Citation
The potential of the pupillary response in business research

Material Information

Title:
The potential of the pupillary response in business research an investigation of methodology and autonomic contamination
Creator:
Bell, Robert Roy, 1945-
Publication Date:
Copyright Date:
1972
Language:
English
Physical Description:
xi, 112 leaves. : illus. ; 28 cm.

Subjects

Subjects / Keywords:
Acoustic data ( jstor )
College students ( jstor )
Linear regression ( jstor )
Marketing ( jstor )
Mental stimulation ( jstor )
Pupil ( jstor )
Pupil size ( jstor )
Questionnaires ( jstor )
Recordings ( jstor )
Research methods ( jstor )
Dissertations, Academic -- Management and Business Law -- UF ( lcsh )
Economics -- Research -- Methodology ( lcsh )
Management and Business Law thesis Ph. D ( lcsh )
Psychometrics ( lcsh )
Pupil (Eye) ( lcsh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis -- University of Florida.
Bibliography:
Bibliography: leaves 93-100.
Additional Physical Form:
Also available on World Wide Web
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
022836019 ( AlephBibNum )
14124093 ( OCLC )
ADB0929 ( NOTIS )

Downloads

This item has the following downloads:


Full Text












THE POTENTIAL OF THE PUPILLAPY' rSPOISE 11N BUSINIIES rESEARCH:

All JNl/ESTI .ATIOil OF IMETHODOLOGi. A D ALTONiO'll COf'lTiN fl.AT lI











By

FOEEPT PO'r ELL












A DISSEPTATIO'I PPEYEI'TED Tni THE GRAL'DUATE
COrUNCIL OF THE lrIIVF%'.iT' OF FLOPIIDA INI FaPTIAL
FULF ILLIENT OF THE t'EriLlTI. 1ENiF FO" THE DEGREE OF
DOCTi 'P IO F I HO ILO'O PH'


UI;YIVEPSJTi' OF FLOFI I.'
173;




























This dissertation i dedicated tro r. paF rints,
Mlr. and Mrs. Al.ah Po, Lel l










AC.t;rO'L EDGE! EIITS


This dissertation is a product of the advice and encouragement of

many people. The author 'ishes to thank everyone who participated in

the study, expressing special thanks to the members of his supervisory

committee. Dr. William For, chairman, has long been a trusted advisor,

teacher, and friend. Dr. Fo,. first introduced the author to the prc.tlem

of psychological measurement, and his constant encouragement and enthusiasm

for this project have been inspirational.

Dr. 'larvin Shaw has provided e-pert guidance throughout the

author's graduate studies. He has served as a source of inforrmatic,'n and

ideas, and has sympathetically led him through the pra-e of [syc'iological

testing and measurementt.

Dr. Walter Hill has provided some of the mo:t rewarding learning

e.'.periences in the author's graduate program. Dr. Hill's willingness

to listen to and explore the ideas of a betrinning researcher is greatly

appreciated.

Another member of the faculty, although not a member of the author's

special committee, deser/Js social recognition and thanks. Without the

interest shown by Dr. William \W1lrot, forrner chairman of the Managemrent

Department, the author would not have entered the graduate program. The

author would lie to e...press his deepest gratitude for the mania hour'

Dr. Wilmot spent giing good advice and counsel.






withoutt the ind iasistanr:e of Dr. Isrret laracan, of the U.F.

r'epartrmarnt of Ps/chiatr the project would ha'm. beenn much more dif-

fi.:ult. The tel e vision DuD1il1orneter he loairned .was one cf the most

valuable instruments used in the stud,,. 'Fimilar , Dr entryy 'Ki patrick

and Dr. Tirel lhalil, of the UI.F. Department of Industrial and S5steims

EnrginerEirin, haie the author's gratitudt for maliMrg the S/stenws Ergine-

ering/Hunan Performanrce Larorator, available.

SeV.eral professors orr.'vided help in the djEsi,'r phases Of the

e. D rnent. DrF Erich Thomn necl' rd ..10hn 1.allace of arinagqem nt, Drs.

W. W. Dam'son and oe. Har-ison of Ophthaimoloa y, Dr. George Barnard of

Psychiatry, and Dr. Pote ,t Is.ac:.on of Psycholo,: ha.e his than ;l for

their assistance.

Tuo student group' in the Colley. of Business Atmirniitiration, the

l A.E.. Club and Delta Siqra Pi, ha.e his oratituie for providing suh-

jectc. The author would li] e to than 'il Robert L.,irns, in particular,

for the effort he put intc acquire" iin volunteerss .

Mr. Alan Copr'e, f the E StFe,s Engineering Laboratory provided

support abo e and be.,:rd the ca311 of dutE,. DurinQo the data collection

pail'cd of the study, Alan spEnrt a lar'-! portion Cf his j3,5, nights,

and we.eIends noCrli on the er ent. e a trirt. H e 3'JtheCr's deeJDest

grati tude.

Several fellow graduate stuJnets hElped in different rhases of

the stud>. In particular, the author would like to thard .Jim Mc'Gregor.

Stu Erownr, I anrgle i'uen. and Fred IL'tt for their contributions. Also,

Mr. E\an Eldridge and the staff of the Eusi ness Admiristration Computing

Laborator ,, and the staff of the C om.munications Department in the

College of E.u:iness have the author's or'atitude.

=1vz











TABLE OF 0Cr:Tl ITS


ACl'l 'L E GrEIrE:;TS
LIST OF TABLES
LIST OF FIGURES
,GSTPACT
CHAPTER 1: liTRO.DUCTIONC
Data Assessment Problems
The Potential of Pupillometrics
Traditional Pupil lmrietri,: Techniques
Pupillornetrics in Cusires;. Research
Scope of the Problem Area
Sumnma r4

CHAPTEP I]: PREVIOUS APPLICATIONS OF THE PU)PILLOIETPil
TECHiiQUL
Historical Develnoments
Attitude Theory
Pupil Size and Mental Activity
Psychiatri .: Pesearch
Emotional Peactions and Pupil Size
Marletini Research
Oissenting Votes
Summary

CHiAPTER Ill: RESEARCH GOALS A'D HIPOTHEFES
Research Goals
HypotheSe;

CHAPTEP IV: PPOCEDUF.E
E:.perimental De :ig
Data Collection
Data Transfer
Data Peduction


Page

ill
vii




1
1


1



i:.









CHtrrTEP V: RF.5ULT JE:
Analysis of PuLrilldar. F:*.action, 4

CHAPTER 'VI: rETHODi'I (,r.iC .L I",FPOVE', ENTS: THE
FE A I I L IT i rF Ci'Or!PlITEF' COlT li'L b.riD
DATP SAi'PLI'G 72

CHAFTEP VII: IMPLICA1TIICS FOF FUTURE PEHEPCH 22
Atti tude Thecor; '.2
Fese'rch on Stimulus F presentation 3
Field Pesearch 84
Laboratory" F'eseairc-h 25

CIIHA TEP VIil: SUJrtVi.Pr AND CON;CLUl l :rf. 87
Gboal: 87
I;ethoid 7
/rnal.sis 9
Results 89
Cos tr.'E:efi t Analysis 90
Conclus ions 91

E:]VELlC,'P PHr ,9

APPENDICIES

APPE;:JDI 1: PPE- AJlO POET-TE FT C IUEST lillAIPES. 10'

.P.FPE lD I II: E IOG FAPHICP L RILIESTIOT I ;N: IE 107
APPENIDI IIl:PPCCEDUiPE FOP:1 1.19

ElIOGPFHICAL S-.ETCF 112


-, 1 -












2-1 Summar, of Pupill .T.-tric Studies 25
4-1 Affective Patings I Indceprrndent Judges) and
Luminosity Scores of Slides 37
4-2 Order of Stimulus Presentation .1
5-1 Prediction Variables Used 51
5-2 Anal.sis of Variance of Ouestionnaire Results
for 4 Groups of Slides 54
5-3 Last Step of Multiplc Steop'.ite Pegression 541
5-4 Stepuise Linear Pc-gression 3-Second Percent
Change in Pupil Size 6
5-5 Last 'Stp of Stepwise Multiple Regression 5S
5-6 Sterpwise Linear Pegression, g-Second Percent Chanr]t 59
Pupil Size
5-7 Correlati nrs between Puril Chringe and S.tirnull 60
5-5 Summar: b,, Stimulus Grrouprs of i3-5.ond Hear trate
Fegr erss crs 61
5-9 Summary,' b/ Indi'.idual Stimuli: Steoise Linpar
Pegressior.s with 9-Second Heartrate a:s Dependent
Vairiables 62
5-10 '.um7mar, by Stiiiul us Groups: '9- second PFrcent Charge
in Heartrate 63
5-11 Stepi-rise Linear Pegression of .-Sr cond Percent Chang..s
in Hood Pressure 61
5-12 Stepwise Linear Pe qre.s ior, of 9-Ceccond Percent Charng:es
in blord Pressure 65
5-13 StepiJise Linear Regress ions of 9-Second Percent Changes
in S in Potential 6
5-14 Pvsults by Hypothesis 69
6-1 PDP 1 /20 ,,stem with Detanhone Linkagr to IE;I 360/65:
Ccst I'[.L-nefit rnal1,'sic : Benefits 78,
6-2 PDP 11,'?ij S.,steom r i i h Da.taphone Linacge to IE.r 360/65:
CostEenefi t Anral sis: Costs 79


-\ i i-










LIST OF FICUPES


Paw


4-1 Procedure 30

4-2 The E-periniental Environrment 31

4-3 Equipment 33

4-4 Pecordi ng Equipm'ent 34

4-5 The Eperinent ir. Froqress 35











IHt L l.il ; l IPL UI IHE PUFILL.'Ht HLi-l.[J! h. 1Hi HU.iJL.: HLP.L i.HM:
PU INVESTIGATIO:l OF :ETHODOLOGY A'!D AUTONIO'lIC CO;TAMlIIAT ION


.By

Fot.ert Poy Bell

August, 1972


Chairman: William 11. FoA
Major Department: Mianagement and FBusiness Law


The purposes of this dissertation were to replicate findings pre'.iously

reported in the field of pupilloinetrics, develop a linear model e:-pla'ni n

the parameters contributing to charges in pucil size, examine the feasibility

of utilizing computer sampling techniques in the collection of ruoillometric

data, and to catalo, physiological responses as coariates of the pupil. A

problem limiting advanced applications of pupillonmtric technique, ha; been

a condition know,,n as autonomic contamination, a phrase describitln the dilation

of the pupil when e.rposed to ansi.ty-arousing stimuli. A goal of the disser-

tation was the utilization of several physiological indicators to define dif-

ferences tbtween pupil responses under autonomic contamination and certain

other conditions.

The study inrolv'Ed presentation of a set of visual stimuli (slide:)

to a group of 29 subjects, and incorporated a randomized block design.

Stimuli were divided into four types: pleasant, unpleasant, neutral, and

an.iety-arousing. Light intensities rere equated for the 16 slides used in

the study. TIo startle stimull there used as hart of the aniety-arousing






set. Environmental factors such as ro:Im light intensity and scund levels

rcre controlled.

A teleiisiion pupi 1 lmeteC'r was used to monitor changes in pupil size.

Other physiological .ariables measured were heart rate, systolic blood

pressure, and skin potential. A dynograph direct-wiring clhart rUecorder

recorded continuous samples on ea:h of the physiological .ariables. These

data were coni erted from analog (c oritinuous)j forrr to digital (discrete)

form, by 3 Eun.err-PF'yrio U'f-', PO r.rocess control computer, which tool. s-amples

of the dita at one-second inter'vais throughout the study. The computer

controlled the timing and presentation of slides during the experiment.

A linear model of the factors contributing to pupil size changes

Was developed for analytical purposes. tep-wise lineer regressions, the

chi-square t,-st, and analyses of verianc uSare use to test the data.

Other data jnalyes involved time-lags and descripticrns of response imai-

ritudes of the phsiologicai :.ariables under differing types of stimuli.

Results r:iised questions about the usefulness of current punillomet-

ric techniques in business research. Variables in the regression riodels

produced multiple correlations of 0.62 and 0.57, respectively, for 9-second

and ?-second percent changes in pupil size. The corresocndling e. plained

variances of less than Jir1 percent indicate that t'lere is a great deal of

reactivity in the pupil '..hich was not e-plaired by tthe variables considered.

Furthermore. the direction of pupil response to the four types of stimuli

did not follow the patterns which have been previously reported by proncnents

of pupillometric techniques.

It does appear that other fi ologcical variablecs e-hi bit s.me

amount of c:nsisLency in their reaction, to an iet.,-arousing stimuli.









covariate with pupil change to irdicate situations in which autonomic

contamination exists.

Although the results tend to question the use of current purillcmetric

techniques in business applications, therE are certain areas of laho'rtory

research where they may still be useful. A cost/benefit study of the feasi-

bility of using a mini-computer in the gathering of physiological data in

the laboratory was performed. Results indicate that the use of computers

as a laboratory tool can be justified in many circumstances.













CHAPTER 1

INTPODU'LT 10O!


In any organize tion, the processes necessary for oncoina acti. it

and success of the enterprise are carried out by human beings. The

study and practice of mrna .eernt. therefore, has implicit in it 6 n:-d

for understanding people--hou they or. how they thinkl, ind how- the;,

can be most effect .ely utilized for progress to Jrrd organization goals.

As Lier't has said, "of all the tasis of mranagem'.nt, mangingr the human

component is the central and rost important task because all else de-

penJs upon ho.u well it is done" (19i 1, p. 197).

Information on r many different aspects of the hurian variable has been

compiled. One of the most elusive, most hard-to-get-at arens of human

behavior lies in the area of ittitudi nal .and eriotional feelings held tb.

people. iet these are unquestioniFly morning thr most significant determin-

ints of human beh.avn ir. The rese' rch described in this p.per was an .attetiiF.

to further refine a ne,. and promising technique for assessing these types

of variatles.


Data Assess:tint Problems


The internal states of individuals art prob tbly amoninq the most dif-

ficult of phenomena to measure. Man.' attempts haje beer man t d. .'eic,

unbibase d :,ar'sticks cf thI emotional and attitudinal resr.ronses of pecole.

Coin.ventional m asures of these' responses such as r.naEer-and-cenc:il question-

-1-











to differentiate "true" reactions from those containing elements of arti-

fact. Ir, many situations, conventional measures are still relatively

accurate and useful. In some situations, how:e.'er, particularly in industry

and in market research, respondents may have scme reason to conceal their

true attitudes and,in these situations, the traditional approach to measure-

rent may be limited in its usefulness.

Lapiere (1924) was one of the first to report a discrepancy between

hot: indi,' duals, in response to a questionnaire, said they would act, and

how they actually behaved. Cook ard Selltiz (1964) prseeited a riore com-

prehensive analysis of the problem of obtaining honest and 'alid question-

naire reactions from respondents. They noted:


Susceptibility of ov.ert response to distortion--that is,
the possibility of discrepancy between private and overt
response--uould seem to be a functions of three character-
istics of the measurementt] instrument: the e tent to
which its purpose is apparent, the etent to which the
irmrlications of specific responses are clear, and the
e'tont to which responses are subject to conici ous control
(p. 222).


The problems of social desirability respondingg with the socially

"correct" answer: Posenthal, 19E.6), demand characteristics (cues which

convey the goals of the study: Orne. 19'9), and evaluation apprehension

(fear of being negatively e.'aluated: Rosenberg, 1969), have also been sug-

gested as contributors to bias in wost tpes of que. tionnaire measures.

Y'ugman (1964). in a marnVeting research study, surveyed results which show,

how social desirability and demand d characteristics interact to distort

several t)pes of questionnaire data.









Other types of methoidologies used in attempts to assess emotional

and attitudinal responses, such as the projective technique; used by Ic-

Clilland. have pro,.ed to be someI:wh.t useful. Their validity coefficients

are usually: lo,-, houe'ser, perhaps due to the fact that scoring of a sub-

ject's responses is so difficult. The tra'inng required to be ahle to

use pi.jective assessment techniques is long and rigorous, and the scoring

methodology at best preserits mTiany areas where errors of riterpr'tatiorn or

other e.*perimenter effects could bias data. The score ron some t pes of

projecti'.e techniques in addition, is to a lairg degree a function of the

vocabularyy of the respondent Itee lMcClellanrd, 19F9. E.Dth question nn.aire

methods and projc,:ti .'e techniques, therefor- are subject to several sources

of assrssmen-t error.


The Potential of Pupillcmetrics


The dc-.elcopiTment of pupillomietric measures of attitudinral/enmotional

response iHess and Po lt, 19V-1: Hes: 1965) presents the hope of circum',ent-

ing some of the errors associated i tith paper-and-pencil questionnaire types

of measures arid pr'cjecti.e techriquc s. The theory of pupillorretrics is

chased on the hyoC'thesis that the pupils dilate in rFrsp.nse to ple-3surable

or favorable stimuli, and coritract in respr.nte to ne'gati.e or distasteful

ones. Sirice the eye is part of the. auttonomic nrer'ous system, inrdi. iduals

presumrlbly carinot consciously control changres in the pupil If it is
1-
The rcsear.-h on conditi iiin r of the pi ;illairy rerporse his produced
equi'.ocal results S.c'veral investigations ((ason, 1912; ct:rier id Eaber,
193. i. Girdern 1941: Crasilneck and -IcCrarn' e, 1956.) reported cositi.e results
in attempLs to condition the pupil It seemT. that just as man. studies, ho':-
e 'er. 'reportedcrjntrar-y findings (Stickle and Cerishme, 1934; I'edell et al.,
1940: Hilgard. et al .. 1949; i'oung, 1954). Even ,.hen conditioning is achieved.
hlo'.e'.er, the process is quite lro.i, aind appears to be limited to a soecific
type of i rdi. idual.









possible to reesure changes in the pupil, and if the logic rmentionEd above

is valid, then it should be possible to obtain a true or objective measure

of attitudinal responses to a stimulus.

Pupillometric assessment should offer several advant.a,.es over other

t.pes of reaction indicators. The response tc. the stimulus is very fast

--usually beginning in less than a second. As mentioned previously, it is

difficult to fal.e a ouoillorietric response, since pupil changes are autonoric

(i.e., not subject to conscious control). A third postsit le ad\antain may

stem from the bi-directionil changes of the pupil--while most physiolocgi-

cal indicators give only an indication of the size of the response, pupil

changes show both size and direction. Previous 'or'k i'ith phl.y:iological

variables, such as that done by Cooper and Pollac. (19.?) on prevuicial

attitudes and the galvanic s in response (GSP) mi.-ht h\ve deri .ed more bene-

fit from the bi-directional pupillar, response approach.


Traditional Pupillor;etric Techniques


The methodology .employed in pupil llometric e,:perinrents ha- been out-

lined by Hess and Polt (196)') and Hess (1965). Eriefl,, a pupill ometric

e-periment requires some type of stimulus. input device, and some type of

device for recording, pupillary ch.anq. Host pupillcmetric e.,perirments have

involved visual stimulus presentation, and have th.refcre utilized slide

projectors as stimulus input devices. Other types cf stimuli which have

been used include liquids (for taste research), still- and motion-oictures

(for market research), sounds (auditor:, research), and several t,pes usin.










cognitive information processing, physical uiorl or startling csunds as

independent variables. I'ost e:ptc 'imenters use rapid frame cameras to tale

pictures Cf the eve, *and leasu ire changes in pupil size frorr t', pictures.

In the most common type cf pup lClmetric stujdy, the subject is seated

before a rectangiul ar boyv i th a vi ev screen in onf end. In ord er to mrini-

F ize rneasuremenr. erroors resulting r from head mc'. cment the siub.ict't head

is usually nimtrobiilized (placed in a chin irest, sometimes with an elastic

str'.;p holding the forehead aqalinst a bar). EquipFment for prc. jecting slides

into the .I e. screen is situated o utsiide the bt-:. A mirror system a,hereby

a motion picture or rapid frrime camera ca r photograph c anges in the eye

is used. The mirror r is placed below the subject line of Sight, at a

4:." anrle, so that a camerer mounted in the side of the boe can photograph

the ey,,. Since the eye is highly reacti.e to light changes, slides are

usually contr'tolled for ll ght intensity, and photographs is performed using

infra red iqu pment. Ihen developed, the photo,?raphs are Tmeasured by hand,

us-ing calipers or millimeter scales or grids.


Pupillom retiric in Cusiness Peseaich


As will be seen in Chapter II, pupillometrics was first emolc.ed as

a business research r method in the field of .jirl'etina research in the i IO 's.

The potential of pupillometrics in this and othcr bhusirne. fields will be

di-cu:sed belov'.


The Perscnnel Function

She personnel function is in area of niaranerial decision makl:in.j which

may benefit from moore accurate information about attitudes end values held








by potential and present employees. This statement must be qualified

b' the possibility that specific icbs nay riot require that prospective

en.ploy'ees' attitudes/,'alues be evaluated. simply because measures of

thEse variables have beer, shcrn to have little predictive validity--i.e..

have litt'l- or no relation to success or the job (Porter and Lailer, 19C,9).

A question that has riot been ,riswered, hoe'ver, it whether the low validity

coefficients are caused by no relationship between the causal and dependent

variables, or because poor measureirent techniques have produced inaccurate

reading of the variables.

There are sone types of managerial positions where specific attitudes

and values have been shoc.'n to be critical ccmponenrts of success (Le.'inrsor,

1964). Certain positions for e'-ample, require that incunitients be able to

perform effectively ir situations involving suhttantial uncertainty and

high risk. many individual: have to assume responsibility for handling

and committing large sums of corporate furins. Other attitudes and values

which may be important include thc.se concerririn minority groups, corpc.rate

ethics, social respcnasi-bility, or tho-e necessary for ,oriinrg in "c.r'ganic"

forms of org.arn iations .

Although pupillometrics is not today at a stage where attitudes and

values such as those mentioned above could be readily uveasured, it is

this type of difficult ime surement which is conrteriplated for advanced

applications of the technique. It riay be that physiological measures,

%hen coupled with other t.,-pes of .assessment, will provide practitioners

with the type of data needed to evaluate candidates along these dimernsions

of "personality".









Certain ethical questions and-questions of acceptability are raised

when industrial applications of phy: ioloqical measuremTent sys tenis are con-

templated. In organizations where high level job candidates are nornially

conducted through soein tpe of psychiatric or psychological evaluation

procedure, pupillomietrics nay be accepted as a matter of cou-re. In

organizations where the technique m ,'y be deemed undesirable, pupillor.mtr:ic

nay still contribute mTaningfully. One of the hopes of de.elopi n. an

easily useable and accurate method for evaluating attitudes and values is

to use it as a tool for validating other types cf de.'ices which rni he

useful or more acceptable for s.aT t ypes of assessment probleFs.

Another application related to the personnel function lri.-s in the

area of human factors ernginEerring.'orl: design. Pupillometrics has been

shown to be a useful indicator of certain types ot stress and arn.iet,

levels (Hess. 196.b), and of noise levels (Nlunnall., et ai., 19E' ). Pupil

response systeT:s. might pro.'e to be valuablee aids in designinq jobs rnd

work' en.,'ror~nments through tle study of their effects on physic.logical

systems .


Mar.tetina Pesearch


Perhaps pupillometrics is one of the oldest kno'.w.n riarketing tools.

Fess (196L,) noted that Chin ee ijade dealers had emp loyed the technique for

centuries, watchingg their client's, eyes to tell when his interest in the

product ':aS highest, then mailing the s..les pitch. More scientific studies

of the pupil in marketing research r-re conducted t, BSrandt (1i-15), who

used ":cular photogra.hy to measure responses to 3dertising messages.









flore recently, pupil dilation and eye lotion for ouoil track)

have been suggested to be useful in measuring package design effective-

ness ('est, 1962), television commercial interest peaks (Lrugman. 19D a),

and product response (Eusiness T'ec, 1967). Fruaman (l'64b)re,'ie'.ed a

series of studies where pupil response was shoun to be indicative of

interest in products such as sterling silvnrware and greeting cards, and

also presented data supporting inter-subject consistency in pupil response

rankings and sales rank data.

Hess and Polt (1966) performed research on taste stimuli which could

be construed as product preference re.erch, jr.d were able to demonstrate

correlations between pupil size and e'.pressed preferences for certain

drinks. The same study also found that both strong positive and strong

iegati..e aversive taste stimuli dilated tie pupil. This finding lends

credence to the hypothesis presented by some researchers that magnitude of

response, not direction, may be the most important indicdctor of feelings.

Halpern (13S7) presented data on pupil responrss to TV commercial and ,ac -

aginr, and noted that contractions were usually found to be associated with

stimuli which "lac the power to interest or arouse the viewer" (p. 7). Hess

(1963a) reviewed additional successful applications of pupillornetrics to

advertising and packaging research.

Scope of the Prcblem Area


Nethodological Improvements

Methodology is a factor limiting the overall usefulness of iost new

techniques. In pupillomentrics, areas needing improvement are methods of

data collection and analysis. Typically., nhotographa of the eye are taken











continuously throulihout a pupi llIme tr'i- e.peririent. These ,hojtograr.hs

rust be developed. and are then mea ured b, h3nd to determine pupil

diameter onr each photoigrrph. This writer has previously demonstrated.

on the ba.is of a simple study, that a porsibilit., of nmalirn major errors

in data interprettio n e.i.s hen .'perim.enters measure Fupil diameter

b, hand (Eell, 1971). This pro.bl-em of e.,.pectancy error i co;ipoiurided

by the sheer nunmter of lme asur.r.ents to be made. Fori e, nmple a study

involving i subjects, with 1. Stimuli (and 15 control stimuli) sho:.-i

for 10 seconds each, Ijith photographs tia.en it 1-second intervals during,

the e: -erinment, would 'require the devel iopment and riessurmigenlt of t6.0O00

plhotographs. Furtherr, a lag time F. ist: between the time the e..per irf'nt

is conducted and thei tire the e..perli:ernter deter, inEs whether his data

are usable--i .e., whether the cariera and film were operating properly.

One of the mnijor goals of the res erch reported in th;i dissertation

i,.6 the jdeveloprernt of computer sa.ni lirn techniques to circumvent the tedious

and perhaps unre li able mieth.od; of data collection mentionrJd above. The

nmethod used in thi. research involveaja TV Punillorr.ter (a video ciaerai.

c:apble: of continuously recording j picture of the eye and rjmeasurirng pupil

di;aneter) tied into a D.inograph recorder (a de',1ice pro'.idinr charts of pupil

and other iphysiological changes) and a Bunter-Pamo FP-330 process control

hybrid cor'iputer. The comORuter sampled dat5 once peir second, and con-erited

the data from arnlog ( conti nuous forri to di ital (disc-re te form through

the use of an analog-to-diitai con.:ie r. The feasibility, and general

appli cabi 1lt., of this technique for laboratory and ron-l ibor ator, settings

is discussed in the paper.













A major problem associated with pupillor.ietrics at thf present state

of the art is caused by reactions of the autonomic nervous system to cer'-

tain types of stimuli. Instead of being a "clean" indicat:.r of a favorabi le

response to stirruli, pupillary dilation is also causeJ by s:,me ve-ry non-

pleasurable stimuli, such as fear and pain (Hess, 1'?62a) and stroinly dis-

tasteful liquids (Hess and Polt, 1966). Since pupil size changes associ-

ated with most stimuli are relatively smr,all, it is highly possible that

autoriomcrc reactions could oive investigators cornmoletel r-eversif-d data from

what the:, expect. A high dearEe of fear associated with an attitude

object, for e,.ample, coulJ, by causing pupil dilation, indicate a .:ry

favorable- ie.spi ne to t.h stiniulus. To Lchieve a h&L.ite r urnd-rstandinu

of why these responses occur, and how they might be "controlled" (partialled

out of the data analysis), a discussion of the autonomic nervous system

is necessary.


The -utornomic nervous system


The word autcnricmic has been defined as "acting independently) of

volition.' Tlhis definition describes the wor[iings of the autcinomic ner.cous

system ( rnS)--"acti ities are largely involuntary, and we- are usually

unaware of them" (Sternbsch, 1966, p. 14). The AIlS is generally concerned

with the regulation of the visceral system of the body, and attenmits to

maintain "homeostatic equillibrium" in the face of '.arying e.tern-l factor;














aff :tinq the body (Cell horn, 1913, c,. 195). The AIIS can be broken down

into two antagonistic :ut-sySten-, the synr.athti: neroius s ,tem (Sr 4 )

and the parasympjtletic nervous system (P'"lC). The '!JS generally' pro'.'ide

emergency resnrrses., 'while the FPJS attemrrts T. mediate or slo'.' dio'.n autc.-

n.mi.: act.i itie and restore normal metabolism ('ternhb.:h, p. 2").


Autornomic innervation of the pupil


The eye and the pu 1il are "doubly boundd, biing innervated by both

the FP and the .1_.. The dilato r iluscles in the pupil are inrvi': ate.jd b.,

the 514,., while the sphincter muscles, which constrict the pupil. a e con-

trolled b by the FPN. (!ilner, 1970. p. 17 .i. The size of the pupil iC con-

trolled by the antago? risticr intr-ractions of the ta.iO sub--.'Sterrs In

qrner.l, e.-itati.in of the is rrpath-tic system cauI.te contr.icti-n ..f the

dil.itor muscle, ..rhich in turn di lat.i the rin il. whi le parjosynrr pathetic

ex., station causes the sphincter mTi tcle to cons' trict the upil ir;-.ossman,

196? p. 170). Se'.eral investg.atorS telio- e there is an inter.iction
















between the P1IS and the Sr:, which c uses dilation. Adle-r (1'50, p. 1 .),

for e'.ample. bel ie'. s that inhi .iti on of the splhiri:ter and crontraction

of the dilatcor muscle Irt toi tthir r to cause puoDil dilatirc.: in mrin, uhile

thr- ,phincter alorine contr cts thc pupi l .

It appearss that .iliticnr of thi- ey? resulting ? from f-ar. howi'..er,

occur not from a joint rworlinn of the SlI; and PF'il. -ut from! a phySi cloir icl

dominance of the S!'. o'ver the r',;i. That is, in timcs rf high statess cf

fePr or pain, the simp.nthetic S) 'tem "tales o.,.er" control of the eIe i rsaac-

scon, et ii 1971, p. 2;E., I'dler, p. 190). This findinri r ill te th tii- ri:

of the e,.pevrimental desi'ln to be dr-scril.:.ld later in thi. pirr.re., it should

be possit-le, if the finding is v'.lid, t, monitor the 'I.'.. anid PFl' and r'b-

ser:.e '.aribles other thar the rpupil ihe- rtrate, tb.looj DIre.C ur e, slin po-

tential. for fe, mple) to define l ,ei'1: of ': iand PF ". a, ti'.'atin'i, arnd

thereby cbtaji n ins i ghts into the puci ll ,-,y resrpon: e. ;Whe.n tot'h .y-,stem

are noriirally aict-i.e, the firndingii h:.poth.;i :i tze: :' Hzes S. hou11 occur. Whh-en

the 515S domain ter- ho;cv'er, w-e sh'IOJld e.'pect findiinss contrary, t.o those

associated with normala" affectie r.-action: of the pupil. If levr el: (of

.activity of other lutonomic .ari .rt -lees such .s the ones mentiron'd .F.c..,*: can

1e .hoi:n t c.:*orre- po',nr d lith the s.,n.*'athetic dori.inanre, then the fear i rnd

pain re-ponwe, can he partialled O'ut of the pupil ret polie t-b :orcurrn-nt 1

moni tor i n, other 3ri.3i]let .







-13-



Summary


It appears that Fupillonm trics is a fruitful tool for conducting

research on the erntional and attitudinal filings hild ) ieorple. The

technique nma, have apolications in riraogenant .rd narl eating resear h.

IMethodological shortcomings anj autonomic cornt !mina-'tion a re two confounil-

ing variables whichh pr'esent1l limit its usefulness. The gol s of the study

reported in this paper are to impro':e the r,.,thodolooy, of pupi ll om'-tr is

anr reduce tle problem of autonomic ccnt eination.











CIIAPTEP 11

FPEVIOUI'S PPLICATIO.C S OF THE PUPILLIE;-ETRI'C TECHNlJUE



Onre reason some, researchers tend to disreg.ird the rupil response

as a valuable tool is that it has been shown to -e a general indicator

of many different typesof processes t'unnally, et al., 1967). As this

literature review, will show'., pupil change has yeepn used as a dependent

\ariable in attitude research, psychiatric resedarich, m.ari.etinr research,

and gustatory. (taste) and auditory research. Pupil size has beer, shown

to be a general indicator of ph:yical and merntal acti..ity, Fnotiona I

arousal, and interest value of visual stimuli.

Nlotwithstnrdirig the "generality" of the Teas ure, Curj-ilometrics

still appears to be a potentially valuable tool ior tisirinss research,

gi,.en proper controls. This chapter wil re.ier- in greater detail the

investigations cn which have contributed toc our present 'rncv1ledge of pupil-

lonetrics.


Historical f'evelooments


The contention that the pupil of the ePe is a reflector of emotion

has be"n held for somiT.e time. As early, as 1765, Fontana descrii-,ed studies

showirnq that pupil dilation folloe.-d the introduction of fearful or Dain-

ful stimuli, e'ern in the pres Cnce of britaht licht. Bender (1933) photo-









graphed sulbjectS' purils in the presence0 of sirul taneous trig''It 1iqht

,and emrcitonal stimuli and also found a "p-,chic" pupil reac-tion.


Attitude Theory


In 1960. an inrr.estigation t., Hess and Folt uncovered a relation-

ship between pupil changee and the interest .alue cof :'isual stimuli.

In this study, the eperim-rters displayed ,hiotoraonhs of male .rnd

female rpinups a taty, a rn.ther and e baty, and a clands:ape scene to

male a;nd female subjects. Fupil diltrion' was laIrest for mer, when ie' -

ing the female pinup, while e the women sho-wed great'- r ouoil response to

the pictures of the taty, the n'lther 3nd tab ., and the male pinun. This

stud rmaled thie Lte nnini cof a concerted effort on the part r.f less

and his associates to study the eye as an indicator cf er,-ticrnl reaction.

In 19?'., Hess surim ariz.d the results of ni.osit rrior studies of

attitude a'rd pupil size. Some of these in"estigatic.rns, using aversive

stimruli such as Dictures of shares, cross-eyed and crippled children, or

strongly unfa .orat.l e political statements, cause'i [upir i constriction of

over 3 percent change in a.'erag'e diameter. Other negjti.e stimuli, suchi

as sips of unpleasant tasting liquids, also caused constriction.

Hess also suggested, on the basii of prenliriinary data collected

during the 1964 political carmpaigins. that the uCunil response might tbe

useful as an indicatori of attitude change (paoe 5).

Puc1il response w.s; sho-.'n to be caret.le of differ'enntiatina betWeen

heterose-ua s and self-adn itite~ l honrce.uals in another studr I'H,:ss

eI tzer and Shlein, 1935). In this r.ilot e'periiment, subjects were shown










slides picturing males arid females in varying states of undress. Four

out of fii'e homose uIls had larger pupil responses to pictures of males

(some up to :35: larger) while five out of fl'.e heterosexuals had larger

pupil responses to pictures of females. A Droblem exists in this tudy,

howe',er, in that the larger pupil response to male pictures on the Dart

of admitted horiosexuals' mayi be the result o.f a type of erpectrncy effect.

Once subjects hai admi tted homosexual interest prior to pr'esr-ntatioun of

the stimunli. the dilatinns could d be phys biological acti.vation1 caused bK

something other than hiiiher sepul interest in the mTa le pictures--some-

thing sir,ilar to the well-I noun skin flush when one becomes a.-'ar he is

being obse v-l.

In : d':rtstin, ld.r',? (17,?5) CLaluated the puril responrie

as a means of reasur in attitudes toward Negroes and found that affect-

relateA1 responses, while significant during e r ly presenitati cns, diminished

during; repeated trials. He did find a significant difference between reari

responses of anti-'legrc. and egalitarian subjects thus provicdirni Euprp.rt

of Hes:s hpothe'.es The repeated trials inv.esticiation als;c prodi.:ed dis-

heiartening resultS concerning the stability of the pupil response, whichh

was freasured to he around 0.50.

Hess '(19I.b) alio described the peculiar results cf the pupil's

fear response, and discusse,1 invpesitigaticns shuing thai t high g alvanic

skin response (GSP); iee Is ere a concomi .tarnt of urn il di lati on from,

shock ocr fearful stimuli. I When GEP iriFoped out", punil dilation turned

to cons tricticn Ip. ,581.. In another ;rti cle (19F.'h a). he su testedd that

fruitful futuLre research should concern itself with ine, stina tions c.f







-17-


the relationship b-et .i en the pupil and ri.n-visual srnrses and t.et -:een

the pupil response ard c.tlher ph,.siol ,? ical rr.e sures.


Pupil Size ,nd 'lental Actijitry,


Data showing that pupil size 'as directly, related to rentl- acti-

vi t, were presented. t.l Htess arnd Pc lt (19 6.1). Thi s e.- oer i ent, r'here

sutiects ,were pre e -nted i i th in.-rtal matheiT tic.al probc lems of .ar,'i ng

degr.Fs of .di ficult, while their pupils itere phc.togr.rphedi shr d,'.-- that

mean pupil dilation wji3 high] co rrel ated ,.;th r .rrblem difficult;,.

1-Jord at. tracltrlne s 'as Eisa o shc',.-n to tb r'Elated toc pupil size when

:i.ibiectt :s er asled to foin r im nrtal iima.-qes rf ab.st i'ct r an1 conCre e it words

(Pa'.io and i:is'.r o n. 19'66). The FP vio anrl S.i.I'pisoi tudy found no relatii'.n-

hip, ho ~"eter, betr4een the rpl'as ntni.s- -un ple.s jtnesi of a ucr'd ind

punil si e (p. 5SE An. he r s tud,' corrob.-.rated the first FPa io and

ips'orn fi'ndir, g in a3 tud on', pupil si-e nrd rimrioiry 1 rIa finding that

"pupil di airiett-:r s r measure of the amrr.unt of material thich is under

acti',e processing at .an tjie" (.ah erirean andE' r.tt 196r,, r. 15: ).

c'ome studies reported in '167 included one t'b I .,rn arnan ard E.ai le,'

repli c tir.g tihe inf.:r.n l tion load end ior.i l I 5 resiul t; repCi'rte1C in

19 6i; ; rep':rt t'y h'.id -d A nderson that ruDil dia.,ete"r mri., t.e a me,.h.n nis

of cerc;ptual rcc.g nition trrethol d. and a stuj:d ty Fes,.ler and !icLauhlin

on the r.elatic.nship between n io.el tiruli and r.upi si ze, all c.f w~ icli

gesnerall, supported the r.su tr.s dlescrlt cd in the aboVe p 'r 'ar, hs Pr. lt

1i 70) descritd another" .-::perient shoniig that subjects e-rtinrg Ir-ater

mental effort had cocresrpond in.-i / greater pur;il dil-ttion.










Two dissertatioiis written at the Urniversi t of O1 lahema (Clar-l ,

1971. and Johnson, 1971) investigated urunillary resrlo'lses duririg short-

term riemiory tasks. Both replicated the finding th t amount of cooiniti.ve

processing ard rpuill.1ry diameter are highly related.


Psychiatric Pe-earch


A stud, in the field of pr'schiatry (Pubin, et al., 196'.) suoG.ested

thleT pupil response ma, tbe a usrcful ps',chiatric diegnortic tool. Pesults

shlo':.ed that characteristic pupil changes in children with cystic fit.rcsis

v'ere significantly different than those of normal chi ldren. The studa,'

also discussed the effect of age or, oupillary reactivit;. noting that

response characteristics tend to change as people- hecorre older.

PF'search rer.orted K,' Sheflin I1969) inr.estigated uriil reactrions

of hnspitalizEd schiz:rphrenics and found that, if pupil dilation is in-

terpreted as a r.ei ire of interest, male paranoid. are not iIIore -e'uall

interested itr m-n than in worir'en--a findinri cntrar, tr. that held bJ y most

psychiatric researchers.

Eoersna, et al.. found that pupillar., response was useful for

studying differences in cognitive information processing irn -educatable

mental retardates aid normal children i1970, D. 143). Purin] the S3am.n

time, Tanci aid Fot.tins invm tigated the [redictab- lity of the r.ur il

response Iusino the Dream Incident Techninue (DIT), the Edijard Per'sonlr

Preference Schedule (EPPS) and the Pepres.-ic.n-.rensi ti action Scale. They

found the [PIT to tb, fa',irly rni"edicti' e of nur.il rsnronse, while scores

on the othie two sc,;l.es sho'.'ea no rel ati 'r.nshir to runil changes .










In another study, Good and Le'in (197r0) fCiled inr an attencrit

to hi:,' that "sen sit, Iers" would e .ert rore rerceorittual '.oilance. orer-

ational ied as cupil dilation than would "r'nre eser,"'. although there'

Jdd co:.rrol'tora te pre ,ious firidin. n on affect ve arousal nr,d puil 1 dila-

tion (c. '21).

More research shoirin' u puipil chrianes iliay t'e useful in clinical

programs was reported b ennedy (1971 ). This tud/ indicated that

pupii dilation coul1 t.":: used as an indicator of the therapeutic irrpact

of treltrcent on.r chronic alcoholics and ma, tbe useful for preJictinq,

r-c idl'is' (rccuirrence of need for treatrmnert).


Einrtional reactions arnd Pupil ize


In the 19l'6t stud discussed earlier. Hess also reported th.at

dilated pupil. in a phCtcgraph of a youn- 00orian cau'-ed grieter crefereice

for that picture than for orre where the woirin's puoilis uere of normal

size. Hicl.k er al., I'1'6 ) in an attempt to rerilicate Hess' findings,

reported that pupil s -Iz .howed- no rel rtion.hip to e'pre.sed attraction,

.lh-ilt facial angle did have ; ianificant effect (D. 3: ). Sim,.s houe,.er

foundL that nuri l 'i ze in pictures had a ;i nifi cant eff ,' t hern 5o... di f-

ferences w 'ere cont rolled. I' arrived suruiects shoiu d .ret.test ilii t ion to

cir.o:i te- -. oicturel tvi th larigrj pur i il and dilatean least to lit e-se,

r.hotograr.is ih 1 rgq urp ils (19 ", D. *~J. Data r, presented bty Etarr

and Jill is (19la.) also supported the dilati crn-.ittractiorn I,-D t.esii .

Ot'.er studies of thie r-elati onshir, bett e n oip'il siz- arnd trhe

ern'Otions w eIr- rer orted inr 19!'6. in a dirss rotation. Guin.-n demonstrated










that emc.tionrlit, (operationali-zed through slide presentations of

emotional wordsds ) did have significant ir.pa:t on puoil size, and that

a significant interaction between time arnd emotionality occurred.

Earlow (1969 and 1970) reported a positive correlaticrn brtteen

expressed preference for photos and punil dilation, arid found that

aversive stimuli did cause constriction of the pupil. In another study,

he found the puoil response to be in perfect agreement ,with ver:, al

preferences in a stud', of political candidate preferences (.arl.w.

1969b).

The Hess (1965) article also reviewed some studies of the rupil

response as an inde of "motivation". In this study Hess and his as-

sociates d pri ved scme subjects of fond for four or f i e hours. The

mean pupil re.-ponse of this gr-c.u to pictures of food was two. and a half

tiTmei larger than the riean response of a control ,rouc, which had e ten

within an hour prior to t.he epcr;iments. whether or not th- terIn "moti-

vation" is appropriate in this crnte:'t, it does arcear that these data

support the He-s and Polt (1960) contention that r-uoil response is an

ind .e:' of interest .alue of visual stif.uli.

Hess and Polt (1966) presented data showgi n that the puoil ma, be

an indicator of taste sensitivity .nd taste orefF.rfence. Five different.

orange beverages, together w- th a control beverage watere) were Qi .en

to subjects. Pupil response was positively correlated %;ith other rank-

ings of beverage preference. Drjole,, and Lehr (196.7), however. criticized

the methodoloa i emolo;.,ed tv Hcss and Folt, noting that controls iornall,

errlo.,ed in gustatory research were not used. and trhat the results may










have therefore bnern biased. The, also noted that order effects coul:1

have t.i.-sed dat In a rcrily, Hcss arin Pelt (19(7i stated that they

hid irn.esti'ated order of ,resentatior as a source of artifact, but

had found no effects in isual. cifactor.i, auditory. or tactile stiTiu-

lation e'-perir enit .


liar -tinr? Pesearcn


]ri the early 19C0's larp.lar a marketingg research orgrniibTiori.

tbe:a;ai irnterest:-d in :omrii.ercial appl icitions of the HE :s pupil resr,.onsej

imeas.urirn ssteri.. The s.tujdie they conrikicted. usually i n conjunction,

with Hess, were rep rted by lIest (1'.2), 1rugrnr I'(194,) ard by Cnonsor

mai az2ine I196<' F briefly, findings \ere that puP il rrsrponre was r elated

to sales rpatterrns of the products t. sted (watches. r.eeting cards. sil 1.er-

wai r ) rid was also r-?lated to couponr returns or n ad..'ertisemrrtstr testes d.

These dat tended to supr'ort the tlhetsis th'at puil rrespunses anrd atti-

tudinal emotiornl resporise were related, ard led to the hope that fur-

thlr research could develoD a new- and useful marletinr tool.

The Marplan studies also sh hoed that the puril res.orins could be

usrd with mrovingq pictures as uell as still frame photo' A continuous

nmea-urcC~.-nt .of the punil, called the pucil rnterest track was recorded

while subjects watched tele, is ion ori .ercif al E/ n.ronito ,r'r ri,, il

dilatirn and ccstr iction. in ,estijq.:'tors w re able to dcet-rflir, ie where

interest pear s p er'e Yhi hest and i:rere intiere' t had waned. trateic

rpl.acement of the main sell in poirt, it war' hoped. could be achie'.ved










by locating the messae!p dirt-ctly alter the high point on the putiil

interest track:. Puril tra'cl a;lso allo.e-d the- makers of the commercial

to ev.'aluate their oroiduct's capaci t to ILeer, vi ere s 'wItchinq that

channel and that co.mmer! ial 'Sn'orsr.' 1'~ ).

As mentioned in Chapter 1, a further use of rpioililorretrics to

marketing researc.he-rs lies in its ability to ce.'ercc.ome social desire-

ability problems associated with t pical written n or c erlbal question-

ing procedure-s. Pupi change ..'a sho in tb t tir:'liarlan studies to h .--

a mathematical l higher relationsir, to sales than dc ver .l ind 1-ritten

sur.eys althc.uuh the differenc-s tire riot alw.Ays stLutistically iq-

nificant (.ruginan, p. 17)

Halpern described the p.otnti;l usefulr1ness of puni 11o.etric tech-

niqu:e in "bt foi'r:-and-,a:ter" iamar -etrin e:.rE rime-nts. He noted (o. 3201

that, often, oh!eri suti.lc:ts ar- calle d urorj, to iTimaLe "after" rIl- rornnt ,

their re plie are bt iased t.y rino.jlrI-d rni their "befoe" resoorises. The

interv iew and questic.nnai rr techniques conrriFonly usdii hy r,.ari etin j re-

searchers also corne under file as discussed in Cha.nte. r i. Halpe-rn

conducted studies shc,'hing that ptrpi 1 dilation was as "ncod" a resooner,

inid--. as w~ere ques tiionnaiirs for these. tFpes of e'neriimenrts, r .'iJ n rot

possessir, the b.i s and social desirahi li ty dra,,ba.: s inhe-rent in ou-s-

ti nna ire and inte,-rview tr-chnique;. Pull d l ti .r, sas a ain slo..wrn to

be- an indicator of interest in p.c': ar esiqn (n. 321).

A "questicorinir article b-, E1laci-ell, f-: al (190 sur.'eyed. the

state of Dupi ll et ric a.rirraiisal in r, irketir r s research, and noted that

the technique was t.b in, used tefor'e being ,:i 11 uirdei stoi-od. They stated










that there eist both "unfounded optimism" and "ill-timed slepti.:ism"

concerrning the cupilloreter_ ana its acnpilicationr and echoed the c.ill

of m.anr ir .est iator for- more research or, the technique anrd its use-

fulne:s.


Dissenting Votes


V.hile these studies were proceedings publications in the field

*.if ophthalmniolo :y appe.ire. d whirh ha .e s rii ficarce for Pur ill o-etric

r'eAe rch. in par ticulr the ocr' s cf Lc.i er; te"nr and Low:enieldd i19%61

and 196.:) made considerahile cortr'ibution to 'no:,1eji e about Duoil func-

tioninil. sr.d sparc ed contro.r .ers, 'ovJer sone of tieiir fin ings. Their

discussion -of the "r.ipll11 5ry refle. dilationn. defined as an increase

irn pupil :1 r e caused t,' sen-.ry :ir er.-jtiran.l stimuli. or bY spoint.irneous

enritloC S :r thoughts. suppor' ts tre affect-,i il ti rr results oF.tainred

b!, Hess. t 'htlialrioologic and phisiological treatises on Fpu[il con-tric-

tion, ho'.e. er, a lm,-.:t unifornly, di sjairee with tie hess assertion that

conr tiiction is affect-rel.itr-d (Lr.,;enfeld, 19C.i). According to Uood-

mansee fi 965), pupillary. constriction can occur for an,' of three rtasr.ns


1. The liqht refle. , hereb' "'an increase of liaht
flu to tlth: retina results in flrow of efferent
impulses from the cciloiTOtotr nucleus to the sri'in. .-
ter muscle, thus acti.'ely co tr ictir. the rpu il."

2. The near-'.'is ion refle. wrhi-h occurs ,hern tle
e.e's shift fr'.,n fc.cusin, oin distant ier.t to
focusing on an otiect less far aw.;y

:.. The relation of the dilator muscle, cause-' ',' y
de.:rease in roual fllraina close on the. heel- of
a high-arou;, l rcfle. di ation condition (cag. 1.










It is this third phenonrenon which Lownifel d lnd oodr.enree hypnthesire

as a possible cause of the cortrictic.n measured by Hess.

Another type of criticism was contributed by lIunnrllv, et al. 1967,!.

They described a series of e>Deriments in which Dupil size 'as shorn to

be related to muscle tension, sound levels. affect, novelty of pictures,

and fear Ithreat of gunshot). They noted that punil si2e seems to be

directly relateW to fear through autononmic contamination (p 154), as

discussed in Chapter 1. The authors conclude. that puril response was

such a genEr-al ied reaction as tc te i alr..est without value for stud,inri

ccrrpl:. ercirtti .ro al stimuli arnd responses.


Summa r


The reccid presented awo: tends to obscure iA, patterns which

ray hrve ererge.d from the past decade of study. A summer, content analysi:

of these articles reviewed in this chapter is pr .enrted in Table 2-1.

Perhaps the mijcr conclusion that car be jdrein at this stage is that the

pupil does respond to stimuli. whether or not it is possible to accurate,

a:seN~ such things a2 attitude and eiroctional reactions is not yet :lear.

nldy studies have presented evidence cf ability to measure attitude_;

others, just as rigorously controlled, have found no consistent relay tiorn-

ship teteern attitudes/emotionw and the pupil re:r.,orse. Pefore the

question can be resolved, a much more precise rnethodolog,'v for measuring

the liariables cf inrteire .t must be developed, and many of the artifacts

nnir associated wiith thr punil response r.ust be catalog e.d and controlled.

The rest of this study is an attenpr to contribute tO these goals.





-25-
T/ALE 2-1

SLiMi',AP' OF F F ILLOMtETFP C STI.I]ES


Puthor

Eell, 1971
ElU1 cLwx ll, -t al ,1')70
Le.ersmra, et al., 1970
Eradshau. 1267
Bradshaw, 1'S'
Doole/ anrd Lehr, 196'
Fr;.ncic and Leile.,, 1963
Guin ia 19i' .,
Halpern, 1967
Hess, 1965
Hess. 1964a
H..ss, 6 1
H-ss arn- FPlt. 19640

Hes and l'alt, 1506.7
H -;s, er.i F'olt, 1967
Hess. et al.. 19'5S
Hir.s, t al 1967
Holmes, 19 7
i.ahii-rn ari [ tty, 1' C
S1 unniarn, 1964a
'ru gnar, 19A. 4b!
F irmel 19E
Landauer Feal rs. 19C'
Lo. rinfr l 19, 1;.
Wlurrnn ll at Il.. 1967
P3. ir, an; Eirri on_,. n i9.6
F:uL.i nr, 1961
Putin. et al ., 1?6.3
Si nrns, 196-
T rnci: and r,bt.i ns, 1970
A.m'wc,, ansr,.- 1965
Television 'lagazine, 1962
['. iness 0e9, 1t'6


1-





I,
C'


Att t.ude Studies
Mental Activ.ity
A rous~ 1
Wa'ri t PeseTarch
Enotion .tudirs
Interest Va 31i Pe:Fcarch
la-te PFe-3erci
Conrli t i -n' i n
Po itr"/ F'ese-arch


1 nrpi C

10. 11
4

2







1, 1
2, 7 11
6




I, 2, _
4






1. 10
4, 10
2, 10, 1 15





71

7, 11



4
10
10, 11
2', 3, 6, 10, 15, 1r
2, 3, 5
I3 11


5., 1?
1. 10
4
4


Peasuriment TechniqiuL;
'!ethr.l l0 ,'
Age Differ-tnce Pec arch
utcnTrc 'i : E a 1 E n.:
FPS,'.hi .try P'e r h; -h
Fear PQsc-rch
Fhr sical '.':.rl Stu.li s
Personal i t tLudi;e










CH.aPTEP Ill

RESEARCH GOALS ANiD HYPOTHESES



This chapter will define the obLectives of the dissertation

research, and will present the specific hypotheses which were tested.

Although they were tested in the null form, the hypctheses listed here

are in the form of expected results. Chapter IV will describe the

methodology of the e-perimEnt, and the collection and analysis of data.


Research Goals


1. The validation and reolication of earlier worl's sho..ing

th-t the pupil res;por, s differentiall, to various stiriul .


2. The developnrnt of a linear model describing the para-

.leters contribution to pupil chan.F-.


3. E.ramrination of several autonomic responses to favorable,

unfavorable, neutir-l, and aro lety-arousing 'tiniuli with

the objective of eliminating autonomic contamination as

a source of error in pupil response data.


4. Examination of an improved n.ethodc.log. for conducting

puFpillo mtric e>oeriments, including conpruter control

of stimulus presentation. analog to digital ccn'version

of ta. and real-tilme sariDling and measurement of the

puri 1 ary response.






-5-?-


5. Collection anri anari is of daai for testing the fol low-

i nj spec i i c I .I.oro.thest


H:,,'othes es


Purpilarv Feaj:rtion


Pleasa;r. stimuli i 1i1 be accor.irpanied b,' drilation, ruhile neutral

stimuli will cause no, chanqng in pupil i.'.


rNesati.e stimuli wi 1 craus a constricti c.n of the p.ji .


Arn ie-ty-' ou irg stirrull w;il di late the oupil.


Hi hlvy rpleasurr.e .: stiruli will orrduce gr ate r r Juni l di l.t i.on

than ill -lss r.ile sa-lit stinrjli.


Autiiorro:ic Conrt.erinat i on


Majni tude ':xv: r i c r-


Eiod pressure charngie will t- simi lar iri re.-ction c hen e.-nosed

to plei.ia a t stimul ii and whern -e posed1 t-, .a n..ie-ty-ar,outsing stir uli.


Heartr a e changes wnll t.e limi lar i when -e, p.ose to. o n..i ety,-a r olsirng

stimuli and pleasant stiir.uli.


St in poteri tit l changes ll t.- sil ilar in react *n r, %lhen e p.:ose

to .n. iet., -.irousiniq and t.: pleasant stilu1i.









Pescr'nse duration


rupillry. dilationr, to fear-arousing sttimuli inll tale loner to

return to initial base 1 .el (at time of stimulus DO-esent.tion) than

will pupil dilation to pleasant. stlnli, for a give.r lcvel of stimulus

intense ity.


Heartrat e will take longer to return to b.ae line in response to

an..iety-arousirng stirr.uli then to ple..ent stimuli. for the same stimulus

level.


Skin rnotential ill take longer to retur to 'asp line in response

to an.iret.,'-arousir ri stirmuli tran to pleasant t i.uli. for the srame

Stimulus level.


blood pressure will take lrnuger to return to base line in response

to an., iety-arousin stir.iuli than to Dleasant :tirruli, for the samie P ieS el

of stimulus intensity .











CHAPTER IV

IFPOC ED U.E



E>oerimental Design


The stud, ..'as conducted under a randomnized 1c.ck design. Each

ut1bect sE.red as a blocl and received each of the 1I. treac~i rits (sti-

muli), 1i th the order, of F, reseritatiorn cUf tr'EatTl'en.ts random' i ed for E Lh

t.1 cl. The desi,-,n '..111 tbe discussed i n rather detail belw.


Sutbijc^.


T..'enrt,-r inr persons sr-r.vc, as subj.ercts in the ': tudl Ali w re

volurteer- mal-e t-eteien ter he a ie t cf 20 arnd 40l, drawn ar laelly from ti'c.

studert .eroiA. i r the Colle e of Fuiinr,, A.dmi nir ti tion (the '1.. .

Club and Ni elt Sirra i1 ). They i rp ie pre -te ted throw. : h que: t i.rnnriire

to det'ermirne an,,- tr'i.c n r prir" ri-edical rprob'C lem1 heart trouble, ec"

deficienriLies etc. which mriht ha'.'e ffectrcd their resro.nses. Tho i

repoerti no. medical probi-rlrr1 werr .:Led to. ser'.'e a: ut iec t The

accc.impi a, yin fig ur-e Shr.ui the flow oCf the pr 'c. '.durI T the su'b ih- tS 'r:lt

thro nh.

At the t e ii of the s tud,, e..ch :.ublect': '.iual acuity r1as tteted

tc. deterniine his ability to di-i-criiriate dert il on li;de stirruli.

3irice one .o eie:t Cf the re'l earch was to de termirne the ft.Aia lillt., of























In
0.

CC



.-- L
C r
01

U U.
a'


- 4 C'
.c r. sI-

u U
IC 'C
44' C- 4-*


L' 4--
L. C

Lu cl-p C
- O '.







(c'
I
LI rC'
al .:- -~
n >,





O r 4444
0T 4-' +-



4- di 1' IC

ci. I-.a

t5 C 3
0 C -- LI
U C (J

0%' 4-4 L',-" ^



r1>


-'f



L
C a' -





r-: U r--

In C-
C -C














4- 0' In
ra *-



L-i
C- Q.I i
C S













a 3 c











a a-
Li CL *


*"1 C^ I


3's


I'1. C
6- I 4-'
4a' C:

u" *' r.

i4 '2 In

IC n.
LiL- In


L. .
SL >,
LI r, Cr '
C U
C. L1
L *u -
C LV'

C'
.L C,



y aj GI C
no 0 0

LJ 4 r
n .10 o
LlL

C.T 0 -












L L .

'o

Sa.
Ui aOC

> aI .
C' 4-' o
r- >, LJ 4-'





C i3 3C





C


CCE
4*
c



s CLi
*r- 4 C




3u U 4'


0 4- *a
C. -1 rL









I n i
.i




UL
*-r C



C *r-
E "



-- 4-


-s
1-9


I


0' I




LI I '0 4-' d'n
O rf re r I
w L E ,L

Cn I- c 3 i'

Q' C- E 3


0
,g-








Oj
Q:


4J


4-' C
I 0O
4L L-

o .C C
CL

4- a] C'


CU a-
n -n



w- :I r
-- C C0.













Lm
'r. r C


OI C E
L. (. ar5 fc
.- ci -

0 a. c

,0 ('

E CL 4-- a

Ca 1 c c


Er' Cl 0 CT


I 2I



t Q 1 r-
r- 12





C 0 L.

'-diE a
(Jr ;F
-4- r- 0
if. LU II .. J


0




CL
r--






LJ









u:ingr p'Jupill oi etric. tecrhnirl-J s in f ,-Ield r, ,.arcli, tlIe '.._r i errl

:.controls nrern itirind abc. v .' re tlhe onl:; one.s u:d trc, iel- ct .uhject .

it w.as Ecsuined that close controls i.o uld hc. l.ac inn in fill r c' -.' arch,

ucina su ject s ;e lct:le from thi'. Inra: l .. t aul t Dc.aiultionr. .nrd thr r' frre

an, labor atory s -Lu usi ig ri,,rwir pec ific control wc uld not be o.nerjiali-

z2ble to thi' t.,pe of pcop.ul tion.


En. i roirrne nr


The st u idies re :onriduct ,:d i r thl', ./ t ,ri: E[l i ne-eri rl,'Hu.,lan

Perf'coriar, nc- L ;a or'ator-,- at the Un i *.er:ir .' c f F ir.' id Li oht 1-v',' 1

periF pher orund, and teirpera t ure tu 'er- all co tr.ii' ll d. The: sne to.,

persons (Lth- invc- nigt:r .arnd a labor' tor.r, technician) Conduct.'e all

r -sionr of the, e-perir ert. 'ccast ionnalll 1, other persons r .t in orn the

study .s; ob, r.'er:.


C")


Figure -2. T ': E *ei 'T.nital nr ir nr.n trt









Apparatus


Equipment layout follow 1d a modification of the design used by

Hcs. (1)65. p. .)jI. Each subject wa: fitted with electrodes for measur-

ing hrartrate and sk in potential. H:c has cho,'n the measuring eouip-

ment, and the basic scope of the stud,' was described. The subject was

then seated in a dentist's chair, which allowed a certain degree of

frFedom, ihile restricting beod, mo.-ner'ent.

The subject looked into a rectangular viewing bo.' measuring

27 cm. by 32 cm. bt, 60 cm. The bc. was erquipor;i 1.1 th a rear-nroji ction

screen in the end opposite the subject. The walls, top, nnd bottom

of the bot were painted flat black to effectli.el, eliminate all out-

side light. A black cloth covered the boe and the sutiect during

each run of the study. The end of the be. at which there subjW-ct was

seated had a 10 t;, 13 cm. opening which circumscribed the subject's

hePd. A chin rest, used to mninimize head nio.ement during the e.'peri-

ment, via recessed in the open rnd of thr' b...

Ti'o lod,l. E tagraphic Model C50 H Carrosel slide piojectors,

located 25 cm. to the rear of the vie-, screen. projected slides on to

the screen. The size c'f the projected injae was reduced to 4 by 6

cm. to reduce eye search of the stimuli. The projectors i'ere connected

to a BR-330 hbryid computer which controlled the timing and presen-

tation of thf slides.












p-~*~*


Fi gue ,4- I.. Equipr, irnt


DuI'iirIq th e .p rirner t, the l; iject :,c.re a :;t of dictattion head-

phc.nie:. The he.dr.hones, c.nrir, ect t r to t( r'e. ,'e- 'oil erIi t pe rccor'der',

prel- 'nted uni f rc m ihite (randei,) IoisE du.-ir,. the -: tud,'.

Just to the left of the sub:j-. t's l;in of :;.jht, a I'hittaler

rlorie l '0, u TV Fupi 1l Tmeter ITVP) iwas situated to tal e coritrtiniOLu I e.d inr s

of the si:e of the subject's left puidil The T;VP is a lc.;l d-: i rcui t

tele.'i Gr i s y s tern, ic i'cor3ip i .rp ti 'ig a hc, o ri, ta'l Es r.nrin r? techniq'.u; .rnd a

signal 'processor uliCrh measurr- arnI dir.pl:ays dupil d.liarmet-r. The TVP

u.s equipped ,ith a t .le '.'i i cn menit cr a-nd remote control s whi ci "l "l ed

the r periiTrert.:r to adjust focuii Ii so that r,,'nCr-r. setting; could he

found for each subject. The eui FIiieit operi'ted in the near inf ra--ed.









with a low' intensity r.ear infrared source illuminating the e.e. With

the exception of one subject, the light source produced no reported

discomfort in this study.


Figure 2--. Pecordin Equiprment


A Eec.kman Type Rl1 Pyn:graph direct rilng recc rdetr. capable rf

rmeasuring up to sit physiological variables simultaneously, was situated

near the maini equipment stand. Sensors on the subject were connected to

the recorder, r hich disoleaed readings on a continuously nmovirin chart and

simrultaneouily transmitted data to the c:,om'uter system described beloi.

Tne recorder contained a timer unit and e.ernt r.arket which were used to



Source: Whitta ter Corporation, Srace Sciences Di ',ision,
Instruction Manual for Series 300 TV Pupi 1 loeters.


4.I i~










note event chariejs such A: the in r' odur tion of rei stiuIli.


-h~ -C


Figure I-.. Th- E. periment rn Proi're' s


The co'n'puter sysl -ten'i or toreli th.- .peri"'nt n t cn.rtinuiOus Y arnd

tranr forrnid the data from dir-ect asr,.lr re-adout of the rph.ysilo, gical

'va riables to d ioit l form. u ing punch.ied parper tal :- a: 5ra output ri d uT:.

The Jdat c.n paper tape %ere th rn Ir arsfrred to r'-r i.et i tape r, l

analyzed us-,n the 1 'I 'CO*6'.5 cc.puter. hi th the .,*-:pti: on cf dat,

tran.rifer prograffmini n). there .wasi no. h'umran, Intr actc iorn w th the data until

after anal /.is lid b:-n completed. The rn a:urrement techninu~i- u-ed in

the ystt-em i.ere .vtaiidati. usii,.j "l licon" r. p:pornr':r Isuch a jartific ia

c.upils o.f 1rno,.n diameter) before the per irrent ,a: begun. Ch r-ses


*a










were measured in tenths of milli eters on a scale of 1.0-10.0, so

the neasureenet system %as sensitive to changes as small as 1 percent

per second.


Stimuli


Stimuli for the studies reported in this dissertation ,ere a set

of blacl and white 35 mr- slides. A series of 58 slides ;?ps ldveloned

for pretestinr. Of these 5E., 16 were chosen for use in the study. In

the slide selection pretest, 1. raters iud'jed the slides using semantic

differentials d -elonred from the mrnifest anriet scai.le and from

evaluative scales used by Osgood, et al. (19'67). A cop cof the rating

questionnaire is included in ADDendi. I of this oaoer. The slides chosen

for the study were those e'.hibitirn high scores al3on .a specific d1rien-

sion and having lo. v'.riab ll it,. The direns:ions of interest .'er.e plea-

sant-unplesarnt, fear/,'n)iety, nd interest value. The slides chosen

are described in Table 4-1.


Scoring the nues.tionnaires


The set of semanrtic differentials shocrn in Appendi I was u.ed to

e-vluate each slide. The set was mriade rup of 2 sutbsets of scales, or

factors.

The pleasart-unp pleasant fac-tor was corpos'ed of the f,: lloWin-g

semantic differentials: Disiusted-elated
Gloom- h-he erful
Ho-tt ile-friendr l
Unp l eas3snt-r Fleas. nt
Uinfa .ora.l e- fa.orble
rNegati've-posi ti .'.





-37-

Table 4-1

Affect ve Faitinriqs ('linden.rrr..'ent Judgres) d i. Lurri nos it, cores of Slides


s'tiriulus Criiitent


i. Picture, rnarl in g [DC
?. Picture, Defori ed Eat..'
. Pi cture, Ferret (aniiri.i I
4. Picture, Diefor. :-d EaL,;
5. Picture, Err aci.t-d Child
6. Picture [leforC ed [. I .
7 Li rin D rawri';n T r .ia r l -
S. Chiarc. al C letch. Girl
9. Ficture, C hair
0. Charcoal Eletch, [;udP
1. F i.tur Srnilin i Girl
Line Drawinq. Ebl rcl
L. Linri-e bra wi r P-ct nrial
14. Picture, Child n Pain
5. Ficture [I,%iig fri Han
L.. Line Drawi ri 'oir, ii

Controi Slides

C' bla rl
(2 Ela r
C: Llanl
C4 Elanrl.
Cs Ll anl.


Judoed Affect"
Fear Fle3as ntness


2:,.

27.9
17.. n



25.0



i5.7
2-0. 1
19.3
1.3
19.5
:''i,. rI
25..

23.6
20.0


51. 1



32.6




24.E
21 .1
13.5

23.



2:9.9
24. d


Pre.dict J
Ps r, nose
TyFpe
Feiar
U.inpleasant


I.pl a Fle sant
Unpl 1 a t
Ir npleasant
I.'eu .ra ,




Pl eas.ant
PFleasa, t


r;eutra 1
I.i nr. 1 e a ,3 Iit
hipl easa I.it
Neutral


Luminosi ,i .
rr.'f r ai l
'irih tness'
-719
-720
-6E.5

-7i1














-'L
-71''


-719
-T.93


-725


Neutral
Neutral


[,L-utrl 1
Neutral


range PF.;s ibil i 1ie: f 'r Pe.cnrse Cat.- r.riL : Fear: 5-35 u .tr -I 1 0.0 ;
P leas.ntrn : 12, fleutra 1 =." ;: in .th :a e:or.rc ies, e hi h sc'C-re im-r'i I,
relat'.i PFeac ti r.s i .e., Hi h F'-r ard Ulrr]e.s. rtness, r- rect i'. ..'.
' a, ured irith '.-strn Li hit fl-T.er rnd trari f r-d tc. d oilta.l forr-. The
r aci e '". t,;, -72'- cor- c.ric.dr. ar, r.ro. riat.l tc t e ,.rig,:. 600u tc.
*'5 fuoC't.n.ld'le: re'pectt ivc- ...






Since each scale had a rarrng of from 1 to 7, the possible range of

scores was 6-42. If a person had chec-.ed all scales at the e.,treme "plea-

sant" side of the scale, the score vjlue for the factor would be 6 Con-

versel; checking all the scales at the negati.e end w..-ould have produced a

factor score of 42. Checking all scales at the "neutral" midpoint 'alue

would have produced a score of 24. The middle roint appears to be a true

"neutral" position on the pleasant-urncleasarit dimension.

The arii et\-arousin factor was corCposed of the semantic differentials:
Unrafraid-afraid
Nervous-ca Irn
Insecure-secure
Tense- r.elyed
The lowest possible an,ipty factor score for a slide was ind tie

hi hest an. iety factC,'o value as :... Unlike the r easant-unpleasant fact',r,

it appears that the an-iet, scale does not yield a tru'- neutral Fpo, iti on, if

by neutral ire mean lack of anxiety. Father, this scale appears to produce a

step-score, ii th degree of an, i .ty increase iig as the score incrir.ases.

A single ser.antic di fferential interestt 1i -uniritere tingI) as used

to ePaluate slides along the "inteEst value" dimension.

To ensure that the groupings of slides used as independent var'iabll

were indeed different from each other, the medians test i:as cc.omutied for each

d:,ad of grouping, using subjects' que tionnairr- data. The groups in each

dyad were si nificantl.,' different from each other at the .01 level ifor the

pleasant-unplea; ant dyad, "' 131 .a9; for the pleasant-neutral d.ad, '*

13i .49; for the plea ant-neutral d), d, *'=1 .31; for the uri e lcasant-rieutr,'l

d/.d, Y. 1F.-1.9; each d.ad tested m: th 1 degree .f frc.edcni:m

After the 1 stimulus slides ware settled o., Kodak 2'91 G,.elatin

neutral Density Filters were 5ddJd to each slid as neeede,, to de.'elop

closely balanCnd l i' It intensity tes. All sl des and control slide cs used

in the study had the sarm intensity, wi thin limits so that any pupill r:.









resrons; occurring could not tbe construct. as a light refle:,. Table 4-1

alsc presented the range of scrarbjltea liht intensities for the set of

stimuli.

St imr lus Presentation

A hlani control slide waeJ projected or the screen while the suh-

ject was being seated in the chair and adjusted to the equipment. .As

soon as all sensors were reading oroperli, a two-mirnute stud, -imilar to

the main stud, iuas conducted tD fariliarize the subject wi.itl the equiJ'-

ment, rirling, and renerral en.:irrornent he would be in during the study.

At the end of toe riinut-s. he was.allo,.,ed to rpet and was asked ahout his

ccmfort. This time .also allowed the erpei-rimenters to male any necessary

're.dju..tments of the eqluipm ent. The subject was then cautioned about

movinr, his head t..c from the headrest. HP uas: 1: asked.. to contair

suallow.irrn and ,most -yeblinkirn to the tim, when control li des were on

the screen.,' since these types of mrveicnts introduced errors into the

lreasurrenrt Syster. He was inforrrmid that, at sonre tine during the eperi-

ment, he night lhe r a buzzer, and that it vas nart of the e..peiriment.

He Lwas cautioned not to io=e ri'aw from the equipment at the :soundr f the

bu-:er.

Tle subject was then aslied to readjust his head into the chin rest,

the camera focus nas checled, a black cloth was pl1ac:d c' er the subject

and over the equipment, and the main e-.pr.-iner mnt was begun. A bl nr.

slide wa: p'rojectc;d on the :.:reen during, tle period described abor'.

The bianl. lide remained on the scr'cn for a period of eighteen seconds

after thie cor puter had tc-n gi.en the. instruction to begin the e. Ieri-

mient. Once gien that instruction, the comr.uter controlle-d both










slide projectors,. and operated them so that no light flash occurred during

chanrge-ov'er from one slide to the ne t. To accomplish this, the ronoutrer

was prograrmmd to start chanqinm the slide to be projected one half sec-

ond before it instructed the other slide projector to remove the s'lij

currently bteinog projected. E.'ery other lide in each project cor w .as a

black slide (one allowing ri. light to Das through) so that vhile one

projector was shioiing a ttiniili;: r- control slide, the other wa sh iowinil

a bltcl lide. This "flip-filop" process was de.-elcoped to allow the re-

duction of thr. light flash nenticried above to a period of les: than 1/20

sr.cond.

Each stimulus slide 'wa represented for 10 seconds and was fr'ilolced

ty a blanrl crnt1rol sli de priesentcd for 20 seconds It \is hoped that the

20-'econd hlani side would all c't. response variables to return, to riormal

levels t.bfore the ne-:t stirmuiluS uss presr-nted. The crder of stimulus

pre-sentatior w'as randomly '.aried to control for order effect Table 4-2

presents the order of each stimulus presentation

The slide runs too'. 4IO secondss i,2 slides, half for 10 s"-cornids ecch

and half for 20 seconds). At the end of this time, the subjectt ,6as given

a "startle" stimulus conisting of an alarm buzzer soundino. The brz:er

was- attached to the wood bottom of the arm rest used bL the subject, arnd

produced both a startling sound and a simultaneous. startlirig .ibr-tion in

the arm rest. Appr-. irr.ately 20 seconds later, t ihe investigator' pushed

the "reledC'e" Switch on th', dentist chair. causing it to mr.'ve very, rapidly.

doinuard, .r'oducing a second startle resLCon:-e. Because of the raIid down-




-41-

TV>le 4-2
Order of Stimulus Presrent.atic'r,

Sub- Stimulu.: u uniber
ject
1iS 07 01 06 0~ : 11 14 12 16 02 n3 1 15 10 05 04 N:; 17 1I
JR 02 04 15 12 01 0- 16 13 14 03 08 06 05 11 10 01 17 15
E, 04 i0 02 09 11 15 07 06 16 10 13 03 15 12 14 01 17 18
.H 11 03 05 16 01 07 14 o0 09 0S 10 04 15 1I 023 13 17 1-
TH 16 09 02' 05 13 10 15 03 08 12 01 04 O' i0 14 11 17 1:
Ofr 16 05 10 15 03 04 12 11 02 14 01 12 07 0'? 08 06 17 15
1S 0i 02 05 10 04 0I 16 06 j03 14 12 13 01 15 0' 11 17 13
E.H 1 05 02 16 12 11 01 0O 10 11 09 15 '3 04 06 07 17 18
JP 01 13 04 03 11 02 iE6 10 12 14 O0 05 09 16 15 07 17 18
EG 06 05 0: 04 13 02 07 12 14 09 11 16 10 15 01 03 17 18
lib 16 13 02 15 11 03 12 04 05 06 01 0' 02 14 10 1il 17 13
FN 14 11 13. 0. 05 03 0' 04 10 15 06 01 09 11 16. 02 1? 1'
E. 10 07 02 13 15 16 09 01 1 OC' 01 14 03 06 05 11 17 10
TD 01 0. 15 13 Oi 16 0; 0 i 0. 11 02 10 14 12 03 01 17 1S:
DP 1i 04 07 l0 n9 16 03 05 13 O. 12 06 10 11 14 02 17 1S
V 10 1 04 15 11 02 02 01 06 05 03 12 13 16 17 18
l)4 07 I6 11 04 13 10 01 03 14 0r: 05 02 09 16 15 12 17 1 i
oD 16 03 06 02 05 12 07 09 14 02 15 01 11 04 10 13 17 1
PS 01 03 04 0' 14 13 10 05 15 02 0.1 16 06 11 11 01 17 18
ET 15 07 01 05 03 12 10 16 01 04 13 019 03 14 11 06 1' 1;
JU 0" 13 09 12 14 0. 06 02 15 11 16 O; I 03 01 04 17 18
P.P 13 16 15 03 07 14 09 06 05 0 12 01 04 10 02 11 17 12
\.. 13 16 15 03 07 14 09 06 05 0. ; 12 01 04 10 C02 11 17 1
Cil 01 05 11 04 09 10 15 C0 03 02 16 06 12 13 1J 07 1: 1I:
IU:' 09' 10 16 0S 04 02 12 15 06 05 03 01 07 04 11 13 17 13
F:L 01 11 05 r0 15 13 04 0? 16 02 07 03 14 10 (6 12 17 12
L I2 05 13 16 02 10 0D 03 01 15 06 04 07 14 11 09 17 12
PE 10 16 13 OE 14 02 12 15 06 05 03 01 07 04 11 09 1' 1;
SE: 11 09 15 04 A0 12 14 16 01 03 06 10 i. 12 05 07 17 15









ward movement, the subject could not maintain his head in the chin rest,

so pun;l data w'as not collected for this second startle response. ThPsn

startle stimuli always occurred at the end of the set of slides, and weir-

always poesentej in the same order.


Data Collection


Method and Pesoonse Variables


One hour was assigned for processing each subject. Upon entering

the Systems Lab, the silubect was introduced to the e'per'imenters and w'as

shon the equipEn.nt. The type of data to be- collected w',.s discussed in

general terms. He was told that there uould be time after the e:,peri-

mernt to .*ie\ the equipui nt in more detail and to see the charted results

of his study Aft.:-r filling out the pretest questionnaire, electrodes

,.ere attached to him and he w'.as seated in the dentist chair. The elec-

trodes were in turn attached to the Dynogra.pn recr.rder. A light-sensisn

pressure clip for measure g iq stol1ic blood pressure was attached to the

subject': right car. The subject as as'ed to place his head in the chin

rest and acclimiate himself to the efpe irr.mntal apparatus.

While the subject w'as aciclinating himself, the e-'perimenters were

adjusting the in;trumentc and focusing the camera on the subject's left

pupil.2 As son as all instruments seemed to be properly functioning,



'An interesting problem occurred when cne subject noted that h-e as
blind in the left e,E. The ca-eira was focused rn the ioght 1 -'e and no
problems occuLrrd as a result of this mir dification.




-4 -


the subj.ie-t twas 11 '..:ed to ,rel a for a minute. He ia then as ed to re-

adjust himself, i.r the 2-minute famirili ariration run was t.epun. After

thi: rur, tne stutiec:t %,'a al 1:wed to rele.' for' about 1 minute. During

this minute, the c i'arusLrlt wear: rotated so that the stir'uluc and conritr'cl

slides sere in pr .sition for the be ilrrinini rof the main E tid.,'. The tape

recorder ia. recy'ced, and all equip',rent '.'as checl.ed for proper func-

tioning The main stud,- fol lc.ed this rest period.


Le.rel of le1 :,ur'enri t

while e slides rere being piresentel, continuous line chart recordings

e,.ee being made on the i,'ynograpli reco,'rder. The Dy:.rogr-aprh had a recordings

cpa;c ity of 6 criannel.. In crder from top to: btttorn, f the reLcrdir n p-per,

the si> .ariables rec:or.'ed ,ere: beat-b,-tbeat heartrte; 60-second he.rt-

r.5 te (e~trap.ljsated each tecrrn ); r.upi l chance: sl.in potential ; blood

pressure; and a light i rten-s i ty r: adCiin for' each slide, t ien by a Wes ton

r'M del 7-1 li iht meter ins tal led insi je the e le irTig t.i:.

The EF-330 computer c :amrpled the corntininuru dat a tt a rate of 1 sail-

ple per second for e;cin of the .ariab le: being recorded, thus pro'.idi n

a digital sample. E'ea t- h ..-beat heartrate w3s nrot saiipled by the coTimpute r'.

A total .,f El ,5.0 .iniples was c.ll ecte, fc.r each subj.. t (512 samples fcor

each of 5 response ,riables ). These date er,'e stored intert ral l, in the

computer until tin- study had beeri completed, nr- were then transferred to

punrchej tape. Later, the dati iere again trarnsferrd this' t ie to ma -

netic: tarn i. al idationi chec:l s ..,'wre made for each data reducti in con-

.'ersion, ensurinri. that error: in tr.ansfer i-ere not maide. Priimar,

data analyvsi w3a done on the discrete data contained on









the nm-inetic tane. The punched tape data and chart data tiere used as

bacd-up data sources. anJ were not subjected to much systematic analysis.


Post-Test Cue:tiornai re


In an attempt to develop convergent ..alidation rf the autonomic

responses (i.e.. evaluate the character of each individual's autonomic

responses from another tyrie of measure) each subject uas as.ed to com-

plete 3 qcuestiornai re ratirin the slides he :iered during the first part

of the e'pe: iiment. COopy of the questionnaire is included in i ,.rlendi..

of the dissertation. This quertionrnire was used a: a major cc'riparison

inde., of affective responses to the slides. The coi relation between this

iide. and the pretest rat iing riade by irndeperndert .judges a.' al so computed.


Debriefino and Eqjuinrmerit F:ecyclinq


Since subjectss iwere members of two student orgariza tions, the pos-

sibility of discussion of the e-periment prior to another subject's parti-

cipation existed. Subject: .'ere aseiJ rnot to discuss the e..per'iment with

anyone, and were told that the irnestigator planned to give a debriefinrg

and suimmiary of the results to the group later in the quarter. Each ut..-

ject was shon the chart output recorded dur-in his r articiration, and

the responses w'ere explained to him.

While the subject w-is l filling out tie post-test questionnaire, the

.e Prinenters were trarsferring the data to punched tape, rear'rangirn

stimuli according to new ranj. nrumbter sequences' resettirng the ccmiruter

and the rest of the equin.nent, and ,reparing for the next subject.








Clata Tra sfer


lihC o'r; utei'r-G I.hr-red de7ita jSir pe .- tra-n; firredi frcrn punched

a i',per t.-,e to ,ir netic tape and th- r to cunched cari ;. Since thi- e.-

[,eri,] 'i nt 'a: a pi lot s tu.-l, n-r, computer :r r'- in r a gr- t arn.:unr, of t in-

W.-5: cori'currd i ri the do.'e .:r-r'r. t r Cf Fil-'T[.r 'IJ and rrnchire ar. iruag q [.r- o.pra srr,,

ftr con','..rrtirn and .1 clidatinri data gathered in c tal (t.r e ') fori onr

the pap:r tar.e to hr-.-,deciiral (base i6) form,- c, tih rniagretic ..:are .:nr

their to. di:ciril (base 1I0) for'i' for arnac ly i a r,i ,re:. ent ti.rn. The;e

conr,'. rsi cr; in'vol d the ilch n I tap e[ t tc-t e or.-rte r t-he IET 1 F 20

ar,.. t:m B.,'E cc u tr rs. Th=j con erir rion froon djiff..rert t:.b : I.

rnadi neceas ar,. bec:u: of t'- uti i ati crn of th'e F.P'-'.0 .c'r.Orirute i and r i .

date- r c:r.r.liir, -,.t- If I-,al tim .a s m;-lii l.i t I"1 L 'i r-m.-rt and

direct a.a'r,-tic tape or-" punch.-d card oultput had h--en fra:ibil for th;s

"tl.o,. their c c.r ers ion pr-,ce:. ,.,u'ld h3a' ie b en 1es s bOthe,-t:r '.

Da tr.anrsfe :.-rd e,.ilucticr. pr'-in ramTil ir ng tool' limic,:t 2 ri.crrtris to

conp '. :t. llo, th.at tht cor,'.ersion padcla3rt have been "det-uajr.d" arnd

.a1 idated, a cor.i et data tranr'fr-r fro 'canir "on- ii ne sann i of 30

.u.bji:ct duij inn1 an e.ri.?r-irirt to decimal da ta ready: fcr sta st itistica

arn ly 1 i can he accomnpl i -hed -in % thian I hc.ur:. I f the '1 stteiT c "Fl) 'C'

hd t*c i-l u'; -d, Cor if a direct ti-i tie-n to thie 3 0'r.E tnro:ti h s.ome t',:

Cf ter.miinr;i l inl a d'. icr- ,r'e u:ed. data co.l d have t.,.r, tranisf, ,red

to a ft'.r r.Iead, for ana'l, i rr..atter of :ieconls fT0roir, the ti, .- the

dite c: llecticn l t :. c:l.c.l ete The id.arnta,:-s ,and dise dv tanta. oe: of such,

t real-tiri *..:ttm, r.ra-ther wl th SoLne co:t -benefit data, Hili be di -

CU -I d i rn a later r- tiC:.n.







-,16-


At each stage of the data con.'ersion process, output data wrre'

compared wiith chart data recorded h:, the Dynogreph recorder to ensur'-

that errors had not b-en made. Several programming errors stemm&ing frori

reversed polarity on the D:nco.gr"phr recorderE*'FP-3 0 irterfacing ineire

detected u.ing this te-chnique. In mri:t caies. these errors r w: re corr-

rectablr- with minor modifications in the con. r' ion prorams.


Data Peduction


Sei.eral forms of data ;rF. ?naijl;'zd. In almost all cases, the

10-second tin:me srpa during whicl each stinrulu w..'~S presented formed the

basic period for analysis. Sincr the EP-3j30 served as both a director

of the ei per-inrnt (chang.in slides at iven tine inter-'al :1 and as a

data recorder, it could not tale sanples during thle second it was chinrg-

ing slides. Co.rsequentl 1, no data sample was talern durin;I the first

second aft r'ia slide change had occurred. This left a 9-s cond aminie

of each .'ari.ble when a stimululidE. sid was r'esrnted, and a 19-second

samplee whien control .lides were presented.

T\:o ba5iC mrealluremenr'j. uere used in drfirni n tries of r-5so.nsc..

To t-st hy.,'potheFes cC n reaionitu-dea the first measure rcc'paed absolute

mrna.imumr during the relevant response period with absolute rn.a..imlums dur-

irng the prior control period. Second, percent change measure ue're in-

corpLorated. Using pupil size as an e;arimol,


SPD PDc
percent chi; ge in puil si;e pip =_ -a c.a
for stimulus on pr-esen'ttion a FD
-_0'3









wlertf PLI = rean Fupil diamr.eter during pre'sontationr
.a a of c ti ulus ind

PD : = mean pupil diameter during control pr"ic.d
cs.a preceedi n i sti u Ilus r, o rn ir'F entatio rn .


Peylre: .i n c:.r ffi .ients I c-re used to test di fferenice:, i n di rc tion c.f

responses, js Cdicussed in later sections.

Two tir.le periods were ral nezed for each stinulu. A 3-sEccnd

tiTe. span ias!. an alyzed because it was felt that the l.ugjerl time periods

ccn'.er tnion 1i1 / used Iight be losing a subs tantial r'r.rt rion of the rele-

v rait Ipup il r'es Cpose du! to a "da iping" of the r'espn;:.rnrs o.r tihe 1c.rn

per iod. Inr this; cat data collected Iii the first 3 seconds of each

stimulus data simple ,were corinpai'ed with dIta :olle:ted in the last 3

seconds pFccCedi rig the s e.rple. The other tin., rieiod in.c1'.ed a 9-ecc nd

analn si., he re data CC ll iected ir the 9-se:cond rie-iod fo'il l i ng a stim;u-

lus jr r'e corparr.-d ith tl data c.: elected in the r? -'.cornd prior tor the

int reduction iof the stiniulus.


Peduc t 1 n iri .,io-r. I e Size


In s'*.er:.1 c ses during the study. one or mnorc, probilesr,, (.CCurrod

in the d ta scni-r.li rg te:hnioues. Several subjects for e.* mple, ad

er.teliri rates which c ..aused murcl oct the data simple to be errorne'.us. In

a fe' other cases, c blod pressure nrd skin potential recording Y,- ste-m

mralfuri..n:ti c'ned durj ing the e-.perimrr.-nt. Causi rl dita cn the:os.e vir'i, bles to

be lost. Consequentl., for ro,. t ,of the anal.":-ss conrduc'teI. the salr le

size i.iw s reduced to .2?. and the nur.iter, of ot-ser,.ati onrs i .iged from'TI 270 to

-17 per ,ariabl:. These will e t noted. in the apiropri ate pla.ces in results

discussed t el C.












CHAPTER V

RESULTS



As discussed in Chapter IV, the e.peririrtnt incorporated a randomized

block design, with 2'i -ubjects serviri as blocks and slide stiruli

seririg as randomized treatments. Most Ocf the analyses discussed in

this chapter utilized all 18 stimuli (16 slides and 2 "startle" stimuli)

as treatments. Case: where the 18 were collapsed into C treatment groups

(pleasant. unpleasant. neutral, and fear-arousing) iill be discussed

scparatel,.


brnal/sis of 'Puoillar; Feactions


A Linear tlodel of Pu ril Pcsronse


In describing the pupil response to i g~.en stimulus cf the typr

used in this disser tation, the folloininrg 1 near rimodel was developed.


PP' = E + 6 .. t + te 't + i '' + + + E + .





where PP = change in cupi 1 size
= regression coefficients
= the initial eciue of the pupil (base level;
= rrrnvircnmr.ental sound cha.ries
: environmental light changes









v, = the orienting refle' natterns of the pupil
*. = color netterns in the visual stirulus
'; = cormple ity of the stimulus
= interest value of the stimulus
pl. e asantnesEunplessantress of the stimulus
s o t liig t intensities in the .'isual s.tirulus
S= C: ramrrblted l ight intensity of the stir'ulu:u
:I 1 endogenr-us subject 'varla.' les
c = error tern


s ca.n be seen. the parameters of this ter,' simplistic model

ermrihasi_: that n rny things contribute to gi'.ven chsinrg in Pupil size.

The model served as a guile for cortitrol' used in the e perirent and for

the de'eloprmernt of tne set of linear predictors used in the statistical

anal., st. Leti. cConitruls anid the utili action of riore accurate predictors

should have caused an increase in the anount of ex,:pained v'ariance in

thr dependent .ariable.

Tre rn::del alsoit enphasi es some firidinigs which are not a51i5ys ,con-

sidered in e.r rerirent.al deMigns and analyses. lost researchers for

e-ample, have not ta.ern ideo3uate account of Wilder's Law of Initial

Val'uies, ,rii ch states that a variable's "...respr'-nse to stirriul tion is a

function :Af tried pre-stimulus ie'v.l..." at which the .aria.ble was oper-

atinri (SterAbach, p. 44). itien usirn percent charge as 5 dependent

variable, an error is introduced rhe-n this la, is nor t colnsid' red. For

e:amp-le, pupil '*ith a diarneter of 7.5 nillimetern has little cnotential

for large dilationr (the familiar .eiirin effect I iile it has a much

higher propensity to constrict in si:e .and return to r.lore nornral









level of operation. A 2 percent dilation, therefore, might be miu.:h

nmcre significant at this level than 10 percent dilatic.n from .3 pupil

which started out at a diaretor of 3 millimeters. Ey using initial

value as a predictor, and incorporating analysis of cvrarirnce, the

m.del can factorr out" of treatment effects that oart caused by initial

values. Sini larily. anx.irety score'. can hbe facto'red outi by cCv'ari -nce

whenr the effects of plea.santness-unpleasar.tress as a a redictor are to

be inresti 3ate. d.

The specific model used tr. test the ;..-er-iriaental date wa s:


PP = + ,' + +' + / + / + ). ++ + + t
S I 2 3 L 5 5 ? 3 .

+ + + + + + ; +
I 1l II I1 2 1 : i 1 ]L I -


+ i i. + 5 .'.
Jo ]


E.4 ;


+ I ...
I' 1


+ '
d/ dl


. + 1 +
1+ 1B IS 1 9 .


+ Y Y + 6 / + f /
2' 2'1. 7: 2: 24; 6


+I c. + + X + ,V +
267 .. 2' t: ; 32 i 3 1 3 1i

4 a 3 i' -+ +



1,here PP. = 3 and 9 second r'ri'cnt change in pupil dirimeter

,-, = 0

r. =: regression coefficients

. ....X were parameters described in Table 5-1

F = an error term.





-51-
Table 5-1
Predicti n Varw blcs UsL-d

Varia ble
Number De criptior, Tyoe
1 3-second pupil percent cnr.-ce dependent
2 9-seconr pupil percent crsn.l dependent
3 initial value moderatc.r
4 slide 1 prCd' ctc.r
5 slide 2 pr'edict.or
6 slide 3 predictor
7 slide 4 predictor
o sli'.: 5 predi tor
9 slide 6 predictcr
10 s idee 7 rp redictor
11 slide predictor
12 slide 9 pr-dictor
13 l ide lu predictor
14 slide 11 predictor
15 slide 11 predic-tor
16 slide 13 predictor
17 slide 14 pre..Iicito
10 slide 15 predi- t.r
19 s ide 1t pr-edictor
20 3-s.,o:,d av.e age-- -lig t i rtensi ty moier- jtor
21 '-secorid average---light i rnteniit ; niodera t'r
i'2 questionnaire score---c.l esratr,-ss predictor
quest on i rre sc,:.re- --interest ,a lu predi ctor
24 interact tion---light initial '.alue pr. dict:,'r
t25 inter.action---irter e.t V'.l. + f ''. pre i.:to:r
26. interaction--- ir, te est '.al. + i'.' e i t
27 interact 'or.---1 light iriten sity ['redic: tor
2 irterectcin----i f.~ predict:or









TablE E-I .orI t.)


Vari.-ble
Lumber ['escri pt ion Tyr.e

29 interection---1-v Interest ,.alu' predictor
30 intera.:ticn---interest value lint predictor
31 interaction---iv fav oredictcr
32 interaction--- '.' interest value predictor
33 i nteracti on--- 1 iiht i nten.i'v predictor
34 inateraction--- iv. 'f; predictor
35 i nter.action--- i v intere- t value predictor
36 questionnaire score---irni et., mFod-erator



Several r'cstr'ictionrs are inherent ir the li-nealr niodel: -ipproach

cdcptcd above. First, th, assu;,, i 'p ion of li ne r'ri t,' and linear effei ct.,

ma:, not be a good appro.li1, n nation of re l i ty fCor ph,sioloc ic, al r'eI .porn.e.

Second. as can be seen in Table 5-1, ana lysis ,ras limit ted tOc the fir'st-

order interactions of th'r mrdel parameters. It doer. appear, ho'-,ever,

that the d-ta fit the par.amnetric requiiem.-nts nrecessar5 fcor the employment

of the statti: c.c l techniiue1 de.criihed below. t' noted earlier, of

rarticiilir concern ,'as the time period btet.,een r ides, since the e' e'l-

mnenter vwi.hej to ne?.ate the impact cf the preiviIout stimulus r' lid: on

the response to the ne t-o.curir'in slide. To test for serial effects,

the Durbin-.'atson te-t for auto-co relation tiree .S.erics in the dati'

ias utilized. Tre Durbin-iat'ion statistitc r ringed from 1.92 to 2 .'0' for

ill of the data used in the analysis' irinlifatring that the 20-second time

period betjee n il de succeeded in el iin itirg the effects of the previous.

re ponse.








Find ingj


A :Spearman r.nl:-c-rder corril tion ('fie:jgel 19-6) ) between the oues-

ti:onirr ire scC.res of i nde-pendent judges and subt.ect's cues tionria ire

s:ore: ,:' '. .orrinute,.l in order to dete,7rin e i the aIreemert t :'. to lide rank:

te tet. r een t u le arn the uu ee ts in the .tIrujy. Tie pleajsantners;-

unple asantn ii;e score i ts the va riable in r.e tii .atei d.

The ub.iects .and piretestt judges tended to aqree -l.,out the emotional

corntert of the sl id-es. ieldinr a correiationi coeif icient ,.f c0..1 (sig-

nificant it th- .[S i', l ). The ranr order of the ;lides in the r,-e-

te-t (indererndt jud-rs grrur, nr.er di offered ty nore than 1 po.iticn

fiorr thi ra.rJ or.:jer of the slide: ir, the post-e.. r rimert.31 (1 ut jie: ts)

nrouL p. Iith o-ire e-cepti or, in the rieu tr Cl :.cat 'Lor the r lides. alo n:s ir,-

tained the same ranl: when cnimpaired by rmdi a3n (nor~noriamietric.) or by the

:era. e: reported in Tatble 5-1. It c.ri t.e inferril frmc-n the hiih degrPe

of inter-rater relitability implied thu t a t,'ce of cr',ss-..alij ticn of

:.timilur tt;, re h. b een .cciomrplishEd.

T. ble ,-2 pre--rit: 3 summary of n41ls. i of i aiarice for the A

*lren r'al qroup.:, of :1 ide: u.ed in the study. The differences here are

alo si'jnifican at the .0IJ le.-el of cornfid-rnce. As .ill t. sr'een later.

how-ever, irid;.. idual slide differences, when jud:e, d ti the deperndenrt ari-

tlble;. do not corfrrm to tho result, hypothes.i:;d or to the re-ult: re-

ported by proponent: of pupillomrt'ic techniqlue-.









Table 5-2

Analysis of Variance of Ouestiornnaire

Pesults for 4 Groups of Slides



Sum of Squares D F llean Square F F:atio

Between Groups 1130.27J2 3 3'6. 7"E.:1 3.3556

Within Grcoups 1lS6:33.04F 1F6 112.2774

Total 197 8.3203 1E9




Tables 5-3 arnd 5-4 present tsumrner, statistics for stepu1ise re res-

sion lth "-second pupil chneri as the de-pendert variable. As car be

seen, the rmoJel has a multiple correltionr coefficient of 0.61, eoplain-

ing 37 percent of t'he variatbility in the dependent :arible Ar analy:1i

of the correlation coefficients associated with the independent and rioder-

ator variables yields rcn consistent relationship betu:rn the 4 stiraulus

types. 1'hEn anal/;ed separately, the startle stimulus was the only one

causing a consistent dilation in the pucil. It doe' aopear that jnalety-

arousing stim-ul do cause significantly greater dilitioris than do Other

tpPs of stimuli. :;o consistent findiniqg were associated with pleasant.

neutral, or unpleasant -tinuli, as the titles shjo .

The 3-seconj time period, used inr the present analysis, may cause

part of the confusinri results obtained. As toodmarnsee (1965) has noted,

the oriierin i and 'lear-'.'ision refle:es of the pupil cause a constriction

in size. It ma; be that subiects allo ed their eyes to lose focus during








Ta.le 5-3

Last Step of multiple Step-'lise F:egression

Multiple F 0.6002

Std. Error of Est. 10.1773


Analysis of Varicnrce

F'eoressi c'n
PF sidual


Sum of 'qu.res

16665. .'3
23127.073


[IHean Squarc F Patic

793.612 6.9 1
111.004


Va\ iatbles in Equ.ation


Std. Error F to Pemo.e


69.02
21.21
150.53
1 Fr. 79
11E.22

?7 .2-1
21.-1
71 .92
7 .277


r0.6 7










0 02
31.E9


-.3.


O. 9i0


0. 41



0.31
S.3-5
0.15





0.44
0.'J2


0.49
0.4s
C0. 4

0.32
0.63
0.20
2.17
0..15


Variable
(lons t-ntt
4


6


9
10
ii
12
14
15
1 C.
17
1.-
19



24


34
-:5.


COEeficient


-46.06
-4.1 ..0
-O.S. 76
-107.3'9


C.:..-.,

43.23
54. _6
-22.1E.
10.32
-1.93
-21.35.
-14.i0,
79.2"


-39.137


0.01


0.15
U. -..





-56-

Table 5-4

Stcpwiise Linear F'eqression
3-Second Percent Change in Pur il Size


Step Varlable Multiple
Lumber Entered F' F'S
1 2 C0.4A6 0.149
2 12 0.490 0.24.1
16 0.523 0.274
4 24 0. r. 3r'i
5. 10 0.55: 0.312
S.4 0.570 0.3?5
7 1 0.5. 52 0.339
9 0.57j 0.351
10 11 0. J7 II..57
11 19 0.599 0.359
12 34 0 600 0. F.1
13 3u 0.6;2 0.363
14 17 0.603 0.36.4
15 36 C0.03 0.3
16 11 0.6'. 0. 3:
17 reMc.'e 0.601 3.4
18 22 0.604 0.365
19 15 0.f04 0.365
20 rrm e 0.C .04 o0. 6'.
21 7 0.605 '. "
22 32 0.605 0. : .
2. 35 r0.605 0. '6
241 re mo .e 0.605 O. 'j.6
25 9 0.605 0.67
26 0. 606 0. '67
27 remove I.606 0.36
2.1 13 0. 6 O.Jr
29 rem-ove i .606 O. 0
30 24 0.606 0.:'6
.1 rem,.e 0.606 0.366
32 14 0.60 7 0. 6.
33 remo..e 0.607 0.;6
34 19 0.607 '. -.,5
35 reci.,-.,e 0.607 r'. 36.8
36 23 0.607 0.369
." 11 0. 61I 0. 6










the 20-seconj control period bet,:en s lides and then had to refocus

rihen ,; n-w slide ,rpeared. The nari r-'.isiori refle:-' associated with re-

focusing would e.,plain at least cart of the rFsults found.

Tables 5-5 and 5-, present the results obtaiind from the regres-

.ion with the 9-second percent change in pupil size ls de, perldent variable.

The rnultipile corr'rlti,:,n coefff icient, and corrieqiiently, the eF planned

.variance, is lower" for this model. Again, the results of' indi idual

slide: rei'es..ic ns are nequivol cal. As tbefc.re, niar ly is of co-, ariarce

should ha..e el-iiriated the effects of light intensri ty and initial alue

on the final results. ThPse firdirngs Iaise questions .-.',,.ut our j tbilit.,

tc. deduce an,' r.,-arniingful conclu-si ons from resu. lts of rjupiliometric

as-essr, ent :tudie The low eil1.ained ..aria iiit., toae other iith the

inicons istFncy associated wi th rr-'pon, ses to thF 4 t r,'es of stimuli, raises

lman que: t i ons abt.out th: usdfulness of pu ii lonetri: tecr-rniques.

The fc.ll inri tables present i-e ult s of step, se li rinear egreres-

sic.ns on rhertrate, blood prressure, ra d slin potential. As would be

e-pecteJ in monel which do r'.t l1o these ; ari. a.les, the rultipie cor-

relation coeffficients iad e*plained .ariance are lower th-n th,'.se for

the pupil rsporns' equE tiions.







.58-


Table 5-1

Last Step of Stepisc Multiole Pegression

Dependent V 'riablc: '-Second Pucil changee

Multiple P 0.r575'
Std. Error of Est. 11.0520


Analysis of Variarnce
PReression
Pesidual


Variable
(Constant 0.0.
4

E
7
8
9
10
12
13
141
15
1E
17
IS
21
23

31
326

3.
35


D F Sumn of SCqu.'re
22 15019.4'26
2.4 3029?2. 547


Variables in Equation


Coefficiert



202.32
-18.2.3
431..?

41L.95
101.59
64.52
114.04
12-.12
7?.31
120.62
291 .01
241 .07
-0.0"

110.46
-13.92
19.92
-0.02
-0.26
-517.a0
-0.:.6


Std. Error


221 .2.5
150.1
13.92
311.09




74.76

,5.24
96.69
7.1 2. .



16.60
15.65
r, .00
0.12
1 .
S1 C,

970.02
0.74


F Patio
5. 50,


mle.rn Squsr"
682.701
122.147


F to Pemove


1 .2

1.72

1 .96
1 .91



T.2J
1.51
2.PP
3?
2 .6.
2. 0I
1.55

1 :
0.27
2.18
1 .29
8.95
4.00

1.34









Table 5-6

Step: ;ise Lirneor" F'.For Sion
9-Ee-.ond Percent Chrnge, Pupil Size


.tep vari ble Multic.1
Number EnterEd P FP.

1 16 0. 18 0.101
2 10 0.416 0.173
S12 0.4l 4.1 0 .198
4 E28 0 .4- 0.227
2.6 u0..501 0.251
13 n..09 0.259
7 19 0.51. 0.266
0 16 0.. 24 0.274
9 9 0..3 0.'281
10 32 r0.i -l 0.2.
ii 21 0.5 9 0.291
12 31 r0..5 7 0.311
i3 33 0.5t.4 0.31?
14 4 0 .567 0.322
15 223 0 .6E, 0.323
16. 17 0..69 0.324
17 e 0.570 0. 24
IS 35. 0.570 0.325
19 14 0.571 0.326
00 0.571 0. 26
21 1'' 0. 71 0.3't
22 E 0.,571 0.32










Table 5-7

Correlatiors Eeteen Puril Chang:e and
3-Second
Percent Change


3 sec -
9 sec t.,
Initial Val
Dog


Anima 1
Baby
Babj
Baby

Triangle.
Girl
Chair
Nude
Girl
Elocl:
Fectarngl-
Child
Hargi ng
IOml'ia r
'tartle


Variable


Stimuli
?-Second
Percent Ch anrq

0. 1
1.00
-0.25



-0.10
-0.02
0.03


-0.27
-0.13
+0.15
0.00
-0.11
0.01
-0.31
-0.01


-0.02
+0.21


1.0
O.?1
-0.31
-0.14
-0.0 s
-0.17
0.00
0.02.
0.04
-0.222
-0.07
40.10
-0.0?
-0.15
-0.05
-0.20
-0.01
-0.1 E6
- 0. 03
.-o K ,














3-5CSecnj

Variable
Step 'lumt er Entered. Peno,.ed
1 &.

2 3

3; 7

4 ,

5 4


Table 5-;

by Stimulus Groups of
Heart ra te Peq r-es S i ns

IMultiple


FP
0. .155

0.172

0.1%
O. 190

0.203
i~i
i, #j7


0.024

0.0,2!'

0.03_:

0.040

0.128


V r i..t.1.: s are- 3 Initial V-lue

4 Ple s;nt Stima.li ir, n rroups)

5 eutir l Stimuli r, i r r up: )

n 1Unpleasanr. timuli in groups)

7 rn, i et,' r .u in Qr St iol i i u n i ( roui )










Tahle 5-5

Sumnart' Ly Individual Stinuili:
Step,.-ise Li near PF'e ressior ,i th
9-Second HEartrate as Deprenjet Vjriable


Step
Number r
1
2
3


5
6
7
F.
9
10
11
1
13
14
15
16
17
12-
19
20
21


Variable
Er.ttered/Rermoved
3
21
13





10
16
17
14
7
.1
20
15
12


11
19
renoe
6
2 C'


lariablr ~ i
Variables 4-:


P.
R
0.4.126
0.445
0.460


0.494
0.494
0.500
0.507
0. 11
0..15
0.517
0.51 ;
0.520
0.520
0.520
0.520
0.521
0.521


0.523
0.522


ilul tlole
FP5
0.181
0.198
0.212
ri. 2; 3


0.2 34
0.z.1


ri.: 5
0.261
0.265
0.367
0.269
0.1'70
0.27')
0.270
0.271
0.2.1
0.272

0.272

0.27'


is Initial Value
1 are irdi .idual sl;ide
and startle stirjull










Tblie E-10

9-Sein ry y Ser tiu Charm n: Gr-ro:

9-S5-cond FPerccent Change in Hir-j'trate


Step NrilUer

1

2

3

4

5


Va ri ,le
Ertered,'Fl'a.irlved
3



6

7

4


Are:

In iti l Value

Pieisant Stimulus Gro'up

re-utr 1 St iimuli groupp

Unpleassant Stimulus Group

n.r! iety-ProuIi ng Stim ulu" i"l' ,rcuD


Multiple


P
0 .4-'7

0. 439

10.444

n.4415

0.J4 -,


F Sr'


0. 19

0. 197

0.19,

2 l l


Vsjritles

3

4-



6 -

? -






-64-


Table 5-11

Suriniary by Stimuli

Stepwise Linrer Pegression of
3-Second Percent ChJqnes i F;Blood Pressure


Variable
Ente rd.'PEF'r cri d

20C
1 1
9
17
19



15
r

.,
1;
18I
16
21






12
12


'*lul t i.le
P P'SO

O.'397 0.009
0.12?7 0.010

0.148 0.021
0.1 5 C0.027
0.174 0.030
0.178 0.031
0.179 0.032
0I. 180 0.0'1
0.180 1 0.032
0.1E1 0.032
0. 182 0.033
I. 12 S '.023
0.123 0.033


0.133 0.033



0.1 3 0.033
0.1 S 0.033

0.14 0.033


Variable 3 is initial value
Variables 4-21 are stimuli


Step Irimbc, r

1


3
4


6
7
E



12
10



14
1F


17
Is
13
19









Tale 5-12

Stepuise Linear Fegressiron of
9-Second Percent Changes in 1Elc:d Pressure

Variable Ilultiple
TlbPr enter td.' ro te:i: .ed P PS(

21 0.115 0.0
15 0. 162 i .0O
8 0.191 0.0
19 0.200 0.0
4 0 .20Z 0.0-
14 0.212 0.0
3 0.214 0.0.
18, 0.215 0.0"
6 0.215 0.0-
10 n.216 n.:o
1' O.21 0.04
9 0.216 0.04






11 0.21 0.0
16 0.217 0. 1
13 0.217 0.04


Step Nur

1
2
3
4
5
6
7



10(
IL
l.k
1 1




16
17

18


Variable 3 is initial v*alu
Variables 4-11 arre stimuli


-I

13
26
-6
40
43
15
1E.
16
16
16

16

16


I7

I.
17
'7






Table 5-13

Stepwise Lineer Pegressions of
9-Second Percent Chanaes in Skin Potential


Step Nunbe,-
1
2
3
4
5
6
7
3
9
10
11
12
13
14
15
16.
17
18
19


VCinable
Entered/Pe o'. ed
11
12
16
17
4
7
14


6
19
21
12


13
13

15
10


Vari. ble 3 it initial value
Variable: 4-21 3,'e S.ti r-uli
Niote: [ecai e of the res:ponr se l:tern in s[in Dotential,
S-second changes were not anrl. 2ed.


ilultic le
P PRE
0. 205 0.042
0.275 0.075
0 298 0.0S9
. 299 0 .0_:9


0.330 0. 090
rj0. nO 0.09C,
0. 300 r0.090


0.301 0 r.r.,

O. 301 O. 09
0. 301 0.090

0. 3:01 0.000''
0 .301 0.0 l

0. 301 0.091

0.301 0.091

0.30'1 0.091
0.301 0.091
0. 30) 0.091









The re;u1lt e aeicri b.--d atxove uwre applied to the h Dotihei'C-

li;t-d in Chaptc, I H.,pothesit, 1, "Ile-asarnt stimruli will be accoT-

pani-d by dilation, while neutral stiriuli will caaj rno .:hanr. e in Durtil

cize", .a. ri .t surprrted. : can be .seer, ir the correl aLion nratri nre-

senrt.d in Table 5-;, all p l asarit :tirrn jli have a slight neatii.'e co: r e-

l ti crn uith 3-seconi rupil char.n lt.thoua h the cor'relat ior.: were Eir:

negqti' : for the ?- record cupil chr.-nrge dat the, .re I- tili neqati e.

Sni ilarl., r, eu ral ;tinrul' i general ly h.ad rnenati ve corre lat or, s. These

re:ut1 s i ir lu e tl u i fi Ta t ir.in o ut cf .,rn.iet., scores tLv analysis of co:-

v arij a rc .

Hr.pothe:i : :, "n gati'. stiiul i will c ause a cor.strlictior, of the

pu.il" as n.t supportr.d, :in:. ti: puril r acti .r, to th s.e stimuli

waj liarer 1A a po ti'e :ch.iange in ize. Hao, t" si; 3, "a. ' iety- air'ui 1 ng

itiriuli .i 11 di late the curil", \ ':s s upp:,rted.

Hypotlhesis; could not be tested due to the inrConsisT.encies inr

the resrorise to p. easart ,t r I irca ri consis te -nt r :r,n to r. le a-

sCarnt ;ti iul i \ s f'urnd, th, dearce-e .,f dilatio.rn could rnot bt cor;Npareo

with dilatiorci;s a : ;.:iatec i th ari ety' stimul i. Hi,'pcthesi 5, "hiihlv

pleF isurat le :stimc uli *,ill podJuce greater pu i l di l3tion than will less

ple tas:it ;t lrj li" \was ro, t s uoportied, 5a tl h cori.,elatjion matri.i. shis .

Scr.r-e poi ti *'. results; wevi oLt.i-.in- ed in th r ari aly, i; of flarnrniltude

co. ari n ace. Hyp-th-sis 7, "t loc.d pressure change, .; 1 be sin il r in

reaction when ri .poFIed rt pl a-.:s r ant : tin,l) i r1 1 r .l;en Ec po.j:ed to ri ety-

ar.:.u:ing ;tiruli", w.as not suJported. !!ii e blood pressure sho..r:ed no









consistent relationship to positive stimuli, it did sho i z positive

relationship to an:n--ct;,--arousing stimuli, with a lag of sw\eral seconds.

Heartrate decreased as a res;:onse to startle stiAuli whilee shoWrin no

consistent response to pleasant sttiuli, so h,'nothesis 8, "heartrate

changes will te similar whern exposed too ain:iet, -arousin.] stimuli and

pleasant stimuli" was not Eurprported. '.kiin potential had the largest

time laq (t.etwreen C and 10 seconds) to startle :stimuli, and had fairlJ

large mnagnitudi c.h nges (un to 70 percent. It shoir-d nc consistent

relation to pleasant stirrull, co Hnothesis 9 was not supported.

l.pothesis 10 could not be tested, again due to the iny..,risi tencyi

in responses to pleasant stimuli. H,pothesis 11 a's su reported, since

no consistent r-lation hFtaeen heartrate .anj nositi':e stimuli were found,

while the average time frcm beginning (of a startle resporine to its end

h.'as 6.2 seconds. H.,pothesis 12 ;a: not TuppoIted, sinCL no consistent

relationship was found for resp.inse latency or return of s;'in potential

to either type of stimuli. Hypothesis 13 was supported, pr'o'ah l; because

of the relationship t 'isting between heartr-ate and L lo.od Cressure.

Table 5-14 presents a summTar, of the hpiCithEscs and the results


obtained.








Table -14


Fa ults t: I','potheP is


:upper td!PF.j ec ted


Puli ilary Peact':n


.1 Pleasant tlirr, uli .ill bte acc=nl'ianied tb,' dilation,
while neutral stimuli i;ll Ci.. u nro change in
pupil size.
2. Il:.ative stimuli will cause a constriction of the
pupil.
3. ,An..ircty-aro;usiri. stirull will dilate the pucil.
4. Anri ty-arcouting. :.ti.uli will c:u;e IrEater di-
Itti nr than will pleasant stinuli cf the rame
inten i ty.
5. Hiqhly ipeasurab, t tiruii will ro'r d.iucoe gro.Ater
pupil dilation than l 1 less1 pleasant stimi uli.
6. Highli negati. st ilul will p'roducE or-e.Ater
pupil conr tricticri thar will iess ne.lati'.e
s t iriul .

'u Itonoric Co rn ita ri3 t i r.,
A rllonitude Cj. riance
'. lood. pressure chari-'no.. ill he ~ si~ ilbar in reac-
tirn whenr e posr ed t. plc;arant :. irruli anrd irher
C-pos ed to an- i ty-arciusing str.iruli.
C. Heart'.te crnii e' will be simi lar .rwhri r po:p.d
to anr iety-arou'.l in stirrul; r.jd pleasant stirnuli.
9. 1 in potential cha ,..i11 :,be similar in reac-
tion ronen r po:.Fd to ajri et, -ar-ou: ing and to
pleasant stiniuli.
B:. Fesp r:r. Dur ati. ri
10. Pupillary dilation ro n-iet*,'-arousino stimul
will t 'a e lone, to r-tiu r to ini tial base 1e .i
(at tine of stimulus rres nitatio:.n) than till ruril
dilation to pleasant c stimuli, for a nien le .cl
of t in u lus irnte-ri ty.


rot suppcr'ted.

not sup.poted
supported





not suoport-dl


not SupOCrcted






not supported

not .urCOr ted


not sur.rrtrd


Until: able with ores4rit data


HypothesaS










Tablt 5-14 (Cont.)


Hynotheses Supported/Pe iec ted

11. Heartrate will tale longer to return to base
line in response to an-ict.-arousing stimuli
than to pleasant stimuli, for the sanre stimulus
level. supported

12. SK.in potential will take longer to return to base
line in response to an. lety-erousinq stirruli than
to pleasant stinrli for the same stimulus level. not supported

13. Blood pressure will tate longer to return te base
line in r'sp nr.se to' an.iet.,-arou'sing stiniuli
than to pleasant stimuli, for the :a',e level Cof
stimulus intensity. supported




Shortcomi nis


It would be useful to attempt to e-,lain the disheartening results

described abo'.e 6~ the product of poor e-periniental dt-siqn ,or analysis.

The controls used in the study, hoev,'er, were at lFast as rigorous as most

reported in the literature. Also, as mentionedd previously, the ddta w,re

collected in 2 forns (chart and digital) with chart data being used to

%alidate digital data at all points of con'.ersion, so data reduction

errors seem to be negligible.

Although the step: taken sDekP well for the data collecCr.d, the e

are several areas where the eperinrent could h '.e been imo.,rcO.ed. Better

care could have been taken to de'elor, a set of slides with more closely

balanced light intensities. Spot intensity on eecli slide should hiv--









been controlled. At the data recording le.fp better rteasures of

11-1ht intensity could ha've been made.

It appears that the scc.r-es on the pre-test questionnaire ram

ha'.3e lid to e rIrorne:uI conrl]us iois c, about c.ne c r" more slides. For e>am-

pie, the s11.ie show ring an attac:l ing pq lice dog sjcoredJ hiuheSt on the

.nrr ir-ty-arous n u Jiirien.-iocri arid wa.si therefore included iri the -rii ety

5st. Sub jects wih. lere debriefedj inforrn'll, ho,ewe-,r tended to ag'jree

thjt tnriIre i '% s little :an.iety,-.3rjuusinl ru llit.,, inr a Islide c.f a.n attacl'-

inrig d j o the questii-irin.aire r scaiF ri, v ha3 e been iiiilei in .

The nme-thoad b. ir ich rc mbl in r s r.;,re rdi ted in thir computer data

c.-ulld also be imripr'.'. id. ThE .it.a r edu:tion priorn's inrc rr.rorted a

-:.*r jutiri.? th.r *. i;ch r .? di ng ur.der .. millir, e er: '.' e1 li in ted

.nad the i:te coint imniredJiatflv preceedinri the eyeb.linr was .utstituterd.

eriih ps the r,:c..t ;apprlFpri ate procedure iU'uld haj br--n the total elijiri-

.1i Cf that dat, point. Alternaiti ely, :e'...- SIT.I thir g t echni qures

are .a. iiab ic- which iiould hev'.e b: rin mo.re ar, rroic.iate th i'il the mI thrlo used.













CHAPTEP VI

IIETHODOL iC,.IC4L I'IPPOVEMIEIITS:
THE FEASIBILITY OF CO PUTEP CONTROL AND DATA SAf.PLING



As discussed in previous chapters, the methods of data collection

ir psycholoc gical and n.cdical research on ph)isioloical variabls .re

often tedious and su-ject to :t-'eral types- of error. Onre of thr most

productive results of this study \:a the demcnstrition that co-rput1r

sampling i a viable and ad,,ante.eous niethio for collectiri dc:ta ,on

physiological *.'ariables. Computers today are capable Cf performing mrn:,

diffiertit functions nd .re availa. le ii rnan forms and si ze:. Their

costs are now at a 1,he1l where rniost rets archers should take tine to

examine their" usefulness.


Adva i ta qes


Several adiantagies accrue to the Iresearchers oho utili-c the

computer. If he can de'.elor a miethocd hererb the computer cari be rro-

,aran .red to contr'Cl the presentation cf stimuli, as in this study, the

experimenter can significantly reduce the uncertainty and inconsistency

with -hich hu.ansr perforni such activities. The comrriuter c:an overcome

limitations cf hurlnn perfor-,mance cap.3'bilities, such as the split-second

changing of slidfc in the, present study. Because of their hi h r.o-









messing speEds ciomputors ari-, cLrapbl' (f perform rin several ope'Fjtionr

simultaneously,I such js the chanirin, of li es anr ri col action of several

data samples it the span of 3a second.

As a data co llector, the c':.r.putter offers the a'l',antages associ-

at.ed with unritiased srpling and recording of results. As the epei-

mrnter has shown earlier (Eeell, 1971), controls oft this tpe must be

incor-pV.rated in ph,'-iological diata-g lthe-ring. methods. Cori-ut.--r also

pro' vi'e' a -ijgnrif-icunt reduce tl[ n in thm_ tinmer it tl e: to pioduc.e cat

in 3 form read. for anale 1is.

Perhaps on o of i he major ad '.aticeI in i:'cc uter h.:idware in reircent

,ears ha tiLeri the mini-crom. pu -e~ These- machires irnco.rprrasti ng solid-

:tace iiir. ocirii ui tr can b' obtjir;n F1 in :i:r th.a. .ill fit 3lnost

ar..sh re and that are hihl, .-.ptl .'. to the r_,ecific re.I qui rem-nts of

tnr purcha.cr. Vendors will, in srone ca-se 3asist in the acquisition

of rqui nrerit confi'uritions t hich :, :t mr'-et th-f needs rf thie u .ser. e.'e' a1al

mini-cc.riputers a e- "con..ers iti rnal" (he,- the ri 1 ity to corinunicat e

v ith larr'.er- scale niachirnet phyvsio .?lic rial resEarcher could obt in

a PDP-11 ri ri i-c'omput(r D (Di it l Equipmr.n-nrt Cor''orationr), for e .nmpl e

to control his eIpFri merit and collect. hi.s data. The machine ciuld be

cor. ne tet d vi a phone linkr to a lcar?. -scale E,: tern such as the IErIl CO/'..

w'h--re direct data i'ea in, anrd araly' is coj ld t.:le r.lace. Thus the

riachinre at the local 1;e--l could be %dapted t, 3 sprecific re-search

need, r.'ilt raiiltatirilng the abil ity to use ler e-cale data ara al is,










packages, and all of the peripheral equipment associated with the

larger scale systems.

Finally, ri.;ny persons associate the word computer ..ith high

costs of operation. while not fully realizing that computer until ization

may reduce ior eliminate other costs related to the pro.iect. This will

be discussed in more detail later in this chapter.


Di sadvantatqes of ComrPuter Lisag


One of the niaor disad'antagnsF of computer usage to behavioral

res'rarchors is the dependence on the "systems man" such us.ge creates.

Lefcore system can bte utilized. hardrar' must tbe acqiui'rcd and made

operaoticnal anJ software (progrnrms) mus t be developed and dehuuged.

Once the systemrn is cperaticnal .nr. changes to be irmJe or any, errors to

be corrected require that someone familiar with the computer be called

ir.

There may also be a tendency to accept the face validity of

cornputer-generated data without questioning the data collection, trans-

fer, and analysis. When using the computer, an e:.perimenter can be far

removed from the data he is collecting, and may miss important happeningrjs

during the e.eoFrirment that could influence the data. ThF use of a com-

puter does not reliiee the e.periliernter of the responsibility for accurate


The use of an earlier model Digital Corporation Hdachine, the
PDP-S. in clinical and diagnnostic nodic l research has been described
earlier by Cov.vey (190,.C .










d-et collection., ari.lysi;, and reporting. An important cornsidPration

is the frequency nd rcomple -;ty of the projects to be underta'i'.enr by the

computer. Coriputers male mostly nieces of furniture- in r research 1 aor-

atcry if the e tent of their usa,3 e does not justify thi-ir" ac. ui;sitio'n.

In uri i'vrsit. sEr.tinrgs, rese rchersl many times share a computer wi th

others w.ho need one in orper to reduce the per-projec:t utilii3ation

:co ts.


Scome Cost-Eenefit C cns idrat i ons


It is 5 rela ti vely simple mai tter to d-iscuss thl' h dvanta.je. 3and

1 mir t .:tir o s of co'rrputer s,':tens in giereir.l1 terms. L.efo're the reco ni-

menridation to puichl ie a 5.,-stem can bte .:cepted orr rejected, ho':.ever,

the \ery difficult tisl: of i ;ss',,ining doll r cists and b-rnefi ts cif th

s.Steni rnust be undert.a' n. !lin;,' cost f'i-ures c ri, be readily obt.airled

fro im pr-ice lists of mariufda turers. Other. ;uch as- the cost of install in

the .,':te and mal irng it operational are less ea.:, to pinpoint. and

ma, require a qreat deal more re r ;earchi rid "uC-sIt im a tl or,". Even thi.se,

ho:everi, may be estiriateid fiorm the files cf ui ers. of s ii ilar equi ment,

or from th.- '" i iUi'.:tuel .

Thi most difficult problem lies in the generation of accuratee

retiriites of the benef ,it to t.: der"i.c- from tlh coTIpiter r .,'tiTe Before

the system m isr installed, it is imipossitble to: foresee al1l of the rp:; ible

.-pplications it n,',:. have. It inaj, also hr ppern thit so:,e .anticiFp ted uses










of the system are not realized after installation. bout th: best

that can tb done is an attempt to foresee as ran,: Dotential advantages

(and limitations) as possible, and to treat these data in a ray that

recognizes the uncertainty under hhich they were generated. For this

reason, this section will treat data it, a r-.rncr proven useful in other

situations iriv.ol i-in uncertainty. In particular, vhenr uncertain nt,

exists, data will be derived from the familiar weighted-average formula

used in PERT and other project planning and control techniques.-

P co:rruter system for collecting and analyzing data of the

type descritbed in this di.sseration, would at a minirrun, coritain elerrents

for acquiring, storinq, and processing data. The hypothetical s.,stemr

described beloi attermpts to re l isticall, neet those requirements il thin

the frame-iorkl of decisions made by a uni er'i ty researcher. it is

assumed that the researcher it norv collecting data manually (as iTos-t

pupillometric researchers do)', transferring it to punched cards and

anal.,iing i by computer.

It is further assumed that meas, urino instru.Trents such as the

television pupil i oneter and devices for assessing heartrate, blood

pressure, ar d 5sin potential are readily available. A. central orocessng

unit .C.P.U. ) dith at least 4-thousan.i (4- ) bit men'.ory vould be required

to store the data. Ana'log-to-digi-.il conversion n requirements call for



llallace. J.E... "lr oroving. communications between s,stenis
analyst and u e-~r Data larniaerene t June 1972, cr,. 21-25.









the purchase cf 6 channels of con,.erters, and the input .output oDera-

tions of the s.sste- require a 4-k 16-bit read/wri te computer core.

The harVdiars de:-cribeJ abo.e form. the nucleus for a system capable

of acqui i nr anJ storing an amount cf information cr. omE.ir- t le to that

gathere.' for each .ubject in this dissertation. If we asi.ure that the

system;. is being desiqnedr for a university t research laboratory, then

the mo,;t efficient use of additional funds for analyiS dvs3i5 t.a eq,'liFp nt

would be the purchase :,f a dataphroe. This rould allow. the .Oriputcr

tC, .:omnT' ic'te with a larqer- .a lae ctrr,. The d!ataphone would tie

the small cc.'puter in with the C P.U. cf th,- larger tma,:hine, thu: greatly

enlar- iinri the stor'a.e capacity :.f the s s,'tern, incr".asing its analytical

rz,pabi i tic:. -ind ccnn tectin? it to all of the r.:-i'p er.l devices on the

larger s.; .stem.

Tabl E -1 de;cri.es the .d llar iestiniates of annual benefits to

be gairoid fr'~ci the new computer system. Table 6-2 d_,scr'ibes estimated

c* ts. Esti mates icre nm.ade by this wri terI ain by prc'fe s.o.r cL'r entlcy

en.aed in psychiatric res-earch Lsinr pupillometrics arn heartr'ate anal.si .

3'mon of the sources cf dollar cr t reductions or- added -enr efi t are men-

ti,:ined iefl, in the ta.bl-s.

Fs can be seen, the simple ,atio of annual rbenefits of the new

system to annual costs is 0.:'0 'Fe'.eral otherr techniqes for analzin

cc t and benefit dat. are currc-nt I, used Arnmcng these, t'e coI t 'ber-~fit

rItio anri paybar' period for the ir..e: ttmenlt in the coiput-r system i.ere

corp j ted.










Tahlr- F,-1

PDP 11/20 C.y, tem iith [lat;, .n r..ne I. i r;tan to IEl 3 ,'6

Lost 'Pencf-it Analv\ i

(Annriru l Fa: iT)

Benefi ts:
Pr-ductiorn PPiss innrP nrt of rlricral t.Aff
(Peduction in staff needed t m, ea'ur-.
phyv;iologicl v.aria blec.)
(Peduction in I, .r.uri.:h costs) p ,

Pecponsi'.enets Fastr.r r',t.a Transfer
(ability to npot errors while u iec:ts
.nri r_-iuir.irrent i confi',uratlionr, are st ll
,. ai l..i.e) 1 rir,r


Accuracy: Increise in Peliabi il ity of Data Sari.li rng
(''lue of increased accuracy reduc,,d need
to 'eperformr e, erimnint; i n:reasie in confi-
drnce in r ge-n ratirn report etc.) 1,0rij
CoirprehenrS i ,en .i es CLFpa ,i 1 it. for Increased
Scope of Data .Acquisition
(valuE cof additional inforrration: ..alue of
new Droic.ts won because of oparide-d
capabilities) 5,CO

Total Est ir.cated Anr. ua r, l Bene; fit r 15,


Cost/ renefit Estiir'i te.r Deri 'ed from teta-Distributed
Estimate-: Und-er Uir.:ertaint :

0 + 4Jr-L + F

where P = retsilnistic
0 = .pt rimistic
ML = ro't likelv










Table 6b-,
PDP 11/20 5..stenm th DatalChone LinnkagR to IE.lI 36.0/6'
Cost/Errnefit Anal;, i
( nn,.al Ea: 1 s i


LOs ts
Jonrl ur rr 1 in
C F .. Furcrhasl
-t 1 6-b1 t r,;-d,'a"-rit' memory
F' rcira ;iiIm r corn:1-cle
Easic mounting bo.:
Po r..' r supril,'
ASP-33 tele .tipe :n control rc. ,00
I i16 l it rcad -orite core 3.1 .li
'. to Di conv'erters, 6 channrr el 2,ri"l
Da t phorir 1 rO
Si te-rr irs :. lotion -debu. i ring 1 ,000

<, 1'9,30
5.-'ear iife innrual Basis 3.FO0

Pe: prr in.]
[ataphone I .o.Jern Frit l I. 1O 'l/ni. ) 1 ,'
S., tfrii pri'O i'arrmi ri ?' 000
3y:terii ri,:iintenjrrar 1 ,0I"l
S;sterim s;.iprirl is-: 500
CFI.I t ill.e,' ,stem :..C, 5,ClO

Toutjl rstirratgd annual o:(st I ..60

Cc,:t hdl' idc-d amrron three ,proje t: 1 ,e. 20



'Co:t'Eeref-t Estim these Derie..d Frorm E.c-t-Di:itr .uted
Est.im ts U-s i r L, Incerta i n.,'
P Ji!L + 0
6
wh r- FP = rsimi t i ; rl = .or in siist i; 'L : mos t 1 i lely

Sou'r:e: 'ioi t l E u- ipnr. nt Cor rF:,'tioin FPDP 11,' 0 F'rice List. rio,..1970










The payback period is defined as:


; Cl
tcI
C( t CC it T
3 j

where T = expected life of the systemm
B = the ith benefit in time period t
CC = the ith recurring cost in time period t
CI = the kth initial cost
P = the time value of money or cc.st of capital


For the data described in Tables '.-1 and 6-2, the pab.a:k periodd is

1.SS y,ears. These data indicate that the ..:;tem ,will pay for itself

in appro,.imi tell, onr, and one-half earsr. Since p 5-..ear life was assumed,

it appears that these results indicate that investment in the syvten is

justified.

The ccst,'benefit ratio is computed from the formula:



r (I3 z CC)
t I Cl
(0 + n)

The data yield a ccst-btnefit ratio of 0.A2-1, indicating that over

the 5-year period the costs associated wih th the system, e'en ..hen con-

sidering the time-valu- of ;none.. invested, are less than orne-half of

the generated benefits. These results also support arnuMients fc.r acqui-

sit ion of the syite.n.






-81-


As is Cobvious, the data presentf-J in Taible 6-1 represent *.'ery

unrcrtrin estimates of benefit t- Hcucver, ev.r if we car assign oly

a Drr.ot.bility oi f C.5,E to the c y.nected occurE-nc of these benefits the

cc.st benefitt ratio .till i les; than 1.0. This result allows scon:e

degree of confidence in the statement that .4 sste.,i carn t-e justified.

In surr:m r.,. it aDpears that researchers should in'.estiq, te Dos-

sitilities cornputer s.,cstens offer as scOurces of moree ad.' arinced E -cerii-

rinrital control 3nr. an l. 'sis. ith-riou h "-pecific cost anrid b nefit figures

depend or inidi:'idu.5l project requi renmentz, the d.ta prir.eri.td above

suggest thit, in some c. aes inestmcrnts in these s, 'i.st' s can I.e eas.il

justified.











CHAPTER Vil

I1PLICAT IONS FOP. FUTJPE PESEAFCH



The results of this studj raise several questions ahout the

direction of future research on pupillometrlcs. If we accept the hyoo-

thesis that the pupil does respond differentially to stimuli and that

it could be a useful indicator of emotional reactions, then .'e are faced

with the apparent need for much ojre basic and much more intensive work

than has been previously reported As with this dissertation, for

ever, "positive" finding reported in the pupillometric li terature,

there seems to bte a corrolar. "negati-e" one. Until this tyoe of

research reaches a stage where general uniformity of results is accom-

plished, it will nct be accepted as a useful techrinue for more advanced

apolicatio.,s.


Attitude Theory


Given the complexity of design necessary for evaluating attitude:

with physiological variables, an effort should be made to determine the

additional information this d:sicn ;ields over other, more Lonventioral,

measures of attitudes. it appears that paper-and-pencil instruments

such as the semantic differential, the Likert technique, or the Tliurtone

technique would yield morE information per input unit of work e''ended

that uould the physiolro ical techniques. At the present state of the art









in pupill -ri etr icL, it al o appears that the e scale: would d yield a

cqrc: ter absolute r..ai ni tude of infi:rnat; t hLC'i h : uld rupil studies.

The generally r'eportd rel iai lity and validity coefficients associ-

a ited \,ith the r. aer-ar.ii-penrcil r.methiods .are higher than tho~e ass oci-

ated with pupillnimetrics (Wloodmarsee, i 'j sc it rppears that for

mou: t asseiL srS'rent probli-er.-, the converitionafl r,,aS ur rinenrt iins trurrer it

imay b-e more appr'opri.ate.

Conmpar tiv e studjie ailing this liner could defi; n- ;ri me r*elEc.ait

attitude doman'in tco he .arimlied.,I thern je..'-iloor aszesrment ins trument: cf

b.-tli t.,:pe roL.tin r the co. t: : f d IS~ oprient in t irne, dollars nranpor, ier' ,

anI other r urcei e..r.ended. Peliahi.li ,, nd ,al-i.it., crefrfi: iir nts

:ul.d he cc.'.ipuled us inr, data g3thEe*' d from "I rne., in group.'. Per unit

c:ontri'LuuticrO and a sol ute contritution'i of infjorrcatior, could h.e cal ulated

ftor" each technique.


PFscearch o n 'tiriulu: Pre nritction


PerhaFrp one of the most limiting fac.to's ir pupiilomer.ict s ii

the mer t:r.d ,C:f joper.atiornalizling independent variab'les throu'. h visual Stipriuli.

1, 1:,O c' tac t he re.- ressior mode l de.e loped in Cihaptps V derr.on t s' tr.

tlis pi ic.tlen fI ny of the fac.:tr. is affecting pupil size .ar';- .a ocia te

with the e.,e a 'r. input medium if r :timuli. Thie near vi ic.r, i refle.., the

oi i'ieting refie., color adaprts tion, i.iht intens'i stv (h.oth -..ot .nd

ScrarTiiled), :tinilu u' ccriple.. ity, .nd the phy;icai locationr of the .viual

s.tinuulu. all cause char,.es in pLpipi 1 ize. if it W.ere po',ss Bl to de'eloCD










a different t,.pe of stimulus inrut, such as auditory stimulation, the

laqrg number of contributors to puioil change :ould be reduced, and

perhaps the effects due to ttimuUlus content could be better isolated.

Beyond this problem, there is the difficulty of projecting ac-

titude objects through the use of visual stimuli. As several subjects

in the present study noted \a ery larce differen..e t is t beti'enr an

attacking police dog and a picture of an qtt-ckinq police dog. ESii-

larly, a picture of a Nlegro rin, not ev'l.e the same t.oe of response that

a living being does.

A very difficult p'e blem must be ovecor.rme before pupill metric

techniques can be applied in areas such as the rcO'isonriel furictiior. To

date. most successful studies hae involved fairly simple, basic atti-

tudes. Host of these could have been c.'alujted equjll / ,ell using ccn-

.'entional techliriiuss. The hope is that some method for presenting comple.

emotional stimuli can be de'velope-d. Of all stimulus t;pes, ccmple,.

visual stimruli are the least preferred, since the comiple. ity requires

greater coping behavior from the I..upil and cause: greater Ee search

of the stimulus, both of whichh confound pupil change data. Smnie mc thod

which does not cause confounding interactions Pith the dependent variable

must be developed for presenting nor.r couple. stir.iul .


Field Fcsearch


In the Iarleting area, in addition to research in stimulus pre-

sentation, much work needs to be done o.n equipim.Pnt aind mohilit,'. P 'r erit










techniques are still v'ry "obtrusive'" and iriLer activ.'E itl. ti resnorn-

dent. The dev',-elopmrnt of a norn-int ict'i -e reasuririg inst ur7in for

research on consumer's re acti r's to p..a'a'e e si no and other promoti onral

techniques would be a imiajor ad/vanr ce.

If r-arletirigg esIarchers ps1 r, to us- Fr pi l m1 ii etrii c a useful

stud, could bt r made using Fortable physiolco ical recordiri. de.ices.

A phy.:si cloqic,: recordl-er, f'or inst arce c, uld tirecord several tr.ac : o, f

data simultaneously, during a iar'l:etling .tijd., The data could th!rn

be tale-n bed to- the labor1oortcr,' arnd rad directly, fro:i the reccrd-'r

into computer. Th.- laq timri'- between data collection, and anal ,sis

c',:uld tbe .ignificintl,,- i educea using this- tec iilque PResearcher. could

u.,,al u.te the tra d .cff '- bet. ei. r u si ng st 1 il -f raTe '.er us ,u i c: i a lrars,

and could de eiop ciit/b Snefit ana l se'. of the irnt uI ri rt ul in th

stud,'.


L ?bo rator',' Fese r:ar'ch


[luIchi r ri. still n eo se Lc. be done:- in l he are.? cof insti urn.ntatic n

in rpuFil 1lometric res: i'earjclh. 1 1. appears t*hat th.1 re is .'ry little question,

other than relati *,e co ts about the c-rcira] super iori t,' of video s stemss

such as t t 1he television pupilloin- ter, e r.ecia li,' ,hnr such U s steriis car,

tbe interfaced rith a computer. whetherer the Sir. n ficanrt addi tiora costs

of thr tele'visicin s-ystesri are lusti fied ,rineriends upon the v'alLe of the

inrforl, nlion 1r. be gather red and the -.a i labi li t,' of s-:pr tin e o. nic 'q en t.










Finally, applications of the mini-ccmputer in physiological

research car now be evaluated. .-'ork needs to be don? on the compatibility

of the small computer systems with larger systems. The information

processing, storage, and input/outout requirements of physiological

data processing systems need to be investigated. With a.dances in

computer techr,olgr;y continuin.- at the current rate, researchers who dc.

not recognize the potential of the computer for their projects may

soon be left with obsolete methods and inferior results.











CHAPTEr' V\II

1.11.i^ f I:,D CCi!CLU .IO1iNS



Gc 1 ls


Th& goal: of this dis er-tttion were to, repr licate fini.ings that

pupil chrqe is a useful measure ocf &emot.l: na ri e:tic''. to .isual sLiiul i,

de.'elop a lin rear nmouel e1 pli niir,g tIhe cnl-tri bit tctrs to r-.u~ l charge de.el op

a cci.puter sariplin r[slhoi i and c'e.,ine the fe i,-i ilit:,' of its use i Ii pupil

change studies, and studio the o their phy,,-iolc. c.ai Co.ariates in pupil

change. It was hor,=d that the recc.rdir n of chanQ-..s in r.otihe r 'a,";ar il

al ni wi r.th pupi 1 il l r. e ~ oulr il ed to i mer ns of 1di -r irii nati ng t et..ceer

pupil dii at ia or.s cause, i b.: p I c ur tt.e stii:uli and d iaii li urt .iur--:l t.,

at iet, -ar .u-. irq s- tii. ul i.


lMethod


Fourteen -ales e.eA..ed as indeperdent rate s cf 57 l ide stirii' .

Of the 57, 16 were .-ho:en for use in the stud,. The 1. were divided into

-1 -isual stimulus .c teoj ries: ple -saiit, nreutr l1, iunc'leisant, rand anr iety-

c.r' feair-':,,i u i nj. The mediar te t wai used- tc. test f r differences

tleteern the sets of stimuli. Each set wa; a s inificarnt'i different from

the cther -et. at the .01 level. It appears, h.or,'.*ver, that the sl de

used as an an- iety-aroui in stirmulus ,.' .actu 11, not t capable c.f afc'u .irngI










anxiety in the subjects use:. Two startle stimuli--s buzzer and a drop

of several inches in the chair subjects sat iri--rere also included in

the an.:iety set.

The study tool place in a laboratory ..'here environmental light

and sound levels could be controlled. Subjects were seated in 3 dentist's

chair, and looked into an apparatus similar to a Hess ['.o.. Slides were

presented in random order on ; screen at the end of the bo" orpo, site

the subject. The tuo startle stimuli always occurred as stimuli number

17 and 18.

A television pupillo:meter measured change in pupil sire. In

addition, heartrate, blood pressure, and slin potential %ere measured.

These variables, tcaetheI with tile liiht interisi t, of the slide screen,

were recorded with a Dyrlrgraph recorder, which Droduced charts of the

data for each subject.

The Eunl:er-Pa computer as coutennercEted with analog-to-digi tal

(onverters to the D.nograph recorder and recorded second-by-second

digital readings for each of the ,.'ariabls, Ft the end of each session

with a subject, a paper tape record of the 2,5GO readings taken on that

subject iwas produced

A Eunler-Playmo process control computer wal connected to to

iodal carousel '50 sliie p--ojectors and controlled the timing of the

slide presentations. Each stimulus slide was preceded by a 20-second

blani sli e. The E:unl.er-Pay:'-. by initiating a change in the projector

containing the stimulus slide before iremToving the controls slire, reduced




Full Text

PAGE 1

THE POTENTIAL OF THE PUPILLARY RESPONSE IN BUSINESS RESEARCH: AN INVESTIGATION OF METHODOLOGY AND AUTONOMIC CONTAMINATION By ROBERT ROY BELL A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1972

PAGE 3

This dissertation is dedicated to my parents, Mr. and Mrs. Alvah Roy Bell

PAGE 4

ACKNOVLEDGEflEriTS This dissertation is a product of the advice and encouragement of many people. The author wishes to thank everyone v/ho participated in the study, expressing special thanks to the members of his supervisory committee. Dr. William Fox, chairman, has long been a trusted advisor, teacher, and friend. Dr. Fox first introduced the author to the problem of psychological measurement, and his constant encouragement and enthusiasm for this project have been inspirational. Dr. Marvin Shaw has provided expert guidance throughout the author's graduate studies. Me has served as a source of information and ideas, and has sympathetically led him through the maze of osychological testing and measurement.. Dr. Walter Hill has provided some of the most rewarding learning experiences in the author's graduate program. Dr. Hill's willingness to listen to and explore the ideas of a beginning researcher is greatly appreciated. Another member of the faculty, although not a member of the author's special committee, deserves soecial recognition and thanks. Without the interest shown by Dr. William Wilmot, former chairman of the Management Department, the author would not have entered the graduate program. The author would like to express his deepest gratitude for the many hours Dr. Wilmot spent giving good advice and counsel. m-

PAGE 5

k'ithout the kind assistance of Dr. Ismet Karacan, of the U.F. Department of Psychiatry, the py^oject v/ould have been much more difficult. The television pupil lometer he loaned was one of the most valuable instruments used in the study. Similarly, Dr. Kerry Kilpatrick and Dr. Tarek Khalil, of the U.F. Department of Industrial and Systems Engineering, have the author's gratitude for making the Systems Engineering/Human Performance Laboratory available. Several professors provided help in the design phases of the experiment. Drs. Erich Thomaneck and John Wallace of Management, Drs. W. W. Dawson and Joe Harrison of Ophthalmology, Dr. George Barnard of Psychiatry, and Dr. Robert Isaacson of Psychology have his thanks for their assistance. Two student groups in the College of Business Administration, the M.B.A. Club and Delta Sigma Pi, have his gratitude for providing subjects. The author would like to thank Mr. Robert Lyons, in particular, for the effort he put into acquiring volunteers. Mr. Alan Copsey of the Systems Engineering Laboratory provided support above and beyond the call of duty. During the data collection period of the study, Alan spent a large portion of his days, nights, and weekends working on the experiment. He has the author's deepest gratitude. Several fellow graduate students helped in different phases of the study. In particular, the author would like to thank Jim McGregor, Stu Brown, Margie Kuenz, and Fred Nutt for their contributions. Also, Mr. Evan Eldridge and the staff of the Business Administration Computing Laboratory, and the staff of the Communications Department in the College of Business have the author's gratitude. =iv=

PAGE 6

TABLE OF CONTFMTS Page i i i ACKNOMEDGEKENTS LIST OF TABLES vii LIST OF FIGURES viii ABSTRACT ix CHAPTER I: INTRODUCTION 1 Data Assessment Problems 1 The Potential of Pupil lometrics 3 Traditional Pupillometric Techniques 4 Pupil lometrics in Business Research 5 Scope of the Problem Area 8 Summary 13 CHAPTER II: PREVIOUS APPLICATIONS OF THE PUPILLOMETRIC TECHNIQUE 14 Historical Developments 14 Attitude Theory 15 Pupil Size and Mental Activity 17 Psychiatric Research 18 Emotional Reactions and Pupil Size 19 Marketing Research 21 Dissenting Votes 23 Summary 24 CHAPTER III: RESEARCH GOALS AND HYPOTHESES 26 Research Goals 26 Hypotheses 27 CHAPTER IV: PROCEDURE 29 Experimental Design 29 Data Collection 42 Data Transfer ^^ Data Reduction -V46

PAGE 7

Page CHAPTER V: RESULTS 48 Analysis of Pupillary Reactions 48 CHAPTER VI: METHODOLOGICAL IMPROVEMENTS: THE FEASIBILITY OF COMPUTER CONTROL AND DATA SAMPLING CHAPTER VI 1: IMPLICATIONS FOR FUTURE RESEARCH Attitude Theory Research on Stimulus Presentation Field Research Analysis Results Cost/Benefit Analysis Conclusions 72 82 82 83 84 Laboratory Research ^^ 87 87 CHAPTER VIII: SUMMARY AND CONCLUSIONS Goals Method ^^ 89 89 90 91 BIBLIOGRAPHY 93 APPENDICES APPENDIX I: PREAND POST-TEST QUESTIONNAIRES 102 APPENDIX II: BIOGRAPHICAL QUESTIONNAIRE 107 APPENDIX III:PROCEDURE FORM ""^^ BIOGRAPHICAL SKETCH "'^^ VI-

PAGE 8

LIST OF TABLES Page 56 2-1 Summary of Pupil lometric Studies 25 4-1 Affective Ratings (Independent Judges) and Luminosity Scores of Slides 37 4-2 Order of Stimulus Presentation 41 5-1 Prediction Variables Used 51 5-2 Analysis of Variance of Questionnaire Results for 4' Groups of Slides 54 5-3 Last Step of Multiple Stepwise Regression 54 5-4 Stepwise Linear Regression 3-Second Percent Change in Pupil Size 5-5 Last Step of Stepwise Multiple Regression 58 5-6 Stepwise Linear Regression, 9-Second Percent Change, 59 Pupil Size 5-7 Correlations Between Pupil Change and Stimuli 60 5-8 Summary by Stimulus Groups of S-Second Heartrate Regressions ^^ 5-9 Summary by Individual Stimuli: Stepwise Linear Regressions with 9-Second Heartrate as Dependent Variables ^^ 5-10 Summary by Stimulus Groups: 9-Second Percent Change in Heartrate ^^ 5-11 Stepwise Linear Regression of 3-Second Percent Changes in Blood Pressure ^^ 5-12 Stepwise Linear Regression of 9-Second Percent Changes in Blood Pressure ^5 5-13 Stepwise Linear Regressions of 9-Second Percent Changes in Skin Potential 5-14 Results by Hypothesis 65 6-1 POP 11/20 System with Dataphone Linkage to IBM 360/65: Cost/Benefit Analysis: Benefits 78 6-2 POP 11/20 System with Dataphone Linkage to IBM 360/65: Cost/Benefit Analysis: Costs 79 vn66

PAGE 9

LIST OF FIGURES Page 4-1 Procedure 30 4-2 The Experimental Environment 31 4-3 Equipment 33 4-4 Recording Equipment 34 4-5 The Experiment in Progress 35 -vm-

PAGE 10

Abstract of Dissertation Presented to the Graduate Council of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THE POTENTIAL OF THE PUPILLARY RESPONSE IN BUSINESS RESEARCHAN INVESTIGATION OF METHODOLOGY AND AUTONOMIC CONTAMINATION By Robert Roy Bell August, 1972 Chairman: William M. Fox Major Department: Management and Business Law The purposes of this dissertation were to replicate findings previously reported in the field of pupillometrics , develop a linear model explaining the parameters contributing to changes in pupil size, examine the feasibility of utilizing computer sampling techniques in the collection of pupillometric data, and to catalog physiological responses as covariates of the pupil. A problem limiting advanced applications of pupillometric techniques has been a condition known as autonomic contamination, a phrase describing the dilation of the pupil when exposed to anxiety-arousing stimuli. A goal of the dissertation was the utilization of several physiological indicators to define differences between pupil responses under autonomic contamination and certain other conditions. The study involved presentation of a set of visual stimuli (slides) to a group of 29 subjects, and incorporated a randomized block design. Stimuli were divided into four tyoes: pleasant, unpleasant, neutral, and anxiety-arousing. Light intensities were equated for the 16 slides used in the study. Tv;o startle stimuli were used as part of the anxiety-arousing -ix-

PAGE 11

set. Environmental factors such as room light intensity and sound levels were controlled. A television pupillometer was used to monitor changes in pupil size. Other physiological variables measured were heart rate, systolic blood pressure, and skin potential. A dynograph direct-wiring chart recorder recorded continuous samples on each of the physiological variables. These data were converted from analog (continuous) form to digital (discrete) form by a Bunker-Raymo BR-330 process control computer, which took samples of the data at one-second intervals throughout the study. The computer controlled the timing and presentation of slides during the experiment. A linear model of the factors contributing to pupil size changes was developed for analytical purposes. Step-wise linear regressions, the chi-square test, and analyses of variance were used to test the data. Other data analyses involved time-lags and descriptions of response magnitudes of the physiological variables under differing types of stimuli. Results raised questions about the usefulness of current pupillometric techniques in business research. Variables in the regression models produced multiple correlations of 0.62 and 0.57, respectively, for 9-second and 3-second percent changes in pupil size. The corresponding explained variances of less than 40 percent indicate that there is a great deal of reactivity in the pupil which was not explained by the variables considered. Furthermore, the direction of pupil response to the four types of stimuli did not follow the patterns which have been previously reported by proponents of pupillometric techniques. It does appear that other physiological variables exhibit some amount of consistency in their reactions to anxiety-arousing stimuli.

PAGE 12

This raises the hope that one of these responses could be used as a covariate with pupil change to indicate situations in which autonomic contamination exists. Although the results tend to question the use of current pupillometric techniques in business applications, there are certain areas of laboratory research where they may still be useful. A cost/benefit study of the feasibility of using a mini-computer in the gathering of physiological data in the laboratory v/as performed. Results indicate that the use of computers as a laboratory tool can be justified in many circumstances.

PAGE 13

CHAPTER 1 INTRODUCTION In any organization, the processes necessary for ongoing activity and success of the enterprise are carried out by human beings. The study and practice of management, therefore, has implicit in it a need for understanding peop1e--how they work, how they think, and how they can be most effectively utilized for progress toward organization goals. As Likert has said, "of all the tasks of management, managing the human component is the central and m.ost important task, because all else depends upon how well it is done" (1971, p. 197). Information on many different aspects of the human variable has been compiled. One of the most elusive, most hard-to-get-at areas of human behavior lies in the area of attitudinal and emotional feelings held by people. Yet these are unquestionably among the most significant determinants of human behavior. The research described in this paper was an attem.pt to further refine a new and promising technique for assessing these types of variables. Data Assessment Problems The internal states of individuals are probably among the most difficult of phenomena to measure. Many attempts have been made to develop unbiased yardsticks of the emotional and attitudinal responses of peoole. Conventional measures of these responses (such as naoer-and-oencil question-1-

PAGE 14

-2naire-type surveys) have been criticized on the basis of their inability to differentiate "true" reactions from those containing elements of artifact. In many situations, conventional measures are still relatively accurate and useful. In some situations, however, particularly in industry and in market research, respondents may have some reason to conceal their true attitudes and, in these situations, the traditional approach to measurement may be limited in its usefulness. Lapiere (1934) was one of the first to report a discrepancy between how individuals, in response to a questionnaire, said they would act, and how they actually behaved. Cook and Selltiz (1964) presented a more comprehensive analysis of the problem of obtaining honest and valid questionnaire reactions from respondents. They noted: Susceptibility of overt response to distortion--that is, the possibility of discrepancy between private and overt response--would seem to be a function of three characteristics of the [measurement] instrument: the extent to which its purpose is apparent, the extent to which the implications of specific responses are clear, and the extent to v/hich responses are subject to conscious control (p. 222). The problems of social desirability (responding with the socially "correct" answer: Rosenthal, 1966), demand characteristics (cues which convey the goals of the study: Orne, 1969), and evaluation apprehension (fear of being negatively evaluated: Rosenberg, 1969), have also been suggested as contributors to bias in most types of questionnaire measures, Krugman (1964), in a marketing research study, surveyed results which show how social desirability and demand characteristics interact to distort several types of questionnaire data.

PAGE 15

Other types of methodologies used in attempts to assess emotional and attitudinal responses, such as the projective techniques used by flcClelland, have proved to be somewhat useful. Their validity coefficients are usually low, hov/ever, perhaps due to the fact that scoring of a subject's responses is so difficult. The training required to be able to use projective assessment techniques is long and rigorous, and the scoring methodology at best presents many areas where errors of interpretation or other experimenter effects could bias data. The score on some types of projective techniques, in addition, is to a large degree a function of the vocabulary of the respondent (see McClelland, 1969). Both questionnaire methods and projective techniques, therefore, are subject to several sources of assessment error. The Potential of Pupillometrics The development of pupillometric measures of attitudinal/emotional response (Hess and Polt, 1960; Hess 1965) presents the hope of circumventing some of the errors associated with paper-and-pencil questionnaire types of measures and projective techniques. The theory of pupillometrics is based on the hypothesis that the pupils dilate in response to pleasurable or favorable stimuli, and contract in response to negative or distasteful ones. Since the eye is part of the autonomic nervous system, individuals presumably cannot consciously control changes in the pupil. If it is The research on conditioning of the puoillary response has nroduced equivocal results. Several investigations (Cason, 1922; Metzner and Baber, 1939; Girden, 1942; Crasilneck and McCranie, 1956) reported oositive results in attempts to condition the pupil. It seems that just as many studies, however, reported contrary findings (Stickle and Crenshae, 1934; Wedell, et al., 1940; Hilgard, et al . , 1949; Young, 1954). Even when conditioning is achieved, however, the process is quite long, and appears to be limited to a specific type of individual .

PAGE 16

possible to rrisasure changes in the pupil, and if the logic mentioned above is valid, then it should be possible to obtain a true or objective measure of attitudinal responses to a stimulus. Pupillometric assessment should offer several advantages over other types of reaction indicators. The response to the stimulus is very fast --usually beginning in less than a second. As mentioned previously, it is difficult to fake a pupillometric response, since pupil changes are autonomic (i.e., not subject to conscious control). A third possible advantage may stem from the bi-directional changes of the pupil--while most physiological indicators give only an indication of the size of the response, pupil changes shov; both size and direction. Previous work with physiological variables, such as that done by Cooper and Pollack (1959) on prejudicial attitudes and the galvanic skin response (GSR) might have derived more benefit from the bi-directional pupillary response approach. Traditional Pupillometric Techniques The methodology employed in pupillometric experiments has been outlined by Hess and Polt (1950) and Hess (1965). Briefly, a pupillometric experiment requires some type of stimulus input device, and some type of device for recording pupillary change. Most pupil lom.etric experiments have involved visual stimulus presentation, and have therefore utilized slide projectors as stimulus input devices. Other types of stimuli which have been used include liquids (for taste research), stilland motion-oictures (for market research), sounds (auditory research), and several types using

PAGE 17

-5cognitive information processing, physical work, or startling sounds as independent variables. Most experimenters use rapid frame cameras to take pictures of the eye, and measure changes in pupil size from the pictures. In the most common type of pupillometric study, the subject is seated before a rectangular box with a view screen in one end. In order to minimize measurement errors resulting from head movement, the subject's head is usually immobilized (placed in a chin rest, sometimes with an elastic strap holding the forehead against a bar). Equipment for projecting slides into the view screen is situated outside the box. A mirror system whereby a motion picture or rapid frame camera can photograph changes in the eye is used. The mirror is placed below the subject's line of sight, at a 45° angle, so that a camera mounted in the side of the box can photograph the eye. Since the eye is highly reactive to light changes, slides are usually controlled for light intensity, and photography is performed using infra-red equipment. When developed, the photographs are measured by hand, using calipers or millimeter scales or grids. Pupil lometrics in Business Research As will be seen in Chapter II, pupil lometrics was first employed as a business research method in the field of marketing research in the 1960's. The potential of pupil lometrics in this and other business fields will be discussed below. The Personnel Function The personnel function is an area of managerial decision making which may benefit from more accurate information about attitudes and values held

PAGE 18

-6by potential and present err.ployees. This statement must be qualified by the possibility that specific jobs may not require that prospective employees' attitudes/values be evaluated, simply because measures of these variables have been shown to have little predictive validity— i .e. , have little or no relation to success on the job (Porter and Lawler, 1959). A question that has not been ^swered, however, is whether the low validity coefficients are caused by no relationship between the causal and dependent variables, or because poor measurement techniques have produced inaccurate reading of the variables. There are some types of managerial positions where specific attitudes and values have been shown to be critical components of success (Levinson, 1964). Certain positions, for example, require that incumbents be able to perform effectively in situations involving substantial uncertainty and high risk. Many individuals have to assume responsibility for handling and committing large sums of corporate funds. Other attitudes and values which may be important include those concerning minority groups, corporate ethics, social responsibility, or those necessary for working in "organic" forms of organizations. Although pupillometrics is not today at a stage where attitudes and values such as those mentioned above could be readily m.easured, it is this type of difficult measurement which is contemplated for advanced applications of the technique. It may be that physiological measures, when coupled with other types of assessment, will provide practitioners with the type of data needed to evaluate candidates along these dimensions of "personality".

PAGE 19

-7Certain ethical questions andquestions of acceptability are raised when industrial applications of physiological measurement systems are contemplated. In organizations where high level job candidates are normally conducted through some type of psychiatric or psychological evaluation procedure, pupillometrics may be accepted as a matter of course. In organizations where the technique may be deemed undesirable, pupillometrics may still contribute meaningfully. One of the hopes of developing an easily useable and accurate method for evaluating attitudes and values is to use it as a tool for validating other types of devices which may be useful or more acceptable for some types of assessment problems. Another application related to the personnel function lies in the area of human factors engineering/work design. Pupillometrics has been shown to be a useful indicator of certain types of stress and anxiety levels (Hess, 1968b), and of noise levels (Nunnally, et al . , 1967). Pupil response systems might prove to be valuable aids in designing jobs and work environments through the study of their effects on physiological systems. Marketing Research Perhaps pupillometrics is one of the oldest known marketing tools. Hess (1955) noted that Chinese jade dealers had employed the technique for centuries, watching their client's eyes to tell when his interest in the product was highest, then making the sales pitch. More scientific studies of the pupil in marketing research were conducted by Brandt (1945), who used "ocular photography" to measure responses to advertising messages.

PAGE 20

More recently, pupil dilation and eye motion (or ouoil track) have been suggested to be useful in measuring package design effectiveness (West, 1962), television commercial interest peaks (Krugman, 196^a), and product resoonse ( Business Week , 1967). Krugman (1964 previewed a series of studies where pupil response was shown to be indicative of interest in products such as sterling silverware and greeting cards, and also presented data supporting inter-subject consistency in pupil response rankings and sales rank data. Hess and Polt (1966) performed research on taste stimuli which could be construed as product preference research, and were able to demonstrate correlations between pupil size and expressed preferences for certain drinks. The same study also found that both strong positive and strong negative aversive taste stimuli dilated the pupil. This finding lends credence to the hypothesis presented by some researchers that magnitude of response, not direction, may be the most important indicator of feelings. Halpern (1967) presented data on pupil responses to TV commercial and packaging, and noted that contractions were usually found to be associated with stimuli which "lack the power to interest or arouse the viewer" (p. 7). Hess (1968a) reviewed additional successful applications of pupillometrics to advertising and packaging research. Scope of the Problem Area Methodological Improvements Methodology is a factor limiting the overall usefulness of most new techniques. In pupillometrics, areas needing improvement are methods of data collection and analysis. Typically, photographs of the eye are taken

PAGE 21

-9continuously throughout a pupillonietric experiment. These ohotogranhs must be developed, and are then measured by hand to determine pupil diameter on each photograph. This v:riter has previously demonstrated, on the basis of a simple study, that a possibility of making major errors in data interpretation exists when experimenters measure pupil diameter by hand (Bell, 1971). This problem of expectancy error is compounded by the sheer number of measurements to be made. For example, a study involving 20 subjects, with 15 stimuli (and 15 control stimuli) shown for 10 seconds each, with photographs taken at 1-second intervals during the experiment, would require the development and measurement of 6,000 photographs. Further, a lag time exists between the time the experiment is conducted and the time the experimenter determines whether his data are useable--i .e. , whether the camera and film were operating properly. One of the major goals of the research reported in this dissertation was the development of computer sampling techniques to circumvent the tedious and perhaps unreliable methods of data collection mentioned above. The method used in this research involved a TV Pupil lometer (a video camera, capable of continuously recording a picture of the eye and measuring pupil diameter) tied into a Dynograph recorder (a device providing charts of pupil and other physiological changes) and a Bunker-Ramo BR-330 process control hybrid computer. The computer sampled data once per second, and converted the data from analog (continuous) form to digital (discrete) form through the use of an analog-to-digital converter. The feasibility and general applicability of this technique for laboratory and non-laboratory settings is discussed in the paper.

PAGE 22

-10Autonomic Contamination of the Pupil Response A major problem associated with pupil lometrics at the present state of the art is caused by reactions of the autonomic nervous system to certain types of stimuli. Instead of being a "clean" indicator of a favorable response to stimuli, pupillary dilation is also caused by some very nonpleasurable stimuli, such as fear and pain (Hess, 1968a) and strongly distasteful liquids (Hess and Polt, 1965). Since pupil size changes associated with most stimuli are relatively small, it is highly possible that autonomic reactions could give investigators completely reversed data from what they expect. A high degree of fear associated with an attitude object, for example, could, by causing pupil dilation, indicate a very favorable respone to the stimulus. To achieve a better understanding of why these responses occur, and how they might be "controlled" (partialled out of the data analysis), a discussion of the autonomic nervous system is necessary. The autonomic nervous system The word autonomic has been defined as "acting independently of volition." This definition describes the workings of the autonomic nervous system (AfIS) — "activities are largely involuntary, and we are usually unaware of them" (Sternbach, 1966, p. 14). The ANS is generally concerned with the regulation of the visceral system of the body, and attempts to maintain "homeostatic equillibrium" in the face of varying external factors

PAGE 23

-n affecting the body (Gellhorn, 1943, d. 195). The ANS can be broken down into two antagonistic sub-systems, the synioathetic nervous system (SNS) and the parasympathetic nervous system (PNS). The SNS generally provides emergency responses, while the PNS attempts to mediate or slow down autonomic activities, and restore normal metabolism (Sternbach, p. 23), Autonomic innervation of the pupil The eye and the pupil are "doubly bound", being innervated by both the PNS and the SNS. The dilator muscles in the pupil are innervated by the SNS, while the sphincter muscles, which constrict the pupil, are controlled by the PNS (Milner, 1970, p. 177). The size of the pupil is controlled by the antagonistic interactions of the two sub-systems. In general, excitation of the sympathetic system causes contractim of the dilator muscle, which in turn dilates the ounil, while parasymnathetic excitation causes the sphincter muscle to constrict the pupil (Grossman, 1967, p. 170). Several investigators believe there is an interaction

PAGE 24

-12between the PNS and the SNS v/hich causes dilation, Adler (1959, p. 176), for example, believes that inhibition of the sphincter and contraction of the dilator muscle act together to cause pupil dilation in man, while the sphincter alone contracts the pupil. It appears that dilations of the eye resulting from fear, however, occur not from a joint working of the SMS and PNS, but from a physiological dominance of the SNS over the PNS. That is, in times of high states of fear or pain, the sympathetic system "takes over" control of the eye (Isaacson, et al., 1971, p. 235; Adler, p. 190). This finding will be the basis of the experimental design to be described later in this paner. It should be possible, if the finding is valid, to monitor the SNS and PNS and observe variables other than the pupil (heartrate, blood pressure, skin potential, for example) to define levels of SNS and PNS activation, and thereby obtain insights into the pupillary response. When both systems are normally active, the findings hypothesized by Hess should occur. When the SNS dominates, however, we should expect findings contrary to those associated with "normal" affective reactions of the pupil. If levels of activity of other autonomic variables such as the ones mentioned above can be shown to correspond with the sympathetic dominance, then the fear and pain responses can be partialled out of the pupil response by concurrently monitoring other variables.

PAGE 25

-13S u mm a ry It appears that pupillometrics is a fruitful tool for conducting research on the emotional and attitudinal feelings held by people. The technique may have applications in management and marketing research. Methodological shortcomings and autonomic contamination are two confounding variables which presently limit its usefulness. The goals of the study reported in this paper are to improve the methodology of pupillometrics and reduce the problem of autonomic contamination.

PAGE 26

CHAPTER II PREVIOUS APPLICATIONS OF THE PUPILLOMETRIC TECHNIQUE One reason some researchers tend to disregard the pupil response as a valuable tool is that it has been shown to be a general indicator of many different typesof processes (Nunnally, et al . , 1967). As this literature review will show, pupil change has been used as a dependent variable in altitude research, psychiatric research, marketing research, and gustatory (taste) and auditory research. Puoil size has been shown to be a general indicator of physical and mental activity, emotional arousal, and interest value of visual stimuli. Notwithstanding the "generality" of the m.easure, ouoillometrics still appears to be a potentially valuable tool for business research, given proper controls. This chapter will review in greater detail the investigations which have contributed to our present knowledge of puoillometrics . Historical Develooments The contention that the puoil of the eye is a reflector of emotion has been held for some time. As early as 1765, Fontana described studies showing that pupil dilation followed the introduction of fearful or painful stimuli, even in the presence of bright light. Bender (1933) photo-14-

PAGE 27

-15graphed subjects' pupils in the presence of simultaneous bright light and emotional stimuli, and also found a "psychic" pupil reaction. Attitude Theory In 1960, an investigation by Hess and Polt uncovered a relationship betv/een pupil change and the interest value of visual stimuli. In this study, the experimenters displayed ohotograohs of male and female pinups, a baby, a mother and a baby, and a landscape scene to male and female subjects. Pupil dilation was largest for men when viewing the female pinup, while the v/omen showed greater ouoil response to the pictures of the baby, the mother and baby, and the male pinup. This study marked the beginning of a concerted effort on the part of Hess and his associates to study the eye as an indicator of emotional reaction. In 1965, Hess summarized the results of most prior studies of attitude and pupil size. Some of these investigations, using aversive stimuli such as pictures of sharks, cross-eyed and crippled children, or strongly unfavorable political statements, caused pupil constriction of over 3 percent change in average diameter. Other negative stimuli, such as sips of unpleasant tasting liquids, also caused constriction. Hess also suggested, on the basis of preliminary data collected during the 1964 political campaigns, that the pupil response might be useful as an indicator of attitude change (page 5). Pupil response was shown to be canable of differentiating between heterosexuals and self-admitted homosexuals in another study, (Hess, Seltzer and Shlein, 1965). In this nilot experiment, subjects were shown

PAGE 28

-16slides picturing males and females in varying states of undress. Four out of five homosexuals had larger pupil responses to pictures of males (some up to 35% larger) while five out of five heterosexuals had larger pupil responses to pictures of females. A problem exists in this study, however, in that the larger pupil response to male pictures on the part of admitted homosexuals may be the result of a type of expectancy effect. Once subjects had admitted homosexual interest prior to presentation of the stimuli, the dilations could be physiological activation caused by something other than higher sexual interest in the male pictures— something similar to the well-known skin flush when one becomes aware he is being observed. In a dissertation, Wocdmrnfee (1C65) evaluated the punil response as a means of mieasuring attitudes toward Negroes, and found that affectrelated responses, while significant during early presentations, diminished during repeated trials. He did find a significant difference between mean responses of anti-Negro and egalitarian subjects, thus providing support of Hess' hypotheses. The repeated trials investigation also produced disheartening results concerning the stability of the pupil response, which was measured to be around 0.30. Hess (1968b) also described the peculiar results of the pupil's fear response, and discussed investigations showing that high galvanic skin response (GSR) levels were a concomitant of ouoil dilation from shock or fearful stimuli. When GSR "dropped out", punil dilation turned to constriction (p. 581). In another article (1968a), he suggested that fruitful future research should concern itself with investigations of

PAGE 29

-17the relationship between the pupil and non-visual senses, and between the pupil response and other physiological measures. Pupil Size and Mental Activity Data showing that pupil size was directly related to mental activity were presented by Hess and Polt (1964). This exoeriment, where subjects were presented with mental mathematical problems of varying degrees of difficulty while their pupils were photographed, shov/ed that mean pupil dilation was highly correlated with problem difficulty. Word abstractness was also shown to be related to pupil size v/hen subjects were asked to form mental images of abstract and concrete words (Pavio and Simpson, 19G5), The Pavio and Simpson study found no relationship, however, betv/een the pleasantness-unpleasantness of a word and pupil size (p. 55). Another study corroborated the first Pavio and Simpson findings in a study on pupil size and memory load, finding that "pupil diameter is a measure of the amount of material which is under active processing at any time" (Kahneman andEeatty, 1966, p. 1585). Some studies reported in 1967 included one by Kahneman and Bailev replicating the "information load and punil size" results reported in 1965; a report by Hudd and Anderson that ouoil diameter may be a mechanism of perceptual recognition threshold, and a study by Peavler and McLaughlin on the relationship between novel stimuli and pupil size, all of which generally supported the results described in the above paragraphs. Polt (1970) described another experiment showing that subjects exerting greater mental effort had correspondingly greater pupil dilation.

PAGE 30

-18Two dissertations written at the University of Oklahoma (Clark, 1971, and Johnson, 1971) investigated ounillary responses during shortterm memory tasks. Both replicated the finding that amount of cognitive processing and pupillary diameter are highly related. Psychiatric Research A study in the field of psychiatry (Rubin, et al . , 1953) suggested that pupil response may be a useful psychiatric diagnostic tool. Results showed that characteristic pupil changes in children with cystic fibrosis were significantly different than those of normal children. The study also discussed the effect of age on pupillary reactivity, noting that response characteristics tend to change as people become older. Research reported by Sheflin (1969) investigated punil reactions of hospitalized schizophrenics and found that, if pupil dilation is interpreted -as a m,easure of interest, male paranoids are not more sexually interested in men than in women--a finding contrary to that held by most psychiatric researchers. Boersma, et al . , found that pupillary response was useful for studying differences in cognitive information processing in educatable mental retardates and normal children (1970, d. 143). During the same time, Tanck and Robbins investigated the predictability of the punil response using the Dream Incident Technioue (DIT), the Edward Personal Preference Schedule (EPPS) and the Repression-Sensitization Scale. They found the DIT to be fairly predictive of nupil r^snonse, while scores on the other two scales showed no releticnshio to ounil changes.

PAGE 31

-19In another study, Good and Levin (1970) failed in an attemot to show that "sensitizers" would exert more perceptual viailancejOnerationalized as pupil dilation than would "repressers", although they did corroborate previous findings on affective arousal and pupil dilation (p. 631). More research showing pupil changes may be useful in clinical programs was reported by Kennedy (1971). This study indicated that pupil dilation could be used as an indicator of the therapeutic impact of treatment on chronic alcoholics, and may be useful for predicting recidivism (recurrence of need for treatment). Emotional Reactions and Pupil Size In the 1965 study discussed earlier, Hess also reported that dilated pupils in a photograph of a young woman caused greater preference for that picture than for one where the woman's pupils were of normal size. Hicks, et al . , (1967) in an attempt to replicate Hess' findings, reported that pupil size showed no relationship to expressed attraction, while facial angle did have a significant effect (p. 388). Simms, however, found thatnupil size in pictures had a significant effect when sex differences were controlled. Married subjects showed greatest dilation to ODPosite-sex pictures with large punils, and dilated least to like-sex ohotogranhs with large pupils (1967, p. 55^). Data presented by Starr and Willis (1967) also supported the dilation-attraction hypothesis. Other studies of the relationshin between oupil size and the emotions were reported in 1966. In a dissertation, Guinan demonstrated

PAGE 32

-20that emotionality (operationalized through slide presentations of emotional words) did have significant impact on punil size, and that a significant interaction between time and emotionality occurred. Barlow (1969 and 1970) reported a positive correlation between expressed preference for photos and puoil dilation, and found that aversive stimuli did cause constriction of the pupil. In another study, he found the pupil response to be in perfect agreement with verbal preferences in a study of political candidate preferences (Barlow, 1969b). The Hess (1965) article also reviewed some studies of the Duoil response as an index of "motivation". In this study Hess and his associates deprived some subjects of food for four or five hours. The mean pupil response of this group to nictures of food was tv-yo and a half times larger than the mean response of a control group which had eaten within an hour prior to the experiments. Whether or not the term "motivation" is appropriate in this context, it does aopear that these data support the Hess and Polt (1960) contention that pupil response is an index of interest value of visual stimuli. Hess and Polt (1966) presented data showing that the pupil may be an indicator of taste sensitivity and taste oreference. Five different orange beverages, together with a control beverage (water) were given to subjects. Pupil resoonse was positively correlated with other rankings of beverage preference. Dooley and Lehr (1967), however, criticized the methodology employed by Hess and Polt, noting that controls normally employed in gustatory research were not used, and that the results may

PAGE 33

-21have therefore been biased. They also noted that order effects could have biased data. In a reply, Hess and Polt (1967) stated that they had investigated order of presentation as a source of artifact, but had found no effects in visual, olfactory, auditory, or tactile stimulation experiments. Marketing Research In the early 1960's, Marplan, a marketing research organization, became interested in commercial applications of the Hess pupil response measuring system. The studies they conducted, usually in conjunction with Hess, were reported by West (1962), Krugman (1964) and by Sponsor magazine (1964). Briefly, findings were that punil resnonse was related to sales patterns of the products tested (watches, greeting cards, silverware) and was also related to coupon returns on advertisements tested. These data tended to support the thesis that puoil responses and attitudinal/emotional responses v/ere related, and led to the hope that further research could develop a new and useful marketing tool. The Marplan studies also showed that the pupil response could be used with moving pictures as well as still frame photos. A continuous measurement of the pupil, called the pupil interest track, was recorded while subjects watched television commercials. By monitoring puoil dilation and constriction, investigators were able to determine where interest peaks were highest and where interest had waned. Strategic placement of the main selling point, it was hoped, could be achieved

PAGE 34

-22by locating the message directly after the high point on the puoil interest track. Puoil track also allowed the makers of the commercial to evaluate their product's capacity to keen viewers watching that channel and that commercial (S ponsor , 1954). As mentioned in Chapter 1, a further use of pupil lometrics to marketing researchers lies in its ability to overcome social desireability problems associated with typical written or verbal questioning procedures. Pupil change was shown by the Marnlan studies to have a mathematically higher relationship to sales than do verbal and written surveys, although the differences were not always statistically significant (Krugman, p. 17). Hal pern described the potential usefulness of pupil lometric techniques in "before-and-after" marketing experiments. He noted (o. 320) that, often, when subjects are called uoon to make "after" responses, their replies are biased by knowledge of their "before" responses, The interview and questionnaire techniques commonly used by marketing researchers also come under fire, as discussed in Chanter 1. Halpern conducted studies showing that pupil dilation was as "good" a response index as were questionnaires for these types of experiments, while not possessing the bias and social desirability drawbacks inherent in questionnaire and interview techniques. Pupil dilation was again shown to be an indicator of interest in package design (n. 321). A "questioning" article by Elackwell, et al . (1970) surveyed the state of pupil lom.etric appraisal in marketing research, and noted that the technique was being used before being well understood. They stated

PAGE 35

-23that there exist both "unfounded optimism" and "ill-timed skepticism" concerning the pupillometer and its applications, and echoed the call of many investigators for more research on the technique and its usefulness. Dissenting Votes While these studies were proceeding, publications in the field of ophthalmology appeared which have significance for pupil lometric research. In particular, the works of Lov/enstein and Lowenfeld (1961 and 1952) made considerable contribution to knowledge about ouoil functioning, and sparked controversy over some of their findings. Their discussion of the "pupillary reflex dilation", defined as an increase in pupil size caused by sensory or emotional stimuli, or by spontaneous emotions or thoughts, supports the affect-dilation results obtained by Hess. Ophthalmologic and physiological treatises on pupil constriction, however, almost uniformly disagree with the Hess assertion that constriction is affect-related (Lowenfeld, 1966). According to Woodmansee (1965), pupillary constriction can occur for any of three reasons 1. The light reflex, whereby "an increase of light flux to the retina results in a flow of efferent impulses from the oculomotor nucleus to the sphincter muscle, thus actively constricting the pupil." 2. The near-vision reflex, which occurs when the eyes shift from focusing on a distant object to focusing on an object less far away. 3. The relaxation of the dilator muscle, caused by a decrease in arousal following close on the heels of a high-arousal, reflex dilation condition (cage 7).

PAGE 36

-24It is this third phenomenon which Lowenfeld and Woodmansee hypothesize as a possible cause of the constriction measured by Hess. Another type of criticism was contributed by Nunnally, et al . (1967). They described a series of experiments in which pupil size was shown to be related to muscle tension, sound levels, affect, novelty of pictures, and fear (threat of gunshot). They noted that pupil size seems to be directly related to fear through autonomic contamination (p. 154), as discussed in Chapter 1. The authors concluded that puoil response was such a generalized reaction as to be almost without value for studying complex emotional stimuli and responses. Summary The record presented above tends to obscure any patterns which may have emerged from the past decade of study. A summary content analysis of these articles reviewed in this chanter is presented in Table 2-1. Perhaps the major conclusion that can be drawn at this stage is that the pupil does respond to stimuli. Whether or not it is possible to accurately assess such things as attitude and emotional reactions is not yet clear. Many studies have presented evidence of ability to measure attitudes; others, just as rigorously controlled, have found no consistent relationship between attitudes/emotions and the pupil response. Before the question can be resolved, a much more precise methodology for measuring the variables of interest must be developed, and many of the artifacts now associated with the puoil response must be cataloged and controlled. The rest of this study is an attempt to contribute to these goals.

PAGE 37

-25TABLE 2-1 SUMMARY OF PUPILLOMETRIC STUDIES Authors Bell, 1971 BlackwGll, Gt al.,1970 Boersina, et al . , 1970 Bradshaw, 1957 Bradshaw, 1958 Dooley and Lehr, 1967 Francis and Kelley, 1959 Guinan, 1969 Halpern, 1967 Hess, 1965 Hess, 1968a Hess, 1968b Hess and Polt, Polt, Polt, Polt, al ToDic Hess Hess Hess Hess, Hicks and and and et et 1960 1964 1966 1967 1965 al., 1967 Holmes, 1967 Kahnenian & Beatty, 1967 Krugman, 1964a Krugman, 1964b Kimmel, 1967 Landauer & Feakes, 1967 •Lowenfeld, 1966 Nunnally, et al . , 1967 Pavio and Simpson, 1966 Rubin, 1951 Rubin, et al . , 1963 Simms, 1967 Tanck and Robbins, 1970 Woodmansee, 1965 Television Magazine , 1962 Business Week, 1967

PAGE 38

CHAPTFR III RESEARCH GOALS AND HYPOTHESES This chapter will define the objectives of the dissertation research, and will present the specific hypotheses which v/ere tested. Although they were tested in the null form, the hypotheses listed here are in the form of expected results. Chapter IV will describe the methodology of the experiment, and the collection and analysis of data, Research Goals 1. The validation and reolication of earlier works showing th?t the pupil responds differentially to various stimuli. 2. The development of a linear model describing the parameters contributing to pupil change. 3. Examination of several autonomic responses to favorable, unfavorable, neutral, and anxiety-arousing stimuli, with the objective of eliminating autonomic contamination as a source of error in pupil resoonse data. 4. Examination of an improved methodology for conducting pupillometric exoeriments, including computer control of stimulus presentation, analog to digital conversion of data, and real-time sampling and measurement of the pupillary resoonse, -26-

PAGE 39

-275. Collection and analysis of data for testing the following specific hypotheses. Hypotheses Pupillary Reaction Pleasant stimuli v/ill be accompanied by dilation, while neutral stimuli will cause no change in pupil size. Negative stimuli will cause a constriction of the pupil. Anxiety-arousing stimuli will dilate the pupil. Highly pleasureable stimuli will produce greater nupil dilation than will less pleasant stimuli. Autonomic Contamination Magnitude covariance Blood pressure changes will be similar in reaction when exnosed to pleasant stimuli and when exposed to anxiety-arousing stimuli. Heartrate changes will be similar when exposed to anxiety-arousing stimuli and pleasant stimuli. Skin potential changes will be similar in reaction when exposed to anxiety-arousing and to pleasant stimuli.

PAGE 40

-?8Response duration Pupillary dilation to fear-arousing stin-uli v/ill take longer to return to initial base level (at time of stimulus presentation) than will pupil dilation to pleasant stimuli, for a given level of stimulus intensity. Heartrate will take longer to return to base line in response to anxiety-arousing stimuli than to pleasant stimuli, for the same stimulus level . Skin potential will take longer to return to base line in response to anxiety-arousing stimuli than to pleasant stimuli, for the sam.e stimulus level. Blood pressure will take longer to return to base line in response to anxiety-arousing stimuli than to pleasant stimuli, for the same level of stimulus intensity.

PAGE 41

CHAPTER IV PROCEDURE Experimental Design The study was conducted under a randomized block design. Each subject served as a block, and received each of the 18 treatments (stimuli), with the order of presentation of treatments randomized for each block. The design will be discussed in greater. detail below. Subjects Twenty-nine persons served as subjects in the study. All were volunteer males between the ages of 20 and 40, drawn largely from two student groups in the College of Business Administration (the M.B.A. Club and Delta Sigma Pi). They were pretested through questionnfire to determine any known prior medical problems (heart trouble, eye deficiencies, etc.) which might have affected their responses. Those reporting no medical problems were asked to serve as subjects. The accompanying figure shows the flow of the procedure the subjects v/ent through. At the time of the study, each subject's visual acuity was tested to determine his ability to discriminate detail on slide stimuli. Since one object of the research was to determine the feasibility of -29-

PAGE 42

-30— +-> sO) rt5 -rSZ I/} JZ «0 TD O :^ (US'— =5 >, O t3 -M 4-> C S-r1/1 to rt3 3 a> ^ o +-> O) o (/) 4J U-) -o SOJ OJ O E >, in m Z5 +-> o "O to o O 4O GJ -u(J r— to U +-> 3 O) c Cn-i-o $CD C -D E •.lO +-> O-a S-rfO CD <0 n CL' i^ cr or fo «j a> ^ o c: +-> r— to -rO JC QJ q; $-a CD c t/1 O" > -•-> .4> to

PAGE 43

-31using pupillometric techniques in field research, the very general controls mentioned above were the only ones used to select subjects. It v/as assumed that close controls would be lacking in field research, using subjects selected from the general adult pooulation, and therefore any laboratory study using more specific controls would not be general izable to this type of population. Environmen t The studies were conducted in the Systems Engineering/Human Performance Laboratory at the University of Florida. Light levels, peripheral sound, and temperature were all controlled. The same two persons (the investigator and a laboratory technician) conducted all sessions of the experiment. Occassionally, other persons sat in on the study as observers. Figure 4-2. The Exoerimental Environment

PAGE 44

-32Apparatus Equipment layout followed a modification of the design used by Hess (1965, p. 48). Each subject was fitted with electrodes for measuring heartrate and skin potential. He was shown the measuring equipment, and the basic scope of the study was described. The subject was then seated in a dentist's chair, which allowed a certain degree of freedom, while restricting body movement. The subject looked into a rectangular viewing box, measuring 27 cm. by 32 cm. by 60 cm. The box was equipoed with a rear-orojection screen in the end opposite the subject. The walls, top, and bottom of the box were painted flat black to effectively eliminate all outside light. A black cloth covered the box and the subject during each run of the study. The end of the box at which the subject was seated had a 10 by 13 cm. opening which circumscribed the subject's head. A chin rest, used to minimize head movement during the experiment, was recessed in the open end of the box. Two Kodak Ektagraphic Model 850 H Carousel slide projectors, located 25 cm. to the rear of the view screen, projected slides on to the screen. The size of the projected image was reduced to 4 by 6 cm. to reduce eye search of the stimuli. The projectors were connected to a BR-330 hybrid computer v/hich controlled the timing and oresentation of the slides.

PAGE 45

33Figure 4-3. Equipment During the experiment, the subject wore a set of dictation headphones. The headphones, connected to a Revere-Wollensak tape recorder, presented uniform white (random) noise during the study. Just to the left of the subject's line of sight, a Whittaker Model 800 TV Pupillometer (TVP) was situated to take continuous readings of the size of the subject's left pupil. The TVP is a closed-circuit television system, incorporating a horizontal scanning technique and a signal processor which m.easures and displays pupil diameter. The TVP was equipped with a television monitor and remote controls which allowed the experimenter to adjust focusing so that proper settings could be found for each subject. The equipment operated in the near infra-red.

PAGE 46

-34with a low intensity near infrared source illuminating the eye. With the exception of one subject, the light source produced no reported discomfort in this study. Figure 4-4. Recording Equipment A Beckman Type RM Dynograph direct wiring recorder, capable of measuring up to six physiological variables simultaneously, was situated near the main equipment stand. Sensors on the subject were connected to the recorder, which displayed readings on a continuously moving chart and simultaneously transmitted data to the computer system described below. The recorder contained a timer unit and event market which were used to ^Source: Whittaker Corporation, Snace Sciences Division, Instruction Manual for Series 300 TV Pupillometers.

PAGE 47

35note event changes such as the introduction of new stimuli Figure 4-5. The Experiment in Progress The computer system monitored the experiment continuously and transformed the data from direct analog readout of the physiological variables to digital form, using punched paper take as an output medium. The data on paper tape v/ere then transferred to magnetic tape and analyzed using the IBM 360/65 computer. With the exception of data transfer programming, there v/as no human interaction with the data until after analysis had been completed. The measurement techniques used in the system were validated using "known" responses (such as artificial pupils of known diameter) before the experiment was begun. Changes

PAGE 48

-36were measured in tenths of millimeters, on a scale of 1.0-10.0, so the measurement system was sensitive to changes as small as 1 percent per second. Stimuli Stimuli for the studies reported in this dissertation were a set of black and white 35 mm slides. A series of 58 slides was developed for pretesting. Of these 58, 16 were chosen for use in the study. In the slide selection pretest, 14 raters judged the slides using semantic differentials developed from the manifest anxiety scale and from evaluative scales used by Osgood, et al . (1967). A copy of the rating questionnaire is included in Appendix I of this paper. The slides chosen for the study were those exhibiting high scores along a specific dimension and having low variability. The dimensions of interest were pleasant-unpleasant, fear/anxiety, and interest value. The slides chosen are described in Table 4-1. Scoring the questionnaires The set of semantic differentials shown in Appendix I was used to evaluate each slide. The set v/as made up of 3 subsets of scales, or factors. The pleasant-unpleasant factor was composed of the following semantic differentials: Disnusted-elated Gloomy-cheerful Hostile-friendly Unpleasant-pleasant Unfavorable-favorable Negative-positive

PAGE 49

-37Table 4-1 Affective Ratings (Independent Judges) and Luminosity Scores of Slides Stimulus Content Judged Affect^ Predicted Luminosity:

PAGE 50

-38Since each scale had a range of from 1 to 7, the possible range of scores v/as 6-42. If a person had checked all scales at the extreme "pleasant" side of the scale, the score value for the factor v/ould be 6. Conversely, checking all the scales at the negative end v/ould have produced a factor score of 42. Checking all scales at the "neutral" midpoint value would have produced a score of 24. The middle point appears to be a true "neutral" position on the pleasant-unpleasant dimension. The anxiety-arousing factor was composed of the semantic differentials: Unafraid-afraid Nervous-calm Insecure-secure Tense-relaxed The lov/est possible anxiety factor score for a slide v/as 5,. and the highest anxiety factor value was 35. Unlike the pleasant-unpleasant factor, it appears that the anxiety scale does not yield a true neutral position, if by neutral v/e mean lack of anxiety. Rather, this scale appears to produce a step-score, with degree of anxiety increasing as the score increases. A single semantic differential (interesting-uninteresting) was used to evaluate slides along the "interest value" dimension. To ensure that the grouoings of slides used as independent variables were indeed different from each other, the medians test was computed for each dyad of grouping, using subjects' questionnaire data. The groups in each dyad were significantly different from each other at the .01 level (for the pleasant-unpleasant dyad, X = 131.49; for the pleasant-neutral dyad, X = 131.49; for the pleasant-neutral dyad, X^=34.31; for the unpleasant-neutral dyad, X^ = 164.49; each dyad tested with 1 degree of freedom). After the 16 stimulus slides were settled on, Kodak l'% Gelatin neutral Density Filters were added to each slide, as needed, to develop closely balanced light intensitites. All slides and control slides used in the study had the same intensity, within limits, so that any pupillary

PAGE 51

-39response occuring could not be construed as a light reflex. Table 4-1 also presented the range of scramblea light intensities for the set of stimuli . Stimulus Presentation A blank control slide was projected on the screen while the subject was being seated in the chair and adjusted to the equipment. As soon as all sensors v/ere reading properly, a two-minute study similar to the main study was conducted to familiarize the subject with the equipment, timing, and general environment he would be in during the study. At the end of two minutes, he was allowed to rest and was asked about his comfort. This time also allowed the experimenters to make any necessary readjustments of the equipment. The subject was then cautioned about moving his head back from the headrest. He was also asked to contain swallowing and most eyebl inking to the time when control slides v/ere on the screen,' since these types of movements introduced errors into the measurement system. He was informed that, at some time during the experiment, he might hear a buzzer, and that it was part of the experiment. He was cautioned not to move away from the equipment at the sound of the buzzer. The subject was then asked to readjust his head into the chin rest, the camera focus was checked, a black cloth was placed over the subject and over the equipment, and the main experiment was begun. A blank slide was projected on the screen during the period described above. The blanic slide remained on the screen for a period of eighteen seconds after the computer had been given the instruction to begin the experiment. Once given that instruction, the computer controlled both

PAGE 52

-40slide projectors, and operated them so that no light flash occured during change-over from one slide to the next. To accomplish this, the computer was programmed to start changing the slide to be projected one half second before it instructed the other slide projector to remove the slide currently being projected. Every other slide in each projector vyas a black slide (one allowing no light to pass through) so that while one projector was showing a stimulus or control slide, the other was showing a black slide. This "flip-flop" process was developed to allow the reduction of the light flash mentioned above to a period of less than 1/20 second. Each stimulus slide was presented for 10 seconds, and was followed by a blank control slide presented for 20 seconds. It was hoped that the 20-second blank slide would allow response variables to return to normal levels before the next stimulus was presented. The order of stimulus presentation was randomly varied to control for order effects. Table 4-2 presents the order of each stimulus presentation. The slide runs took 480 seconds (32 slides, half for 10 seconds each, and half for 20 seconds). At the end of this time, the subject was given a "startle" stimulus, consisting of an alarm buzzer sounding. The buzzer was attached to the wood bottom of the arm rest used by the subject, and produced both a startling sound and a simultaneous startling vibration in the arm rest. Approximately 20 seconds later, the investigator pushed the "release" switch on the dentist chair, causing it to move very rapidly downward, producing a second startle response. Because of the raoid down-

PAGE 53

-41 •

PAGE 54

-42ward movement, the subject could not maintain his head in the chin rest, so pupil data was not collected for this second startle response. These startle stimuli always occured at the end of the set of slides, and v;ere always presented in the same order. Data Collection Method and Response Variable s One hour was assigned for processing each subject. Upon entering the Systems Lab, the subject was introduced to the experimenters, and was shown the equipment. The type of data to be collected was discussed in general terms. He was told that there would be time after the experiment to view the equipm.ent in more detail and to see the charted results of his study. After filling out the pretest questionnaire, electrodes were attached to him and he was seated in the dentist chair. The electrodes were in turn attached to the Dynograph recorder. A light-sensing pressure clip for measuring systolic blood pressure was attached to the subject's right ear. The subject was asked to place his head in the chin rest and acclimate himself to the experimental anparatus. While the subject was acclimating himself, the experimenters were adjusting the instruments and focusing the camera on the subject's left pupil. ^ As soon as all instruments seemed to be properly functioning. An interesting problem occurred when one subject noted that he v/as blind in the left eye. The camera was focused on the right eye, and no problems occurred as a result of this modification.

PAGE 55

-43the subject wasellowedto relax for a minute. He was then asked to readjust himself, and the 2-minute familiarization run vjas begun. After this run, the subject was allowed to relax for about 1 minute. During this minute, the carousels v/ere rotated so that the stimulus and control slides were in position for the beginning of the main study. The tape recorder was recycled, and all equipment v/as checked for proper functioning. The main study followed this rest period. Level of Measurement While slides were being presented, continuous line chart recordings were being made on the Dynograph recorder. The Dynograph had a recording capacity of 6 channels. In order from top to bottom of the recording paper, the six variables recorded v;ere: beat-by-beat heartrate; 60-second heartrate (extrapolated each second); pupil change; skin potential; blood pressure; and a light intensity reading for each slide, taken by a Weston Model 748 light meter installed inside the viewing box. The BR-330 computer sampled the continuous data at a rate of 1 sample per second for each of the variables being recorded, thus providing a digital sample. Beat-by-beat heartrate was not sampled by the comouter. A total of 2,560 samples was collected for each subject (512 samples for each of 5 response variables). These data were stored internally in the computer until the study had been completed, and were then transferred to punched tape. Later, the data were again transferred, this time to magnetic tape. Validation checks were made for each data reduction conversion, ensuring that errors in transfer were not made. Primary data analysis was done on the discrete data contained on

PAGE 56

-44the magnetic tape. The punched tape data and chart data v/ere used as back-up data sources, and v.'ere not subjected to much systematic analysis. Post-Test Questionnaire In an attempt to develop convergent validation of the autonomic responses (i.e., evaluate the character of each individual's autonomic responses from another type of measure) each subject v/as asked to complete a questionnaire rating the slides he viev/ed during the first part of the experiment. A copy of the questionnaire is included in Aopendix I of the dissertation. This questionnaire was used as a major comparison index of affective responses to the slides. The correlation betv/een this index and the pretest ratings made by independent judges v;as also computed, Debriefing and Equipment Recycling Since subjects were members of two student organizations, the possibility of discussion of the experiment prior to another subject's participation existed. Subjects v/ere asked not to discuss the experiment with anyone, and v/ere told that the investigator planned to give a debriefing and summary of the results to the group later in the quarter. Each subject was shown the chart output recorded during his participation, and the responses were explained to him. VJhile the subject was filling out the post-test questionnaire, the experimenters were transferring the data to punched tape, rearranging stimuli according to new random number sequences, resetting the comouter and the rest of the equipment, and preparing for the next subject.

PAGE 57

-45Data Transfer The computer-gathered data san:p1e was transferred from punched p?per tape to magnetic tape and then to ounched cards. Since this experiment v.'as a pilot study on computer sampling, a great amount of time was consumed in the develonment of FOHTFlAN and machine language programs for converting and validating data gathered in octal (base 8) form on the paper tape to hexadecimal (base 16) form on the magnetic tape and then to decimal (base 10) form for analysis and presentation. These conversions involved the Mohawk tape-to-tape converter, the IBM 1620 and System 360/65 computers. The conversions from different bases were made necessary because of the utilization of the BR-330 computer and its data recording sy^t^n^, if real-time sampling with IBM equipment and direct magnetic tape or punched card output had been feasible for this study, the conversion process would have been less bothersome. Data' transfer and reduction programming took almost 2 months to complete. Now that the conversion packages have been "debugged" and validated, a complete data transfer from an "on-line" samole of 30 subjects during an experiment to decimal data ready for statistical analysis can be accomplished in less than 3 hours. If the System 360/65 had been used, or if a direct tie-in to the 360/65 through some tyoe of terminal linkage device were used, data could have been transferred to a form ready for analysis in a matter of seconds from the time the data collection was completed. The advantages and disadvantages of such a real-time system, together with some cost/benefit data, will be discussed in a later section.

PAGE 58

-46At each stage of the data conversion process, output data were compared with chart data recorded by the Dynograph recorder to ensure that errors had not been made. Several programming errors stemming from reversed polarity on the Dynograph recorder/BR-330 interfacing were detected using this technique. In most cases, these errors were correctable with minor modifications in the conversion programs. Data Reduction Several forms of data were analyzed. In almost all cases, the 10-second time span during which each stimulus was presented formed the basic period for analysis. Since the BR-330 served as both a director of the experiment (changing slides at given time intervals) and as a data recorder, it could not take samples during the second it was changing slides. Consequently, no data sample was taken during the first second after a slide change had occurred. This left a 9-second sample of each variable when a stimulus slide was presented, and a 19-second sample when control slides were presented. Two basic measurements were used in defining types of resoonses. To test hypotheses on magnitude, the first measure compared absolute maximums during the relevant response period with absolute maximums during the prior control period. Second, percent change measures were incorporated. Using pupil size as an example, PD PD percent change in puoil size _ pQp ^ xa cxa_ for stimulus x^ on presentation a^ ^'^cxa

PAGE 59

-47where PD = mean pupil diameter during presentation ^^ a_ of stimulus x^ and PD = mean pupil diameter during control period ^^^ preceeding stimulus x^ on presentation a^. Regression coefficients v/ere used to test differences in direction of responses, as discussed in later sections. Two time periods were analyzed for each stimulus. A 3-second time span was analyzed because it was felt that the longer time periods conventionally used might be losing a substantial portion of the relevant pupil response due to a "damping" of the response over the long period. In this case, data collected in the first 3 seconds of each stimulus data sample were compared with data collected in the last 3 seconds preceeding the sample. The other time period involved a 9-second analysis, where data collected in the 9-second period following a stimulus were compared with data collected in the 9 seconds prior to the introduction of the stimulus. Reduction in Sample Size In several cases during the study, one or more problems occurred in the data sampling techniques. Several subjects, for example, had eyeblink rates which caused much of the data sample to be erroneous. In a few other cases, blood pressure and skin potential recording systems malfunctioned during the experime.'nt, causing data on those variables to be lost. Consequently, for most of the analyses conducted, the sample size was reduced to 22, and the number of observations ranged from 270 to 470 per variable. These will be noted in the appropriate places in results discussed below.

PAGE 60

CHAPTER V RESULTS As discussed in Chapter IV, the experinent incorporated a randomized block design, with 29 subjects serving as blocks and slide stimuli serving as randomized treatments. Most of the analyses discussed in this chapter utilized all 18 stimuli (16 slides and 2 "startle" stimuli) as treatments. Cases v/here the 18 v/ere collapsed into 4 treatment groups (pleasant, unpleasant, neutral, and fear-arousing) will be discussed separately. Analysis of Puoillary Reactions A Linear Model of Pupil Response In describing the pupil response to a given stimulus of the type used in this dissertation, the following linear model was developed: PR = B +8X +SX +eX +SX +6X +BX +6X +6X 11 22 33 itit 55 66 77 88 + BX+SX +SX +e 9 9 10 10 1 1 J 1 where PR = change in pupil size 6. = regression coefficients X = the initial value of the pupil (base level) 1 X = environmiental sound changes 2 X3 = environmental light changes -48-

PAGE 61

-49X^ = the orienting reflex patterns of the pupil X5 = color patterns in the visual stimulus Xe = complexity of the stimulus X7 = interest value of the stimulus Xe = pleasantness/unpleasantness of the stimulus X9 = spot light intensities in the visual stimulus Xio = scrambled light intensity of the stimulus Xii = endogenous subject variables e = error term As can be seen, the parameters of this very simplistic model emphasize that m;.ny things contribute to a given change in pupil size. The model served as a guide for controls used in the experiment and for the development of the set of linear predictors used in the statistical analysis. Better controls and the utilization of more accurate predictors should have caused an increase in the amount of explained variance in the dependent variable. The model also emphasizes some findings which are not always considered in experimental designs and analyses. Most researchers, for example, have not taken adequate account of Wilder's Law of Initial Values, which states that a variable's "...response to stimulation is a function of the pre-stimulus level..." at which the variable was operating (Sternbach, p. 44). When using percent change as a dependent variable, an error is introduced when this law is not considered. For example, a pupil with a diameter of 7,5 millimeters has little potential for large dilation (the familiar ceiling effect) while it has a much higher propensity to constrict in size and return to a more normal

PAGE 62

-50level of operation. A 2 percent dilation, therefore, might be much more significant at this level than a 10 percent dilation from a pupil which started out at a diameter of 3 millimeters. By using initial value as a predictor, and incorporating analysis of covariance, the model can "factor out" of treatment effects that oart caused by initial values. Similarly, anxiety scores can be factored out by covariance when the effects of pleasantness-unpleasantness as a oredictor are to be investigated. The specific model used to test the experimental data was: PR = B^ . B^\ S^X . B^X . s^X . B X . s^X . b X . b X '' 55 66 77 68 + S9X9 + SX +SX +SX +BX + R Y +S15X1S+ SX +SX +BX +6X +RX '' 16 1' 17 18^8 \s\, ^0 20 +B21X21+ SX +BX +BX +BX +BX 22 22 23 23 ^2/2. ^5%5 ^6 26 +627X27+ SX +SX +8X +BX +BX 2« 28 29 29 30 30 ^3/31 %2%2 +S33X33+ SX +6X +BX +r 3'' S** 35 35 35 36 Where PR = 3 and 9 second percent change in pupil diameter Bo = 6^. = regression coefficients ^r"^36 ^'^^^ parameters described in Table 5-1 e = an error term.

PAGE 63

-51Table 5-1 Prediction Variables Used Variable Number 1 2 3 4 5 6 7 8 9 10 n 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 Description Tyoe 3-second pupil percent change dependent 9-second pupil percent change dependent initial value moderator slide 1 predi slide 2 predi slide 3 predi slide 4 predi slide 5 predi slide 6 predi slide 7 predi slide 8 pred slide 9 predi slide 10 pred slide 11 pred slide 12 pred slide 13 pred slide 14 pred slide 15 pred slide 16 ctor ctor ctor ctor ctor ctor ctor ctor ctor ctor ctor ctor ctor ctor ctor predictor 3-second average---light intensity moderator 9-second average---light intensity moderator ctor ctor ctor ctor ctor ctor questionnaire score---pleasantness predi questionnaire score---interest value predi interacti interacti interacti interacti interacti on---light + initial value predi on---interest val. + fav. predi on — interest val. + iv predi on — iv light intensity predi on — iv + fav predictor

PAGE 64

-52Table 5-1 (Cont.) Variable Number Description Type 29 interaction — iv interest value predictor 30 interaction — interest value x lint predictor 31 interaction---iv x fav oredictor 32 interaction — iv x interest value predictor 33 interaction---light inten./iv predictor 34 interaction---iv/fav predictor 35 interaction---iv/interest value predoctor 35 questionnaire score — anxiety. moderator Several restrictions are inherent in the linear models approach adopted above. First, the essunption of linearity and linear effects may not be a good aoproximation of reality for physiological responses. Second, as can be seen in Table 5-1, analysis v/as limited to the firstorder interactions of the model parameters. It does appear, hov/ever, that the data fit the parametric requirements necessary for the employment of the statistical techniques described belov;. As noted earlier, of particular concern v/as the time period between slides, since the experimenter wished to negate the impact of the previous stimulus slide on the response to the next-occuring slide. To test for serial effects, the Durbin-U'atson test for auto-correlation (time series in the data) was utilized. The Durbin-l.'atson statistic ranged from 1.92 to 2.09 for all of the data used in the analysis, indicating that the 20-second time period betv/een slides succeeded in eliminating the effects of the' previous response.

PAGE 65

-53Findings A Spearman rank-order correlation (Siege! , 1956) between the questionnaire scores of independent judges and subject's questionnaire scores was computed in order to determine the agreement as to slide rank between the judges and the subjects in the study. The pleasantnessunpleasantness score was the variable investigated. The subjects and pre=test judges tended to agree about the emotional content of the slides, yielding a correlation coefficient of 0.81 (significant at the .05 level). The rank order of the slides in the pretest (independent judges) group never differed by more than 1 position from the rank order of the slides in the post-experimental (subjects) group. With one exception in the neutral category, the slides also maintained the same rank when compared by medians (nonparametric) or by the averages reported in Table 5-1. It can be inferred from the high degree of inter-rater reliability implied that a tyne of cross-validation of stimulus type has been accomplished. Table 5-2 presents a summary of analysis of variance for the 4 general groups of slides used in the study. The differences here are also significant at the .05 level of confidence. As will be seen later, hov/ever, individual slide differences, when judged by the dependent variables, do not conform to the results hypothesized or to the results reported by proponents of pupil lometric techniques.

PAGE 66

-54Table 5-2 Analysis of Variance of Questionnaire Results for 4 Groups of Slides Sum of Squares D F Mean Square F Ratio Between Groups 1130.2742 3 376.7581 3.3556 Within Groups 18638.0469 166 112.2774 Total 19768.3203 169 Tables 5-3 and 5-4 present summary statistics for stepwise regression with 3-second pupil change as the dependent variable. As can be seen, the model has a multiple correlation coefficient of 0.61, explaining 37 percent of the variability in the dependent variable. An analysis of the correlation coefficients associated with the independent and moderator variables yields no consistent relationshio between the 4 stimulus types. VJhen analyzed separately, the startle stimulus was the only one causing a consistent dilation in the puoil. It does aopear that anxietyarousing stimuli do cause significantly greater dilations than do other types of stimuli. No consistent findings were associated with pleasant, neutral, or unpleasant stimuli, as the tables show. The 3-second time period, used in the present analysis, may cause part of the confusing results obtained. As I'oodmansee (1965) has noted, the orienting and near-vision reflexes of the pupil cause a constriction in size. It may be that subjects allowed their eyes to lose focus during

PAGE 67

-55Table 5-3 Last Step of Multiple Step-Wise Regression Multiple R 0.6082 Stci. Error of Est. 10.5773 Analysis of Variance D F Sum of Squares Mean Square F Ratio Regression 21 16665.863 793.612 6.961 Residual 249 28387.078 114.004 Variables in Equation Variable Coefficient Std. Error F to Remove (Constant 0.0) 4 -46.06 69.02 0.44 6 44.60 81.21 0.30 7 -88.76 150.53 0.34 8 -107,39 180.79 0.35 9 -69.47 118.22 0.34 10. 8.38 21.34 0.15 11 43.83 71.92 0.37 12 54.86 71.27 0.59 14 -28.15 37.41 0.56 15 40.32 60.27 0.44 16 -1.93 11.62 0.02 17 -21.35 39.67 0.28 18 -14.40 20.41 0.49 19 79.28 117.27 0.45 22 6.87 9.94 0.47 23 -39.13 63.11 0.38 24 -0.01 0.02 0.63 32 0.07 0.15 0.20 34 7.22 4.89 2.17 35 0.35 0.90 0.15

PAGE 68

-56Table 5-4 Stepwise Linear Regression 3-Second Percent Change in Pupil Size Step

PAGE 69

57the 20-second control period between slides, and then had to refocus when a new slide appeared. The near-vision reflex associated with refocusing would explain at least part of the results found. Tables 5-5 and 5-6 present the results obtained from the regression with the 9-second percent change in pupil size as dependent variable. The multiple correlation coeffficient, and consequently the explained variance, is lower for this model. Again, the results of individual slide regressions are equivocal. As before, analysis of covariance should have eliminated the effects of light intensity and initial value on the final results. These findings raise questions about cur ability to deduce any meaningful conclusions from results of pupillometric assessment studies. The low explained variability, together with the inconsistency associated with responses to the 4 types of stimuli, raises many questions about the usefulness of pupillometric techniques. The following tables present results of stepwise linear regressions on heartrate, blood pressure, and skin potential. As would be expected in models which do not lag these variables, the multiple correlation coeffficients and explained variance are lower than those for the pupil response equations.

PAGE 70

_58Table 5-5 Last Step of Stepwise Multiple Regression Dependent Variable: 9-Second Pupil Change Multiple R

PAGE 71

-59Table 5-6 Step Number 1 2 3 4 5 6 7 8 9 10 n 12 13 14 15 16 17 18 19 20 21 22 Stepwise Linear Regression 9-Second Percent Change, Pupil Size Variable Entered

PAGE 72

-60Variable 1. 3 sec % 2. 9 sec % 3. Initial Val 4. Dog 5. Baby 6. Animal 7. Baby 8. Baby 9. Baby 10. Triangle 11. Girl 12. Chair 13. Nude 14. Girl 15. Block . 16. Rectangle 17. Child 18. Hanging 19. Woman 20. Startle Table 5-7

PAGE 73

-61Table 5-8 Summary by Stimulus Groups of 3-Second Heartrete Regressions Step Number

PAGE 74

-62Table 5-9 Summary by Individual Stimuli: Stepv.'ise Linear Regression with 9-Second Heartrate as Dependent Variable Step Number

PAGE 75

-63Table 5-10 Summary by Stimulus Groups 9-Second Percent Change in Heartrate Step Number

PAGE 76

-64Table 5-11 Summary by Stimuli Stepwise Linear Regression of 3-Second Percent Chanqes in Blood Pressure Step Number

PAGE 77

-65Table 5-12 Stepv/ise Linear Regression of 9-Second Percent Changes in Blood Pressure Step Number

PAGE 78

-66Table 5-13 Stepwise Linesr Regressions of 9-Second Percent Changes in Skin Potential Step Number

PAGE 79

67The results described above were applied to the hypotheses listed in Chapter II. Hypothesis 1, "pleasant stimuli will be accompanied by dilation, while neutral stimuli will cause no change in oupil size", was not supported. As can be seen in the correlation matrix presented in Table 5-7, all pleasant stimuli have a slight negative correlation with 3-second pupil change. Although the correlations were less negative for the 9-second pupil change data, they are still negative. Similarly, neutral stimuli generally had negative correlations. These resutls include the factoring out of anxiety scores by analysis of covariance. Hypothesis 2, "negative stimuli will cause a constriction of the pupil", was not supported, since t!ic puoil reecticn to these stimuli was largely a positive change in size. Hypothesis 3, "anxiety-arousing stimuli will dilate the pupil", was supported. Hypothesis 4 could not be tested due to the inconsistencies in the response to pleasant stimuli. Since no consistent response to pleasant stimuli was found, the degree of dilation could not be compared with dilations associated with anxiety stimuli. Hypothesis 5, "highly pleasurable stimuli will produce greater pupil dilation than will less pleasant stimuli", was not supported, as the correlation matrix shows. Some positive results were obtained in the analysis of magnitude covariance. Hypothesis 7, "blood pressure changes will be similar in reaction when exposed to pleasant stimuli and when exposed to anxietyarousing stimuli", was not supported. While blood pressure showed no

PAGE 80

68consistent relationship to positive stimuli, it did shov/ a positive relationship to anxiety-arousing stiinuli, with a lag of several seconds. Heartrate decreased as a response to startle stimuli, v/hile showing no consistent response to pleasant stimuli, so hypothesis 8, "heartrate changes will be similar when exposed to anxiety-arousing stimuli and pleasant stimuli", was not supported. Skin potential had the largest time lag (between 6 and 10 seconds) to startle stimuli, and had fairly large magnitude changes (up to 70 percent). It showed no consistent relation to pleasant stimuli, so Hypothesis 9 was not supported. Hypothesis 10 could not be tested, again due to the inconsistency in responses to pleasant stimuli. Hypothesis 11 was supported, since no consistent relation between heartrate and positive stimuli were found, while the average time from beginning of a startle response to its end was 6.2 seconds. Hypothesis 12 was not supported, since no consistent relationship was found for response latency or return of skin potential to either type of stimuli. Hypothesis 13 was supported, probably because of the relationship existing between heartrate and blood pressure. Table 5-14 presents a summary of the hypotheses and the results obtained.

PAGE 81

-69Table 5-14 Results by Hypothesis Hypotheses Pupillary Reaction 1 Supported/Rejected Pleasant stimuli will be acconipanied by dilation, v/hile neutral stimuli will cause no change in pupil size. 2. Negative stimuli v/ill cause a constriction of the pupil . 3. Anxiety-arousing stimuli will dilate the pupil. 4. Anxiety-arousing stimuli will cause greater dilation than will pleasant stimuli of the same intensity. 5. Highly pleasurable stimuli will produce greater pupil dilation than will less pleasant stimuli. 6. Highly negative stimuli will produce greater pupil constriction than will less negative stimuli . Autonomic Contamination A. Magnitude Covariance 7. Blood pressure changes will be similar in reaction when exposed to pleasant stimuli and when exposed to anxiety-arousing stimuli. 8. Heartrate changes will be similar when exposed to anxiety-arousing stimuli and pleasant stimuli. 9. Skin potential changes will be similar in reaction when exposed to anxiety-arousing and to pleasant stimuli . B. Response Duration 10. Pupillary dilation to anxiety-arousing stimuli will take longer to return to initial base level (at time of stimulus presentation) than will puoil dilation to pleasant stimuli, for a given level of stimulus intensity. not supported not supported supported not suDported not supported not supported not supported not supported Untestable with present data

PAGE 82

-70Table 5-14 (Cont.) Hyootheses Supported/Rejected 11. Heartrate v/ill take longer to return to base line in response to anxiety-arousing stimuli than to pleasant stimuli, for the same stimulus level. supported 12. Skin potential v/ill take longer to return to base line in response to anxiety-arousing stimuli than to pleasant stimuli, for the same stimulus level. not supported 13. Blood pressure v/ill take longer to return to base line in response to anxiety-arousing stimuli than to pleasant stimuli, for the same level of stimulus intensity. supported Shortcomings It v/ould be useful to attempt to explain the disheartening results described above as the product of poor experimental design or analysis. The controls used in the study, hov/ever, were at least as rigorous as most reported in the literature. Also, as mentioned previously, the data were collected in 2 forms (chart and digital) with chart data being used to validate digital data at all points of conversion, so data reduction errors seem to be negligible. Although the steps taken speak v/ell for the data collected, there are several areas where the experiment could have been improved. Better care could have been taken to develop a set of slides with more closely balanced light intensities. Spot intensity on each slide should have

PAGE 83

•71been controlled. At the data recording level, better measures of light intensity could have been made. It appears that the scores on the pre-test questionnaire may have led to erroneous conclusions about one or more slides. For example, the slide showing an attacking police dog scored highest on the anxiety-arousing dimension, and was therefore included in the anxiety set. Subjects who were debriefed informally, however, tended to agree that there was little anxiety-arousing quality in a slide of an attacking dog, so the questionnaire score mv have been misleading. The method by which eyeblinks were edited in the computer data could also be improved. The data reduction programs incornorated a subroutine \;herehy each reading under 2.5 millimeters was eliminated and the data Doint immediately preceeding the eyeblink was substituted. Perhaps the m.ost appropriate procedure would have been the total elimination of that data point. Alternatively, several smoothing techniques are available which would have been more aonrooriate than the method used,

PAGE 84

CHAPTER VI METHODOLOGICAL IMPROVEMENTS: THE FEASIBILITY OF COMPUTER CONTROL AND DATA SAMPLING As discussed in previous chanters, the methods of data collection in psychological and medical research on physiological variables are often tedious and subject to several types of error. One of the miost productive results of this study v.'as the demonstration that computer sampling is a viable and advantageous method for collecting data on physiological variables. Computers today are capable of performing many different functions, and are available in many forms and sizes. Their costs are now at a level v/here most researchers should take time to examine their usefulness. Advantages Several advantages accrue to the researchers v/ho utilize the computer. If he can develop a method whereby the computer can be programmed to control the presentation of stimuli, as in this study, the experimenter can significantly reduce the uncertainty and inconsistency with which humans perform such activities. The computer can overcome limitations of human performance capabilities, such as the splitsecond changing of slides in the present study. Because of their high oro-72-

PAGE 85

73cessing speeds, computers are capable of performing several operations simultaneously, such as the changing of slides and collection of several data samples in the span of a second. As a data collector, the computer offers the advantages associated with unbiased sampling and recording of results. As the experimenter has shov/n earlier (Bell, 1971), controls of this type must be incorporated in physiological data-gathering methods. Computers also provide a significant reduction in the time it takes to produce data in a form ready for analysis. Perhaps one of the major advances in computer hardyv^are in recent years has been the mini -computer. These machines, incorporating solidstate microcircuitry, can be obtained in sizes that will fit almost anywhere, and that are highly adaptive to the snecific requirements of the purchaser. Vendors will, in some cases, assist in the acquisition of equipment configurations which best meet the needs of the user. Several mini -computers are "conversational" (have the ability to communicate) with larger scale machines. A physiological researcher could obtain a PDP-11 mini-computer (Digital Equipment Corporation), for example, to control his experiment and collect his data. The machine could be connected via phone link to a large-scale system such as the IBM 360/65 where direct data reading and analysis could take nlace. Thus the machine at the local level could be adapted to a specific research need, while maintaining the ability to use large-scale data analysis

PAGE 86

74packages, and all of the peripheral equipment associated v/ith the larger scale systems. Finally, many persons associate the v/ord computer vn'th high costs of operation, while not fully realizing that computer utilization may reduce or eliminate other costs related to the project. This v;ill be discussed in more detail later in this chapter. Disadvantages of Computer Usage One of the major disadvantages of computer usage to behavioral researchers is the dependence on the "systems man" such usage creates. Before e system can be utilized, hardware must be acquired and made operational, and software (programs) must be developed and debugged. Once the system is operational, any changes to be made or any errors to be corrected require that someone familiar with the computer be called in. There may also be a tendency to accept the face validity of computer-generated data without questioning the data collection, transfer, and analysis. When using the computer, an experimenter can be far removed from the data he is collecting, and may miss important happenings during the experiment that could influence the data. The use of a computer does not relieve the experimenter of the responsibility for accurate The use of an earlier model Digital Corporation Machine, the PDP-8, in clinical and diagnostic medical research has been described earlier by Covvey (1970).

PAGE 87

•75data collection, analysis, and reporting. An important consideration is the frequency and complexity of the projects to be undertaken by the computer. Computers make costly pieces of furniture in a research laboratory if the extent of their usage does not justify their acquisition. In university settings, researchers many times share a computer with others who need one in order to reduce the per-project utilization costs. Some Cost-Benefit Consideration s It is a relatively simple matter to discuss the advantages and limitations of computer systems in general terms. Before the recommendation to purchase a system can be accepted or rejected, however, the very difficult task of assigning dollar costs and benefits of the system must be undertaken. Many cost figures can be readily obtained from price lists of manufacturers. Others, such as the cost of installing the system and making it operational, are less easy to pinpoint, and may require a great deal more research and "guesstimation". Even these, however, may be estimated from the files of users of similar equipment, or from the manufactuer. The most difficult problem lies in the generation of accurate estimates of the benefits to be derived from the computer system. Before the system is installed, it is impossible to foresee all of the possible applications it may have. It may also happen that some anticipated uses

PAGE 88

76of the system are not realized after installation. About ths best that can be done is an attempt to foresee as many ootential advantages (and limitations) as possible, and to treat these data in a way that recognizes the uncertainty under which they were generated. For this reason, this section will treat data in a manner proven useful in other situations involving uncertainty. In particular, when uncertainty exists, data will be derived from the familiar v;eighted-average formula used in PERT and other project planning and control techniques. ^ A computer system for collecting and analyzing data of the type described in this disseration, would at a minimum, contain elements for acquiring, storing, and processing data. The hypothetical system described below attempts to realistically meet those requirements within the framework of decisions made by a university researcher. It is assumed that the researcher is now collecting data manually (as most pupillometric researchers do), transferring it to punched cards and analyzing it by computer. It is further assumed that measuring instruments such as the television pupillometer and devices for assessing heartrate, blood pressure, and skin potential are readily available. A central processng unit (C.P.U.) with at least 4-thousand (4-k) bit memory would be required to store the data. Analog-to-digital conversion requirements call for ^Wallace, J.B., "Improving communication between systems analyst and user'j Data Management , June 1972, pp. 21-25.

PAGE 89

77the purchase of 6 channels of converters, and the input/output operations of the system require a 4-k 16-bit read/v;rite computer core. The hardware described above form the nucleus for a system capable of acquiring and storing an amount of information comparable to that gathered for each subject in this dissertation. If we assume that the system is being designed for a university research laboratory, then the most efficient use of additional funds for analysis data equipment would be the purchase of a datanhone. This would allow the computer to comimunicate with a larger-scale system. The dataphone would tie the small computer in with the C.P.U. of the larger machine, thus greatly enlarging the storage capacity of the system, increasing its analytical capabilities, and connecting it to all of the peripheral devices on the larger system. Table 6-1 describes the dollar estimates of annual benefits to be gained from the new computer system. Table 6-2 describes estimated costs. Estimates were made by this writer and by a professor currently engaged in psychiatric research using pupillometrics and heartrate analysis, Some of the sources of dollar cost reductions or added benefits are mentioned briefly in the tables, As can be seen, the simple ratio of annual benefits of the new system to annual costs is 0.30 . Several other techniques for analyzing cost and benefit data are currently used. Among these, the cost/benefit ratio and payback period for the investment in the computer system v/ere computed.

PAGE 90

78TablR 6-1 PDP 11/20 System v/ith Datanhone l.inkane to IBM 360/65 Cost/Penefit Analysis (Annual Basis) Benefits: Reduction ^ Reassionpent o'f Clerical Staff (Reduction in staff needed to measure physiological variables) (Reduction in keypunch costs) % 8,000 Responsiveness: Faster Hata Transfer (ability to snot errors while subjects and equionient confiaurations are still available) 1 ,oon Accuracy: Increase in Reliability of Data Sampling (value of increased accuracy: reduced need to reperform experiment; increase in confidence in generating reports, etc.) 1,000 Comprehensiveness: Capability for Increased Scope of Data Acquisition (value of additional information; value of nev; projects v/on because of expanded capabilities) 5,000 Total Estimated Annual Benefit $15,000 Cost/Benefit Estimates Derived from Beta-Distributed Estimates Under Uncertainty: F + 4f^L + P " " G where P = pessimistic = optimistic ML = most likely

PAGE 91

•79Table 6-2 PDP n/20 System with Dataphone Linkage to IBM 360/65 Cost/Benefit Analysis (Annual Basis) Costs Nonrecurring C.P.U. Purchase 4-K 16-bit read/v/rite memory Programmer console Basic mounting box Power supply ASR-33 teletype and control $ 10,800 4 K 16 bit read/write core 3,000 A to D converters, 6 channels 2,700 Dataphone 1 ,800 System installation/debugging 1,000 $ 19,300 S-year life Annual Basis 3,860 Recurring Dataphone Modem Rental ($100/mo.) 1,200 System programming 2,000 System maintenance 1,000 System supplies 500 CPU time/System 360 5,000 Total estimated annual cost 13,560 Cost divided among three projects 4,520 Cost/Benefit Estimates Derived From Beta-Distributed Estimates Under Uncertainty: P P + 4ML + ^ ~ 6 where P = pessimistic; = optimistic; ML = most likely Source: Digital Equipment Corporation: PDP 11/20 Price List, Nov. 1970

PAGE 92

-80The payback period is defined as: I". E ( Z B. . Z CC. J/T . j J,t 1,1 where T = expected life of the system B = the jth benefit in time period t CC = the ith recurring cost in time period t CI = the kth initial cost R = the time value of money or cost of capital For the data described in Tables 6-1 and 6-2, the payback oeriod is 1.58 years. These data indicate that the system vn'll pay for itself in approximately one and one-half years. Since a 5-year life v/as assum.ed, it appears that these results indicate that investment in the system is justified. The cost/benefit ratio is computed from the formula: I {ZB I CC) ^ ,/ Z CI . (1 + R)^ / The data yield a cost-benefit ratio of 0.424, indicating that over the 5-year period the costs associated with the system, even when considering the time-value of money invested, are less than one-half of the generated benefits. These results also support arguments for acqui sition of the system.

PAGE 93

-81. As is obvious, the data presented in Table 6-1 represent very uncertain estimates of benefits. However, even if we can assign only a probability of 0.5Q to the expected occurence of these benefits, the cost/benefit ratio still is less than 1.0. This result allows some degree of confidence in the statement that a system can be justified. In summary, it appears that researchers should investigate possibilities computer systems offer as sources of more advanced exoerimental control and analysis. Although specific cost and benefit figures depend on individual project requirements, the data presented above suggest that, in some cases, investments in these systems can be easily justified.

PAGE 94

CHAPTER VII IMPLICATIONS FOR FUTURE RESEARCH The results of this study raise several questions about the direction of future research on puoillometrics. If we accept the hyoothesis that the pupil does respond differentially to stimuli and that it could be a useful indicator of emotional reactions, then we are faced with the apparent need for much more basic and m.uch more intensive work than has been previously reported. As with this dissertation, for every "positive" finding reported in the pupillometric literature, there seems to be a corrolary "negative" one. Until this type of research reaches a stage where general uniformity of results is accomplished, it will not be accepted as a useful technique for more advanced apoli cations. Attitude Theory Given the complexity of design necessary for evaluating attitudes with physiological variables, an effort should be made to determine the additional information this design yields over other, more conventional, measures of attitudes. It appears that paper-and-pencil instruments such as the semantic differential, the Likert technique, or the Thurstone technique would yield more information per input unit of work exoended that would the physiological techniques. At the present state of the art -82-

PAGE 95

-83in pupil! ometrics, it also appears that these scales would yield a greater absolute magnitude of information than v/ould oupil studies. The generally reported reliability and validity coefficients associated with the paoer-and-pencil methods are higher than those associated with pupillometrics (Woodmansee, 1965), so it appears that for most assessment problems, the conventional measurement instruments may be more appropriate. Comparative studies along this line could define some relevant attitude domain to be sampled, then develoo assessment instruments of both types, noting the costs of development in time, dollars, manpov/er, and other resources expended. Reliability and validity coefficients could be computed using data gathered from "known groups". Per unit contribution and absolute contribution of information could be calculated for each technique. Research on Stimulus Presentation Perhaps one of the most limiting factors in pupillometrics is the method of operationalizing independent variables through visual stimuli A look back to the regression models developed in Chapter V demonstrates this problem. Many of the factors affecting pupil size are associated with the eye as an input medium for stimuli. The near vision reflex, the orienting reflex, color adaptation, light intensity (both spot and scrambled), stimulus complexity, and the physical location of the visual stimulus all cause changes in pupil size. If it were possible to develop

PAGE 96

-8-1a different type of stimulus input, such as auditory stimulation, the large number of contributors to nuoil change would be reduced, and perhaps the effects due to stimulus content could be better isolated. Beyond this problem, there is the difficulty of projecting attitude objects through the use of visual stimuli. As several subjects in the present study noted, a very large difference exists between an attacking police dog and a picture of an attacking police dog. Similarly, a picture of a Negro may not evoke the same type of response that a living being does. A very difficult problem must be overcome before pupillometric techniques can be applied in areas such as the personnel function. To date, most successful studies have involved fairly simple, basic attitudes. Most of these could have been evaluated equally well using conventional techniques. The hope is that some method for presenting complex emotional stimuli can be developed. Of ell stimulus types, complex visual stimuli are the least preferred, since the complexity requires greater coping behavior from the pupil and causes greater eye search of the stimulus, both of which confound pupil change data. Some method which does not cause confounding interactions with the dependent variable must be developed for presenting more complex stimuli. Field Research In the marketing area, in addition to research in stimulus presentation, much work needs to be done on equipment and mobility. Present

PAGE 97

85techniques are still wery "obtrusive" and interactive with tlie resoondent. The development of a non-interactive n-ieasuring instrument for research on consumer's reactions to package design and other promotional techniques would be a major advance. If marketing researchers plan to use pupil lom.etrics, a useful study could be made using portable physiological recording devices. A physiological recorder, for instance, could record several tracks of data simultaneously, during a marketing study. The data could then be taken back to the laboratory and read directly from the recorder into a computer. The lag time between data collection and analysis could be significantly reduced using this technique. Researchers could evaluate the tradeoffs between using still -frame versus video cameras, and could develop cost/benefit analyses of the instruments used in the study. Laboratory Research Much work still needs to be done in the area of instrumentation in pupillometric research. It appears that there is very little question, other than relative costs, about the general superiority of video systems such as the television pupillometer, especially when such systems can be interfaced with a computer. V.'hether the significant additional costs of the television system are justified depends upon the value of the information to be gathered and the availability of supporting equipment.

PAGE 98

-86Finally, applications cf the mini -computer in physiological research can now be evaluated. Work needs to be done on the compatibility of the small computer systems with larger systems. The information processing, storage, and input/outout reauirements of physiological data processing systems need to be investigated. With advances in computer technology continuing at the current rate, researchers who do not recognize the potential of the computer for their projects may soon be left with obsolete methods and inferior results.

PAGE 99

CHAPTER VIII SUMMARY AND CONCLUSIONS Goals The goals of this dissertation were to replicate findings that pupil change is a useful measure of emotional reactions to visual stimuli, develop a linear model explaining the contributors to pupil change, develop a computer sampling method and examine the feasibility of its use in pupil change studies, and study the other physiological ccvariates in pupil change. It was hoped that the recording of changes in other variables along with pupil change would lead to a means of discriminating between pupil dilations caused by pleasurable stimuli and dilations caused by anxiety-arousing stimuli. Method Fourteen males served as independent ratei^ of 57 slide stimuli. Of the 57, 15 were chosen for use in the study. The 16 were divided into 4 visual stimulus categories: pleasant, neutral, unpleasant, and anxietyor fear-arousing. The medians test was used to test for differences between the sets of stimuli. Each set was significantly different from the other sets at the .01 level. It appears, however, that the slide used as an anxiety-arousing stimulus was actually not capable of arousing -87-

PAGE 100

88anxiety in the subjects used. Tv.'o startle stimuli--a buzzer and a drop of several inches in the chair subjects sat in--were also included in the anxiety set. The study took place in a laboratory where environmental light and sound levels could be controlled. Subjects were seated in a dentist's chair, and looked into an apparatus similar to a Hess Box. Slides were presented in random order on a screen at the end of the box opposite the subject. The two startle stimuli always occurred as stimuli number 17 and 18. A television pupillometer measured change in pupil size. In addition, heartrate, blood pressure, and skin potential were measured. These variables, together with the light intensity of the slide screen, were recorded with a Dynograph recorder, which produced charts of the data for each subject. The Bunker-Raymo computer was connected with analog-to-digital converters to the Dynograph recorder and recorded second-by-second digital readings for each of the variables, At the end of each session with a subject, a paper tape record of the 2,550 readings taken on that subject v/as produced. A Bunker-Raymo process control computer v/as connected to two Kodak carousel 850 slide projectors and controlled the timing of the slide presentations. Each stimulus slide was preceeded by a 20-second blank slide. The Bunker-Raymo, by initiating a change in the projector containing the stimulus slide before removing the controls slide, reduced

PAGE 101

the light-flash and time required to change slides to under 1/10 second. At the end of the slide sequence, the computer automatically sounded the buzzer. Approximately 20 seconds later, the experimenter released the vertical hold of the chair the subject was sitting in, causing it to "fall" rapidly for about 5 inches. Analysis The paper taoe record was converted to magnetic tape, and the magnetic tape data v/e re analyzed. For analytic purposes, a linear model was used to describe the parameters affecting Dupil change. Stepwise linear regressions were computed for each of the physiological variables, with the regressions using analysis of covariance to negate the effects of light intensity changes and the effects of base level --the "initial value" of the dependent variable before stimulation. In addition, correlations between each slide and the percent change in pupil size were computed, using both a 3and a 9-second measurement of pupil change. Analysis of variance of pupil change was computed for the 4 groups of stim.uli. Other, more general analyses involved an investigation of the time lags associated with the physiological variables, and an analysis of consistencies in response magnitudes. Results Although 36 variables, including first-order interactions', were included in the stepwise regression analysis, and although the model's

PAGE 102

-gopredictability of pupil change was significant at the ,05 level, the exolained variance in pupil size v;as less than 40 percent. Furthermore, although analysis of variance for the groups of stimuli was significant at the .05 level, the individual correlations of slides with the dependent variables were low and surprisingly inconsistent. Pleasant slides did not produce consistent pupil dilation. In fact, just the opposite was true in many situations. Using pupil change as a predictor, researchers would have failed to identify the sets of slides chosen by the independent raters. Since post-test questionnaires filled out by subjects tended to agree with the independent judges as to the classification of slides, the hypothesis that pupils dilate to pleasant stimuli and constrict to negative stimuli does not find support in these data. Because of the discrepancies in direction of nuoil change, certain hypotheses about the relative effects of pleasant and anxiety-arousing stimuli on pupil dilation could not be tested with the data gathered. It does not appear, on the basis of response latency data and magnitude data, that heartrate would be a useful covariate indicating an anxiety-arousing stimulus. Further work could be done on the length of time from stimulation to maximum response as a possible indicator of strength of reaction. Cost/Benefit Analysis A brief cost/benefit study was conducted to determine the feasibility of using mini -computers as real-time data sampling instruments in the type of research reported in this dissertation. Many advantages, of

PAGE 103

.91. computer usage were cited, including increased data reliability, increased experimental control, faster turn-around of data, and the potential for expansion of the scope of research projects. Costs, inefficient usage, and the dependence on systems programmers were discussed as drav^backs to the adoption of computer-based techniques. Although specific data were not available in every category, the experimenter, a laboratory technician, and a full-time researcher made estimates of somie of the costs and benefits associated with computer usage. Estimating techniques for decisions under uncertainty were used to calculate final figures. Results indicate that there are significantly more dollar benefits than dollar costs, and that, at least in some instances, computers can be justified as laboratory instruments. Conclusions With the large amount of data collected in this study, comprehensive analysis has just been initiated by the dissertation report. The disheartening results so far seem to indicate that the basic theory oroposed by Hess and others is open to significant question. It appears at this stage as if the results reported previously have been contaminated due to lack of adequate consideration of such factors as stimulus visual complexity, initial levels of response variables, interactions of response variables, and the myriad of within-subject contributors to physiological changes . Also, for the puoillom.etric technique to be useful to business researchers, a great deal more consistency in the response variables must

PAGE 104

-92be found. This study, together with others such as those by Woodmansee (1965) and NunnaHy, et al . (1967) tendsto indicate that this may be a very difficult task. In addition, it appears that visual stimuli are less-than-optimal for use in pupil studies. The use of the eye as both an input and an output mechanism is of questionable value. Many of the "contaminants" of the pupillary response to psychological stimuli &re associated with visual presentation. The care with which stimuli must be produced, together with the problem of operationalizing complex emotional stimuli through pictures, indicate that new directions of research on methods of stimulus presentation are needed. Although concentrated analysis has yet to be done, it appears that all other physiological variables studied hold some hope of serving as indicators of the autonomic contamination reported by Hess. Their usefulness, of course, presupposes pupil dilation to both pleasant and fear-arousing stimuli—a supposition not supported by the findings reported here. The ease with which computer data sampling was accomplished indicates that this technique could prove very useful to researchers. The efficient use of a computer, however, requires a well-equipped laboratory with sophisticated measuring devices such as the television pupillometer. In view of the present state of the pupil lometrics art, however, it appears reasonable to assume that researchers will have to use advanced techniques to avoid the collection and presentation of spurious results.

PAGE 105

BIBLIOGRAPHY Adier, A., P hysiology of t he Eye. St. Louis: The C. V. Mosby Comoany, 1959. Aronson, E., "Some Antecedents of Interpersonal Attraction", in J. Arnold and D. Levine (eds.), Nebraska Symposium on Motivatio n, 1969, 17, pp. 143-173. Ax, A. P., "The Physiological Differentiation Betv;een Fear and Anger in Humans", in T. K. Landauer, Readings in Physiologic al Psychology. New York: McGraw-Hill, 1967. ' Bahn, C, "The Economy of Scientific Selection", Personnel Journa l, August, 1970. Barlow, J. D. , Pupillary Size as an Index of Preference , Ph.D. Diss., Louisiana State University, 1969. , "Pupillary Size as an Index of Preference in Political Candidates", Perceptual and Motor Skill s. 1969, 28, pp. 587-590, _, "Pupillary Size as an Index of Preference", Perceotual and Motor Skills , 1970, 31, pp. 331-336, Bell, Robert R., "Experimenter Expectancy in Pupillometric Research", Perceptual and Motor Skills , 1971, 3_3, p. 174. Bender, W.R., "The Effect of Pain and Emotional Stimuli and Alcohol on Pupillary Reflex Activity", Psychological Monographs , 1933, No. 198. Berscheid, E., and E. Walster, Inter personal Attra ction. Menlo Park, California: Addi son-Wesley, 1969. Blackwell, R. D., J. S. Hensel , and B. Sternthal , "Pupil Dilation: What Does it Measure?" Journal of Advertising Research , 1970, 20, p. 4. Boersma, F., K. Wilton, R. Barham, and W. Muir, "Effects of Arithmetic Problem Difficulty on Pupillary Dilation in Normals and Educatable Retardates", J. Experimental Child Psychology , 1970, 9, pp. 142-155. Bradshaw, J. L., "Pupil Size as a Measure of Arousal During Information Processing", Nature, 1967, 216^, pp. 515-516. Bradshaw, J. L., "Load end Puoillary Changes in Continuous Processing Tasks", British Journal of Psychology, 1968, 59_, dd. 265-271. -93-

PAGE 106

-94Brandt, H. F., The Psychology of Seeing . New York: Philosophical Library, Inc., 1945. Cartwright, D., "The Nature of Group Cohesiveness" , in D. Cartwright and A. Zander, Group Dynamics , New York: Harper and Row, 1968. Cason, H., "The Conditioned Pupillary Reaction" J. Experimental Psycholog y, 1922, 5, pp. 108-146. Chowdry, K, , "Selection of Executives and Administrators: Implications of Recent Research", Personnel Journal , 1969, 48(2), p. 102. CI ark , \'L R . , A Pupil lographic Study of Short Term Memory Search , Ph.D. Diss., Oklahoma State University, 1971. Cook, S., and C. Selltiz, "A Multi-Indicator Approach to Attitude Measurement", in M. Fishbein (ed.). R eadings in Attitute Theory and Measurement , New York: John Wiley and Sons, 1967. Cooper, J., and D. Pollack, "The Identification of Prejudicial Attitudes bv the Galvanic Skin Response", Journal of Social Psycholog y, 1959, 50, pp. 241-245. Crasilneck, H.B. and E.J. McCranie, "On the Conditioning of the Pupillary Reflex", Journal of Psycholo gy, 1955, 42, pp. 23-27. Davis, R. C, A. Buchwald, and R. Frankman, "Autonomic and Muscular Responses, and Their Relation to Simple Stimuli", Psy. Monograohs , Gene ral and Applied , 1955, 69, p. 20. Domm, D., and J. Stafford, "Personnel: Behind the Tim.es", Personnel Journal , July 1970. Dooley, R. P. and D. J. Lehr, "Critique of a Pupillary Response Experiment", P erceptual and Motor Skills , 1967, 25, pp. 603-4. Duke-Elder, S., System of Ophthalmology. Vol. IV. The Physiology of the Eye and Vision , St. Louis: The C. V. Mosby Co., 1968. Dunnette, M. D., Personnel Selection and Placement . Belmont, California: Wadsworth Publishing Co., 1967. Fiedler, F. E., "Styles of Leadership", in E. P. Hollander and R. G. Hunt, Current Perspectives in Social Psychology , New York: Oxford University Press, 1967. Francis, R. D. and M. R. Kelly, "An Investigation of the Relationshio Between Word Stimuli and Optical Pupil Size", Australian Journal of Psycholog y, 1969, 21_. 2.

PAGE 107

-95Gellhorn, E., Autonomic Regulations . New York: Interscience Publishers. 1943. Ghiselli, E. and N. Haire, "The Validation of Selection Tests in the Light of the Dynamic Character of Criteria", Personnel Psychology , N. A., 1959. Girden, E., "The Dissociation of Pupillary Conditioned Reflex Under Erythroidine and Curare", Journal of Experimental Psychology , 1970, 30, pp. 631-534. Good, L. R. end R. H. Levin, "Pupillary Responses of Repressers and Sensitizers to Sexual and Aversive Stimuli", Perceptual and Motor Skills , 1970, 30, p. 631. Graham, C. H. (ed.), Vision and Visual Perceotion, New York: Wiley and Sons, 1965. Gross, A. C, "Accessibility and the Small Comouter", Datamation , Nov. 15, 1971, pp. 42-48. Grossman, S. P., A Textbook o f Physiological Psychology , New York: Wiley and Sons, 1967. Guinan, J. F., An I nvestigati on of the Relationship E et vjeen Puoil Size and Emot i onal Words , Ph.D. Diss., Michigan State University, 1967, Halpern, R. S., "Applications of Pupil Response to Before-and-After Experiments", J. Marketing Research , 1967, 4, pp. 32-37. Hampton, D., C. Sumner, and R. Webber, "Psychological Factors in Organizational Behavior", in J. Hutchinson, R eadings in Manager.e nt Strategy and Tactics . New York: Holt, Rinehart and Winston, 1971. Heise, D. R., "Semantic Differential Profiles for 1000 Most Frequent English Words", Psychological Monographs , 1965, 79_, No. 8. Hess, E. H., "Attitude and Pupil Size", Scientific American , 1965, 212, p. 46. _, "Pupillometrics" in F. M. Bass, C. King, and E. Pessemier (eds. ) , A pplications of the Sciences in Marketing Research . New York: Wiley, 1968a. ' , "Pupillometric Assessment", Research in Psychotherap y, 1968b, 3, pp. 573-83. _, and J. Polt, "Pupil Size as Related to Interest Value of Visual Stimuli", Science , 1960, J32.. P349,

PAGE 108

-96_, and J. PoU, "Pupil Size in Relation to Mental Activity During Simple Problem Solving", Science, 1964, U3, p. 13. , and J. Polt, "Changes in Pupil Size as a Measure of Taste Differences", P erceptual and Motor S kills, 1966, 23, p. 451. _, and 0. Polt, "Reoly to 'Critique of a Pupil Response Experiment'", Perceptual' and Motor Skills . 1967, 25^, p. 659. ^ A. Seltzer, and J. Shlien, "Pupil Responses of Heteroand Ho'iTiosexual Males to Pictures of Men and Women: A Pilot Study", J. Abnormal Psychology , 1965, 3_, p. 168. Hicks, R., T. Raeney, and L. Hill, "Effects of Pupil Size and Facial Angle on Preference of Photographs of a Young Woman", Perce ptual and M otor Skills , 1967, 24, op. 388-390. Hilgard, E. R. , C. E. Dutton, and J. S. Helnick, "Attempted Pupillary Conditioning at Four Stimulus Intervals", Journal of Experimental Psychology, 1941, 29, pp. 89-103. Holmes, D., "Pupillary Response, Conditioning, and Personality", Journal of Personality and Social Psychology , 1967, 5, pp. 98-103. Hudd, L. D. and J. P. Anderson, "The Relationship Between Puoil Size and Recognition Threshhold", Psychonomic Science . 1967, 9_(8), pp. 471-478. Isaacson. R., R. Douglas, J. Lubar. and L. Schmoltz, ^Primer of Physiological Psychology . New York: Harper and Row, 1971, Johnson, D. A.. Pupillary Responses During a Short T erm Memory Task With Instr uctions to Forget . Ph.D. Diss., Oklahoma State University, 19/1. Kahneman, D., D. Beatty, and I. Pollack, "Perceptual Deficit During a Mental Task", Science . 1967, 157_. pp. 218-219. Kahneman, D. and J. Beattv, "Pupil Diameter and Load on Memory", Science, 1966 » 154, p. 1583-1585. Kahneman, D., L. Onuska, and R. Wolman, "Effects of Grouping on the Pupillary Response in a Short-Term Memory Task", Quarterly Journal of Experimental Research in Personality . 1968, 20, pp, 309-311, Kahneman. D., W. Peavler, and L. Onuska, "Effects of Verbalization and Incentive on the Puoil Response to Mental Activity", Canadian Journal of Psychology . 1968, 22. pp. 186-196.

PAGE 109

•97Kennedy, 0. A., Pupil lometrics as a n Aid in theA ssessme nt of Motivation , Impact of Trea tme nt, and Prognosis of Chronic Alcoholics , Ph.D. Diss., University of Utah, 1971. Kimmel , H, , "Instrumental Conditioning of Autonomically Mediated Behavior", Psycholog i cal B ulletin, 1967, 5_, pp. 337-345, Koons , P. B., "Canonical Analysis", in H. Borko, (ed.) Comnuter A pplications in the Behav ioral Sciences , Enqlewood Cliffs, M.J.: PrenticeHall, 1962, p. 266." Korman, A. K. , "The Prediction of Managerial Performance: A Review", Personnel Psychology , 1968, 21, (3). Krugman, H., "In the Eye of the Peholder", Sponsor, December 28, 1964, pp. 25-29. Krugman, H., "Some Applications of Pupil Measurement", J. Marketing Re search , 1964, j^, pp. 15-19. Lacey, J. I., "The Evaluation of Autonomic Responses: Tov/ard a General Solution", Annals Nev/ York Acade m y of Sciences , 1956, 67_, pp. 123-164. Landauer, A,, and R. Eeakes, "A Simple Electronic Apparatus to Record Continuously Changes in Pupillary Size", Perceptual and Motor Skills . 1967, 24, pp. 283-286. Lang, P. J., "Autnomic Control", Psycholo gy Toda y, October, 1970, 4_, 4, pp. 137-186. Lapiere, R. T., "Attitudes Versus Actions", Social Forces , 1934, 13, pp. 230-237, reprinted in M. Fishbein, ed., Attitude Theory and Management . New York: John Wiley and Sons, 1967. Levinson, H., Executive Stress . New York, Harper and Row, 1964. Likert, R., "New Foundations for the Art of Management", in J.Hutchinson, Readings in Management Strategy and Tactics , New York: Holt, Rinehart and Winston, 1971 , Lowenfeld, I. E. "Pupil Size", Surv ey of Ophthalm ology. 1956, JJ_, p. 735. Lov/enstein, 0., and I. E. Lov/enfeld, "Disintegration of Central Autonomic Regulation during Fatigue and its Reintegration of Psychosensory Controlling Mechanisms: Pupil lographic Studies", Journal of Nervous and Mental Diseases , 1952, 115, pp. 1-22.

PAGE 110

-98Lowenstein, 0., and I. Lov/enfeld, "Mechanisms of Pupillary Reflex Dilation", American Journal of Opthalmology , 1961, S]_, p. 644. Lowenstein, 0. and I. Lowenfeld, "The Pupil" in The Eye , Hugh Davson, ed. London: Academic Press, 1962. McClelland, D. C, "To'ward a Theory of Motive Acquisition", in L. Cummings and W. Scott, eds.. Readings in Organizatio n al Behavior and Human Performance , Georgetown, Ontario: Irwin-Dorsay, Ltd., 1969. Metzner, C. A., and L. E. Baber, "The Pupillary Response Conditioned to Subliminal Auditory Stimuli", Psychological Bulletin , 1939, 36, p. 625. Milner, P. M., Ph ysiological Psychology . New York: Holt, Rinehart and Winston, 1970. Miner, J. B., Personnel Psychology , Toronto: Collier-MacMillan, Ltd., 1969. Newcomb, T. U., "Stabilities Underlying Changes in Interpersonal Attraction", in D. Cartright & A. Zander, G roup Dynamics , New York: Harper and Row, 1968. Nunnally, J., P. Knott, A. Duchnowski, and R. Parker, "Pupillary Response as a General Measure of Activation", Perception and Psychophysics , 1967, 2, pp. 149-155. Orne, M. T., "Demand Characteristics and Quasi-Controls" , in R. Rosenthal and R. Rosnow, A rtifact in Behavioral Research . New York: Academic Press, 1969. Osgood, C, J. Suci , and P. Tannenbaum, The Measurement of Meaning . Urbana: University of Illinois Press, 1967. Pavio, A. and H. Simpson, "The Effect of Word Abstractness and Pleasantness on Pupil Size During an Imagery Task", Psychonomic Science , 1966, 5^, pp. 55-56. Peavler, W. S. and J. P. McLaughlin, "The Question of Stimulus Content and Pupil Size", Psychonomic Science , 1967, 8, pp. 505-506. Pen rod, J. P., The U tilization of E.N.G. in the Assess m ent of Auditory Sensitivity , M.A. Thesis, University of Florida, 1968. Polt, J. M. "Effect of Threat of Shock on Pupillary Response in a Problem Solving Situation", Perceptual and Motor Skills , 1970, 31, pp. 587-593.

PAGE 111

-99Porter, L., and E. Lawler, "Properties of Organization Structure in Relation to Job Attitudes and Job Behavior" in L. Cummings and W. Scott, Readings in Organization B ehavior and Human Performance . Georgetown, Ontario: Irwin-Dorsey, Ltd., 1969. Prahalis, C. P., "Personality Tests are a Joke Because..." Sales Mananement 102 , (1), 1969. ^Rosenberg, M., "The Conditions and Consequences of Evaluation Apprehension", in R. Rosenthal and R. Rosnow, Artifact in Behavioral Research , 1969. Rosenthal, R., Experimenter Effe c ts in Behavioral Research . New York: Appleton-Century Crofts, 1%6. Rosenthal, R., and R. Rosnow, Artifact in Behavioral Research . New York: Academic Press, 1969. Rubin, L. S. , G. J. Barbero, W. S. Chernick, and M. S. Sibinga, "Pupillary Reactivity as a Measure of Autonomic Balance in Cystic Fibrosis", J. Pediatrics , 1963, 63, po. 1120-1129. S h e f 1 i n , J . ,_.An Application of Hess' Pupillometric Procedure_to_a_Psych i a tD-£i^L''lation^A_n_ADproadi Utj_lj_zjjK)_SGX_ua]_S^timjj2i. Ph.D. Diss., Purdue University, 196*9'. Shull, F., A. Delbecq, and L. Cummings, Organizational Decision Mak ing. New York: McGraw-Hill, 1970. ^ Siegel, S.', Nonparametric Statistics for the Behavioral S ciences. New York: McGraw-Hill, T956. Simms, T., "Pupillary Response of Male and Female Subjects to Pupillary Differences in Male and Female Picture Stimuli", Percept. & Psychophysics, 1967, 2, pp. 553-555. Stass, J. W., and F. Willis, Jr., "Eye Contact, Pupil Dilation, and Personal Preference", Psychonomic Science , 1967, 7, p. 375. Steckle, L. C,, "Two Additional Attempts to Condition the Pupillary Reaction", Journal of General Psychology , 1936, 15, pp. 369-377. Steckle, L., C, and S. Crenshae, "An Investigation of the Conditioned Iridic Reflex", Journal of General Psychology , 1934, ri_, pp. 3-23. Sternback, R. J., Principles of PsvchoDhvs iologv. New York: Academic Press, 1966. " ' Tanck, R. H. and P. R. Robbins, "Puoillary Reactions to Sexual, Aggressive and Other Stimuli as a Function of Personality", J . Projective Techniques and Personality Assessment . 1970, 23_ (4), p. 277.

PAGE 112

-100Veldman, D. J., Fortran Programming for the Behavioral Sciences . Nev/ Yorl^: Holt, Rinehart & Winston, 1967. Wallace, J. B. , "Improving Communication Between Systems Analyst and User", Data Management , June 1972, p. 21-25. Wedell, C.H., F. V. Taylor, and A. Skolnick, "An Attempt to Condition the Pupillary Response", Journal for Experimisntal Psycholog y, 1940, 27, p. 517. West, D. v., "In the Eye of the Beholder", Television Magazine , April 1962, pp. 60-63. Wilder, J., "The Law of Initial Values of Neurology and Psychiatry: Facts and Problems", Journal of Nervous and Mental Diseases , 1957, 125, pp. 73-86. Woodmansee, J. J., Jr., An Evaluation of Pupil Response as a Measure of Attitudes Toward Negroes , Ph.D. Diss., University of Colorado, 1965. Young, F. A., "An Attempt to Obtain Pupillary Conditioning with Infrared Photography", Journal of Experimental Psychology , 1954, 48, pp. 62-68. Zuckerman, M. "The Development of an Affect Adjective Check List for the Measurement of Anxiety", J. Consulti ng P sychiatry , 1960, 24, pp. 457-462. Business Week , "Admen try 'Eye-Spy' to Read Buyer's Mind", August 19, 1957. "Minicomputers for Real-Time Applications", Datamation , March 1969, pp. 39-61 Sponsor Magazi ne, "... In the Eye of the Beholder", December 28, 1964, pp. 25-29. Television Magazine , "In the Eye of the Beholder", Aoril, 1962, pp. 60-63.

PAGE 113

APPENDICES

PAGE 114

APPENDIX I: Preand Post-Test Questionnaires

PAGE 115

-103Slides Pretest Slide number: Describe it:_ Look at what is going on in the picture. How does it make you feel? Use the middle of each scale as a "neutral" Dosition, with your reaction getting stronger as you move farther tov/ard the end points of the scale. Place one mark on each line. How does this slide make you feel? Afraid Unafraid Calm Nervous Disgusted Elated Secure Insecure Gloomy Cheerful Tense Pel axed Hostile Friendly Upset Contented How would you evaluate the slide in terms of the following characteristics? Pleasant Unpleasant Unfavorable Favorable Interesting Uninteresting Negative Positive

PAGE 116

Post-Test Questionnaire

PAGE 117

-105To be completed iminediately after viewing slides Your Name Number We v/ould like you to look at copies of the slides you have just seen. We v/ould like to know how each of them made you feel when you viev;ed them a few minutes ago. Please take as much time as you want, and fill out one answer sheet for each slide. When you have finished evaluating that slide, advance the projector to the next one and fill our an answer sheet for it.

PAGE 118

-105Answer Sheet Slide Number: Describe the slide: How did this slide make you feel? Use the middle of each scale as a "neutral" position, with your reaction getting stronger as you move farther toward the end points of the scale. Place one mar^ on eadi line. How did this slide make you feel? Afraid Unafraid Calm Nervous Disgusted Elated Secure Insecure Gloomy Cheerful Tense Relaxed Hostile Friendly Upggt Contented How would you evaluate the slide in terms of the following characteristics? Pleasant Unpleasant Unfavorable Favorable Interesting Uninteresting Negative Positive

PAGE 119

APPENDIX II: BIOGRAPHICAL QUESTIONNAIPsE

PAGE 120

-108QUESTIONNAIRE PLEASE FILL THIS OUT BEFORE BEGIMING THE EXPERIMENT YOUR NAME PHONE NUMBER YOUR AGE DO YOU WEAR CORRECTIVE LENSES _(Yes/No) DO YOU HAVE 20/20 VISION WHEN CORRECTED [Yes/No) HAVE YOU EVER BEEN HOSPITALIZED FOR HEART TROUBLE OR EYE PROBLEMS (YES/NO) IF YES, WHICH

PAGE 121

APPENDIX III: PROCEDURE FORM

PAGE 122

--noSUBJECT NUMBER_ DATE READY . . . COMPUTER CAMERA RECORDER EKG SPR BLOOD PRESSURE ... PHOTOMETER . . . CHARTS ... PAPER TAPE . . . SNELLEN CHART ... FILL OUT QUESTIONNAIRE ... EXPLAIN EQUIPMENT AND WHAT WE ARE MEASURING ... HOOK UP TO APPARATUS . . . EKG SPR RESP BLD PRESS ... MENTION BUZZER-PART OF EXPT.-TRY NOT TO PULL AWAY HOLD DOWN EYE BLINK AND SWALLOWING BLINK EYES DURING BLANK SLIDES IF POSSIBLE ADJUST LEVELS ATTACH EARPHONES . . . ASK IF COMFORTABLE ... DARKEN ROOM 2 MINUTE RUN TO FAMILIARIZE WITH WHAT WILL BE HAPPENING START TAPE AND CHART . . . BEGIN PHOTOMETER RUN ... PREPARE FOR STARTLE

PAGE 123

•Ill END OF RUN TURN CAMERA TO STANDBY ASK SUBJECT TO FILL OUR QUESTIONNAIRE ON SLIDES CLEAN ELECTRODES RECYCLE TAPE RECORDER CHAIR TO DOWN POSITION DUMP DATA ON TAPE RECYCLE COMPUTER CHECK PAPER TAPE LEVEL CHECK CHART PAPER LEVEL REORDER SLIDES COMMENTS

PAGE 124

BIOGRAPHICAL SKETCH Robert. Roy Bell was born in Vincennes, Indiana, September 23, 1945. He attended grammar and high school in St. Francis ville, Illinois. In 1961, Mr. Bell moved with his parents to Cocoa, Florida, where he completed high school in 1963. In the fall of 1963, he enrolled in Brevard Junior College at Cocoa, and received the A. A. degree in 1956. In 1966, Mr. Bell enrolled in the University of Florida, where he received ihe Bachelor of Science in Business Administration degree, with honors, in 1969. He was a meinber of a team which represented the U.F. College of Business Administration at the Emory University Intercollegiate Business Ganies in 1969. During summer breaks, Mr. Bell worked for Trans World Airiineb at the John F. Kennedy Soace Center, Florida. In March 1969, Mr. Bell was admitted to graduate work at the University of Florida, and received the Master of Arts degree in 1970. His studies were financed through a graduate assistantship in the Bureau of Economic and Business Research in the College of Business Administration. The title of his master's thesis was "An Investigation of Methodologies for the Measurement of Emotional Charging". In 1970, Mr. Bell was awarded a National Defense Education Act Fellowshin to continue his studies toward the Doctor of Philosophy degree. While completing his studies and writing his dissertation, fir. Bell taught undergraduate courses in the Department of Management and Business Law. Mr. Bell is a member of the Academy of Management and the American Institute for Decision Sciences. He has accepted an appointment as Assistant Professor of Management at the University of North Florida. -112-

PAGE 125

I certify that I have read this study and that in my ooinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. iam M. Fox, Chairrpdn Professor of Management c(nd Industrial Relati^s I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Walter A. Hill, Associate Professor of Management I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. *1larvTFTr~Shaw Professor of Psychology This dissertation v/as submitted to the Department of Management in the College of Business Administration and to the Graduate Council, and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. August, 1972 Dean, Graduate School

PAGE 126

|W '^ ifkk)p. DM^^. 1 ^ 897. f1