H. (1991). Russian and Soviet social psychology. Canadian Psychology. 32:284-301.

Taylor, F. W. (1971). The principles of scientific management. Harper and Row: New York, NY.

Venda, V. F. (1975). Engineering psychology and design of information display systems. Mashinostroenie:

Moscow.

Venda, V. F. (1980). Voies nouvelles pour une theorie de l’apprentissage, In: Present et futur de la psychologie

du travail. EPA: Paris. 586-594.

Venda, V. F. (1982). Engineering psychology and design of information display systems. Mashinostroenie:

Moscow.

Venda, V. F. (1986). On transformation learning theory. Behavioral Science, 31:1-11.

Venda, V. F. (1990). Hybrid intelligence systems: Evolution, psychology, ergonomics. Moscow: Mashinos-

troenie.

Venda, V. F., & Venda, Yuri V. (1991). Transformation dynamics in complex systems. Journal of Washington

Academy of Science, 81:163-184.

Venda, V. F. & Venda, Y. V. (1994). Dynamics in ergonomics, psychology and decisions. Norwood, NJ.

Warren, W. H. (1984). Perceiving affordances: visual guidence of stair climbing. Journal of Experimental

Psychology, 10:683-703.

Webster’s Third New International Dictionary (1976). G. & C. Merriam Co. Springfield, MA.

Woodworth, R. (1938). Experimental Psychology. Holt: New York, NY.

Zarakovski, G. M., Korolyov, B. M., Medvedev, V. I., & Shlaen, P. Y. (1977). Introduction to ergonomics.

Sovetskoe Radio Publishers: Moscow.

Zinchenko, V. P. & Munipov, V. M. (1989). Fundamentals of ergonomics. Progress: Moscow.

Journal of the Washington Academy of Sciences,

Volume 83, Number 1, Pages 32-78, March 1993

Human Abilities and Age-Related Changes

in Driving Performance’

Robert E. Llaneras, Robert W. Swezey, John F. Brock

InterScience America, Leesburg, Virginia

and

William C. Rogers

American Trucking Associations Foundation, Trucking Research Institute,

Alexandria, Virginia

ABSTRACT

Literatures concerning fourteen human abilities are reviewed with respect to changes

which occur with aging, and in the context of driving performance. Particular emphasis is

devoted to relationships between the abilities and commercial vehicle driving. The ability

literatures reviewed are: static visual acuity, dynamic visual acuity, contrast sensitivity, use-

ful field of vision, field dependence, depth perception, glare sensitivity, night vision, reaction

time, multilimb coordination and physical proficiency, control precision, decision-making,

selective attention, and attention sharing.

Introduction

In the United States, it has been recently reported that 22 percent of the

licensed drivers are 55 and older, while 10 percent are 65 and older (Rothe,

' This work was conducted under Federal Highway Administration Contract No. DTFH61-93-C-00088

with the American Trucking Associations Foundation, Trucking Research Institute. The authors would like to

acknowledge the important contributions of Mr. Robert Davis, Ms. Teresa Doggett, and Mr. Nathan Root of

the Federal Highway Administration (FHWA) Office of Motor Carriers, Dr. Harold Van Cott of Van Cott and

Associates, Dr. Peter Hancock of the University of Minnesota, and Mr. David Willis of the AAA Foundation

for Traffic Safety, for providing information documents and reviewing drafts; and Ms. Lisa Swezey of ISA for

the research and administrative support she provided. Finally, we wish to acknowledge the important contri-

butions of Dr. Mark Barnes, formerly of InterScience America, who developed the earlier review document

(Barnes, Llaneras, Swezey, Brock, and Rogers, 1994) from which this article was adapted. That comprehensive

report is available from the ATA Foundation, 2200 Mill Road, Alexandria, VA, 22314-4677, USA.

32

_ ABILITIES, AGE, AND DRIVING PERFORMANCE 33

1990). Rothe has also estimated that by the year 2000, approximately 28 percent

of drivers will be 65 and older. The number of older drivers 1s increasing in part

because of medical advances; more people are living longer and enjoying better

health later in life (Santrock, 1985); however, other reasons related to social

changes are also contributing to this trend. A greater proportion of persons in

younger generations have learned to drive. People now have more leisure time

than previously, and thus may have an increased desire to continue driving

(Jette and Branch, 1992). In short, driving remains an important facet of the

older person’s lifestyle. However, a concern also exists that older drivers pose

safety problems to both themselves and to others. There is evidence that: a) older

drivers are overrepresented in certain types of accidents, and b) older people

tend to overrate their driving abilities (Fox, 1989). The highway safety industry

is thus, faced with finding ways to allow older drivers to continue having driving

privileges, while at the same time ensuring road safety.

Studies that have investigated age and accident characteristics have shown

that older drivers tend to be disproportionately involved in accidents involving

head-on collisions, left turns, parking or backing, and right angles (Maleck and

Hummer, 1986). In addition, older drivers have been found to be overly repre-

sented in accidents in urban settings, and accidents involving right-of-way, im-

proper turning, and the disregard of signals (Maleck and Hummer, 1986; All-

gier, 1965, Huston and Janke, 1986). Accidents involving older drivers are more

apt to occur at slower speeds, at intersections, involve multiple vehicles, and

occur under good weather conditions, as compared to younger drivers (Waller,

House, and Stewart, 1977). When taking into account the number of miles

driven per year however, accident rates increase with drivers in their late 50s,

and accelerate beyond that (Cerelli, 1989; Waller, 1991; Williams and Carsten,

1989). This increase occurs despite indications that many older drivers inten-

tionally avoid driving at night, in heavy traffic, and in other demanding condi-

tions (Yee, 1985). Although information such as this provides a snapshot of the

accident patterns associated with older drivers, a shortcoming is that it does not

provide much insight into what is actually causing the accidents.

One approach to identifying potential causal factors, is to identify human

performance abilities that are associated with accidents. This logic presumes

that accidents are related to particular ability deficits. Briefly, the premise un-

derlying the human ability approach to analyzing driving safety, is that success-

ful driving requires a combination of perceptual, psychomotor, and cognitive

abilities. Drivers must continuously monitor the environment in order to avoid

obstacles and to ensure lateral control. Coordination among arms, legs, neck,

and head are needed for steering and shifting. Strength, in combination with

quickness, is needed for circumstances involving manual shifting or sudden

34 LLANERAS ET AL.

braking movements. Drivers must maintain attention, shift and/or share atten-

tion, and constantly make decisions. The extent to which drivers rely on particu-

lar abilities depends upon demands from both the highway environment and

vehicle. Successful driving hinges on the individual’s ability to meet both these

demands, and the consequences that result when they are not met.

This article summarizes research which addresses how changes in perceptual,

psychomotor, and cognitive abilities influence driving performance. It is a por-

tion of a much larger and more comprehensive report (Barnes, Llaneras, Swe-

zey, Brock, and Rogers, 1994). Of particular interest were findings related to

abilities needed to safely operate commercial trucks. As discussed in the review,

however, research pertaining specifically to the commercial truck driving in-

dustry is sparse as compared to that related to conventional vehicle drivers

generally. Therefore, the literature reviewed herein was not restricted to articles

involving commercial vehicle driving.

Literature which addresses the following fourteen driving-related abilities are

reviewed herein. These abilities were selected on the basis of their known rela-

tionships with driving performance.

Perceptual Abilities — static visual acuity

— dynamic visual acuity

— contrast sensitivity

— useful field of vision

— field dependence

— depth perception

— glare sensitivity

— night vision

Psychomotor Abilities — reaction time

— multilimb coordination and physical

proficiency

— control precision

Cognitive Abilities — decision-making

— selective attention

— attention sharing

Although these abilities are commonly discussed in the literature as indepen-

dent, realistically they are interrelated. That is, there are common underlying

mechanisms that cause them to change in similar ways (Greene and Madden,

1987). Much of this is due to the fact that abilities reflect different aspects of a

common physical structure and sensory system. Consequently, changes that

occur within one sensory system (e.g., the visual system) will impact several

‘ABILITIES, AGE, AND DRIVING PERFORMANCE 35

abilities. Furthermore, changes to the central nervous system could potentially

have an impact on all abilities. Some normative changes to the central nervous

system and sensory systems occur with advancing age, which in turn affect

abilities. Neural processing becomes less efficient as a result of neural atrophy

and slowing (Birren, 1965). In addition, reduced blood flow results in a decline

of muscle and sensory function. In essence, many of the ability changes that are

discussed in this review can be traced back to these fundamental changes.

Discussion of Driving-related Abilities

Perceptual Abilities

Some normal age-related changes to the structure of the eye result in broad

changes to all visual abilities (Weale, 1963). For example, the cornea, which acts

to focus images onto the retina, becomes increasingly more rigid and yellow as

new cellular layers are added. As a result, older adults are less able to bring close

objects into focus; the average 60 year old person is unable to focus on objects

closer than 40 inches (Helander, 1987). Because of yellowing, less light pene-

trates the cornea, and that which does, undergoes some degree of scattering. By

age 50, the amount of light reaching the retina is reduced by about 50% (Weale,

1963). Consequently, older adults are less able to see in the presence of dim

illumination or to cope well with glare (Fozard, Wolf, Bell, McFarland, and

Podolsky, 1977). In addition, they require greater contrast between a target and

its background, as compared to young persons (Blackwell and Blackwell, 1980).

Also, the blood supply to the retina diminishes in older adults (Weale, 1963). As

a result, visual receptors are slower to regenerate, peripheral vision becomes

restricted, and the size of the blind spot increases.

Shinar and Schieber (1991) have offered some general remarks regarding the

effect of the normal aging process on visual abilities. They concluded that: 1) all

visual functions deteriorate with increasing age, 2) the amount, rate, and onset

of deterioration vary widely between individuals and functions, 3) significant

deterioration in static acuity does not appear before the age of 60, but more

complex processes begin to decline earlier, and 4) performance differences be-

tween individuals increase with age. Arguably, the most important perceptual

abilities associated with driving are visual abilities. Drivers are required to con-

tinuously scan the roadway, focus on relevant objects, and make distance judg-

ments in the presence of varying illumination levels. It has been estimated that

85 to 95 percent of the sensing clues in the driving task are visual (Malfetti and

Winter, 1986).

36 LLANERAS ET AL.

Static Visual Acuity

Static visual acuity has traditionally been defined as the ability to resolve

details of a stationary target in a well illuminated environment (Sturr, Kline,

and Taub, 1990). Visual acuity varies for different locations on the retina; it is

best near the straight-ahead or foveal field (near 0 degrees), but drops off rapidly

toward the periphery (Schmidt and Connolly, 1966). The area of best acuity

corresponds with the highest concentration of cone receptors, and extends less

than ten degrees away from the foveal center (Coran, Porak, and Ward, 1984).

Although visual acuity declines in all persons under dim illumination, re-

search has indicated that significant age-related declines in static visual acuity

become evident around age 45 (Decina and Staplin, 1993), and rapidly acceler-

ates after the age of 60 (Burg, 1966; Pitts, 1982; Laux and Brelsford, 1990). For

example, although 85% of individuals age 35 to 44 have acuity levels 20/20 or

better, only 32% of individuals age 65 to 74 have similar acuity levels (U. S.

Department of Health, Education, and Welfare, 1977). Fortunately, static vi-

sual acuity can be easily corrected by contact lenses or glasses. Coincidentally, it

is also the one measure that all state licensing agencies depend upon to detect

visual deficiency (National Highway Traffic Safety Administration, 1986).

Relationships to driving performance. Static acuity has been found to have

weak, but consistent relationships to traffic accidents and conviction rates

(Burg, 1971; Burg, 1964; Burg, 1967; Henderson and Burg, 1974; Shinar, 1977).

For example, Burg (1967) reported that three static visual tests considered as a

composite had the second strongest relationship with accidents behind dynamic

visual acuity. The three correlations, based on a sample of 17,000 drivers, —.129

(screen static acuity), —.076 (Ortho-rater), and —.053 (Snellen) were small but

statistically significant. In other studies using fewer subjects, Henderson and

Burg (1974) found a significant relationship between static acuity and accident

rates only for drivers aged 25-49. Hoffstetter (1976; cited in Bailey and Sheedy,

1988) provided evidence that drivers with poor static acuity were more likely to

be accident prone; that is, they were involved in multiple accidents over short

time periods. Drivers in the lowest quartile of static acuity measurements were

twice as likely to have had three accidents, and 50 percent more likely to have

two accidents in the previous twelve months, than those with measurements

above the median.

Rogers and Janke (1992) compared driving records of heavy-vehicle drivers

having substandard static acuity with those that were unimpaired. Substandard

static acuity subjects were categorized into two groups: moderately impaired

(corrected acuity between 20/40 and 20/200 in the worse eye, 20/40 or better in

the good eye), and severely impaired (corrected acuity worse than 20/200 in the

worse eye, 20/40 or better in the good eye). Results showed that as a group,

ABILITIES, AGE, AND DRIVING PERFORMANCE 37

visually impaired drivers had a higher incidence of accidents and convictions

than did unimpaired drivers. However, the incidence of accidents between mod-

erately impaired and unimpaired drivers did not differ significantly, regardless

of age. Older drivers in this sample had /ower conviction and accident rates,

despite having poorer visual acuity on average. Although age ranges and correla-

tion coefficients were not reported, this would appear to contradict other data

that reports poorer vision and higher accident rates for older drivers. In another

study which included 236 truck and bus drivers, Henderson and Burg (1973)

failed to find any significant relationships between static acuity and three acci-

dent measures.

Other studies have linked visual acuity to non-accident driving performance

measures. Relationships have been demonstrated between acuity and improper

lookout behavior (Shinar, McDonald, and Treat, 1978), and the ability to detect

the distance of an object on the road in the presence of oncoming car headlights.

However, Kline, Ghali, Kline, and Brown (1990) found no difference in the

distance at which young, middle-aged, and elderly drivers were able to see high-

way signs, regardless of illumination levels.

Because it is easy to measure and relevant to driving (TRB, 1988), static acuity

is routinely checked by all states before issuing an initial driver's license (Bailey

and Sheedy, 1988). However, its importance applies primarily to instances in

which the vehicle is stopped or moving at a slow rate, such as at intersections or

in parking lots. Unlike real visual scenes, the stimuli used to measure static

visual acuity are typically small and of high contrast (Ginsburg, 1980; Owsley,

Sekuler, and Boldt, 1981; both cited in Evans and Ginsburg, 1985). It is not,

therefore, surprising to find wide variations, among the research describing the

effects of static visual acuity on driving performance and accident involvement

rates.

Dynamic Visual Acuity

Dynamic visual acuity is the ability to resolve the details of a moving target.

Static acuity determines the upper potential of dynamic acuity, since dynamic

acuity also relies on foveal vision (Shinar, 1977). Most studies, however, report

little or no correlation between these measures (Burg, 1968; Henderson and

Burg, 1974). The weak association between static and dynamic visual acuity has

been found to decrease with advancing age (Burg, 1968). Four major points

concerning dynamic visual acuity as summarized by Burg (1964) are that: 1)

Acuity for a moving target deteriorates with increased target velocity, 2) It is

improved by increasing exposure time, illumination, and by practice, 3) It is

better when the target is foveal rather than peripheral, and 4) It varies widely

between individuals with essentially the same static visual acuity.

38 LLANERAS ET AL.

In general, all visual acuity deteriorates rapidly as the image moves away from

the foveal region of the retina (Shinar, 1977). Declines begin when the lateral

rate of the target approaches 20 degrees/second. Above 30 degrees/second,

smooth eye pursuit movements lag behind the target and more complex eye

movements (e.g., saccadic movements) are needed to keep the image in the

foveal region. With respect to aging, dynamic visual acuity has been found to

decline at an earlier age than does static acuity, and this deterioration accelerates

more rapidly after age 50 (Burg, 1966). Shinar (1976; cited in Shinar, 1977) has

placed the onset age of rapid decline to be slightly later, at about age 55. Age-re-

lated changes are greater as the angular velocity exceeds about 30 degrees/se-

cond. At this speed, acuity becomes dependent on more complex perceptual

and neural processing as well as fine ocular movements, which partially explains

why most studies have found weak (Burg and Hulbert, 1961) or no correlation

(Henderson and Burg, 1974) between dynamic and static acuity. Since older

persons demonstrate a systematic reduction in the ability to execute smooth

pursuit eye movements (Sharpe and Sylvester, 1978), these additional demands

may determine why age-related declines in dynamic acuity begin earlier than for

static acuity.

Relationships to driving performance. Dynamic visual acuity has been

found to be positively associated with the average number of miles a person

drives (Retchin, Cox, Fox, and Irwin, 1988); and has been the ability most

related to accident involvement in several correlational studies (Burg, 1968;

Shinar, 1977; Laux and Brelsford, 1990). As with static acuity, however, these

relationships have been consistent, but weak. The reasons for this are also simi-

lar; dynamic acuity is a constant and important ability used when driving, but is

not the only important visual ability involved in driving. Shinar, Mayer, and

Treat (1975) noted that drivers found recently to be at fault in an accident had

poorer dynamic visual acuity than a group of persons who had not been in an

accident for two years. Henderson and Burg (1974) found that among profes-

sional drivers over 50 years of age, the best 10% with respect to dynamic visual

acuity had a lower accident rate than the mean, while the worst 10% had a higher

accident rate.

As indicated by the literature, driving-related activities that appear to rely on

dynamic acuity include reading street signs while in motion, locating road

boundaries when negotiating a turn, and making lateral lane changes. In these

situations, greater speeds are associated with poorer acuity and narrowing of the

high acuity portion of the visual field. As the visual field becomes more complex

and dynamic, other abilities are increasingly important. This is especially true

when drivers must detect objects in instances where: a) the relevant visual field

area is large, b) many relevant objects are present in the visual field, and c)

ABILITIES, AGE, AND DRIVING PERFORMANCE 39

angular velocities of objects within the visual field are high. Under these condi-

tions, other abilities such as head and neck flexibility, field dependence, and

attentional factors take on greater importance. Most studies in this area have

relied upon accident data as a criterion, which, because many variables contrib-

ute to accidents, does not result in strong conclusions concerning the impor-

tance of individual factors. Although dynamic visual acuity has been consis-

tently related to increased accident involvement, the magnitude of this

relationship has not been strong enough to change licensing policy (Bailey and

Sheedy, 1988). Leibowitz, et al. (1993) have speculated that although dynamic

visual acuity is important to performance, stronger relationships with driving

performance must be established before policy decisions can be made with

confidence.

Contrast Sensitivity

Contrast sensitivity is the ability to detect variation in sine-wave patterns (i.e.,

adjacent light and dark regions) as a function of closeness of the neighboring

regions (Decina, Breton, and Staplin, 1991). Contrast sensitivity tests measure

both the response to sharply defined black-on-white targets and those with

grayer, less distinct edges. By measuring the full range of spatial frequencies and

regional contrasts, a comprehensive measurement of visual capability can be

assessed, allowing performance to be summarized as a plot relating target sensi-

tivity across the spatial frequency ranges (Evans and Ginsburg, 1985). Perfor-

mance with respect to higher spatial frequencies (measured in cycles per degree)

has been found to predict real-world target detection, such as detecting simu-

lated aircraft targets (Ginsburg, Evans, Sekular, and Harp, 1982) and reading

road signs (Evans and Ginsburg, 1985). Response to low spatial frequencies has

been linked to visual performance under poor viewing conditions (Ginsburg,

Easterly, and Evans, 1983; Ginsburg, 1980; Ginsburg, 1981).

Evans and Ginsburg (1985) reported a comparison of contrast sensitivity

between a group of 19-30 year-olds and a group of 55-79 year olds. The data

indicated a significant decrement for the older group in the middle and upper

frequency ranges (i.e., more detailed stimuli). Similarly, Schieber (1988) re-

ported that sensitivity in these same ranges begins after age 40. Age-related

declines have also been found for dynamic contrast sensitivity (Scialfa, Garvey,

Gish, Deering, Leibowitz, and Goebel, 1988; Scialfa, Garvey, Tyrrell, and Lei-

bowitz, 1992), sensitivity under low levels of illumination (Sloane, Owsley, and

Alvarez, 1988), and sensitivity in the presence of glare (Leibowitz, Tyrrell,

Andre, Eggers, and Nicholson, 1993).

Relationships to driving performance. Although these data make a case for

the importance of contrast sensitivity with respect to driving visual require-

40 LLANERAS ET AL.

ments, lack of confirming data has limited its potential usefulness as a predictor

of driving performance. During the course of this review, only one study was

identified which reported data concerning relationships between contrast sensi-

tivity and driving performance. Decina and Staplin (1993) failed to find a rela-

tionship between contrast sensitivity measures and accident rates in a 3.67 year

period following the measurement. However, a composite measure combining a

broad contrast sensitivity measure, binocular visual acuity, and horizontal vi-

sual field measurement, was related to crash involvement for drivers aged 66

and older. The authors asserted that including contrast sensitivity in visual

measures used for licensing, could contribute to the identification of the highest

risk older drivers.

Testing practices regarding contrast sensitivity face challenging issues. De-

cina, Breton, and Staplin (1991), have asserted that full ranges of contrast sensi-

tivity testing require more time than is typically available. In the case of licens-

ing, added administration time and cost would be undesirable (Bailey and

Sheedy, 1988). Further, a lack of normative data raises problems with specifying

the criterion level that separates normal from abnormal performance, deter-

mining the size of sample measures needed for accurate testing, and establishing

expected reliability ranges (Legge and Rubin, 1986).

Useful Field of Vision

A measure of the visual field which has received much recognition 1n recent

research is the useful field of vision (UFOV). This has been defined as the area of

the visual field that is useful for acquiring information during a brief glance

(Sanders, 1970). UFOV is determined by measuring the peripheral field

surrounding a straight-ahead fixation point, in which a subject can detect stim-

uli (in the periphery) while simultaneously performing a task in the forward

view. It is believed to accurately represent performance of typical search behav-

iors, since the individual’s attention remains in the forward view.

One factor that contributes to the size of the useful field of vision, is the size

and sensitivity of the peripheral field—an area which tends to become less

sensitive with increasing age (Wolf, 1967; Haas, Flammer, and Schnieder, 1986;

Drance, Berry, and Hughes, 1967; Johnson and Keltner, 1983). These studies

have shown that when parts of the visual field are lost as a result of normal aging,

it is the peripheral area that is affected first. Although peripheral vision is an

important component of UFOV, the latter measure is believed to be more useful

for predicting visual performance in real activities, since it relies somewhat on

attentional ability (Walker, Sedney, Wochinger, Boehm-Davis, and Perez,

1993).

UFOV has been shown to decline with advancing age. Changes have been

ABILITIES, AGE, AND DRIVING PERFORMANCE 41

found for extrafoveal acuity (Cerrella, 1985), determining locations of stimuli in

the presence of distracting stimuli (Sekular and Ball, 1986), and for identifying

letters (Scialfa, Kline, and Lyman, 1987). In addition, Ball, Beard, Roenker,

Miller, and Griggs (1988), found that increasing the difficulty of the task in the

straight-ahead field had a detrimental effect only on a group of older subjects.

Scialfa, et al. (1987) concluded that older adults appear to take smaller percep-

tual samples from the visual scene and scan these samples more slowly than do

young adults. As a result, the size or intensity of stimuli presented in the periph-

ery needs to be increased if older adults are to see them as quickly. In summary,

research indicates that the UFOV is influenced by: a) the size of the visual field,

and b) ability to make use of information within the visual field.

Relationships to driving performance. Several attempts have been made to

link aspects of the visual field to driving performance. Regarding visual field

size, Johnson and Keltner (1983) found that drivers with visual field loss in both

eyes had accident and conviction rates more than twice as high as those with no

significant loss. Others have also found associations, but only in similar circum-

stances where persons had field size losses greater than would be expected with

normal aging (Burg, 1968; Shinar, 1977; Council and Allen, 1974). Walker, et

al. (1993), have pointed out that one reason peripheral vision has been only

weakly associated with driving performance, is that peripheral target detection

has typically been measured as a primary task. It has only been in cases of

extreme abnormalities in the size of the visual field that influences on driving

performance have been detected.

The UFOV measure has shown promise as a better predictor of driving safety

than field size alone. Walker, et al. (1993), have found similar patterns with the

UFOV and driving performance as have been found by other researchers study-

ing non-driving tasks. Using a part-task driving simulator, they found that an

older driver group (range = 60-65 years old) showed a narrowing effect on

UFOV as a result of increased difficulty of the primary task, as indicated by

reaction times.

Ball, Owsley, Sloane, Roenker, and Bruni (1993) examined the relationship

between several measures of visual processing (e.g., eye health, visual function,

mental status, and UFOV) and the number of crashes that each participant was

responsible for in a five year period. They reported a correlation of .46 between

UFOV and accidents. Based on the set of correlations between all variables, they

suggested a model for predicting crash frequency. According to their model,

central vision, eye health, peripheral vision, and mental status, all affect UFOV

directly. The size of the UFOV in turn, had direct effects on crash frequency.

Mental status had a direct effect on both UFOV and crash frequency. Although

age was included as a predictor variable in the study, the change of UFOV asa

42 LLANERAS ET AL.

function of age was not reported. The Ball, et al. (1993) study represents a

different approach to examining accident precursors. The majority of other

studies have treated each ability as being independent and as having a direct

effect on driving performance. In this case, the model reflects partialled effects

and interactions among abilities. It is possible that this method may depict more

accurate relationships between abilities and performance.

Field Dependence

Field dependence refers to the ability of a person to perceive relevant targets

within an embedded context (Shinar, McDowell, Rackoff, and Rockwell, 1978).

It reflects a type of stable cognitive style—a way of perceiving which is domi-

nated by the overall organization of the field. Field independence, on the other

hand, reflects a style in which parts of the field are perceived as discrete from the

organized background (Witkin, Lewis, Hertzman, Machover, Meissner, and

Wapner, 1954). This ability is typically measured by having a subject scan a

complex figure and detect a prescribed target. Persons who are best able to

extract salient information from complex backgrounds are characterized as field

independent, while those less able to do so are field dependent.

Field dependence is distinct from UFOV in that it involves more than target

detection. Rather, it requires the use of scanning strategies and making sense of

multiple information sources in the visual scene (Shinar, et al., 1978). Most

research indicates that people become more field dependent with increasing age

(Shinar, et al., 1978; Ranney and Pulling, 1990; Manivannan, Czaja, Drury, and

Ip, 1993). Loo (1978) reported a small, but insignificant relationship between

age and field dependency.

Relationships to driving performance. Field dependency measures have

demonstrated some predictive value with respect to driving performance mea-

sures such as recognizing hazards, recognizing road signs, controlling skidding

vehicles, and driving in high speed and high density trafic (Goodenough, 1976).

In driving studies that did not consider age differences, field-independent

drivers have been found to have lower accident rates (Harano, 1970), have

quicker braking reaction times (Olson, 1974; Barrett and Thornton, 1968), have

higher deceleration rates during simulated emergency situations (Barrett,

Thornton, and Cabe, 1969), and show better headway maintenance patterns,

and better control a skid-prone car after an initial trial (Olson, 1974), than field

dependent drivers. Arthur, Barrett, and Alexander (1991), included field depen-

dence as part of a meta-analysis examining the roles of cognitive, personality,

and demographic factors on vehicular accident involvement. Their research

resulted in twelve correlation coefficients related to field dependence. The mean

overall correlation was .151, while 65.07 percent of the variance in these studies

ABILITIES, AGE, AND DRIVING PERFORMANCE 43

was left unaccounted for by moderating variables. These authors determined

field dependence to be a weak predictor of accident involvement, and attributed

this to be due in part to the three different assessment tests that were used across

studies (e.g., the Group Embedded Figures Test, the Portable Rod-and-Frame

Test, and the Hidden Figures Test).

We are aware of two studies that have examined age differences in conjunc-

tion with field dependency and driving performance. Mihal and Barrett (1976),

examined the relationships between several perceptual and information-process-

ing abilities and accident rates of utility company drivers. One ability was field

dependence, measured by the Portable Rod and Frame Test and the Embedded

Figures Test. Both measures were significantly correlated with the number of

accidents over a five year period. As an additional statistical manipulation, the

sample was divided into a younger group (25 to 43 years) and an older group (45

to 64 years). For every significant predictor variable (i.e., Rod and Frame, Em-

bedded Figures, complex reaction time, selective attention), the relationship

with accidents was greater for the older group.

Shinar, McDowell, Rackoff, and Rockwell (1978) examined the relationship

between field dependence and on-the-road visual search behavior. They noted

an age-related decline in time needed to identify information. Field dependent

drivers were characterized as needing longer eye fixations to gather relevant

information. In addition, they were less able to adapt to changing perceptual

requirements involved in curve negotiation. Although there appears to be some

support for a relationship between field dependence and driving performance,

the small number of studies in this area limit the confidence of this assertion.

Depth Perception

Depth perception is defined as the ability to judge the distance, and changes in

distance, of an object (Burg, 1964). Depth perception cues can be one of two

types: physiological and environmental (Coran, Porak, and Ward, 1984). Most

physiological measures stem from information about the shape of the eye. When

shifting one’s gaze to a closer or further distance, the lens changes shape to focus

the image. The muscular tension used to change the curvature of the lens pro-

vides a cue as to the distance of an object. When objects are distant however (50

feet and beyond), the shape of the lens does not change dramatically, and is

therefore less helpful as a measure. In addition, because the lens hardens and

ocular muscles weaken in older adults, physiological measures of depth become

less effective with age. The second source of depth perception cues are those

present in the visual field. When judging the distance of stationary objects, cues

such as surface texture gradients, relative heights of objects, linear angles, and

retinal disparity are used (Goldstein, 1989). For judging the distance of objects

4 LLANERAS ET AL.

in motion, cues in the form of changes to the relative size of the images (i.e.,

expansion or contraction) and stimulation to adjacent receptor cells also contrib-

ute to depth judgments (Shinar, 1977).

In general, individuals are more sensitive (i.e., threshold levels are lower) to

targets that are in the presence of photopic illumination levels (Rock, 1953; cited

in Henderson and Burg, 1974) when they are allowed unlimited viewing time

(Rock, 1953; Leibowitz and Lomont, 1954; Leibowitz, 1955; each cited in Hen-

derson and Burg, 1974), and that take up less than 2—3 degrees of space on the

retina (Shinar, 1977). Furthermore, sensitivity to central movement in-depth

(i.e., targets moving directly at an observer from the front) is greater than periph-

eral movement in-depth (i.e., targets moving toward an observer from a side

angle) for all ages. Nevertheless, the ability to make accurate depth judgments

and detect changes in depth, decreases with advancing age (Bell, Wolf, and

Bernholtz, 1972; Henderson and Burg, 1973, 1974; Shinar, 1976). This pattern

holds true for both peripherally and centrally presented targets (Shinar, 1976).

Further, glare sensitivity and visual acuity under dim illumination have both

been shown to limit the ability to detect depth (Yanik, 1986; Rock, 1953; cited

in Henderson and Burg, 1974). Both of these abilities also deteriorate with age

and are important when driving at night. Age differences in depth perception

begin to be noticeable before age 40, increase significantly by age 50, and con-

tinue to increase thereafter (Bell, Wolf, and Bernholtz, 1972). These patterns

were replicated by Henderson and Burg (1974). Other research related to angu-

lar movement detection has been sparse, but existing data suggests that the range

of individual differences is large (Henderson and Burg, 1974). This, combined

with intuitive relevance to driving tasks, indicate that these measures may serve

as a good discriminator for driving performance.

Relationships to driving performance. The ability to judge the distance be-

tween one’s vehicle and other vehicles moving at approximately the same rate is

one of the most critical driving abilities (Shinar, 1977). Central angular detec-

tion ability is important when steering to follow a desired path and avoiding

vehicles in a traffic string (Henderson and Burg, 1974). Detection of angular

motion in the periphery may be the first clue that other vehicles are close or that

hazards are appearing from the side. Boyar, Couts, Joshi, and Klein (1985)

found that following too closely was the second most common mistake that led

to commercial vehicle accidents in 1983. Similar conclusions have been reached

for conventional vehicles (Barrett, Alexander, and Forbes, 1973). Judging dis-

tance through angular movement may be a critical contributor to these types of

incidents.

Judging in-depth motion is made difficult by the fact that when no lateral

displacement occurs, the primary depth cue is the expansion or contraction in

_ ABILITIES, AGE, AND DRIVING PERFORMANCE 45

the image size of other vehicles (Hills, 1980). It appears that older drivers have

difficulty controlling vehicles in this type of situation (Ranney and Pulling,

1990). Using a battery of closed-course driving and laboratory tests, Ranney and

Pulling found that older drivers (74 to 83 years old) made more gap execution

errors (1.e., struck objects/excessively slow speed) and more gap judgment errors

than did a group of younger drivers (30 to 51 years old). In Henderson and

Burg’s (1973) study of truck drivers, those who had central movement in-depth

thresholds greater than 16 minutes of arc/second for large targets, and greater

than 12 minutes of arc/second for small targets had higher accident rates than

the population mean. However, Henderson and Burg (1974) failed to replicate

these findings. In the 1974 study, only ability to detect depth of a large expand-

ing target (1.e., simulated movement of a car on a collision course) was asso-

ciated with accident rates for adults over 25 years of age.

The ability to perceive depth where lateral movement occurs is believed to be

an important determinant in driving tasks involving traffic merging and cross-

ing through intersections (Staplin and Lyles, 1992). To assess age-related depth/

motion perception ability changes, Staplin and Lyles (1992) had subjects esti-

mate how long it would take them to reach specified points in their path given a

constant speed (time-to-collision). In one study, a 100 percent threshold eleva-

tion was found for 70-75 year old drivers, as compared to those 20-29 years old.

This was assessed using a simulation task requiring drivers to identify car pass-

ing events based on tail light positions. Hills (1975) found age-related decre-

ments with motion perception when vehicles were moving closer, but no age

differences when vehicles were moving away. A second study by Hills and John-

son (1980) found that older drivers were likely to underestimate the speeds of

oncoming vehicles. Also, judgments by older drivers about the “‘last safe point”

at which to make a left turn in front of an oncoming vehicle, were found to

correspond to a constant distance, whereas younger drivers took into account

the speed of the oncoming vehicle. Staplin and Lyles concluded that an increase

in angular velocity is an important clue needed for motion detection and that

this ability declines with age. Other data also suggest that this may be true, since

older drivers report that they have problems making left turns against oncoming

traffic, and merging into a traffic stream (Malfetti and Winter, 1987).

Depth perception does indeed appear to be a promising variable for predicting

driving performance, as indicated by both simulated driving tasks and correla-

tional studies. Examining the tasks that have been studied, it appears that depth

perception is an important contributor of information leading to decisions. One

promising finding in the literature is that practice, with or without feedback, was

able to improve ability to detect movement of angular targets (Johnson and

Leibowitz, 1974; cited in Shinar, 1977). The effects of practice were still detected

Ad LLANERAS ET AL.

three months later. A test-retest study by Henderson and Burg (1974), con-

firmed this finding. Thus, depth perception may be an ability that declines with

age, but may be rejuvenated via remedial attention.

Glare Sensitivity

Glare has been defined as brightness within the field of vision that is sufh-

ciently greater than the luminance to which the eyes are adapted (McCormick

and Sanders, 1982). Glare is mostly debilitating to the region within 10 degrees

of the line of sight, since this region is instrumental for seeing detail (Henderson

and Burg, 1974). In the driving environment, two types of glare, veiling glare

and spot glare, are relevant (Shinar, 1977). Veiling glare is a uniform luminance

masking, while spot glare is characterized by a region(s) of concentrated lumi-

nance (e.g., headlights).

Due to yellowing of the lens and clouding of the intraocular fluid, older people

tend to experience a veiling luminance that imposes upon the retinal image

when in bright environments (Drance, Berry, and Hughes, 1967). As a result,

they are less tolerant of glare, and consequently are less apt to detect stimuli in

the visual field in the presence of imposing bright illumination. Leibowitz,

Tyrrell, Andre, Eggers, and Nicholson (1993) examined age differences in both

static and dynamic contrast sensitivity in the presence and absence of glare.

Glare significantly reduced performance for both static and dynamic measures

across all spatial frequencies. In addition, performance was degraded across all

ages. However, glare caused a greater detriment for a group of older adults

(mean age of 68.3 years) than for a group of younger adults (mean age of 25.7

years).

Other researchers have confirmed age-related declines for contrast sensitivity

and static acuity in the presence of glare. However, age differences vary some-

what across studies, making a determination of an onset age unclear. Allen and

Vos (1967), and Burg (1967), examined contrast sensitivity in the presence of

glare and determined that performance remained little effected for various tar-

get sizes and background illuminations until the mid to late 40s, after which

decrements began to accelerate. In addition, Burg reported that the time needed

to recover from glare exposure follows a similar pattern. Shinar (1976) measured

acuity in the presence of both veiling and spot glare. In contrast, this data

suggested that significant changes do not begin to appear until around age 65.

Further evidence that glare effects are not significant across age groups was

provided by Finlay and Wilkinson (1984). They found no differences in glare-

related performance effects for an older group (ages 43 to 56 years) as compared

to a young group (19 to 24 years). The lack of consistency between studies may

be heightened by large individual differences in glare sensitivity. While patholo-

ABILITIES, AGE, AND DRIVING PERFORMANCE 47

gies such as cataracts and glaucoma may have only a small impact on acuity

(depending on their location), they still may increase light scattering in the eye

(Shinar, 1977).

Relationships to driving performance. Although proper lighting can be ef-

fective in increasing visibility for all drivers (Mortimer, 1988), older drivers

require extra illuminance and contrast to be able to see adequately (Sivak,

Olson, and Pastalan, 1981). Furthermore, older drivers have less ability to toler-

ate glare, such as that produced by automobile headlights (Mortimer, 1988).

Henderson and Burg (1974) found that 50 year-old drivers who had the lowest

10 percent of visual acuity in the presence of veiling and spot glare, had accident

rates higher than that of the population mean. Pulling, Wolf, Sturgis, Vaillan-

court, and Dolliver (1980), studied the acceptable level of oncoming headlight

illuminance using a simulated driving task. Prior to age 70, performance slowly

decreased and individual differences were large, but after age 70 the ability to

tolerate glare decreased rapidly. This problem is maximized at night, when

illuminance from headlights reduces the contrast of objects to their background

(Attwood, 1979).

Despite evidence that glare sensitivity and recovery are important age-related

visual changes, few studies have investigated the relationship between glare and

actual driving performance. Headlight glare was attributed as a potential ‘“‘envi-

ronmental” causal factor in approximately 2.3% of night accidents in one inves-

tigative study (Indiana University, 1975; cited in Mortimer, 1988). In another,

glare was mentioned as a factor in 3 of 30 nighttime incidents where drivers were

run off the road (Boyce, Hochmuth, Meneguzzer, and Mortimer, 1987; cited in

Mortimer, 1988). Further, glare was named as a contributing factor in 13 per-

cent of accidents in which an adverse environment was involved (Sabey and

Stoughton, 1975; cited in Mortimer, 1988). The time needed to recover from

glare was included in Burg’s (1971) study, and was found to correlate weakly

with accident rates. Other studies, however, have failed to find a direct relation-

ship between glare sensitivity measures and driving performance (Shinar, 1977;

Wolbarsht, 1977; Burg, 1967).

One possible reason that glare is not implicated as a significant factor more

often is that only drivers with the poorest tolerance to glare are affected to the

point where accidents occur, as Henderson and Burg (1974) found. A second

explanation might be that glare is not a constant part of the driving environ-

ment. In other words, abilities such as static and dynamic acuity are always

relied upon, and thus are more likely to be associated with accident occurrence.

Because glare sensitivity and recovery are only salient in the presence of glare

within about 10 degrees of the line of sight, and because sources of glare are not

always present, the relationship with all accidents is understandably weak.

48 LLANERAS ET AL.

Night Vision

Visual degradation resulting from lower illumination levels has been found to

increase with advancing age (Laux and Brelsford, 1990; Forbes and Vanosdall,

1973; Shinar, 1976; Henderson and Burg, 1974). Further, these studies indicate

that the decline in static acuity under reduced illumination conditions, occurs

earlier in life and progresses more quickly than does declines due to acuity in the

presence of daytime illumination levels. Sturr, Kline, and Taub (1990), for

example, tested static acuity of young (18-25 years) and older (60-87 years)

persons under six luminance levels, ranging from photopic (daytime) to me-

sopic (nighttime). In this study, very little differentiation between age groups

occurred under the highest level of illumination (245.50 cd/m). However,

under lower illumination levels, distinctions appeared. Richards (1977) esti-

mated that the average nighttime illumination level on urban roads was about

1.03 cd/m?. At a level just above this (2.45 cd/m7), large differences exist be-

tween the 60-64 group and older groups. Based upon these data, Sturr, Kline,

and Taub concluded that 65 1s a critical point, after which visual acuity becomes

increasingly poorer under low illumination.

In the case of nighttime vision, acuity depends upon adaptation, the ability of

the eye to adapt its sensitivity to incident light changes (Haig, 1941). During

dark adaptation, the size of the pupil and sensitivity of the retina change in order

to prevent underlighting and overlighting of the retina. Acuity improves rapidly

after one moves to a darkened state. This initial response pattern is due to rapid

adaptation by the cone receptors and pupillary expansion (Wald, 1968). The

slower, but more dominant shift, reflects adaptation by the rod receptors, lo-

cated toward the periphery of the retina. These receptors, and associated neural

processing, produce visual responses useful for detecting general light patterns,

as opposed to fine detail (Barlow, 1982). Approximately 5—8 minutes are needed

before the light-adapted eye switches from resolving visual information with

cones to the more sensitive rods (Shinar, 1977). Rods then continue to slowly

gain sensitivity.

The shape of the adaptation function does not change significantly with age;

however, the extent of sensitivity achieved appears to decrease as one ages

(McFarland, et al, 1960). As a consequence, older people have less sensitivity at

any given time during adaptation. It is therefore more beneficial to older persons

to avoid environments requiring complete adaptation or large contrasts in illu-

mination. The ideal situation would be to always have at least low-level illumina-

tion present and no sources of bright light which require the visual receptors to

readapt to darkened environments.

Relationships to driving performance. Visual abilities that are important

during daytime driving are also important during nighttime driving. Although

_ ABILITIES, AGE, AND DRIVING PERFORMANCE 49

rods are most sensitive in low illumination, adaptation of foveal vision is critical

when performing everyday activities where objects must be detected, such as

when walking into a darkened room (Shipley, 1974). The adaptation of the rods

become important when circumstances require peripheral vision (e.g., detecting

pedestrians). Adaptation is required frequently within the nighttime driving

environment due to extreme changes in illumination and to the presence of

headlights from other vehicles. Henderson and Burg (1974) made the point that

dark adaptation proceeds at a slower rate than the natural decline in outdoor

illumination after sunset. During this transitionary period, peripheral vision is

most affected. This is one reason that twilight is a dangerous time. for driving,

especially when driving into sunlight.

As with photopic static visual acuity, studies examining acuity under mesopic

conditions have revealed weak relationships between acuity and driving perfor-

mance. No relationship was found between low illumination acuity (Henderson

and Burg, 1973) or dark adaptation (Burg, 1964) and accident involvement.

Henderson and Burg (1974) found no overall relationship between low illumina-

tion acuity and accident rates. However, by separating the data by various

accident risk levels, they determined that acuity under low illumination was a

good measure for discriminating drivers with poor overall vision who experi-

enced higher rates of accidents. Shinar (1975), compared low illumination

acuity scores for drivers judged to have committed recognition errors while

involved in an accident, with a group that did not commit recognition errors. As

a group, drivers committing errors attained poorer acuity scores. In a survey of

pedestrian accidents, Hazlett and Allen (1968; cited in Shinar, 1977) found that

drivers did not see the pedestrian that they struck significantly more at night

(87%), than during the day (11.8%).

Weak relationships between night vision abilities and accident rates may

result from failing to distinguish between accidents occurring during the night-

time versus daytime hours, or from the tendency of drivers with night vision

problems to drive less often or more cautiously (Shinar, 1977). Of the studies

mentioned, only Hazlett and Allen’s took into account the time of day that the

accident took place. Without controlling for this, the chance of finding strong

relationships between nighttime acuity and driving performance is severely re-

stricted.

In a study that distinguished between daytime and nighttime accidents,

Shinar (1977), found that acuity under dim illumination conditions was specifi-

cally associated with nighttime accidents and was one of the best predictors of

accident involvement with respect to older drivers. Sivak, Olson, and Pastalan

(1981), found age differences in sign reading distances under scotopic condi-

tions. Sixty year-old drivers needed to be up to 35 percent closer to read signs as

50 LLANERAS ET AL.

compared to 25 year old drivers with equal photopic acuity. These researchers

and others (Sturr, Kline, and Taub, 1990) have claimed that a separate testing of

acuity under low levels of illumination is warranted for licensing, based upon

the weak relationship between photopic and scotopic acuity.

Summary of Perceptual Abilities

Sensory and perceptual changes appear to be an inevitable part of aging.

Different abilities change at different rates and changes vary widely across indi-

viduals; some abilities decline steadily throughout the life-span, while others

appear to change at critical ages. Distinct patterns begin to emerge as early as

forty years of age, and by age fifty an assortment of declining perceptual abilities

can be identified. Table 1 provides a synopsis of studies examining perceptual

abilities in association with driving performance. In early studies, weak relation-

ships were found between ability variables and accidents or convictions. Dy-

namic visual acuity tended to be associated with these measures most consis-

tently, with some associations found for static visual acuity, glare recovery, and

field size.

According to Burg (1971) potential explanations as to why weak relationships

have generally resulted from this type of research are that:

@ Many factors influence driving performance,

@ There may be disparity between an individual’s capability and the degree to which it

is used,

@ Tests may measure characteristics that are not necessarily related to functions used in

driving, and

@ Studies may have other shortcomings related to sampling and performance measures.

Although Burg made these remarks in connection with visual abilities, in

essence they apply to all abilities covered in this review. Driving depends upon a

myriad of factors, making it difficult to attribute performance to any single

ability. This does not imply that each individual ability is unimportant. In fact,

some evidence of performance effect was found for each visual ability. With this

in mind, many studies have attempted to make importance comparisons by

including visual measurements. In some cases, the number of abilities included

in these studies were limited to the abilities that devices were capable of measur-

ing. In others, it was limited by the goals of the researchers. Data from early

studies, along with advances in testing engineering, have lead to improved func-

tional measures, such as contrast sensitivity, depth perception, and useful field

of vision. These measures have been studied less often, but the strength of results

obtained with these measures, suggests that they may be increasingly useful for

predicting driving performance.

ABILITIES, AGE, AND DRIVING PERFORMANCE 51

Table 1.—Summary of Perceptual Abilities and Driving Performance

Ability/Author(s)

Static Visual Acuity

Burg (1964; 1967), Henderson

& Burg (1974) and Shinar

(1977)

Henderson & Burg (1974)

Hoffstetter (1976)

Burg (1971)

Rogers and Janke (1992)

Henderson & Burg (1973)

Shinar, McDonald, and Treat

(1978)

Kline, Ghali, Kline, and Brown

(1990)

Dynamic Visual Acuity

Retchin, Cox, Fox, and Irwin

(1988)

Burg (1968), Shinar (1977) and

Laux & Brelsford (1990)

Shinar, Mayer, and Treat (1975)

Henderson & Burg (1974)

Contrast Sensitivity

Decina & Staplin (1993)

Useful Field of Vision (UFOV)

Johnson & Keltner (1983)

Burg (1968), Shinar (1977) and

Council & Allen (1974)

Walker, Sedney, Wochinger,

Boehm-Davis, and Perez

(1993)

Ball, Owsley, Sloane, Roenker,

and Bruni (1993)

Owsley, Ball, Sloane, Roenker

and Bruni (1991)

Field Dependence

Harano (1970)

Research Findings

Found consistent, but weak, relationships between static acuity and

traffic accident and conviction rates.

Identified significant relationships between static visual acuity and

accident rates occurred for a restricted population: drivers aged

25 to 49.

Found that drivers with poor static acuity were more likely to be

involved in multiple accidents.

Indicated that significant relationships between static acuity and

conviction rates occurred only for female drivers.

Reported that drivers with poor static visual acuity had significantly

higher conviction rates, but no differences in overall accident rates.

Failed to identify any significant relationship between static acuity

and accident rates.

Found a significant relationship between static visual acuity and

improper lookout behavior.

Failed to find differences in the ability to detect highway signs,

regardless of illumination levels, across young, middle aged, and

elderly drivers.

Found a significant relationship between dynamic visual acuity and

the number of miles driven.

Indicated that dynamic visual acuity was the ability most highly

correlated with accident involvement.

Indicated that drivers found to be at fault in accidents are more

likely to have poorer dynamic visual acuity than comparable

groups of drivers.

Found that professional drivers who were over age 50, and among

the top 10 percent with respect to dynamic visual acuity, had

lower than average accident rates, while the bottom 10 percent

had higher than average accident rates.

Found no significant relationships between contrast sensitivity and

accident rates; however, when included as part of a composite

measure, contrast sensitivity was related to the incidence of

accidents in drivers age 66 and older.

Found that accident and conviction rates for drivers with visual field

loss were approximately double those of individuals with no

significant loss.

Found that accident and conviction rates were associated with

UFOV only when the loss was greater than would be expected

with normal aging.

Found that increasing task difficulty significantly reduced the useful

field of vision of older drivers (age 60 to 65).

Indicated that UFOV was significantly related (r = .46) to crash

frequency.

Found a correlation of .36 between UFOV and accidents in a

sample of 53 older drivers.

Found field independent drivers had lower accident rates than field

dependent drivers (age differences not considered).

52

Table 1 .—Continued

Ability/Author(s)

LLANERAS ET AL.

Research Findings

Olson (1974) and Barrett &

Thornton (1968)

Barrett, Thornton, and Cabe

(1969)

Olson (1974)

Mihal & Barrett (1976)

Shinar, McDowell, Rackoff,

and Rockwell (1978)

Depth Perception

Henderson & Burg (1973)

Ranney & Pulling (1990)

Hills & Johnson (1980)

Staplin & Lyles (1992)

Hills (1975)

Glare Sensitivity

Sivak, Olson, and Pastalan

(1981)

Mortimer (1988) and Attwood

(1979)

Henderson & Burg (1974)

Shinar (1977), Wolbarsht

(1977) and Burg (1967)

Burg (1971)

Night Vision

Henderson & Burg (1973; 1974)

Burg (1964)

Hazlett & Allen (1968)

Shinar (1977)

Found braking reaction times were significantly correlated with field

dependence; field independent drivers had quicker braking

reaction times than field dependent drivers.

Found that while driving during simulated emergency situations,

field independent drivers were able to decelerate significantly

faster than field dependent drivers.

Indicated that field independent drivers were able to maintain better

headway patterns and control a skid prone car than field

dependent drivers.

Found that field dependency, in a sample of utility company drivers,

was significantly correlated with accident frequency over a five

year period. This relationship was greater for older drivers (age 45

to 64) than younger drivers (age 25 to 43).

Indicated that field dependent drivers required longer eye fixations

to gather relevant information, and demonstrated less adaptive

eye fixation patterns in response to changing roadway conditions

than field independent drivers.

Found that truck drivers with in-depth thresholds greater than 16

minutes of arc/sec for large targets, and 12 minutes of arc/sec for

small targets had higher than average accident rates.

Indicated that when forced to rely on a single depth cue (the

expansion and contraction of the image size) older drivers (age

74 to 83) made more gap execution and judgment errors than a

group of younger drivers (age 30 to 51).

Found that older drivers were more likely to underestimate the

speed of oncoming vehicles, and base the last safe point at which

to make a turn in front of an oncoming vehicle on a constant

distance rather than the speed of the oncoming vehicle.

Found older driver’s (age 70 to 75) time-to-collison estimates were

elevated by a magnitude of 100 percent as compared to younger

drivers (age 20 to 29); older drivers significantly over-estimated

the time-to-collison with an oncoming vehicle.

Found age-related decrements in the ability to detect motion when

vehicles were moving closer, but no age differences in this ability

when the vehicles were moving away.

Indicated that older drives required extra illuminance and contrast

in order to see adequately.

Found that older drivers are less able to tolerate glare (such as that

produced by automobile headlights) than younger drivers.

Found that 50 year-old drivers who had poor visual acuity (in the

lower 10 percent range) in the presence of glare had higher

accident rates.

Failed to find relationships between glare sensitivity measures and

driving performance.

Found glare sensitivity to correlate weakly with female accident rates.

Found no relationship between low illumination acuity and

accident rates.

Failed to find a relationship between dark adaption and accident

involvement.

Found that the majority of drivers involved in pedestrian accidents

failed to detect pedestrians more at night; 87 percent failed to

detect pedestrians at nighttime versus 12 percent during the

daytime.

Found that acuity under low illumination levels was a leading

predictor of older driver nighttime accident involvement.

‘ABILITIES, AGE, AND DRIVING PERFORMANCE 53

Psychomotor Abilities

Psychomotor abilities change throughout adulthood in much the same way as

perceptual abilities. In general, psychomotor abilities decline with advancing

age, but these changes vary widely, both across abilities and among individuals.

Some changes to physical structures of the body underlie changes to individual

psychomotor abilities. Fundamentally, older adults are smaller in most aspects

of body size than younger adults (Stoudt, 1981). In addition to changes in the

neural systems, changes also occur to musculoskeletal and cardiovascular sys-

tems. As a result, older adults have less muscle strength and are less able to

maintain maximum muscular effort (Santrock, 1985). States (1985), has sug-

gested that some age-related changes to bones, cartilage, ligaments, and muscles

impair the musculoskeletal system’s capability to perform driving activities, and

_ that licensing standards should be developed for strength and joint range-of-

motion.

Several psychomotor abilities appear to be important for driving. Most nota-

bly, multilimb coordination and control precision are necessary when maneu-

vering, braking, and manually shifting (Stelmach and Nahom, 1992). These

“sets” of movements in turn require many individual movements. Essential

musculoskeletal skills needed to perform driving movements include ankle and

plantar flexion, knee extension, hip flexion and extension, hand grip strength,

wrist flexion and extension, and shoulder flexion and extension (Stock, Light,

Douglass, and Burg, 1970). These physical proficiency abilities are sometimes

measured individually, but for the sake of brevity, they will be discussed here,

only to the extent that they contribute to more general movements. Further,

reaction time is important when responding to potentially hazardous situations,

such as wind gusts or objects appearing in the roadway (Wierwille, Casali, and

Repa, 1983; Cox, 1989).

Reaction Time

One of the most widely studied aspects of motor performance is reaction time

(Kausler, 1991). It is most often defined as the time elapsed between the appear-

ance of a signal and the execution of a person’s response movement. The num-

ber of studies investigating reaction time is enormous and it is impractical to

include all of them in this review (see Welford, 1977). As an example of age-re-

lated changes found in reaction time studies, Hodgkins (1962), measured the

reaction time of 400 females between the ages of six and eighty-four. Reaction

time improved from childhood until about the age of twenty, remained constant

until about age twenty-six, then gradually declined. Between the twenties and

seventies there was a 43 percent increase in reaction time.

Researchers have distinguished between phases comprising reactive move-

ments. For this discussion, it is appropriate to distinguish between premotor

54 LLANERAS ET AL.

activities and motor activities. Premotor activities refer to response preparation,

selection, and programming (Stelmach and Nahom, 1992). These activities are

largely cognitive in nature, and are intimately related to decision-making. Mo-

tor activities are the actual physical movements that occur. Stelmach and Na-

hom (1992) reported a sample of studies which illustrate simple and choice

reaction time differences between young and older adults (Clarkson, 1978; Jor-

dan and Rabbitt, 1977; Larish and Stelmach, 1982; Szafran, 1951; Weiss, 1965).

Each study reported slower reaction times for the older group of adults. Across

studies, the older groups averaged approximately 21 percent greater simple reac-

tion times and approximately 38 percent greater choice reaction times com-

pared to young groups. Some studies made separate measurements for premotor

and motor phases and reported a similar trend; older adults took longer to

complete both phases.

Some alternative explanations have been offered to account for these age-re-

lated changes. Stelmach and Goggin (1989), proposed that changes in reaction

time go beyond the simple process of neural slowing. Rather, they contended

that older adults are less able to utilize automated motor programs, and instead

must rely on feedback control processes. As a result, reactive movements may be

more deliberate for older individuals than for younger ones. Consequently,

responses become more variable as well as slower. Salthouse (1985) added that

older adults may be more concerned about avoiding mistakes, and therefore

reaction time is slowed to ensure better accuracy. Alternatively, Rabbitt (1979)

concluded that age-related slowing is related to loss of precise control over the

speed at which responses can be made, or to loss of fine differentiation between

fast’ and ’slow’ responses (1.e., larger intra-individual variance).

In summary, the aging research is fairly consistent in the view that reaction

time declines with increasing age. Some evidence suggests that the pace of the

aging process can be altered somewhat. First, physical activity may slow the

aging effect. Spirduso (1982), found exactly this, when comparing the reaction

times of age-matched physically trained older subjects with physically untrained

older subjects. Reaction times have also been improved through training by

imposing time limitations on responses (Baron and Mattila, 1989). Finally, it

has been noted by several researchers that allowing older persons to anticipate

having to react, may help to speed responses.

Relationships to driving performance. Individual experiments studying age-

related differences in reaction time during driving situations have provided a

wide range of results. Much research has focused on reaction times involving

braking movements. Olson and Sivak (1986) examined brake reaction/move-

ment time in an on-the-road driving task. Response latencies were greater for an

older driver group (ages 50-84), than for a younger group (ages 18-40) when

' ABILITIES, AGE, AND DRIVING PERFORMANCE a5

braking in response to a light attached to the front hood of the car. However, no

response differences were found when subjects responded to an object suddenly

appearing in the roadway.

Lerner (1993), compared on-the-road brake perception-reaction times for

drivers in three age groups: 20-40, 65-69, and over 70 years old. Subjects drove

their own vehicles on actual roads and were required to make an emergency

response to a barrel appearing in the roadway. No age difference in braking time

was found, although only about one-half of the subjects used braking as part of

their response. Thus, assuming that no accidents occurred, it is questionable

whether this task truly represented an emergency event. However, other studies

have also failed to find braking reaction time differences between younger and

older drivers under conditions of unexpected roadway hazards (Korteling, 1990;

Olson and Sivak, 1986). Without discounting the prevailing evidence that infor-

mation processing slows with age, Korteling (1990) suggested that age differ-

ences are moderated by longitudinal driving practice.

Retchin, Cox, Fox, and Irwin (1988) addressed the role of practice and driving

reaction time by comparing reaction times of older persons who were either

frequent drivers, infrequent drivers, or nondrivers. In a simulated task, subjects

released an accelerator control and applied a brake control in response to a

traffic light change. They found that nondrivers took significantly more time to

make these movements than other groups, but that all three groups performed

slower than a group of younger subjects. Practice has been shown to improve

response selection ability in older adults in non-driving tasks. Clark, Lanphear,

and Riddick (1987) had a group of older adults play a video game for two hours

per week for seven weeks. This group later performed better on a response

selection task when they were required to press a button in response to a stimu-

lus using the opposing hand (i.e., incompatible response). Since younger persons

were not included in the study, it is not known whether the practice effects were

related to age.

In the above discussion of age-related reaction time changes, it was noted that

older drivers might benefit by being able to anticipate having to make actions.

Studying reaction time under cued conditions, Staplin, Janoff, and Decina

(1985) found that driving movement responses are only slightly slower for older

drivers when responses are preplanned. In the Olson and Sivak (1986) experi-

ment, no significant age differences in perception-reaction times occurred when

drivers knew that they would soon need to brake.

One potential problem with the majority of these past studies, is that reaction

time has been examined as an isolated event. Considering the limited input and

responses, simple reaction time studies are likely to be a poor predictor of the

complex driving task (Mihal and Barrett, 1976). Although some studies used

56 LLANERAS ET AL.

contrived driving situations, it is unlikely that each of these situations demanded

quick reactions. In a real driving environment, drivers are continually required

to react to many sources of environmental stimuli. Existing driving reaction

time research does not provide a great deal of evidence as to whether older

drivers are more likely to be hindered when faced with multiple choices of

action, or whether they are less likely to recover from mistakes made during

response initiation, as shown in non-driving situations.

Multilimb Coordination and Physical Proficiency

Multilimb coordination is the ability to coordinate movements of two or

more limbs, such as in moving equipment controls (Fleishman and Quaintance,

1984). In the case of driving a vehicle, coordinated movements are needed for

shifting, steering, braking, and accelerating (Cox, 1989). Especially demanding

are circumstances where several movements must be performed simultaneously

or in rapid sequence. Flexibility of the trunk and neck also contribute to coordi-

nated movements, since actions such as merging, changing lanes, and backing

require the driver to scan a large portion of the 360 degree visual field (Hancock,

Caird, and White, 1990).

Multilimb coordination is actually a combination of several more molecular

abilities. It requires some cognitive effort in the form of monitoring and feed-

back processes (Godthelp, Milgram, and Blaaw, 1984), as well as the physical

actions that make up the set of movements. In addition to limb movements,

strength and flexibility of the trunk and neck may also affect coordination while

driving (Stelmach and Nahom, 1992). Some biomechanical measures, which

contribute to coordinated driving movements have been shown to decline with

age. Larsson, Grimby, and Karlsson (1979) found that maximum knee extensor

isometric strength, extension velocity, and dynamic strength all increase

through the age of 29 years, remain stable through the age of 40, and then

decline beyond that age. A slightly different pattern was observed by Murray,

Gardner, Molinger, and Sepic (1980). In their sample, static and dynamic

strength for both knee flexion and extension movements declined in men from

age 20 onward. Laux and Brelsford (1990) reported the following relationships

between age of a sample of active drivers and anthropometric measures believed

to be driving-related:

Measure Correlation with age

Grip strength (left hand) nate |

Grip strength (right hand) —.42

Neck flexibility (left side) me

Neck flexibility (right side) —.44

Torso flexibility (left side) 2

Torso flexibility (right side) Ou

| ABILITIES, AGE, AND DRIVING PERFORMANCE 57

Taken together, these data indicate that the maximum efforts that older per-

sons are able to produce deteriorate. Maximum strength will rarely be required

while driving, except in abnormal situations, such as a tire blowout or stopping

abruptly (Sanders, 1981). The effect of age-related strength loss is unclear, but

probably not critical for most driving tasks. Flexibility measures, on the other

hand, may indicate a more pertinent problem for driving. Coordinated move-

ments that require the driver to exceed comfortable movement boundaries

(reaching, turning) may be adversely affected (Ostrow, Shaffron, and McPher-

son, 1992). In general, data suggest that older adults may have some coordina-

tion deficit when tasks are demanding (e.g., require a great deal of speed). Under

normal circumstances, or when persons are experienced, however, these differ-

ences may not surface.

Relationships to driving performance. Despite the logical connection be-

tween multilimb movements and driving, few studies have investigated it for-

mally. In those that have, small or nonexistent age-performance relationships

have resulted. Most studies have included measurements of strength or flexibil-

ity, rather than coordination. Cox (1989), included measures for range of mo-

tion, muscle strength, head and trunk control, grip strength, reaction time,

proprioception, and light touch and localization in a test battery. None of these

measures were found to be useful for predicting actual in-car driving perfor-

mance for a group of adults over 65 years old.

Laux and Brelsford (1990) compared grip strength, trunk flexibility, and neck

flexibility measures with a set of self-reported driving performance measures.

Poorer grip strength for both hands was significantly related to higher frequen-

cies of bumping into something with the front bumper, running over a curb, and

getting honked at by other drivers (interpreted as a possible driving norm viola-

tion). Height was also related to each of these measures, indicating that shorter

and weaker subjects were reporting more driving problems. Interestingly, grip

strength has been found to correlate with the number of miles an adult drives in

two separate studies (Laux and Brelsford, 1990; Retchin, Cox, Fox, and Irwin,

1988). In the Laux and Brelsford study, the ability to turn one’s head to the left,

important when checking the blind spot caused by a vehicles’ door frame, was

also related to being honked at. Other researchers have also noted that some

older drivers have the habit of failing to look to the rear when changing lanes or

backing up (Malfetti and Winter, 1986). One likely contributor to this habit is

that older persons compensate for chronic stiffness or pain in the upper torso

and neck by not making this critical movement. Indeed, data shows that older

drivers may have more problems making these movements. In one sample of

older drivers, 35% reported problems with arthritis, and 21% found it difficult to

turn their heads in order to look to the rear when driving (Yee, 1985). In addi-

tion, McPherson, Ostrow, Shaffron, and Yeater (1988), found that older drivers

58 LLANERAS ET AL.

(60-75 years) had less shoulder and torso/neck flexibility than younger drivers

(20-35 years). Thus, although only slight relationships between abilities related

to making coordinated movements and accident involvement are apparent,

there is some indication that these abilities could potentially limit movements

necessary for safe driving.

In summary, evidence of age differences in making coordinated driving move-

ments is inconclusive. In the case of normal driving situations, physical limita-

tions probably play a minor role, since steering, braking, and accelerating move-

ments become well learned with experience (Staplin, Janoff, and Decina, 1985;

cited in Stelmach and Nahom, 1992). It is unclear whether coordination plays a

more important role in demanding circumstances, where cognitive demands are

also great (e.g., making sharp turns, turning at intersections).

Control Precision

Control precision refers to the ability to accurately adjust the controls of a

machine or vehicle. This involves the degree to which controls can be moved

quickly and repeatedly to exact positions (Fleishman and Quaintance, 1984). In

the case of driving, the gear shift, clutch, steering wheel, brake pedal, and acceler-

ator must all be moved to more or less exact locations. Some movements may

require quickness or force; thus, control precision 1s closely related to coordina-

tion and reaction time.

Most research addressing coordinated movement and control precision has

measured performance in terms of reaction time. Past research has provided few

clues as to how coordinated control movements, as required for driving a vehi-

cle, change with age. Two experiments studied age differences in the ability to

move a lever. Singleton (1955), studied rapid lever movement patterns and

found age-related differences during separate phases of movement. The task

required subjects to move a lever rapidly from side to side in an 18-inch slot.

Overall, older subjects made movements 29 percent slower than younger sub-

jects. No age differences were evident during the first quarter of movement,

suggesting that age was not a factor in initial acceleration. During the last quarter

of movement, older subjects actually had faster movement. The overall slowing

was a result of older subjects taking longer to make movements at the middle

and for changing direction at the ends. Thus, the older subjects reached slower

maximum movement times and used the end stops to arrest movements. In

contrast, younger subjects made more graded decelerations which resulted in

overall quicker responses.

Another experiment by Singleton (1954), employed a decision-making task

requiring movements similar to those used when manually shifting gears in a

vehicle. Subjects sat with a joystick located between their knees. At the appear-

ABILITIES, AGE, AND DRIVING PERFORMANCE 59

ance of a signal, they were to pull the joystick straight back. At the end of this

movement another signal indicated whether to move the Joystick to the left or

right. Both time to move the joystick and time spent at points where the joystick

changed direction increased with age. Thus, quickness of control movements

was shown to decline somewhat with increasing age.

Relationships to driving performance. Little evidence exists to indicate the

effects of age-related differences with respect to making precise movements of

vehicle controls. The experimental tasks used by Singleton (1954, 1955) may

offer some generalizability. In these experiments, older adults were found to

move levers and joysticks in a less graduated manner; they had slower move-

ment times and relied on end stops of the path to arrest their movement. It

would be useful to know if older drivers also demonstrate cumbersome move-

_ ment when shifting gears, pressing brake pedals, etc. Staplin, Janoff, and Decina

(1985; cited in Stelmach and Nahom, 1992) compared the driving skills of

young and old adults, and determined that older drivers perform movements

only slightly slower than young drivers when they were preplanned. They con-

cluded that when driving activities are well learned, the effects of the aging

process are minimal.

Sanders (1981), included a task related to driving that required a combination

of control precision and strength. Truck and bus drivers applied peak and sus-

tained isometric forces to a truck steering wheel using three different hand

positions. The purpose was to determine how many current drivers could apply

the force needed to control the vehicle in the case of a tire blowout. It was

estimated that seven percent of the drivers would not have been able to apply

enough force to maintain control. Because measurements were taken under

expected conditions and subjects were not required to brake, it is likely that the

actual number of control failures could be higher. Age was not considered in this

study, however, in light of evidence that arm and grip strength decline with age

(Johnson, 1982; Forbes and Reina, 1970; MacLennan, Hall, Timothy, and

Robinson, 1980; Laux and Brelsford, 1990), it is possible that fewer older drivers

would be able to maintain control.

In brief, any age-related changes concerning control precision probably play a

minor role in driving performance. Drivers with practice may show only slight

decrements in the ability to make correct movements. Demands for making

quick and forceful movements have been lessened by some design innovations.

Finally, it appears that the consequences associated with making poor control

movements may be less severe relative to poor performance with other abilities

(e.g., reaction time). Failing to move the transmission smoothly or exhibiting

poor braking or steering patterns may reflect poor driving skill, but are less likely

to result in significant numbers of incidents.

60 LLANERAS ET AL.

Summary of Psychomotor Abilities

In summary, a variety of psychomotor abilities are obviously needed to drive

a vehicle, and some data suggest that they decline with age. The influence of

psychomotor changes on age-related accident patterns, however, is minor rela-

tive to the influence of decision-making, perception, and cognitive abilities

(Welford, 1977). On a positive note, one reason that physical abilities are less

critical, is that physical demands have been reduced by the introduction of

innovations such as automatic transmissions, power steering and power brakes.

As a result, fewer controlling movements (e.g., shifting) are required and less

force is necessary to make those that do exist. Indeed, older drivers may take

advantage of such innovations. As Lerner (1993) noted, almost all of the older

drivers’ vehicles studied were equipped with automatic transmissions as com-

pared to about two-thirds of the younger drivers’ vehicles.

Table 2 provides a synopsis of studies that examined psychomotor abilities in

relation to driving performance. The majority of studies centered around reac-

tion time, due to its critical nature. Although little driving research addressed

control precision and multilimb coordination, these abilities were included in

this review because they are relied on so frequently, and because they may

represent a critical distinction between abilities needed for driving conventional

vehicles and those needed for driving commercial vehicles. Psychomotor de-

mands could be greater when driving trucks or buses due to differences in

physical features of these vehicles. The majority of commercial vehicles are

equipped with standard transmissions, which require frequent repetitive shift-

ing. Head, neck, and trunk movements may be more substantial due to the need

to monitor larger viewing angles. Reactive steering and braking movements

may also require more limb movement and strength. Finally, anticipation and

reaction time is liable to be even more critical when driving larger vehicles, since

it takes longer to stop a larger vehicle’s forward motion. On the other hand,

psychomotor demands associated with driving commercial vehicles may be

offset by driver experience. For example, Godthelp (1986), concluded that expe-

rienced drivers are able to use their knowledge of vehicle handling to lessen their

reliance on visual feedback. Also, Staplin, et al., (1985) found that braking,

steering, and accelerating activities may become well learned and show resis-

tance to age-related reaction time slowing. Because of this uncertainty, psycho-

motor abilities have been included in the scope of this review.

Cognitive Abilities

Cognitive abilities are an integral part of performance of everyday complex

activities. Cognition has been defined as the ability to know and understand the

ABILITIES, AGE, AND DRIVING PERFORMANCE 61

Table 2.—Summary of Psychomotor Abilities and Driving Performance

Ability/Author(s)

Reaction Time

Olson & Sivak (1986)

Lerner (1993)

Korteling (1990)

Retchin, Cox, Fox, and Irwin

(1988)

Staplin, Janoff, and Decina

(1985)

Multilimb Coordination &

Physical Proficiency

Cox (1989)

Laux & Brelsford (1990)

Retchin, Cox, Fox, and Irwin

(1988) and Laux &

Brelsford (1990)

McPherson, Ostrow,

Shaffron, and Yeater

(1988)

Malfetti & Winter (1986) and

Yee (1985)

Control Precision

Singleton (1955)

Singleton (1954)

Research Findings

Found that response latencies for older drivers (age 50 to 84) were

significantly greater than for younger drivers (age 18 to 40) when

braking in response to a light attached to the front hood of a car.

However, no age-related differences in reaction time were found when

drivers knew they would soon need to brake.

Found no age differences in braking times in a comparison of on-the-road

brake perception-reaction times for drivers in three age groups—20 to

40, 65 to 69, and over 70.

Failed to find differences in braking reaction times between younger and

older drivers under conditions of unexpected roadway hazards.

Indicated that older non-drivers took significantly more time to release an

accelerator control and apply a brake control in response to traffic light

changes than older frequent or infrequent drivers. However, older

drivers, regardless of driving experience, performed slower than a group

of younger adults.

Found that movement responses are only slightly slower for older drivers

when responses are preplanned.

Found no relationship between actual in-car driving performance for a

group of adults over 65 and measures of range of motion, muscle

strength, head and trunk control, grip strength, reaction time,

proprioception, and touch and sound localization.

Found poor grip strength was related to higher incidence of degraded

driving performance; including bumping into something with the front

bumper, running over a curb, and getting honked at (all self-reports).

Found grip strength was correlated with the number of miles driven by

adult drivers.

Found older drivers (age 60 to 75) had less shoulder and torso/neck

flexibility than younger drivers (age 20 to 35).

Found older drivers experienced difficulty and/or failed to turn their heads

and look in the rear when changing lanes or backing-up.

Found age was not a factor in initial acceleration during rapid lever

movements; however, older subjects had slow maximum movement

times and used end points to arrest their movements.

Found that the speed associated with control movements in a simulated

gear shifting task declined with age; older subjects took longer to move

a joystick and spent more time at points where the joystick changed

directions than younger subjects.

demands of safe driving, and to be able to react to situations in an appropriate

manner (Irwin, 1989). Driving performance, in particular, is maintained

through a constant stream of small decisions and less frequent larger decisions

(Decina, Breton, and Staplin, 1991). In order to make decisions, drivers must be

able to focus and divide attention on information sources. In this respect, atten-

tion has an intimate relationship with perceptual abilities; attention is needed to

62 LLANERAS ET AL.

focus on relevant information, but in turn will be limited by the quality of the

information that the sensory systems provide (Foley and Moray, 1987).

Like perceptual and psychomotor abilities, cognitive abilities undergo age-

related changes. Cognitive difficulty, often referred to as workload, will be high-

est when there are uncertainties about the environment and when tasks are

physically difficult (Hancock, Caird, and White, 1990). Investigations of acci-

dents and fatalities of older drivers show that cognitive factors, namely, errors of

omission (failing to take some action), and inattention are significant contribut-

ing factors (Malfetti and Winter, 1986). This appears to be largely a problem of

failing to commit undivided, concentrated attention to the driving task. Fell

(1976), has estimated that 60 percent of crashes involving older drivers occur as

a result of cognitive factors; while Shinar (1978), has estimated that 25%-50% of

accidents are a result of driver inattention. Inattention is a likely culprit in many

rear end collisions, since drivers have better sensitivity to movement toward

them as opposed to away from them, indicating a reduced role of perceptual

abilities (Staplin and Lyles, 1992).

Decision-Making

Closely related to choice reaction time is decision-making—the ability to

judge when a situation requires action and to take appropriate action. As dis-

cussed, older adults show slower performance when faced with reaction time

tasks involving multiple response alternatives (Singleton, 1954, 1955). Differ-

ences in response selection ability may be an important contributor to age-

related declines on highly reactive decision-making tasks (Kausler, 1991). Mak-

ing decisions while driving involves some degree of selective attention, since

drivers must process some perceptual information while ignoring other infor-

mation (Staplin, 1990; cited in Staplin and Fisk, 1991). Age-related decrements

related to perceptual abilities will thus effect the speed and accuracy of the

information intake (e.g., highway geometry, traffic signs, other vehicles), and

therefore will also slow decision-making (Staplin and Fisk, 1991).

Under uncertain conditions, cognitive effort referred to as control processing,

is needed to process incoming perceptual information and compare it to knowl-

edge and decision rules stored in memory in order to determine an action to take

(Schneider and Shiffrin, 1977). Control processing is slow, effortful, and is ad-

versely affected by demanding circumstances (Hancock, Caird, and White,

1990).

Relationships to driving performance. If, in fact, older adults suffer signifi-

cant declines in perception and cognitive processing (Salthouse, 1990a), it

should follow that driving performance will be poorer in situations involving

greater decision-making. Research indicates that situations that require decision

et

' ABILITIES, AGE, AND DRIVING PERFORMANCE 63

making, namely left-turns, parking and backing, and right-angle turns, are prob-

lematic for older drivers (Maleck and Hummer, 1986). In addition, older drivers

have been found to have difficulty with high-density intersections (Drury and

Clement, 1978), despite the fact that they avoid driving at night, in heavy traffic,

and in unfamiliar situations where perceptual performance is more difficult

(Shinar and Schieber, 1991; Laux and Brelsford, 1990). Thus, it seems reason-

able to expect that declining decision-making ability contributes to the problems

of the older driver in demanding situations such as signalized intersections

(Staplin and Fisk, 1991).

As mentioned, one driving situation that older drivers have particular diffi-

culty with is making left-hand turns through intersections. When making a

left-hand turn, drivers must pre-plan the path of the turn, perceive relevant

traffic signs and/or lights, determine the right-of-way status of other vehicles,

visually monitor the status of moving vehicles, make a decision to actuate the

turn, and monitor the visual scene while making the turn. This represents one of

the most demanding driving tasks and one of the few situations where signifi-

cant response planning is required.

Staplin and Fisk (1991) conducted two experiments where younger (ages

18-49), and older (ages 65-80), drivers made decisions about the right-of-way

status as quickly as possible in several simulated left-hand turn driving situa-

tions. In the first experiment, older drivers made slower decisions than younger

drivers in each type of situation, but no differences occurred with respect to

decision accuracy. The second experiment included a dynamic presentation

with cued response, intended to make the task more complex. Under these

conditions, older drivers also made relatively more mistakes. The introduction

of advanced information before the intersection aided performance of both

groups equally. Thus, adding additional information ahead of the intersection

did not help to alleviate age differences.

In the case of decision-making when merging or turning at intersections, there

is a fair amount of uncertainty in the form of visually hidden information,

changing right-of-way status, and the intent of other drivers. Sivak, Flannagan,

and Olson (1987) examined reaction time in relation to uncertainty (likelihood

of having to make a simulated braking response). Although reaction times were

slowest during conditions of relatively high uncertainty, no significant differ-

ences resulted with respect to age comparisons.

Other studies have investigated decision-making away from intersections and

found age-related differences. van Wolffelaar, Rothengatter, and Brouwer

(1990; cited in Korteling, 1990) found that older drivers needed 50% more time

than young drivers to observe and decide whether they could safely merge into

traffic on a road. Flint, Smith, and Rossi (1988) found that older persons do well

64 LLANERAS ET AL.

when staying 1n one lane, but that they make more inappropriate responses and

have greater response delays when having to make quick decisive reactions.

Practice may offer some potential for improving cognitive performance while

driving. Lucas, Heimstra and Spiegel (1973) demonstrated improvement in

making judgments about the last safe moment for passing a lead vehicle in the

face of an oncoming vehicle, by providing feedback during a training session.

Age differences were not reported, but it seems that training could be an effec-

tive measure for reducing age-related decision-making decrements in driving

situations. Salthouse (1990b) summarized two general findings in the effects of

practice on cognitive functioning. First, it appears that both young and old

adults improve their performance with additional experience. Thus, if it is desir-

able to improve the absolute level of functioning, then experience may be bene-

ficial. Second, there is no evidence that age differences can be eliminated after all

individuals have received comparable amounts of practice or training. Thus,

given equal amounts of experience, age differences in cognitive performance

will persist.

In summary, the driving literature reveals several situations that require deci-

sion-making ability (e.g., turning at intersections, merging into traffic). Al-

though decision-making is not a continuous activity, it is fairly critical, since

mistakes can easily result in collisions. Few studies have examined age-related

decision-making in the context of driving, but those that have, provide consis-

tent evidence that older drivers have difficulty making roadway decisions

quickly and accurately when tasks are demanding.

Selective Attention

Attention can be considered as a type of psychological energy required to

perform effortful mental work. Evidence suggests that people possess a limited

attentional capacity (Hasher and Zacks, 1979; Kahneman, 1973), and that it

becomes more limited with increasing age (Hasher and Zacks, 1979; Craik and

Simon, 1980). Selective attention, sometimes referred to as focused attention, 1s

the ability to concentrate on a task one is performing, despite boredom or

distracting stimuli (Fleishman and Quaintance, 1984). It includes focusing on

relevant stimuli while ignoring that which is irrelevant, and sometimes involves

switching attentional resources to appropriate stimuli at appropriate moments.

Although research has provided evidence for age-related declines in atten-

tional capacity (e.g., memory), findings related specifically to selective attention

are less clear. Results are mixed, and vary depending upon the type of task

studied. Rabbitt (1965), demonstrated an age-related decline in the ability to

ignore irrelevant stimuli in a card sorting task. Manivannan, Czaja, Drury, and

Ip (1993), Plude and Hoyer (1986), and Farkas and Hoyer (1980) have found

ABILITIES, AGE, AND DRIVING PERFORMANCE 65

similar changes using letter and symbol detection tasks. On the other hand,

Madden and Nebes (1980) found no age difference in the ability to perform a

visual search task under cued (not requiring selective attention) or noncued

(requiring selective attention) conditions.

Other studies have also addressed auditory selective attention and found age

changes for some conditions. Panek and Rush (1981), employed a dichotic-

listening task, in which stimuli were presented to each ear simultaneously. Sub-

jects repeated the target messages (e.g., numbers and letters) as soon as they

heard them, while ignoring irrelevant stimuli. The number of errors increased

for older adults. In conclusion, it appears that older adults are adversely affected

by irrelevant stimuli when visual search is required or when irrelevant stimuli

must be processed along with relevant stimuli. Data stemming from non-search

tasks and auditory tasks are inconclusive, but suggest potential age-related de-

cline.

Relationships to driving performance. Selective attention is likely important

for making constant speed and position adjustments and remaining prepared to

make reactive movements. It is also important whenever a driver must direct

attention to changes in the roadway. In this role, attentional limitations may

exacerbate the effect of declining sensory skills (Sekuler, Kline, and Dismukes,

1982). Older drivers have been noted to make mistakes of omission, such as

failing to yield and running red lights (Planek and Fowler, 1971), and not mak-

ing avoidance responses before an accident (Sussman, Bishop, Madnick, and

Walter, 1982). Planek and Fowler (1971) have claimed that many of the acci-

dents involving older drivers are directly attributable to reduced ability to ignore

irrelevant stimuli in older adults.

Selective attention was found to be the cognitive/human characteristic vari-

able most related to vehicle accident involvement in Arthur, Barrett, and Alex-

ander’s (1991) meta-analysis. It was deemed to be a “‘moderately favorable”

predictor; the overall mean correlation was .257, while 42.52 percent of the

variance remained unaccounted for. Again, these authors attributed some pre-

dictive weakening to the fact that two different measures were used across stud-

ies (e.g., the Auditory Selective Attention Test and the Dichotic Listening Test).

Arthur and Doverspike (1992) found a significant correlation between auditory

selective attention in an investigation of personal characteristics and self-

reported accident rates. Kahneman, Ben-Ishai, and Lotan (1973) investigated

the relationship between an auditory selective attention test and accident rates

of professional bus drivers. The test required the listener to monitor concurrent

messages delivered simultaneously to both ears. After the presentations, a tone

was presented to one of the ears indicating which of the two messages to report.

The ability to reorient attention to relevant stimuli was believed to reflect an

66 LLANERAS ET AL.

important attentional component to driving. Data showed a significant rela-

tionship between the number of errors on the test and accidents during that year.

Mihal and Barrett (1976) included the same auditory selection measure in a

study of utility company driver accident rates. They also found a moderate

correlation between selective attention and accidents. The relationship between

attentional deficit and accident rate was greater for older drivers as compared to

younger drivers.

Avolio, Kroeck, and Panek (1985) gathered three measures of visual and

auditory selective attention (omission errors, intrusion errors, and switching

errors), as well as a field dependency measure, and correlated these with 10-year

accident experience for commercial drivers from a utility firm. Correlations

between all of the attention measures except visual (intrusion errors) were signif-

icant. Correlations among the attention measures were also high, suggesting that

they may be measuring some common aspects of the same fundamental ability.

In summary, the number of driving studies which include measures of selec-

tive attention is small. However, the strength of the correlations in studies relat-

ing selective attention to accidents is encouraging. The data are especially im-

pressive considering that it has been difficult for any ability to be strongly

associated with accidents. In the case of selective attention, it is important to

examine associations with other abilities. As mentioned, attention in general has

intimate ties with other abilities, most notably the useful field of vision and

depth perception.

Attention Sharing

Attention sharing (also commonly referred to as divided attention) is the

ability to shift one’s attention back and forth between two or more sources of

information (Fleishman and Quaintance, 1984). Attention sharing decrements

stem from an inability to process all important information that is present in a

situation. One of the clearest results in the aging literature prior to around 1980,

was the finding that older subjects are more penalized when they must divide

their attention (Craik, 1977). However, these early divided attention studies

failed to properly control for irrelevant age differences. Specifically, age-related

differences in emphasis people place on each task, and ability to perform individ-

ual tasks of varying complexity were not controlled. In dual-task situations,

performance on both tasks typically exhibit performance decrements (Salt-

house, 1982). It is uncertain, however, as to how each individual is allocating

attentional resources between tasks.

Some recent studies have controlled the effects of emphasis and complexity

by using Performance Operating Curves (POCs) in which performance on one

task is plotted against performance on another task across several conditions

ABILITIES, AGE, AND DRIVING PERFORMANCE 67

involving relative emphasis of the two tasks (Somberg and Salthouse, 1982;

Salthouse, Rogan, and Prill, 1984; Ponds, Brouwer, and van Wolffelaar, 1988;

McDowd and Craik, 1988). In each study, emphasis was controlled by manipu-

lating the level of reward for performance on each task, while complexity was

controlled in all but the first study by including tasks of varying difficulty.

Somberg and Salthouse (1982) reported no age differences in dual-task perfor-

mance involving two simple keypressing tasks. On the other hand, McDowd

and Craik (1988) using auditory monitoring and visual identification tasks,

Salthouse et al., (1984) using two memory tasks, and Ponds, et al., (1988) using

simulated road display and visual counting tasks, found significant age-related

declines. The common conclusion from these controlled” studies is that age-

related differences in ability to shift attention between tasks occurs when great

emphasis is placed on one of the tasks and when at least one of the tasks is

complex.

Relationships to driving performance. With respect to driving, it is assumed

that maintaining visual attention in the forward field is typically the primary

task. Attention will sometimes be shared with other tasks, such as routine steer-

ing and shifting, listening to conversations or radios, visually scanning for haz-

ards, or avoiding vehicles entering into the driving path. The amount of effort

required to perform tasks in tandem depend upon: a) demands placed on percep-

tual, physical, or cognitive resources, b) compatibility of tasks (Kantowitz,

1974), and c) experience performing the tasks (Shiffrin and Schneider, 1977).

Ponds, Brouwer, and van Wolffelaar (1988) studied divided attention in a

simulated driving task, using POCs to control for both task difficulty and empha-

sis. Subjects guided the path of an automobile along a simulated roadway in the

presence of obstacles. A secondary task required them to determine the number

of dots that appeared in a visual display. The results indicated that the detrimen-

tal effect caused by dividing attention between tasks was greater for a group of

older adults (mean age = 68.6 years) as compared to young and middle-aged

adults. No difference occurred between the young and middle-aged adults.

Brouwer, Waterink, van Wolffelaar, and Rothengatter (1991), studied di-

vided attention using dual visual tasks in a driving simulator. One task was a

compensatory lane-tracking task, while the other was a self-paced visual analysis

task. POCs were not used to control for emphasis; however, an attempt to adjust

for individual differences was made by representing each person’s dual-task

score as a function of their performance on an adaptive single-task score. Older

adults showed a decreased ability to divide attention, as shown by lane tracking

performance and accuracy with respect to the visual analysis task. The impair-

ment with respect to visual analysis task accuracy was less when responses were

made verbally, as opposed to manually; indicating that requiring persons to

68 LLANERAS ET AL.

make two responses using the same body parts may contribute to performance

decrements for older persons.

McKnight and McKnight (1991) studied the effects of performing several

common secondary tasks during a simulated driving task. Subjects responded to

traffic situations while placing a cellular phone call, carrying on a simple cellular

phone conversation (simply talking), carrying on a complex cellular phone con-

versation (solving problems), tuning a radio, or driving with no distractions. All

of the distractions led to significant increases in both situational response fail-

ures and response time. Complex conversations were most detrimental, while

simple conversations were least detrimental. Response failures and response

times approximately doubled for a group of older adults (ages 51-80), as com-

pared to younger adults when making calls or having simple conversations.

In this experiment, drivers were required to perform the secondary task. Thus,

it appears that drivers sacrificed driving performance in response. Staplin and

Fisk (1991), however, determined that drivers ignored signs if their messages

were not of sufficient conspicuity and legibility to be perceived and understood

at a glance. Speculatively, if Staplin and Fisk's suggestions are correct, and

age-related perceptual deterioration is taken into consideration, it may be pre-

dicted that older drivers would increasingly ignore environmental stimuli that

they were unable to easily process. This illustrates the importance of examining

dual-task performance across levels of emphasis.

In summary, the literature provides some fairly conclusive evidence that older

adults have difficulty sharing attention between two tasks, especially when at

least one task is complex or greatly emphasized. The few studies using simulated

driving tasks provide some indication that this pattern may also apply to driving

situations. It should be expected that when driving demands are light, the sec-

ondary task can easily be attended. It is in circumstances where driving demands

become great, that drivers, and especially older drivers, are likely to fail to

process all important information. Urban settings often require the use of di-

vided attention, and older drivers have been shown to be overrepresented in

these accidents (Maleck and Hummer, 1986).

Summary of Cognitive Abilities

Table 3 provides a synopsis of studies examining the relationship between

decision-making, selective attention, and divided attention with driving perfor-

mance.

In summary, both the aging and older driver literature for decision-making,

selective attention, and attention sharing, indicate that capacity limitations exist

and are more limited in older adults. Age-related performance differences are

even more substantial under demanding circumstances.

ABILITIES, AGE, AND DRIVING PERFORMANCE 69

Table 3.—Summary of Cognitive Abilities and Driving Performance

Ability/Author(s)

Decision-Making

Singleton (1954)

Staplin & Fisk (1991)

van Wolffelaar, Rothengatter, and

Brouwer (1990)

Flint, Smith, and Rossi (1988)

| Sivak, Flannagan, and Olson

(1987)

Lucas, Heimstra and Spiegel (1973)

Selective Attention

Planek & Fowler (1971) and

Sussman, Bishop, Madnick, and

Walter (1982)

Kahneman, Ben-Ishai, and Lotan

(1973)

Mihal & Barrett (1976)

Avolio, Kroeck, and Panek (1985)

Attention Sharing

McKnight & McKnight (1991)

Brouwer, Waterink, van

Wolffelaar, and Rothengatter

(1991)

Ponds, Brouwer, and van

Wolffelaar (1988)

Research Findings

Found that older adults made slower decisions at points where

directional movements were required in a lever moving task.

Examined right-of-way and left-hand turn situations and found

that older drivers (age 65 to 80) took significantly longer than

younger drivers (age 18 to 49) to make decisions in both

situations, but detected no differences in terms of accuracy.

When the tasks were increased in difficulty, however, older

drivers made relatively more mistakes than younger drivers.

Indicated that older drivers needed 50 percent more time than

younger drivers to observe and decide whether they could safely

merge into traffic.

Found that older drivers make more mistakes when the task

requires them to make quick decisions as compared to situations

where these responses are not necessary (e.g., staying in a single

lane).

Found no significant age-related differences in the ability to make

decisions in the presence of uncertainty as measured by braking

reaction times.

Indicated that feedback improves decision-making, as

demonstrated in improved judgments about the last safe

moment for passing a lead vehicle in the presence of oncoming

traffic. (Age effects were not investigated)

Found older drivers tend to make mistakes of omission, such as

failing to yield and running red lights, and failing to make

avoidance responses before an accident.

Found a significant relationship between auditory selective

attention and accident rates of professional bus drivers.

Indicated that the relationship between auditory selective attention

and accident rates of utility company drivers was significantly

greater for older drivers than younger drivers.

Found significant relationships between measures of visual and

auditory selective attention and accident rates of commercial

drivers over a ten year period.

In a study investigating the effects of common secondary tasks

(talking, tuning the ratio, etc.) on simulated driving

performance, these researchers found significant increases in

response failures and response times as a consequence of these

distractions. Deficiencies were approximately doubled for older

drivers (age 51 to 80) than for younger drivers when making

calls or having simple conversations.

Found older adults showed a decreased ability to divide attention,

as measured by lane tracking performance and visual accuracy.

Indicated that when guiding a vehicle along a simulated roadway

in the presence of obstacles, older drivers (mean age 69 years)

had significantly more difficulty dividing their attention

between the roadway and a secondary visual display than

younger drivers.

70 LLANERAS ET AL.

The strength of effects related to driving performance support the call for

additional basic research on how driving performance is affected by attention,

information processing, and problem solving across the population of drivers

(TRB, 1988). However, two problems were apparent in the attentional studies,

and these must be overcome in order to be able to generalize the data properly.

First, the relative contributions of separate measures must be assessed. For

instance, several selective attention measures were claimed to have been related

to accidents in Avolio, et al.’s (1985) study. However, without knowing what the

associations between these measures were, it is difficult to know their unique

predictability. In other words, they may be (and likely are) measuring some of

the same process. Second, emphasis and complexity must be controlled in dual-

task situations. Without knowing how individuals are placing emphasis on each

task, it is difficult to generalize findings to a real situation, such as driving.

Conclusions

One theme that has been reiterated throughout this review is that driver

abilities are a critical component of the vehicle-driver-roadway system. The

requirements to perceive, think, and take action must meet the demands of each

driving circumstance in order to ensure safety. The reviewed literature has pro-

vided a considerable amount of evidence that older drivers are less able than

younger drivers to meet those demands. Although it is convenient to classify

driver data by an individual’s age, not all individuals will perform consistently

with an age group’s norm; in fact, much of the literature reviewed here reported

that variability in performance increased with advancing age. Thus, although

perceptual, psychomotor, and cognitive functions tend to deteriorate with in-

creasing age, the amount, rate, and onset of these degradations vary widely

among individuals and functions.

As driving-related abilities decline, risks associated with driving become

greater. Older drivers in the general population tend to compensate for these

degradations by changing their driving habits; they drive fewer miles, and avoid

driving at night, at high speeds, and in bad weather. Commercial truck drivers,

however, are less apt to have the flexibility to choose the circumstances under

which they drive. Thus, it is conceivable that ability deficits may have a larger

impact on commercial driving performance than conventional driving. On a

positive note, older commercial drivers may possess greater experience levels

than does the general driving population. Consequently, they may be more

efficient in terms of acquiring environmental information and acting on it, than

less experienced drivers; in effect, countering ability degradations.

.

|

|

‘

ABILITIES, AGE, AND DRIVING PERFORMANCE 71

One potential solution for bridging the gap between driver abilities and driv-

ing demands is to introduce interventions which counteract the gap. In order to

produce effective interventions, it appears logical to address areas where they

may lead to the greatest performance improvement. To do this, interventions

must therefore address abilities that (1) decline with age, and (2) are important

to driving. Ideally, it would be helpful to make definitive conclusions about the

relative contribution of each ability reviewed toward performance problems of

older drivers. Unfortunately, it is difficult to make such comparisons since most

studies in the area have either restricted their scope to a single driving-related

ability, expressed driving performance in different terms, and/or failed to report

the relationships between variables. It is evident that driving requires coordi-

nated input from many abilities. There was evidence that all of the abilities

discussed in this review contributed to performance in some way. Thus, it is

likely that no single ability will account for a major proportion of the total

influence associated with driving performance.

References

Allen, M. J., & Vos, J. J. (1967). Ocular scattered light and visual performance as a function of age. American

Journal of Optometry and Archives of the American Academy of Optometry, 44:717-727.

Allgier, E. (1965). Accident involvement of senior drivers. Traffic Digest and Review, 3(3):17-19.

Arthur, W., Jr., Barrett, G. V., & Alexander, R. A. (1991). Prediction of vehicular accident involvement: A

meta-analysis. Human Performance, 4(2):89-105.

Arthur, W., Jr., & Doverspike, D. (1992). Locus of control and auditory selective attention as predictors of

driving accident involvement: A comparative longitudinal investigation. Journal of Safety Research,

23(2):73-80.

Attwood, D. A. (1979). The effects of headlight glare on vehicle detection at dusk and dawn. Human Factors,

31:35-45.

Avolio, B. J., Kroeck, K. G., & Panek, P. E. (1985). Individual differences in information-processing ability as

a predictor of motor vehicle accidents. Human Factors, 27(5):577-587.

Bailey, I. L., & Sheedy, J. E. (1988). Vision Screening for Driver Licensure. In Transportation in an Aging

Society. Transportation Research Board Special Report 218(2). National Research Council: Washington,

DC.

Ball, K. K., Beard, B. L., Roenker, D. L., Miller, R. L., & Griggs, D. S. (1988). Age and visual search:

Expanding the useful field of view. Journal of the Optical Society of America, 5:2210-2219.

Ball, K., Owsley, C., Sloane, M. E., Roenker, D. L., & Bruni, J. R. (1993). Visual attention problems as a

predictor of vehicle crashes in older drivers. Investigative Ophthalmology & Visual Science, 34(11):3110-

3123;

Barlow, H. B. (1982). Physiology of the Retina. In H. B. Barlow and J. D. Mollon (Eds.), The Senses.

Cambridge University Press: Cambridge.

Barnes, M. E., Llaneras, R. E., Brock, J. F., Swezey, R. W., & Rogers, W. C. (1994). Human Abilities and

Driving Performance: A Review of the Literature With Respect to Older Commercial Vehicle Drivers. Fed-

eral Highway Administration (Contract No. DIFH61-93-C-00088). Available through the American

Trucking Associations Foundation, Trucking Research Institute, Alexandria, VA.

Baron, A., & Mattila, W. R. (1989). Response slowing of older adults: Effects of time-limit contingencies on

single- and dual-task performances. Psychology and Aging, 4:66-72.

Barrett, G. V., Alexander, R. A., & Forbes, J. B. (1973). Analysis of performance measurement and training

requirements for driver decision making in emergency situations. Rochester, New York: University of

Rochester, Management Research Center, Report No. DOT-HS-800 867.

Barrett, G. V., & Thornton, C. L. (1968). Relation between perceptual style and driver reaction in an emer-

gency situation. Journal of Applied Psychology, 52:169-176.

Barrett, G. V., Thornton, C. L., & Cabe, P. A. (1969). Relation between Embedded Figures Test performance

and simulator behavior. Journal of Applied Psychology, 53:253-254.

ha LLANERAS ET AL.

Bell, B., Wolf, E., & Bernholtz, C. D. (1972). Depth perception as a function of age. Human Development,

3:77-88.

Birren, J. E. (1965). Age changes in speed and behavior: Its central nature and physiological correlates. In A. T.

Welford and J. E. Birren (Eds.), Behavior, Aging, and the Nervous System. Springfield, IL: Charles C.

Thomas.

Blackwell, O. M., & Blackwell, H. R. (1980). Individual responses to lighting parameters for a population of

235 observers of varying ages. Journal of the Illuminating Engineering Society, 9:3-13.

Boyar, V. W., Couts, D. A., Joshi, A. J., & Klein, T. M. (1985). Identification of preventable commercial

vehicle accidents and their causes. U. S. Department of Transportation (Contract No. DTFH61-84-C-

00034). Washington, DC.

Boyce, D., Hochmuth, J. J., Meneguzzer, C., & Mortimer, R. G. (1987). A Survey of Run-Off-the-Road

Events and Accidents. In Cost-Effective 3R Roadside Safety Policy for Two-Lane Rural Highways. Final

Report. University of Illinois at Urbana-Champaign.

Brouwer, W. H., Waterink, W., van Wolffelaar, P. C., & Rothengatter, T. (1991). Divided attention in

experienced young and older drivers: Lane tracking and visual analysis in a dynamic driving simulator.

Human Factors, 33(5):573-582.

Burg, A. (1971). Vision and driving: A report on research. Human Factors, 13(1):79-87.

Burg, A. (1968). Vision and driving: A summary of research findings. Highway Research Record 216.

Burg, A. (1967). The relationship between vision test scores and driving record: Additional findings. (Report

67-24). Los Angeles: University of California, Department of Engineering.

Burg, A. (1966). Visual acuity as measured by dynamic and static tests. Journal of Applied Psychology,

50(6):460-466.

Burg, A. (1964). An investigation of some relationships between dynamic visual acuity, static visual acuity, and

driving record. (Report 64-18). Los Angeles: University of California, Department of Engineering.

Burg, A., & Hulbert, S. F. (1961). Dynamic visual acuity as related to age, sex, and static acuity. Journal of

Applied Psychology, 45:111-116.-

Cerella, J. (1985). Age-related declines in extrafoveal letter perception. Journal of Gerontology, 40:727-736.

Cerelli, E. (1989). Older drivers, the age factor in traffic safety. Department of Transportation Study Report

No. DOT-HS-807 402.

Clark, J. E., Lanphear, A. K., & Riddick, C. C. (1987). The effects of videogame playing on the response

selection processing of elderly adults. Journal of Gerontology, 42:82-85.

Clarkson, P. M. (1978). The effect of age and activity level on simple and choice fractionated response time.

European Journal of Applied Physiology, 40:17-25.

Coran, S., Porak, C., & Ward, L. M. (1984). Sensation and Perception (2nd Ed.). Academic Press: Orlando.

Council, F. M., & Allen, J. A. (1974). A study of the visual fields of North Carolina drivers and their relationship

to accidents. Chapel Hill, NC: University of North Carolina Highway Safety Research Center.

Cox, J. L. (1989). Elderly drivers’ perceptions of their driving abilities compared to their functional motor

skills and their actual driving performance. In Taira (Ed.), Assessing the driving ability of the elderly: A

preliminary investigation. The Haworth Press: New York.

Craik, F. I. M. (1977). Age differences in human memory. In J. E. Birren and K. W. Schaie (Eds.), Handbook

of the psychology of aging. New York: Van Nostrand Reinhold.

Craik, F.I.M., & Simon, E. (1980). Age differences in memory: The roles of attention and depth of processing.

In L. W. Poon, J. L. Fozard, L. S. Cermack, D. Arenberg, & L. W. Thompson (Eds.), New Directions in

memory and Aging. Hillsdale, NJ: Erlbaum.

Decina, L. E., Breton, M. E., & Staplin, L. (1991). Visual disorders and commercial drivers. Report No.

DTFH61-90-C-00093. Office of Motor Carriers, Federal Highway Administration: Washington, DC.

Decina, L. E., & Staplin, L. (1993). Retrospective evaluation of alternative vision screening criteria for older

and younger drivers. Accident Analysis & Prevention, 25(3):267-275.

Drance, S. M., Berry, V., & Hughes, A. (1967). Studies of the effects of age on the central and peripheral

isopters of the visual field in normal subjects. American Journal of Ophthalmology, 63:1667-1672.

Drury, C. G., & Clement, M. R. (1978). The effect of area, density, and number of background characters in

visual search. Human Factors, 20:597-602.

Evans, D. W., & Ginsburg, A. P. (1985). Contrast sensitivity predicts age-related differences in highway-sign

discriminability. Human Factors, 27(6):637-642.

Farkas, M.S., & Hoyer, W. J. (1980). Processing consequences of perceptual grouping in selective attention.

Journal of Gerontology, 35:207-216.

Fell, J. C. (1976). A motor vehicle accident causal system: The human element. Human Factors, 18:85-94.

Finlay, D., & Wilkinson, J. (1984). The effects of glare on the contrast sensitivity function. Human Factors,

26(3):283-287.

Fleishman, E. A., & Quaintance, M. K. (1984). Taxonomies of human performance: The description of human

tasks. New York: Harcourt Brace Jovanovich, Academic Press, Inc.

‘ABILITIES, AGE, AND DRIVING PERFORMANCE 73

Flint, S. J., Smith, K. W., & Rossi, D. G. (1988). An evaluation of mature driver performance. Technical

Report. Santa Fe: Traffic Safety Bureau, Transportation Programs Division. New Mexico Highway and

Transportation Department.

Foley, P., & Moray, N. (1987). Sensation, perception, and systems design. In G. Salvendy (Ed.), Handbook of

Human Factors. John Wiley and Sons: New York.

Forbes, C. B., & Reina, J. C. (1970). Adult lean body mass declines with age; some longitudinal observations.

Metabolism, 19:653.

Forbes, T. W., & Vanosdall, F. E. (1973). Low-contrast vision under mesopic and photopic illumination.

Highway Research and Record #440, 29-37. Washington, DC.

Fox, M. D. (1989). Elderly drivers’ perceptions of their driving abilities compared to their functional visual

perception skills and their actual driving performance. In Taira (Ed.), Assessing the driving ability of the

elderly: A preliminary investigation. The Haworth Press: New York.

Fozard, J. L., Wolf, E., Bell, B., McFarland, R. A., & Podolsky, S. (1977). Visual perception and communica-

tion. In J. E. Birren and K. W. Schaie (Eds.), Handbook of the psychology of aging. New York: Van Nostrand

Reinhold.

Ginsburg, A. (1981). Spatial filtering and vision: Implications for normal and abnormal vision. In L. Proenza,

J. Enoch, & A. Jamplosky (Eds.), Applications of psychophysics to clinical problems (pp. 70-106). New

York: Cambridge University Press.

Ginsburg, A. (1980). Proposed new vision standards for the 1980s and beyond: Contrast sensitivity. In Proceed-

ings No. 310 of the Advisory Group for Aerospace Research and Development (pp. 123-128). Toronto,

Canada: Advisory Group for Aerospace Research and Development.

Ginsburg, A. P., Easterly, J.. & Evans, D. W. (1983). Contrast sensitivity predicts target detection field

performance of pilots. Proceedings of the Human Factors Society—27th Annual Meeting, pp. 269-273.

Ginsburg, A. P., Evans, R., Sekular, R., & Harp, S. (1982). Contrast sensitivity predicts pilots’ performance in

aircraft simulators. American Journal of Optometry and Physiological Optics, 59:105-109.

Godthelp, H. (1986). Vehicle control during curve driving. Human Factors, 28(2):211-221.

Godthelp, J., Milgram, P., & Blaaw, G. J. (1984). The development of a time-related measure to describe

driving strategy. Human Factors, 26(3):257-268.

Goldstein, E. B. (1989). Sensation and Perception (3rd Ed.). Wadsworth Publishing Company: Belmont, CA.

Goodenough, D. R. (1976). A review of individual differences in field dependence as a factor in auto safety.

Human Factors, 18:53-62.

Greene, H. A., & Madden, D. J. (1987). Adult age differences in visual acuity, stereopsis, and contrast

sensitivity. American Journal of Optometry and Physiological Optics, 64:749-753.

Haas, A., Flammer, J., & Schnieder, U. (1986). Influence of age on visual fields of normal subjects. American

Journal of Ophthalmology, 101:199-203.

Haig, C. (1941). The course of rod dark adaption as influenced by intensity and duration of pre-adapting to

light. Journal of General Physiology, 24:735-751.

Hancock, P. A., Caird, J. K., & White, H. G. (1990). The use of driving simulation for the assessment, training,

and testing of older drivers. Report HFRL NIA 90-01. National Institute of Aging.

Harano, R. M. (1970). Relationship of field dependence and motor-vehicle-accident involvement. Perceptual

and Motor Skills, 31:272-274.

Hasher, L., & Zacks, R. T. (1979). Automatic and effortful processes in memory. Journal of Experimental

Psychology: General, 108:356-388.

Hazlett, R. D., & Allen, M. J. (1968). The ability to see a pedestrian at night: Effect of clothing, reflectorization

and driver intoxication. HRB Record 216:13-22.

Helander, M. G. (1987). Design of visual displays. In G. Salvendy (Ed.), Handbook of Human Factors. John

Wiley & Sons: New York.

Henderson, R. L., & Burg, A. (1974). Vision and audition in driving. Final Report, DOT-HS-009-1-009, Santa

Monica, Calif.: Systems Development Corporation.

Henderson, R. L., & Burg, A. (1973). The Role of Vision and Audition in Truck and Bus Driving. Report

TM(L)-5260/000/00. Santa Monica: Systems Development Corporation.

Hills, B. L. (1980). Speed and minimum gap acceptance judgments at two rural junctions. Report SR137.

Department of the Environment, Transport and Road Research Laboratory, Crowthorne, Berkshire, En-

gland.

Hills, B. L. (1975). Some studies of movement perception, age and accidents. In Proceedings of the First

International Congress on Vision and Road Safety, pp. 65-80. Paris: Routiere Internationale.

Hills, B. L., & Johnson, L. (1980). Speed and minimum gap acceptance judgements at two rural junctions.

Report SR515. Department of the Environment, Transport and Road Research Laboratory, Crowthorne,

Berkshire, England.

Hodgkins, J. (1962). Influence of age on the speed of reaction and movement in females. Journal of Gerontol-

ogy, 17:385-389.

74 LLANERAS ET AL.

Hoffstetter, H. W. (1976). Visual acuity and highway accidents. Journal of the American Optometric Associa-

tion, 47:887-893.

Huston, R. E., & Janke, M. K. (1986). Senior driver facts. Report CAL-DMV-RSS-86-82, 2nd ed. California

Department of Motor Vehicles: Sacramento, CA.

Indiana University (1975). Tri-level Study of the Causes of Traffic Accidents. Report DOT-HS-801-335.

Institute for Research in Public Safety, Indiana University, Bloomington.

Irwin, L. (1989). Elderly drivers’ perceptions of their driving abilities compared to their compared to their

cognitive skills and driving performance. In Taira (Ed.), Assessing the driving ability of the elderly: A

preliminary investigation. The Haworth Press: New York.

Jette, A. M., & Branch, L. G. (1992). A ten-year follow-up of driving patterns among the community-dwelling

elderly. Human Factors, 34(1):25-31.

Johnson, T. (1982). Age-related differences in isometric and dynamic strength and endurance. Physical Ther-

apy, 62(7):985-989.

Johnson, C. A., & Keltner, J. L. (1983). Incidence of visual field loss in 20,000 eyes and its relationship to

driving performance. Archives of Ophthalmology, 101:371-375.

Johnson, C. A., & Leibowitz, H. W. (1974). Practice, refractive error, and feedback as factors influencing

peripheral motion thresholds. Perception and Psychophysics, 15:276-280.

Jordan, T. C., & Rabbitt, P. M. A. (1977). Response times to stimuli of increasing complexity as a function of

ageing. British Journal of Psychology, 68:189-201.

Kahneman, D. (1973). Attention and Effort. Englewood Cliffs, NJ: Prentice-Hall.

Kahneman, D., Ben-Ishai, R., & Lotan, M. (1973). Relation of a test of attention to road accidents. Journal of

Applied Psychology, 58(1):113-115.

Kantowitz, B. H. (1974). Double stimulation: In B. H. Kantowitz (Ed.), Human Information Processing:

Tutorials in Performance and Cognition. Hillsdale, NJ: Lawrence Erlbaum Associates.

Kausler, D. H. (1991). Experimental Psychology, Cognition, and Human Aging. Springer-Verlag: New York.

Kline, T. J. B., Ghali, L. M., Kline, D. W., & Brown, S. (1990). Visibility distance of highway signs among

young, middle-aged and elderly observers: Icons are better than text. Human Factors, 32(5):609-619.

Korteling, J. E. (1990). Perception-response speed and driving capabilities of brain-damaged and older drivers.

Human Factors, 32(1):95-108.

Larish, D. D., & Stelmach, G. E. (1982). Preprogramming, programming, and reprogramming of aimed hand

movements as a function of age. Journal of Motor Behavior, 14:322-340.

Larsson, L., Grimby, G., & Karlsson, J. (1979). Muscle strength and speed of movement in relation to age and

muscle morphology. Journal of Applied Physiology, 46:45 1-454.

Laux, L. F., & Brelsford, J., Jr. (1990). Age-related changes in sensory, cognitive, psychomotor and physical

functioning and driving performance in drivers aged 40 to 92. AAA Foundation for Traffic Safety: Washing-

ton, DC.

Legge, G. E., & Rubin, G. S. (1986). Contrast sensitivity function as a screening test: A critique. American

Journal of Optometry and Physiological Optics, 63:265-270.

Leibowitz, H. W. (1955). The relation between the rate threshold for the perception of movement and lumi-

nance for various durations of exposure. Journal of Experimental Psychology, 49(3).

Leibowitz, H. W., & Lomont, J. F. (1954). The effect of luminance and exposure time upon perception of

motion. WADC Technical Report 54-780.

Leibowitz, H. W., Tyrrell, R. A., Andre, J. T., Eggers, B. G., & Nicholson, M. E. (1993). Dynamic Visual

Contrast Sensitivity. AAA Foundation for Traffic Safety: Washington, DC.

Lerner, N. D. (1993). Brake reaction times of older and younger drivers. Proceedings of the Human Factors

and Ergonomics Society 37th Annual Meeting, pp. 206-210.

Loo, R. (1978). Individual differences and the perception of traffic signs. Human Factors, 20(1):65-74.

Lucas, R., Heimstra, N., & Spiegel, D. (1973). Part-task simulation training of driver’s passing judgments.

Human Factors, 15(3):269-274.

MacLennan, W. J., Hall, M. R. P., Timothy, J. I., & Robinson, M. (1980). Is weakness in old age due to muscle

wasting? Age and Aging, 9:188-192.

Madden, D. J., & Nebes, R. D. (1980). Aging and the development of automaticity in visual search. Develop-

mental Psychology, 16:377-384.

Maleck, T., & Hummer, J. (1986). Driver age and highway safety. Transportation Research Record 1059 (pp.

6-12). Washington, DC: National Research Council, Transportation Research Board.

Malfetti, J., & Winter, D. (1987). Safe and Unsafe Performance of Older Drivers: A Descriptive Study. AAA

Foundation for Traffic Safety: Falls Church, VA.

Malfetti, J. L., & Winter, D. J. (1986). Drivers 55 plus: Test your own performance. AAA Foundation for

Traffic Safety: Washington, DC.

Manivannan, P., Czaja, S., Drury, C., & Ip, C. M. (1993). The impact of age on visual search performance. In

Proceedings of the Human Factors and Ergonomics Society 37th Annual Meeting. Human Factors and

Ergonomics Society: Santa Monica, CA.

ABILITIES, AGE, AND DRIVING PERFORMANCE 75

McCormick, E. J., & Sanders, M. S. (1982). Human Factors in Engineering and Design (5th Ed.). McGraw-

Hill: New York.

McDowd, J. M., & Craik, F. I. M. (1988). Effects of aging and task difficulty on divided attention perfor-

mance. Journal of Experimental Psychology: Human Perception and Performance, 14:267-280.

McFarland, R. A., Domey, R. G., Warren, A. B., & Ward, D. C. (1960). Dark-adaptation as a function of age: I.

A statistical analysis. Journal of Gerontology, 15:149-154.

McKnight, A. J.. & McKnight, A. S. (1991). The effect of cellular phone use upon driver attention. AAA

Foundation for Traffic Safety. Washington, DC.

McPherson, K., Ostrow, A., Shaffron, P., & Yeater, R. (1988). Physical fitness and the aging driver. AAA

Foundation for Traffic Safety, Washington, DC.

Mihal, W. L., & Barrett, G. V. (1976). Individual differences in perceptual information processing and their

relation to automobile accident involvement. Journal of Applied Psychology, 61(2):229-233.

Mortimer, R. G. (1988). Headlamp performance factors affecting the visibility of older drivers in night driving.

In Transportation in an Aging Society. Transportation Research Board Special Report 218(2). National

Research Council: Washington, DC.

Murray, M. P., Gardner, G. M., Molinger, L. A. and Sepic, S. B. (1980). Strength of isometric and isokinetic

contractions, knee muscles of men aged 20-86. Physical Therapy, 60:412-419.

National Highway Traffic Safety Administration (1986). State and provincial licensing systems: Comparative

data. Washington, DC: U.S. Department of Transportation.

Olson, P. L. (1974). Aspects of driving performance as a function of field dependence. Journal of Applied

Psychology, 61:229-233.

Olson, P. L., & Sivak, M. (1986). Perception-response time to unexpected roadway hazards. Human Factors,

28(1):91-96.

Ostrow, A. C., Shaffron, P., & McPherson, K. (1992). The effects of a joint range-of-motion physical fitness

training program on the automobile driving skills of older adults. Journal of Safety Research, 23(4):207-

219.

Owsley, C., Ball, K., Sloane, M. E., Roenker, D. L., & Bruni, J. R. (1991). Visual/perceptual/cognitive

correlates of vehicle accidents in older drivers. (Paper No. 910368). Washington, DC.: Transportation Re-

search Board.

Owsley, C., Sekular, R., & Boldt, C. (1981). Aging and low-contrast vision: Face perception. /nvestigative

Ophthalmology and Visual Science, 21:362-368.

Panek, P. E., & Rush, M. C. (1981). Simultaneous examination of age related differences in the ability to

maintain and reorient auditory selective attention. Experimental Aging Research, 7:405-416.

Pitts, D. G. (1982). Visual acuity as a function of age. Journal of the American Optometric Association,

53:117-124.

Planek, T. W., & Fowler, R. C. (1971). Traffic accident problems and exposure characteristics of the aging

driver. Journal of Gerontology, 26:224-—230.

Plude, D. J., & Hoyer, W. J. (1986). Age and the selectivity of visual information processing. Psychology and

Aging, 1:4-10.

Ponds, R. W. M., Brouwer, W. H., & van Wolffelaar, P. C. (1988). Age differences in divided attention in a

simulated driving task. Journal of Gerontology, 43:151-156.

Pulling, N. H., Wolf, E. S., Sturgis, S. P., Vaillancourt, D. R., & Dolliver, J. J. (1980). Headlight glare

resistance and driver age. Human Factors, 22(1):103-112.

Rabbitt, P. M. A. (1979). How old and young subjects monitor and control responses for accuracy and speed.

British Journal of Psychology, 70:305-311.

Rabbitt, P. M. A. (1965). An age-decrement in the ability to ignore irrelevant information. Journal of Gerontol-

ogy, 20:233-238.

Ranney, T. A., & Pulling, N. H. (1990). Performance differences on driving and laboratory tasks between

drivers of different ages. Transportation Research Record 1281. Transportation Research Board, National

Research Council: Washington, DC.

Retchin, S. M., Cox, J., Fox., M., & Irwin, L. (1988). Performance-based measurements among elderly drivers

and nondrivers. Journal of the American Geriatrics Society, 36:813-819.

Richards, O. W. (1977). Effects of luminance and contrast on visual acuity, ages 16-90 years. American

Journal of Optometry and Physiological Optics, 54:178-184.

Rock, M. L. (1953). Visual performance as a function of low photopic brightness levels. Journal of Applied

Psychology, 37(5):412—427.

Rogers, P. N., & Janke, M. K. (1992). Performance of visually impaired heavy-vehicle operators. Journal of

Safety Research, 23:159-170.

Rothe, J. P. (1990). The Safety of Elderly Drivers: Yesterday’s Young in Today’s Traffic. Transaction Pub-

lishers: New Brunswick.

76 LLANERAS ET AL.

Sabey, B. E., & Stoughton, G. C. (1975). Interacting Roles of Road Environment, Vehicle and Road User in

Accidents. U.K. Transport and Road Research Laboratory, Crowthorne, Berkshire, England, September.

Salthouse, T. A. (1990a). Working memory as a processing resource in cognitive aging. Developmental Re-

view, 10:101-124.

Salthouse, T. A. (1990b). Influence of experience on age differences in cognitive functioning. Human Factors,

32(5):551-569.

Salthouse, T. A. (1985). Speed of behavior and its implications for cognition. In J. E. Birren and K. W. Schaie

(Eds.), Handbook of the Psychology of Aging (2nd Ed.), pp. 400-426). New York: Van Nostrand Reinhold.

Salthouse, T. A. (1982). Adult cognition. New York: Springer-Verlag.

Salthouse, T. A., Rogan, J. D., & Prill, K. A. (1984). Division of attention: Age differences on a visually

presented memory task. Memory & Cognition, 12:613-620.

Sanders, A. F. (1970). Some aspects of the selective process in the functional field of view. Ergonomics,

13:101-117.

Sanders, M. S. (1981). Peak and sustained isometric forces applied to a truck steering wheel. Human Factors,

23(6):655-660.

Santrock, J. W. (1985). Adult Development and Aging. Wm. C. Brown Publishers: Dubuque, IA.

Schieber, F. (1988). Vision assessment technology and screening older drivers: Past practices and emerging

techniques. In Transportation in an Aging Society. Transportation Research Board Special Report 218(2).

National Research Council: Washington, DC.

Schmidt, I., & Connolly, P. L. (1966). Visual considerations of man, the vehicle, and the highway. SAE Report

SP-279.

Schneider, W., & Shiffrin, R. M. (1977). Controlled and automatic human information processing: I. Detec-

tion, search, and attention. Psychological Review, 84:1-66.

Scialfa, C. T., Garvey, P. M., Gish, K. W., Deering, L. M., Leibowitz, H. W., & Goebel, C. C. (1988).

Relationships among measures of static and dynamic visual sensitivity. Human Factors, 30:677-687.

Scialfa, C. T., Kline, D. W., & Lyman, B. J. (1987). Age differences in target identification as a function of

retinal location and noise level: Examination of the useful field of view. Psychology and Aging, 2:14-19.

Sekular, R., & Ball, K. (1986). Visual localization: Age and practice. Journal of the Optical Society of America

A, 3:864-867.

Sekuler, R., Kline, D., & Dismukes, K. (Eds.). (1982). Aging and Human Visual Function. New York: Alan R.

Liss.

Sharpe, J. A., & Sylvester, T. O. (1978). Effects of aging on horizontal smooth pursuit. Investigative Ophthal-

mology and Visual Science, 17:465-468.

Shiffrin, R. M., & Schneider, W. (1977). Controlled and automatic human information processing: II. Percep-

tual learning. automatic attending, and a general theory. Psychological Review, 84:127-—190.

Shinar, D. (1978). Driver performance and individual differences in attention and information processing:

Volume I. Driver inattention. (Tech. Report DOT-HS-8-801819). Washington, DC: U.S. Department of

Transportation, National Highway Traffic Safety Administration.

Shinar, D. (1977). Driver visual limitations, diagnosis, and treatment. (Tech. Report DOT-HS-5-01275).

Bloomington: Indiana University.

Shinar, D. (1976). The effects of age on simple and complex visual skills. Paper presented at the Annual

Convention of the Western Psychological Association, Los Angeles.

Shinar, D. (1975). Practice effects on simple and complex visual skills. Paper presented at the Annual Meeting

of the American Academy of Optometry.

Shinar, D., Mayer, R. E., & Treat, J. R. (1975). Reliability and validity assessments of a newly developed

battery of driving-related vision tests. Proceedings of the 19th Annual Meeting of the American Association

for Automotive Medicine. San Diego, Calif.

Shinar, D., McDonald, S. T., & Treat, J. R. (1978). The interaction between driver mental and physical

conditions and errors causing traffic accidents: An analytical approach. Journal of Safety Research, 10:16-

23.

Shinar, D., McDowell, E. D., Rackoff, N. J., & Rockwell, T. H. (1978). Field dependence and driver visual

search behavior. Human Factors, 20(5):553-559.

Shinar, D., & Schieber, F. (1991). Visual requirements for safety and mobility of older drivers. Human

Factors, 33(5):507-519.

Shipley, T. (1974). Thresholds and resolution in Human vision: A new approach to night vision testing.

Human Factors, 16(1):56-64.

Singleton, W. T. (1955). Age and performance timing on simple skills. In Old Age in the Modern World:

Report of the 3rd Congress of the International Association of Gerontology, pp. 221-231. Edinburgh: Living-

stone.

Singleton, W. T. (1954). The change of movement timing with age. British Journal of Psychology, 45:166-172.

ABILITIES, AGE, AND DRIVING PERFORMANCE aT

Sivak, M., Flannagan, M., & Olson, P. L. (1987). Brake lamp photometrics and automobile rear signalling.

Human Factors, 29(5):533-540.

Sivak, M., Olson, P., & Pastalan, L. (1981). Effects of driver’s age on nighttime legibility of highway signs.

Human Factors, 23(1):59-64.

Sloane, M. E., Owsley, C., & Alverez, S. L. (1988). Aging, senile miosis and spatial contrast sensitivity at low

luminance. Vision Research, 28:1235-1246.

Somberg, B. L., & Salthouse, T. A. (1982). Divided attention abilities in young and old adults. Journal of

Experimental Psychology: Human Perception and Performance, 8:651-663.

Spirduso, W. W. (1982). Physical fitness in relation to motor aging. In Mortimer, F. Pirozzolo, & G. Maletta

(Eds.), Aging motor system (pp. 120-151). New York: Praeger.

Staplin, L. (1990). Diminished driver capability and the use of traffic control devices (Report FHWA/RD-90/

061). Washington, DC: U.S. Department of Transportation.

Staplin, L., & Fisk, A. D. (1991). A cognitive engineering approach to improving signalized left-turn intersec-

tions. Human Factors, 33(5):559-572.

Staplin, L., Janoff, M., & Decina, L. (1985). Reduced lighting during periods of low traffic density (Final

Report: FHWA Contract DTFH61-83-C-00056). McLean, VA: Turner-Fairbank Highway Research

Center.

Staplin, L., & Lyles, R. W. (1992). Age differences in motion perception and specific traffic maneuver prob-

lems. Transportation Research Record 1325. Transportation Research Board, National Research Council:

' Washington, DC.

States, J. D. (1985). Musculoskeletal system impairment related to safety and comfort of drivers over 55. In J.

W. Malfetti (Ed.), Needs and Problems of Older Drivers: Survey Results and Recommendations (pp. 63-76).

Falls Church, VA: AAA Foundation for Traffic Safety.

Stelmach, G. E., & Goggin, N. L. (1989). Psychomotor decline with age. In W. W. Spirduso and H. M. Eckert

(Eds.), Physical activity and aging (american Academy of Physical Education Papers No. 22, pp. 6-18).

Champaign, IL: Human Kinetics Books.

Stelmach, G. E., & Nahom, A. (1992). Cognitive-motor abilities of the elderly driver. Human Factors,

34(1):53-66.

Stock, M. S., Light, M. O., Douglass, J. M., & Burg, F. D. (1970). Licensing the driver with musculoskeletal

difficulty. Journal of Bone and Joint Surgery, 52-A(2):343-346.

Stoudt, H. W. (1981). The anthropometry of the elderly. Human Factors, 23(1):29-37.

Sturr, J. F., Kline, G. E., & Taub, H. A. (1990). Performance of young and older drivers on a static acuity test

under photopic and mesopic luminance conditions. Human Factors, 32(1):1-8.

Sussman, E. D., Bishop, H, Madnick, B., & Walter, R. (1982). Driver inattention and highway safety. In

Transportation Research Record 1047 (pp.40-48). Washington, DC: National Research Council, Transpor-

tation Research Board.

Szafran, J. (1951). Changes with age and with exclusion of vision in performance at an aiming task. Quarterly

Journal of Experimental Psychology, 3:111-118.

TRB (1988). Transportation in an Aging Society. Transportation Research Board Special Report 218(2).

National Research Council: Washington, DC.

U.S. Department of Health, Education, & Welfare. (1977). Monocular visual acuity of persons 4-74 years in

the United States, 1971-1972 (Publication 201, Series II). Washington, DC: U.S. Government Printing

Office.

van Wolffelaar, P. C., Rothengatter, J. A., & Brouwer, W. H. (1990). Elderly drivers’ traffic merging decisions.

In G. R. Cunningham (Ed.), Vision in Vehicles III: Proceedings of the 3rd International Conference on

Vision in Vehicles. Amsterdam: North-Holland.

Wald, G. (1968). The molecular basis of visual excitation. Science, 127:222—226.

Waller, P. F. (1991). The older driver. Human Factors, 33(5):499-506.

Waller, P. F., House, E. G., & Stewart, J. R. (1977). An analysis of accidents by age. Chapel Hill: University of

North Carolina, Highway Safety Research Center.

Walker, J., Sedney, C., Wochinger, K, Boehm-Davis, D. A., & Perez, W. A. (1993). Age differences in the

useful field of view in a part-task driving simulator. Proceedings of the Human Factors and Ergonomics

Society 37th Annual Meeting. Human Factors and Ergonomics Society: Santa Monica, CA.

Weale, R. A. (1963). The Aging Eye. New York: Harper & Row.

Weiss, A. D. (1965). The locus of reaction time change with set, motivation, and age. Journal of Gerontology,

20:60-64.

Welford, A. T. (1977). Motor performance. In J. E. Birren, and K. W. Schaie (Eds.), Handbook of the Psychol-

ogy of Aging. Van Nostrand Reinhold: New York.

Wierwille, W. W., Casali, J. G., & Repa, B. S. (1983). Driver steering reaction time to abrupt-onset cross-

winds, as measured in a moving-base driving simulator. Human Factors, 25(1):103-116.

78 LLANERAS ET AL.

Williams, A. F., & Carsten, O. (1989). Driver age and crash involvement. American Journal of Public Health,

79:326-327.

Witkin, H. A., Lewis, H. B., Hertzman, M., Machover, K., Meissner, P. B., & Wapner, S. (1954). Personality

through perception. New York: Harper.

Wolbarsht, M. L. (1977). Tests for glare sensitivity and peripheral vision in driver applicants. Journal of Safety

Research, 9:128-139.

Wolf, E. (1967). Studies on the shrinkage of the visual field with age. HRB Record 164, 1-7.

Yanik, A. J. (1986). Aging factors that affect the driving task. In Proceedings of the Fourth International

Conference on Mobility and Transport for Elderly and Disabled Persons (pp.747-757). Ottawa, Ontario,

Canada: Transport Canada.

Yee, D. (1985). A survey o the traffic safety needs and problems of drivers ages 55 and over. In J. L. Malfetti

(Ed.), Drivers 55+: Needs and problems of older drivers: Survey results and recommendations (pp. 96-128).

Falls Church, VA: AAA Foundation for Traffic Safety.

From the Editor

The Journal of the Washington Academy of Sciences has enjoyed a long and

venerable history. With this issue we begin our 83rd year of continuous publica-

tion. Volume 82 closed with a total of 222 pages. We might hope to publish as

many in volume 83, but a paucity of manuscript submissions over the past few

years has caused us to fall behind in our normal publishing scheduled. Future

issues may also be thinner than usual.

The cost of publication continues to increase, though not dramatically, for

our Journal. Currently such costs run slightly in excess of $64 per page. The

society absorbs this cost by asking only that those who can afford through their

grant or by other means pay for their page charges. The current format of the

Journal was developed with a view to economy, so we continue to be able to take

advantage of the finest printing technology and materials available.

The content of the papers published in the Journal reflects the subject matter

~ areas represented by manuscripts received. There is no editorial attempt to mold

areas of emphasis in, or to exclude others from, the Journal. The Journal of the

Washington Academy of Sciences is devoted to reporting results of original

research, critical reviews, and historical articles that contribute significantly to

our knowledge of all areas of science and engineering. Shorter communications

and reviews articles are also published occasionally.

We practice positive editorship. If your manuscript is judged by our reviewers

and Board of Editors to have scientific merit and to be suitable for inclusion of

the Journal, the editorial office will work with you. We realize that your manu-

script emerges from intensive efforts. In turn, we feel that the additional time

taken to offer what we hope are helpful, constructive, and detailed suggestions in

matters of style and journal format are justified. Of course, any energies ex-

pended on you part as author(s) in these matters will help to speed things along

and are much appreciated!

You are invited to submit manuscripts for publication consideration. We

hope that you will encourage your colleagues working in any area of science and

engineering to submit. Instructions to contribute were revised in 1993 and pub-

lished in Vol. 82. No. 1, March 1992 Journal, or can be obtained by writing to the

Editor:

Dr. Bruce F. Hill

Mount Vernon College

2100 Foxhall Road N.W.

Washington, D.C. 20007

The standard of scientific quality that has become the hallmark of the Journal of

the Washington Academy of the Sciences has depended, and will continue to

depend, on your contributions as authors.

79

i

PG lis miu) cy al: ast} oy ‘ecole ai att an109% i aie bast ihe

OF

ORR ESTEE Jovkal ir) «aso

i ph ity I est BG Tite Hi

SES AER OTT OCE AER EE ' RM al vine AOE Vote sell oer penta fy 9

Ky Dt ) *

a Ce a yy Ay ory Pee rie ae i)

Fey WERE OG DAR DOS ORS : B

; j

i eh Ye ee ok fe Po. oak atotahe oh red

y Mus oy i. ¥ "TOR AT Sd ou | ¢ nt a ait

a a , a iT } |e 5 4 :

| Mah Fe) PENT WE SONS ine

‘ p ; oH ’ rr ns Fi ‘ ' cn

+ ee $9 ae Oe i tvtriegan Bhs ae ee 8 Pie i 2 ;

L ay RES CAR AE GAS DHE g SOR Mee On, a? Ree RRL Pe ee rit

, bey Oey hee z TB oe hal Cao pis i thy ! v

j SheMet hy PON LYS: ORCI BaD) BE Barty: :

+ ur cee hee i ety : j Pat 7-3 hud $ ;

} Bead heey: Pry ri Loe BUG Rix Pome Tee eee ore : -t

Re : \ td Pee ad LNG MKT Fae ad ; if Pel Apts, try ay

“ r hae Ad hl a Oe Pe ae Pay iaale 6 ee 9 yf

bite, leak ‘ ve

v1 ree vs ‘ fy

Pye ae ‘i? ne 4 ye P : a © se \

a af We I ey eed ended 4 ak ye f ee if) 3

ri Be, t hay adhe a. P's uf feuy , i ¥4 - a rey bie sb Ay avehOor

. Aha, belie ee aU ee A ae Se Dae aE Pe i e i

i

ate. Ro htae B ed Pata : ; Hil ohp se rel vane Ye as: ‘ Pan

bogs EV ST TO A) eA ea vi Oe Fer eee ipa RE EIN vf 1 tT lovee Bib , salaaies Me:

i bynd a oe Peres . wt aa Py ne Pity Akane bpotry Pieehe bike ni t oh | de is oy rh, —

EES OO ATR. Ree Oe le SR Oo CANON te aoe

? tJ ; ; SO J .

i A , \ ( 3 ‘ }

h oe ' Pe j ee A 1 f Ps : p ay % } ‘

‘ i r ' + i be) | ¥ \ ed fy ‘ n .s

ne ; | bain | ag at i , At yin

a3 Bool & bevoms ead eee gio? ey WiTabeoA, notanisiea

“HOHE ity Ce Het nape vi ‘biderowlaysey wheats

mb mina sae Sal pia, as aS ite indms.tiaist Five oui

v

5

—~,

ee

_

—

tr. =

“ +h.

= Sack

=

-

4 my yp iP rtieilas &

Die. ma ai, firs vdeo: mosaa ve Jenny

| Saale Lie aie yA OP haar ia

yea Be Eee Of Vet

J 4 1 7

‘ eB Lr Aye

oiinny cauning tesgt pty 10 §

™~

~

poe

ts

=

he

=

~

=

=

+

=

“

a wae idee ny : i

f : { 1 i j

a ; a) hue it en y H hes } bash yey ag ¥e ;

sey t % S\

f i ¢ ) br Pinay | : ‘ ‘i

: { fr * 2 ' F

oy ‘ ur Ae Aa Rae {

\

i 7 :

ny ek 8 ; ‘ y ;

asl | , Be a, | ede te ke ood -

PET 5) 28 ; een i iy +} . rye ra ‘ rm

u nk ; 1 } y I i . , a 4h i ‘

! : i

ee ON ee uct Me Oe se oT ew ae Bi ih i «i

es iy t] ‘ Phe Waa se ORE Pree Bee th ici ie ei | vl i : By ‘

. | ee MEE SRR AAS ERE ET a OR SS LEO TCH Sa YHOo 4 ae ir wot

+

a hip ‘ i Che) a ¥ j ¥ ,

i by : ee oF | 4 k 4 a. un des i hg *

POG tie yA Sg rgery BES €

OF eins en tireveline

eB hy me (aha ied cae Pe SEAN onl wren ven Ces Pe ; i i

Be a g Nd ENP GGA ED Dabo ORI tS AS aes WAS) ehh |

ny

2

{

f

i 4

x

{ ae

S

4 {

a0) na fie | ity i cass baba OK nfo i me soon oe usd toy

DELEGATES TO THE WASHINGTON ACADEMY OF SCIENCES,

REPRESENTING THE LOCAL AFFILIATED SOCIETIES

Philasophicall Society Of Washington .....).....0..0.0 06. 0c cede scence eves Thomas R. Lettieri

Aintacopolorical Society of Washington ......... 2.0.0.6... 20.. ce dae seed enes Belford Lawson III

Peirce SOCIETY OF WaSMINPION Ub ilo. bes siecle sae cnciec eed enule cesses Kristian Fauchald

Muemical Society OF WaSHINSTOM 6. c. 2.266 he eee bela eee ck eee hen bee si Elise A. B. Brown

Emomolorical Society of Washington ..................6.... cece ence es F. Christian Thompson

eR COCTADIMIC SOCICLY, 4). occu occa de nek bod mene cbaadcenmanae Stanley G. Leftwich

Bee OGICEY Ol NV ASMINPLON Seca oe. sas ilss'e eid aiene Gees tals ee de dinpe ven cee s VACANT

micata society of the District of Columbia .............05 00... ca ceases een teens John P. Utz

Peemees) society of Washington, DC .... 2... cca cee ec ee ees Thomas G. Manning

MEME SOCICEY OF VW aASMINGTON |. 2. <janjoc emia tne eels deans case eedavaasecce eas Muriel Poston

Society of American Foresters, Washington Section ..................0ee neces Eldon W. Ross

SEE PME ICICI Ol: FE TIOTNEGES wh 5 oloic.s oe ao 6 2 oo ark ey ied ain, mr aicile wale ciel Wie sled os Alvin Reiner

Institute of Electrical and Electronics Engineers, Washington Section ........ George Abraham

American Society of Mechanical Engineers, Washington Section ......... Clayton W. Robson

Helanmthological Society of Washington .............0. 00 ccc cece eset eee eeeecenes VACANT

American Society for Microbiology, Washington Branch .................. Herman Schneider

Society of American Military Engineers, Washington Post ................. William A. Stanley

American Society of Civil Engineers, National Capital Section .............. John N. Hummel

Society for Experimental Biology and Medicine, DC Section .............. Cyrus R. Creveling

Peer imcimational, Washington Chapter ..............0. 00050 0c cee s een eees Pamela S. Patrick

American Association of Dental Research, Washington Section ............. J. Terrell Hoffeld

- American Institute of Aeronautics and Astronautics, National Capital

ESE EE oy oR IETS OS SOI ae an ee Reginald C. Smith

American Meteorological Society, DC Chapter ...............000 000 cece ends A. James Wagner

memeecience Society Of WaShINnetON ..... 4.0.2... 66. de ce cece cece ee ccccccens To be determined

Acoustical Society of America, Washington Chapter ........................ Richard K. Cook

paenican) Nuclear Society, Washington Section ..2...0...0.......0. cee eee ewes Kamal Araj

Institute of Food Technologists, Washington Section ....................0.0 eee Roy E. Martin

American Ceramic Society, Baltimore-Washington Section .................. Curtis A. Martin

MME AMEE EEN 0) hued oso! aieiei isla egy Ud bie als bia bla weg ule'e oh vein ate slaty «6 Regis Conrad

hy asmiepion Enstory of Science Club ... 0.0.00... 0000 ccc ce cece eee ees Albert G. Gluckman

American Association of Physics Teachers, Chesapeake Section ............. Robert A. Morse

Optical Society of America, National Capital Section ...................... William R. Graver

American Society of Plant Physiologists, Washington Area Section ............. Steven J. Britz

Washington Operations Research/Management Science Council .............. John G. Honig

Instrument Society of America, Washington Section ...................00005 Donald M. Paul

American Institute of Mining, Metallurgical and Petroleum Engineers,

REP MEEPT SC CUMOM i ear eee eu eka oe OLS ou Wh Harold Newman

Prenat Capital ASITONOMECE 2). isd ijc ck cs cc cleee cece cele cceeccecuns Robert H. McCracken

Mathematics Association of America, MD-DC-VA Section ................. Sharon K. Hauge

sizes of Columbia Institute of Chemists ...........0....000.00000080000e William E. Hanford

District of Columbia Psychological Association ..............cc ccc cece e cece eeeees Ron Wynne

i suimeion Paint Technology Group)... .... 2.6.66 cee ieee ele tcc ee sne sues Lloyd M. Smith

American Phytopathological Society, Potomac Division .................... Kenneth L. Deahl

Society for General Systems Research, Metropolitan Washington

CL ESSTTETT OG ORIG AAI TE COPEL TAS ie Ale Ru 8 ER SR a David B. Keever

Fiaman Factors Society; Potomac Chapter... ......000.2.0..60c00. 0 cease Thomas B. Malone

American Fisheries Society, Potomac Chapter ..........00.¢50000.ccecee ees Dennis R. Lassuy

Association for Science, Technology and Innovation ....................000.000e Ralph I. Cole

Pet ery SOCIO ORICA SOCICLY 206i ode eo hela ceelya idee Goldwioaeldd ced 4's Ronald W. Manderscheid

Institute of Electrical and Electronics Engineers, Northern Virginia

SS ONT OURO MERU A UCHR EU RE SN TARAS RO Gn OU nd Blanchard D. Smith

Association for Computing Machinery, Washington Chapter ............. Charles E. Youman

Pee SIR LOM) SUALISHICHl SOCICEY a. ecpts sce waeitiemen se tioded welded Uslsla dlasaae sere Nancy Flournoy

Society of Manufacturing Engineers, Washington, DC Chapter ............... James E. Spates

Institute of Industrial Engineers, National Capital Chapter ................... James S. Powell

Delegates continue to represent their societies until new appointments are made.

Washington Academy of Sciences 2nd Class Postage Paid

2100 Foxhall Road, NW at Washington, DC

Washington, DC 20007-1199 and additional mailing offices.

Return Postage Guaranteed

wel /

WH VOLUME 83

Number 2

Jour nal of the June, 1993

WASHINGTON

ACADEMY... SCIENCES

ISSN 0043-0439

Issued Quarterly

at Washington, D.C.

CONTENTS

Articles:

EDUARDO SALAS, JANIS A. CANNON-BOWERS &

ELIZABETH L. BLICKENSDERFER, “‘Team performance and training

PARSALGHE WIMCRAMMPHPTIMCIDIES 26 Se crs siieiciids sees eye a cps nie sis rule.ee dale eld mae bole we

DEBORAH A. BOEHM-DAVIS, ROBERT W. HOLT &

ROBERT D. PETERS, “Effects of different data base formats on information

REET ONES [0 Os ROMP SN ISL ol a Ck aa pa RO a eR

Ce

SMITHSON AR

\

NOV U 2 1998

LIBRARIES

Washington Academy of Sciences

Founded in 1898

EXECUTIVE COMMITTEE

President

John H. Proctor

President-Elect

Rev. Frank R. Haig, SJ

Secretary

Thomas R. Lettieri

Treasurer

Norman Doctor

Past President

Stanley G. Leftwich

Vice President, Membership Affairs

Cyrus R. Creveling

Vice President, Administrative Affairs

Grover C. Sherlin

Vice President, Junior Academy Affairs

Marylin B. Krupsaw

Vice President, Affiliate Affairs

Thomas W. Doeppner

Board of Managers

James W. Harr

Clifford M. Krowne

Herbert H. Fockler

Nina M. Roscher

William B. Taylor

Neal F. Schmeidler

REPRESENTATIVES FROM

AFFILIATED SOCIETIES

Delegates are listed on inside rear cover

of each Journal.

ACADEMY OFFICE

2100 Foxhall Road, N.W.

Washington, D.C. 20007

Phone: (202) 337-2077

EDITORIAL BOARD

Editor:

Bruce F. Hill, Mount Vernon College

Associate Editors:

Milton P. Eisner, Mount Vernon Col-

lege

Albert G. Gluckman, University of

Maryland

Marc Rothenberg, Smithsonian Insti-

tution

Marc M. Sebrechts, Catholic Univer-

sity of America

Edward J. Wegman, George Mason

University

The Journal

This journal, the official organ of the Washing-

ton Academy of Sciences, publishes original

scientific research, critical reviews, historical

articles, proceedings of scholarly meetings of

its afhliated societies, reports of the Academy,

and other items of interest to Academy

members. The Journal appears four times a

year (March, June, September, and De-

cember). The December issue contains a di-

rectory of the current membership of the

Academy.

Subscription Rates

Members, fellows, and life members in good

standing receive the Journal without charge.

Subscriptions are available on a calendar year

basis, payable in advance. Payment must be

made in U.S. currency at the following rates:

U.S. and Canada... 3) =.45 eee $25.00

Other countries. 2.23.02. eee 30.00

Single copies, when available ....... 10.00

Claims for Missing Issues

Claims will not be allowed if received more

than 60 days after the day of mailing plus time

normally required for postal delivery and

claim. No claims will be allowed because of

failure to notify the Academy of a change of

address.

Notification of Change of Address

Address changes should be sent promptly to

the Academy Office. Such notification should

show both old and new addresses and zip

codes.

POSTMASTER: Send address changes to

Washington Academy of Sciences, 2100 Fox-

hall Road, N.W. Washington, DC 20007-

1199.

Journal of the Washington Academy of Sciences (ISSN 0043-0439)

Published quarterly in March, June, September, and December of each year by the Washing-

ton Academy of Sciences, 2100 Foxhall Road, N.W., Washington, DC, 20007-1199. Second

Class postage paid at Washington, DC and additional mailing offices.

Journal of the Washington Academy of Sciences,

Volume 83, Number 2, Pages 81-106, June 1993

Team Performance and Training Research:

Emerging Principles

Eduardo Salas

Janis A. Cannon-Bowers

Elizabeth L. Blickensderfer

Naval Air Warfare Center Training Systems Division, Orlando, FL

ABSTRACT

It is widely recognized that teams play a vital role in modern society. Successful teams are

characterized by high levels of coordination and efficiency. In efforts over the past two

decades to understand teams, researchers have devoted considerable energy to investigating

team performance and team training. This paper reviews both past and current themes and

issues in team performance and training research. Issues addressed include: theoretical mod-

els of team performance, measurement issues in team research, and team training tech-

niques. In addition, we offer an initial set of theoretically and empirically driven principles of

team performance and training that can guide our thinking about the nature of these com-

plex phenomena. Finally, we discuss additional areas of team performance which demand

further research.

Introduction

For those who have been on a high performance team, be it on the athletic

field, in a surgery ward, in an aircrew, in a military confrontation, or in a fire

fighting squadron, the word “teamwork” inspires a sense of exhilaration. To

those observing such teams, the extensive demands imposed upon each team are

obvious. In fact, it is the crucial tasks which teams typically perform that have

led to an increasing amount of team research over the past 50 years.

The broad question this work seeks to examine is: how do teams function

while performing these critical tasks? To understand team functioning, one

81

82 SALAS, CANNON-BOWERS, AND BLICKENSDERFER

must understand how individuals interact, coordinate, communicate, exchange

information, and adapt to task demands. In addition to the task demands, one

must assess how the team is organized, what individuals bring to the task, and

how organizational characteristics influence team performance. Each of these

factors represent a piece of the puzzle that is team performance.

Team performance, then, is comprised of a complex set of factors. The com-

plexity in describing team performance creates difficulties in research. In fact,

there are difficulties in simply defining “‘team”’. In this paper, a team is defined

as a set of two or more individuals who must interact cooperatively and adap-

tively in pursuit of shared, valued objectives. Further, team members have

clearly defined, differentiated roles and responsibilities, hold task relevant

knowledge, and are interdependent (1.e., must rely on one another in order to

accomplish goals) (Dyer, 1984; Orasanu & Salas, 1993; Salas, Dickinson, Con-

verse, & Tannenbaum, 1992). The remainder of this paper applies only to this

kind of team.

In the past few decades, a great deal of time, money, and effort have been

directed at the investigation of team functioning. Until recently however, little

concrete knowledge on the nature of teamwork or effective team training ex-

isted. The purpose of this paper is to review the team performance and training.

literature. We accomplish this by first providing a historical perspective. That is,

we have chosen themes 1n past work that we believe are fundamental to under-

standing contemporary team performance and training research. Following this

brief review, more recent team performance and training issues are discussed.

Next, we offer an initial set of theoretically and empirically driven principles

that can guide our thinking about the nature of team performance and training.

Finally, we discuss additional areas which demand further research.

Before beginning the review, we should acknowledge that we have not over-

looked the work that has been performed in the “‘small group” research arena.

While we believe that this research has contributed significantly to our knowl-

edge of small group dynamics and phenomena, it is only tangential to our focus

on teams. That is, small group research employs a looser definition of the term

“group” than we use to define teams. In particular, we reserve the term “team”

for those groups that are interdependent and working toward a common goal;

neither of which have been required as central to the definition of groups. In

short, we have borrowed from small group literature as appropriate, and we

chose only citations that are relevant to and consistent with our purpose. For

those interested, see Levine and Moreland (1990) for a full review of small group

research. It should be noted that, although the vast majority of research cited in

this paper was conducted on military teams, we believe that these results general-

ize to other teams with similar characteristics (i.e., interdependencies among

i eal

TEAM PERFORMANCE AND TRAINING RESEARCH 83

members, complex tasks, and stressful environments). Therefore, the conclu-

sions we draw in this paper should have applicability to: fire fighting teams,

surgical teams, emergency medical teams, aviation crews, law enforcement

teams, air traffic control teams, and the like.

Research in the 1960s and 1970s

Interpersonal Skills

Although group research has existed for the greater part of this century, re-

search focused specifically on teams generally did not appear until the 1960s.

We will begin this review with the literature from that decade, as that work has

_ provided the major underpinnings of the recent team research. The 1960s litera-

ture represents a mixture of team and group studies, with both early and more

recent team researchers having utilized the group literature to formulate hypoth-

- eses for team performance. One example of the influence of group literature can

be found in the hypothesis that the level of team members’ interpersonal skills

influenced team performance. Tuckman (1965), for example, advocated that

good interpersonal relationships between small group members must be formed

before they are able to focus effectively on performance of the team task. Al-

though Tuckman focused on “groups” as opposed to “‘teams’’, this proposition

directed research attention to “team’”’ concepts such as coordination, interac-

tion, and communication.

While the interpersonal skill proposition seemed applicable to group research,

other variables such as individual ability and skill began to attract attention also.

_ This was part of a gradual shift in attention away from interpersonal skills and

toward the effects of individual skills and abilities on team performance. It is

important to note, however, that Tuckman’s ideas were not discounted entirely.

In fact, his ideas were later incorporated into models of team performance (e.g.,

Morgan et al., 1986; discussed later in this paper).

Individual Ability and Skills: Individual training or team training?

As noted above, another area of interest in the 1960s-70s was whether or not

the skill levels of individual team members influenced overall team perfor-

mance and if individualized or team based training would be most effective. A

number of studies demonstrated that teams with members who had higher

individual task proficiency, abilities, and skills performed better (Gladstein,

1984; Kabanoff & O’Brien, 1979; Steiner, 1972). For example, researchers

found that the accuracy and speed of team performance were positively related

to the average skill level of individual team members (Kabanoff & O’Brien,

84 SALAS, CANNON-BOWERS, AND BLICKENSDERFER

1979; Klaus & Glaser, 1970; Terborg, Castore, & Deninno, 1976 ). Similarly,

research also found that teams with high individual skill levels reached criterion

performance with less training than did teams with average or poor individual

skill levels (Bouchard, 1969; George, Hoak & Boutwell, 1963; Hall & Rizzo,

1975; Klaus & Glaser, 1965; Tziner & Eden, 1985).

Other researchers argued that individual skill level is only one part of team

performance and, as with interpersonal skill, not the whole story. For example,

Bass & Barrett (1981) and Terborg et al. (1976) reported that the degree of

positive relationship between average skill level and performance quality was

small. In fact, Terborg et al. (1976) reported individual skill level as accounting

for a mere three percent of the variance in performance. As was discussed in a

number of literature reviews in the late 1970s, individual skill was apparently

not the only variable involved in team performance (Denson, 1981; Nieva,

Fleishman, & Reick, 1978; Wagner, Hibbits, Rosenblatt, Schulz, 1976).

Although results generally pointed to the importance of individual skills and

abilities, questions still existed as to why similar studies did not show the same

effect. Nieva et al. (1978) suggested that contingencies exist which differentially

influence the effectiveness of individual versus team based training. For exam-

ple, Nieva et al. (1978) proposed that the interaction requirement of the task

determines whether group or individual training is more effective. Thus, task

characteristics may effect the level of influence that individual skill will have in

team performance and, in turn, what type of training is most effective. The most

influential task characteristic might well be the level of interaction required by a

team task.

The influence of both individual and team proficiency on team performance

suggests that team members must concentrate on both individual skills and

team skills. When considering team performance in this light, a team task could

be considered “dual tasking’’. That is, team members must perform their indi-

vidual tasks at the same time as communicating and interacting appropriately

with their team members. In balancing their team duties along with their indi-

vidual duties, individual skills may be sacrificed or performed inefficiently and

result in a performance decrement. This type of performance decrement is

called process loss and occurs when team member efforts are wasted or dupli-

cated in the course of meeting coordination and communication behaviors

required for team performance. One of the first proponents of process loss,

Steiner (1972) attributed the small relationship between average skill level of the

team and overall team performance to individual team members’ inability to

balance the load between individual task and team responsibilities. This inabil-

ity to balance individual task performance with team duties suggests that train-

ing teams in both individual skills and team skills is a necessary part of team

training.

TEAM PERFORMANCE AND TRAINING RESEARCH 85

ANTECEDENTS

CONDITIONS TEAM PERFORMANCE

INDIVIDUAL TASK

MEMBER RESOURCES PERFORMANCE

EXTERNAL CONDITIONS TASK CHARACTERISTICS AND

IMPOSED ON THE TEAM DEMANDS

TEAM CHARACTERISTICS

TEAM PERFORMANCE

FUNCTIONS

Fig. 1. Conceptual model of team performance (adapted from Nieva, Fleishman, and Reick, 1978)

Once research had revealed that a number of variables were involved in team

training (i.e., individual skill levels, task characteristics), the next step was to link

the various pieces together into a conceptual model. Several theorists in the late

1970s constructed conceptual models of team performance. As an illustration,

Nieva et al. (1978) posited a model which highlighted the research focus on the

individual skills/team skills question. This model is depicted in Figure 1.

Looking at Figure 1, the two major elements of team performance are task

behaviors by individuals and task functions at the team level. The authors

maintained that individual task performance or team member interactions may ©

predominate, but that in most tasks, performance is probably determined by

both components. Figure | also shows that external conditions, such as the

organizational context, precede team performance. Additional precursors to

team performance are individual skills, abilities, and personality characteristics.

Finally, team characteristics such as size, group cohesiveness, intra-and inter-

team cooperation, communications standards and communication networks,

in addition to task characteristics and demands, also affect team performance.

While the Nieva et al. (1978) model reflected the emphasis on the individual

skills/team skills question, it was not until the 1980s that research would begin

to identify specific team skills that contribute to team performance.

Research in the 1980s

Taskwork and Teamwork

While the early research in the 60s and 70s indicated that effective team

performance was a combination of factors, subsequent research has focused

86 SALAS, CANNON-BOWERS, AND BLICKENSDERFER

more clearly on achieving an empirical distinction between “teamwork” and

“‘taskwork”’. Teamwork skills have been defined as those skills that related to

functioning effectively as a team member (Cannon-Bowers, Salas, & Converse,

1993). Taskwork skills, on the other hand, have been defined as those skills that

related to the execution of the task or the mission itself (e.g., operating equip-

ment, following procedures) (Cannon-Bowers et al., 1993). Results prompting

this conclusion first appeared in the late 1980s and have been a prominent

building block for subsequent research.

Considerable evidence supports the teamwork/taskwork division. First, a

number of researchers have shown that behaviors which are related specifically

to team functioning and which are independent of the task at hand are highly

important to team outcomes (e.g., Oser, McCallum, Salas, & Morgan, 1989;

Stout, Cannon-Bowers, Salas, & Morgan, 1990). Oser et al. (1989), for example,

found that several team behaviors significantly correlated with more effective

team performance. These included: members offering praise to one another for

doing well on a task, members suggesting to one another to recheck work for

errors, and members providing suggestions on the best way to locate an error.

Second, team process variables (e.g., communication, coordination, compen-

satory behavior) have been shown to influence team effectiveness (e.g., Stout et

al., 1990). In other words, the process (e.g., communication patterns, coordina-

tion) that a team goes through in achieving a particular outcome influences that

outcome. Third, effective teamwork behavior has appeared as fairly consistent

across tasks (e.g., McIntyre & Salas, in press). McIntyre and Salas (in press)

examined three teams who operated in similar military environments, but

whose tasks were distinctly different. One team was tasked with providing gun-

fire support to ground troops (naval gunfire support team), another team was

responsible for detecting, tracking, and defending against enemy submarines

(anti-submarine warfare team), while the third team was tasked to detect, track,

and defend against enemy surface and subsurface vessels (guided missile teams).

McIntyre and Salas (in press) identified common teamwork behaviors that all

teams exhibited. Thus, it appears that crucial teamwork behaviors can be iso-

lated from other task-related behaviors and are consistent across tasks.

Evolution of Teams

Another area of emphasis in the mid- 1980s examined the evolution and matu-

ration of operational teams (Glickman et al., 1987; Guerette, Miller, Glickman,

Morgan, & Salas, 1987; McIntyre, Morgan, Salas, & Glickman, 1988; Morgan,

Glickman, Woodward, Blaiwes, & Salas, 1986). The focus was to investigate the

ways that teams perform in the “naturalistic”? environment, and the research

involved real U. S. Navy command and control teams. McIntyre et al. (1988)

TEAM PERFORMANCE AND TRAINING RESEARCH 87

presented a description of the evolutionary process of the observed teams. A

summary of this description follows.

Phase 1: Team members explored individual roles and how roles meshed with their

teammates. Team members helped each other learn the task.

Phase 2: Members continued to seek role clarity on the team; asked many questions,

checked with each other; top ranked teams began to show different behavioral

patterns which indicated that they were “‘maturing”’ with respect to both task

and team.

Phase 3: Top ranked teams showed distinctly different behavioral patterns. For in-

stance, top-ranked teams showed a tendency to coordinate the gathering of

critical information for successful operation and tended to be ready to provide

required information while lower-ranked did not.

Phase 4: The behaviors which characterized top-ranked teams in Phase 3 became more

characteristic of teams in general. Importantly, team members showed a ten-

dency to ensure that their intended message was received. Lower-ranked

teams were less inclined to ask for feedback or help. Perhaps less trust existed

in the lower-ranked teams.

Phase 5: Top-ranked teams generally continued development of teamwork competen-

cies. Lower-ranked teams tended to plateau.

Phase 6: All teams became fuller-functioning teams and were much less stifled by

former prescribed group norms. Less distinction between top-ranked teams

and non-top-ranked-teams appeared.

The evolutionary perspective employed a longitudinal rather than cross-sec-

tional approach to teams and provided researchers the opportunity to monitor

team development (1.e., team skill acquisition) over time which, in turn, helped

researchers to better understand the development of taskwork and teamwork

skills. The approach also produced a number of implications for the design and

development of team training.

Tuckman (1965) also observed an evolutionary trend which divided group

development into several stages. Tuckman labeled the stages ““Forming”’,

“Storming’’, ““Norming’’, and “Performing”. Morgan et al. (1986) later incorpo-

rated Tuckman’s stages into their Team Evolution and Maturation (TEAM)

model. The model is depicted in Figure 2. Morgan et al. (1986) suggested that

teams progress through developmental phases which depend to a certain extent

upon members’ experience as a team, individual expertise, task characteristics,

and the environmental context. Not all teams are expected to pass through all

phases. This model also distinguished between task-oriented skills and team

skills training. ““Taskwork skills” training was considered to be training geared

toward imparting task skills such as understanding task requirements, discover-

ing operating procedures, and acquiring task information and other task ori-

ented issues. ““Teamwork skills” training was considered to be training aimed at

88 SALAS, CANNON-BOWERS, AND BLICKENSDERFER

“FIRST MEETING TRANSITION COMPLETION

(Beginning of lite cycle) PHASE | (Mid-point of life cycle) PHASE Il (End of life cycle)

**PRE-FORMING FORMING STORMING NORMING PERFORMING-| REFORMING PERFORMING-I| CONFORMING

Review of

| Accompllehmante

peneiP ivan

of Task Recycle

Assignments

|

|

Orientation |

TASKWORK MESS l Emotional

|

|

Response to

Taek Demands Open Exchange | Withdrawal

|

f Relevant l from Task

Completion and

|

coe

SKILLS

en)

Interpretations Emergence of : A Fare

Solutions ramework pie Deli

tcl) ompletion 30) e ney of

: sdsel 10) axe an Adinstmentta

Peepemen netinemient of Roles Erylonmental

TEAMWORK oes Beisisanees

Skiers intragroup Cohesion

Conflict

Testing |

Roles Demands

)

Dependence |

Investigation

of Group ii) |

“Adopted from Gersick (1985)

**Adopted from Tuckman (1965)

| o

|

|

| Remembering

Group

ENVIRONMENTAL DEMANDS AND CONSTRAINTS

(Social and Organizational Context)

Fig. 2. The Team Evolution and Maturation model (adapted from Morgan et al., 1986)

the behavioral interactions and attitudinal responses of team members that

must develop such as coordination, adaptation to varying situational demands,

mutual performance monitoring, and effective communication.

The 1980s’ study of teams over time and the empirical identification of the

taskwork/teamwork division were significant accomplishments for team re-

search. This work set the stage for the development of team performance mea-

surement tools which allowed investigators to identify specific cognitive and

behavioral components of taskwork and teamwork.

Teams in the 1990s

Identifying Team Skills

Research demonstrating that taskwork differed from teamwork enabled re-

searchers to identify and measure taskwork. However, identifying specific team

skills has proven highly difficult (Cannon-Bowers et al., 1993). A number of

researchers have developed general teamwork skill classifications. For example,

one strategy to team skill classification is a behavioral approach. This technique

was adopted by researchers for the U. S. Navy. The Navy sought to identify the

skill dimensions that comprise effective teamwork and the particular behaviors

within each of these dimensions that result in effective flight crew performance.

Importantly, this tactic emphasized trainable skills. A full description of the

‘TEAM PERFORMANCE AND TRAINING RESEARCH 89

approach used to determine which teamwork behaviors were related to aircrew

coordination appears in Prince and Salas (1993).

Briefly, approximately 60 behavioral statements were gathered from reviews

of past aircrew coordination literature and team training literature, observations

of team development, interviews of job experts, and surveys of the aviation

communication. Based on independent classification by job experts these behav-

iors were then arranged under seven dimensions: Mission Analysis, Assertive-

ness, Adaptability/Flexibility, Situational Awareness, Decision Making, Leader-

ship, and Communication. These aircrew coordination dimensions were then

decomposed into their knowledge, skills, and attitudes, and a training program

was developed. Importantly, validation efforts followed the identification of

behaviors. These included a comparison of the perceptions of the importance of

aircrew coordination behaviors of aviators from diverse wing communities

(Stout, Prince, Baker, Bergondy, & Salas, 1992). In other words, the seven behav-

ioral skill dimensions were the skills that subject matter experts considered vital

to effective flight performance. Further Prince, Brannick, Prince, & Salas (1992)

found evidence of discriminant validity between the dimensions which indi-

cated the multi-dimensionality of team processes. Helmreich and Foushee

(1993) have also contributed similar validation efforts.

Along with the work by Prince, Salas and their colleagues, a number of other

researchers—focused on a variety of environments—have also approached the

team issue from a skill-based perspective. This work has generated a variety of

team skill listings. An effort to synthesize these listings, Cannon-Bowers, Tan-

nenbaum, Salas, and Volpe (in press) consolidated a number of the different

teamwork skill labels and definitions found in the literature into the following

dimensions: adaptability, shared situational awareness, performance monitor-

ing and feedback, leadership/team management, interpersonal, coordination,

communication, and decision making. Associated subskills were also generated

for each skill dimension. Adaptability, for example, is defined as, ““The process

by which a team is able to use information gathered from the task environment

to adjust task strategies through the use of compensatory behavior and realloca-

tion of intra-team resources” (p. 42). The list of alternate labels for adaptability

were flexibility, compensatory behavior, and dynamic reallocation of function.

In this definition alone, Cannon-Bowers et al. (in press) drew from the work of a

number of authors (Alexander & Cooperband, 1965; Johnston & Briggs, 1968;

McCallum, Oser, Morgan, & Salas, 1989; McIntyre & Salas, in press; McIntyre

et al., 1988; Oser et al., 1989; Oser, Prince, & Morgan, 1990; Streufert & No-

gami, 1992). The lengthy list of cited works indicates the large number of team-

work classification/categorization schemas present in current team literature.

By sorting through the multitude of definitions and synthesizing them into one

90 SALAS, CANNON-BOWERS, AND BLICKENSDERFER

succinct list, Cannon-Bowers et al. (in press). demonstrated the links between

the different approaches to team skill classification and helped clarify the behav-

ioral definition of ““teamwork’’. Thus, some success has occurred in identifying

certain behavioral team skills (e.g., assertiveness). However, the abstract nature

of a number of other team process skills (e.g., adaptability, situational aware-

ness) has created difficulties in specifying the underlying abilities necessary to

perform the skills.

Shared Mental Models

Identifying a core set of teamwork skills is only a necessary first step in under-

standing team performance. In order to enhance our understanding of team-

work, the knowledge, skills, and abilities utilized in effective team performance

must be specified. However, specification remains elusive with respect to certain

teamwork behaviors such as coordinating action, adapting to changing task

conditions, and anticipating the needs of the task and team. Effective coordina-

tion demands that team members understand when specific behaviors are neces-

sary. This may depend on the task itself or characteristics, needs, and duties of

other team members (Prince, Chidester, Cannon-Bowers, & Bowers, 1992). In

light of this, researchers have focused increasingly on the knowledge structures

and perceptual skills necessary for a team to successfully perform a particular set

of behaviors. That is, if it is difficult to train a particular team skill (such as

adaptability) outright, it may be possible to effect the skill indirectly by first

identifying the knowledge structures and perceptual skills underlying the team

skill and, further, designing training to foster appropriate knowledge structures

and improve perceptual skills. For example, consider the team skill dimension

“adaptability”. “Adaptability” requires team members to predict and anticipate

actions of other team members and to adjust their own behavior accordingly

(Cannon-Bowers et al., 1993). A focus on the knowledge structures of the team

members could show what team members think and when they think it which,

in turn, may help us to understand how teammates predict, anticipate and adapt

effectively to each other’s needs. The focus on knowledge structures in teams

grew out of research into mental models (Cannon-Bowers et al., 1993; Orasanu,

1990; Noble, Grosz, & Boehm-Davis, 1987).

The mental model approach may be useful for understanding and training

teamwork skills (Cannon-Bowers & Salas, 1990). A number of researchers argue

that team members must hold an accurate understanding of various require-

ments of team performance if they are to perform effectively as team members

(Adelman, Zirk, Lehner, Moffett, & Hall, 1986). When team members possess

accurate and equally detailed conceptualizations of the requirements for team

functioning, they can be said to have shared mental models (Cannon-Bowers,

TEAM PERFORMANCE AND TRAINING RESEARCH 91

Salas, & Converse, 1990; Klein, 1989; Rouse, Cannon-Bowers, & Salas, 1993).

Unfortunately, the identification of methods to measure individual mental mod-

els, and the degree of overlap among mental models of team members, has

proven to be a highly difficult problem (Salas et al., 1992). Such an identification

system is needed to determine features of mental models and the impact of

features on the performance of task and teamwork behaviors.

A Comprehensive Model

The latest theoretical thought and empirical findings are reflected in updated

models of team performance. For example, Salas et al. (1992) developed an

integrative model based on previous team research and theory which incorpo-

_ rates the multiple forces affecting team performance. After an extensive review

of the earlier team performance models (i.e., Gersick, 1988; Gladstein, 1984;

Hackman, 1983; Morgan et al., 1986; Nieva, et al., 1978), Salas and his col-

leagues synthesized these models into a single framework. The framework con-

ceptualizes team performance as the outcome of dynamic processes reflected in

the coordination and communication patterns that teams develop over time. In

this conceptualization, the model takes a more comprehensive view of team

performance than have previous models. The Salas et al. (1992) framework is an

input, throughput, output model. It makes explicit the links between the organi-

zational and situational context, task characteristics, work structure, individual

characteristics, team characteristics, and team processes. Driving the model is

the contention that teams do not function in isolation. Rather, numerous fac-

tors both internal and external to the team, feed into each other, interact, and

affect a team’s performance. Tannenbaum, Beard, and Salas (1992) later refined

this model, and an adaptation of that version is presented in Figure 3. The

following section is based on the Tannenbaum et al. (1992) description.

Looking at Figure 3, it is apparent that many variables interact to influence

team activity. Influencing the entire input, throughput, output system is the

organizational and situational context. Work teams operate within a particular

environmental context (Hackman, 1983; Nieva et al., 1978), and any number of

factors in the organizational climate exist that could influence a team’s behav-

ior. For instance, group versus individual based reward systems, level of coopera-

tion or competition between members, and cooperation among departments

are all environmental factors that can have an indirect or direct influence on

team performance at any time.

Upon further examination of the model, we can see that fae characteristics

and individual characteristics both affect work structure. Task characteristics

include task complexity and task type. As mentioned earlier, previous work

92 SALAS, CANNON-BOWERS, AND BLICKENSDERFER

ORGANIZATIONAL & SITUATIONAL CHARACTERISTICS

Reward Systems Management Control Organizational Climate Intergroup Relations

Resource Scarcity Levels of Stress Competition Environmental Uncertainty

THROUGHPUT OUTPUT

TASK WORE STRUCTURE A Neh ay eS

CHARACTERISTICS

* Work Assignment ° New Norms

» Task Organization * Team Norms TEAM PROCESSES ° New Roles

° Task Type ° Communication * New Communication

° Structure Patterns

OSs eae > (SEE * New Processes

* Communication

° Conflict Resolution

Decision Meking

© Problem Solving TEAM

* Boundary Spanning

INDIVIDU AL ee PERFORMANCE

CHARACTERISTICS CHARACTERISTICS

* Task ESAs

© General Abilities * Power Distribution

° Motivation * Member Homogensity

° Err

Attitudes Team Resources i

* Persomality ° Climate - Team TEAM

° Me. * Cohesiv

ntel Models eaz INTERVENTIONS

INDIVIDU AL

CHANGES

© Imfivid uel T mining

* Team Tmining

° Team Building ° Task ESAs

e Attitudes

* Motivation

° Mental Models

Feedback

Fig. 3. Integrated model of team performance and training (adapted from Tannenbaum, Beard, & Salas, 1992)

contended that team interdependence influenced team task performance (Nieva

et al., 1978), and a meta-analysis of team research found that task difficulty

accounted for a significant amount of variance in team performance (Tannen-

baum, Dickinson, Salas, & Converse, 1990). Individual characteristics include

task knowledge, skills and abilities, in addition to general abilities, motivation,

attitudes, personality and mental models. As we have already indicated, it has

been shown repeatedly that individual abilities and proficiencies account for

some degree of team performance (Tannenbaum et al., 1990). Additionally,

personality may factor into team performance. For example, Driskell, Salas, &

Hogan (1987) argued that teams composed of certain personality traits will

perform best on certain tasks.

Both task characteristics and individual characteristics influence work struc-

ture, which includes work assignment, team norms, and communication struc-

ture. For example, who is allowed or required to speak with whom can influence

team performance (Naylor & Dickinson, 1969). In addition, teams build norms

regarding the way work is to be performed, and these norms can exert a large

influence on team processes and subsequent performance (Hackman, 1987).

Task and individual characteristics also affect team characteristics, which in-

clude power distribution, member homogeneity, team resources, team climate,

TEAM PERFORMANCE AND TRAINING RESEARCH 93

and cohesiveness. The makeup of a team, such as a team’s homogeneity, has

been shown to be related to team performance (Gunderson & Ryman, 1967). In

addition, team cohesiveness reflects a team’s sense of belongingness and sense of

teamness and has been recognized as a critical team characteristic (Tannen-

baum et al., 1990). Task characteristics, individual characteristics, work struc-

ture, and team characteristics make up the input stage of the model.

Although not explicitly stated by the Tannenbaum et al. (1992) model, we

contend that a number of other influences exist between task characteristics,

individual characteristics, team characteristics and work structure. First, we

argue that task characteristics can also influence individual characteristics. Fur-

ther, we argue that team characteristics can influence work structure, and that

individual characteristics can affect task characteristics. In addition, we also

- argue that team characteristics and individual characteristics mutually influ-

ence each other. In sum, we contend that the four input components (team

characteristics, individual characteristics, task characteristics, and work struc-

ture) all interact, and the result of those interactions, in turn, influences the

throughput and output stages. Future work should focus on these consider-

ations.

The result of the interaction between task, individual, and team characteris-

tics, and work structure produces a combined effect on the throughput stage of

the model affecting team processes. These are the intragroup and intergroup

actions that transform resources into a product (Gladstein, 1984). Team pro-

cesses include coordination, communication, conflict resolution, decision mak-

ing, problem solving, and boundary spanning. Considerable research has identi-

fied a number of processes that factor into team performance (see Tannenbaum

et al., 1992). In addition, team interventions such as individual training, team

training, and team building also affect team processes. Training interventions

can focus on the development of individual skills or on the development of team

skills (Sundstrom, Perkins, George, Futrell, & Hoffman, 1990). Team building

can include goal-setting, interpersonal training, and role negotiation.

The results of team processes feed into the output stage. The output stage

includes team changes (e. g., new norms, new rules, new communication pat-

terns, new processes), team performance (e. g., quality, quantity, time, errors,

costs), and individual changes (e. g., task KSAs, attitudes, motivation, and men-

tal models). While team performance is the primary output, Tannenbaum et al.

(1992) also consider changes in the team and changes in the individual team

members as outputs. For example, team changes may include greater or lessor

cohesiveness. Individual changes may include improved or decreased skills,

attitudes, or motivation. Finally, the team’s performance can serve as feedback

94 SALAS, CANNON-BOWERS, AND BLICKENSDERFER

and subsequently affects the input block of task, individual, team characteris-

tics, and work structure—the cycle begins once again.

Measurement

One challenge of team research has been to develop powerful, reliable, and

valid measurement techniques for measuring team performance. This has not

been an easy task. The distinction between individual and team tasks creates one

measurement challenge unique to team training. This is exacerbated by the

division between behavioral processes and performance outcomes. Another dif-

ficulty is that team performance evaluation thus far has been largely subjective

with the usual assessment technique being observation based appraisals of team

processes (Baker & Salas, 1992). Thus, the search for objective performance

measurement has been of particular importance.

A full discussion of measurement issues is beyond our scope here. Therefore,

we will focus briefly on the major issues relevant to team performance measure-

ment. The goal of team performance measurement must be to assess the process

that a team employs in task performance, as well as to assess the outcome of that

process. This is an important, but often overlooked distinction. While it is

important to measure the outcomes of team effort, it 1s equally essential to

understand the specific behavior that led to those outcomes. That is, measure-

ment must provide information that indicates why processes occurred as they

did and how those processes are linked to particular outcomes.

Also, team performance measures must consider performance at the individ-

ual level as well as at the subteam (i.e., teams within a team) and overall team

levels. It is important to make this distinction to emphasize the nature of the

individual skill versus team skill contribution to overall performance. For feed-

back purposes, it is essential to understand whether particular aspects of perfor-

mance can be attributed to individual or team behavior.

In addition to these requirements, a comprehensive system to measure perfor-

mance should serve several purposes if it is to be useful in training and managing

team performance (Cannon-Bowers, Salas, & Grossman, 1991). First of all,

team performance measures must be able to describe team performance accu-

rately; that is, measures must be sensitive enough to document the moment-to-

moment interactions and changes in performance. While this may seem rather

obvious, in many team situations the ability of the measurement system to

describe performance is complicated due to the dynamic, interactive tasks and

environments characteristic of team situations.

Team performance measures must also provide an evaluation of team perfor-

mance. Specifically, measures must distinguish between effective and ineffec-

tive processes, strategies, and teamwork behaviors. The implication of distin-

,

’

P

|

TEAM PERFORMANCE AND TRAINING RESEARCH 95

guishing between effective and ineffective is that a standard or index of perfor-

mance can be developed as a means to gauge the performance of a given team.

Also, team performance measures must capture the quality of teamwork skills

displayed by the team.

Finally to be most useful, team performance measures must be diagnostic. As

mentioned earlier, the measurement tools must provide information that indi-

cates why processes occurred as they did and how particular processes are linked

to certain outcomes. If these objectives can be achieved, it will be possible for

researchers to understand fully the nature of team performance, and practi-

tioners will be able to provide teams with feedback necessary to improve future

performance. Given these multiple purposes, it is likely that a team performance

measurement system will consist of several components. These measurement

components or “tools” can be developed and employed individually or in com-

bination in order to address specific research questions or practical problems

(e.g., providing feedback, assessing training effectiveness, and etc.).

In sum, it can be seen that an ideal team performance measurement system

must have the following characteristics: 1) provide measurement at both team

and individual levels; 2) assess the quality of team processes as well as outcomes;

3) focus attention on teamwork skills (as an adjunct to individual competen-

cies); 4) provide data that can be used to describe, evaluate, and diagnose team

performance (Cannon-Bowers, Salas, & Grossman, 1991). Given such stringent

requirements, it is clear that a team performance measurement system will

actually consist of several measurement tools, each of which assesses a subset of

the dimensions of team performance noted above.

One such effort addressed the subset dimensions of the team process issues.

Glickman, McIntyre, and their colleagues have worked at defining teamwork

components and developing teamwork measures (Dickinson et al., 1992; McIn-

tyre & Salas, in press; Morgan et al., 1986). The latest effort from this group

resulted in a conceptual framework for developing team process measures of

decision making performance (Dickinson et al., 1992). Briefly, Dickinson and

his colleagues first generated components of teamwork: team orientation, team

leadership, communication, monitoring, feedback, backup behavior, and coor-

dination. Based on these components, three techniques for teamwork measures

were presented: behavioral observation scale, behavioral summary scale, and

behavioral event format.

The behavioral observation scale is used to rate the frequency of teamwork

behaviors exhibited by a particular team and its members. The format consists

of seven scales. Each scale represents a particular component of teamwork and

incorporates 6-12 items which reflect that component. Each item is rated on a

5-point scale according to its frequency of occurrence (Almost Always to Almost

96 SALAS, CANNON-BOWERS, AND BLICKENSDERFER

Never). Teamwork behaviors between any two or more members are included.

Thus, the sheer number of teamwork behaviors displayed is rated. The behav-

ioral summary scale, on the other hand, can also be used to rate the degree of

teamwork displayed, but these scales do not contain multiple items. Instead, the

observer rates each component of teamwork only once. The observer rates the

team’s skill level on each respective component according to a 5 point scale

ranging from “Hardly Any Skill” to ““Complete Skill’.

The third measurement technique from Dickinson et al. (1992), the behav-

ioral event format, is used to observe and code team performance in structured

environments. Critical events must be first identified by subject matter experts.

Because of this, the behavioral event format is scenario specific such that for

each critical event in the scenario there are expected behaviors. The observer

then checks whether the expected and appropriate teamwork behaviors have

occurred. The observer also lists any additional behaviors that appear and indi-

cate the teamwork component of those behaviors. While at present we cannot

do more, we recommend that some future effort be directed at developing

measurement methods that could be used for both “‘real’’, on-going perfor-

mance as well as for training purposes.

Overall, developing team performance measures has been a difficult area, but

progress has been made. Numerous issues remain to be addressed, including: the

most appropriate level of analysis (i. e., the individual or team level), the quality

of outcomes and processes associated with team performance, and the search for

tools that describe, evaluate, and diagnose team performance.

Team Training Techniques

Perhaps most important to the on-going quest of understanding team perfor-

mance is the goal of designing effective team training strategies. Cannon-

Bowers, Salas, & Grossman (1991) argued that tactical decision making teams in

complex environments are faced with scenarios characterized by rapidly un-

folding events, multiple plausible hypotheses, high information ambiguity, se-

vere time pressure, sustained operations, and severe consequences for errors. In

addition to military teams, surgical, fire fighting, and even astronaut teams face

scenarios with similar characteristics. To perform effectively in the high stress

environment, team members must learn to coordinate their action so that they

can gather, process, integrate, and communicate information in a timely and

effective manner (Hall, Dwyer, Cannon-Bowers, Salas, & Volpe, 1993). There-

fore, team training interventions must maximize the use of instructional designs

that will allow teams to maintain performance under stressful conditions (Hall

et al., 1993).

‘TEAM PERFORMANCE AND TRAINING RESEARCH 97

With both individual skills and team skills being part of team performance, a

need for both individual and team training exists (Salas et al., 1992). However, a

number of authors have noted that team training usually emphasizes instruction

of individual skills within a team setting, regardless of the nature of the team task

(Briggs & Johnston, 1967; Converse, Dickinson, Tannenbaum, & Salas, 1988;

Meister, 1976; Salas et al., 1992). One reason behind the lack of teamwork skill

training may be that researchers have not yet given team training practitioners

proven tools with which teamwork skills can be trained.

However, recent research has taken a step in this direction with an increased

focus on specific team training interventions. Smith and Salas (1991), for exam-

ple, developed a training technique with which to train one behavioral team-

work skill, assertiveness. Smith and Salas (1991) found that subjects who re-

ceived assertiveness training including lecture with modeling and role play

practice exhibited significantly more assertive behavior in the team task than did

subjects who received training including only lecture or lecture with modeling.

Most importantly, subjects who received assertiveness training that included

role play were the only subjects who differed significantly from the no-training

control group. Thus, role play appears necessary in training the team skill asser-

tiveness. The Smith and Salas training method appears ready for use by practi-

tioners.

Employing cross-training using positional rotation as a team training tech-

nique has also received a growing amount of attention. Positional rotation can

be conceptualized as a type of job rotation among team members. This method

of cross-training provides team members with an understanding of the basic

knowledge necessary to successfully perform the tasks, duties, and/or positions

of the other team members. Additionally, cross-training gives team members an

overall framework of the team task and how each particular individual’s task is

important to the team task (Travillian, Volpe, Cannon-Bowers, & Salas, 1993).

Cross-trained teams have achieved team process ratings and team outcome

scores higher than those teams without such training (Travillian et al., 1993). In

sum, cross-training appears to be a viable instructional strategy for teams.

Other recent research has focused on summarizing and extracting team train-

ing guidelines from the empirical research and ensuring that those guidelines are

accessible for use by both researchers and practitioners. For instance, Swezey

and Salas (1992) presented a comprehensive listing of team training guidelines,

while Burgess, Salas, and Cannon-Bowers (1993) presented a listing of tech-

niques for effective team leadership. The growing accessibility of such guidelines

should help to increase the training of teamwork skills.

Research emphasis on specific team training techniques, in addition to results

summarized into straight-forward guidelines, should help practitioners begin to

98 SALAS, CANNON-BOWERS, AND BLICKENSDERFER

integrate teamwork skill training into their own training programs. Cross-train-

ing and modeling/role-playing are two recently emphasized team training tech-

niques that should help practitioners to begin integrating teamwork skill train-

ing into team training programs.

Other Team Training Issues

A recurrent issue in team training design concerns the sequence of team

training. That is, trainers must consider whether mastery of individual skills

should occur first followed by mastery of team skills or vice-versa. Research

generally advocates that team training is most effective and efficient when indi-

vidual skill mastery is completed before team training (Biggs & Johnston, 1967;

Daniels, Alder, Kanarick, Gray & Feuge, 1972; Denson, 1981; Johnston, 1966;

Klaus & Glaser, 1970). The rationale for training individual skills first is that if

individual skills are not developed fully, the team may not be able to perform the

task successfully no matter how effective their team skills may be. In addition, if

the taskwork skills are not developed prior to the introduction of the teamwork

skills, training for teamwork skills may interfere with team members’ learning of

taskwork skills. 3

An issue in team training that has not yet been resolved is performance feed-

back for team tasks. Although team research has acknowledged the importance

of feedback (Dyer, 1984), many questions regarding feedback in team training

exist. Feedback in a team environment should 1) enable each team member to

perform his/her individual task, 2) demonstrate the contribution of an individ-

ual’s performance to the performance of other members and, 3) demonstrate the

contribution of an individual’s performance to the performance of the team as a

whole. Because of the multiple levels that team feedback can address, giving

team feedback is not entirely straightforward. For instance, feedback enhances

performance on the aspect of performance about which feedback 1s provided. In

team tasks, this translates to feedback focusing on the team skills or on the task

skills, and the trainer must decide where to focus the feedback (Alexander &

Cooperband, 1965; Chapman, Kennedy, Newell, & Biel, 1959). In addition,

problems arise when giving overall team outcome feedback. For example, if

high and low performing individuals receive the same positive team feedback,

the low performers will likely not realize that their performance needs improve-

ment (Nadler, 1979). An unintended consequence of giving team level feedback

without respect to the relationship of individual performance to the team perfor-

mance is that incorrect behaviors may be reinforced. This, in turn, may result in

no improvement and may well wash out the impact of team feedback on the

performance of both the individual and team overall.

One issue inherent in team feedback is sequence—when, what type, and how

‘TEAM PERFORMANCE AND TRAINING RESEARCH 99

much feedback should be given (Salas et al., 1992). Briggs and Johnston (1967)

found that providing feedback on one aspect of the task during early training

sessions and increasing this feedback to refer to several aspects as training pro-

gresses helped teams focus on different aspects of the task. Klaus and Glaser

(1970) also emphasized timing of feedback. They suggested that individual per-

formance feedback should be given during early training and that overall team

performance feedback should be given during later phases of training. While it is

easy to understand that feedback will likely need to change focus at different

points in team training, the most effective uses of feedback in team training tasks

is still not well understood and numerous issues have yet to be addressed.

Clearly, effective team training is an important goal of team performance

research. Researchers agree that individual skills should be trained before team

skills. However, questions still exist on the most effective team training tech-

niques, and how to use feedback as a team training tool.

Principles of Team Performance and Training

We define “principle” as an underlying truth about a human phenomenon.

In team performance and training research, it is important not only to elicit

principles but also to communicate these principles to both the research commu-

nity and practitioners. After the review of team training and performance, a

number of principles can be derived.

Principles of Team Performance

Principle 1: Teamwork skills are distinct from taskwork (individual) skills (Mor-

gan et al., 1986).

As McIntyre and Salas (in press) summarized, two “tracks” of skill exist in

team tasks: the “‘taskwork”’ track involves those skills necessary for individuals

to perform their own task or function. On the other hand, “teamwork” skills are

those necessary to be an effective team member.

Principle 2: Teamwork consists of a series of related behaviors (McIntyre &

Salas, in press).

After separating taskwork from teamwork, teamwork then divides into re-

lated behaviors. McIntyre and Salas (in press) argued that these include, for

example, monitor own performance, perform self-correction of errors, provide

task and motivational reinforcement, adapt to unpredictable situations, predict

each others behavior, and use closed-loop communication (see also Oser et al.,

1989). Effective teams exhibit these behaviors more frequently than do less

effective teams.

100 SALAS, CANNON-BOWERS, AND BLICKENSDERFER

Principle 3: Teams evolve (mature) over time (Glickman et al., 1987).

As team members learn about each other and the task, team members pro-

gress from working as individuals into working as a team (Glickman et al.,

1987). Effective feedback parallels this evolution. That is, at certain times in

team training, individually focused feedback may be the most useful, while at

other points in the training process general team feedback may be the most

beneficial.

Principle 4: “Mature” teams have members who anticipate each others’ needs

(Glickman et al., 1987).

When team members become familiar with each other’s knowledge, skills,

abilities, attitudes, motivation, preferences, style, etc., they are able to better

anticipate the task, informational, and interpersonal needs of teammates. Ac-

cording to Cannon-Bowers et al. (1993), this is the basis of shared team mental

models. 3

Principle 5: Effective teams have a strong sense of “teamness”’ (Glickman et al.,

1987).

Experienced teams realize when their team performs an operation which

utilizes teamwork behaviors, and experienced teams can identify operations

which will require teamwork behaviors.

Principle 6: “Mature” teams do not need to rely on overt communicate as much

to perform effectively (Orasanu, 1990; Rouse et al., 1993).

Teams with shared mental models don’t need to communicate as much under

high workload as do teams with less developed shared knowledge. Communica-

tion will decrease when teams have a higher amount of shared information; they

have shared expectations and intentions and can anticipate each other’s behav-

ior. This has been referred to as “‘implicit”’ coordination (see Kleinman & Ser-

faty, 1989). |

Principle 7: Effective teams can adjust their strategy under stress (Kleinman &

Serfaty, 1989).

According to Kleinman & Serfaty (1989), effective teams employ “implicit”

coordination strategies under high workload conditions. This is an example of

how teams might change their behavior in response to task demands. Other such

strategies include: load balancing, performance monitoring, feedback (see Can-

non-Bowers et al., in press).

Principle 8: Some teamwork skills are generic (Cannon-Bowers et al., in press).

Although tasks may require certain task specific skills, some basic teamwork

skills appear to be a part of effective teamwork across situations. Therefore, it

may be possible to improve team performance by training team members in

“generic” teamwork skills. However, more specific (task-related) team training

is probably necessary in many cases (Cannon-Bowers et al., in press).

TEAM PERFORMANCE AND TRAINING RESEARCH 101

Principle 9: Effective teams optimize resources (Cannon-Bowers et al., in press).

Teams working effectively learn to self-correct. At the same time, team

members can also compensate for each other. That is, if one member does not

perform well, the team will learn to work around the problem. The emphasis for

effective teams is on teamwork process—working well as a team.

Principle 10: 4 number of external and internal factors influence team perfor-

mance (Salas et al., 1992).

Teams do not operate in isolation. The organizational and situational con-

text, task characteristics, work structure, individual characteristics, team char-

acteristics, and team processes all influence team effectiveness (Tannenbaum et

al., 1992). It is the interaction among the external and internal factors that make

each team unique.

These principles are a necessary step in understanding team training. How-

ever, once principles are delineated, we must operationalize those principles

into specific training techniques.

Principles of Team Training

Principle 1: Individual proficiency must precede team proficiency (Stout, Salas,

& Carson, 1994).

If team members do not learn the task skills, team performance will remain

low. Training time must be allowed for members to master their individual task

skills.

Principle 2: Team training must diagnose and remediate team performance

(Cannon-Bowers et al., 1991).

In order to effect a behavioral change, a necessary first step is to identify the

deficiencies in team members’ skills, knowledge, and attitudes. To accomplish

this, measurement and diagnostic mechanisms must exist that allow team in-

structors to identify and attribute team performance problems. Once having

done this, training interventions can be used to ameliorate the particular perfor-

mance problems.

Principle 3: Team training must allow for: information exchange, demonstration

of teamwork behaviors, practice, and feedback (Prince et al., 1992).

As with all types of training, team training must follow the basic requirements

of sound training design (Salas, Burgess, & Cannon-Bowers, in press). For team

training, these phases of team training should be based on a careful analysis of

the teamwork demands of the task.

Principle 4: Team training must emphasize the nature of interdependency (Swe-

zey & Salas, 1992).

Without interaction, team skills will likely remain undeveloped. Scenarios

structured around mutual dependency promote the use of team skills. More-

102 SALAS, CANNON-BOWERS, AND BLICKENSDERFER

over, feedback regarding teamwork skills is impossible if the task does not de-

mand sufficient interdependency among team members.

Principle 5: Team training must emphasize teamwork skills.

This relates directly to the previously mentioned teamwork/taskwork divi-

sion (Morgan et al., 1986). Team training that emphasizes only individual skills

will not ultimately improve performance. Team skills are part of team perfor-

mance and must be addressed in training.

Principle 6: Team training must create systematic opportunities to practice (sce-

nario development) (Prince et al., 1992).

Prince, Oser, Salas, and Woodruff (1993) summarized that the findings of

multiple researchers which suggested that team training requires the develop-

ment of structured scenarios which provide the opportunity to perform and

receive feedback about-critical team actions. Further, Hall et al. (1993) pre-

sented a strategy for developing team scenarios with varying degrees of stress.

Future Research

While team research has made progress in the area of team performance and

team training over the past few decades, much remains to be done. In particular,

the following research questions need to be addressed:

What is the nature and content of shared mental models or knowledge struc-

tures?

As mentioned earlier, in efforts to identify the subtleties behind team pro-

cesses, recent work has turned to the mental model construct. However, at this

time, no concrete evidence for shared mental models has appeared. Researchers

must now focus on providing evidence of shared mental models as well as

techniques designed to measure the degree of similarity of the knowledge struc-

tures in team members’ minds.

Can reliable, systematic, usable team performance and diagnostic measures

be designed and developed?

Until team researchers have satisfactory measurement tools, team perfor-

mance and training research will never reach its full potential. Once developed,

such a “‘tool-box’’ could be used by researchers and practitioners alike and

would add considerable authority to team research and training.

What is the impact of selected individual and organizational factors on team

performance?

As presented earlier, team performance models acknowledge the significant

impact that individual and organizational factors have on team performance.

However, the models do not explain specifically which influences will affect

TEAM PERFORMANCE AND TRAINING RESEARCH 103

which aspects of team performance. Further, the models do not explain benefi-

cial versus detrimental influences on team performance.

Summary and Conclusion

Since the 1960s, the area of team performance and training research has

evolved dramatically. Early studies focused on the importance of individual

abilities and skills to the overall team task. These studies were followed by efforts

focused on identifying a difference between task skills and team skills (e.g., the

taskwork versus teamwork and evolution of teams concepts described previ-

ously) and the focus on trainable skills (e.g., the team skill dimensions). Only

recently have theorists emphasized the realm of both individual and organiza-

tional factors that influence team performance. Current research reflects this

emphasis, and there are an increasing number of efforts aimed at explicating the

relationship between team performance and a variety of concepts (e.g., organiza-

tional climate, leadership, and efficacy issues).

Another potentially fruitful research area which has only recently come into

focus concerns cognitive aspects of team performance such as shared knowledge

structures among team members and complex team skills such as team decision

making. In addition to the need for further theoretical underpinnings this area

also demands investigation of more refined measurement techniques. If team

research is to benefit from the mental model construct, valid and reliable mea-

sures for tapping these structures are necessary.

At present, the literature is beginning to provide useful guidelines for manag-

ing and training teams. Specifically, a solid base of empirically based research is

available. In addition, team researchers have begun to present their findings in

the form of explicit recommendations and guidelines of team training tech-

niques useful to practitioners. This strongly suggests that the field has matured

significantly over the last 20 years.

Finally, the team performance and training literature appears to be coming

into focus on the ultimate mission: improving team performance. That is, re-

searchers and practitioners alike have begun integrating laboratory results with

real world environments. Clearly, team training research is alive and well. A

solid body of empirical research both supports this assertion and lays out the

questions which we still need to address.

Author Notes

The views expressed herein are those of the authors and do not reflect the organization to which they are

affiliated. We would like to thank Tom Eggemeier, John Burns, and an anonymous reviewer for their helpful

104 SALAS, CANNON-BOWERS, AND BLICKENSDERFER

comments on this paper. Portions of this paper were presented at the A.P.A. Division 21 mid-year meeting,

Washington, D.C., 1993.

References

Adelman, L., Zirk, D. A., Lehner, P. E., Moffett, R. J., & Hall, R. (1986). Distributed tactical decision

making: Conceptual framework and empirical results. ZEEE Transactions on Systems, Man, and Cyber-

netics, SMC-16, 794-805.

Alexander, L. T., & Cooperband, A. S. (1965). System Training and Research in Team Behavior. (TM-2581).

Santa Monica, CA: System Development Corporation. (DTIC No. AD620606).

Baker, D. P., & Salas, E. (1992). Principles for measuring teamwork skills. Human Factors, 34, 469-475.

Bass, B. M., & Barrett, B. V. (1981). People, work, and, organizations. Boston: Allyn and Bacon.

Bouchard, T. J. Jr. (1969). Personality, problem-solving procedure and performance in small groups. Journal

of Applied Psychology Monographs, 53, 1-29.

Briggs, G. E., & Johnston, W. A. (1967). Team Training. (NAVTRADEVCEN-1 327-4 AD-660 019). Techni-

cal Report No. 1327-4). Orlando, FL: Naval Training Device Center.

Burgess, K. A., Salas, E., & Cannon-Bowers, J. A. (1993). Training team leaders: “More than meets the eye”.

Paper presented at the Eight Annual Conference of the Society for Industrial and Organizational Psychol-

ogy, San Francisco, CA.

Cannon-Bowers, J. A. & Salas, E. (1990, April). Cognitive psychology and team training : Shared mental

models in complex systems. Symposium presented at the Fifth Annual Conference of the Society for Indus-

trial Organizational Psychology, Miami, FL.

Cannon-Bowers, J. A., Salas, E., & Converse, S. A. (1990). Cognitive psychology and team training: Training

shared mental models and complex systems. Human Factors Society Bulletin, 33, 1-4.

Cannon-Bowers, J. A., Salas, E., & Converse, S. A. (1993). Shared mental models in expert team decision

making. In J. Castellan, Jr. (Ed.), Current issues in individual and decision making (221-246). Hillsdale, NJ:

Erlbaum.

Cannon-Bowers, J. A., Salas, E., & Grossman, J. D. (1991, June). Improving tactical decision making under

stress: Research directions and applied implications. Paper presented at the International Applied Military

Psychology Symposium, Stockholm, Sweden.

Cannon-Bowers, J. A., Tannenbaum, S. I., Salas, E., & Volpe, C. E. (in press). Defining team competencies:

Implications for training requirements and strategies. In R. Guzzo and E. Salas (Eds.) Team effectiveness

and decision making in organizations. San Francisco: Jossey Bass.

Chapman, R. L., Kennedy, J. L., Newell, A., & Biel, W. C. (1959). The system research laboratory’s air defense

experiments. Management Service, 5, 250-269.

Converse, S. A., Dickinson, T. L., Tannenbaum, S. I., & Salas, E. (1988, April). Team training and perfor-

mance: A meta-analysis. Paper presented at the annual meeting of the Southeastern Psychology Association,

New Orleans, LA.

Daniels, R. W., Alder, D. G., Kanarick, A. F., Gray, T. H., & Feuge, R. L. (1972). Automated operation

instruction in team tactics (NAVTRADEVCEN-70-c-03 10-1, AD-736 970). Orlando, FL: Naval Training

Device Center.

Denson, R. W. (1981). Team training: Literature review and annotated bibliography (AFHRL-TR-80-40, A

099994). Wright-Patterson AFB, OH: Logistic and Technical Training Division, Air Force Human Re-

sources Laboratory.

Dickinson, T. L., McIntyre, R. M., Ruggeberg, B. J., Yanushefski, A. M., Hamill, L. S., Vick, A. L. (1992). A

conceptual framework for developing team process measures of decision-making performance. (unpub-

lished Tech Rep.). Orlando, FL: Naval Training Systems Center.

Driskell, J. E., Salas, E., & Hogan, R. (1987). A Taxonomy for Composing Effective Naval Teams (Tech. Rep.

No. TR87002). Orlando, FL : Naval Training Systems Center.

Dyer, J. L. (1984). Team research and team training: A state of the art review. Human Factors Review,

285-323.

George, C. E., Hoak, G., & Boutwell, J. (1963). Pilot studies of team effectiveness (Research Memorandum

No. 28, AD-627, 217). Ft. Benning, GA: U.S. Army Infantry Human Research Unit, Human Resources

Research Office.

Gersick, C. J. G. (1985). Time and transition in work teams: Towards a new model of group development.

Unpublished manuscript, University of California, Los Angeles.

Gersick, C. J. G. (1988). Time and transition in work teams: Towards a new model of group development.

Academy of Management Review, 31, 9-14.

Gladstein, D. (1984). Groups in context: A model of task group effectiveness. Administrative Science Quar-

terly, 29, 499-517.

Glickman, A. S., Zimmer, S., Montero, R. C., Guerette, P. J., Campbell, W. J., Morgan, Jr. B. B., & Salas, E.

TEAM PERFORMANCE AND TRAINING RESEARCH 105

(1987). The evolution of teamwork skills: An empirical assessment with implications for training. Technical

Report NTSC 87-016. Arlington, VA: Office of Naval Research.

Guerette, P. J., Miller, D. L., Glickman, A. S., Morgan Jr, B. B. & Salas, E. (1987). Instructional Processes

and Strategies in Team Training. (Tech. Rep. No. NTSC 87-017). Orlando, FL: Naval Training Systems

Center, Human Factors Division.

Gunderson, E. K. & Ryman, D. (1967). Group homogeneity compatibility, and accomplishment. (Report No.

NNNRU-67-16). San Diego, CA: Navy Medical Neuropsychiatric Research Unit.

Hackman, J. R. (1983). A normative model of work team effectiveness (Technical Report No. 2). New Haven,

CT: Yale University.

Hackman, J. R. (1987). The design of work teams. In J. Lorsch (Ed.), Handbook of organizational behavior

(pp. 315-342). New York: Prentice-Hall.

Hall, E. R., & Rizzo, W. A. (1975). An assessment of U.S. tactical team training (TAEG-18, AD-AO11 452).

Orlando, FL: Naval Training Equipment Center.

Hall, J. K., Dwyer, D. J., Cannon-Bowers, J. A., Salas, E., & Volpe, C. E. (1993). Toward assessing team

tactical decision making under stress: The development of a methodology for structuring team training

scenarios. Proceedings of the 15th Annual Interservice/Industry Training Systems Conference (pp. 87-98).

Washington, DC: National Security Industrial Association.

Helmreich, R. L., & Foushee, H. C. (1993). Why crew resource management? Empirical and theoretical bases

of human factors training in aviation. In E. L. Weiner, R. L. Helmreich, and B. B. Kanki (Eds.), Cockpit

resource management (pp. 3-45). San Diego, CA: Academic Press.

Johnston, W. A., and Briggs, G, E. (1968). Team performances a function of team arrangement and work

load. Journal of Applied Psychology, 52 (2), 89-94.

Johnston, W. A. (1966). Transfer of team skills as a function of type of training. Journal of Applied Psychology,

50, 102-108.

Kabanoff, B., & O’Brien, G. E. (1979). Cooperation structure and the relationship of leader and member

ability to group performance. Journal of Applied Psychology, 64, 526-532.

Klaus, D. J., & Glaser, R. (1965). Increasing team proficiency through training.5.Team learning as a function

of member learning characteristics and practice conditions (AIR-E-1-4/65-TR, AD-215 621). Pittsburg, PA:

American Institute for Research.

Klaus, D. J., & Glaser, R. (1970). Reinforcement determinants of team proficiency. Organizational Behavior

and Human Performance, 5, 33-67.

Klein, G. A. (1989). Recognition-primed decisions. In W. B. Rouse (Ed.), Advances in man-machine systems

research (Vol. 5, pp 47-92). Greenwich, CT: JAI Press.

Kleinman, D. L. & Serfaty, D. (1989). Team performance assessment decision making. Proceedings of the

Symposium on Interactive Networked Simulation for Training (pp.22—27). Orlando, FL.

Levine, J. M. & Moreland, R. L. (1990). Progress in small group research. Annual Review Psychology, 41,

585-634.

McCallum, G. A., Oser, R., Morgan, B. B., Jr., & Salas, E. (1989, August). An investigation of the behavioral

components of teamwork. Paper presented at the Annual Meeting of the American Psychological Associa-

tion, New Orleans, LA.

McIntyre, R. M., & Salas, E. (in press). Team performance in complex environments: What we have learned,

so far. To Appear in R. Guzzo and E. Salas (Eds.), Team effectiveness and decision making in organizations.

Jossey Bass: San Francisco.

McIntyre, R. M., Salas, E., Morgan, B., & Glickman, A. (1989). Team research in the 80’s: Lessons learned.

Technical Report. Orlando, FL: Naval Training Systems Center.

Meister, D. (1976). Team functions. In D. Meister (Ed.), Behavioral foundations of system development. New

York: Wiley and Sons.

Morgan, B. B. Jr., Glickman, A. S., Woodward, E. A., Blaiwes, A. S., & Salas, E. (1986). Measurement of

team behaviors in a Navy environment. (Technical Report No. NTSC TR-86-014). Orlando, FL: Naval

Training Systems Center.

Nadler, D. A. (1979). The effects of feedback on task group behavior: A review of the experimental research.

Organizational Behavior and Human Performance, 23, 309-338.

Naylor, J. C. & Dickinson, T. L. (1969). Task Structure, work structure, and team performance. Journal of

Applied Psychology, 53, 167-177.

Nieva, V. F., Fleishman, E. A., & Reick, A. (1978). Team dimensions: Their identity, their measurement and

their relationships. Final Technical Report, Contract No. DAH19-78-C-0001, Washington, D. C.: Ad-

vanced Research Resources Organization.

Noble, D., Grosz, C., & Boehm-Davis, D. (1987). Rules, schema, and decision making. (TR R-125-87).

Vienna VA: Engineering Research Association.

Orasanu, J. (1990, July). Shared mental models and crew decision making. Paper presented at the 12th

Annual Conference of the Cognitive Science Society, Cambridge, MA.

106 SALAS, CANNON-BOWERS, AND BLICKENSDERFER

Orasanu, J., & Salas, E. (1993). Team decision making in complex environments. In G. Klein, J. Orasanu,

and R. Calderwood (Eds.). Decision making in action: Models and method. 327-345. Norwood, NJ: Ablex

Publishing Corp. In press.

Oser, R. L., McCallum, G. A., Salas, E., & Morgan, B. B., Jr. (1989). Toward a definition of teamwork: An

analysis of critical team behavior (NTSC Technical Report No. 89-004). Orlando, FL: Naval Training

Systems Center.

Oser, R. L., Prince, C., & Morgan, B. B., Jr. (1990, October). Differences in aircrew communication content as

a function of flight requirement: Implication for operational aircrew training. Paper presented at the 34th

Annual Meeting of the Human Factors Society, Orlando, FL.

Prince, A., Brannick, M. T., Prince, C., & Salas, E. (1992). Team Process Measurement and Implications for

Training. Proceedings of the Human Factors Society 36th Annual Meeting (pp. 1351-1355). Santa Monica,

CA: Human Factors Society.

Prince, C., Chidester, T. R., Cannon-Bowers, J. A., & Bowers, C.A. (1992). Aircrew coordination: Achieving

teamwork in the cockpit. In R. W. Swezey, and E. Salas (Eds.), Teams : Their training and performance,

(329-353). Norwood, NJ: Ablex.

Prince, C., Oser, R., Salas., & Woodruff, W. (1993). Increasing hits and reducing misses in CRM/LOS

scenarios: Guidelines for simulator scenario development. The International Journal of Aviation Psychol-

ogy, 3, 69-82.

Prince, C. & Salas, E. (1993). Training and research for teamwork in the military aircrew. In E. L. Wiener,

B. G. Kanki, & R. L. Helmreich (Eds.), Cockpit resource management (pp. 337-366). Orlando, FL: Aca-

demic Press.

Rouse, W. B., Cannon-Bowers, J. A., & Salas, E. (1993). The role of mental models in team performance in

complex systems. JEEE Transactions on Systems, Man, and Cybernetics, 22 (6), 1296-1308.

Salas, E., Burgess, K., A., & Cannon-Bowers, J. A. (in press). Training effectiveness research: The tools of the

trade. In J. Weiner, (Ed.) Research techniques in human engineering. Englewood, NJ: Princeton & Hall.

Salas, E., Dickinson, T.-L., Converse, S. A., & Tannenbaum, S. I. (1992). Toward an understanding of team

performance and training. In R. W. Swezey and E. Salas (Eds.), Teams: Their training and performance (pp.

3-29). Norwood, NJ: Ablex.

Smith, K. A., & Salas, E. (March, 1991). Training assertiveness: The importance of active participation. Paper

presented at the 37th annual meeting of Southeastern Psychological Association, New Orleans, LA

Steiner, I. D. (1972). Group process and productivity. Orlando: Academic Press.

Stout, R. J., Cannon-Bowers, J. A., Salas, E., & Morgan, B. B., Jr. (1990). Does crew coordination behavior

impact performance?. Paper presented at the 1990 Annual Meeting of the Human Factors Society, Orlando,

je

Stout, R. J., Prince, C., Baker, D. P., Bergondy, M. L., Salas, E. (1992). Aircrew coordination: What does it

take? Proceedings of the Thirteenth Biennial Psychology in the Department of Defense Symposium, 133-

137!

Stout, R. J., Salas, E., & Carson, R. (1994). Individual task proficiency and team process behavior: What’s

important for team functioning. Military Psychology, 6, 177-192.

Streufert, S., & Nogami, G. (1992). Cognitive complexity and team decision making. In R. W. Swezey & E.

Salas (Eds.), Teams: Their training and performance (127-152). Norwood, NJ: Ablex.

Sundstrom, E., Perkins, M., George, J., Futrell, D. & Hoffman, D. A. (1990). Work-team context, develop-

ment, and effectiveness in a manufacturing organization. Presented at the Fifth Annual Conference of

Society for Industrial and Organizational Psychology, April, Miami.

Swezey, R. W., & Salas, E. (1992). Guidelines for use in team-training development. In R. W. Swezey and E.

Salas (Eds.), Teams: Their training and performance (219-245). Norwood, NJ: Ablex.

Tannenbaum, S. I., Beard, R. L., & Salas, E. (1992). Team building and its influence on team developments.

In K. Kelley (Ed.) Issues, Theory, and Research in Psychology (pp. 117-153). Amsterdam: Elsevier.

Tannenbaum, S. I., Dickinson, T. L., Salas, E., & Converse, S. A. (1990). .A meta-analysis of team performance

and team training. Unpublished manuscript, State University of New York at Albany.

Terborg, J. R., Castore, C., & DeNinno, J. A. (1976, May). A longitudinal field investigation of the impact of

group composition on group performance and cohesion. Journal of Personality and Social Psychology, 34

(5), 782-790.

Travillian, K. K., Volpe, C. E., Cannon-Bowers, J. A., Salas, E. (1993). Cross-training highly interdependent

teams: Effects on team process and team performance. Proceedings of the 37th Annual Human Factors and

Ergonomics Society Conference (pp. 1243-1247), Santa Monica, CA: Human Factors Society.

Tuckman, B. W. (1965). Developmental sequences in small groups. Psychological Bulletin, 63, 384-399.

Tziner, A., & Eden, D. (1985). Effects of crew composition on crew performance: Does the whole equal the

sum of the parts? Journal of Applied Psychology, 70, 85-93.

Wagner, H., Hibbits, N., Rosenblatt, R., & Schulz, R. (1976). Team training and evaluation strategies: A state

of the art review. (SR-ED—76-11). Alexandria, VA: Human Resources Research Organization.

Journal of the Washington Academy of Sciences,

Volume 83, Number 2, Pages 107-123, June 1993

Effects of Different Data Base Formats

on Information Modification

Deborah A. Boehm-Davis,' Robert W. Holt, and Robert D. Peters

George Mason University, Psychology Department, Fairfax, Virginia

ABSTRACT

This research examined the effects of three different data base formats on the ability of

users to modify information in the data base. Graphical, tabular, and verbal forms of a

thesaurus data base were constructed, along with questions that required users to modify the

data base. Three question formats, each compatible with one of the forms of the data bases,

were designed — graphical, tabular, and verbal. The data indicate that users are faster and

more accurate in modifying the data base when the format of the information in the data

base matches the format of the information in the modification instructions. While the

importance of matching data base format to likely modifications may seem obvious, it

would appear that the designers of most current data base systems have not taken this into

account.

Since the advent of the information age, increasing amounts of information

have come to be stored in data bases. However, research into the best ways of

storing and presenting the information in the data base has not followed at the

same pace. Although research has been conducted on this issue, there is no

consensus on the effects of specific information presentation formats.

In the domains of problem solving and decision-making, most of the research

has focused on comparing the speed and accuracy of performance using tabular

versus graphic formats (e.g. see Lalomia & Coovert, 1987; and Powers, Lashley,

Sanchez, & Shneiderman, 1984). Investigators, however, have failed in their

attempts to clearly demonstrate the general superiority of one format over an-

other. For example, after reviewing 29 studies, DeSanctis (1984) reported that

12 studies found tables superior to graphs, 7 found graphs superior to tables, and

10 found no difference between the two formats. She concluded that the best

' Requests for reprints should be sent to Deborah A. Boehm-Davis, George Mason University, Psychology

Department, Fairfax, VA 22030-4444.

107

108 BOEHM-DAVIS, HOLT, AND PETERS

method of displaying information may be a function of the task to be per-

formed.

Unfortunately, knowing that the task is an important constraint on the form

of information display does not provide, a priori, a method for deciding what

form information presentation should take in a particular instance. That is,

none of the studies discussed so far addresses the more theoretical question of

why one might expect a particular form of information presentation to be better

than another in a given situation. Recently, Vessey and Galletta (1991) have

proposed the notion of “‘cognitive fit’, which they describe as “resulting from

matching the characteristics of the problem representation to those of the task’’.

In this paper, Vessey and Galletta propose a problem solving model that takes as

input the problem representation, the problem solving task, and the problem

solving skill of the user. Their argument is that to the extent that the representa-

tion of the problem (in the interface) matches the characteristics of the problem

solving task, performance will be improved. Although their recent research

(Sinha & Vessey, 1991; Vessey, 1991; and Vessey & Galletta, 1991) has only

partially supported the model, it provides a comprehensive view in which pre-

vious research can be accomodated. It also suggests the importance of matching

the format of input information to the format in which the information must be

interpreted. This general notion has been supported by a number of previous

studies in the literature (Bennett, 1987; Boehm-Davis, Holt, Koll, Yastrop, &

Peters, 1989; Durding, Becker, & Gould, 1977; Peters, Yastrop, & Boehm-Da-

vis, 1987; Wickens & Scott, 1983; and Wright & Fox, 1972).

The studies by Durding, Becker, and Gould (1977) and by Bennett (1987)

provide specific evidence for the importance of stimulus-response compatibil-

ity. Durding, Becker, and Gould examined how people organize information.

They presented people with sets of data organized into schemes that were either

consistent or inconsistent with their natural structure. People were generally

faster and more accurate at remembering and organizing information when

they were presented with a structure in which to put their responses that

matched the original organization of the words than when they were presented

with a less compatible structure. These researchers argue that the conceptual

structure of a data base should conform to the semantic relationships among the

data elements.

This argument is supported by Bennett (1987) who examined transfer-of-

training performance of users working with a perceptual data base system. His

results showed that training with a display that is consistent with the target

display facilitated performance and that training with an inconsistent display

inhibited performance.

The research by Wickens and Scott (1983), Wright and Fox (1972), Peters,

DATA BASE FORMATS AND INFORMATIONAL RETRIVAL 109

Yastrop and Boehm-Davis (1987) and Boehm-Davis, Holt, Koll, Yastrop and

Peters (1989) speak more directly to the importance of matching the format of

information presentation to the characteristics of the task to be performed.

Wickens and Scott examined the performance of students on a complex deci-

sion task requiring the integration of separate pieces of information. Their re-

sults showed that performance was best when the displays showed the data in an

integral form which could be used directly as the basis for judgment. Wright and

Fox, using different forms of information display in tables, also found that

performance was best when the table contained integral information that could

be used directly in the decision-making task.

Boehm-Davis, Holt, Koll, Yastrop, and Peters (1989), using both database

format (spatial, tabular, or verbal) and question type (spatial, tabular, or verbal)

as independent variables, found that users were faster and more accurate at

information retrieval when the format of the information in the database

matched the type of information needed to answer the question. In addition, an

item analysis of the retrieval tasks revealed an interaction between type of re-

trieval task and data base format. For example, searching for the number of

occurrences of a specific item was facilitated by using a tabular database, while

searching for the shortest route between two cities was facilitated by using a

graphic database. This finding was replicated using a modified version of the

original data base (Peters, Yastrop, & Boehm-Davis, 1987).

The present study was designed to extend the previous work in this area. Our

last study (Boehm-Davis, Holt, Koll, Yastrop, & Peters, 1989) focused on

searching through a data base to locate information. However, most data bases

are dynamic, not static and users enter the new information into the data base.

We therefore felt it important to examine the effect of display format on the

ability of users to modify information in the data base. This study addresses that

issue by asking users to modify information in a data base of a given format

when the information to be modified is presented in a format that is either

consistent or inconsistent with the format of the data base itself.

Method

Design

The experimental design used in this experiment was a 3 X (3 X 2 X 3 X 2)

design. The between-subjects factor was the format of the data base the subject

used (graphic, tabular, or verbal). The within-subjects factors were format of the

question (graphic, tabular, or verbal), item modified (word or relationship), type

110 BOEHM-DAVIS, HOLT, AND PETERS

of modification (add, modify, or delete), and time (whether it was the first or

second time they made a particular type of modification).

Participants

The participants in this study were 36 students at George Mason University.

The students received either payment or course credit for their participation in

the study. Unfortunately, after data collection was complete, the data from one

subject were lost; the final analyses in this paper are based on the remaining 35

subjects.

Materials

Data Bases. A thesaurus data base developed by Boehm-Davis, Holt, Koll,

Yastrop, & Peters (1989) was used in this study. The thesaurus data base con-

tained relational information for 21 words, including broader terms, narrower

terms, and related terms to the target word. Three versions of the data base were

used — a graphic version, a tabular version, and a verbal version. Portions of

each of the three versions of the data base are illustrated in Figure 1.

Questions. A set of questions was written for use in this study. These questions

required modification of some type of information contained in the data base.

These questions required the subjects to make a change (add, modify or delete)

to an item (word or relationship) in the data base. This resulted in six types of

questions: adding a word, modifying a word, deleting a word, adding a relation-

ship, modifying a relationship, and deleting a relationship. Nine examples of

each type of question were generated to create a set of fifty-four questions.

Examples of the six types of questions are shown in Table 1. 7

The questions consisted of two parts: (1) computer-based questions that al-

lowed the change information to be entered into the data base and (2) paper-

based descriptions of the changes to be made. The computer-based questions

had to be modified slightly for use with each of the three data base formats;

however, care was taken to keep the questions as similar as possible. For exam-

ple, a sample question modifying a relationship, tailored for each type of data

base, 1s shown in Table 2. The accompanying paper documentation for the

changes also had to be modified to represent each of the three question formats

(1.e., graphic, tabular, and verbal). The three question formats are shown in

Figure 2 for modifying a relationship.

Procedure

Experimental sessions were conducted on an IBM PC. Initially, the partici-

pants were introduced to the data base with which they would be working. Each

participant was given a paper copy of one format of the data base (graphic,

DATA BASE FORMATS AND INFORMATIONAL RETRIVAL 111

(a) Spatial Form of the Data Base.

Heavy line connects Broader Term (above) to Narrow Term (below);

Thin line connects Related Terms.

Key:

communication

sychology

cybernetic homeostasis

advertising

(b) Tabular Form of the Data Base.

Key: _B: Term is above Broader Term for term on left;

N: Term is above Narrower Term for term on left;

R: Term above is Related Term to term on left.

communication

(c) Verbal Version of the Data Base.

Key: BT: Broader Term (s) for the key term;

NT: Narrower Term (s) for the key term;

RT: Related Term (s) to the key term.

communication NT: advertising, cybernetics, language, non_verbal_com

RT: feedback, language arts, linguistics, media, speech.

cybernetics BT: communication

NT: feedback

RT: homeostasis

feedback BT: cybernetics, homeostasis, psychology

RT: communication

Fig. 1. Portions of the graphical, tabular, and verbal versions of the thesaurus data base.

tabular, or verbal) to examine. Then, the subjects were presented with three

blocks of trials (one for each format of question, presented in random order).

Within each block, the first 6 questions were practice trials; they represented one

112

BOEHM-DAVIS, HOLT, AND PETERS

Table 1.—Sample questions in the verbal format

Add Word

Modify Word

Delete Word

Add Relationship

Modify Relationship

Delete Relationship

Add to the data base the word that is circled on your change form.

To add the word, you will only need to indicate where it should

be placed into the data base.

Enter the number of the word that the new word follows.

In the data base, replace the word that is crossed out with the word

shown next to it. To replace the word, you will only have to

give the location of the old word.

Enter the number of the word to be replaced:

Delete the circled word from the data base.

Enter the number of the word to be deleted:

Add to the data base the relationship shown by the circles on your

entry form. To add this relationship, you will need to give the

numbers of the two words and the type of relationship.

Enter the number corresponding to the key term:

Enter the row number for the kind of relationship you are adding:

Enter the position number showing which word you will insert the

new word after (from the list of words to the right of the key term):

In the database, change the relationship between the two words

indicated to the relationship shown within the circle. To change

this relationship, you only need to give the numbers of the two

words and the type of relationship.

Enter the number corresponding to the key term:

Enter the number corresponding to the word it is related to:

Enter the choice number that reflects the modified relationship

between the two words: (1) BT (2) NT (3)RT:

Delete from the data base the relationship that is crossed out on

your entry form. To delete this relationship, you will only need

to give the numbers of the two words that are related.

Enter the number corresponding to the key term:

Enter the location of the word it 1s related to:

modification of each type (1.e., adding, modifying, or deleting either a word ora

relationship). Following the six practice questions, twelve experimental ques-

tions were presented, two of each type of modification.

An interactive data collection system recorded the participants’ responses

throughout the session, and the time required for each response. After all of the

questions had been answered, the participants completed a questionnaire.

Results

Performance Measures

The two major performance measures in this study were the time or latency of

the response and the accuracy of the response. Response latencies were mea-

sured in hundredths of a second. Where multiple questions had to be answered

in order to implement a modification (i.e., when modifying relationships), the

average time required to answer the questions was calculated for use in the

analyses. Both of these dependent variables were analyzed bya3 X (3 X2X3X

DATA BASE FORMATS AND INFORMATIONAL RETRIVAL 113

Table 2.—Verbal, graphic, and tabular formats of the modify relationship question

Modify Relationship (Graphic Format) In the database, change the relationship between the two words

indicated to the relationship shown within the circle. To

change this relationship, you only need to give the numbers

of the two words and the type of relationship.

— Enter the location of the first word:

— Enter the location of the word it is related to:

— Enter the choice number that reflects the modified relationship

between the two words: (1) = = = = (2) --------- :

Modify Relationship (Tabular Format) In the database, change the relationship between the two words

indicated to the relationship shown within the circle. To

change this relationship, you only need to give the numbers

of the two words and the type of relationship.

— Enter the row number of the key term:

—- Enter the column number of the word it is related to:

—- Enter the choice number that reflects the modified relationship

between the two words: (1) B (2)N (3)R:

Modify Relationship (Verbal Format) In the database, change the relationship between the two words

indicated to the relationship shown within the circle. To

change this relationship, you only need to give the numbers

of the two words and the type of relationship.

— Enter the number corresponding to the key term:

— Enter the number corresponding to the word it is related to:

— Enter the choice number that reflects the modified relationship

between the two words: (1) BT (2) NT (3)RT:

2) repeated measures analysis of variance. The between-subjects factor was the

format of the data base (graphic, tabular, and verbal). In addition to format of

question (graphic, tabular, or verbal), the type of item modified (word or rela-

tionship), type of modification (add, modify, or delete) and time (first or second

presentation of the type of modification) formed additional within-subjects fac-

tors.

Response Latency. There was a main effect for data base format (F (2, 33) =

6.17, p< 0.01). The average latency of response for the graphic data base (13.56

sec) was shortest, with the tabular data base (14.62 sec) taking somewhat longer,

and the verbal data base taking the longest (19.01 sec), as can be seen in Figure 3.

This suggests that the graphic data base was “‘easiest”’ to use overall.

The main effects for question format and time were not significant at the 0.01

level (F'(2, 66) = 2.22, F (1,33) = 0.99). Main effects were significant for the type

of modification (add, modify, or delete) being made (F (2,66) = 27.52, p < 0.01)

and for the item (word or relationship) being modified (F (1,33) = 35.67, p <

0.01); see Figures 4 and 5. Making an addition to the data base required the most

time (19.32 sec), followed by making a modification (15.09 sec) and making a

deletion (12.78 sec). Modifying a word (average time per response = 18.00 sec)

took longer than modifying a relationship (average time per response = 13.46

sec). This result may have been an artifact of our data reduction methodology.

The relationship questions required answering two or three questions while the

114 BOEHM-DAVIS, HOLT, AND PETERS

communication

non_verbal_com ——DJ advertising

(a) | Graphic Form of the Modify Relationship Question (relationship shown on this paper is

the new relationship between the words).

_advertising

non_verbal_com

psychology

(b) Tabular Form of the Modify Relationship Question (relationship shown on this paper is

the new relationship between the words).

nanes

non_verbal_con Ba

NT:

RT: (“advertising >

psychology

(c) Verbal Form of the Modify Relatonship Question (relationship shown on this paper is

the new relationship between the words).

Fig. 2. Sample paper-based questions for modifying a relationship using graphical, tabular, and verbal

format of questions.

word questions required answering only one question. It may be that answering

the first question takes more time than answering the subsequent questions; this

would lead to a shorter time overall for the modifications that required answer-

ing several questions (i.e., the relationship questions).

DATA BASE FORMATS AND INFORMATIONAL RETRIVAL 115

TIME TO COMPLETE MODIFICATION TASK

20

O

3]

Ly 18

ea]

Ss

ke

a 16

z

e)

I

ea

=

fx] 14

4

12

GRAPHIC TABULAR VERBAL

TYPE OF DATABASE

Fig. 3. Main effect of data base format (in seconds).

The critical interaction between data base format and type of question, which

can be seen in Figure 6, was also significant (F (4,66) = 2.68, p < 0.04). In

addition, several of the other two-way interactions were found to be significant.

The item (word or relationship) by type of modification (add, modify, or delete)

TIME TO COMPLETE MODIFICATION TASK

20

18

O

w

y

isa)

S 16

e=!

~

2)

eae

3

12

ADD MODIFY DELETE

TYPE OF CHANGE

Fig. 4. Main effect of type of modification (add, modify, or delete) in seconds.

116 BOEHM-DAVIS, HOLT, AND PETERS

TIME TO COMPLETE MODIFICATION TASK

20

18

16

14

REACTION TIME (SEC)

12

WORD RELATION

TYPE OF MODIFICATION

Fig. 5. Main effect of type of item modified (word or relationship), in seconds.

interaction (Figure 7), was significant (F (2,66) = 34.53, p < 0.01). This interac-

tion showed that the amount of time required to modify a relationship was

constant across types of modification (add, modify, or delete, mean time =

TIME TO COMPLETE MODIFICATION TASK

BY TYPE OF DATABASE

24

22

% 20

2

¢

H 18

a

(e)

|

G 16

2 TASK

* GRAPHIC

14 4 TABULAR

@ VERBAL

12

GRAPHIC TABULAR VERBAL

TYPE OF DATABASE

Fig. 6. Interaction between data base format and format of question, in seconds.

DATA BASE FORMATS AND INFORMATIONAL RETRIVAL 117

TIME TO COMPLETE MODIFICATION TASK

30

IME (SEC)

N

wn

N

So

4 WORD

@ RELATIONSHIP

REACTION -

un

So

ADD MODIFY DELETE

TYPE OF CHANGE

Fig. 7. Interaction between type of item modified (word or relationship) and type of modification made

(add, modify, or delete), in seconds.

13.46 sec); however, the time required to add a word (25.34 sec) to the data base

was greater than that required to either modify (16.88 sec) or delete (11.78) a

word.

Accuracy. Data base format had an effect on accuracy (F (2,33) = 26.18, p <

0.01). As can be seen in Figure 8, performance on the graphic (99.3%) and

ACCURACY BY TYPE OF DATABASE

100

a

ie 9s

eal

<

=

UO

inte 39.0

85

GRAPHIC TABULAR VERBAL

TYPE OF DATABASE

Fig. 8. Main effect of data base format for accuracy.

118 BOEHM-DAVIS, HOLT, AND PETERS

ACCURACY OF MODIFICATION TASK

BY TYPE OF DATABASE

Ve)

18)

ACCURACY (%)

We)

oO

TASK

* GRAPHIC

85 4 TABULAR

@® VERBAL

80

GRAPHIC TABULAR VERBAL

TYPE OF DATABASE

Fig. 9. Interaction between data base format and format of question for accuracy data.

tabular (99.1%) data bases was almost perfect, while performance on the verbal

data base was reduced (90.3%). However, this main effect must be interpreted

with caution. The poorer overall performance of the verbal data base arises from

the combination of verbal data base with a graphic question, as can be seen in

Figure 9. This interpretation is supported by the significant interaction between

data base format and question format (F (4,66) = 4.51, p < 0.01). The critical

interaction also supported the main hypothesis that accuracy would be higher

when the form of the data base matched the form of the question. —

None of the other main effects were significant for the accuracy variable.

However, several two-way and three-way interactions were significant. An in-

teraction (Figure 10) was found between type of modification (add, modify, or

delete) and format of question (F (4,132) = 2.49, p < 0.05). For the verbal and

tabular questions, modifying an item (word or relationship) was more error-

prone than adding or deleting an item. For the graphical questions, adding an

item was much more error-prone than either modifying or deleting an item.

As in the latency data, the item (word or relationship) by modification type

(add, modify, or delete) interaction was significant (F(2,66) = 10.10, p < 0.01);

however, the form of the interaction was slightly different (see Figure 11). This

analysis showed that more errors were made when adding a word than when

adding a relationship; however, more errors were made when modifying a rela-

tionship than when modifying a word. When deleting words and relationships,

an equal number of errors was made.

DATA BASE FORMATS AND INFORMATIONAL RETRIVAL 119

ACCURACY BY TYPE OF CHANGE AND TASK TYPE

100

oo

eakels:

a

s

>

UO

U

c 90 TASK

* GRAPHIC

4 TABULAR

@® VERBAL

85

ADD MODIFY DELETE

TYPE OF CHANGE

Fig. 10. Interaction between type of modification (add, modify, or delete) and format of question for

accuracy data.

A three-way interaction between data base format, question format and type

of modification (add, modify, or delete, F (8,132) = 2.54, p < 0.02) suggests that

when working with a graphic data base, modifications and deletions are more

ACCURACY BY TYPE OF CHANGE

100

95

beet

>

3

Sr Tne5

U

cH

80 4 WORD

@ RELATIONSHIP

75

ADD MODIFY DELETE

TYPE OF CHANGE

Fig. 11. Interaction between type of item (word or relationship) modified by type of modification (add,

modify, or delete) for accuracy data.

120 BOEHM-DAVIS, HOLT, AND PETERS

difficult when working from a verbal question; when working with a tabular

data base, additions are more difficult than modifications and deletions when

working from a verbal question for all types of questions; when working with a

verbal data base, adding an item (word or relationship) is extraordinarily error-

prone when working from a graphic question.

An examination of the three-way interaction of data base format, modifica-

tion type (add, modify, or delete), and item modified (word or relationship, F

(4,66) = 5.05, p < 0.01) suggests that the two-way interaction seen in Figure | 1 is

due primarily to the pattern of results obtained with the verbal data base. For the

graphic and tabular data bases (Figure 10), the accuracy scores are all close to

ceiling; however, for the verbal data base, the pattern of results is similar to that

in Figure 11. That is, adding a word is more error-prone than adding a relation-

ship; modifying a relationship 1s more error-prone than modifying a word; and

deleting words and relationships are equally error-prone.

Discussion

Several basic results emerge from this investigation into the effect of data base

format on the ability to modify information in that data base. First, the results

support the hypothesis that the nature of the modification task to be performed

exerts a significant influence on the best form of information display. Perfor-

mance, as measured both by response time and by percent correct, was best

when the format of the data base presentation matched the type of information

to be modified.

Second, the graphic data base was generally easier to use than either the

tabular or verbal data bases. Comparing these results with the results for accu-

racy, we find that participants were more often correct with the graphic and

tabular forms of the data base than with the verbal form of the data bases. Unlike

our previous study (Boehm-Davis, Holt, Koll, Yastrop and Peters, 1989) where

the forms of the data base that facilitated the quickest responses were not neces-

sarily the same forms as those that produced more accurate responses, here the

data base that produced the quickest responses was also the one that produced

the most accurate modifications. These combined results suggest that for an

interactive data base task with this type of content (relational data about words),

the graphic form was best.

Finally, the results suggest that specific types of modifications may be easier or

more difficult to make. The latency data showed a main effect of type of modifi-

cation (add, modify, or delete) which suggested that adding items to a data base

is more difficult than modifying or deleting an item. This is interesting in light of

DATA BASE FORMATS AND INFORMATIONAL RETRIVAL 121

the fact that adding an item to our data base did not require any more entries

than deleting an item, and that modifying an item took more steps (although the

data were calculated on the average time per input, not on total time). Further,

the interaction between the format of the question (graphic, tabular, or verbal)

and type of modification (add, modify, or delete) in the accuracy data suggests

that there are specific task factors at work. This would not be surprising accord-

ing to Vessey’s (1991) notion of “‘cognitive fit’, if different tasks tap different

underlying representations. However, it is not immediately apparent from an

examination of our three types of modifications why they would be suited to

different forms of external representation, or what the underlying variables

accounting for these results might be.

In summary, the data suggest that users are faster and more accurate in modi-

fying information in the data base when the modification to be made is pre-

sented in the same format as that of the information in the data base. However,

the interactions found in this study also suggest that the picture is not that

simple. These interactions suggest that there are task-specific variables that influ-

ence the general pattern of results. This suggests that, in future research, we

broaden the types of tasks examined, and perhaps look for dimensions underly-

ing tasks that might be used to predict performance.

Implications for Data Base Design

These results suggest some implications for designing data bases. First, the

interaction between question format and data base format suggests that the

nature of the searches that are likely to be performed on the data base should be

considered when choosing an interface format. Specifically, the results suggest

that designers should conduct task analyses to determine the nature of a typical

retrieval situation.

However, it is likely that more than one type of question will be asked in any

given data base. One solution would be to take an averaging approach, that is, to

choose the type of data base format that serves the average type of retrieval

question. However, the overall speed-accuracy tradeoff in the search task sug-

gests that one format for all question types may not be the best route to take, at

least for static data bases.

Another possible solution would be to use multiple display formats. However,

that leaves open the question of who would choose the format to be displayed.

One alternative would be to give users the option. This assumes users know their

best options. Although one of our earlier studies (Boehm-Davis, Holt, Koll,

Yastrop, & Peters, 1989) suggested that this may be the case, another study

(Peters, Yastrop, & Boehm-Davis, 1988), and earlier research by Vicente,

Hayes, and Williges (1987) suggested that more basic cognitive and spatial fac-

122 BOEHM-DAVIS, HOLT, AND PETERS

tors are more important. The alternative would be to have the computer decide

which format to display. This could be done either by determining what task is

being performed or by following a user profile.

Each of these options comes with a cost and the benefit to be derived from

tailoring data bases. If a builder has a captive audience, that is, one that must use

the system, the tradeoff appears to be between the increased cost to develop the

system and the risk of increased errors, response time, and reduced user satisfac-

tion in using the system. Although users may not have a choice of which system

to use, they always have the choice of whether or not to use the system at all. For

users who do have a choice of which system to use, the tradeoff appears to be

between the increased design cost and later loss of business. Thus, it may be

more important for developers to design a flexible interface when they are in a

competitive market place. However, for both groups of users, total system per-

formance is related to usability. That is, system performance can be defined as a

function of the ability to make effective use of a system multiplied by how much

someone actually uses the system. Using this metric, performance is related to

the amount of effort that goes into the original design, whether or not the user

has a choice about when to use the system.

In examining the value of this research, it is important to remember that the

data show that spatial skills are an important component of performance, even

for small data bases. When one moves to larger data bases, where the task of

navigating to the right location in the data base is added to the task of locating

information locally, it seems likely that the problem of individual differences

would be greatly magnified. In summary, the research suggests that designers

need to consider the use to which a data base will be put during system design,

and to recognize that individual differences play a great role in information

retrieval tasks.

Acknowledgements

This research was supported by the Office of Naval Research, Engineering

Psychology Group (Contract #N00014-85-K-0243). The views expressed in this

paper are not necessarily those of the Office of Naval Research or the Depart-

ment of Defense.

REFERENCES

Bennett, K. B. (1987). Mental models, conceptual model, and the design of graphic displays. Unpublished

manuscript, personal communication.

Boehm-Davis, D., Holt, R., Koll, M., Yastrop, G., and Peters, R. (1989). Effects of different data base formats

on information retrieval. Human Factors, 31:579-592.

DATA BASE FORMATS AND INFORMATIONAL RETRIVAL 123

DeSanctis, G. (1984). Computer graphics as decision aids: Directions for research. Decision Sciences, 15:463-

487.

Durding, B. M., Becker, C. A., and Gould, J. D. (1977). Data organization. Human Factors, 19:1-14.

Lalomia, M. J., & Coovert, M. D. (1987). A comparison of tabular and graphical displays in four problem-

solving domains. SIGCHI Bulletin, 19:49-54.

Peters, R., Yastrop, G., & Boehm-Davis, D. A. (1988). Predicting information retrieval performance. In

Proceedings of the Human Factors Society Annual Meeting, 32:301-305.

Powers, M., Lashley, C., Sanchez, P., & Shneiderman, B. (1984). An experimental comparison of tabular and

graphic data presentation. /nternational Journal of Man-Machine Studies , 20:545-566.

Sinha, A., & Vessey, I. (1991). Cognitive fit: An empirical study of recursion and iteration. Pennsylvania State

University. Unpublished manuscript.

Vessey, I. (1991). Cognitive fit: A theory-based analysis of the graphs versus tables literature. Decision

Sciences, 23:219-240.

Vessey, I. & Galletta, G. (1991). Cognitive fit: An empirical study of information acquisition. Information

Systems Research, 2:63.

Vicente, K. J., Hayes, B. C., & Williges, R. C. (1987). Assaying and isolating individual differences in search-

ing a hierarchical file system. Human Factors, 29:349-359.

Wickens, C. D., & Scott, B. D. (1983, June). A comparison of verbal and graphical information presentation in

a complex information integration decision task. (Technical Report EPL-8 3-1/O NR-83 -1) Champaign, IL:

University of Illinois at Urbana-Champaign.

Wright, P., & Fox, K. (1972). Explicit and implicit tabulation formats. Ergonomics, 15:173-187.

| A eld AR MA

yar viel (rats reese * ies

ae Oe C ue

| Li: a sean ‘ hai vm

Peek, 2 OV Vea +4 phe Le tats fry spat Gt) © (ya

> a3

Sa ay

varie ft

"

¥¢ Creeley: : eer nt

} } j : ’ b

A

3 5)

% 1

r iy ' *

i PY | fe

ory 4

Pin, a LAE Ib Oe

{ ‘ phan ;

Wa hp4 Bes Dina

a * A eee

+ aura rm is ity iat

: ; by # i Bo 0%, Ck ) WFR

Hi Minit Are ron dioptabet. 4 88

On! ‘a é ye

{ ? i)

is

Journal of the Washington Academy of Sciences,

Volume 83, Number 2. Pages 125-131, June 1993

No Arrow of Time

Joe Rosen!

Department of Physics, The Catholic University of America, Washington, DC 20064

and

Department of Physics and Astronomy, University of Central Arkansas,

Conway, AR 720357

ABSTRACT

The fallacy of “‘the flow of time” and “the arrow of time” is pointed out, and a better

notion of time, in terms of becoming, is offered.

Introduction

Much, indeed very much, has been written and said about the various aspects

of time, ranging through the philosophical, the psychological, the artistic, the

social, and so on to the technical physical. For a sampling one might refer to the

published proceedings of the International Society for the Study of Time, whose

eighth conference was held in the summer of 1992. For a taste of the technical

physical approach see Zeh (1989), for example.

The present article is offered as an addition to that print deluge, because many

of us who are occupied with science, but are not devoting special thought to

time, are being misled by much of the more publicized expositions about time.

And that is especially acute with regard to the notion expressed by the terms

“arrow of time,” “direction of time,”’ and expressions in similar vein, the notion

that time could “flow one way” rather than “‘flow the opposite way.”

Let us start then with a brief discussion of “‘the flow of time.”

' On leave from the School of Physics and Astronomy, Tel-Aviv University, 69978 Tel-Aviv, Israel.

? Present address.

125

126 ROSEN

The Flow of Time

We perceive change in ourselves and in our environment. (Perception itself

involves change.) We use the term “‘time’”’ to indicate that change in general. We

might (and I do) define time as the dimension of change (Rosen, 1991, Chapter

7), or as the possibility, or potentiality, or capacity, of nature for change.

Now, it is very common to liken time to a flow. One view is passive: events

flow toward us from the future, we perceive them in the present, then they

recede from us into the past. The “‘motion”’ is from the future to the past. A

simile is a river of events flowing past a stationary perceiver standing on the

shore.

Another view is active: we flow from the past to the future, experiencing

events along the way. The “motion” here is from the past to the future. A

corresponding simile is the perceiver in a boat, floating along with the river

current past the stationary events along the shore. That might remind physicists

of the “‘block universe” picture, in which one’s consciousness and sense of

present is supposed to creep along one’s world line.

Time in Physics

Physics is completely lacking a flow of time. Our sense of time’s passage has

no representation in physics. Time in physics is a parameter, commonly de-

noted by ¢, that can take on a range of values. But there is nothing in physics that

makes f¢ take any specific value or any sequence of values.

Thus we can “run” our mathematical models “‘forward”’ in time, by consider-

ing their behavior as ¢ takes on an increasing sequence of values, or “backward”

in time, as ¢ assumes a decreasing sequence. That brings us to time reversal

symmetry for models.

Time Reversal for Models

A model is said to possess time reversal symmetry if its set of behaviors for

increasing sequences of ¢ values is the same as its set for decreasing sequences.

That can be checked for equations by changing the sign of ¢ and of all odd-order

time derivatives and seeing whether the resulting equation possesses the same set

of solutions as the original one.

For example, Newtonian mechanics with elastic forces 1s a time reversal sym-

metric model. Any process obtained for a decreasing sequence of t values can be

obtained, with a suitable choice of initial conditions, for an increasing sequence.

The diffusion equation, on the other hand, is not time reversal symmetric.

The process of concentration that results for a decreasing sequence of ¢ values is

not a behavior obtained for an increasing sequence. (It could be obtained by

NO ARROW OF TIME 127

changing the sign of the diffusion coefficient, but the sign of the diffusion coefh-

cient is an ingredient of the model.)

Time Reversal for Nature

That was time reversal symmetry (or asymmetry) for models. Time reversal

symmetry for nature operates closer to the real world. (1) First, let TS denote the

state that is the time reversal transform of state S, where TS is obtained from S$

by changing the sign of all generalized velocities or of the quantum phase, as the

case may be. If a state S is not characterized by any generalized velocity or

quantum phase, then TS and S will be identical. (2) Second, consider any pro-

cess of some physical system by which an initial state J evolves into a final state

F, I — F. (3) Third, consider the process that develops from the state TF, the

- time reversal transform of final state F, when TF is taken as the initial state of the

same physical system. (4) Finally, if state TF evolves into the time reversal

transform of state J, T/, 1.e., if TF — TJ, for all states J of the system, then the

system is said to possesses time reversal symmetry. (The situation is actually

more complicated. See Rosen (1994b).)

Time reversal symmetry of a physical system can thus be expressed as the

validity of the following diagram for all states J of the system (Rosen, 1983,

Chapter 6):

Time reversal

I Tl

Natural — Natural

Evolution Evolution

| Time reversal

TF

Or in everyday terms, a physical system is time reversal symmetric, if for any

process of the system a movie of that process projected in reverse depicts a

possible process of the system.

An example of a time reversal symmetric system is any isolated system that

does not involve neutral kaons (or K mesons), when the system is considered

microscopically. On the other hand, such a system that does involve neutral

kaons is not time reversal symmetric.

A sufficiently complex system, when considered macroscopically (where ma-

crostates are equivalence classes of microstates), can be time reversal asymmet-

ric. That is macroscopic irreversibility and is related to thermodynamic consid-

erations.

128 ROSEN

As for the Universe as a whole, since physics deals only with the single Uni-

verse we have and the Universe is expanding, it seems physically meaningless to

consider whether contraction is a possible process for the Universe.

Arrows of Time

The term “arrow of time” is used to refer to any phenomenon that is time

reversal asymmetric. Here are extremely brief descriptions of some arrows of

time.

The most obvious arrow of time is the psychological arrow of time, our sub-

jective time sense. We remember the past, are intensely aware of the present,

and anticipate the future. We do not remember the future and do not anticipate

the past. |

Then there is the thermodynamic arrow of time, the time reversal asymmetry

of sufficiently complex systems considered macroscopically. This arrow of time

can by coupled with the psychological one by the claim that the stable recording

of memory traces is an entropy increasing process, so that we cannot but per-

ceive entropy increase in our surroundings.

The submicroscopic world does not offer an arrow of time, except through the

neutral kaon. But that effect is generally considered too small to have significant

effect on the ordinary scale.

The cosmic arrow of time is the expansion of the Universe, in the sense that

the Universe is in fact expanding and not contracting, independent of any con-

sideration of whether contraction is possible for it. Some deem this to be the

master arrow of time from which all others follow, but that approach is not

universally subscribed to.

Other arrows of time have been pointed out (Zeh, 1989).

As commonly understood, what seems to be the message of all this is that,

except for phenomena involving neutral kaons, the submicroscopic world pos-

sesses no arrow of time, and one direction of time is just as valid as the other. At

larger scales differentiation of temporal directions enters the picture, and a

fortiori our observation of the world around us imposes sharp discrimination

between directions of time. Some would add that it is the expansion of the

Universe that underlies all that and fundamentally endows time with a direc-

tion.

No Arrow of Time

That appears to be what is commonly understood. But as I wrote in the

Introduction, that is misleading. Time has no arrow, no direction. The very

notion of directionality has no relevance for time. While “the flow of time” is a

NO ARROW OF TIME 129

beautiful metaphor, it must not be stretched to the extent of thinking of time as

flowing one “way” rather than another.

I think the arrow fallacy is a result of conceptual spatialization of time’s

“flow,” against which Capek (1961, Chapter 11), among others, strongly warns

us. After all, although a river actually flows in one direction, it can be imagined

flowing in the opposite direction. So when the “flow” of time is likened, as it so

commonly is, to the flow of a river, it is all too easy to fall into the fallacy of

ascribing to time a direction of flow and the possibility of flow in the opposite

direction. And so it is also with the “‘block universe” picture; if we imagine we

“move” along world lines, then why not “move” in the opposite direction?

But time, as I mentioned in Section 2, is nature’s dimension, or possibility, or

potentiality, or capacity, of change. That is it. Simple, in a way. Events occur;

_ things change. The undoing of a change is also a change, an event or sequence of

events. A time reversed process is also a process. None of that is time “running in

reverse.’ Time does not run anywhere. Events occur; things change. That is

time.

Indeed, we quantify time and find it useful to represent time on an axis in

4-dimensional space-time. But nevertheless time is not a spatial dimension.

Even the theories of relativity do not completely unify time with space. In fact

they actually emphasize the distinction of time from space; the light-cone struc-

ture of space-time, reflected in the indefinite metric, assigns to time a different

character from that of space. Just because we can move along the x axis and can

move just as well in the negative x direction as in the positive, does not imply

that we must then necessarily ““move’”’ in time or, moreover, that it is necessarily

meaningful to ““move”’ in one “‘direction”’ rather than in the other.

Becoming

Well, if time does not run anywhere, does not really flow, and has no direc-

tion, how then might we better grasp the notion of time? We can do very well by

focusing on change, as I stated in Section 2. And I think the following reasoning

iS persuasive.

We human beings have evolved to be very well adapted to the world we live in,

since we are on the whole thriving quite successfully. Thus the way we perceive

and conceive of the world cannot be terribly mismatched with what is really

going on there, at least at the ordinary scales of lengths, time intervals, speeds,

ete.

Now, our temporal conception is of an ordered sequence of events, of

changes, involving: (1) the past, consisting of those events that have happened,

some of which we remember; (2) the present, involving the events currently

happening; (3) and the future, consisting of whatever will occur. Of the three, the

130 ROSEN

present is the only perceived reality. The past is our mental construct of what we

remember to have happened together with whatever else might have happened

even if we do not remember it. The future is our anticipation of more events to

occur after whatever is happening now.

We perceive a present of occurring events, changes, which is well termed

“becoming.” Our temporal perceived reality is of becoming. Normally we are

constantly aware of the world’s becoming, including our own becoming and our

awareness as part of that becoming. (I assume that the self-referential circularity

of our awareness of our being aware is deeply involved in our time sense.)

So that should reflect what is really going on in the world, at least for ordinary

scale time intervals (sufficiently longer than, perhaps, the Planck time of about

10°* s and sufficiently shorter than, say, the Hubble time of some 10!° y—I

strongly suspect that our intuition is at least as poorly adapted to the extra large

scale as it is to the extra small). Thus time should be thought of as a universal

wave of becoming, the combined becomings at all locations in space. Our con-

sciousness, as basically a physical phenomenon, participates in the universal

wave of becoming, and, one might say, “rides the wave.”

The universal wave of becoming is not to be understood as a universal wave of

simultaneity! We know better, from the special theory of relativity. Simultaneity

is a convention that is operationally defined and observer dependent. The uni-

versal wave of becoming is not operational and is observer independent. It is a

good way of thinking about time that is less likely to lead to the fallacy of an

arrow, or a direction, for time.

Besides its much better representing what time is all about, this view of time in

terms of (nondirectional) becoming, of change, rather than in terms of direc-

tional flow, has the following additional advantage. Many questions about time,

all those questions concerned with the issue of time’s “‘arrow,” time’s “‘direc-

tion,” become nonquestions, and the picture simplifies enormously. It is not

that the questions are swept under the rug. Rather, they are validly dismissed as

irrelevant. The situation is very much like that of astronomical epicycles. When

the motions of the planets were grasped in terms of ellipses rather than circles, all

questions involving epicycles went straight to the trash heap of irrelevancy. And

similarly for questions about the composition, form, and color of ghosts, when,

as long as there 1s no objective evidence for them, ghosts are generally assumed

not to exist.

In my holistic moods I like to think of the becomings at the various locations

in space, which all together constitute the universal wave of becoming, as some-

how linked, perhaps coordinated, via a deeper, nontemporal and nonspatial

level of reality underlying space-time. That level might be partially accessible via

the quantum. I have elaborated on such ideas elsewhere (Rosen, 1994a).

NO ARROW OF TIME 131

Acknowledgments

I would like to express my deep thanks to Lawrence Fagg and to Avshalom

Elitzur for many interesting and useful discussions about time.

References

Capek, M. (1961). The Philosophical Impact of Contemporary Physics. Van Nostrand: Princeton, NJ.

Rosen, J. (1983)..A Symmetry Primer for Scientists. Wiley: New York, NY.

Rosen, J. (1991). The Capricious Cosmos. Macmillian: New York, NY.

Rosen, J. (1994a). Time, c, and nonlocality: a glimpse beneath the surface? Physics Essays, (in press).

Rosen, J. (1994b). Universal microirreversibility and indeterminism in classical dynamical systems. J. Wash.

Acad. Sci., (in press).

Zeh, H.-D. (1989). The Physical Basis of the Direction of Time. Springer-Verlag: Berlin, Germany.

was

‘nee

RG) Picue Lan

aS Tey

s

‘2

thes 4

ee A “

1%

r

6 M

ia _

"

x

i

‘

’ ee |

x >

os

Vi wre

Lass * e

DELEGATES TO THE WASHINGTON ACADEMY OF SCIENCES,

REPRESENTING THE LOCAL AFFILIATED SOCIETIES

Piilasopiaical: SOGGY OF WASHINGION, 22. icc. ac cc ee cece ceed dee daewes Thomas R. Lettieri

PEnopelofical SOctety Of WaSHINGLOM: 26)... 6.6 ss sc ccc e ee dees eesnetececes Jean K. Boek

Bralomical SOciety Of WasmINGtOM 6.5 Mb. L.ncicecc ee cee cae eee dase deasescues Kristian Fauchald

erred SOCIETY OF VV aSMINPIOM 2262 so. eda Sees eo eicibn was beaaboenceuss Elise A. B. Brown

Entomological Society of Washington ..... 2.2.02... ce eee eee e ee ele F. Christian Thompson

EE mErEMEECOCEAPINC SOCICLY icc ci codec dics cc dec bewaSecescneeddaeeseuws Stanley G. Leftwich

PEA SMCICLY OF VWWASHINGLOM 2021) 6c) csiiaveis cs cars c cee daca deed osiieaee ssid vce ees VACANT

mest socicty of the District of Columbia... 00... 50.000. ccc ence e cease cnencee John P. Utz

Emel smciery Of Washington,DC 2: 2. ccc ss cc eck te cck seec as seennncneaneeuesd VACANT

MME TIESOCICLY, Of WASMINGTON oie. ca occ ccc cae cece e occ caceeesstasbaceses Muriel Poston

Society of American Foresters, Washington Section ...............0. cece cece Eldon W. Ross

MEIER SOCICLVIO! FNMEINCEIS 0. 6.52 cis. ccinlc asle hides doses sdecsiotabieeceubawes Alvin Reiner

Institute of Electrical and Electronics Engineers, Washington Section ........ George Abraham

American Society of Mechanical Engineers, Washington Section ............ Daniel J. Vavrick

Mommarioispical Society Of Washington ..... 050.50... secs sea ccccce ee ccecedcees VACANT

_ American Society for Microbiology, Washington Branch ................0.ce0ceeeees Ben Tall

Society of American Military Engineers, Washington Post ................. William A. Stanley

American Society of Civil Engineers, National Capital Section ..................... VACANT

Society for Experimental Biology and Medicine, DC Section .............. Cyrus R. Creveling

Peimeroational, Washington Chapter ...............0.0.00csccccsccccccees Richard Ricker

American Association of Dental Research, Washington Section ............. J. Terrell Hoffeld

American Institute of Aeronautics and Astronautics, National Capital

maT es Cre ae Rete Esc wth alegre deg 6i4 » Wie s 4 Hats Reginald C. Smith

Pumerican Meteorological Society, DC Chapter ..................0c0ceceeees A. James Wagner

Besmeenee SOclely Of WashiNgtOm ... of. cal. coke cc ec cen e eevee sssceeccnces To be determined

Acoustical Society of America, Washington Chapter .....................05. Richard K. Cook

panerican Nuclear Society, Washington Section ....2............000.eee cece cece es Kamal Araj

Institute of Food Technologists, Washington Section ..................0000eeee Roy E. Martin

American Ceramic Society, Baltimore-Washington Section .................. Curtis A. Martin

ene OEIC DY eo 2 eer et a Man gk 24’ olab adseew aug aeene’e nas Regis Conrad

Moasietonuaistory Of Science Club ............ 0.0.0.0. cece ecw eeceeeee Albert G. Gluckman

American Association of Physics Teachers, Chesapeake Section ............. Robert A. Morse

Optical Society of America, National Capital Section ...................... William R. Graver

American Society of Plant Physiologists, Washington Area Section ............. Steven J. Britz

Washington Operations Research/Management Science Council .............. John G. Honig

insimment Society of America, Washington Section ...................0000cee sense VACANT

American Institute of Mining, Metallurgical and Petroleum Engineers,

OEE UDG IA STOUT Me APs beet re ne Anthany Commarota Jr.

eeememeapital ASIroOMOMETS .. 2). 26.256 oo cc cae cece eccccieecccaveoeews Robert H. McCracken

Mathematics Association of America, MD-DC-VA Section ................. Sharon K. Hauge

Pistncehor Columbia Institute of Chemists ...............2...es0.0sce0000- William E. Hanford

Mister of Columbia Psychological Association ............¢...2....0.-0: Marilyn Sue Bogner

Srsmarrouraimnt bechnolopy Group 1... o.oo ee oll elec cece eee e ccs Lloyd M. Smith

American Phytopathological Society, Potomac Division .................... Kenneth L. Deahl

International Society for the System Science, Metropolitan Washington

“ ILSSYEEIE Ga had Aho a RIM CB) aillonthssg SIU oe RENO elated OT David B. Keever

inuman Factors Society, Potomac Chapter .... 0.0... 2.0 66..eccceeeseecccees Thomas B. Malone

pimerican risheres Society, Potomac Chapter ...............¢s0.....0000e: Dennis R. Lassuy

Association for Science, Technology and Innovation ...................5. Clifford E. Lanham

emia SOCIOIOPICAl SOGICLY 3.2546 os avis ccccics oes s secu wes cuclesnseess Ronald W. Manderscheid

Institute of Electrical and Electronics Engineers, Northern Virginia

VEE TUE 1, ARIE ORES IAC ECTS dn Uk NSN ee ee aI 0 Blanchard D. Smith

Association for Computing Machinery, Washington Chapter ............. Charles E. Youman

Pe asuimeronr statistical SOGIety. 2.) bs Sau ko cee cee cn cS ckeucnees ci sesacesdus cece David Crosby

Society of Manufacturing Engineers, Washington, DC Chapter ............... James E. Spates

Institute of Industrial Engineers, National Capital Chapter ................ Neal F. Schmeidler

Delegates continue to represent their societies until new appointments are made.

Washington Academy of Sciences 2nd Class Postage Paid

2100 Foxhall Road, NW at Washington, DC

Washington, DC 20007-1199 and additional mailing offices.

Return Postage Guaranteed

wsiy7

N [++ VOLUME 83

Number 3

ae our nal of the September, 1993

WASHINGTON

ACADEMY .. SCIENCES

ISSN 0043-0439

Issued Quarterly

at Washington, D.C.

CONTENTS

Articles:

JOE ROSEN, “Universal microirreversibility and indeterminism in classical

nM ANAICAN SY SEMIS: 4 |o betcyiam sein ve pleco sieica. « 2! 6 + eines ie oR acl oe CRE ao RO 133

JAMES A. BALLAS, “Interpreting the language of information sound” ......

JOHN M. REISING, TERRY J. EMERSON & KRISTEN K. LIGGETT,

SDisplayins: information lin fUpUTe COCKPIUS. (2. wack Ge dicks sree ce he wiee see ec eas

Washington Academy of Sciences

Founded in 1898

EXECUTIVE COMMITTEE

President

John H. Proctor

President-Elect

Rev. Frank R. Haig, SJ

Secretary

Thomas R. Lettieri

Treasurer

Norman Doctor

Past President

Stanley G. Leftwich

Vice President, Membership Affairs

Cyrus R. Creveling

Vice President, Administrative Affairs

Grover C. Sherlin

Vice President, Junior Academy Affairs

Marylin B. Krupsaw

Vice President, Affiliate Affairs

Thomas W. Doeppner

Board of Managers

James W. Harr

Clifford M. Krowne

Herbert H. Fockler

Nina M. Roscher

William B. Taylor

Neal F. Schmeidler

REPRESENTATIVES FROM

AFFILIATED SOCIETIES

Delegates are listed on inside rear cover

of each Journal.

ACADEMY OFFICE

2100 Foxhall Road, N.W.

Washington, D.C. 20007

Phone: (202) 337-2077

EDITORIAL BOARD

Editor:

Bruce F. Hill, Mount Vernon College

Associate Editors:

Milton P. Eisner, Mount Vernon Col-

lege

Albert G. Gluckman, University of

Maryland

Marc Rothenberg, Smithsonian Insti-

tution

Marc M. Sebrechts, Catholic Univer-

sity of America

Edward J. Wegman, George Mason

University

The Journal

This journal, the official organ of the Washing-

ton Academy of Sciences, publishes original _

scientific research, critical reviews, historical —

articles, proceedings of scholarly meetings of

its affiliated societies, reports of the Academy,

and other items of interest to Academy

members. The Journal appears four times a

year (March, June, September, and De-

cember). The December issue contains a di-

rectory of the current membership of the

Academy.

Subscription Rates

Members, fellows, and life members in good

standing receive the Journal without charge.

Subscriptions are available on a calendar year —

basis, payable in advance. Payment must be

made in U.S. currency at the following rates:

U.S. and Canada

Other countries. 25...4.55 eee

Single copies, when available .......

eeceree ee ec eee ee we ee oe

Claims for Missing Issues

Claims will not be allowed if received more ~

than 60 days after the day of mailing plus time

normally required for postal delivery and ©

claim. No claims will be allowed because of —

failure to notify the Academy of a change of ~

address.

Notification of Change of Address

Address changes should be sent promptly to ~

the Academy Office. Such notification should —

show both old and new addresses and zip

codes.

POSTMASTER: Send address changes to —

Washington Academy of Sciences, 2100 Fox- —

hall Road, N.W. Washington, DC 20007- —

1199.

Journal of the Washington Academy of Sciences (ISSN 0043-0439)

Published quarterly in March, June, September, and December of each year by the Washing-

ton Academy of Sciences, 2100 Foxhall Road, N.W., Washington, DC, 20007-1199. Second

Class postage paid at Washington, DC and additional mailing offices.

Journal of the Washington Academy of Sciences,

Volume 83, Number 3, Pages 133-141, September 1993

Universal Microirreversibility and

Indeterminism in Classical

Dynamical Systems

Joe Rosen!’

Department of Physics, The Catholic University of America, Washington, DC 20064

and

Department of Physics and Astronomy, University of

Central Arkansas, Conway, AR 720357

ABSTRACT

In order to help make the ideas of classical microirreversibility and chaos more accessible

to interested nonspecialists, a narrative style discussion is presented. It is shown, based on an

operational understanding of time reversal symmetry, that irreversibility, i1.e., time reversal

asymmetry, is a property of the microevolution of classical dynamical systems in general.

Yet approximate microreversibility may be exhibited for sufficiently short durations. The

cause underlying universal microirreversibility is the generally divergent evolution of classi-

cal dynamical systems and the consequent sensitivity of the evolution to initial conditions.

Divergent evolution is responsible also for the indeterministic, i.e., chaotic, behavior of those

systems. .

Introduction

This article offers some thoughts in narrative form on determinism, indeter-

minism, chaos, and macroscopic and microscopic reversibility and irreversibil-

ity. No essentially new results are presented. But this way of looking at things,

and especially the symmetry considerations involved in this way of looking at

things, helps clarify what is going on and makes the ideas more accessible to. the

interested nonspecialist.

' On leave from the School of Physics and Astronomy, Tel-Aviv University, 69978 Tel-Aviv, Israel.

? Present address.

133

134 JOE ROSEN

The main aim of this article is to show, based on an operational understand-

ing of time reversal symmetry, that irreversibility, 1.e., time reversal asymmetry,

is a property of the microevolution of classical dynamical systems in general.

Even so, approximate microreversibility, 1.e., approximate time reversal sym-

metry of microevolution, may be exhibited for sufficiently short durations of

evolution. The cause underlying universal microirreversibility is the generally

divergent character of the evolution of dynamical systems and the evolution’s

consequent sensitivity to initial conditions. Divergent evolution is responsible

also for the indeterministic, 1.e., chaotic, behavior of those systems.

The presentation proceeds as follows. In the next section we look into the

meaning of time reversal symmetry in general. In the section entitled ““Macro-

irreversibility” there is a succinct discussion of the macroirreversibility of

dynamical systems possessing an equilibrium macrostate. We see how the gener-

ally divergent evolution of dynamical systems brings about their indeterminis-

tic, 1.e., chaotic, behavior in the section entitled ““‘Determinism Lost’’. That such

systems may nevertheless exhibit approximate determinism for sufficiently

short durations is shown in the section entitled ““cDeterminism Approximately

Regained’’. After the discussion of macroirreversibility in the beginning of the

article as preparation, we see in the last section, entitled ““Microirreversibility”’

that the criterion for time reversal symmetry, understood operationally, cannot

be met in general even for microevolution, whatever the dynamics of the system.

Thus microirreversibility is a universal feature of dynamical systems. However,

approximate microreversibility may be exhibited for sufficiently short dura-

tions. |

For the purpose of our discussion we consider the following model dynamical

system: a large classical, ostensibly deterministic system, isolated from its

surroundings, describable both in macroscopic and in microscopic terms. As an

example, think of an isolated quantity of ideal gas. For simplicity of presentation

we usually take the natural evolution of the system to be discrete: initial state >

final state. We assume that the system possesses a unique equilibrium macro-

state, which is a certain macrostate into which the system evolves from any

initial macrostate. An equilibrium macrostate evolves, of course, into itself.

A macrostate corresponds to a class of microstates. I.e., if the system is in any

microstate of such a class, it will be in the corresponding macrostate. Thus

microstate space decomposes into classes, of which each class but one corre-

sponds to a unique macrostate, while the exceptional class contains all micro-

states to which no macroscopic description is appropriate.

Of microevolution and macroevolution the former is the more fundamental,

giving rise to the latter through course-graining, i.e., through our ignoring cer-

tain details of microstates and considering only equivalence classes of them. Just

UNIVERSAL MICROIRREVERSIBILITY AND INDETERMINISM 135

how that comes about in technical detail 1s, it seems, still under investigation by

the experts in the field.

Time Reversal Symmetry

We start by defining the time reversal transformation of states, denoted T: For

any classical (macro or micro) state S, the corresponding time reversed state TS

is obtained from S' by reversing (1.e., changing the sign of) all generalized veloci-

ties (i.e., the time derivative of generalized coordinates, such as the ordinary

velocity of a point particle, the angular velocity of a rigid body, the time deriva-

tive of a field intensity, etc.) characterizing S, while leaving all other properties of

S unchanged. (Thus the time reversal transform of a time reversal transform is

the original state itself; TT.S = S.) If state Sis not characterized by any general-

_ 1zed velocity, then TS and S will be identical.

Now, for any classical dynamical system consider a general evolution,

whereby initial (macro or micro) state J evolves into final state F. Consider the

time reversal transform of final state F, TF, as an initial state. If TF evolves into

the time reversal transform of initial state J, TY, 1.e.,if TF — TY, for all states J of

the system, then the system’s evolution possesses time reversal symmetry and

the system is said to be reversible (Rosen, 1983, Section 6.1).

Note that a necessary condition for reversibility is that the evolution be non-

convergent, 1.e., that different states always evolve into different states. To see

that, assume that different states J and J’ both evolve into the same state F. Now

consider the time reversal transform of state F, TF, as an initial state. In a

deterministic system TF will evolve into a unique final state, which thus cannot

be both T/ and TI’, since those states are different, as J and /’ are different by

assumption. Therefore reversibility is precluded by nonconvergent evolution.

Macroirreversibility

Thus whether or not the microevolution of our model dynamical system is

assumed to be reversible, its macroevolution is irreversible, as follows from the

assumed convergence property, that every initial macrostate evolves into the

same (equilibrium) macrostate, where the latter evolves into itself. The simulta-

neous possession of both microreversibility and macroirreversibility is not for-

bidden by nature, since there indeed exist physical systems that possess both

microreversibility and macroirreversibility, such as a sample of gas. That has

been taken as a dilemma, a contradiction, a problem. How can reversible mi-

croevolution give rise to irreversible macroevolution (Balescu, 1975, Section

ee)?

In order to understand how that comes about, let us rephrase the question in

the light of our above discussion: How does course-graining give rise to conver-

136 JOE ROSEN

gent macroevolution, even when the fundamental microevolution is noncon-

vergent? The resolution of that depends crucially on the fact that for our model

to represent the real world we must also course-grain 1n time, 1.e., our observa-

tions must each be of truly macroscopic duration, must be time-smoothing.

What happens is the following.

Let us consider the system continuously in time and microscopically. Let it

start in some microstate belonging to the class corresponding to some nonequi-

librium macrostate. As time proceeds the system “‘wanders”’ through microstate

space, from microstate to microstate to microstate.. . . Each microstate along

the trajectory belongs to the class corresponding to one or another macrostate or

possibly to no macrostate. As we well know, statistical analysis shows that the

population of the equilibrium macrostate class is overwhelmingly greater than

that of all the other classes combined. Thus, as the system evolves, it spends

overwhelmingly more time in its equilibrium macrostate class than in any other

macrostate class or in the class of no macrostate (Balescu, 1975, Chapter 4).

Since the system starts in a nonequilibrium macrostate class, some time must

pass before it enters the equilibrium macrostate class for the first time. The

characteristic time duration for that to happen 1s called the “relaxation time” of

the system. The relaxation time for an ordinary sample of gas, for example, is

about 107'! s (Balescu, 1975, Section 13.2). So after a time interval of the order

of the relaxation time the system is in the equilibrium macrostate class and stays

there for a long time before wandering out.

The characteristic time duration for remaining in the equilibrium macrostate

class once entering it can be called the “excitation time” of the system. The

excitation time for an ordinary sample of gas, for example, is probably longer

than the age of the universe. (““You should live so long,”’ Boltzmann is supposed

to have told Zermelo (Lebowitz, 1983).) When the system finally does ““wander”’

out of the equilibrium macrostate class, it will stay out for a time interval of the

order of the relaxation time, only to return to the equilibrium class and remain

there again for a time interval of the order of the excitation time. Thus the

fraction of time the system spends out of macroscopic equilibrium is of the order

of the ratio of its relaxation time to its excitation time. That fraction is less than

about 10°”? for an ordinary sample of gas, for example. It follows then that a

macroscopic duration for observations is any duration that is much longer than

the relaxation time. For such observations the evolving system is always found

to be in equilibrium, with average relative error of the order of the above ratio.

Determinism Lost

Until fairly recently the commonly held conception of the character of the

evolution of isolated classical dynamical systems was that the evolution is in

UNIVERSAL MICROIRREVERSIBILITY AND INDETERMINISM 137

general nondivergent. This means that for a given system similar initial states,

1.€., initial states that are close to each other in state space, generally evolve into

similar (close) final states, where “similar” and “‘close”” have whatever signifi-

cance is appropriate to the system under consideration. Or, considered continu-

ously, evolution trajectories in state space that start off close to each other

generally remain close for their entire duration.

It had traditionally been considered safe in the context of classical physics to

assume that it is possible in principle to set up an isolated classical system in any

of its possible states. At the same time it had been recognized that in practice this

is not exactly true, due to uncontrollable residual imprecision. (And not merely

in practice, since quantum theory sets limits on the precision of setting up

classical states, such as states characterized by both position and momentum.)

_ But, in line with the assumed generally nondivergent evolution, that impreci-

sion had been taken as inconsequential, since it had been assumed that even if

we did not manage to set up exactly the state we wanted but only a very similar

one, the system would anyway evolve into a state very similar to the one it

ideally would have evolved into.

Yet fairly recent developments in the field of dynamical systems have led to

the belief among those familiar with the field that the evolution of isolated

classical dynamical systems is generally characterized by divergence rather than

by nondivergence. Similar initial states of the same system, even if they differ

merely on the order of quantum uncertainty, do not in general evolve into

similar final states but into widely differing ones. Trajectories in state space

generally diverge in time (Horton, Reichl, & Szebehely, 1983, especially Lebo-

witz, 1983, and Misra and Prigogine, 1983, there, and Prigogine and Stengers,

1984). That effect is also variously called “instabilities,” “‘sensitivity to initial

conditions,” or “stochasticity of the motion” (Lebowitz, 1983). Nondivergent

evolution might occur in special cases.

An important implication of that for physics is that determinism, and hence

also predictability, are dead. And we are talking about classical physics! It is true

that laws and theories predict results and so imply determinism. But physics is

concerned first and foremost with phenomena. And in the operational dealing

with phenomena we are forced to realize that determinism is no longer with us.

How is that? Determinism means that, with suitable definition of state, the

initial state of any isolated dynamical system uniquely determines its final state.

Operationally this means that every time we set up a dynamical system in the

same initial state / it will evolve into the same final state F. But we cannot set up

a system in any initial state J more than once (and even then it will not be exactly

in the state we might have intended to set up). The best we can do is to set up a

system in a sequence of similar initial states J, J’, 7”, . . . , which evolve into

138 JOE ROSEN

final states F, F’, F”,. . . , respectively. As long as nondivergent evolution was

assumed, we thought we could confirm determinism by checking whether final

states F, F’, F”,. . . were similar. But it now appears that, as similar as initial

states J, I’, I”, . . . are made to be, even if they differ merely on the order of

quantum uncertainty, final states F, F’, F”,. . . willin general be widely differ-

ent. How, then, determinism? Thus classical trajectories in state space are in

general unobservable (Prigogine and Stengers, 1984, p. 264).

And even more than that. We know that an isolated system is but an idealiza-

tion; we know there are influences that cannot be totally screened out (and

perhaps also those that cannot even be attenuated by distance). So I prefer the

term “‘quasi-isolated”’ to “isolated,” where quasi-isolated means isolated as best

we can. Then, even if it were possible to set up a quasi-isolated dynamical system

in precisely the same initial state more than once, random influences from the

surroundings, as weak as they might be, generally would, due to the amplifying

effect of divergent evolution, preclude the system’s evolving into even similar

final states. Hence indeterminism again.

From the preceding discussion in this Section it follows that chaotic behavior

is the general fate of quasi-isolated dynamical systems. By Poincaré’s theorem a

truly isolated dynamical system will eventually evolve into a state as similar to

its initial state as desired. If the evolution were nondivergent, the system would

then go through a cycle of evolution very similar to the previous one, since it

would then be starting off again in a state very similar to its initial state. Hence its

characteristic behavior would in general be quasi-periodic, exhibiting Poincaré

recurrences (Lebowitz, 1983, and Balescu, 1975, Section 3.3). Indeed, a truly

isolated dynamical system’s behavior might even be exactly periodic. That

would happen if after some finite time it returned to its initial state exactly. The

evolution, however, is generally divergent, and similar is just not good enough.

Even when the system eventually evolves into a state very similar to its initial

state, its ensuing evolution will in general bear no resemblance to the previous

cycle. Furthermore there do exist anti-isolatory factors. Hence, no periodicity,

no quasi-periodicity, no Poincaré recurrences—only chaos.

Determinism Approximately Regained

If determinism is indeed dead, and predictability along with it, how is it that

we can do science? (Do not forget that reproducibility and predictability form

the cornerstones of the foundations of science.) And actually it seems very

reasonable that without determinism in nature living beings, and a fortiori intel-

ligent beings such as scientists, could not even exist. Consider the degree of

determinism that is necessary for our survival as individuals and as the human

race, for us to have stable, reliable memories as individuals and as a society, and

UNIVERSAL MICROIRREVERSIBILITY AND INDETERMINISM 139

so on. Indeterminism implies lack of predictability. And without predictability

we could not reliably control our bodies or anything else and could not survive

our interactions with our unpredictable environment and with each other. Yet

we exist and we do science. How does that accord with the discussion of the

preceding section?

Just as for the course-graining that brings about macroirreversibility, time is

the essential issue here. Although initially close trajectories in microstate space

will in general eventually diverge, some time may pass before they actually do so

appreciably. That time interval can be called the “chaotization time.” Thus the

chaotization time of a dynamical system is the characteristic time duration from

the start of its evolution until the onset of its chaotic behavior. A system’s

chaotization time might depend, even very strongly, on the values of system

parameters, such as flow speed of a fluid or temperature difference, for example.

(Nondivergent evolution is characterized by infinite chaotization time.) For

time intervals sufficiently shorter than its chaotization time, a dynamical sys-

~ tem’s behavior is deterministic to a sufficiently good approximation. Thus the

determinism we find in nature indicates that there do exist dynamical systems

whose chaotization times are very long compared with the durations of our

observations.

An additional source of determinism could also lie with the ability of open

systems far from macroscopic equilibrium to remain far from equilibrium while

generating order within themselves (i.e., decreasing their entropy, but at the

expense of increased entropy of their surroundings) (Prigogine and Stengers,

1984, Chapter 6, and Prigogine, 1980). Merely maintaining order, let alone

producing it, requires deterministic evolution. So such systems seem to be con-

currently “generating” determinism, actively extending their chaotization

times, presumably at the expense of reduced chaotization times of their

surroundings. Living beings are such systems.

Microirreversibility

Our discussion of macroirreversibility in the section bearing that title does not

present anything new. It does show succinctly how large dynamical systems

possessing a macroscopic equilibrium state are found to be irreversible when

considered macroscopically (by course-graining, in time as well). And that is

valid whether the microevolutions of the systems are reversible or not. The

discussion of macroirreversibility is intended to serve as preparation for our

discussion of microirreversibility, where we will see that in a certain sense mi-

croevolution, too, is irreversible whether it is reversible or not.

Our presentation of time reversal symmetry in the section of that title should

140 JOE ROSEN

be understood operationally as saying: If we have a process of a dynamical

system whereby state J evolves into state F, ] > F, we should set up state TF, the

time reversal transform of state F (by reversing all generalized velocities asso-

ciated with state F), and let it evolve, TF — _X. If state X is the same as state T/,

the time reversal transform of state J, the process is time reversal symmetric, or

reversible. If that is found to hold true for more and more states J of the system,

we suspect, then gain confidence in, time reversal symmetry of the evolution

and thus reversibility of the system. That is independent of whether we have a

law or theory for the system’s evolution. If we do have a law in the form of a set

of equations, time reversal symmetry is expressed by invariance of the set of

solutions of the equations under the substitution of —t for t, —d/dt for d/dt, —v

for v, etc.

Now, what we will see is that the operational criterion for time reversal sym-

metry cannot be met in general. And that is true even if there exists a time

reversal symmetric law or theory of microevolution, such as the classical theory

of electrodynamics. That is the meaning of the above oxymoronic declaration

that microevolution is irreversible whether it is reversible or not.

Traditionally, assuming nondivergent evolution for classical dynamical sys-

tems, we had thought it inconsequential for time reversal symmetry that, as

mentioned in the section “Determinism Lost”’’, states cannot be set up precisely

to specification. Thus, by the operational understanding of time reversal sym-

metry, even if we could not precisely set up state TF for the purpose of letting it

evolve and comparing the result with state TJ, as long as we set up a state

sufficiently similar to TF, it would evolve into a state sufficiently similar to the

ideal result so that its comparison with state TJ would be meaningful. But it now

appears that evolution is divergent in general. So setting up a state merely

similar to TF is no longer good enough, since the result of its evolution will in

general be a state bearing no resemblance to the state that would have resulted if

TF itself had evolved.

Thus, whatever the underlying dynamics, the operational criterion for time

reversal symmetry cannot in general be met. Try as we may to get things running

in reverse, the system will never go back again.

Nevertheless, the possibility of approximate determinism allows the possibil-

ity of approximate time reversal symmetry. As long as we limit our observations

of a dynamical system to durations sufficiently shorter than its chaotization

time, we will be able to meaningfully apply the operational criterion for time

reversal symmetry. Thus, only for durations sufficiently shorter than its chaoti-

zation time can a dynamical system exhibit even approximate microreversi-

bility.

UNIVERSAL MICROIRREVERSIBILITY AND INDETERMINISM 141

References

Balescu, R. (1975). Equilibrium and Nonequilibrium Statistical Mechanics. Wiley: New York, NY.

Horton, Jr., C. W., Reichl, L. E., & Szebehely, V. G. (eds.) (1983). Long-Time Prediction in Dynamics. Wiley:

New York, NY.

Lebowitz, J. L. (1983). Microscopic dynamics and macroscopic laws. Jn Horton, Jr., C. W., Reichl, L. E., &

Szebehely, V. G. (eds.) Long-Time Prediction in Dynamics. (pp. 3-19). Wiley: New York, NY.

Misra, B., & Prigogine, I. (1983). Time, probability, and dynamics. Jn Horton, Jr., C. W., Reichl, L. E., &

Szebehely, V. G. (eds.) Long-Time Prediction in Dynamics. (pp. 21-43). Wiley: New York, NY.

Prigogine, I. (1980). From Being to Becoming. Freeman: San Francisco, CA.

Prigogine, I., & Stengers, I. (1984). Order Out of Chaos. Bantam: New York, NY.

Rosen, J. (1995). Symmetry in Science: An Introduction to the General Theory. Springer: New York, NY.

Journal of the Washington Academy of Sciences,

Volume 83, Number 3, Pages 143-160, September 1993

Interpreting the Language

of Informational Sound!

James A. Ballas

Navy Center for Applied Research in Artificial Intelligence

Naval Research Laboratory, Washington, DC

ABSTRACT

Sound offers advantages for information systems, in the delivery of alerts, duration infor-

mation, for encoding of rapidly incoming information, for reaction time enhancement, for

background monitoring, and for representing position in 3-D space around the person. To

assist in utilizing these advantages, background information on auditory capabilities and

design guidelines are available. This paper discusses ways of conveying information using

non-speech audition, including the limitations of present applications of auditory signals,

the basis of these limitations, recent developments in the field including encoding of ur-

gency, presenting 3-D audio and using sounds of real events in computer systems. In order to

conceptualize the use of informational sound, analogies to language are presented and de-

scribed. While these analogies have clear limitations, they provide a useful framework. Spe-

cifically, sounds are used analogously as exclamations, for deictic reference both to place and

to entities, as simile and metaphor, and for symbolic reference. The incorporation of every-

day sounds as symbols for computer processes is examined in detail. Issues in this applica-

tion include the integration of the sound with a concurrent visual stimulus, and the identifia-

bility of the sound. Recent research on causal ambiguity of everyday sounds is presented.

Introduction

Modern computer, aircraft, process control, and C? systems have information

available that cannot be delivered to the operator without careful consideration

to the encoding of this information including the modality for the information

and the formatting and sequencing of information within the modality. The

modality that is of interest here is audition. Auditory signals can be used to

present a variety of information including status of equipment, status of a dy-

namic process, and initiation or completion of events. A primary application is

"Presented at the mid-year meeting of Division 21 of the American Psychological Association, March 5,

1992. Preparation of this paper was supported in part by the Naval Research Laboratory. The author’s

research on sound identification was supported by the Office of Naval Research through the Perceptual

Sciences Program.

143

144 JAMES A. BALLAS

as an alerting signal to direct attention to critical information. Auditory signals

are also used to deliver information through another channel, increasing the

amount of information that can be delivered concurrently. In using auditory

signals, it is helpful to recognize applications in which sound has potential

advantages over visual presentation of information. Applications in which

sound is at an advantage including alerting information (Posner, Nissen, and

Klein, 1976), encoding of rapid incoming information (Posner, 1967), reaction

time enhancement (Colavita, 1974), information monitored in the background,

and information intended to represent position in 3-D space around the opera-

tor. Jenkins (1985) summarizes the benefits of acoustic information over visual

information, particularly in natural settings. The advantages include unobtru-

sive monitoring, no requirement for an external energy source if natural events

are producing the sound, provision of information about the cause of the sound

and its source in space, and interrupt capability because sound does not require

oriented receptors for effective delivery of the information. In order to imple-

ment these applications, guidelines for the use of complex sound in systems

design have been presented by Sorkin (1987) and by Mulligan, McBride, and

Goodman (1985), and general information about audition is available in Haw-

kins and Presson (1986), Scharf and Houtsma (1986), Green (1988) and Hirsh

(1988).

Both Sorkin (1987) and Mulligan, McBride, and Goodman (1985) describe

the characteristics of the auditory channel and guidelines for its usage. Sorkin

(1987) addresses factors that must be considered in establishing the level, pitch,

duration, shape, and temporal pattern of the sound. In addition, he covers the

design of binaural sounds and complex coding for sounds. Mulligan, McBride,

and Goodman (1985) provide algorithms that assist the designer in designing

auditory signals especially in ways to enhance detectability of signals in noise

and to increase loudness without increasing signal level.

Coverage of audition is available in several sources, often from a particular

perspective. Hawkins and Presson (1986) focus on topics related to the capacity

to process auditory information including attention and memory, and factors

that mediate processing capacity such as noise and aging. Green (1988) and

Scharf and Houtsma (1986) cover psychophysical performance in detection and

discrimination of intensity and frequency, sound localization, and perception of

loudness and pitch. Hirsh (1988) organizes his coverage of audition into single

sounds, sound sequences, and speech, covering the important perceptual attrib-

utes of each type of sound. These sources provide extensive and excellent cover-

age of auditory perception.

This paper is an overview of how sound is used in systems to convey informa-

tion. As used here, the term informational sound refers to the broader meaning

INFORMATIONAL SOUND 145

of information rather than the precise meaning from information theory.

Readers interested in the latter approach should consult the early work by Pol-

lack and Ficks (1954), Sumby, Chambliss and Pollack (1958), and Chaney and

Webster (1966). Here the term means any non-speech sound that conveys infor-

mation relevant to the completion of a task, a notion presented by Burrows

(1960). The approach here is to present examples of different types of sound and

explore the benefits and limitations of these sounds. The objective is to docu-

ment the wide range of sound that has been put to use in systems including

recent developments. Broader coverages of sound including speech, music, natu-

ral sound and sounds of modern life are available in Truax (1984) and Schafer

(1977). The focus here is on sound in human-machine systems that typically

interest applied-experimental psychologists.

The need to document sound usage derives in part from the ephemeral nature

of sound. In contrast to visual displays and text, which are routinely printed,

distributed, and saved, sound is transient and must be recorded as it occurs.

Even when recorded, it is not distributed conveniently to others. Thus the usage

of sound is known sometimes only through anecdote. A good example is the

often told story of the expert auto mechanic who can diagnose engine problems.

Although the story has probably been around since the invention of the auto-

mobile, it is only recently that descriptions of the sounds such a mechanic might

use and tape recordings of those sounds for training purposes have been avail-

able (Home Mechanix, 1986). Even with this example, the information is not in

the scientific literature.

Analogies to Linguistic Elements

Analogies are useful devices to present relationships in a meaningful manner.

In order to present a framework for the variety of non-speech sound that is used

in information systems, I have adopted analogies from linguistics. The approach

is certainly not new. Comparisons between speech and music have a long his-

tory, and a similar description for sounds used in computer interfaces was pre-

sented by Gaver (1989), although the terminology I use is somewhat different.

He proposed that the relationship between a sound and its meaning could be

symbolic, metaphorical, or iconic. He defined symbolic relationships as being

arbitrary. Metaphorical relationships rely on similarities across the different

domains. Iconic relationships are based upon physical causation. In this paper,

using analogies to linguistic units, I classify sounds in terms of functionality

rather than on the basis of relationships. However, an important aspect of func-

tionality is the type of relational information that is provided by the sound, and

so there is overlap between the scheme that I use and the one presented by

Gaver. For example, I discuss sounds that function as similes and metaphors,

146 JAMES A. BALLAS

Table 1.—Summary of Linguistic Analogies to Informational Sound

Linguistic Element Informational Sound Example

Exclamation Alerting tone

_Deictic reference to place 3-D audio tone

Deictic reference to an entity Indicator tone

Onomatopoeic Sign Auditory icon

Simile Earcon

Metaphor Audio monitor

Polysemous word Ambiguous sound

and sounds that designate an event by simulating the actual sound of the event.

Two sources that offer extended discussion of the relationships between speech,

music and natural sounds are Handel (1989) and Truax (1984). The analogies I

use are only for illustrative purposes. Analogies at another level are discussed in

Ballas and Howard (1987). A summary of the analogies is presented in Table 1.

The following definitions of linguistic elements are offered: _

Exclamation is a sudden, vehement utterance or outcry.

Deictic reference is a demonstrative, a pointing-out device, such as the pro-

nouns “there,” “here,” “‘this,”’ “those.” Reference can be made to a place, an

entity, or a time. Interpretation of the reference requires knowledge of the

speaker’s position and time.

Simile is a figure of speech in which one thing is likened to another, dissimilar

thing.

Metaphor is a figure of speech in which one thing is likened to another,

dissimilar thing by being spoken of as if it were that thing. Dead metaphors are

figures of speech which have lost their metaphoric function and are used to

denote a particular concept, such as “ship of state,’ ““cold person.”

Polysemy means multiplicity of meanings.

A sign is a unit of language such as a word that means, stands for, designates,

or denotes something to an interpreter.

Onomatopoeic means imitative or echoic.

A symbol is something that stands for or suggests something else by reason of

relationship, association, convention, or accidental but not intentional refer-

ence.

An index is a sign whose specific character is causally dependent on the object

to which it refers but independent of an interpretant.

Sounds as Exclamations

The general alerting tone is a type of exclamation functioning similarly to the

word “‘hey!”’. As exclamation, the alert works well. However, designs that lead to

INFORMATIONAL SOUND 147

high false-alarm rates may have adverse consequences on overall system perfor-

mance (Sorkin and Woods, 1985). An important point about the meaning of

exclamations is that they are uninterpretable out of context. The same applies to

alert tones. Although alerts serve a critical function, there are some problems in

their usage. A disadvantage of simpler types of signals is that they are subject to

noise masking and do not have the spectral complexity or redundancy of speech

that can help to offset the effects of noise masking. Doll and Folds (1986)

reported that some of the signals being used would be hard to discriminate in

high workload and stressful conditions and recommend research on enhancing

the distinctiveness and masking resistance of auditory signals. |

A problem with alerts in commercial aircraft has been the increase in the

number of alerts (Veitengruber, 1978). Ironically, the subsystem that has seen

the greatest growth in alerts is the automatic flight control system (AFCS).

According to Veitengruber, the number of alerts in this subsystem increased at

about twice the rate of any other subsystem between 1965 and 1970. He also

found that pilots agreed unanimously that any further increase in the number of

alerts would be unacceptable. The increasing reliance on alerting signals in

systems that had introduced automation is needed to direct attention to the

status of systems that were not under continuous operator control.

The recent development of techniques to encode urgency may alleviate some

of the problems in implementing alerting tones. Edworthy, Loxley and Dennis

(1991) examined the role of both spectral and temporal parameters in conveying

urgency. They identified nine parameters that contribute to perceived urgency

and showed how selected combinations of these parameters could convey varied

levels of urgency. The parameters include spectral and envelop properties of

sound bursts as well as temporal and melodic patterns across several bursts that

are joined to form an urgency alarm. The important contribution from their

findings is they have shown that urgency need not be encoded by increasing

amplitude, an approach which has several drawbacks. It should be kept in mind

that the encoding they have used provides relative levels of urgency. The signals

may not be identified as alerts, especially the low urgency ones, unless there is

prior exposure and learning.

Sounds for Deictic Reference to Entities

The alerting tone can also be designed as an indicator for a specific problem,

process, or subsystem. In this usage, it provides deictic reference to an entity.

Ideally, the tone should be quickly and easily identifiable. Unfortunately, there

are problems in achieving this goal. Doll and Folds (1986) compared the audi-

tory signals used in a variety of aircraft and found no standardization. They

found also that a relatively large number of signals were being used, making it

148 JAMES A. BALLAS

difficult for the crew to recall the meanings of the messages. These problems

arise not only because guidelines are not being applied, but also because stan-

dard signals do not exist. Design guidelines discuss general principles for audi-

tory signal design, especially the design of tones that vary in loudness or pitch.

Unfortunately the reliance on loudness and pitch changes in designing a signal

limits the application and/or requires substantial training to learn the arbitrary

relationship between the signal and its meaning. Therefore, the guidelines gener-

ally recommend audio signals for simple, short messages, and especially for

alerting signals. Design guidelines generally suggest limiting the sounds to a few

levels that are highly discriminable in one dimension. As a result, the usage of

signals is limited.

The increased usage of tones in systems has resulted in the proliferation of

auditory signals and produced auditory information systems that ironically re-

move the very advantage of the auditory modality in presenting alerts. With just

a few tonal signals, the meaning of the tones can be determined quickly assum-

ing adequate training on the signals. However, with the proliferation of alerting

tones, the advantage of the tone as a specific alert is lost. With proliferation, the

tones become generic alerting signals, unable to generate a unique interpreta-

tion. For example, at Three Mile Island, 150 alarms went off in the first few

seconds of the accident. These alarms were not coded for priority and the opera-

tors could only acknowledge them with a single switch.

Sounds for Deictic Reference to Place

The recent development of techniques to present spatial audio to a user effec-

tively and relatively simply introduces the possibility of putting tones in 3-D

space and using them to function similarly to demonstrative pronouns such as

“here” or “there.” The device that is cited most commonly is the Convolvotron

(Wenzel, Stone, Fisher, and Foster; 1990), so called because it uses mathemati-

cal convolution to filter a sound digitally in the same way that the pinnae do.

This filtering includes the effects of interaural differences in time and amplitude

that accompany changes in the spatial location of a sound source. The effect is a

perception of a sound source originating outside the head even though the sound

is delivered through headphones, which would normally produce source origina-

tion within the head. The convolution is specific to the user’s pinnae shape and

is determined empirically.

Although the Convolvotron is limited in the number of sources that can be

presented, and may not produce some of the subtleties of free-field sound such

as reverberant fields, the sounds of self-movement sound (the sound of clothing

as you walk and turn your head, etc.), and the filtering effects of objects in the

environment, the perception of spatial audio is dramatic. The technique has

INFORMATIONAL SOUND 149

been verified for static source localization especially for source azimuth. Best

results are obtained if the digital filtering is based upon an individualized func-

tion, although Wenzel et al. (1990) suggest that results with a generalized

transfer function are sufficiently accurate.

The Convolvotron will change the stimulus ‘“‘on the fly’ depending on the

listener’s current head position in order to maintain a fixed position in space for

the source of the sound. Sorkin, Wightman, Kistler, and Elvers (1989) examined

the effect of the movement-correlated sound for localization, comparing this

condition to localization with the head position fixed and to localization with

head movement required but the sound is maintained at a fixed position relative

to the head. The last condition employed sound that had cues supporting the

perception of an external source, but the location of the source was fixed relative

to the head. They found that localization in azimuth was best when the sound

cues were coupled to head movement and the source remained fixed to an

external location. There was no difference in the three conditions for elevation

location.

The effectiveness of spatial cues has been demonstrated recently for detection

tasks by Perrott, Sadralodabai, Saberi and Strybel (1991). They presented visual

targets for detection with and without a concurrent sound located at the same

position as the visual stimulus. They found that the positional cue improved

joint detection and identification of the visual stimulus and that the improve-

ment increased as the stimulus became more peripheral and as the number of

distractor stimuli increased. A concurrent sound even provides some improve-

ment when the stimulus appears in the line of sight.

A different approach to produce deictic reference to place was used by Ed-

wards (1989) to communicate the position of the cursor relative to objects on a

computer screen. As the cursor moved from one object to another, the tone

changed in pitch increasing from left to right and from bottom to top. In addi-

tion, the edges of the screen were marked by a distinctive tone. The user could

get a speech message about the currently-picked object with a mouse click.

Edwards found that blind users did not use the pitch level to locate the cursor

but did count the tone changes to locate the cursor.

Sounds as Similes

There are many examples of sounds used as similes in systems. To act as a

simile, the sound must be used as a comparison to some property or parameter,

and the basis of this comparison should not be completely arbitrary.

One of the better known examples of sound simile that is effective is the

portable radiation monitor which is similar to the geiger counter. Contrary to

what you might expect, most modern radiation monitors employ a visual dis-

150 JAMES A. BALLAS

play and even those that add an auditory signal emphasize the use of the visual

display (Tzelgov, Srebro, Henik, and Kushelevsky, 1987). However, Tzelgov et

al. found that the auditory signal was better than the visual display or the dual-

mode system in a search task. In a detection task, there were no differences

between the single modes and no differences between single modes and dual

mode. They interpreted the results in terms of a visual bias effect which directed

the operator’s attention from monitoring other aspects of the task.

Simile is also effective in the design of devices for the visually impaired.

Bindal, Saksena and Singh (1983) developed a sonic weighting balance which

uses Changes in both frequency and amplitude to indicate the scale level and the

point of balance, respectively. In order to weigh an item, the user rotates a dial

producing both an increase in frequency and amplitude. At the point of balance,

there is a sharp drop in amplitude for the current frequency. As the user moves

away from the point of balance, the frequency continues to increase, as well as

the amplitude (from the lower level). Tests with a small group of subjects indi-

cated that most errors with this device were less than 1%.

Sometimes the simile is indirect but still effective. An example is the acoustic

traffic signal developed by Poulsen (1982) for blind pedestrians which delivers

WAIT and WALK signals. The requirements for a pedestrian signal include

good localization, discrimination from other street sounds, being audible to

elderly people with hearing impairment, low annoyance, not disorienting to

guide dogs, attenuated by windows, and reliable performance. These require-

ments led to the development and implementation of two sounds that capture

some aspects of the difference between walking and waiting. The walk signal

uses a shorter sound and repeats it more rapidly (200 ms pulsed square or

sawtooth wave at 880 Hz repeated at 2.5 Hz, with a 200 ms gap between sounds)

compared to the wait signal (400 ms wave at the same frequency repeated at 0.5

Hz, with a 1.6 sec gap between sounds).

The simile can also be very elaborate. Blattner, Sumikawa, and Greenberg

(1989) have developed a system of representing messages to the user through

short musical sequences. They introduce the concept of audio icons, which they

call earcons. These are auditory signals that provide information and feedback

to the computer user about the status and functioning of the computer system.

Blattner et al. restrict the audio cues to tones and tone sequences that change in

pitch and loudness. Their system includes a method of representing the hierar-

chical structure implicit in computer messages. For example, they suggest that

message families such as errors, prompts, system messages, and editing messages

can be signaled by a family sequence followed by a sequence that represents the

specific message within the family. The focus of their approach is clearly on

INFORMATIONAL SOUND 151

musical tones, but they mention the possibility of using natural sounds to repre-

sent events and processes in the computer system.

Effective earcons ought to be based upon commonly understood metaphors

for sound. Walker (1987) investigated the choices of visual metaphors for sound

parameters. In the sound domain he looked at frequency, waveform, amplitude,

and duration. Choices that the subjects were given in the visual domain to

match changes in these auditory parameters included size, shape, pattern, and

vertical or horizontal position. He found consistent support for four matches:

frequency with vertical placement, waveform with pattern, amplitude with size,

and duration with horizontal length. However, consistency of these matches was

related to musical training and age, and lesser so to cultural and environmental

factors.

Sound simile must also consider the effect of psychophysical transformations.

Mappings between aural dimensions and visual dimensions, and between aural

dimensions and data should be founded on the established literature on cross-

modality matching (e.g., Baird and Noma, 1978).

The effectiveness of sounds that rely on simile is still an open issue. A recent

study by Barfield, Rosenberg and Levasseur (1991) found that the addition of

earcons to commands did not improve performance with either command-

based or iconic-based menus. They measured three aspects of performance:

time to complete the command, memorability of icons or commands, and

memorability of the top-level branch for specific menu items. They used tones

for sounds, and modified the pitch of the tone to indicate the menu level for the

executed command or icon. The tones sounded for about one-half second. This

implementation is consistent with the general guidelines proposed by Blattner,

Sumikawa and Greenberg (1989). However, Barfield et al. point out that the

pitch of the tones represented the menu level, not the individual commands,

and thus was not indicative of individual command content. Furthermore, he

also points to the problem discussed by Gaver of designing auditory sounds for

functions that do not have a representative sound. An intriguing example of

sound simile are the sounds used in a children’s drawing program called Kid Pix.

The drawing program provides an assortment of drawing tools to the child, and

there is usually a sound associated with the employment of each tool. For exam-

ple, drawing with the pencil produces a scratching sound. As the pencil width

increases, the scratching sound is increased. Other tools produce visual effects,

and the sounds reflect the visual effect. The visual effect and the sound seem

compatible on an intuitive level. It would be interesting to analyze the visual

effects and the sounds to determine the basis of the compatibility. I expect that

we would be able to explain very little of the apparent compatibility.

152 JAMES A. BALLAS

Sound as Metaphor

A metaphor is an extension of the simile to an identity relationship. Rather

than saying that one thing 1s like another, a metaphor states that the first thing is

the other. Metaphor can also be used as a class inclusion statement (Glucksberg

and Keysar, 1990), but that is not how it is being used here.

The reason that metaphor is introduced here is that there are some examples

of sounds that take simile beyond its normal usage. These sounds convey infor-

mation about a hidden process in a way that supports an identity between the

sound and the process. The result is that the listener 1s provided with informa-

tion about the process itself. The two examples I cite both involve an extension

of hearing beyond the normal audible range. In contrast, when sounds are used

as simile, the listener receives information about a value of a variable or parame-

ter from the level of the sound within a particular dimension. In metaphor, the

way I am using it, the perception 1s focused on the process being represented; in

simile the perception is focused on the acoustic parameters.

Two examples can be cited. The first is the use of Doppler ultrasonic monitor-

ing to diagnose decompression sickness (Butler, Robinson, Fife, and Sutton,

1991). When a diver decompresses too rapidly, bubbles become more prevalent

in the bloodstream and begin to increase in size. Ultrasonic waves are reflected

by these bubbles and the echo is transferred into the audible range. Skilled

listeners can assess not only the increase in the number of bubbles, but also the

increase in the size. The sound delivered to the listener is artificial, but is deter-

mined by the physical properties of the bubbles reflecting the waves. The sound

is present ultrasonically and the equipment brings it into the audible range. The

second example comes from Beizer (1984) who recommends the use of an audio

monitor to assess software performance. The audio monitor is simply a radio

(AM) placed near the computer to pick up electromagnetic emissions. Algo-

rithms running within the machine produce different types of emissions and

thus different types of sound. For example, loops will produce pitched sounds

with higher pitches coming from tighter loops. Computer load will change the

intensity of the sound. Updating the CRT display produces a noticeable change

in the sound.

In both of these examples auditory perception involves an interpretation of

acoustic parameters as is the case in interpreting sounds that act as similes. The

difference is that with these metaphorical sounds the interpreter is listening to

sounds that are causally related to the physical properties in a nonarbitrary

manner. In fact, the relationship between the sound and the phenomena is

determined by the physical processes in the phenomena, not by a scheme of

matching specific data values to specific acoustic parameters.

INFORMATIONAL SOUND 153

Onomatopoeic Sounds as Signs of Events

One way to offset the limitations of tonal signals, but take advantage of the

benefits of an auditory signal, is to use sounds that imitate real sounds of events.

There are many examples of real sounds in computer systems that provide

information about events. These include the sounds made by a disk drive, both

by the movement of the read/write head and the movement of the disk. Any

computer user who has heard a disk “thrashing” (often caused by excessive

swapping of data between memory and disk) will not forget the sound, and the

initial worry that something serious is wrong with the disk. Formatting of a disk

produces a distinctive sound, which can change as the formatting proceeds and

the head is accessing different sectors. Finally, a disk drive read/write failure

produces a distinctive sound if the disk driver software has been programmed to

make several attempts. Experienced users probably come to rely upon the infor-

mation in these sounds to monitor the status of the disk system. But these

sounds and the meaning of them are not typically documented even though they

provide useful information. Two types of information are available through

sound: unseen activity and the status of components. The examples of how

sound indicates the status of a disk drive are all examples of sound revealing the

status of unseen activity.

Real sounds can also be used in system evaluations. An example of how sound

has been used to analyze an accident comes from Air Florida flight 90. The flight

recorder recorded the sound of the throttles being pushed up to full power

seconds before the plane hit the 14th Street bridge in Washington, DC. This

determination was based upon an analysis of the blade passing frequency

(Vance, 1986). The pilots had the engine throttle set at 75% power because of

faulty instrument readings.

Recent research into everyday sound perception has provided insights into

the usage of event-based sounds in systems design. Gaver (1989) developed a

sonic interface that uses sounds called auditory icons in a computer interface.

These sounds have an intuitive basis to their meaning. These are sounds of

everyday events and are used to represent the same event in the computer

system. For example, in a windowed computer display, the user must often

move objects around the display from one window to another. An everyday

sound that suggests this event directly is the sound of a object being dragged

across a surface. Gaver concluded that a prototype interface using auditory

icons was effective in increasing the user’s flexibility to gain information about

the system and in directly engaging the user in the system. Auditory icons may

work better than current signals because they would naturally represent the

intended meaning.

As with simile, good examples of sound usage are found in educational soft-

154 JAMES A. BALLAS

ware. For example, a software program called The Playroom presents the sounds

of a bird chirping, an animal thumping its tail on the floor, clothes rustling, and

the sound of a drawer opening when the child selects the appropriate object on

the computer display. These sounds mimic the actual sounds that would be

produced by the movements of the objects.

However, the implementation of auditory icons would be limited to sounds

that have a direct or metaphorical relationship to the events being represented.

The advantage of auditory icons is that intuitive or natural knowledge is the

basis for the user’s interpretation of the sound. This advantage would exclude

conveying system messages that have no equivalent sound or which do not have

a counterpart event in the everyday world. Gaver (1989) suggests that in these

cases a sound be constructed which seems related to the system message either

by analogy or through metaphor. Although this solution would help, there

would still be system messages that could not be signaled by an auditory icon

because an equivalent, analogous, or metaphorical sound is unavailable.

Sound Ambiguity

An issue that arises in using onomatopoeic sounds is how well the sounds can

be identified, and what conditions influence the accuracy of identification. Un-

fortunately, the meaning of real sounds can be ambiguous even with experi-

enced operators. For example, in the crash of Delta Flight 1141 in Dallas,

August 31, 1988, the pilots heard the sound of an engine compressor stall (which

is like a car backfiring) several times. They interpreted this as an indicator of

engine failure. The cause was a disruption of airflow to the engines, because of

the unusually high pitch of the aircraft. They may have hesitated engaging full

power because of the misdiagnosis of the cause of these sounds.

A series of studies by myself and colleagues have examined factors related to

the identification of single sounds presented in isolation and the identification of

single sounds presented within a sequence of other sounds. The identification of

sounds presented in isolation is related to a number of factors. In a study of 41

brief everyday sounds, Ballas (1993) found that identification time was corre-

lated with causal uncertainty and was also related to ecological frequency and

certain acoustic properties. Furthermore, the degree to which the sound

matched a mental stereotype also affected identification time.

When sounds were placed in sequences of other sounds to assess context

effects, the interpretation of ambiguous sounds was influenced in expected direc-

tions by the context (Ballas and Mullins, 1991). In a signal detection analyses,

we found that the context produced consistent effects on response bias but had

little effect on measures of sensitivity. The magnitude of the response bias was

increased when a free identification paradigm was used, compared to a forced

INFORMATIONAL SOUND 155

choice paradigm, but the direction of the response bias was consistent in both

paradigms. These results support the importance of considering sound ambigu-

ity when using everyday sounds to convey information.

One way to illustrate sound ambiguity is to cluster sounds on the basis of

response similarity. The clustering would reveal alternative identifications that

might occur from a sound. Using data from the identification of 41 brief sounds

(Ballas, 1993), a hierarchical cluster analysis was conducted using an index of

causal similarity calculated from a matrix of overlapping identification re-

sponses. Specifically, the identification responses for the 41 sounds were com-

bined and sorted by similar response and by sound. Altogether, 1795 identifica-

tion responses were sorted into categories of similar events. Restricting the

categories to those that occurred for at least two sounds resulted in a total of 66

categories. A matrix was formed of 66 response categories by the 41 sounds, with

the entries a binary notation of the occurrence of an event category used to

identify a sound. From this matrix, a response similarity matrix (half of a 41 by

41 matrix) was generated by counting the number of event categories that pairs

of sounds had in common. Response similarity was computed as follows:

Si, = 1/(e; + 1)

S;; = response similarity for sound 7 and sound j

é,;, = number of events cited in common for sounds / and j when / is not

equal to 7, number of events cited for a sound when 7 is equal to /.

These data were used as similarity data in a cluster analysis. Both single

linkage and complete linkage solutions produced two large clusters of the

sounds, one composed mostly of impact sounds and the other composed of

water, signaling, and continuous sounds. The complete linkage solution is

shown in Figure |. In both solutions, the first four clusters formed are identical.

The sounds within clusters in both solutions have obvious acoustic similarities.

In order to determine whether ~ octave profiles capture this similarity, the 5

octave spectral values for a sound were treated as vectors, normalized, and the

cosine between every pair of octave vectors computed and correlated with the

distance measure defined above. The correlation was significant because of the

large number of pairs (r = —0.21, p < 0.01, m = 820), but the variability in the

response overlap data accounted for by octave similarity is less than 5%, R* =

0.04. Thus the similarity between sounds in this clustering is only weakly related

to spectral properties. Temporal properties would probably be more important.

Clustering produced groupings of water sounds and impact sounds and re-

vealed identification confusions. Greater confusions would be expected within

clusters that are formed to the left of the figure, where similarities are high (e.g.,

156

Gunshot indoors

Telephone hung up

Cork pop

Tree chop

File drawer closed

Door latched

Gunshot outdoors

Fireworks

Stapler

Automatic rifle

Car backfire

Electric lock

Door closed

Door opened

Jail door closed

Footsteps

Door knock

Hammering

Clog footsteps

Clock ticking

Bugle

Car horn

Foghorn

Bell buoy

Toilet flush

Bacon frying

Water bubbling

Oar rowing

Water drip

Church bell

Touch tone

Doorbell

Telephone ring

Light switch

Boat whistle

Car ignition

Lawn mower

Power saw

Sub dive horn

Sawing

Cigarette lighter

JAMES A. BALLAS

eececes

Maximum Distance Between Clusters

Fig. 1. Complete linkage clustering of identification similarity data for 41 sounds.

tree chop and a corkpop). These types of confusions would be expected. How-

ever, according to the cluster analysis, the sound of hanging up a phone may be

confused with the sound of a gunshot indoors, which would not be expected.

Although this clustering is limited to the set of 41 sounds, it does illustrate the

types of sounds that might be confused.

Combining Functionality

Several of the most recent developments in aural delivery of information

combine the functions described above, especially in combining spatial refer-

ence with another function. For example, the improved detection and identifi-

cation of visual stimuli found by Perrott et al. (1991) is actually achieved by a

combination of an exclamation and a sound providing reference to place. An

example of combining simile and spatial reference was developed by Smith,

Bergeron and Grinstein (1990). Their system maps data parameters to aural

dimensions and provides stereo separation capability, so that the sound can be

INFORMATIONAL SOUND 157

moved horizontally between the speakers, as well as cues for apparent distance.

Moreover, they integrate the aural presentation of information with visual dis-

plays that provide “texture” cues related to the data parameters, and thus use

sound to complement and extend the simile presented visually. This example

raises important questions about how aural information is integrated with con-

current visual information, when both modalities are employing simile or meta-

phor.

Modality Integration

The last issue to be discussed concerns the integration of sound with other

media, particularly visual images. In any type of system except a high-fidelity

simulation in which the aural and visual information is veridical with the true

_ system, there will be a conceptual relationship between the stimulus properties

and the information to be represented. This relationship can be described from

several perspectives. Concept formation is involved, and if the stimulus cues are

highly abstract, then traditional literature on concept formation may be rele-

vant. However, the more typical case is one in which the mappings in the aural

and visual modalities are not arbitrary but based upon effective simile. Even

then, there are several issues involved. Often the information to be displayed is

dynamic and represented by events and objects. Treisman (1986) suggests that

fundamental differences between vision and audition exist in the definition of

objects and events. Visual objects are physical structures; visual events involve

some movement or change in the physical structure. Auditory objects are not

defined as easily. They could be thought of as the source of the sound, the sound

itself, or properties of the sound. Furthermore, the source could be thought of as

either the events or the objects producing the sound.

The difficulty in defining an aural object and the lack of a clear parallel

between the concept of object and event in the two modalities may not arise if

the intention is to simply put a sound out that is redundant with the visual

image. This can be assured only when there is a unequivocal ecological linkage

between the image and the sound. Unfortunately, for many of the events that

occur in using computer systems, there is no ecologically valid sound and/or

image. For example, compilation of code is a process for which no visual image

or sound exists. In these cases, the similes used in the two modalities would have

to be integrated. Treisman (1986) suggests that cross-modal integration occurs

in two ways. First, the representations within each dimension could be trans-

lated into separate unimodal objects which would then be integrated. For exam-

ple, the image could represent the three dimensional structure of an object and

the sound could represent its resonance when struck. The two object representa-

tions could be combined to indicate whether the object is a block of wood or

158 JAMES A. BALLAS

steel. Alternatively, the representations could be integrated prior to the defini-

tion of an object. For example, the height of a bar and the pitch of a tone could

be integrated to indicate relative position, which would then be used to represent

the value of a parameter in a spreadsheet.

There is limited information available on the integration of information from

different modalities that is relevant to the design of informational sound (see

Welch and Warren, 1986 for an overview). Vision may dominate in spatial

orientation and shape and size perception. Audition may dominate in temporal

perceptions such as duration. When a sound is simply added to complement a

visual image, it may function as a label and have the potentially negative effects

that have been found with labeling. For example, Rankin (1963) found that

labeling a set of abstract figures was detrimental to the performance of some

tasks such as drawing the figures from memory and fitting the figures together as

a jigsaw puzzle. Labeling did improve the serial recall of the figures.

Conclusion

Although there are still formidable issues in the application of sound in mod-

ern systems, advances 1n presenting sound have been made in a number of areas.

The development of systematic approaches to encoding urgency may lead to the

development of standardized alerting sounds. The development of 3-D audio

technology that has a firm scientific basis should lead to enhanced methods of

representing events and objects in space in a wide variety of applications. The

addition of sounds to computer interfaces has produced new ways of using

sound and stimulated analyses and research on the symbolic effectiveness of

different types of sound. These developments have broadened the concept of

how sound can convey information. Analogies to linguistic devices help to

convey the variety of informational functions that non-speech sound can sup-

port. There is much more to be learned. The scattered examples of innovative

uses of sound presented in this paper are suggestive of a large variety of interest-

ing, effective ways of using sound. Further documentation of sound usage would

be helpful to both designers and researchers. Part of the documentation should

address how sound is being integrated with other modalities, especially in newer

computer interfaces and virtual reality systems which have flexibility in the

design of audio and video stimuli.

References

Baird, J. C., & Noma, E. (1978). Fundamentals of scaling and psychophysics. New York: Wiley.

Ballas, J. A. (1993). Common factors in the identification of an assortment of brief everyday sounds. Journal

of Experimental Psychology: Human Perception and Performance, 19(2), 250-267.

INFORMATIONAL SOUND 159

Ballas, J. A., & Howard, J. A., Jr. (1987). Interpreting the language of environmental sound. Environment

and Behavior, 19, 91-114.

Ballas, J. A., & Mullins, T. (1991). Effects of context on the identification of everyday sounds. Human

Performance. 4, 199-219.

Barfield, W., Rosenberg, C., & Levasseur, G. (1991). The use of icons, earcons, and commands in the design of

an online hierarchical menu. JEEE Transactions on Professional Communication, 34(2), 101-108.

Beizer, B. (1984). Software system testing and quality assurance. New York: Van Nostrand Reinhold.

Bindal, V. N., Saksena, T. K., & Singh, G. (1983). A novel sonic weighing system for the blind. Applied

Acoustics. 16, 347-354.

Blattner, M. M., Sumikawa, D. A., & Greenberg, R. M. (1989). Earcons and icons: Their structure and

common design principles. Human-Computer Interaction, 4, 11-44.

Burrows, A. A. (1960). Acoustic noise, an informational definition. Human Factors, 3, 163-168.

Butler, B. D., Robinson, R., Fife, C., & Sutton, T. (1991). Doppler detection of decompression bubbles with

computer assisted digitization of ultrasonic signals. Aviation Space and Environmental Medicine, 62, 997-

1004.

Chaney, R. B., & Webster, J. C. (1956). Information in certain multidimensional sounds. Journal of the

Acoustical Society of America, 40, 447-455.

Colavita, B. F. (1974). Human sensory dominance. Perception and Psychophysics, 16, 409-412.

Doll, T. J., & Folds, D. J. (1986). Auditory signals in military aircraft: Ergonomic principles versus practice.

Applied Ergonomics, 17, 257-264.

Edwards, A. D. N. (1989). Soundtrack: An auditory interface for blind users. Human-Computer Interaction, 4,

45-66.

Edworthy, J., Loxley, S. & Dennis, I. (1991). Improved auditory warning design: relationship between warn-

ing sound parameters and perceived urgency. Human Factors, 33, 205-232.

Gaver, W. W. (1989). The sonicfinder: An interface that uses auditory icons. Human-Computer Interaction, 4,

67-94.

Glucksberg, S. & Keysar, B. (1990). Understanding metaphorical comparisons: Beyond similarity. Psychologi-

cal Review, 97, 3-18.

Green, D. M. (1988). Audition: Psychophysics and perception. In R. C. Atkinson, R. J. Herrnstein, G.

Lindzey, & R. D. Luce (Eds.), Steven’s handbook of experimental psychology (pp. 377-408). New York:

Wiley.

Handel, S. (1989). Listening: An introduction to the perception of auditory events. Cambridge, MA: MIT Press.

Hawkins, H. L. & Presson, J. C. (1986). Auditory information processing. In K. R. Boff, L. Kaufman, & J. P.

Thomas (Eds.), Handbook of perception and human performance (chap. 26). New York: Wiley.

Hirsh, I. J. (1988). Auditory perception and speech. In R. C. Atkinson, R. J. Herrnstein, G. Lindzey, & R. D.

Luce (Eds.), Steven’s handbook of experimental psychology (pp. 377-408). New York: Wiley.

Home Mechanix (1986). What’s that noise. Home Mechanix, May, 81-107.

Jenkins, J. (1985). Acoustic information for objects, places and events. In W. Warren & R. Shaw (Eds.),

Persistence and change (pp. 115-138). Hillsdale, NJ: Erlbaum.

Mulligan, B. E., McBride, D. K., & Goodman, L. S. (1987). A design guide for nonspeech auditory displays.

Pensacola, FL: Naval Aerospace Medical Research Laboratory.

Perrott, D. R., Sadralodabai, T., Saberi, K. & Strybel, T. Z. (1991). Aurally aided visual search in the central

visual field: Effects of visual load and visual enhancement of the target. Human Factors, 33, 367-388.

Pollack, I. & Ficks, L. (1954). Information of elementary auditory displays. The Journal of the Acoustical

Society of America. 26, 155-158.

Posner, M. I. (1967). Characteristics of visual and kinesthetic memory codes. Journal of Experimental Psychol-

ogy, 75, 103-107.

Posner, M. I., Nissen, M. J., & Klein, R. M. (1976). Visual dominance: An information-processing account of

its origin and significance. Perception and Psychophysics, 83, 157-171.

Poulsen, T. (1982). Acoustic traffic signal for blind pedestrians. Applied Acoustics. 15, 363-376.

Ranken, H. B. (1963). Language and thinking: Positive and negative effects of naming. Science, 141, 48-50.

Schafer, R. M. (1977). The tuning of the world. New York: Knopf.

Scharf, B. & Houtsma, A. J. M. (1986). Audition II: Pitch, localization, aural distortion, pathology. In K. R.

Boff, L. Kaufman, & J. P. Thomas (Eds.), Handbook of perception and human performance (chap. 26). New

York: Wiley.

Smith, S., Bergeron, R. D., & Grinstein, G. G. (1990). Stereophonic and surface sound generation for explor-

atory data analysis. In Proceedings of CHI 90 human factors in computing systems (pp. 125-132). New

York: ACM.

Sorkin, R. D. (1987). Design of auditory and tactile displays. In G. Salvendy (Ed.), Handbook of human factors

(pp. 549-576). New York: Wiley.

Sorkin, R. D., Wightman, F. L., Kistler, D. S., & Elvers, G. C. (1989). An exploratory study of the use of

movement-correlated cues in an auditory head-up display. Human Factors, 31, 161-166.

160 JAMES A. BALLAS

Sorkin, R. D. & Woods, D. D. (1985). Systems with human monitors: A signal detection analysis. Human-

Computer Interaction, 1, 49-75.

Sumby, W. H., Chambliss, D., & Pollack, I. (1958). Information transmission with elementary auditory

displays. Journal of the Acoustical Society of America. 30, 425-429.

Treisman, A. (1986). Properties, parts, and objects. In K. R. Boff, L. Kaufman, & J. P. Thomas (Eds.),

Handbook of perception and human performance (chap. 35). New York: Wiley.

Truax, B. (1984). Acoustic communication. Norwood, NJ: Ablex.

Tzelgov, J., Srebro, R., Henik, A. & Kushelevsky, A. (1987). Radiation detection by ear and by eye. Human

Factors, 29(1), 87-98.

Vance, J. J. (1986). Blind Trust. New York: Morrow and Co.

Veitengruber, J. E. (1978). Design criteria for aircraft warning, caution and advisory systems. Journa of

Aircraft, 15, 574-581.

Walker, R. (1987). The effects of culture, environment, age, and musical training on choices of visual meta-

phors for sound. Perception and Psychophysics. 42, 491-502.

Welch, R. B., & Warren, D. H. (1986). Intersensory interactions. In K. R. Boff, L. Kaufman, & J. P. Thomas

(Eds.), Handbook of perception and human performance (chap. 25). New York: Wiley.

Wenzel, E. M., Stone, P. K., Fisher, S. S., & Foster, S. H. (1990). A system for three-dimensional acoustic

“visualization” in a virtual environment workstation. In Proceedings of the IEEE Visualization 90 Confer-

ence (pp. 329-337). San Francisco: IEEE.

Journal of the Washington Academy of Sciences,

Volume 83, Number 3, Pages 161-177, September 1993

Displaying Information in Future Cockpits

John M. Reising, Terry J. Emerson, and Kristen K. Liggett

Cockpit Integration Division Wright-Patterson AFB, Ohio

ABSTRACT

Rapid developments in display technologies and artificial intelligence software are dramat-

ically affecting cockpit design. Beginning with the F-18 and continuing through the F-15E,

cockpits have changed from presenting information on mostly electro-mechanical instru-

ments to utilizing electro-optical controls and displays for the presentation of information

required by the pilot. With the advent of large screen flat panel displays, the entire instru-

ment panel may some day consist of one display surface where both pictorial and alphanu-

meric formats will be displayed. It is very possible that some of the information will be

displayed in three dimensions. The future may also bring the virtual cockpit where pilots

have a 360 degree look around capability, and the instrument panel, as it is known today, is

replaced by images of the instruments projected on the pilot’s visor. The purpose of this

paper is to trace the history of cockpit displays, to discuss the potential impact that several

emerging display concepts will have on future cockpits, and to cite the results of several

experiments involving some emerging display concepts.

In order to appreciate how information will be displayed in future aircraft, it is

necessary to understand the type of information that is normally or traditionally

displayed in aircraft cockpits, as well as the past and present methods of display-

ing this information. Performing a very top level task analysis of the pilot’s job in

a single seat, military aircraft will provide us with an understanding of the type

of information to be displayed. Many aircraft have more than one crewmember,

but it is felt that the most difficult job of information portrayal is in single seat

aircraft.

History of Cockpits

The primary role of pilots is to control the aircraft, although in commercial

aircraft this job is often taken over by the flight management system. Control

information includes attitude and power indicators. Even if pilots are not man-

ually controlling the aircraft, they still need to know the flight status; therefore,

performance information is of prime importance. An aircraft in flight is always

161

162 REISING, EMERSON AND LIGGETT

proceeding to a particular destination so navigation information is also crucial

to pilots. Pilots are also concerned with the health of the systems aboard the

aircraft. While the systems are performing normally, system status information

is of prime importance; if problem situations occur, emergency information is

crucial. For military aircraft, additional information is required to keep pilots

abreast of the enemy and the threat that they may pose to them. This knowledge

is called tactical situation information. The challenge for the cockpit designer is

to present all of this information in a manner which does not overload pilots’

information processing capability. In addition, the manner of information pre-

sentation is often affected by the particular technologies (controls and displays)

available at the time. For the purposes of this short historical account, cockpit

technology will be broken into two areas—the electro-mechanical (E-M) and

the electro-optical (E-O).

Electro-mechanical Instrument Provided Information

The use of E-M instruments can be traced to some very early days of flight,

roughly 1920, and a Cathode Ray Tube (CRT) E-O display was flight tested by

United Airlines in 1937. However, the “era” of E-M cockpits is generally ac-

cepted to be from the introduction of vertical tape instruments seen in Figure 1,

(late 1950s), until the Navy F-18, with its multi-purpose CRTs, was introduced

in the early 1980s. E-M instruments are in extensive use today, but are gradually

being replaced by E-O displays. In most cases, the E-M instruments are powered

by electricity and have a display driver that is a combined electrical and me-

chanical device. The driver converts an electrical signal to movement of a dis-

play element, such as a pointer, over the display surface or to movement of the

entire display surface, such as an attitude ball. A small servomechanism is often

employed as the driver. Many E-M instruments are single purpose or dedicated

to the display of a single flight parameter, such as engine RPMs. The tape

displays referred to earlier are dedicated displays, although they use one instru-

ment case to contain several drive mechanisms for several different parameters.

For example, in Figure 1, the vertical tape instrument on the left has individual

drivers for mach number, airspeed, and acceleration, while the tape instrument

on the right displays rate-of-climb, vernier altitude, and gross altitude. Similarly

attitude and navigation instruments display more than one parameter.

Impact of the Electro-Optical Cockpit

There has been a dramatic reduction in the use of single-purpose controls and

displays in modern aircraft crewstations. For example, the Boeing 747-400 (the

latest version of this aircraft) contains 365 lights, gages and switches; the pre-

vious models of the 747 contain 971 of these devices (Avionics, 1988). The

DISPLAYING INFORMATION 163

AIRSPEED RATE OF CLIMB

MACH NUMBER VERNIER ALTITUDE

ACCELERATION GROSS ALTITUDE

¢

mM

ro)

ll

Serres

ae

it > “Taal \

q 5

a

3

ripe

on

zn

COMMAND

MACH KNOTS

110

COMMAND

2992 32300 =

TARGET

& : (39300

ot e J Wes

Fig. 1. Electro-mechanical Flight Displays.

primary cause for this reduction was the substitution of multifunction E-O

displays, which can include many pages of information, for the single purpose

displays. The reason multifunction displays were required was that the spare

cockpit real estate was already very scarce in the earlier cockpit versions, and

with an increase in the number of cockpit monitoring/controlling systems that

needed to be included in the cockpit, a new display design philosophy was

needed. Also, with the entire cockpit real estate filled with single purpose dis-

plays and controls, the time for the pilot’s scan of the displays and controls can

become excessive. The incorporation of E-O displays and controls in the cockpit

can help solve the limited real estate problem with the use of multifunctional

displays. The 747-400 contains six 8-inch color CRTs and three multifunction

control/display units. This same trend has become evident in military cockpits

as well, especially in fighter aircraft. A view of Figure 2, an F-111 E-M cockpit,

and Figure 3, an F-18 cockpit with three multifunction displays, illustrate the

trend from single-purpose displays and controls to their multifunctional coun-

terparts.

When the Navy introduced the F-18 into the inventory, the aircraft had a

164 REISING, EMERSON AND LIGGETT

Fig. 2. F-111 Cockpit.

profound impact on cockpit control/display (C/D) research. With the multi-

functional capability, emphasis in cockpit research shifted into the area of infor-

mation processing, specifically, how to best “‘package”’ display formats and con-

trol menus so that pilots don’t drown in an overflow of data or get lost in the

bowels of a very complicated control logic structure. In addition, there was an

ability to change the way information could be presented since the flexibility of

the CRTs allowed many different variations of the information formats—a

capability not available with E-M instruments.

Early Electro-Optical Display Provided Information

In the early military and civilian E-O cockpits, the format of the information

presented was, by and large, the same as that presented on the E-M instruments.

There are several reasons for this. The first is that the introduction of E-O

displays into the cockpit was a radical departure from previous designs; there-

fore, in order to prevent “‘culture shock’’, the decision was made to emulate the

same picture on the CRT that the pilot was used to seeing on his E-M instru-

DISPLAYING INFORMATION 165

dose

wow, /

@-s e

‘

es

$

Fig. 3. F-18 Cockpit.

ments. Another reason that more advanced display formats were not used 1s that

the symbol generator (the computer that draws the format) was not capable of

drawing more complicated pictures. However, current developments in air-

borne graphics generators will make more detailed or advanced formats possi-

ble. Still another reason why the E-O formats didn’t differ from the E-M instru-

ments is that the advanced formats had not been developed to the stage where

pilots had complete confidence in them. As pilots become more experienced

with E-O displays, cockpit designers can explore their potential by developing

formats that represent a total paradigm shift from replicating the E-M instru-

ments on the CRT. ;

Types of Advanced Technology and their Potential Payoff

Before discussing in detail the computer-generated formats which will appear

in future aircraft, it is necessary to briefly review the hardware and software

166 REISING, EMERSON AND LIGGETT

advances which will affect both the type of information presented and the man-

ner in which it is shown. The most important technological developments are

flat panel displays and artificially intelligent software.

Flat Panel Displays

The term “flat panel’? comes about because these devices occupy a much

smaller relative depth than does a CRT. Another term, “matrix addressed dis-

plays’, is used because of the manner in which the individual picture elements

(pixels) are activated. The pixels are arranged in an X, Y matrix, as are the

electrical connections (drivers) which activate them. When an electrical signal is

sent across a particular row and column, the pixel at the intersection of the row

and column is activated (Haralson, Reising, & Ghrayeb, 1989). When matrix

addressed displays reach maturity, they are expected to offer several advantages

over CRTs. The most publicized advantage is in the area of reliability; the

displays are expected to be an order of magnitude more reliable when compared

to CRTs. While the reliability of airborne CRTs is often in the hundreds of

hours, flat panel displays are expected to be in the thousands.

Matrix addressed displays are different from the CRT in one fundamental

aspect, which is very important for information presentation in the cockpit—

the surface size of the display, through continued development, has the potential

to increase until one display covers the entire instrument panel (300 sq. in.). The

largest CRTs in fighter aircraft have a surface area of approximately 40 sq. in.

(Olson, Arbak, & Jauer, 1991). Therefore, by using flat panels, the canvass on

which to paint the display picture becomes an order of magnitude larger, and the

designer can now contemplate display formats which were not possible with

older technology.

Artificial Intelligence

In recent years, a number of authors have introduced the concept of providing

assistance to the pilot through expertise residing in a combination of conven-

tional and artificially intelligent software. (Small, Lizza, & Zenyuh, 1989). This

software combination has been referred to as an electronic flight engineer, a

pilot’s associate, and an electronic crewmember (EC). Regardless of what it is

called, this concept is importanf to the display of cockpit information since it

will have a major impact on both what is displayed and how the information is

presented. Hereafter we will refer to the concept as the EC.

EC enhancement of pilot’s situation awareness. Situation awareness (SA) can

be defined as the crew’s knowledge of both the internal and external states of the

aircraft, as well as the environment in which it is operating. The internal state of

the aircraft refers to the “health” of its utility systems, such as hydraulics, electri-

DISPLAYING INFORMATION 167

cal, and fuel; and to its mission equipment, such as radar and weapons. In order

for pilots to be aware of systems’ status, the systems have to be monitored. If

pilots have to do this, it can require a significant amount of their time, and it can

also cause boredom since it is not a very challenging task. However, the EC can

do this very efficiently and keep pilots well informed as to the situation by

presenting status information upon pilot request or automatically in the case of

failure—and the EC never gets bored.

The EC, through sensors located in the aircraft’s skin and at the flight control

surfaces, will also have in-depth knowledge of the external states of the aircraft.

Because of the EC’s tremendous processing throughput, it can update its knowl-

edge base orders of magnitude faster than can pilots. For example, the EC can

monitor the systems assigned to it at a 30 Hz rate, something impractical for

_ pilots. It can determine if flight control surfaces have been damaged and decide

if flight control reconfiguration is required to give pilots the best performance

possible. The impact of the damage on mission success can also be displayed to

pilots.

The aircraft’s external environment is especially important since it directly

affects safety (location of terrain or threats) and mission success (weather condi-

tions or target location). The EC’s sensors will play a crucial role in providing a

clear picture of this environment. The EC, through the use of digital data bases

combined with pictorial graphics and stored photographs, will be able to provide

pilots with previews of their waypoints and target areas. In addition, this capabil-

ity will allow pilots to look ahead and preview the run-in to the target long before

they actually arrive in the area. This kind of knowledge is crucial to giving pilots

the SA they need to stay ahead of the mission.

EC fused data. The EC could easily inundate the pilot with data since it has

such a high processing rate. In order to avoid this problem, the EC must com-

bine (fuse) data into higher level information packages. It is at this higher level

that the EC will communicate its information to pilots; the form of the commu-

nication could be verbal through a voice interactive system, or visual through

computer generated, pictorial formats. The form of the communication will

depend on which of the pilot’s processing resources are least loaded at the time,

and the type of information to be conveyed. The fused information is crucial to

SA because it allows pilots to operate at an executive level and deal with only the

most important mission-level decisions. Specific examples of formats which will

display fused information are discussed in the following section.

Advanced Graphic Formats

The formats reviewed in this section are related to the previously discussed

information needed by pilots. Specifically, control and limited performance

168 REISING, EMERSON AND LIGGETT

FLIGHT PATH

MARKER

HEADING

INDICATOR

FOLLOW-ME

AIRCRAFT | ALTIMETER

AIRSPEED ON me

INDICATOR e°%e SS

oo te Pca Cea ; P

e 1756 e r)

@ e e

DME 12.8

on

Fig. 4. Pathway-in-the-Sky Display.

information is provided by the Pathway-in-the-Sky; navigation and tactical situ-

ation information is provided by the Tactical Situation Display; and systems’

status and emergency information is provided by the Crew Alerting and Sys-

tems’ Status Display. An additional feature which has been added to a number

of these display formats is 3-D stereo which will provide the formats with added

realism.

Pathway-in-the-Sky

The heart of the advanced flight display is the Pathway-in-the-Sky (Hoover,

Cronauer, & Shelly, 1985). The Pathway can take the place of both the flight

path angle scale and the flight path marker symbology currently used on head up

displays (HUDs). The Pathway consists of a series of blocks configured to resem-

ble a highway (Figure 4). In addition, a “follow me” aircraft appears at a particu-

lar distance above the path and acts as both a speed and altitude cue when pilots

fly in formation with it. The advantage of the path is that it gives pilots a means

of determining what their 3-dimensional route will be like in the future and how

to maintain their commanded airspeed in a very natural manner, by flying in

formation with the follow me aircraft. Thus pilots can view the path in the

distance and anticipate the turns, climbs, and dives; whereas, today’s HUD

displays only depict the route location at the present time and do not provide

knowledge about future maneuvers.

DISPLAYING INFORMATION 169

Fig. 5. Tactical Situation Display.

Tactical Situation Display (TSD)

The TSD (Figure 5) portrays to pilots the EC’s fused data regarding both

navigation and tactical information. Through the use ofa pictorial format which

chunks the data, the TSD can reduce information overload and aid decision

making. The TSD combines, in a perspective view, the aircraft symbol with

terrain data and threat data. It gives pilots a look at the overall tactical situation

as it is developing before them out to the horizon, e.g., 20 miles away. There is a

pathway extending ahead of the aircraft which shows where the aircraft will go if

pilots allow it to follow the pre-planned path. One of the issues faced in design-

ing a perspective display is the elevation and angle of the viewpoint (McCleary,

Jenks, & Ellis, 1991). The viewpoint chosen for initial perspective displays was |

mile behind and 1000 feet above the pilot’s own aircraft; the lookdown angel