INVESTIGATION OF AFFECTIVE FACIAL EXPRESSIONS
THROUGH SLOW MOTION AND NORMAL SPEED
VIDEOTAPE TECHNIQUES

Disser’tation for the Degree of Ph. D.
MICHIGAN STATE UNIVERSITY
DAVID J. INMAN
1976

III III III IIIIIIIIII

3 1393

. LIBRARY

sun
UM

 

This is to certify that the

thesis entitled

Investigation of Affective Facial
Expressions through Slow Motion and
Normal Speed Videotape Techniques

presented by

David J. Inman

has been accepted towards fulfillment
of the requirements for

Ph.D. degree in Counseling,

Personnel Services and
Educational Psychology

AM; (We

Major professor

Date August U, 1976

 

G7 639

 

Copyright by
David J. Inman

1976

“swam av -

m: L51 on.

THROCS

Pasial ex:

';.;5 A .
“"9“ to later

 

Iflj/L'l [I

ABSTRACT
INVESTIGATION OF AFFECTIVE FACIAL EXPRESSIONS

THROUGH SLOW MOTION AND NORMAL SPEED
VIDEOTAPE TECHNIQUES

BY

David J. Inman

Facial expressions have often been thought to be
related to internal emotional states, but previous studies
have had mixed success in determining the nature of this
relationship. The present study investigates the ability
of trained persons to identify emotions from spontaneous
facial expressions using the following six affective
categories: interest-excitement, enjoyment-joy, surprise—
startle, distress-anguish, anger-rage, and contempt-
disgust.

Facial expressions were recorded on videotape, and
the effect of tape speed (normalvs. slow motion) were
varied in two presentations (time 1 and time 2). Two
dependent measures wenerecorded: the accuracy of identi-
fication of emotions from facial expressions and the overall
number of emotions observed.

A videotape composed of a series of five second
segments was produced; in each segment a subject was filmed

while spontaneously displaying a recognizable facial

expression. 2m
:::iorI so that (-
:ze'sixth normal
meaty-nit,
Eritigan State '5
aggressions of I:
;::ies. Sgbjec‘
With each group
teas-i the nor;
13:51 SPEEd ta;
23:; three vie1
:73 group f0;r ‘
Two deper.

n
q.
U~

:centificati

Q
:R‘V

amber of a

r.
y:

0‘ “I

. «ESE two I“
My

3:.

‘Eetapes, he
Tale of the 0‘
The pr 2.
1. Tran

 

David J. Inman

expression. An identical videotape was edited in slow
motion so that each segment was slowed to approximately
one-sixth normal speed.

Twenty-nine students in the College of Education at
Michigan State University were trained to identify facial
expressions of emotions in each of the six affective cate—
gories. Subjects viewed the stimulus tapes in four groups,
with each group viewing the material twice. Group one
viewed the normal speed tape twice; group two viewed the
normal speed tape followed by the slow motion version;
group three viewed slow motion followed by normal speed,
and group four viewed the slow motion version twice.

Two dependent variables were recorded: accuracy
of identification of the predominant affect in each segment,
and number of affects seen during a segment. A multivariate
analysis of variance for repeated measures analyzed the
effects of tape speed and test-retest procedures on each
of these two dependent variables. A post hoc questionnaire
was mailed to all peOple who rated the tapes. Items on the
questionnaire measured raters' state of mind while viewing
the tapes, the quality of.the tapes themselves, and the
value of the overall study.

The primary results of this study were as follows:

1, Trained subjects canaccurately identify
facial expressions in both slow motion and at normal speed in

each of the six emotional categories.

2. In gef‘
cf facial expre;
of stations the:

1
'1

-.AA
\ .D-N
rib»!
o

3. Trait.
;:essio:s in 51

a:::rately tha r.

‘HA ‘
~ -3"
vvvvv
.

 

David J. Inman

2. In general, trained raters who view a videotape
of facial expressions in slow motion record a greater number
of emotions than if they view that same videotape at normal
speed.

3. Trained raters who View a videotape of facial ex-
pressions in slow motion do not identify expressions more
accurately than if they View that same videotape at normal
speed.

4. If a slow motion tape is viewed first, raters do
not record a greater number of facial expressions in later

trials.
5. If a slow motion tape is viewed first, then raters

identify emotions from facial expressions more accurately
in later trials.
6. Raters who view a videotape in slow motion during

the retest rate the value of the overall study lower than

raters who view the normal speed tape during retest.

These results are discussed in relation to several

theories of facial expression. It is theorized that playing

Vldeotapes of facial expressions in slow motion sensitizes

Observers to the nuances of minute facial muscular movement
and fibereby facilitates more accurate identification of

emOtimn from facial expression.

 

 

in pa

 

 

INVESTIGATION OF AFFECTIVE FACIAL EXPRESSIONS
THROUGH SLOW MOTION AND NORMAL SPEED

VIDEOTAPE TECHNIQUES

BY

17""
‘

David J: Inman

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
College of Education

1976

To Linda,
thousand inc on:
aggreciate the

zone believed

 

DEDICATION

To Linda, who has lovingly endured through a
thousand inconveniences, and who will never fully
appreciate the importance of her constancy. It helps

to be believed in.

ii

Without t
would never ha'.
rationed here

To Bill I
export served
:ts.

To Sam P
ozzy life wil

T0 Cecil
Ker”: and WI'
je"'50:“3-Ier11: as

T0 Alber
influence on r

T0 Bob }
and EXPerinen:
'Cid be noth;

Md fin:

oatr ibut e d t} I

 

45:4" ,

“31' BO:
- |
tics .

‘t' Mlke T

he students

 

ACKNOWLEDGMENTS

Without the help of a great many people, this project

would never have been possible. A few of them are

mentioned here.

To Bill Hinds whose unfailing encouragement and
suppxort served as much needed sustenance during many trying
times.

To Sam Plyler, whose gentle but powerful influence
CHIIny life will long be cherished and remembered.

To Cecil Williams, who freely gave his time and
energy, and who significantly contributed to my growth and
development as a counselor.

To Albert Rabin, whose teachings had a great

influence on me during my graduate studies.

To Bob Wilson, without whose knowledge of statistics
and experimental procedures, this noble edifice of research

would.be nothing but a clapboard shanty.

And finally, to the many people who graciously

contributed their time and energy to this project: Roger

Landvoy, Bob Tomory, Fritz Johnson, Larry Serna, Fran
StOtt, Mike Talmo, Michael Gieser and the twenty-

nine students who rated the videotapes.

iii

(I)

I 1‘ OF TABLES

to

O
(I)
--—l

or FIGURES

1:5? or APP?"*“

U.HJ .

'3‘ Duv-

..i... I...
I. Iioaoozr

Theor
En:
Evo
Uni

ht

Need
Purpcl
085i:
HYPO:
Overy

3' REVIew

Intr:
Elicg
En"
Inf
Cor
RECO:
ArI
St.
F1;
Fa
Pa'
51'
Co,
Sub<.

E .

 

”I1 "I
{v n: 0'

PO

 

 

TABLE OF CONTENTS

LIST OF TABLES O O O O O O O O O O O O O O C O O C . Vii

LIST OF FIGURES . . . . . . . . . . . . . . . . . . ix

LIST OF APPENDICES . . . . . . . . . . . . . . . . . x

CHAPTER Page

I. INTRODUCTION . . . . . . . . . . . . . . . . 1

Theory . . . . . . . . . . . . . . . . . . 3

Emotions and Facial Expressions . . . . . 4

Evolution of Facial Expressions . . . . . 7
Universals and Deception in Facial

Expressions . . . . . . . . . . . . . . 10

Need . . . . . . . . . . . . . . . . . . . 15

Purpose . . . . . . . . . . . . . . . . . . 16

Definition of Terms . . . . . . . . . . . . l6

Hypotheses . . . . . . . . . . . . . . . . 18

Overview . . . . . . . . . . . . . . . . . 18

II. REVIEW OF THE LITERATURE . . . . . . . . . . 19

Introduction . . . . . . . . . . . . . . . l9

Eliciting Variables . . . . . . . . . . . . 21

Environmental Elicitors--Situations . . . 2l

Internal Elicitors--Posing . . . . . . . 27

Conclusion . . . . . . . . . . . . . . . 29

Recording and Presentation Procedures . . . 30

Artificial Modalities . . . . . . . . . . 30

Still Photographs . . . . . . . . . . . . 31

Films . . . . . . . . . . . . . . . . . . 32

Facial Components . . . . . . . . . . . . 33

Facial and Body Cues . . . . . . . . . . 35

Slow Motion Techniques . . . . . . . . . 36

Conclusion . . . . . . . . . . . . . . . 38

Subject and Rater Variables . . . . . . . . 38

Development of Facial Expressions . . . . 39

Facial Expressions in Blind Children . . 40

Facial Expression and Personality . . . . 41

Race and Sex Differences . . . . . . . . 44

iv

FZLD’PR

ywu cu

HI.

DESIGN I

Cros
Effe
Con:
Depen;
The
The
Ele:
Concl:

The S
The P
The
The
Video
Recal
Judgi
Tape
Rater
Rater
Desi;
Post
Hypo:
Analy

 

’A\ I
_

 

2':

2

O
~o r

 

CELPQPTER

IIJI.

IVH

Cross Cultural Studies . .
Effects of Rater Training .
Conclusion . . . . . . . .
Dependent Measures . . . . .
The Dimension Approach . .
The Category Approach . . .
Electromyography . . . . .
Conclusion . . . . . . . . .

DESIGN OF THE STUDY . . . . . .

The Subject Sample . . . . .
The Primary Interview . . . .
The Structured Interview .
The Unstructured Interview
Videotape Apparatus . . . .
Recall Interview . . . .
Judging Interview Tapes .
Tape Editing . . . . .
Rater Sample . . . . .
Rater Training . . . .
Design . . . . . . . . . .
Post Hoc Rater Questionnaire
Hypotheses . . . . . . . . .
Analysis of the Data . . . .

ANALYSIS OF RESULTS . . . . . .

Introduction:
and Analysis . . . .
Subject Self Report . .

Expert Judge Agreement . . .
Revision of Design . . . .
Stimulus Tape Composition . .
Analysis of Dependent Measures
Accuracy of Rater Responses
Reliability of the Scales . .
Design I: Accuracy Data . .
Design Over Subjects . . .
Design Over Measures . . .
Frequency of Rater Responses
Design II: Frequency Data .
Design Over Subjects .
Design Over Measures .
Post Hoc Questionnaire .
Hypotheses . . . . . . .
Additional Findings . . .
Summary . . . . . . . .

Overview of the Design

Page

45
49
50
51
52
53
57
58

61

61
63
64
65
65
66
67
69
69
70
72
75
76
77

80

80
81
84
88
89
90
90
98
100
100
102
104
105
105
109
110
115
119
121

fl
5

F‘K'H
~-"‘.R
A-.. cu
Y Y (‘J‘ ‘1
J0 UFu-floRll

Some:
Concl.
Eff
c:
E“.
F.
The

Acc

The

r—a
I)

¥~

t"
'7.

“U {/7 tr: 7

H

’3

V ('7
-

O—-—4 (1 a 7 '.‘
—-

 

CWLPUPTER

‘1. SUMMARY, CONCLUSIONS AND DISCUSSION .

Summary . . . . . . . . . . . . .
Conclusions . . . . . . . . . . .

Effect of Tape Speed.on Identification

Of motion 0 O O O O C O O 0

Effect of Tape Speed on the Number of

Facial Expressions Observed

The Effect of Tape Speed on the Change
in the Accuracy of Rater Performance

Over Time . . . . . . . . . .
The Effect of Tape Speed on the
in the Number of Affects Over
Discussion . . . . . . . . . .
Validity of Stimulus Segments
Subject Self Report . . . . .
Accuracy of Rater Judgments .
Confusions In Identification
The Effect of Rater Fatigue
Test Ceiling Effect . . . .
Limits of the Study . . . .
Eliciting Circumstances .
Subject Sample . . . . .
Rater Sample . . . . . .
Dependent Measures . . .
Implications of the Study .
Theoretical Implications . .
Clinical Implications . . . .

Implications for Future Research

REFERENCES . . . . . . . . . . . . . .

APPENDICES O O C O C O O O O . O O C 0

vi

Change
Time

Page

123

123
125

125

126

127

127
128
128
130
131
133
134
135
136
136
317
139
140
141
141
145
148

151

hfie

II

II

43

A3

4.4

Literatl
not;'

Stim:

Identif
Visua

 

Table

2.].

2.2

4.4

4.5

4.6
4.7

LIST OF TABLES

Page

Literate Cultures, Percentage that Same

Emotion Is Identified with Same

Stimulus within Each Group . . . . . . . . 47
Identification of Emotion Photographs by

Visually Isolated Preliterate Observers . . 48
Emotion Categories Proposed by Five

Investigators . . . . . . . . . . . . . . . 55
Flow Chart of Procedures . . . . . . . . . . 62
Design of the Study . . . . . . . . . . . . . 74
Number of Facial Expressions Reported by

Subjects in Each of Eight Affective

Categories for the Two Interview

Conditions . . . . . . . . . . . . . . . . 82
Number of Instances of 75 Percent Agreement

in the Two Interview Conditions for Each

of the Eight Affective Categories . . . . . 86
Number of Answers and Standard Deviations in

Each Affective Category for Each Scale at

Time 1 and at Time 2 . . . . . . . . . . . 91
Coefficient Alpha for Inter-item Reliability

of Each of the Six Affective Scales . . . . 99
Spearman Rank Order Correlations for the

Relative Frequency of Identifications on

Each Affective Category for the Two

Presentations of the Stimulus . . . . . . . 99
Design I: Accuracy Data . . . . . . . . . . 101
Anova for Design Over Subjects, Accuracy

Data 0 O O O O O I O O O O O O O O O O O O 102

vii

9.51
e3u$e

L8 Anova f
Meas:
Accur
L9 Tl x P1

Affec'

 

 

 

Tkat>le

I4..8

4.:10

4.211

4.16
4.17
4.18

Anova for Interaction of Design Over
Measures with Design Over Subjects:
Accuracy Data . . . . . . . . . . . .

T1 x P1P2 Effect: Anova for Six
Affective Categories . . . . . . . .

Design II: Frequency Data . . . . . .

Anova for Design Over Subjects:
Frequency Data . . . . . . . . . . .

T2 Effect: Anova for the Six Affective
Scales . . . . . . . . . . . . . . .

Tl x T2 Effect: Anova for the Six
Affective Scales . . . . . . . . . .

Anova for Design Over Measures x Design
Over subject 0 O O O I O O O O I O 0

T1 Effect: Anova for the Six Affective
scales 0 O O O O O O O O O O O O O 0

Post Hoc Questionnaire . . . . . . . .
Anova For Post Hoc Rater Questionnaire

Anova For Post Hoc Rater Questionnaire:
Quality of Tapes . . . . . . . . . .

Anova For Post Hoc Rater Questionnaire:
Value of the Study . . . . . . . . .

viii

Page

103

104

106

107

107

108

109

110
111
112

112

112

Fizure

4.1

4.2

J.-
O
W

4.4

4.5

4.6

4,7

4,3

 

 

 

Figure

1.1

4.1

4.2

4.3

4.4

4.5

4.6

4.9

LIST OF FIGURES

Model Integrating Universal Facial
Expressions, Culturally Learned Elicitors,
and Display Rules . . . . . . . . . . . . .

Total Subject Emotions Reported in Each of
Eight Affective Categories . . . . . . . .

Number of Affects Identified by Subjects
in Each of Eight Categories in the Two
Interview Conditions . . . . . . . . . . .

Total Number of Instances of 75 Percent
Agreement of Expert Judges in Each of
Eight Affective Categories . . . . . . . .

Mean Number of Answers and 95 Percent
Confidence Interval in Each of Six
Affective Categories for All Scales at
Time 1 and Time 2 as Compared to Level
of Accuracy Expected by Chance . . . . . .

Summary of Accuracy Data: Percent of
Correct Answers in Each Category and
Highest Frequency Wrong Answer in Each
Category . . . . . . . . . . . . . . . . .

Tl x T2 Interaction Effect, Design Across
Measures, Frequency Data: Mean Percent of
Items With More Than One Answer Per
Segment at T1 and T2 for Normal Speed and
Slow Motion . . . . . . . . . . . . . . . .

Plot of Mean Scores for Rater "State of
Mind" for Speeds at Time 1 and Time 2 . . .

Plot of Mean Scores for Overall Value of the
Study for Speeds at Time 1 and Time 2 . . .

Plot of Mean Scores for Quality of the Video-
tapes for Speeds at Time 1 and Time 2 . . .

ix

Page

11

83

85

87

92

97

108

113

114

114

PEPEIDIX

A. Sign C;
on A:

B. Compos:

C“ 535160:

Affe-
D SUbjec~
5. Slbj. 9‘
Con
Inte:
8' subjEC
H. POrm U
Inte
1' Compos
J. compo:
SL131?
K' Speed
Bets
MQtiI
L. Rater
N. Rater
n, Rater
0' Scheya
R
Affec\

 

LIST OF APPENDICES

APPENDIX
A. Sign Up Sheet for Participation in Research
on Affective Communication . . . . . . .
B. Composition of Subject Group . . . . . . .
C. Subject Release Form for Study on
Affective Communication . . . . . . . . .
D. Subject Information Form . . . . . . . . .
E. Subject Instructions: Structured Interview
condition O O O O O O O O O O O O O O O O
F. Subject Instructions: Unstructured
Interview Condition . . . . . . . . . . .
G. Subject Instructions: Recall Interview . .
H. Form Used to Record Information From Recall
InterView O O O O O O O O O O O O O O O O
I. Composition of Stimulus Tape Segments . . .
J. Composition of Stimulus Tape Segments--
smary O O O O O O O O O O O O O O O O O
K. Speed of a Sample of Segments and Pauses
Between Segments Chosen From the Slow
Motion Tape O O O O O O O O O O O O O O O
L. Rater Sample Composition . . . . . . . . .
M. Rater Information Form . . . . . . . . . .
N. Rater Release of Information Form . . . . .
0. Schematic of Facial Musculature . . . . . .
P. Affect Recognition Training: Categories of

Emotion O O O O O O O O O O O O O O O O O

X

Page

161

162

163

164

165

166

167

168

169

171

172
173
174
175

176

177

.FPEXDIX
Q. Answer
R. Answer

5. Preli
Quest'

U- Data

V. Data ...‘
PerC':
AESWE

i. POSt

“1“
AJ‘,

Vt.

 

 

APPENDIX Page

Q. Answer Sheet: Affect Recognition Training . 181
R. Answer Sheet: Affect Recognition Study . . . 184
S. Preliminary Summary of Findings Accompanying
Questionnaire in a Sealed Envelope . . . . 189
T. Post Hoc Rater Questionnaire . . . . . . . . 191

U. Data Matrix for Accuracy Data: Mean Number
of Right Answers in a Category . . . . . . 192

V. Data Matrix for Frequency Data: Mean
Percent of Items with More Than One
Answer Recorded . . . . . . . . . . . . . . 193

W. Data Matrix for Post Hoc Rater Questionnaire 194

xi

:s‘fincs and 1

3°"?! Specifi-

“55- The E

Site of enot
aEI-Eat deal
Pressions are
states. who
Rah" a tr
3"“: umbrella

 

CHAPTER I
INTRODUCTION

The problem of how to analyze, measure, or record
emotion is one of the most complex areas in the field of
psychology. Human beings commonly react to their inner
feelings and emotions, and yet experimental methods and
scientific technology have for the most part failed to un-
cover specific patterns of behavioral or physiological re-
sponses that reliably correspond to these subjective experi-
ences. The focus of this study is to delineate patterns of
facial behavior which correspond to these subjectively felt
emotions.

It is commonly believed that the face is the primary
site of emotional communication, and that people can discern
a great deal from facial expressions. Moreover, facial ex-
pressions are commonly thought to be associated with inner
states. Who has not heard such expressions as, "down in the
mouth," "a troubled brow," or proverbs like, "let a smile be
your umbrella" and "the eyes are the window to the soul."
Phrases such as these would be meaningless if inner feelings
were not thought to be associated with facial expressions.

One relevant question would appear to be "Why are the

actions 50 C:
trough the f'I

£bfin<I
in its locati:
attributes re.

The face is t:

3-:O'v'er' the'
:ate and Com:
_..-H I
“l‘a‘rles are 8.

Enable parts I

toe muscular
But “ha
3‘ I

racial mus

C
.or the man

i?
.‘it lower an‘
a less c

7‘37 "

3.3.les is tr
2?

.3 (1971)
432$:th Of '
fies) they
:«SCE‘d in a
M .
“fit

corn
.qu‘dhir

=§9ress'

 

emotions so commonly thought to be communicated specifically
through the faCe?"

Ekman (1972), notes that the face is both commanding
in its location and complex in its musculature; these
attributes render it an excellent medium of communication.
The face is the locus of many sensory organs-~eyes, ears,
nose and mouth; and being situated on top of the torso, the
face commands more attention than other parts of the body.
Moreover, the muscles of the face form one of the more intri-
cate and complex muscular systems in the body. The facial
muscles are small, highly mobile, and attached to easily
movable parts. Thus the great potential for movement and
the muscular complexity of the face also commands attention.

But what is the purpose of this highly complex system
of facial muscles? Of course some muscles are necessary
for the manipulation of the mouth and the sensory organs,
but lower animals move these organs very efficiently with
a less complex facial muscular system (Huber, 1931; Gregory,
1929). It seems possible that a partial function of facial
muscles is to communicate information about the organism.
Izard (1971) reported that when Rhesus monkies were denied
movement of facial muscles (through cutting the facial
nerves) they were attacked twice as often as normals when
placed in a new colony. Thus the face of the normal monkies
nmmt.communicate some information which serves to inhibit

aggression toward them in other monkies of the same species,

and thus are l

The fac;
:icate infor:;'
nobile set of
tion to occur

and the conte:

zest facial e:

in the Verbal I

 

and thus are less likely to be attacked.

The facial muscle system may then function to commu—
nicate information to other organisms, and a complex and
mobile set of facial muscles is necessary for this communica—
tion to occur. The process by which the face communicates
and the content of this communication are the topics of
most facial expression research. Ekman (1969) noted that
the purpose of most research in the area of non-verbal
communication is to gain information which is not available
in the verbal channel. Most researchers in this area have
felt that the face is the primary source of emotional commu-
nication. The importance of facial behavior was eloquently
summed up by Duschenne in 1862 when he stated, ”When the
human mind is perturbed, the face becomes a vivid picture in
which the passions are rendered with much tenderness and
energy, in which each motion of the mind is reflected by a
line and each action by a vivid impression which precedes
the expression of will. It uncovers and reflects by varying

signs the images of our most secret emotions."

Theory

There are several issues involved in explaining how
facial expressions might be embued with emotional meaning.
The first issue involves the development of a theory that
explains how emotions are associated (on a physiological
level) with facial expressions. Another issue is how such

facial expressions have developed in an evolutionary context.

The final iss

variability o

anions and I

The fir
aezhanisms by
expressions.
theorists who
back as a con
Cannon (1927)
Ei'eal Skeleta

this concept

Patti

CUlar is

 

U,“
N:

body mils
I‘L ‘

"3"3‘88 whiC
Ja‘nsl E
.\_ theOr

‘T‘I'n ,
l”aver: Jan-I
33:16
3 a
S 1
Sir

 

4

The final issue involves the question of why there is such

variability of research success in this area.

Emotions and Facial Expressions

The first issue to be reviewed is the theoretical
mechanisms by which emotions become associated with facial
expressions. The discussion here is confined to those
theorists who provide some basis for skeletal muscular feed-
back as a component of emotional experience. James (1884):
Cannon (1927) , and Arnold (1960) , all emphasize the importance of
general skeletal feedback in producing emotions. Tomkins extends
this concept and argues that facial muscular feedback in
particular is essential to emotional experience.

William James, perhaps the most famous originator of
an affective theory, states that "the bodily changes follow
directly the perception of the existing facts and that our
feelings of the same changes as they occur IS emotion"
(James, 1884). Thus in order for an emotion to be experi-
enced, the environmental situation must first be perceived,
the body must then react to this perception, and the bodily
changes which ensue are experienced as emotion. Reviews of
James' theory (Cannon, 1927; Wenger, 1950) have focused
pmimarily on visceral changes and their effect on emotion.
However, James also included the skeletal or voluntary
nmscles as important in emotional experience. Since the
skeletal muscles include the facial system, it is possible

to extend James' theory to include facial reactions to

stimuli. The

 

along with 0:

 
  

:‘Ie experienc
Other a
.‘uoldl also

an integral E

 

 

 

352-35. I103." eve
insular and

fi‘.
- I:
1%..‘81' “Co‘s:

stimuli. Thus feedback from the facial muscular system
along with other visceral and muscular changes would produce
the experience of emotion in James' system.

Other affective theorists (Cannon, Lindsay, and
Arnold) also include feedback from the skeletal muscles as
an integral part of the experience of emotion. Unlike
James, however, these theorists do not believe that
muscular and visceral feedback is the sole cause of emotions.
Rather, muscular feedback is seen as an important but not
essential component of the experience. Sensations which
are construed as emotions are felt to arise from a variety
of sources: thalamic structures (Cannon, 1927), cortical
processes (Arnold, 1960) or reticular structures (Lindsay,
1957). None of these authors discuss facial muscles
specifically, but each has implied possible facial muscular
involvement under the general heading of skeletal muscular
feedback.

Sylvan Tomkins' theory of emotion (Tomkins, 1962, 1963,
1964) is significant in its treatment of the face as an
essential part of affective experience. According to
Tomkins, affects are the primary motives of human beings.
He distinguished affects from drive stimuli in that drives
provide information about deficiencies in the physiological
system, while affects provide motives for responding to these
deficiencies. Regardless of what information an individual

receives about his drives, he is not likely to act on it

".ttil he has 5

 

ample, if at
iily, he is ll
ineating (lfl'l
:iiile of a h.

V u
‘ o
4‘ '5‘,th
u‘ l‘Jhd h '.'
-|. “J 8 ll.

 

imes li.e..
rectify this

=2»-

x-r\terr0r

~A ‘C

tsgerEntiE

aw

:57-

. “e"lced ‘
1

£3:

until he has some affect associated with the drive. For
example, if an individual experiences hunger while sitting
idly, he is liable to eat because he becomes interested
in eating (interest-excitement is defined by Tomkins as an
affect). On the other hand, if this individual is in the
middle of a heated argument, and is experiencing anger, then
although he might obtain the same information from his
drives (i.e., that he is hungry), he will not behave to
rectify this situation until he has achieved closure on his
anger, thus permitting him to attend to the hunger drive.
According to Tomkins then, one does not eat merely because
one is hungry: one must also be interested in eating.
Tomkins defines eight specific motivational states
which he calls affects (Tomkins, 1963, 1964). Each of these
affects he has labeled with a dual term which symbolizes
both the mild and the severe manifestations of this parti-
cular affective state. The emotions defined by Tomkins
include: interest-excitement, enjoyment-joy, surprise-
startle, distress-anguish, anger-rage, contempt-disgust,
fear-terror, and shame-humiliation. These eight states are
differentiated by varying rates of neural firing over time,
and each have specific reward values, being innately ex-
perienced as either positive or negative conditions. The
positive affects (joy, interest, and surprise) are innately
rewarding experiences, and the negative states (fear, dis-

gust, anger, distress, and shame) are innately unpleasant

II

and negative E
::-:i‘:ational 5;
internal SYS’C'i
re triggered
Perhaps
:‘tecry is his
of affective ‘
internal inte:
:crnri-‘ticate a:
finer and inw I
3': Steers or
1364). The e
”:1 affect is

3&5 a positiy

mie in that
.* +
"‘e of the
in

 

usxal exPre I
tin.

~j‘cte

Pfe~

z;31

and negative experiences. Thus, through a system of
motivational affective states, Tomkins has contructed an
internal system of affective rewards and punishments which
are triggered by events in the environment.

Perhaps the most significant aspect of Tomkins'
theory is his belief that the human face is the primary site
of affective behavior. "Affects are not private obsure

internal intestinal responses but facial responses that

 

communicate and motivate at once publicly outward to the
other and inwards to the one who smiles or cries or frowns
or sneers or otherwise expresses his affects" (Tomkins,
1964). The expression of affects has then a dual outcome:
an affect is at once an internal differentiated state which
has a positive or negative motivational value, and simul-
taneously it is a facial expression which has communication
value in that it informs the environment of the internal
state of the organism. To Tomkins, the face serves as the
mirror of the emotional state of the individual, and

communicates this internal state to the world at large.

Evolution of Facial Expressions

 

Charles Darwin, in his work The Expression of Emotion

 

in Man and Animals was the first theorist to explain how

 

facial expressions may have become associated with emotional
states (Darwin, 1872). Utilizing Duschenne's work on ex-
pressions (Duschenne, 1862), Darwin treated facial expres-

sions as observable behavioral correlates of internal

(N‘-

:SSiC
e?»

U
A.‘

Lie SDECle

5

.'H
1 Y A

'b.

.
yl‘?
it
§
Darw
I!“
hasn‘t-A.-
Av
p...
.u‘

'e for
arxin EX'

6“
L. .'
“Film: I

My

V

‘V
.A'
n

ctlve 5

Q
n ~fl
...‘.u S
vy“
I
L. .
ru-
5
U

V

St:1
“a

Aiv
“Us

 

 

 

3101:

x1"

affective states. He believed that these external displays
had survival value by communicating to various members of
the species the internal state of the organism.

Darwin's recognition that facial affects had survival
value for a species was vital to his understanding of human
behavior. In The Expression of Emotion in Man and Animals
Darwin examined such states as weeping, laughter, love,
devotion, anger, contempt, disgust, surprise, fear, and
shame. It is noted that many of these affective categories
investigated by Darwin are those which Tomkins mentioned
fifty years later. Thus Darwin had made a lasting contri-
bution to the field long before emotions and facial
expressions were popular t0pics of research.

Sixty years after Darwin, Plutchik forewarded a
theory of emotional development. In his book entitled
The Emotions: Facts, Theories, and a Model, Plutchik (1962)
postulated several ideas. First of all, there exist only a
small number of pure emotions; other more complex affects
are combinations of the few primary affects. Moreover,
Plutchik believed that the primary affects differed both in
physiology and behavior. Thus the affects themselves were
merely hypothetical constructs defining a variety of physio-
logical and behavioral responses. According to Plutchik,
the primary affects consisted of fear, disgust, wonder, anger,
elation, and tenderness. Each of these emotions is associa-

ted with a behavioral "instinct" which has survival value

 

 

;ecies it
survival v
learning t
I’m that

Plut
corresponj

fer the e:

for both individual members of a species and for the
species itself. The affect of fear, for instance, has
survival value in the instinct of flight. Obviously,
learning to be afraid of a situation and learning to flee
from that situation are both essential elements of survival.
Plutchik deliniates the following instincts which
correspond to different emotions: repulsion is the instinct
for the emotion of disgust, curiosity for the emotion of won-
der, pugnacity for the emotion of anger, self-assertion for
the emotion of elation, and parental instincts for the emo-
tion of tenderness. Each emotion is associated with an
instinct or pattern of behavior which occurs in response to
environmental situations and contributes to the survival of
the individual and the prolongation of the species.

Irenaus Eibl-Eibesfeldt in his book,Love and Hate

 

identifies various facial expressions of emotion which differ
from either Plutchik or Darwin (Eibl-Eibesfeldt, 1970).
According to the author, affective facial displays have two
functions: mediating intraspecific aggression,and consoli-
dating bonding behavior among members of a species. For
example, behavioral displays of deference among members of

a species short circuit the aggressive behavior of dominant
members which, if carried to an extreme, would be harmful

to the group. To communicate surrender by a facial

deference display causes further violence on the part of

the dominant member to be unnecessary, thereby maintaining

10

dominance heirarchies with little physical harm to group
members. Correspondingly, the author believes that the
communication of positive affect among members of the species
consolidates bonding behavior. For example, a smile accom-
panied by an eyebrow flash (a quick raising of the eyebrows)
is a positive affect display which is mutually reinforcing
for the participants. This type of communication serves to
tighten the interpersonal bonds between members of a

species, and this bondedness has lasting survival value

(Eibl-Eibesfeldt, 1970).

Universals and Deception in Facial Expressions

 

Thus far the question of how certain recognizable
facial expressions are associated with specific emotions
has been examined from both a theoretical and evolutionary
perspective. However, the question of why facial expressions
sometimes do not correspond to these emotional states
has not yet been discussed. In this regard Ekman's model
of facial displays and Haggard and Issacs' theory of Micro-
momentary Expressions will be examined.

Ekman believes that certain affects occur in all
cultures (Innate Affect Programs). However different cultur-
al norms alter the situations that trigger these affects
(Elicitors) and the way these affects are displayed

(Display Rules). In his work entitled Universals and

 

Cultural Differences in Facial Expressions of Emotion (1973),

 

Ekman forwards a model which takes into account both the

11

consistency of facial expressions across cultures, and the
inaccuracy some studies have found in the identification of
emotions from facial expressions. In Ekman's model (see
Fhmne LJJ there are four elements: the Elicitors, the Innate

Affect Program, the Display Rules and the Consequences.

MLML

 

 

 

Q
CONS! HCES
] Imvuv ““
RULES
hem
VACOAL AFFECI FIDGRNI
'0'3 llutor Adaptive
ELICI Fuel-l Niel“ In P. as
run-on Assoclaclon llth ""'
enz:.0" ... Verbal. Vocal
Ex fatlon. vavlnoes olsgust
H-:::I.s Anger Sadness thsloioglcol
Etc 50'PF'IO Interest
' Four
Intmfly
OolntonsUOy
neutralize

Mask

   

J

CUUWLLY VARIABLE
(In part)

Source: "Universals and Cultural Differences in Facial
Expressions of Emotion" (Ekman, 1972).

Figure Ll. Model Integrating Universal Facial Expressions,
Culturally Learned Elicitors, and Display

Rules.

The Innate Affect Programs which Ekman discusses
involve several facial expressions which are associated with

affective states across several different cultures. Although

 

 

 

"'e cross

-r A
‘. u
“-33 .er

\

rz’JSt me“

h.“ ‘ ‘
4.003190

g:

.fic.

12

the cross cultural comparison studies will be discussed in
greater detail in Chapter II, Ekman believes that affective
states are identifiable across cultures through common
facial expressions, and these expressions are not modified
by cultural learning. This is not to say that the events
which elicit the facial expression are constant across
cultures, but the expression itself, as indicative of an
affective state, is not modifiable through cultural or
interpersonal experience. For example a smile accompanied
by minute muscular movements around the eyes is indicative
of a joyful state across cultures. The situations which

stimulate this joyful state may vary, but the expression of

 

the upturned corners of the mouth and the muscular movements
around the eyes have the same meaning in every culture
studied (Ekman, 1973).

The Elicitors involve learned situations which

 

stimulate a particular emotional response: these can be
learned both culturally and interpersonally. An example of
cultural learning of eliciting circumstances is that for
most members of American culture, the devouring of raw,
uncooked meat, might stimulate the affect of disgust. On
the other hand, members of another culture might not feel
disgust when eating raw meat since their cultural norm pro-
scribing the cooking of meat differs from our own. Thus
for the same eliciting conditions (i.e., the eating of

raw meat), the emotional and facial response might be very

 

 

 

if ferent

't‘e'mers:

as those <
scribed c:
the familj

lfltEfgers:

ll’lg me

Uh g

~
“‘ERS‘C
is .h .

:

s

13

different because of the individual's cultural learning.
Interpersonally learned elicitors operate in the same way
as those culturally learned, except they are not the pro-
scribed cultural norm. These elicitors may be learned in
the family, through peer interaction, or through almost any
interpersonal situation.

The concept of Display Rules is Ekman's most interest-

 

ing contribution to a theory of facial expressions. Display
Rules are processes through which the innate affect program
is modified so that the display of the affect is less

obvious to an observer. For example, a male may have learned
to experience fear when confronted with a particular object
in the environment. However he may also lave learned that

it is not appropriate for males to display a fearful expres-
sion publically. A male in this culture might then utilize
display rules to alter the expression of fear and thus
disguise his expression of the affect.

The display rules that Ekman describes include
intensifying, de—intensifying, masking, and neutralizing.
Intensifying is the facial exaggeration of the emotion which
is being experienced. De-intensifying is diminuation of a
facial expression so that it might not be noticed so
readily by observers. Neutralizing involves inhibition of
facial musculature so that the expression is suppressed.
Masking involves displacing one facial expression with

another which the person has learned as more acceptable in

 

 

this Particu
{2195: the i
in SUCh a we
accurately I

Ekman'
crass-CUR“
and the ext:
within a P51
facial ex;r!

internal af

is to filte
learn the t
environment

A dis
Haggard and
shortening

at u
men prose

I‘VRNA
Jugcs, bot
nsrn

Mal spee

oservabl

“Women
173i

“9 Ekman
Vie'

"ed as t
I
Coxr

14

this particular situation. Through these four display
rules, the individual is able to alter the affect program
in such a way that the innate facial behaviors do not
accurately represent the felt experience.

Ekman's theory of display rules explains both the
cross-cultural identification of certain facial expressions
and the extreme variability of learned expressions even
within a particular culture. He concludes that although a
facial expression is sometimes an accurate reflection of an
internal affective state, it may be at times a deceptive
message. The problem in many studies of facial expressions
is to filter out facial deception in order that one may
learn the true facial eXpression which is elicited by an
environmental situation.

A display rule not mentioned by Ekman but studied by

Haggard and Issacs (1969) might be called compacting--a

 

shortening of the time in which a facial expression occurs.
Their procedure was to play films of a therapy hour to
judges, both at normal speed and in slow motion (one-fifth
normal speeéD. The authors found that many expressions
occur in a fraction of a second and were therefore almost
unobservable at normal speed; they dubbed this phenomenon
"Micromomentary Expressions" (Haggard and Issacs, 1969).
Using Ekman's model, micromomentary expressions could be
‘viewed as the consequence of a display rule operating to

"compress" unacceptable affect into a form which is not

face in

,.

iazzard

ine so

?
5

~an

A
p-
5..

. n"
"V b-\4’..

.
‘ «.321

5C

‘An
‘w._

uj CC‘

I

 

 

 

15

visible to the observer. Thus rather than altering the
face in a particular way in order to mask an affect,
Haggard and Issacs suggest that the facial eXpression of
affect might not itself be altered, but merely compacted in
time so that the occurrence of the facial expression would
be almost unobservable.

From this discussion of the theoretical background
of facial expressions, several key points emerge. The
physiological system of affects in the human organism is
set up in such a way that muscular feedback from facial
expressions could be important to the felt experience of
emotion. Secondly, in our evolutionary history, the
communication of affect through facial expression may have
had survival value. Finally, although several universally
recognized facial expressions seem to exist, these expres—
sions can be modified in many ways to mislead observers

and cause inaccuracies in studies of facial expressions.

Need

 

Although theie is evidence to suggest that facial
expressions are associated with emotions in still photo-
graphs (Ekman, 1972; Izard, 1971), no previous study has
determined whether spontaneous facial expressions recorded
on videotape can be accurately identified with the categories
used in the cross-cultural studies. Moreover although some
initial work has been done (Haggard and Issacs, 1969),

a controlled study has yet to examine the effect of observing

d

‘-
§

:t

g. as

a.ter
accu

.3

 

 

 

16

a videotape of facial expressions in slow motion. Finally,
if as Haggard and Issacs suggest, slow motion speed changes
observers' perception of facial expression, does this change
alter the perception of later tapes played in the normal
speed condition. The present study seeks to determine the

answers to these important questions.

Purpose
The purpose of the present study is to determine
whether the emotional meanings of facial expressions can be
accurately identified on videotape under a variety of con-
ditions. Variables which are thought to influence judgments

are the eliciting circumstances, speed of_the videotape,

 

 

and the order in which different speeds are presented. The

 

number of facial expressions identified as well as the

accuracy of the identifications are both dependent measures

 

in the present study.

Definition of Terms

 

Eliciting circumstances are defined as those conditions

 

under which the subject displays a facial expression. In
this study, two types of eliciting circumstances are
employed: a structured intervew and a relatively unstruc-
tured interview. During the structured interview, the sub-
ject is presented with photographs of facial expressions
which have previously been validated, and is asked to dis-

cuss the emotions depicted in these photographs. During this

the the SL1}:
expressions
tired inter:
can categori
ccndition he:
experinehte:
for later v:

The v;
interview cr
+3199 is pla;
feet/Segofld
at an avera

The P.

u. .
.49 flrSt 1:

 

17

time the subject is being video-taped so that his own facial
expressions can be later viewed and rated. In the unstruc-
tured interview, the subject is asked to define and label his
own categories of emotions and discuss them, but in this
condition he is not provided with photographs or with the
experimenter's list of labels. This interview is also taped
for later viewing

The videotapes of facial expressions made in the two
interview conditions are re-played to raters in two

videotape speeds. In the slow motion condition, the video-

 

tape is played at approximately 1/6 normal speed, or 10
feet/second; in the normal speed condition, the tape moves
at an average of 60 feet per second.

The presentation order refers to two treatments. In

 

the first treatment, normal speed videotape is shown first
followed by slow motion, and in the second treatment, the
slow motion tape is shown first, followed by the normal
speed tape .

The effects of these three variables--eliciting cir-
cumstances, videotape speed, and presentation order--will
be recorded in terms of two dependent measures. The first

is the accuracy of the rater's identification of facial

 

expressions. the second dependent measure is the number
of facial expressions observed by a rater during a segment

of the tape.

18

Hypotheses

 

The first hypothesis is that a slow motion videotape
will allow observers to judge facial expressions more
accurately, and will allow observers to see a greater number
of facial expressions.

The second hypothesis is that observers will record
a greater number of eXpressions and have greater accuracy
in identifying these expressions when viewing the unstruc-
tured interview than when viewing the structured interview.

The third hypothesis is that when the slow motion
videotape is seen first, the subsequent videotape will be
perceived more accurately and a greater number of facial
expressions will be observed than when the normal speed tape

is seen first.

Overview

 

In Chapter II the literature pertinent to the area of
facial expressions and emotions will be reviewed. In
Chapter III the design of the study will be described
including the sample employed, the measures used, the speci-
fic hypotheses to be tested, and the data analysis techniques
employed. In Chapter IV the analysis of the results will be
presented, and in Chapter V the study will be discussed and
the conclusions will be reviewed along with a consideration

of implications for future research.

 

“v. wt Wu a; ad a x .n .
l l 3. .1 .L C. C
E e 3. C S J u:

an“ r .7. .3 e ... a3

A. t a S c . r u... C.

 

CHAPTER II

REVIEW OF THE LITERATURE

Introduction

 

The popularity of research on facial expressions has
varied widely during this century. Ekman (1972) cites a
number of studies on facial affect during the late 19203 and
early thirties followed by a dearth of studies for 15 years
thereafter; Possible reasons for this change in pOpularity
are many. Early studies by Landis (1924, 1929) strongly
suggested that there was no consistent relationship between
facial expression and emotional state. Although these
results were soundly rebutted by later researchers (Murphy,
Murphy and Newcomb, 1937; Davis, 1934), the field of
psychology thereafter immersed itself in a behavioral
period where there was little interest in investigating such
invisible,and therefore hypothetical,states such as emotions.
After 1950, improvements in recording techniques, along with
the realization of the flaws in early studies, led to a new
look at facial expressions. Moreover, at this time
psychology was entering a period wherein the topic of
emotion was more acceptable in experimental research

(Ekman, 1972).

19

 

 

 

The
research 1
to the la<
The of the
Eraser an;

of facial

generaliz

20

The improvement in respectability of facial expression
research has not deterred any of its critics from pointing
to the lack of success in many facial expression studies.
One of the shortcomings in facial research, pointed out by
Bruner and Tanguri (1964), is that the results in the field
of facial research have not been "additive;" the knowledge
gained from one study has not contributed to the performance
of other studies. Though much research has been done, each
study seems to exist in isolation and it is difficult to
generalize the conclusions to other studies. In searching
for an explanation (fl? this phenomenon, Bruner and Tanguri
note that a variety of techniques have been employed to
stimulate and record subjects' emotional responses: still
photographs, motion pictures, drawings, transcripts, vocal
stimuli, and models. Moreover, different methodologies
have been used to measure dependent variables: dimensions,
categories, free vs. forced choice, and electromyography.
Because of this variety of procedures, studies of facial
expression have met with mixed success, and it was not
until the late 19605 that reliable and effective studies of
facial expression were performed. The early discouragement
of researchers on the tOpic of emotions was expressed by
Duffy (1941) who suggested that because of the problems in
measuring emotions, the concept should be abandoned altoge-
ther and other concepts develOped where the meanings would
not be so idiosyncratic, and research would prove more

feasible.

21

The following review of the literature will be organ-
ized by procedures rather than by historical development.
The aspects of procedures to be examined will include:
eliciting conditions, recording procedures, subject and
rater variables, and dependent measures. Through this format
it is possible to isolate both what has been learned from
previous research and what has yet to be studied definitive-

ly on the topic of facial expressions of emotion.

Elicitingariables

 

The "eliciting circumstances" is a term borrowed from
Ekman (1972) which refers to a procedure or set of condi-
tions which are employed by the experimenter to stimulate
or elicit the facial expressions to be portrayed. There are
two types of eliciting circumstances employed in most
studies: when the subject is asked to react to an environ-
mental situation, he is responding spontaneously to external
stimuli. On the other hand, when the subject is instructed
to pose an emotion, he is responding to internal, pro-
prioceptive stimuli in order to produce a facial expression.
The type of instructions used--spontaneous vs. posed--seems
to have an effect on the elicited facial expression, and

therefore these two conditions will be considered separately.

Environmental Elicitors--Situations

 

Eliciting conditions relying on environmental stimuli

were first employed to study facial expressions by Feleky

 

 

(1914);
of situa
Althougl
:al IES‘
were co:

=V“'ess:

blA-t‘
.

f51‘

\- Eng
‘1-
"Qt n

22

(1914); he photographed one hundred subjects in a variety
of situations spontaneously expressing a number of affects.
Although this study was basically exploratory, and statisti-
cal tests were not performed, Feleky believed that emotions
were consistently associated with several facial
expressions.

Like Feleky, Charles Landis employed environmental
eliciting circumstances to obtain emotional responses
from subjects (Landis, 1924: Landis, 1924; Landis, 1929).
In studying "normal emotional behavior" rather than posed
expressions, Landis was interested in whether people
could accurately label facial expressions which were
elicited by environmental stimuli. His procedure was to
present the subject with a variety of situations and then
to photograph any facial expression changes he observed.
Faces of subjects were smeared with burnt cork to make them
more observable. Observers were then asked to match these
photos with the situation they thought elicited that facial
expression. Examples of Landis' situations were: "listening
to popular music, truth or falsehood telling, decapitation
of a live rat, and technical music" (Landis, 1924). Landis
found no consistent association between facial expression
and environmental situations; moreover, no facial expres-
sion was found to be correlated with the verbal report of
felt emotion. The conclusion drawn from these studies was

that "emotion" was more in the perception of the nature of

23

the situation and was not a differentiated state. Landis
believed that facial expressions were learned reactions
corresponding to particular environmental situations.
Because individual learning is so variable, emotions could
not be accurately judged from facial expressions alone
(i.e., without knowledge of the environmental situation).

Landis' conclusions were controversial, and many
criticisms of his procedure were forthcoming. Davis (1934),
in his review of Landis' studies, cited the accumulative
effect of the series of eliciting circumstances as a possible
confounding variable. For example, a subject may have first
been asked to decapitate a live rat (to elicit the affect of
disgust) and later been told an amusing story by the same
experimenter (to elicit amusement). The disgust response
from the previous condition may have generalized to the
later condition, thereby invalidating the later eliciting
condition. A three hour series of such a diverse array of
treatments for a single subject could have profound effects
on the expressions elicited.

Fois-Whittman (1930) cited the effect of individualized
learning on the facial expressions recorded by Landis.
Emotional responses to particular stimuli are usually
learned, and thus not generalizable across subjects. A
particular situation (classical music listening, for example)
may elicit the affect of joy in one subject while eliciting

the affect of boredom in another. Ekman (1972) also cites

24

this criticism as limiting the validity of Landis' findings.
The same situation may have elicited different affects
across subjects, and thus a consistent association between
a facial expression and an eliciting situation would be
unusual.

Murphy, Murphy, and Newcomb (1937) have argued that
Landis' procedures encouraged the masking of facial expres-
sions. Subjects' faces were marked with burnt cork,
possibly causing self-consciousness about facial movement.
Moreover, all of the subjects employed by Landis were
psycholoqists with professional status to maintain, and all
of them were personal aquaintances of the experimenter.
These subjects, the authors argue, may have been motivated
to mask or obscure specific emotions in certain situations.

Ekman (1972) notes that Landis used still photographs
rather than film; some research (Haggard and Issacs, 1964)
indicates that certain facial expressions occur so quickly
that they are unrecordable except through the use of slow
motion equipment. Moreover, Landis makes no mention of how
some photographs were chosen for presentation in the study
while others were screened out. Ekman argues that the
procedures for screening photos or film is important to the
validity of the findings. Ekman's final criticism of Landis'
work falls into the category of dependent measures, i.e.,
how results are recorded, what labels are used for facial

expressions, and how these labels are chosen. Landis

 

allowed 5
inroduci

The large

exact t8}

Da'
here we
Of 131';

#5

5 verns

H:
V"
.

25

allowed subjects to freely choose their own labels, thus
introducing a variety of semantic problems into his study.
The large number of emotional labels to choose from renders
it improbable that subjects would independently agree on an
exact term for describing an emotional state.

Davis (1934) reviewed Landis' data and argued that
there were statistical problems with the original method
of analysis. He stated that a number of facial reaction
patterns could have been obtained from the self-report data
if the proper statistical procedure had been utilized.
Furthermore, he argued that some facial expressions are
specific to situations and others have common features which
generalize across situations. Upon reviewing Davis' paper,
Landis stated that the chief conclusion flint there is also
a correspondence between a subject's self report and facial
response) was essentially correct; thus he revised his original
position (Davis, 1934).

In an attempt to put Landis' conclusions to a further
test, Fois-Whittman (1930), studied the effect of situational
cues on the judgment of facial expression. He concluded
that there was consistently "a unique and distinctive
pattern . . . of muscular involvements" for every emotion
studied. In contrast to Landis, Fois Whittman utilized
pg§2g_subjects in his research and thus there were no possi-
ble confounding variables in his environmental eliciting

circumstances.

.a'

‘1‘;—

.7
Lo- .u
a

to obse

‘

g S C l w... ".1. n .
«a.» .3 ad .e . d s... Q»
A: C . «G .t t a»

J r .. e 7‘. a .3"

r u. 0» L . sun“ ~ .4 .s e

0 T“ L N J . . a. .2
5» 9 s :u .q 4 I u thU

 

A

5x:

26

After Fois-Whittman, several studies have compared
the effect of presenting photographs of facial expressions
to observers with and without situational cues (Munn, 1940;
Frijda, 1958; and Watson, 1972). Both Munn (1949) and
Frijda (1958) found that although situational cues were
helpful in identification, emotions can be judged from
facial expressions alone without the help of the situational
cues. Watson (1972) found that cues utilized by observers
to judge emotions depended on the intensity of the facial
expression and the environmental situation involved. When
the face was neutral, the emotion was judged from the
situation; when the situation was neutral, the emotion was
judged from the face (Watson, 1972).

These three studies may in part explain Landis'
negative findings. If there is a complimentary interaction
between observers' use of facial and situational cues in
judging emotion, then the presentation of ambiguous facial
expressions (by employing faulty eliciting conditions) would
fpggg_the observer to respond to the situational cues alone.
Under better eliciting circumstances, however, subjects would
produce less ambiguous facial expressions, and observers
might be able to judge emotions from these expressions alone.

Recently, two studies have explored the effect of
environmental eliciting stimuli on spontaneous facial
expressions (Gubar, 1966; Lanzetta and Kleck, 1970). In
both studies, subjects were observed in two treatment condi-

tions: one treatment group was punished with an electric

1132271.

A

or not

'1 t
t u.
l 0
vi. an.”
.1 .
kg u.
a: '

 

t
rnal

\v

eel
(flare .

V

e
stio
C.

Til-av“.
‘oAb " a.
. .
5"“
‘U
J
‘.‘
«.45

27

shock, and the other was not. The observer's task was to
watch a videotape of the subject's face and to tell whether
or not the subject was anticipating a shock. Both studies
concluded that observers could accurately judge the shock
from the non-shock conditions through the videotape of the
facial expressions. Thus, the use of better eliciting
circumstances and the simplification of the judging task

(by limiting responses to two categories) improved observers'
ability to accurately judge the meaning of facial expression

without knowledge of the situation.

Internal Elicitors--Posing

 

In procedures relying on internal elicitors, the
subject responds not to an external situation but to inter-
nal proprioceptive cues or emotional states. Two basic
questions are usually asked in this research: 1) can
emotions be posed and recognized accurately, and 2) do
internal and environmental elicitors of the same affect
produce different facial responses.

The major work on the topic of internal elicitors
was done by Coleman (1949) and had a dual purpose: to
explore whether posed expressions could be accurately
recognized, and to determine whether some areas of the face
are more crucial to recognition than others. Utilizing
16mm film, Coleman recorded expressions elicited by both
external and internal stimuli; frames were then chosen from

the film and viewed singly by judges.

 

 

Col
of elicit
Sane exp:
spontanec
when sub:
were more

in others

8.70.1033

28

Coleman found no clear pattern in comparing the types
of eliciting circumstances or the importance of facial areas.
Some expressions were more accurately judged when they were
spontaneous, while others were identified more accurately
when subjects were asked to pose them. Certain expressions
were more accurately identified through eye movement, while
in others the mouth was more important. Overall, posed
emotions were more accurately identified when observers were
allowed to view the mouth region than when they viewed the
eyes. Coleman concluded that personality factors, facial
region, and type of stimulus condition all affected the
accuracy with which observers judged facial expressions
of emotion (Coleman, 1949).

In a similar study, Thompson and Meltzer (1964) asked
50 subjects to pose each of ten affects and administered
the California Psychological Inventory to them. They found
as did Coleman, that no specific personality variables were
correlated with posing ability; the authors did conclude
however, that an affect can be posed by subjects and recog-
nized accurately by observers. Thompson and Meltzer,
agreed with Coleman in concluding that facial expressions
could be posed accurately, but the level of accuracy is
related to many personality and situational variables.

The previous two studies have shown that posing ability,
type of emotion to be posed, and region of the face all

affect the accuracy of raters' judgments. However, both

Clo-‘AvL‘Q

m

csefl
3192C

U
u

L
tolere

'

 

“v.

A:
«4 u

expat;
‘2.“- -
‘L

5‘;

29

studies concluded that facial expression can indeed be
posed and judged with some degree of accuracy. Moreover,
there seem to be differences between the facial conse-
quences of posed and spontaneous elicitors but as yet the

exact nature of these differences is not clear.

Conclusion

 

In reviewing the two kinds of eliciting circumstances,
Ekman (1972) makes several recommendations to correct errors
in previous research procedures. More than one type of
eliciting circumstance should be selected so that the
generalizations from the study may be more broadly extended.
He points out that many studies have been limited because
experimenters have not attempted to include more than one
eliciting circumstance. This limited generalizability
in many facial studies may contribute to the "non-additivity"
mentioned in the area of facial research.

Ekman's second recommendation is that the experimenter
attempt to ascertain whether the eliciting stimulus does
indeed elicit the intended emotion. Researchers who attempt
to match facial expression with situations should account
for the effects of social learning in determining the facial
response. Thus an attempt should be made either to choose
a stimulus which would elicit the same emotional response
across subjects, or to tailor the eliciting circumstance to

the subject's previous learning and experience.

,5 er‘;
s‘;-".
".b- ‘

-
V

 

e
Ru
‘3
Pm

30

Finally, Ekman contends that an attempt should be
made to explain the relationship between the eliciting circum-
stance and the naturally occurring emotion. Thus if posing
is employed, the researcher should state how this type of
stimulus would effect the facial response. Even in sponta-
neous conditions, the experimental situation alters the
subject's perception of the eliciting circumstance, so the

facial response can never truly be "natural."

Recording and Presentation Procedures
The studies reviewed in this section focus on the

impact of presentation and recording variables on the accu-
rate recognition of facial expressions. Facial expressions
may be recordeélfor viewing by observers in many different
ways: sketches, models, photographs, films, and videotapes.
Moreover, different types of recorded data have been pre-
sented in various studies: areas of the face, full face,
face alone, and face and body together. Finally, special
techniques will be discussed which alter the speed of the
occurrence of the recorded event; for example, slow motion

videotape.

Artificial Modalities

 

The problem of how to capture facial expressions for
viewing by observers is subtle and complex, and many
procedures have been used with varying degrees of success.

Early studies of facial expression (Langfield, 1918; Buzby,

197

. 1.4 u a \ 1
.3 .3 E l: L l
U r e .L 3 .1 t. .. l
5; O .Hu .. a. A. .C L...
.3 5 . L t a: ~ 3 m... .l. «2

AE

9}".
stanc
3f

 

 

31

1924) were forced to rely on artificial recording methods
such as sketches or three dimensional models to present
facial expressions to observers. Boring and Titchner

(1925) improved this procedure by devising a model with
replacable features. More recent studies (Cuceloglu, 1972;
Harrison, 1964; and Royal, 1959) have utilized simplified
sketches of faces to investigate facial expressions. These
studies, however, are all beset with the same flaw; in order
for a study to generalize about the spontaneous expression,
the face must be recorded in a realistic way. Recognition
studies that use artificial presentation procedures may
generalize facial expressions as perceived in that particular
modality, but it is difficult to draw conclusions about

naturally occurring facial expressions from this data.

Still Photographs

 

As the use of still photographs became more common,
recording and presentation procedures improved. Goodenough
and Tinker (1931) presented still photographs of facial
expressions with written descriptions of eliciting circum-
stances to observers. They found that recognition was best
when the photo and the description matched. When the two
modalities conflicted, some observers judged the emotion by
the photograph and others by the written description. The
authors concluded that both facial expression and knowledge

of the situation aid in the recognition of emotion.

fro:

 

Sag en u “we V s r
a\~ ‘L A)“ H...“ 11!...
u a: C 6 AM.»

a“! RU;
1N .
.0 Aw
\. x h

 

32

LoCastro (1972) performed an interesting variation of
Goodenough and Tinker's techniques, presenting the photo-
graphs and written statements along with audio recordings.
He also separated the sound of the voice (tonal content)
from the information (verbal content) provided by the
recordings. For "love" and "anger," verbal and vocal
modalities seemed more crucial to recognition, while for
"sadness" and "joy," the facial channel was more important

than either the vocal or the verbal.

mass

In the preceding studies, only still photographs were
used to present facial expressions to observers. Two
studies (Frijda, 1953; and Dusenbury and Knower, 1938) have
compared still photographs to films in judging of facial
expression. Both studies reported that films of facial
expressions were judged significantly more accurately than
still photographs. These studies present clear evidence
that films are a more effective modality for the presentation
of facial expressions and it is unfortunate that due to the
ease of recording and presenting still photographs, many
studies of facial expressions have used this less effective
modality. It might also be noted that since films are able
to portray facial movement in a more natural way, it is also
more possible to generalize beyond the experimental situation

when employing the film modality.

33

Several studies have investigated what attributes
of film allow it to be accurately judged. Levitt (1964)
presented subjects portraying six emotions in visual, audio,
and audio-visual modes. Contrary to his expectation, they
found that emotional meanings were communicated more
accurately in the visual mode than in the audio. Howell
and Jorgensen (1969) compared pleasant and unpleasant
emotions in the same three modes and included written
transcripts. For pleasant affects, they found that observ-
ers were more accurate when viewing the audio—visual mode
and less accurate in audio-only and transcripts. Trans-
cripts, however, were superior for some unpleasant affects.
Burns and Brier (1973) added the tonal dimension to this
type of study and found that audio-visual with either type
of sound mode was superior to visual-only, and visual-only
was superior to audio-only or transcripts. Thus it
seems clear that the audio-visual modality is superior to
all others in accurately representing emotional behavior.
This superiority seems to stem from the cues available in
the visual content, the verbal content, and the tonal
content. Visual-only is the next best modality, followed

by still photographs and finally transcripts.

Facial Components
The preceding section has dealt only with how facial
expressions were presented to observers: through photos,

through models, or through films. Another aspect of the

Au

.u.
~N u

:0 l
-;y-‘
Olvo‘fi

 

C .
C L
y.

“,1

Hrs-
Va...

1‘

.a L

.Ll—

CL

all

E

H...

\
l
a“:

fin
..
H. ‘

 

34

presentation procedure involves the Eypg of material
presented. Does the full face or only a part of the face
need to be presented? Do body movement and other non-

verbal cues heighten accurate judgment of facial expressions?
And finally what non-verbal cues are necessary and sufficient
to the accurate recognition of emotion?

Duschenne (1862) in an early study of facial compo-
nents, divided facial muscles into groups according to
their importance in facial expressions. Dunlap (1927)
investigated the same phenomenon by substituting various
mouth features in facial expressions, but like Duschenne,
the study was primarily descriptive and no experimental
procedures were employed. Coleman (1949) also studied
differential importance of areas of the face and found a
complex interaction between facial area observed, the type
of affect involved, and the eliciting conditions employed.

An interesting deve10pment in studying the importance
of different facial areas is the Facial Affect Scoring
Technique by Ekman, Friesen, and Tomkins (1971). Dividing
the face into three areas--brow, eyes, and mouth--the authors
presented various combinations of these to observers, and
found that certain areas are crucial to the recognition of
some emotions, while other emotions seem to involve the

entire face. For example, in Unmaskingthe Face (1975),

 

Ekman and Friesen note that the brow is the most crucial

area in the emotion of sadness, but the involvement of the

5““

EUuH

CI LIE

 

T. .C .01 .14
.4.“ ~ u a“ N.»
a& .v' vRIH «Nd
NH ~k~ L ~ .6 .

[us

.
\...' "

Ker

35

mouth intensifies the expression of the affect. On the
other hand, the affect of disgust employs only the lower
half of the face, the upturned lip being all that is

needed to recognize this expression.

Facial and Body Cues

 

The previous studies examined facial cues which are
crucial for accurate recognition of facial expression.
Another series of studies investigates a related issue:
does facial expression interact with other non-verbal cues
to provide a more accurate representation of emotional
meaning. Several studies (Ekman, 1974; Ekman, 1975;

Ekman and Friesen, 1967; Ekman and Friesen, 1969) conclude
that if observers are shown videotapescfi'body posture and
facial expressions separately, the Eype of affect is judged

primarily in the face while the intensipy of the affect is

 

judged from both the head and the body. In a recent study
(Ekman and Friesen, 1974) observers watched face and body
cues while subjects are motivated to conceal their true
affective responses. Untrained observers were able to pick
up the deception from body cues more easily than from facial
cues, but trained observers were able to detect deception
from facial cues alone. The authors conclude: "the informa-
tion is there on the face if the observer knows what to

look for." Thus there is some evidence to support the

contention that observers can pick up general information

S 3.
3.

.l n I l I .‘l " c v C ‘Fl‘4 ANN
. .A E E o}. a v . t . A... 3 ~ .. r m: S t t v3. t r S
3 S O . . .l n: 7 f. a .. e a o .3 e. T. d C a t .. “w. E \m‘
. v .. #1 "A I a). on L .1. a... v. r. v... ..-.. an. » . L... O S E h. v: r .5
. 8 RD 6 he (1 a It .il .2 L .a C. e a . .. a . s . .1 l. 0 8. so. It

«a

 

 

36

about affective states from bodily cues, but trained observ-

ers can glean this information from facial cues alone.

Slow Motion Techniques

 

While many non-verbal variables have been studied
extensively, only a very few studies have investigated the
brief duration of some facial expressions. Landis and Hunt
(1937) mention that while the magnification of space is
an accepted scientific procedure, magnification of time
(through slow motion playback techniques) has been studied
little. In their investigation of the startle pattern,
Landis and Hunt (1939) found that the small facial muscles
are the fastest to respond to stimuli, while the larger
groups of muscles responded at a much slower rate. For
example, the eyeblink of a startle response occurs in an
average of 4/1000 of a second after the stimulus while the
movement of the head does not occur until 8/1000 afterward,
and the full body movement does not occur until much later.
Thus the extreme speed of spontaneous facial responses points
to the capacity of the face to transmit information so quickly
as to be undetectable except with high speed recording
techniques.

The only research which has been performed specifically
on the slow motion filming and recognition of facial
expression is Haggard and Issacs "Micromomentary Facial Ex-
pressions as Indicators of Ego Mechanisms in Psychotherapy"

(Haggard and Issacs, 1966). The authors define

the e

5:19;):

q
\r-L. [V t

3"!

 

 

37

micromomentary expressions (MME's) as "facial expressions that
that are so short-lived that they seem to be quicker than

the eye." In the study, an average of two and one half

times more facial expressions were seen when a stimulus

film was viewed in slow motion than when viewed at normal
speed. Haggard and Issacs felt that MME's did not occur
continuously, but were related to areas where the individual
experienced emotional conflict. They seem to serve as "trial
balloons" by which a patient displays an affect subliminally
and finds out whether he can experience a particular
emotional state without fear of rejection from the therapist
(personal communication, Earnest A Haggard, 1975).

Although Haggard and Issac's study is interesting,
there are several procedural flaws which mar the results.
First, only two subjects were employed, and the authors
never state how many judges or raters participated. No
definition of “facial expression" is made, and no schema
for attaching facial expressions with emotional meaning is
forewarded. Moreover, inter-judge reliability across
segments was extremely variable, and finally, the eliciting
situation--a therapy hour--was too complex to be useful.

In spite of these flaws, however, Haggard and Issacs present
an interesting proposition: that significantly more facial

expressions are observed at slow motion than at normal speed,
and that the types of expressions observed in slow motion are

different from those observed at normal speed.

F.

pIC

 

 

38

Conclusion

To conclude this section of recording and presentation
procedures it seems first that films or videotapes are
superior modalities for observers to accurately label
emotional states. The accuracy of the visual-only modality
is similar to audio-visual but lacks the tonal and content
cues in the voice. Types of emotions are most accurately
recognized from facial expressions, while intensity of
emotional arousal is most readily discerned from other
non-verbal cues. Finally, the facial musculature seems
capable of responding at great speed and viewing films in
slow motion seems to increase the number of expressions

perceived.

Supject and Rater Variables

 

Subject variables are defined as aspects of the person
portraying the emotion that influence the quality of the
facial expression. Rater variables include aspects of
raters (those who observe the subject portraying the emotion)
which might influence the accuracy of their judgment. This
section examines differences in both subjects and raters
which influence the quality of their performance. Included
in this section are developmental differences in posing and
judging, differences in expression of blind and sighted
children, personality differences in posing and judging,
race and sex differences in posing and judging, cross-

cultural recognition of facial expression, and finally, the

m:

WJ

 

“U.

‘H‘

 

39

effects of rater training on the judgment of facial

expressions.

Development of Facial Expressions

 

Some studies on the development of facial expressions
have investigated the extent to which infants respond to
adult facial behavior. Two studies (Wilcox, 1969; Wilcox
and Clayton, 1968) found that infants responded to facial
behavior no more than any other complex and interesting
simulus. However, in a recent study, Goren, Sarty and Wu
(1974) have disputed these conclusions and have obtained
evidence that infants fixate on stimulus patterns resembling
facial features significantly more often than other patterns
of equal complexity. Moreover, two studies (Kreutzer and
Charlesworth, 1973; Bfihler and Hetzer, 1928) found that at
the age of 5 to 8 months infants begin to discriminate
between positive and negative facial expressions. Thus there
is evidence that even for young infants, the human face is
an important source of stimulation in the human environment.

In investigating preschoolers' facial expressions,
Goodenough (1972) asked university students to judge
photographs of children expressing Spontaneous emotion,
and attained accuracy at six times chance levels. A similar
study was performed by Hamilton (1972) with positive results,
although the accuracy of the judgments did not approach

that of the previous study.

 

 

ing botl
found tl

sent as

40

Kwint (1934), Gates (1925) and Hoshino (1969) each
studied the development of facial expressions in older
children (ages 3-14). These studies emerged with essentially
the same finding: that children's ability to judge facial
expressions improved markedly with age. Furthermore, study—
ing both posing and judging ability, Odom and Lemond (1972)
found that posing ability also shows a trend toward improve-
ment as the child matures, though at a slower rate than
accuracy of judgment. It seems clear from these studies that
the ability to pose and judge facial expressions improves as

the child becomes more mature.

Facial Expressions in Blind Children

In order to try to differentiate the learned and
innate components of facial expressions, a series of
studies have been performed on the facial expressions of
blind children. Freedman (1964) noted that blind infants
"exhibited reflex-like action" until the age of five or six
months, with expressions becoming more prolonged and indi-
vidualized thereafter. Both Thompson (1941) and Fultcher
(1943) found that blind and seeing children showed similar
expressions of sadness, joy, and anger while very young.
However, in contrast to sighted children, the frequency of
occurrence of these expressions decreased in blind children
as they matured. Thus although the mechanisms for facial
expression in blind and sighted children appear to be

similar in infancy, some amount of visual stimulation seems

 

 

to be ne
authors
learned

facial e

totiona
to influ
{1974) i
raters b
exPressi
thEir 0w
there Wa
35 the r
the e'fiot
evidEnCE
the rate

a $0th

41

to be necessary to maintain these mechanisms. Many
authors conclude from these studies that there is both a
learned and an innate component in the development of

facial expressions.

Facial Expression and Personality

 

A number of studies have attempted to alter the
emotional state of those judging facial expression in order
to influence the accuracy of their ratings. Schiffenbauer
(1974) induced various degrees of emotional arousal in
raters before they judged still photographs of facial
expressions. Raters showed a marked tendency to attribute
their own affective state to the photograph. Moreover,
there was a high positive correlation between the degree
of the rater's emotional arousal and the intensity of
the emotional state attributed to the photograph. This
evidence suggests that the immediate emotional state of
the rater has some effect on the type of emotion seen in
a photograph and the intensity of the state perceived.

Several researchers have tried to correlate per-
sonality variables with the ability to pose emotions. Drag
and Shaw (1967) looked for differences in posing ability
correlated with both situational anxiety and the personality
of the power. No correlation could be found between amounts

of anxiety experienced or personality traits with posing

 

 

42

ability. However it was discovered that some affects were
easier for all subjects to pose than others. This finding
coincides with the earlier research by Coleman (1949) who
concluded that there was a complex interaction between
personality of the poser, facial region observed, and type
of emotion posed in determining the accuracy of recognition
of any given posed expression.

The production and recognition of facial expression
as related to diagnostic categories of mental illness is
a frequent topic in the literature. Dougherty, Bartlett,
and Izard (1969) found that normals were significantly
more accurate than schizophrenics in identifying facial
expressions. Shannon (1971), however, found no overall
differences between normals, schizophrenics and depressives,

but did discover differences in recognition of specific

 

emotional states. Schizophrenics and depressives were
less accurate than normals in identifying anger, and
depressives were less accurate in identifying fear than
either of the other two groups. Using more general measures
of pathology, Chambers (1961) found a positive relationship
between the ability to accurately identify facial expres-
sions and college adjustment during the freshman year.
Several studies have attempted to correlate facial
expression recognition with scales on established personality
Ineasures. Zlatchin (1955) presented facial photographs

tachistoscopically in diminishing exposure times and

 

 

attempt
with sc

tory an

no sign
Carlson
ability
tive Pe

the ir r

Persona

43

attempted to relate differences in recognition thresholds
with scores on the Minnesota Multiphasic Personality Inven-
tory and the Gough Adjective Checklist. Because of the low
reliability of the recognition of the photographs, however,
no significant correlations were found. In a similar study,
Carlson and Levy (1973) attempted to relate recognition
ability to Meyers Briggs Types. They found that the Intui-
tive Perceiving types were significantly more accurate in
their recognition than the Sensing Judging Types.

Most of the preceding studies of facial expression and
personality have attempted to correlate recognition accuracy
with either diagnostic labels or test scores. Varying
this procedure, Cohen and Rau (1972) showed photographs of
facial expressions to groups of subjects previously identi-
fied as normal and depressed. The subjects' task was to
pick the photograph that best described their own emotional
state. The photographs which were chosen discriminated the
depressed group from the normal group and also distinguished
levels of severity of depression within the depressed group.
Thus through the use of a different procedure, the authors
have confirmed a relationship between facial expression re-
cognition and personality adjustment.

In another creative approach to the problem of facial
expressions, Tomkins and McCarter (1964) showed still
photographs of eight affects to a group of subjects, develop—

ing what they called "Affect Sensitivity Contours" for each

 

 

sabj
attr

e1h0a

44

subject. Their assumption here was that rather than
attributing errors to unknown variables, each mistake
embodied a reflection of the personality configuration.
Thus each subject's pattern of errors in identification was
reflected in his "Affect Sensitivity Contour." Whether this
measure correlates with any of the more established personal-
ity tests has not been determined.

As studies such as these are compounded, it would
seem reasonable to conclude that most researchers have
found some relationship between facial expression recogni-
tion and personality. However, because of the procedural
variations in these studies, it is difficult to emerge with
a specific overall conclusion. Again the non-uniform
procedures have contributed to the non-additivity of facial

research in this area.

Race and Sex Differences

 

There is consistent evidence in the literature that
females are slightly more accurate than males in the posing
of facial expressions (Buck, Savin, Miller, and Caul, 1969,
1972, 1974; Black, 1969; and Gitter, 1972). Some suggestion
is made (Buck, Miller, and Caul, 1974) that this difference
is due to a greater amount of activity in female faces.
According to these authors, because female faces are more
mobile and expressions more pronounced, it is easier to

judge emotions from them.

45

Findings are mixed as to whether females are also

more accurate in the recognition of facial expressions.

Some studies (Buck, Miller, and Caul, 1974; Black, 1969

and Gitter, 1972) have found no differences between males
and females in the accuracy of perceiving. Yet Kozel (1969)
showed that although there was no overall superiority in
the recognition scores of women, they were significantly
more accurate than males in four out of seven affects
judged.

In investigating the effects of race on portrayal and
recognition of emotion, there have also been mixed results.
Two studies (Gitter, 1972; and Black, 1969) found that white
posers were judged more accurately than blacks. In a very
similar study, however, Kozel (1972) found no such effect.
In terms of the recognition of facial expressions, the few
studies which have been performed have found blacks signi-
ficantly more accurate than whites in their judgments (Black,

1969 and Gitter, 1972).

Cross Cultural Studies

 

The controversy over innate and learned components
of facial expressions during the past ten years has spurred
a very important series of cross-cultural studies. Theorists
such as Birdwhistle (1963) have proclaimed that there are no
innate facial behaviors and that expressions are acquired
only through social learning. Studies will be reviewed

Ihere that put this assumption to an experimental test.

”as?
uv *“
1

v

or n

Y;

n».
D.»

L
a.ta

are
“3-:

 

18C

3833‘

DhOtC
2.1
Litu

S

.‘I‘A
W’IQ‘J
C .(

 

46

Vinake (1949, 1955) studied recognition of emotion in
both Caucasian and Oriental faces, and concluded that
no matter what nationality was judging (Japanese, Chinese
or American) there were no differences in the accuracy of
identification. Dickey and Knower (1941) compared Mexican
and American children in identifying photographs of facial
expressions. Higher than chance levels of accuracy were
attained for both groups, but the Mexican group was slightly
more accurate than the American group in the task. Ekman
(1971) showed stress films and neutral films to observers in
four groups. Americans living in America, Americans living
in Japan, Japanese living in America and Japanese living in
Japan. He videotaped the subjects facial responses to these
films and showed these videotapes to observers. Accurate
discrimination between stress and non-stress conditions were
made by observers for subjects in all groups, showing that
regardless of the nationality of the subject, the face was
judged in the same manner.

Ekman (1972) and Izard (1971) showed a series of
photographs to observers in a total of thirteen different
cultures (see Table 2.1) and found high degrees of correct
identification in each of the cultures. One could conclude
from this data that observers from all of the cultures
studied perceived facial expressions as having the same
qualities of emotional meanings. Some critics argued,

however, that there is a significant transfer of knowledge

\
lump“
~
.-
d
N
fi
H
PA
a.
N
5
I.
w
-H.
MHV~
. n)"
4~
w
I.“
y‘
H N
N
N
....v
~
A
4N
C
.... ~

on

a.
.—
1
.fl
.~
.«
a
3
m
.ud.
J
H
d
.-
_
u
n
.n
d
u,
.w
m .x
-
=
_..m
U;
F
—
J
u
.3
a
r
,U
.-
.. nu.
.—
“a
n
.7
d
a
. a
W
I)...
H
P
LA
.IV
d
1w
5
I
~
A
.~
m.
.E
h
H
:dJ
U
n
m4
~.
.U
p
.. on
my
I
._ .U
~
,~
n. U
Paxfl
)5
9i.
. L

 

 

47

 

 

Amhmac caegm “mousom
mum: nonmfiq
woman.
o o o o o m m m m N m N m m m m cane “mayo
$30035
mo Honesz
so mad mas ow mm om mm om mm he as mms mm mm «ems was mum>ummno
no nonasz
me mm om mm an mm mm am on on an so an me am He saw
mm 3. mm mm .8 em mm em 2. we mm mm em .8 S 8 568.8
\umdmmfio
as me we mm mm mm Hm om mm as mm mm as mm mm mm omega
Hm mm mm mm em an as om mm am om mm om om me am mmflumusm
mm we me up He mm we we no mm as am no we mm as name
em as om am em so mm mm so em we mm mm em mm em madam
m
an...“ HHN wmwc 0m®G £8 3mg ENE SmHH .m.D .m.D G00.

.m.: Iguana mango Imam undue neous ufiuua.xoouo mmflsm annoys

team uuoo new

 

~58th a £553

UHMNH

Hogans a engage

 

 

macao.gumm

55.3 935nm who 53 83383 E 8.39m 88m umfi omficmonom 68330 38.33 dm 3nt

 

 

among l
countel
and shc
in New
laden s

ooserve

F.“ 1 a

A

“DLE A

48

among highly technological and civilized cultures. To
counter this argument, Ekman took his study one step further
and showed his series of photographs tO'adults and children
in New Guinea, a "visually isolated" culture. Using emotion-
laden stories rather than labels, Ekman found that these
observers could accurately identify the expressions studied

(see Table 2.2).

Table 2J2 Identification of Emotion Photographs by Visually
Isolated Preliterate Observers

 

 

Percent Choice of the Emotion
Expected in Terms of Agreement
With the Judgments of Western

Emotion Category
Described in the

 

 

Story Literature Culture Observers
Adults 1 Children

Happiness 92 92
Sadness 79 81
Anger 84 90
Surprise 68 98
Fear

from Anger/Dusgust or Sadness 80 93

from Surprise 43 __
Number of Observers 189 130

 

Source: Ekman & Friesen (1972)

 

Tl
least si
tai'ed-
disgust.
relatior
ccurs j

than lea

49

This research presents conclusive evidence that at
least six facial displays are recognizable in all cultures
studied-~happy, sad, fear, anger, surprise, and contempt/
disgust. This pan-cultural accuracy seems to indicate a
relationship between facial expressions and emotions which
occurs in all peoples and which seems to be an innate rather

than learned component of the human repitiore.

Effects of Rater Trainipg

 

In seeking to ascertain the extent to which learning
can improve the recognition of facial expressions, two
studies have investigated the effect of training on the
accuracy of raters. An early study (Jenness, 1932) con-
cluded that training significantly improved rating scores,
but also found a negative correlation between initial
accuracy and subsequent amount of improvement across
training sessions. In other words, if the rater was accurate
to begin with, then further training was not very helpful:
training was most beneficial to raters who had initial diffi-
culty in assigning the correct labels to facial expressions.

In a similar study, Guilford (1929) trained judges
every other day over a period of ten days, and found that
this type of procedure improved overall accuracy greatly.
Moreover, the rating roups became more uniform as the train-
ing progressed, with poor initial raters showing much
greater improvement over time. It is clear that Guilford's

results correspond with Jenness' findings showing that

 

 

50

training significantly improves the accuracy of judgments,
especially for raters whose beginning scores are particu-

larly poor.

Conclusion

 

In reviewing the findings on subject and rater
variables, several key points emerge. First it would
seem that certain facial expressions are innate human
behaviors and occur without prior learning. Evidence
supporting this argument is found in studies of infants,
blind children, and cross-cultural research. Secondly,
although there seem to be differences in the area of sex
and race for posing and judging emotions, all groups studied
were able to recognize facial expressions at above chance
accuracy, thus the similarity of these diverse groups is far
more striking than are the differences between them.
Finally, it is possible for observers to recognize these
facial expressions without prior learning; observers who are
trained in the recognition of facial expressions are signi-
ficantly better at the identification task. It is reasonable
to conclude therefore that even though innate facial expres-
sions of emotion can be identified without prior learning,
some training tends to sharpen the identification of these

expressions.

 

Thi

neasuring

expressi
such as
of desC'

n
QiC.

enplgyj

electr‘

dEflOns
on fre
1931;

0‘3 lal

51

Dependent Measures

 

This section explores a number of methodologies for
measuring or describing facial expressions of emotion.
One popular device for the measurement of facial expressions
is the assignment of a written or verbal label to each
expression. These labels may be in the form of categories
such as "happiness, fear, love, etc." or dimensions composed W"
of descriptors such as "active-passive, pleasant-unpleasant,

etc." Other techniques will be examined which, rather than

 

employing labels, measure facial activity directly through
electrical activity of the facial muscles.

The importance of labeling methodologies is indirectly
demonstrated by the poor success of studies which have relied
on free labeling of facial expressions by subjects (Kramer,
1931; Landis, 1924). If subjects are allowed to apply their
own labels to facial expressions, the number of possible
choices is so great it would be exceedingly rare for
subjects to agree on one label. Davitz (1969) in The

Language of Emotion, found hundreds of words which could

 

describe an emotional state, with each of these having its
own shade of meaning. To escape from the problem of free
labeling, most researchers have relied either on the dimen-
sion approach or the category approach to describe the

‘ expressions.

 

 

HOUjWO
SCEOIS
'conte

EXCIES

52

The Dimension Approach

 

In a series of studies, several investigators employed
the dimension approach to study facial expressions.
Schlosberg (1941) investigated a scale for the description
of facial expressions first proposed by Woodworth in
1938. The affects in Woodworth and Schlosberg's model were:
happiness, surprise, fear, anger, disgust, and contempt. F
These affects were arranged along a continuous dimension,
each blending into the next along the scale. Although ' [
Woodworth first proposed that the scale was linear;

Scholsberg's results showed it to be circular, with the
"contempt" expression blending with the "happiness"
expression to form a complete circle. Schlosberg also
showed that the facial expressions in the study could be
described on scales of attention-rejection and pleasantness-
unpleasantness (Schlosberg, 1941).

Other experimenters investigated Woodworth's model
(Secord, Dukes, and Bevan, 1954, 1956; Engen, Levy, and
Schlosberg, 1958) and essentially replicated the earlier
findings: that it is possible to identify facial expres-
sions using Woodworth's scale, and that expressions can also
be distinguished on dimensions of attention-rejection, and
pleasantness-unpleasantness. Moreover, some evidence was
compiled as to the cross-cultural reliability of these

findings (Triandis and Lambert, 1958).

tiate

than <

tensi<

Phili}

third

<~ FLA
6 La.» ‘1).
r; r l.
«\u. a: .l
v... 3.. .» n
E , .. P

Sernat

1 . CC
C. U
at e
.au r
t. a

2‘

Q»
in .
shk

Q» fiU Q» aw)
». a: as so
I AN.» Q\ r
a: a \I.. u. e.
at -l a

53

Many other investigators have attempted to differen-
tiate facial expressions by asking subjects to place
them on several dimensions. Most have included a pleasant-
unpleasant dimension and an activity dimension ranging from
tension to sleep (Abelson and Sermat, 1962; Frijda and
Philipszoon, 1963; Bartlett, 1971; and Gladstones, 1962).
Each study has discovered a third dimension; however, this R‘—
third factor has been labeled differently in each study: 5

expressionless-mobile (Gladstones, 1962), self assurance . +‘

 

(Bartlett, 1971), intensity vs. control (Frijda and
Philipszoon, 1963) and attention-rejection (Abelson and
Sermat, 1962).

From this data it has been established that both
the pleasant-unpleasant dimension and the activity dimension
are useful in making discriminations between facial expres-
sions. The third dimension also seems important, but the

findings are ambiguous as to its composition.

The Category Approach

 

Another series of studies of facial expressions have
employedmdiscrete emotional labels to describe facial
expressions. The basic difference between the dimension
approach and this approach is that while the former
assumes a relationship between different emotions, the
category approach assumes that each affect is distinct and
independent of all others. As an example of the category

approach, Tomkins and McCarter (1964) showed 69 still

 

 

eight <
surpri:
conten;
lcCarte

correle

54

photographs of facial expressions to 24 subjects and

asked them to place each photo in one of the following

eight categories: interest-excitement, enjoyment-joy,
surprise-startle, distress-anguish, anger-rage, fear-terror,
contempt-disgust, and shame-humiliation (Tomkins and
McCarter, 1966). The investigators obtained an average
correlation of +.858 between the affect posed and the affect
judged, and they concluded that it was therefore possible

to accurately judge the meanings of facial expressions using
this particular set of affective labels.

Ekman (1972) has compared results from five major
studies of affective categories, including Tomkins and
McCarter, and has emerged with a synthesis which formed the
basis for the categories included in his very successful
cross-cultural work (see Table 2.3). He argued that all
major studies of facial expressions included labels repre-
senting each of the following categories: happiness,
surprise, fear, sadness, anger, disgust-contempt, and

interest. As mentioned before, the success of Ekman's

cross cultural work points to the utility of this categoriza-

tion scheme. Moreover, Izard's cross cultural work (1971)
employed Tomkins' and McCarter's set of eight emotional
labels, thereby corresponding to seven of Ekman's categories
'with the addition of shame. In conclusion it seems that
there have emerged at least six distinct categories of

facial expressions: happiness, surprise, fear, sadness,

find-rm u viii-IV"

all

2.2.. a . . s. w J 7.1.) C H

 

3..) H ..~

\aka

~J ... e we. at: e a. eta . ~

1.....8 w uh A-v~.m..~ ,u FfiJ

:n» m d .32}:

M.) um.‘

.4. ~ a a 2.x.

 

55

tha .cmfixm "mousom

 

 

 

 

 

 

 

 

coHumcHEuoumo comm coflu
- mmoccuo mom Iooafiuo o
Momcc nooofi wmmmnsum Homo” nomad .mew¢a
oo:m>oc:¢
nomad coaaom
nflmomoo
BOHHOm ousom moauw
smflomcd 3ouwom mcaummmsm
mmoaomm pom mmoooom mammuo mmouumflo mmoco>amco .
Eooonom . m
Hemommo
Houume
Home “mom Hmmmom uwmwoe Room Room
m m coflmconoummm
vooEuMWMHsmm mauumum ucoecmwooumm
omwumuom omfluouom ucoEonam mmflumuom wmwwmwME¢ omwumuom
mmfiumnsm . w
HounoomH oofluuoz
omao
mmocfloomm homo: woo woo moo mmocflmmmm
unmeaomom nuns:
ucoswomsm mmommoo m>om
mnemooao uoflso
moomooameoo
vmma
nmmma moma Noma mmma
ommomoum confine ooommo Houuouoz xenousHm nuuo3o003

a mcflxeoe

 

 

muoummfiumo>cH o>wm ho oomomoum

moeuomouou noduosm MVN canoe

 

5...: C ..~ ..~ :Chu

M -N 3.5:...»38

 

56

Amrmav sosxm "mousom

.ooumwa coon o>on moomumm

msHoEHum noon mo mucoEmosn mo mfimxamcm mop cw oomuoEo nofln3 mowuomouoo Hades

.ooumfia coon o>o£ momhaocm muoo

mo momma ooucu m.ooommo mo 03H umooa um ca venom muoz gowns mofluomwuoo Ado;

 

 

 

 

 

 

 

 

 

 

Emwowpdoxm
ousoomoH
acouH
moaum
umuuflm muowxcm
nonhuman coHuoflaweom cowumuomuoocH
ano huflm osmnm moomuoooofi
coaummaoauc<
. umououcH uooamuaoxm . . .
umonoucH coflucouud . mocmuoomxm
mucouoomxm umououcH mmoco>flucouu¢
mafinuooq
cuoom moanumoq umeucoo
new so mEo co .
\wmsmwaoo umommflo poewucoo wmsmwaoo umommflo
. umsomflo . mmocoeommuwa umommwo
oocowoocd
voma
omwma. meme Nme mmma
emmodoum «.meflnum accommo umuumooz xenousam nunogeooz

w mcwxeoe

 

 

82328 M . N 833.

 

anger, an
two categ

in the li

Dir
the only
expressi
it is pc
05 iacie
Schwart;
that me‘
tion of

ObVious

“invent

57

anger, and contempt-disgust. The validity of the remaining
two categories, interest and shame, also has some support

in the literature.

Electromypgraphy

 

Dimensions and category labels have in the past been
the only existing method to observe and measure facial
expressions. With the advent of electromyography, however,
it is possible to measure directly the electrical activity
of facial muscles as the expressions themselves occur.
Schwartz, perhaps the pioneer in electromyography, believes
that methodologies emphasizing observation and categoriza-
tion of facial expressions lack sensitivity to many less
obvious facial changes. The Electromyograph (EMG) measures
these small facial movements by recording changes in the
electrical potential of certain facial muscles. Schwartz
found different electrical potentials in the frontalis and
corrugator muscles depending on the emotional content of
the subject's imagery (Schwartz, Fair, Greenberg, and
Friedman, 1973). Moreover, this effect was present even when
no visible facial expression was evident. Schwartz thus
concluded that the EMG was sensitive to facial changes which
were invisible to observers (Fair, Schwartz, Greenberg,
Klerman, and Gardner, 1974). Although the use of the EMG
is still in a preliminary stage, it seems possible to cir-

cumvent many of the problems inherent in both the dimension

58

and category methods by directly measuring facial electri-

cal activity through this technique.

Conclusion

 

In this chapter, several procedural aspects of facial
expression studies have been reviewed. Certain conclusions
can be drawn from each of these sections and these have been
considered in planning the present experiment.

It has been demonstrated that since expressions of

facial emotion are elicited by a wide range of stimuli, it

 

is necessary to choose an eliciting condition with care so
that the intended affect is indeed elicited. Asking subjects
to pose facial expressions avoids these considerations, yet
there is some evidence to suggest that posed expressions
differ systematically from those elicited spontaneously.

In this study, spontaneous expressions are generated by
asking the subjects to produce their own eliciting circum-
stances through the use of imagery. In this way it is
possible to study spontaneous facial expressions without
encountering the problems inherent when the experimenter
sets the eliciting circumstances.

The next problem to be confronted is how to present
these facial expressions to raters. Although the vast ma-
jority of studies have employed still photographs, it has
been shown here that films or videotapes are significantly
superior to photographs in the accuracy of identification.

There is also some suggestion that the speed of facial

expre

«nu

51
5V

.

EX CIE

..
h

oanT

N I
e
u."
0
(All.

grout

 

'9 he
aCCU:

U

LaPes

respor

 

59

expressions when viewed on videotape influences the accuracy
of ratings. In this study, the same videotape of facial
expressions is shown to raters both in slow motion and in
normal speed._

The third aspect of these facial expression studies
involves the composition of the subject and rater groups.
Differences in observers' ability to identify has been
found between races, between sexes, between age groups,
between personality types and between cultures. Thus far,
however, the similarities in the accuracy of these diverse
groups are far more striking than the differences that exist
between them. This type of data points strongly to a
capacity to recognize and identify facial expressions
possessed by all human beings which cuts across cultural,
national, sexual, and age-related boundaries. On the other
hand, it has been shown fairly conclusively that although
human beings are capable of accurately identifying some
facial expressions, training in this area has also proved to
be helpful, particularly for those who are initially less
accurate than average. In this study, all raters have
been trained.and have attained a level of at least 80
percent accuracy for still photographs before viewing video-
tapes.

The methodology for describing and recording raters'
responses is an important part of any study of facial

expressions. Many studies have shown that either the use

 

 

 

of dir

fullv
J

60

of dimensions or categories enable observers to success-
fully discriminate between facial expressions. A synthesis
of the major studies has emerged with the categories of
happiness, surprise, sadness, anger, fear, and contempt-
disgust. The categories of interest and shame are also
included very successfully in some research. In this study,
Tomkins schema of eight affective categories has been
employed. The affects studied are: interest-excitement,
el'ljoyment-joy, surprise-startle, distress-anguish, anger-rage,
contempt-disgust, fear-terror, and shame-humiliation. This
group of categories includes all those studied successfully

in the cross cultural research.

 

CHAPTER III
DESIGN OF THE STUDY

This chapter is divided into the following twelve
sections: the subject sample, the primary interview, the
videotape apparatus, the recall interview, judging of inter-
view tapes, editing of the interview tapes, the rater
sample, training of raters, the post hoc rater questionnaire,
the design of the study, the hypotheses, and finally, the
analysis of the data. A flow chart of the procedure of the

study is presented in Table 3.1.

The Subject Sample

 

Eighteen volunteer subjects were obtained through sign-
up sheets (see Appendix A) from a large class, "The Indivi-
dual and the School" (ED-200), in the College of Education at
Michigan State University. Subjects were informed before
they signed up that they would be participating in a study
which involved making videotapes of people discussing their
feelings and emotions. They were informed of the approxi-
Inate time commitment involved.and told that they would
participate in a videotaped interview followed by a recall
session where the contents of the interview would be

reviewe d.

61

 

62

 

 

ouflmc
Icowumooo
Houmm
com “mom

 

 

 

aosum

 

on» m0
cmflmoo

 

 

 

meanaom
mama

mudooooum mo uuonu 30am

 

 

 

mnemooo
woos
3ow>uouoH

 

 

3mfl>uoucH
Hamoom

 

 

.H.m manna

 

 

3oe>uoucH l

muofiaum

 

 

 

 

 

 

 

 

 

mcficwoue
umumm

 

 

 

onEwm
Houmm

 

 

 

 

 

oomuomow>

 

monoummmd

 

onEmm
noonnsm

 

 

 

 

so:

.u‘u

lute

63

Ages of subjects ranged from nineteen to forty-seven
with the average age being twenty-seven. The average level
of education completed is sophomore in college, and 59 per-
cent of the subjects were female with 41 percent of the
subjects being male. Fourteen of the subjects were white

and three were black (see Appendix B).

The Primary Interview

 

When the subject arrived for participation in the
study, the experimenter again informed him of the video-
taping procedure and asked him to fill out a release form
and a short series of informational questions (see Appendices
C and D). Although the subject was going to be videotaped,
the experimenter explained that the verbal content of the
interview was not being recorded. Thus, everything said
in the interview remained strictly confidential and was not
considered material for the study. Moreover, the identity
of the subject was kept completely confidential through the
use of a number system. Although subjects were aware that
their non-verbal behavior was being recorded, at no time
did the experimenters mention the term "facial expression"
or make any reference to the focus of the study. If subjects
asked the purpose of the study, they were asked to postpone
their questions until after the primary interview and recall
session were completed. The primary interviewer in both
the structured and unstructured interview was a male Ph.D.

candidate in the College of Education at Michigan State

 

Unive:
in an
atnos

was I

the s

COEdI

the

She

64

University. He was instructed to respond to the subjects
in an interested, accepting manner in order to create an
atmosphere of trust during the interview. The interviewer
was naive as to the hypotheses of the study.

The first 9 subjects scheduled were interviewed in
the structured condition and the last 8 in the unstructured

condition.

The Structured Interview

In the structured interview condition, the interviewer
first read the instructions to the subject (see Appendix
E). He told the subject that he or she would be shown a
set of photographs of facial expressions, and that each of
these expressions represented a particular type of emotion.
After a photo was shown to the subject, the interviewer
verbally identified the emotional category represented in
the photo. He then interviewed the subject on how he or
she experienced that particular emotion. Questions asked
for each category were: how familiar the subject was with
that emotion, how he identified it, where he felt it in
his body, and finally what situations tended to trigger it.
‘When the subject was finished talking about one emotion, the
interviewer presented another photograph and another affec-

‘tive Category was discussed with the same topic areas

 

1The photographs used were obtained courtesy of
Carroll Izard, Vanderbilt University.

 

 

[EVE

"

inte
aceE
the

Vene

44‘

65

reviewed. The photographs portrayed expressions of each
of these eight affective categories: interest-excitement,
enjoyment-joy, surprise-startle, distress-anguish, anger-
rage, contempt—disgust, fear-terror, and shame-humiliation.
It was these eight categories that the subject was asked

to discuss or explore in terms of his own experience.

The Unstructured Interview

 

In the unstructured interview, the basic purpose was
the same as the structured condition: the subject was asked
to discuss his experience with several different affective
states. In this condition, however, the subject was asked
to make his own list of emotions anéidiscuss them (see
Appendix F). Subjects were asked how familiar they were with
the emotions they chose, how they identified them, where they
felt them in their bodies, and what situations triggered
the particular emotions. The difference here was that the
interviewer did not provide photographs or the concurrent
affective categories for discussion by the subject. Rather
the subject was expected to generate these without inter-

vention by the interviewer.

Videotape Apparatus

 

While the subject was being interviewed in either the
unstructured or structured condition, videotaping was taking
place behind a one-way mirror. Two Sony video-cameras were

employed; one with a 25-22.5-90mm zoom lens was trained on

 

 

th

IE

(I)
n a

the

tar
- F

66

the subject's head and neck region. The other camera with

a 16 mm normal lens was trained on a digital timer with a

read-out in seconds. Both cameras were fed into a SONY

SPECIAL EFFECTS GENERATOR (SEGl) which showed the subject's

face on the majority of the screen while the read-out from

the digital clock was recorded in the upper right-hand

corner. The recording of the digital timer allowed great y-
accuracy in locating the segments of the interview tape for i
the editing and judging procedures to follow. The video- L

tape was recorded on an AMPEX 3600 Video-Tape Recorder.

Recall Interview

 

After the primary interview was concluded, the sub-
ject was introduced to the recall interviewer and the tape
was rewound. The recall interviewer was also a male Ph.D.
candidate in the College of Education at Michigan State
University, and was naive as to the hypotheses of the study.

The recall interviewer then read another set of
instructions to the subject (see Appendix G). The subject
was told that they would be viewing together a videotape of
the previous interview. The subject's task was to view the
tape and to identify places where he saw himself as experi-
encing an emotion. All subjects during the recall interview
were asked to use the same affective categories as were used
in the structured interview condition. If a subject was
unable to fit his emotional label into one of the eight cate-

gories, the response was not used in the study. .

67

The recall interviewer was instructed to view the
videotape along with the subject and to record the points
in time on the tape where the subject identified an emo-
tional response. He thus recorded the type of response in
one of the eight affective categories and pinpointed its
location on the tape by recording the corresponding digital
clock read out (see Appendix H).

Before leaving, the subject was informed of the
purposes of the study and was given the opportunity to

discuss any reactions he had to the procedure.

Judging Interview Tapes

 

While previous sections of this chapter examined how
the facial expressions were elicited and recorded through
the interview procedure, this section deals with how the
most valid segments were chosen from the raw data of the
interview tapes. A rating panel composed of four expert
judges was chosen on the basis of the judges' familiarity
with Izard's affective states (Izard, 1971), and with re-
search and literature on facial expressions. The interview
tapes were played to the judges, and the judges were asked to
identify the emotion on the subject's face at those points
on the tape chosen by the subjects themselves during the
recall interview. An average of 4 seconds was bracketed
around the recorded time to allow for errors by the

subjects or by the recall interviewer.

 

 

(W

'47

I?

68

An example might be helpful to make this part of the
procedure clear. For instance the subject identified a
point on the interview tape where he thought he looked happy.
The recall interviewer recorded this label and, through the
digital readout in the corner of the screen, also recorded
the point on the tape at which it occurs. Two seconds of
viewing time were added to the segment preceding the time !
identified by the subject. Another two seconds were added
following the identified time. Thus a total of five seconds
of videotape were viewed, with the exact second identified
by the subject occurring in the middle of the segment. If
the readout time then was recorded at 5 minutes and 30 sec-
onds, the judges would have viewed the tape segment from 5
minutes 28 seconds to 5 minutes 33 seconds, and would have
recorded which of the 8 affective categories was in their
opinion represented on the segment.

For purposes of this research, the cirterion for valid-
ity was the agreement of 75 percent of the judges on the
affective category of a 5 second segment. Thus if 75 percent
of the judges agreed on the emotion depicted in a five sec-
ond segment, this segment was acceptable for use in the
stimulus tapes. Two affective categories (fear-terror and
shame-humiliation) were eliminated from the study due to an
inadequate number of acceptable segments portraying these two
emotions. The materials included in the stimulus tapes then
consisted of segments which were identified in the following

categories: interest-excitement, enjoyment-joy, surprise-

69

startle, (fistress-anguish, anger-rage and contempt-dis-
gust. The rationale for the elimination of fear-terror and

shame-humiliation will be further discussed in the following

chapter.

Tape Editing

 

One hundred and one segments in the remaining sex cate-
gories were randomly edited onto a master tape. A random
numbers table (Glass and Stanley, 1971) was employed for
this part of the procedure. An AMPEX 3650 Video Editor
was employed for all editing. Segments were spaced so that
there was a lO-second pause between two segments to allow
raters time to record their responses. An AMPEX 5680
Video Editor was employed to make a duplicate tape with the
segments slowed to an average of 31 seconds (one-sixth
speed) for the slow-motion condition. In the slow-motion
tape, an attempt was made to retain the intervals between
the segments at an approximate length of 10 seconds;
because of taping and editing errors, however, the actual in-
terval was closer to 9 seconds. (For specific composition

and speed of the stimulus tapes, see Appendices I, J and K.)

Rater Sample
The procedure thus far has yielded two stimulus tapes
(one in slow motion and one at normal speed) on which 101
segments of facial expressions occur with 10 second inter-
vals between them. The next step in the procedure is to

select raters who will label facial eXpressions when they

70

view the segments. The rater group consisted of 29
volunteers from education classes in "Group Leadership"
(ED 882) and in "Teacher Interaction Skills" (ED 882)
offered by the College of Education at Michigan State Uni-
versity. The average age of the raters was 31 and all ra-

ters in the study were graduate students (see Appendix L).

Rater Training

 

The first part of the training program began after the
rater had filled out the information sheet and release form
(see Appendices M and N). The program consisted of an intro-
duction to the six affective categories in the study, and
the appearance of these emotions in facial expressions. A
schematic of facial musculature was presented to acquaint
the rater with facial muscle groups (see Appendix 0). Each
of the six categories of emotions was then explained in
detail including synonyms for the affective terms, sensa-
tions associated with each state, possible elicitors, the
theoretical function of the affect, and finally, the facial
appearance of each state (see Appendix P). During the
discussion of an affect, a slide was shown depicting the
facial expression of that affect.2 The rater was also
introduced to the set of abbreviations which were to be used
in place of writing out the names of the emotional catego-

ries on the scoring sheet (see Appendix Q).

 

2All slides used in this section of training were
obtained courtesy of Carroll Izard, Vanderbilt University.

71

After the discussion of the six categories of emotion,
a set of 18 slides3 were shown to the raters and they were
asked to attempt to identify each photo using one of the six
emotional categories discussed previously. The raters at
this point were allowed to view these slides as long as they
wished. After the task was completed, the slides were
reshown along with the correct answers, and the raters were
encouraged to discuss any confusions or questions they might
have had at this time.

The raters were then administered an accuracy test
composed of 36 slides of spontaneous expressions.4 Each of
these slides was exposed for only 5 seconds; there was a
10-second interval between the showings of the slides so the
raters could record their responses on the answer sheet.
Since the timing of this test was identical to the timing
of the normal speed videotape, the results of these slides
were used to determine the initial accuracy of the rating
groups.

During the next training session, the results of the
first accuracy test were returned and the raters received
feedback on their scores. They were then shown videotaped
segments of the posed affects which were presented in the

same way as the stimulus tapes (S-second eXposures, followed

 

3All slides used in this section of training were
obtained courtesy of Carroll Izard, Vanderbilt University.

4Slides were produced from videotapes obtained courte-
sy of Bob Wilson, College of Education, Michigan State
‘University.

72

by 10-second rest). Because of the posed eliciting condi-
tions, however, these responses were not scored for accuracy.
The purpose of this section of the procedure was to
familiarize the raters with the task of viewing videotapes.
The final part of the training task was a second
accuracy test which was identical in form to that mentioned
previously.5 Slides of the spontaneous expressions were
again exposed for 5 seconds followed by lO-second rests.
Raters were asked to identify the facial expression using
the 6 emotional labels. When each rating group member
established a criterion level of at least 80 percent
accuracy in identification of the slides, then they were

ready to rate the stimulus material.

Design

This design was intended to test the effects of tape
speed (normal vs. slow) and test-retest procedures on the
raters' perception of spontaneous facial expressions.
There were two dependent measures employed: the number of
facial expressions observed (M1), and the accuracy of
observers' identification in each of the six categories
(M2).

Each group of observers viewed the same videotape

‘Uwice (see Table 3.2). Group one saw the normal speed tape

 

5Slides were produced from videotapes obtained
courtesy of Bob Wilson, College of Education, Michigan
State University.

73

followed by the slow motion speed tape; group three saw
slow motion followed by normal; and group four saw slow
motion followed by slow motion. A minimum of one week
elapsed before a group was presented with the second tape.
Observers who viewed the slow motion version of the tape
were allowed a rest break of 10 minutes after 45 minutes
of viewing time since the slow motion task was found to be
a very fatiguing one.

Responses were recorded on forms similar to those
used during the training session (see Appendix R). If the
raters wished to record more than one affect for a segment,
they were instructed to write the "most intense and pre-
dominant" affect first, and the other affect(s) second,
third, and so on. Thus the order in which the rater
recorded the affects was not necessarily the order in which

they appeared on the tape.

74

Table 3.2. Design of the Study.

 

 

 

 

 

 

 

 

 

 

 

 

\DGJQOSU'ibwwl-‘w

 

H
O

 

H
H

 

U)
p...
U)
H
N

 

l—'
U.)

 

|—'
h

 

H
U!

 

|-"
Ch

 

H
\l

 

H
00

 

l-"
\D

 

N
O

 

N
H

 

N
N

 

N
U)

 

N
b

 

 

N
U'l

 

N
0‘

 

N
\l

 

N
on

 

 

 

 

 

 

 

 

N
\D

 

 

 

 

 

 

 

 

 

75

 

Abbreviations: 81 = normal speed

82 = slow motion

T1 = speed at time 1

T2 = speed at time 2

M1 = number of facial expressions
per segment

M2 = number of right answers per
category

P1 = first presentation

P2 = second presentation

A1 = interest-excitement

A2 = enjoyment-joy

A3 = surprise-startle

A4 = distress-anguish

A5 = anger-rage

A6 = contempt-disgust

Post Hoc Rater Questionnaire

 

Two months after the stimulus tapes had been viewed
by the raters, a summary of the study and a questionnaire
was mailed to each person who had rated the tapes (see
Appendices S and T). This questionnaire consisted of 12
items in three categories: the rater's state of mind while
viewing the stimulus tapes, the quality of the tapes them-
selves, and the rater's evaluation of the overall study.
Each item on the questionnaire consisted of a seven point
scale on a continuum such as "interested-bored," or
"relaxed-tense." The purpose of this quesionnaire was to
determine whether confounding variables were operating as a
function of the composition of the rater groups or as a

function of the stimulus tapes themselves.

 

1:1:

L“?

(:1:

76

Hypotheses

 

Six research hypotheses were tested, each postulating

a relationship between speed of the videotape and the

accuracy of frequency of raters' responses. The six

hypotheses

Hypothesis

are as follows:

1: Trained observers will record a greater

 

Hypothesis

number of facial expressions when viewing the
stimulus tape at slow motion than when it is

viewed at normal speed.

2: Trained observers will identify facial

 

Hypothesis

expressions more accurately when viewing a
stimulus tape in slow motion than when it is

viewed at normal speed.

3: Trained observers will record a greater

 

Hypothesis

number of facial expressions when the unstruc-
tured interview is played in slow motion than
when the structured interview is played in slow

motion.

4: Trained observers will identify facial
expressions more accurately when the unstruc-
tured interview is played in slow motion than
when the structured interview is played in

slow motion.

77

Hypothesis 5: When the slow motion videotape is vieweci

 

at time 1, trained observers will record a
greater number of facial expressions at presen-

tation 2 than they did at presentation 1.

Hypothesis 6: When a slow motion videotape is viewed

 

at time 1, trained observers will more accurately
identify facial expressions at presentation 2

than they did at presentation 1.

For each of these hypotheses, the alpha level is set at

p = less than .05.

Analysis of the Data

 

The analysis of the data took place in several steps.
The primary rater response was scored as right or wrong,
corresponding to the answer key developed through the
responses of the expert judges. Each affective category
was considered a separate scale, and the total number of
right answers per scale was designated as one of the depen-
dent variables, measure 1 (M1).

The accuracy of raters' judgments was calculated for
the primary response and compared to the accuracy expected
by chance alone. To determine inter-item reliabilities,
coefficient alpha was calculated for each of the affective
scales. Spearman Rank Order correlations were also calcu-
lated to determine the reliability of the patterns of

responses for each scale across the two presentations of

78

the stimulus tapes.

Measure 1, the total number of right answers per
affective category was then analyzed in a design across
subjects and a design across measures. The design across
subjects tested the effects of the speed at time 1 (T1),
the speed at time 2 (T2), and the interaction between these
speeds. The speeds compared are normal speed (F) and
slow motion (S). The design across measures tested the
effects of presentation 1 (P1) and presentation 2 (P2) on
each of the six affective scales (A1 through A6). Again
the dependent variable was the number of right answers on
each affective scale.

Measure 2 concerned the number of facial expressions
recorded by raters for each segment. It was calculated by
dividing the number of segments in which more than one
facial expression was seen, by the total number of segments
or items on a scale. The dependent variable then was the
percent of items in a scale where more than 1 affective
facial expression was observed.

The designs utilized for measure 2 were the same
as those used for measure 1. The across subjects design
tested the effects of speed at time 1, spee<lat time 2 and
the interaction of speeds. The across measures design
tested the effects of presentation 1 and presentation 2 on
the frequency of responses across the six affective

categories.

79

The results for both measure 1 and measure 2 were
annalyzed in a multivariate approach to repeated measures
axnalysis of variance incorporating multiple measures taken
art.each measure point. In this design it was possible to
analyze not only the main effects in each of the designs
1311t also the interaction between the designs themselves.
{Pile variables studied in the overall design thus includes:

For M1 (accuracy data) and M2 (frequency data):

Tl speed at time 1

T2 speed at time 2

T1 x T2 = speeciat time 1 x speed at time 2

Pl/PZ = test-retest

Tl x P1/P2 = speed at time 1 x test-retest

T2 x Pl/PZ = speed at time 2 x test-retest

Tl x T2 x P1/P2 = speed at time 1 x speed at time 2 x
test-retest

Tflae significance level for all of the effects studied was

Set at p< .05.

CHAPTER IV
ANALYSIS OF RESULTS

Introduction: Overview of the Design and Analysis

Because of the complexity of the design, the data
in the study is reported in several steps. The first area
covered is Subject Self Reports: how subjects identified
their own facial expressions using each of Tomkins' eight
categories. The second section, Expert Judge Agreement,
focuses on the agreement of the subject with a panel of
6Xpert judges.

The outcome of the Expert Judge Agreement data was
twofold: the design was altered to provide a stronger test
of the main hypotheses at the expense of two minor hypo-
theses, and the segments appropriate for the stimulus tapes
Were chosen along with the "right answer" key for these
Segments.

The simulus tape was presented to trained raters, and
two measures (M1 and M2) were recorded and analyzed. The
a<:<:uracy of the primary rater responses (M1) was compared
witzh chance levels of accuracy; also Coefficient Alpha and
Spearman Rank Order correlations were calculated to deter-
mine inter-item and test-retest reliability. Finally, a

multivariate approach to analysis of repeated measures

80

81

designs which incorporates multiple dependent variables at
each of the points of repeated measurement was employed for
both the accuracy and frequency measures. The design thus
tested the effect of the following variables: speed at
time 1 (normal vs. slow motion), speed at time 2 (normal
vs. slow motion), test-retest, and six affective categories
(interest, joy, surprise, distress, anger and contempt-
disgust). The dependent measure which was analyzed for
each of these effects was the accuracy of the raters'
responses.

The second dependent measure analyzed (M2) consisted
c>f the proportion of videotaped segments where the rater
Jrecorded seeing more than one emotion. This frequency data
was analyzed in the same type of analysis of variance for

repeated measures as was performed for the accuracy data.

Finally, the outcomes of these two designs were

Irealated to the remaining hypotheses of the study and the
Statistical decision was made as to whether or not the
1111111 hypotheses should be rejected, and the directional

a l ternatives accepted .

Subject Self Report
The dependent measure in the subject self-report data
is the total number of responses the subjects gave in each
affective category while reviewing the tape of their own

primary interview. No mention is made here of the validity

82

of subject responses but merely of the frequency of
responses in each category.
From Table 4.1, it is clear that some emotions were

labeled more often than others. Interest was the category
most often labeled, followed by distress and joy. The
categories of anger and contempt were used by subjects
less frequently than interest, distress, or joy and the
categories of surprise, fear and shame were labeled quite
infrequently. Shame-humiliation was the category least
utilized by subjects, with a total of only 12 instances of
shame labeled by subjects in all of the recall sessions.
fPhese data are presented graphically in Figure 4.1.
Tkable 4.1. Number of facial expressions reported by

subjects in each of eight affective categories
for the two interview conditions.

 

Affect I1 12 Total
Interest 52 50 102

Joy 43 30 73

Surprise 24 3 27

Distress 28 58 86

Anger 33 14 47

Contempt-

Disgust 39 7 46

Fear 19 8 27

Shame 10 2 12

\

structured

Abbreviations: I1

unstructured

83

mwfluommumo
m>fluommm< usmflm mo comm cw cmuuommm mcofiHOEm uomnnsm Hobos

.H.v musmflm

 

 

 

 

 

 

macaw ummm .ucoo .024 .umao ..Hsm hon .ucH

» +( (43%;- w. -2 w. 4 wt 1
(an
16v
m
.Wm
m

/// _

lam

 

coa

snoacqns Kg panzodau snoesgv :0 Jaqmnu

84

When subject responses were divided between the two
interview conditions, there was a marked contrast in the
subject response frequencies (see Figure 4.2). Subjects
in the unstructured interview condition made markedly
fewer responses in the categories of surprise, anger,
contempt, and fear. The only category in which subjects
in the unstructured. condition made more responses was the
category of distress-anguish, and here they made over
twice the number of reSponses than had the subjects in the
structured interview condition. In only two categories did
Subjects in the interview conditions produce comparable
response frequencies: interest, the most popular category,

and shame, the least popular category.

Expert Judge Agreement

 

After the subject self report responses were recorded,
a panel of expert judges viewed those portions of the tape
chosen by subjects and made independent judgments of the
affect present. The data compiled here (Table 4.2) are the
InHuber of instances in which 75 percent of the judges
agreed on the category label regardless of the subject's
labeling of the segment. The greatest amount of judge
agreement was in the categories of interest, joy, and
distress. Surprise, anger and disgust yielded a moderate
amOUnt of judge agreement, and fear and shame yielded the
fewest number of instances where 75 percent of the judges

agreed on the category.

85

mcownflocoo 3ofl>uwucH 039 map ca mmfluommumu
uamflm mo comm an muuwnnsm >3 nmflmaucmcH muomumm mo umnEsz

mfiwzm umom .umou .mmd .uMHQ .Hsm won
4‘ a _ _

» l
d

      

{I
«pmusuosuumcsv / \ /

N wmmcm //

  
   

.~.4 museum

.udH
_

0H

ON

ow

 

sqoafqns Kq peqxodeu sqoeggv go Jeqmnu

86

Table 4.2. Number of Instances of 75 Percent Agreement
in the Two Interview Conditions for Each of
the Eight Affective Categories

 

Affect I1 12 Total
Interest l9 14 33
Joy l9 13 32
Surprise 7 3 10
Distress 13 19 32
Anger 7 ll 18
Disgust 6 12
Fear 2

Shame 4

I1 structured interview

12 unstructured interview

In comparing the two types of interview conditions,
ble same type of pattern of judge agreement occurred
Esee Figure 4.3). Interest, joy, and distress yielded
Efeatest judge agreement and fear and shame the least.
sanerally, judges were less in agreement on segments taken
tfom the unstructured interview condition with the excep-
i.ons of the categories of distress and anger. For these
~70 affects, especially distress, the unstructured inter-
i.eW’yielded more agreement among the judges than did the

: ructured interview .

_ “In," ‘

.hu‘fin
}

87

mmfluoowumu m>Huommm¢ ucmflm mo comm Ce mmmpwh
uumdxm mo ucmsmmumm unwouwn mm mo moocmumcH mo nonesz amp 9

o . o . 0 %Oh
“mom .ucou ocd umHQ How

.m.v musoflm

 

62H
mEmnm _ — _ _
T J _ _
..... V. .Mhfl.. . O
0.. \ /. ....
.\ /.. .. .. um
/../ .. .
.. z I; .. \./ .
. \ .
. (WV . I.
n/ \.. 2
o CO . o. /
.. JV/I \
“.0 . pwv . / I.
e .. «z . .
{W . I» . / u
no .. . 13
x,» 60 .. /
n 9 . /
w 6., .
k «I
“If I II! II.
x, . . 1
on
1mm

 

quemeaxbv queoxed SL go saoueqsux go Jeqmnu

88

Revision of Design

 

Because of the low overall number of instances of
75 percent judge agreement, two major revisions in the
design were instituted. Because of the small number of
subject recall responses and the small number of instances
of 75 percent judge agreement in the affective categories
of fear—terror and shame-humiliation, these categories were
eliminated and the two interview conditions (11 and 12)
were merged. The stimulus tape then contained segments
from both the structured and unstructured interview con-
ditions and the data were not analyzed in terms of this
variable.

The elimination of the categories of fear-terror and
shame-humiliation has no effect on the overall design. All
hypotheses can be tested utilizing six categories rather
than the original eight. The elimination of the interview
condition as a variable does have important effects on the
design and the tests of two hypotheses. These hypotheses
involved the prediction that there would be differences on
each of the measures (accuracy and frequency) owing to
the type of interview condition employed for that subject.
111 this revised design, these hypotheses are untestable.

The strength of the revised design is that it allows
(file remaining hypotheses to be tested with greater reliabi-
Lity; Because each of the remaining scales has at least

nirue items (most have 15 or 20) the hypotheses regarding

89

the effect of tape speed and test-retest can be more relia-
bly tested. It is these remaining hypotheses that con-

stitute the core of the study, and thus the elimination of
the two hypotheses regarding interview conditions does not

alter the main thrust of the research.

Stimulus Tape Composition

The stimulus tape was composed of 101 segments but
because of editing errors and distortions in the videotape
output due to faulty equipment, the number of useable seg-
ments was reduced to 88. Of these, 65 percent involved
female subjects and 35 percent involved males. Fifty-two
Percent were drawn from the structured interviews and 48
Percent from the unstructured interviews. Segments were
edited in a random order onto the master tape through the
use of a random numbers table (Glass and Stanley, 1970) .
For an item by item breakdown of the segments on the master
tape, see Appendix I and J.

Two criteria are used to deVelop the right answer key
for the segments on the master tape: 75 percent agreement
of judges, and the better than chance agreement of
subjects with judges. Segments were chosen on the basis
of 75 percent of agreement of expert judges, regardless of
the subject's label. Moreover, in 60 percent of the chosen
Segments, the subjects also agreed with the label assigned
by the judges. The percent of agreement expected by chance

alone is 12.5 percent. Thus the subject's agreement with

90

the judges for the segments used in the stimulus tape is
almost five times the level expected by chance. The high
judge agreement coupled with better than chance agreement
Ibetween subjects and judges makes a strong case for the

validity of the segments employed.

Analysis of Dependent Measures

 

The stimulus tapes were viewed by a sample of 29
trained raters. For this task, an answer form was developed
which recorded two measures: first, the raters' judgments
of the primary affect in the segment, and second, the
number of facial expressions seen in that segment. It is
these two measures that form the core of the data in the

study .

Accuracy of Rater Respgnses

 

Measure 1 concerns the accuracy of rater identifica-
tiixbns of facial expressions on the master tape. To calcua
late this measure, the number of right answers in each
affective category was computed for each subject.

It was of interest to find out not only with what
frequency people answered correctly on segments keyed for

each of the affective scales, but also to find out in
‘Vlléit: way errors were cammitted. Table 4.3 and Figure 4.4
‘iflllllstrate the frequency distribution of responses given to
items on each of the affective scales. Each chart in
Figure 4.4 represents all items keyed to the correct

r a o '
es£><>nse--one of the 81X affective categories.

 

91

 

 

 

 

 

 

 

 

 

Table 4.3. Number of Answers and Standard Deviations in
Each Affective Category for Each Scale at
Time 1 and at Time 2

Int. Joy Sur. Dist. Ang. Con.
T1 3?: 9.586 .759 .488 3.75 1.38 .86
SD 2.57 1.09 7.36 1.70 1.29 1.06

Interest _
T2 X 10.48 .828 .483 3.06 .97 .69
SD 2.49 1.04 .74 1.80 1.15 1.07
Tl X 3.10 13.35 .62 .31 .35 .17
SD 2.21 2.72 1.18 .54 .55 .47

JOY _
T2 X 2.80 13.52 6.21 3.79 .14 .10
SD 2.14 2.65 .98 6.8 .36 .56
T1 X 1.55 .276 4.72 .17 .14 .10
SD 1.21 .46 1.36 .38 .35 .41

Surprise

T2 3? 1.69 .138 4. 59 .27 .17 .10
SD 1.56 .44 1.72 .59 .47 .56
T1 3(— 3.86 .55 .24 7.52 2.03 2.72
SD 2.42 .73 .56 3.35 2.71 2.16

Distress _
T2 X 4.31 .244 .345 7.48 2.20 1.89
SD 2.58 .497 .614 3.08 2.23 1.86
T1 X 3.59 .55 .24 2.28 6.31 2.79
SD 2.16 .69 .51 1.56 2.88 1.89

Anger _
T2 X 3.86 .48 .41 2.0 6.79 2.28
SD 2.60 .69 .87 1.73 3.22 2.03
T1 X 3.97 .76 .17 2.24 1.24 4.72
SD 1.89 .64 .38 1.21 1.24 1.73

Contempt _
T2 X 4.35 .72 .17 1.80 1.17 4.52
SD 1.91 .53 .47 1.29 1.22 2.03

92

mocmsu an pmuommxm acousoom mo Hm>mq ou omummeoo mm
m mass can a mafia um mmamom see you mmeuommumo m>euommme xem mo
comm ca Hm>umucH mocmoflmcou unwound mm pom mumsmgs mo Hmafisz sums .¢.v musmfim

 

 

 

noncommmm mo Hwnfisz com: m.m u mocmso
as me me ea ma NH Ha OH m m h o m e m‘\ m H
_ _ a _ (a. a _ _ _ _ 14. _ _ _ : _ .IWT.o
.V IIII+. s
ulll+rll o
_ _ m NH.
_ .||I+ e
IIILII. _ H acme
_ ||I1I o umueoxm
.III+I jmr \ummumu
ill— ID. IGH
_ .flwm H9
_ Il¢l h
.I+II. _ fiw

 

 

 

 

93

Umscaucoo .v.¢ mucosa

noncommmm
mo Honesz cam: m.H u wosmnu

e m m e m m H k

_ m a H a q a _ +-U
_ ..«
_ :Tna we
_ _+e
_ H mauumum
.1o

 

 

 

 

 

 

 

 

 

 

 

_ x
_ .+< rmmflumusm
_ +n_ H9
III+|
_ n
IIW1._ H
noncommmm mo Honasz com: o.m n mocmsu
ma ha ma ma «a ma NH Ha ca m m h o m e m\ N H
____4__4__4__l__44_
_ e
_ _
_ “#WW «6
IIIIITII: _
.IAHHI: H woe
_ .+LP \ucme
_ .+Iuw I>0ngm
[Ta
_
nI+III _

 

 

 

 

poncflucou .v.v musmflm

noncommmm Ho Honesz cows
ma ma «a ma NH Ha ca m m h m m w m «e H

 

 

 

94

 

 

 

 

 

 

 

 

 

 

 

 

_ _ _ _ _ _ _ _ (4 all _ all 4l4+tl. a o
.1 _ e
IT: a ma
_ m
. b
_ _ -W mmmm
__ mm
_ _ \H as
.II+II. n_ as
_ le
_ 1+! o
IIITI._ (H
noncommmm mo umnasz cams mm.m u mono
h m w m
_ _ A 4‘ fi_lr ,0
4l|+|| 4
1 _ NB
[Trm
_ km
W _ JW masons
n .
u _ AW \mmmuumwo
1 E.
_ m
_ n
(LII. -H

 

 

95

 

 

UmscHucoo .¢.v musmHm
Hm>umucH mocopHmcoo usmoumm mm was some n ll+ll
Hm3mcm uownnoo mm UmumcmHmmp whommumo n t
umsmmelumeucoo u U
momulnmmcm u d
anszMImmmHume n a
wHuHmumlmmHHmudm u m
acmluco8%oflcm u b
ucwfimuHoxmlummHmucH u a
m 08H» H mm.
H mEHu u H9 "mGOHumH>mHQQ¢
noncommmm mo Honesz cam: mH.m u mocmsu
MH NH HH OH m m h m m w m m H
14‘ fl _ (H _ _ s I+I+I+. 1| 41 .H
_ IITI. s
I_IT Wm: NH.
m.
rll
_ .+ CMV umsmmHo
i. ||+||I _ H \umfiwu
.IIITII _ TN acoo
_ IT M:
(\..—II, 10! H8
kw
_ 3+ nun!
llTl _ H

 

 

 

 

 

96

Thus it becomes evident not only how often subjects
answered correctly (identifying interest on an "interest"
item for example) but also how often subjects answered in
each of the incorrect categories (for example answering
"joy" for an "interest" item).

The first and most important aspect of this data is
that raters identified the segments with an accuracy which
was much greater than chance would predict. Although many
times the average levels of accuracy were low, in each
category they were significantly different from chance
(p < .05). Joy was the most accurately identified emotion,
followed by interest, surprise, distress, anger, and finally,
contempt. For five of these six categories, the second
highest score was consistently in the category of interest-
excitement. For the scale in which interest was the right
answer, the second highest score is that of distress-
anguish. In three of the six categories, the second highest
score was significantly different from chance. However, in
only one category (contempt-disgust) did the second highest
score (interest) approach the accuracy level of the "right
answer" score. Figure 4.5 summarizes the results of the
accuracy data for both the correct answers in each category
and the highest frequency of wrong answers in each

category.

muomwumo comm cH Hm3msd mcou3 hocmsvmum unmanm can muommumu
”mama momuaoos mo humEEsm .m.¢ musmHm

20mm CH mumzmcd uomuuou mo usmoumm

Hm3msm mcou3

 

 

.ucH .HCH .ucH .ucH .ucH .umHD mocmsvmnm ummanm
.coo .mc¢ .umHQ .Hsm moo .ucH mumsmcm uanm

mH. n mocmuudooo wocmzo I. oH

‘|" I’ d
1‘. 1.

Hm3mcm mcoua \.\ :4) m
wocmoomnm ummcmHm .\ T on u
q
Hm3mcm usmHm l S

as
m
I m
cm a
S
cm M”
a
I 1
on S

cm

cm

r

 

00H

98

Reliability of the Scales

Two types of reliability were calculated for each
of the six affective scales. Coefficient Alpha was calcu-
lated to determine the inter-item correlations between the
items in the scales. Spearman Rank Order Correlations were
computed for the relative frequency of each possible
response in each of the scales across the two testingS- Thus
the overall pattern of right and wrong answers for each
scale was compared over two presentations.

The inter—item correlations between items in each
scale is presented in Table 4.4. From this data it is
clear that the inter—item reliability within each scale
was quite variable. The highest reliability was in the
distress-anguish scale, followed by enjoyment-joy, anger-
rage, and interest—excitement. Two scales which had
markedly lower reliabilities were contempt-disgust, and
surprise-startle. It is possible that the low reliability
of the surprise scale was due to the small number of items
(n = 7) which were included.

The Spearman Rank Order Correlations (see Table 4.5)
for the two presentations of the stimulus tapes were much
higher than the coefficient alphas. To obtain these
correlations, the mean number of responses in each category
were calculated and ranked from highest to lowest. These
rankings are compared at the two presentations of the tape.

Thus, the Spearman correlations provide information as to

99

Table 4.4. Coefficient Alpha for Inter-item Reliability
of Each of the Six Affective Scales

 

 

 

Affect Reliability
Interest .53
Joy .70
Surprise .28
Distress .71
Anger .62
Contempt .32

 

Table 4.5. Spearman Rank Order Correlations for the
Relative Frequency of Identifications on
Each Affective Category for the Two Presen-
tations of the Stimulus

 

 

Spearman Rank

 

AffeCt .Order Correlations
Interest 1-00
Joy .74
Surprise -81
Distress -54
Anger 1.00

Contempt 1.00

 

100

the consistency of the overall pattern of responses on
each scale over time. The correlations for the scales of
interest, anger, and contempt are very high (1.00), and
the correlations for joy and surprise are moderately

high (.74 and .81). Distress was the only category with a
relatively low correlation between the rankings at the two

presentations (.54).

Design I: Accuracy Data

 

Design Over Subjects

 

The design over subjects tests overall performance of
raters by employing the sum of the number of right answers
in both presentations of the stimulus materials (see Table
4.6). Each rater is scored for the number of right answers
and the total right/wrong score is computed for each scale.
The effects tested in this design are the time 1 effect
(T1), the time 2 effect (T2) and the interaction effects
between time 1 and time 2 (T1 x T2) for each of the six
affective categories (A1 - A6). The analysis of variance
for the design over subjects is presented in Table 4.7 (see
Appendix U for cell means). In this design, neither the T1
effect, the T2 effect nor the T1 x T2 effect were signi-
ficant at the .05 alpha level. The cell means obtained for

design 1 are presented in Appendix U.

Table

.101

4.6. Design I: Accuracy Data

 

 

 

 

 

P1 P2
T1 T2 R
Al A2 A3 A4 A5 A6 A1 A2 A3 A4 A5 A6

F n=8

S n=8

F =8
S S n=8
T1 = speed at time 1 P1 = first presentation
T2 = speed at time 2 P2 = second presentation
F = normal speed Al-A6 = affective scales
S = slow motion R = raters

Effects Tested in Table 4.6.

 

T1 - Time 1 Effect. The overall performance of raters is

 

influenced by the speed of the videotape they view
first regardless of the speed of the second videotape.

T2 - Time 2 Effect. The overall performance of raters

 

T1*T2

will be influenced by the speed of the videotape they
view second, regardless of the speed of the first
videotape.

- Time 1 Time 2 Interaction. That the overall per-

 

formance of raters will be influenced both by the speed
of the videotape they view at time 1 and the speed
of the videotape they view at time 2.

T1*P1P2 - Time 1 x Test-Retest Interaction. That the change

 

in raters' performance will be influenced by the speed
of the videotape which they view at time 1.

T2*P1P2 - Time 2 x Test-Retest Interaction. That the change

 

in raters' performance will be influenced by the speed
of the videotape which they view at time 2.

T1*T2*P1P2 — Time 1 x Time 2 x Test-Retest Interaction.

 

That the changefiin raters' performance willibe influ-
enced both by the speed of videotape they view at
time 1 and the speed of videotape which they view at
time 2.

102

Table 4.7. Anova for Design Over Subjects, Accuracy Data

 

 

 

Source df F p Sig
Grand Mean 6 303.39

T1 6 .545 .7674 ns

T2 6 1.232 .3317 ns

Tl x T2 6 1.157 .3673 ns

Design Over Measures

 

The design over measures tests the change in performance
between the two presentations of the stimulus materials,
P1 and P2 across the six affective categories (Al - A6).
The dependent measure is the number of right answers per

affective scale at either P1 or P2. The effect tested in

103

this design is the PLP2 or test-retest effect. Presented
along with the design over measures is the interaction
between this design with the design over subjects. Thus

the effects tested in this analysis include the test-retest
effect (P1P2), the interaction between the time 1 effect and
test-retest (Tl x P1P2), interaction between the time 2 ef-
fect and test-retest (T2 x P1P2), and the interaction
between time 1, time 2 and test-retest (Tl x T2 x test-re—
test). The analysis of variance for the design over
measures and the interaction between designs is presented

in Table 4.8.

Table 4.8. Anova for Interaction of Design Over Measures
with Design Over Subjects: Accuracy Data

 

 

 

Source df F p Sig
P1P2 6 1.07 .447 ns
T1 x P1P2 6 3.01 .029 p < .05
T2 x P.P2 6 1.16 .363 ns
T1 x T2 x P.P2 6 .82 .567 ns

 

In this analysis, the test-retest effect, the T2 x test-re-
test effect and the T1 x T2 x test-retest effect fail to
reach significance at the .05 alpha level. The T1 x test-
restest effect however was significant at alpha < .05. In
Table 4.9, this effect is analyzed for each of the affective

scales.

104

Table 4.9. T1 x P1P2 Effect: Anova for Six Affective

 

 

 

Categories

Affect nggged F p Sig
Interest 21.25 3.89 .05 p < .05
Joy 22.69 2.59 .11 ns
Surprise 3.78 .99 .32 ns
Distress .83 .11 .74 ns
Anger 54.24 3.98 .05 p < .05
Contempt 1.008 .35 .55 ns

 

Frequency_of Rater Responses (M2)

 

Measure 2 involves the number of rater responses to
segments of the stimulus materials. Raters were allowed to
record as many answers as they saw expressions. While the
first measure concerned the accuracy of the primary answer,
measure 2 concerns only the number of answers recorded for
segments, regardless of the accuracy of later answers.

Each item was scored "0" if one answer was recorded
and "1" if more than one answer was recorded per segment.
The number of items with more than 1 answer was counted for
each scale and divided by the total number of items in the
scale. This dependent measure (recorded for each scale)
is, then, the prOportion of items for which a rater recorded

more than 1 answer per item.

105

Design II: Frequency Data

Design Over Subjects

 

The design over subjects for the frequency data is
identical to the design over subjects for the accuracy data
(see Table 4.10). The dependent variable, however, is
measure 2, the overall proportion of items where raters
recorded more than one answer, computed for each of the six
affective scales and summed over the two presentations.

The effects tested in this design are the time 1 effect (T1),
the time 2 effect (T2) and the interaction effect between
time 1 and time 2 (T1 x T2) for each of the six affects

(A1 - A6). The analysis of variance for the design over
subjects is presented in Table 4.11. The cell means

obtained for design 2 are presented in Appendix V.

106

Table 4.10. Design II. Frequency Data

 

 

 

 

 

P1 P2
T1 T2 R
Al A2 A3 A4 A5 A6 A1 A2 A3 A4 A5 A6

F F n=8

S n=
S F n=8

S =8
T1 = speed at time 1 P1 = first presentation
T2 = speed at time 2 P2 = second presentation
R = normal speed Al-A6 = affective scales
S = slow motion R = raters

Effects Tested in Table 4.10.

 

T1 - Time 1 Effect. The overall performance of raters
is influenced by the speed of the videotape they view
first, regardless of the speed of the second videotape.

T2 - Time 2 Effect. The overall performance of raters
will be influenced by the speed of the videotape they
view second, regardless of the speed of the first
videotape.

T1*T2 - Time 1 Time 2 Interaction. That the overall
performance of raters will be influenced both by the
speed of the videotape they view at time 1 and the
speed of the videotape they view at time 2.

P1P2 - Test-Retest Effect. That the performance of the
raters will change with a second exposure to the
stimulus tape.

T1*P1P2 - Time 1 x Test-Retest Interaction. That the
change in raters' performance Will be influenced by the
speed of the videotape which they view at time 1.

T2*P1P2 - Time 2 x Test-Retest Interaction. That the
change in raters' performance will be influenced by the
speed of the videotape which they view at time 2.

T1*T2*P1P2 - Time 1 x Time 2 x Test-Retest Interaction.

That the change in ratersT performance will Be in-
fluenced both by the speed of videotape they view at
time 1 and the speed of videotape which they view at
time 2.

 

 

 

 

 

 

 

107

 

 

 

Table 4.11 Anova for Design Over Subjects: Frequency Data
Source df F p Sig.
Grand Mean 6 17.96
T1 6 .34 .9034 ns
T2 6 2.70 .0436 p < .05
T1 x T2 6 3.12 .0251 p < .05

 

In this analysis the T1 effect was not significant. However,
the T2 effect and the interaction effect between T1 and T2

were both significant at alpha '<.05. .In Tables 4.12 and
4.13 these significant effects are analyzed for each of the
six affective scales. The interaction effect (T1 x T2) is
graphically depicted in Figure 4.6.
T2 Effect:

Table 4,12. Anova for the Six Affective Scales

 

 

 

Source Mean Squared F p

Interest .01 .30 .58 ns
Joy .07 1.40 .24 ns
Surprise 1.49 8.44 .007 p ‘<.05
Distress .30 .41 .52 ns
Anger .15 1.55 .22 ns
Contempt .01 .19 .66 ns

 

Table 4.13.

T1 x T2 Effect:
Scales

108

Anova for the Six Affective

 

 

 

Source Mean Squares F p

Interest .39 6.60 .01 p‘<.05
Joy .71 14.34 .0009 p <.05
Surprise 2.50 14.1 .001 p<:.05
Distress 1.22 14.8 .0008 p'<.05
Anger 1.01 10.1 .004 p:<.05
Contempt .10 12.07 .001 p‘<.05

 

Mean Percent of Items With
More Than One Answer Per

Segment

Figure 4.6.

 

 

.70_
\ O
.60_ .\ //////
-\ ’//// T2 = Slow
.50 __r \ //,--,
' \
. 40 /_/
—-1 ‘/ \
«fl'
. 30 _J \-
\\ T2 = Fast
.20 _ \
\s
.10__1
4 l
Slrow Fast
T1

T1 x T2 Interaction Effect, Design Across

Measures,

Frequency Data:

Mean percent of

Items With More Than One Answer Per Segment
at T1 and T2 for Normal Speed and Slow Motion

109

Design Over Measures

The design over measures for the frequency data is
identical to the design over measures employed for the
accuracy data. In this design, the change in performance
between the two presentations of the stimulus materials
(P1 and P2) is tested for each of the six affective scales
(A1 - A6). The dependent variable is the proportion of
items in a scale where the rater recorded more than one
answer during the first presentation (P1) and during the
second presentation (P2). In analysis (see Table 4.14),
only the T1 x test—retest effect was significant at the .05
alpha level. In Table 4.15 this effect was analyzed across

the six affective scales

Table 4.14. Anova for Design Over Measures x Design Over

 

 

 

Subject
Source df F p Sig
P1P2 6 1.33 .28 ns
Tl x P1P2 6 4.73 .003 p <,05
T2 x P1P2 6 2.23 .082 ns

T1 x T2 x P1P2 6 1.34 .287 ns

 

110

 

 

 

Table 4.15. _Tl Effect: Anova for the Six Affective Scales

V riable Mean F '

a Square P 519
Interest .10 3.33 .08 ns
Joy .25 15.84 .00006 p < .05
Surprise 1.5 24.22 .0001 p < .05
Distress .13 5.18 .0317 p < .05
Anger .67 12.00 .002 p < .05
Contempt .12 3.3 .08 ns

 

Post Hoc Questionnaire

 

The results of the post hoc rater questionnaire

(see Table 4.16) were analyzed on 3 dimensions:

1) rater

state of mind (R), 2) tape quality (T) and 3) overall

value of the study (8).

A multivariate analysis of

variance for repeated measures was performend to test each

of these three effects (for design outline and cell means,

see Appendix W).

Employing the .10 alpha level as the

criterion for significance, significant effects are found

in the "rater state of mind," the "tape quality" and the

overall "value of the study" (see Tables 4.17, 4.18, 4.19).
T2 effects were found in both the
the "value of the study" data.
were found for the "quality of the tapes" and the "value of
the study" data.

the tapes" data.

"rater state of mind"

T1 x T2 interaction effects

A Tl effect was found in the "quality of

and

111

Table 4.16. Post Hoc Questionnaire’

 

 

T1 T2 r R T S

 

 

 

speed at time 1
speed at time 2
normal

slow'motion

raters

rater state of mind
tape quality

T1

T2

F

S

r

R

T

8 value of the study

Effects Tested in Table 4.16

 

T1 - Tl Effect. That the speed of the tape at time 1
influenced the rater's scores.

T2 - T2 Effect. That the speed of the tape at time 2
influenced raters' scores.

T1*T2 - Time 1 x Time 2 Interaction. That both the speed
of the tape at time 1 and the speed of the tape at
time 2 influenced raters' scores.

 

 

 

112

Table 4.17. Anova For Post Hoc Rater Questionnaire
Rater State of Mind

 

 

 

Source df F p Sig
Grand Mean 6 168.39
T1 6 .7794 .5980 ns
T2 6 2.51 .06 p <.10
T1 x T2 6 .4868 .81 ns

 

T1 Time 1; T2 = Time 2

Table 4.18. Anova For Post Hoc Rater Questionnaire:
Quality of Tapes

 

 

 

Source df F p Sig
Grand Mean 4 170.22
T1 4 2.41 .087 p <.10
T2 4 1.76 .18 ns
T1 x T2 4 2.50 .07 p <.10

 

T1 = Time 1; T2 = Time 2

Table 4.19. Anova For Post Hoc Rater Questionnaire:
Value of the Study

 

 

 

Source df F p Sig
Grand Mean 2 374.9
T1 2 .0312 .9694 ns
T2 2 3.96 .036 p <.lO
T1 x T2 2 3.80 .039 p <.10

 

T1 = Time 1; T2 = Time 2

113

Plotting the results from the Post Hoc Rater
Questionnaire (see Figures 4.7, 4.8, 4.9), groups who saw
the slow motion videotape at time 2 in general scored lower
on all three dimensions than groups who saw normal speed
at time 2. Moreover, raters in the slow-slow group
usually rated the study lower than either of the other three
groups. Thus the tendency of raters was to rate their
state of mind, the tape quality, and the value of the study

as poorer if the slow motion condition was viewed at time 2

 

than if normal speed was viewed at time 2. Moreover, on
each of these dimensions, the group who viewed s1 w motion
both at time 1 and time 2 rated the study lower than the

other three rater groups.

5 .3 T2 = Fast

4 \ T2 = Slow

Mean Scores

 

 

:1 5’

T1

Figure 4.7. Plot of Mean Scores for Rater "State of
Mind" for Speeds at Time lwand Time 2

114

Fast

l

1
F3

Slow

Mean Scores
00

 

 

T1

Figure 4.8. Plot of Mean Scores for Overall
Value of the Study for Speeds
at Time 1 and Time 2

Fast

'“‘ aw T2 = Slow

Mean Scores

 

 

T I
F S
T1
Figure 4.9. Plot of Mean Scores for Quality of
the Videotapes for Speeds at Time 1
and Time 2

115

Hypotheses

 

Hypothesis 1. Null Hypothesis: Trained observers will

 

 

record the same number of facial expres-
sions when viewing a stimulus tape in slow
motion as when it is viewed in normal speed.

Alternative Hypothesis: Trained observers

 

will record a greater number of facial ex-
pressions when viewing a stimulus tape in
slow motion than when it is viewed in normal
speed.
Examining the design over subjects, for the frequency data
(measure 2) we find the multivariate F value of the T1 x T2
interaction effect was 3.12 (see Table 4.11). This value
was significant at the .05 alpha level. Therefore the
null hypothesis for the tape speed effect on the frequency
of rater responses was rejected.
Examining the interaction effect more closely,
it would be expected that if 51 w motion viewing of the
videotape renders a greater number of affects observable,
then the slow-slow group would rank highest in number of
affects seen; the slow-fast and fast—slow groups would be
next, and finally the fast-fast group would be lowest. The
actual ranking was slow-fast, fast-slow (nearly equal)
followed by slow-slow and finally fast-fast (see Figure 4.6).
The Spearman Rank Order correlation between the expected and

actual order of the groups is .4. In addition, in this same

116

analysis, the T2 effect was significant at the .05 alpha
level. This effect expands upon the information available
from the interaction effect and the rank orderings. If the
slow motion tape was shown to raters at time 2, then

raters recorded a greater number of affects in terms of
their overall scores than if the normal speed tape was
shown at time 2. The slow-slow and the fast-slow groups,
therefore, viewed a greater number of affects than did the
fast-fast or slow-fast groups. This is additional evidence
that there were substantial differences in the number of
affects recorded by raters owing to the speed at which the
videotape was vieweCL Thus, moderate evidence exists for
the rejection of the null hypothesis and acceptance of

the directional alternative.

Hypotheses 2. Null Hypothesis: Trained observers will

 

 

identify facial expressions with the same
accuracy when viewing a stimulus tape in slow
motion as when it is viewed in normal speed.

Alternative Hypothesis: Trained observers

 

will identify facial expressions more accu-
rately when viewing a stimulus tape in slow
motion than when it is viewed in normal
speed.
The multivariate F value for the T1 x T2 interaction
in the design over subjects for the accuracy data was 1.157.

This value was not significant at the .05 alpha level (see

Table 4.7).

117

The null hypothesis for the tape speed effect

on the accuracy of rater responses was therefore not

rejected.

Hypothesis 3.

 

Null Hypothesis: Trained observers will
record the same number of facial expressions
when the unstructured interview is played in
slow motion as when the structured interview
is played in slow motion.

Alternative Hypgthesis: Trained observers
will record a greater number of facial expres-
sions when the unstructured interview is
played in slow motion than when the structured

interview is played in slow motion.

In the revised design, the variable of interview

condition (I1 and I2) was eliminated. This design there-

fore provided no test of the hypothesis of the effect of

interview condition on the number of rater responses.

Hypothesis 4.

 

Null hypohesis: Trained observers will

 

identify facial expressions with the same
accuracy when the unstructured interview is
played in slow motion as when the structured
interview is played in slow motion.

Alternative Hypothesis: Trained observers

 

will identify facial expressions more accuv

rately when the unstructured interview is

118

played in slow motion than when the struc-
tured interview is played in slow motion.
In the revised design, again the variable of inter-
view condition (Il and 12) was eliminated. This design
therefore provided no test of the hypothesis of the effect

of interview conditions on the frequency of rater responses.

Hypothesis 5. Null hypothesis: When the slow motion

 

stimulus tape is viewed at time 1, trained
observers will record the same number of
facial expression at presentation 2 as they
had at presentation 1.

Alternative hypothesis: When a slow motion

 

stimulus tape is viewed at time 1, trained
observers will record a greater number of
facial expressions at presentation 2 than
they had at presentation 1.

Examining the interaction between the design over
measures and the design over subjects for the frequency
data (see Table 4.18), the F value for the T1 x P1P2 inter-
action is 4.73. This value was significant at the .05
alpha level. However, the effect found was in the opposite
of the direction predicted. In other words, raters who
viewed the normal speed tape at time 1 showed a greater
amount of improvement at time 2 than those who saw slow

motion at time 1.

‘

119

Hypothesis 6. Null hypothesis: When a slow motion

 

 

stimulus tape is viewed at time 1, trained
observers will identify facial expressions
with the same accuracy at presentation 2 as
they had at presentation 1.

Alternative hypothesis: When a slow motion

 

stimulus tape is viewed at time 1, trained

observers will identify facial expressions

more accurately at presentation 2 than they
had at presentation 1.

Examining the interaction between the design over
subjects and the design over measures for the accuracy data
(see Table 4.8) the multivariate F value for the T1 x P1P2
interaction is 2.01. This value was significant at the .05
alpha level. Therefore the null hypothesis for the inter-
action between the time 1 effect and the test-retest
effect on the accuracy of rater responses was rejected and

the alternative hypothesis was supported.

Additional Findings

 

The effects tested by the hypotheses in this study
were a core subset of all effects in the overall design.
Along with the tested hypotheses, this design explored
several effects which contribute to the body of knowledge
concerning tape speed and its effect on the accuracy and

number of facial expression identification.

120

T1 Effect: The effect of the speed of the videotape
at time 1 on the overall performance of the raters. For
neither the accuracy nor the frequency data did this effect
reach significance. Thus the speed of the videotape at
time 1 does not affect overall performance of the
raters on eight of the dependent measures.

T2 Effect: The effect of the speed of the videotape
at time 2 on the overall performance of the raters. This
effect was significant in terms of the number of emotions
observed, but did not reach significance in the accuracy
of the identification. Thus although the speed of the video-
tape at time 2 improves the overall number of affects
observed, it does not influence the accuracy of the identi-
fication.

P1P2 Effect: The Test-Retest Effect. On both the

 

accuracy and frequency measures, this effect failed to

reach significance. Thus, regardless of speed of the video-
tape, a second viewing of the same stimulus does not improve
scores in either of the two measures.

T2*P1P2 Effect: Time 2 x Test-Retest Interaction.

 

The change in raters' performance in two viewings of the
stimulus tape was not significantly influenced by the
speed of the videotape at time 2.

T1*T2*P1P2 Effect: Time 1 x Time 2 x Test-Retest.
The change in raters' performance in two viewings of the

stimulus tape was not significantly influenced by the

121

interaction between the speed of the videotape at time 1

and the speed of the videotape at time 2.

Summary
Hypotheses were tested regarding 1) effect of tape

speed on overall performance and 2) the effect of tape
speed at time 1 on change in performance at time 2. The
dependent variables included the accuracy of the primary
rater responses and the number of rater responses per seg-
ment. Multivariate analyses of variance were performed
for these two types of data and in two of the four cases,
the null hypothesis was rejected at the .05 alpha level.
It was found that raters saw a significantly greater
number of facial expressions when viewing the tape in slow
motion (p‘<.05). However the accuracy of their ratings
of the slow motion tape was not significantly different
from their ratings of normal speed tape. If raters viewed
the videotape in slow motion first, their scores on the
second viewing showed significant improvement in the
accuracy of their primary response (p <.05). In terms of
the number of facial expressions recorded however, viewing
the slow motion videotape first was not associated with

improvement on the second trial. Because of the revision

of the design, the hypotheses regarding the effect of inter-

view condition on ratings was not tested. Several

additional findings which were tested in the design but

122

which were not included in the hypotheses were also

presented.

A summary of the study, the conclusions drawn and
a discussion of the results is presented in

Chapter V.

CHAPTER V
SUMMARY, CONCLUSIONS AND DISCUSSION

In this chapter, a summary of the study is presented
and the conclusions of the data analysis are examined.
Various aspects of the study are discussed including:
subject self report data, accuracy of rater judgments,
rater fatigue, test ceiling, limits of the study, and

finally, implications of the research.

Summary

This study investigated the ability of trained raters
to perceive spontaneous facial expressions on videotape
and to label them accurately in six affective categories.
The categories employed were: interest-excitement, enjoy-
ment-joy, surprise-startle, distress-anguish, anger-rage
and contempt-disgust. Also investigated were the number of
spontaneous facial expressions which raters perceived in
the stimulus materials.

Although facial expressions have been judged in many
studies, most of these have not employed videotape to
record spontaneous expression. Moreover, the effect of
viewing facial expressions in slow motion has not as yet
been extensively researched by most authors. The purpose
of the present study then is to test the effect of l) tape

123

\

124

speed and 2) test-retest procedures on both the number of
facial expressions observed and the accuracy of the percep-
tion of the primary affect in each of six emotional
categories.

A stimulus tape composed of a series of 5 second
videotaped segments was produced with each segment portray-
ing a recognizable type of facial expression.

A second stimulus tape was produced which was identi-
cal to the first except that each segment was slowed to
approximately one-sixth normal speed. The validity of the
emotion in each of the segments was determined by a panel
of expert judges.

Twenty-nine graduate students enrolled in courses in
the College of Education at Michigan State University were
trained to identify facial expressions for each of the six
affective categories. Raters participating in the study
attained a level of 80 percent accuracy in identifying
emotions from still photographs of facial expressions before
viewing the stimulus tapes.

Each group of raters viewed the stimulus tapes two
times. The speed of the tape (normal or slow motion) and
the order in which the speeds were presented was varied
among groups. Group one viewed the normal speed tape at
both time 1 and time 2. Group two viewed the normal speed
tape at time 1 and the slow motion tape at time 2. Group
three viewed the slow motion tape at time 1 and the normal

speed tape at time 2. Finally, group four viewed the slow

125

motion tape both at time 1 and time 2.

There were two dependent variables employed in the
study: accuracy of identification of emotions from facial
expressions, and the number of affects recorded for each
segment. A multivariate approach to analysis of variance
for repeated measures was employed to test several effects
on both the accuracy and frequency data for raters' over—
all performance and their change in performance over time.
The effects tested were 1) speed at time 1, 2) speed at
time 2, and 3) interaction between speeds time 1 and time 2.
Also tested was the effect of the test-retest procedure on
the accuracy of identification and the number of responses.

A Post Hoc Questionnaire was mailed to raters two
months after their participation in the study. The quesion-
naire explored three areas: the rater's state of mind while
viewing the tapes, the quality of the tapes themselves, and
the rater's evaluation of the overall study. The results
of the questionnaire were analyzed in the same type of a
multivariate analysis employed for both the accuracy and

number data mentioned previously.

Conclusions

 

Effect of Tape Speed on Identification of Emotion

 

The multivariate analysis of repeated measures for
the design over subjects led to the conclusion that there
was no significant difference between accuracy of raters'

judgments at normal speed and accuracy of judgments at the

126

slow motion Speed, at the .05 alpha level. Raters there-
fore did not perceive and label facial expressions of emo-
tion more accurately when they viewed the slow motion
videotape.

Effect of Tape Speed on the Number
of Facial Expressions Observed

 

 

The multivariate analysis of repeated measures for the
design over subjects led to the conclusion that there was a
significant difference between number of facial expressions
seen at normal speed and number of facial expressions seen in
slow motion. This effect was significant at the .05 alpha
level. The group who viewed the normal speed tape twice saw
the fewest number of affects. The next fewest number was
seen by the slow-slow group, and the greatest number seen
by both the slow-fast and the fast-slow groups. This rank
ordering is as predicted by the hypothesis with the excep-
tion of the poor performance by the slow-slow group.

The performance of the slow-slow group may be due to
the effect of rater fatigue. It is possible that viewing the
slow—motion tape two consecutive times had an effect on the
rater's willingness to record more than one response per
segment of the tape. This hypothesized effect in the slow-
slow group may have caused them to see fewer affects at
time 2 than they had at time 1.

Moreover, the results of the post-hoc questionnaire

indicate the tendency of all raters who saw the slow motion

127

tape at time 2 to rate the value of the study and the
quality of the tapes lower than other groups (p‘<.10).

This effect was especially apparent for the slow—slow

group who were generally more critical of the quality of the
tapes themselves and the value of the overall study.

The negative attitude of the slow-slow group may have
affected the motivation of the raters and thus explain their
relatively poor performance in the number of expressions
they observed in slow motion during the second viewing.

The Effect of Tape Speed on the Change ip
the Accuracypof Rater Performance Over Time

 

 

The multivariate analysis for the interaction between
the design over subjects and the design over measures led
to the conclusion that if the slow motion tape was viewed
first, there was a significant improvement in the accuracy
of later ratings. This difference was significant at the
.05 alpha level. Thus if raters were shown a slow motion
tape at time 1, the accuracy of their identifications of
facial expressions at time 2 would be significantly better
than if they had first viewed the normal speed tape.

The Effect of Tape Speed on the Change in
the Number of Affects Over Time

 

 

The multivariate analysis for the interaction
between the design over subjects and the design over
measures led to the conclusion that if subjects viewed the
slow motion tape first, there was not a significant improve-

ment in the number of affects observed at a later viewing.

128

Subjects who viewed the slow motion tape failed to see a
greater number of affects during the second viewing of the
tape, regardless of the speed of the second viewing. As
with the others, the alpha level for the test of this
effect was p = less than .05.

It is possible that a ceiling effect may have influ-
enced the test of this hypothesis. It has already been
shown that raters observed more affects in a slow motion
videotape regardless of whether it is seen first or second.
If raters observe the maximum number of affects possible
in slow motion at time 1, further improvement would there-
fore be impossible at time 2 even if the second tape was
also observed in slow motion. On the other hand, if the
normal speed is viewed first, then improvement at time 2 is
possible, especially if the slow motion tape is played. The
speed at which the tape is played at time 1 then dictates
the amount of improvement possible at time 2. Thus the
tendency for raters to reach a ceiling when rating slow

motion may have undermined the test of this hypothesis.

Discussion

 

Validity of Stimulus Segments

 

The type of emotion portrayed in each of the stimulus
segments was decided by a panel of expert judges. The
subject self-report data was not taken into account to

determine the emotion portrayed in each segment.

129

The decision to determine the validity of the segments
through the ratings of expert judges was made on the basis
of two arguments. First, subjects were not trained in the
use of the categories or in the identification of facial
expressions, even their own. A greater number of in-
accuracies could then be expected had the self report data
alone been employed to determine the validity of the
stimulus segments.

The second argument for use of segments agreed upon
by expert judges was that subjects may be motivated to
consciously or unconsciously defend against unacceptable
affect in their self reports. Identifying certain unplea-
sant emotional states in their own tapes during the recall
session would not only disclose these emotions to the exper-
imenter, but also insure a permanent record of their own
display of these emotions on videotape. Studies of per-
ceptual defense (Forrest and Lee, 1962; McGinnies, 1949;
Erikson, 1951; Postman, Bruner and McGinnies, 1948) suggest
that emotionally unacceptable stimuli is less likely to be
reported as perceived than other stimuli. Moreover, even if
an unacceptable emotional state were perceived, the subject
may be consciously resistant to disclosing this emotion to
the interviewer, who is a relative stranger. Thus, because
of the subjects' lack of training in identifying emotions,
and possible motivation to either conceal or repress

unacceptable affective states, the validity of the emotion

130

in the stimulus segments was determined by a panel of expert

judges and not by subject's self report.

Subject Self Report

 

The three affects that the videotaped subjects were
willing to identify most often in themselves during the
recall interviews were interest, joy, and distress. Anger
and contempt were labeled less often, and surprise, fear,
and shame were rarely labeled by subjects. These results
broadly correspond to Izard's cross cultural research on
affects (Izard, 1971). For the American culture, interest
and joy were most often cited as emotions people believed
they "understand the best." Moreover, distress was cited as
the negative emotion that people in America experience most
frequently. Fear and shame were the two emotions least
frequently reported by American subjects (Izard, 1971).
Thus the emotions elicited by the interviews reflected the
general cultural norms for experiencing and expressing
emotions in the American culture.

Striking differences were found between the emotions
that subjects recognized in themselves in the two interview
conditions. Subjects in the structured interview, who were
shown photographs and given labels by the experimenter,
reported a greater number of emotional states.in 6 out of
8 affective categories. The only category in which subjects
in the unstructured interview reported a greater number of

affects was that of distress. It is possible that

131

introducing the subjects to categories and labels during the
structured interview training had the effect of training
them to label more specifically their affective states
during the recall interview (especially in the negative
categories). Subjects who did not have this information may
have made poorer discriminations between the negative
emotional states and many have placed more emotions which
they experienced as unpleasant into the general category of
"distress." Thus the structured interview may have in-
volved a covert training effect which allowed subjects to
more accurately label their emotional states during the

later recall interview.

Accuracy of Rater Judgments

 

Upon reviewing the accuracy with which raters identi-
fied emotions from subjects' facial expressions, one clear
fact emerges: that raters were able to identify sponta-
neous facial expressions of emotions at levels of accuracy
which were significantly above chance. This accuracy of
identification was consistent both for normal speed and
slow motion videotapes. Moreover, although there were
differences in the levels of accuracy with which various
emotions were identified, all emotions studied were identi-
fied by raters at levels well above chance.

Raters were able to most accurately identify the
emotion of enjoyment-joy while the emotions of contempt—dis-

gust and anger-rage were identified least accurately of the

132

group. These findings generally correspond with results of
various cross-cultural studies in that enjoyment-joy is

the most accurately identified of all emotions (Ekman,

1972; Izard, 1971; Dickey and Knower, 1941). The negative
affects--distress-anguish, anger-rage, and contempt-disgust--
are generally found to be less accurately identified by
observing in many cultures (Ekman, 1972; Izard, 1971). It is
significant here that the studies mentioned have employed
still photographs of posed subjects for identification by
raters. The present study examines videotaped expressions
which are elicited spontaneously and this procedure has
produceciresults similar to the earlier studies.

Although raters were able to identify facial expres-
sions at accuracy levels which differed from chance, the
actual mean accuracy levels obtained for most emotions was
relatively low (interest = 59 percent, joy = 74 percent,
surprise = 66 percent, distress = 44 percent, anger = 41
percent and contempt = 35 percent). This low level of over-
all accuracy may be a product of the difficulty of the
rating task itself. Raters were allowed to View only a
brief segment of the videotape and the stimulus during this
time was an animated, complex series of muscular movements.
Identifying emotion from this type of stimulus is a more
difficult task than when still photographs freeze the face
in one position and allow the rater to closely study the
muscular configuration. Thus the complexity of the video—

taped stimulus is a possible explanation for the relatively

133

low levels of rater accuracy obtained in the present study.

Confusions In Identification

 

Tomkins (1965) discusses the concept of common con-
fusions in rater accuracy which are caused by underlying
personality variables altering the perceptual and decision-
making processes of the rater. Although this type of
notion is difficult to verify, the data in this study do
not support the idea of common confusions occurring across
raters. In three cases (distress, anger, and contempt)

a second affect (interest-excitement) was identified at
levels higher than chance. Only in one case (contempt-
disgust) was the affect of interest cited nearly as often
as the correct answer.

The fact that interest—excitement was the most fre-
quently cited "wrong answer" in five of the six emotional
categories pants to the possibility that the interest
category was often employed by raters as the equivalent of
a "neutral" or "don't know" response. The relatively low
level of facial activity in the expression of interst-
excitement and their relatively neutral connotations of the
emotional state may have disposed raters to employ this
category as an answer when they became confused or unsure.
These results suggest that rather than systematically
confusing certain affective states, raters tended to employ
the category of interest-excitement when they became con-

fused or unclear as to the emotion portrayed.

134

The Effect of Rater Fatigue

 

The overall difficulty of the rating task was reflec-
ted in the low overall accuracy of ratings and in the
inflated use of the interest-excitement category. This
difficulty may have also been reflected in a state of
fatigue which developed in raters who were asked to View the
slow motion tape. Although the complexity of the rating task
itself may have been eased through the use of slow motion,
the increased rating time necessitated by this condition may
have produced a state of fatigue which caused raters' per-
formance to deteriorate. The effect of rater fatigue was
evident in the slow-slow group where raters recorded fewer
affects at time 2 then they had at time 1. One rater in this
group stated, "I think there were people who just gave up.
After a while, we just started taking a stab in the dark."

Not only did the length of the slow motion rating task
seem to affect raters, but the quality of the tapes them-
selves appeared to be a factor. Because of the age of the
tapes and the poor quality of the slow motion reproduction
process, a noticable flicker was apparent on the slow motion
tape and this appeared to disturb many raters. In fact,
several raters who viewed the tape in slow motion at time 1
did not return for the second viewing.

It is not surprising then that on the Post Hoc Rater
Questionnaire, those individuals viewing the slow motion
tape at time 2 tended to rate the quality of the tapes, and

the value of the overall study lower then the other groups.

135

It is felt that this lower evaluation of the study reflects
a generally more negative attitude experienced by raters in
these groups, and that this attitude was induced primarily
by the experience of rating the slow motion videotape. It
is felt that this negative attitude may have affected the
performance of these groups in rating the facial expressions,
especially on the number of responses recorded in the slow-

slow group.

Test Ceiling Effect

 

Another facet of the slow motion results was that in
terms of the number of expressions observed, raters'
responses quickly reached a ceiling which dramatically af-
fected the data in this condition. For example, if a
finite number of affects are observable in a segment, and
if the rater watches this segment in slow motion at time 1,
then it is likely that the rater will observe all possible
affects in the segment. Then if at time 2, slow motion is
again observed, the fatigue effect would predict that
scores would deteriorate somewhat. If on the other hand,
the normal speed tape is observed at time 1 then improve-
ment is possible at time 2, especially if the slow motion
tape is played. The test ceiling would thus predict that if
a rater sees slow motion at time 1, he will either stay the
same or do less well (due to fatigue) at time 2; if normal
speed is shown at time 1, then the rater should stay the

same or improve at time 2. Indeed, reviewing the data

136

for these groups, the test ceiling effect and the fatigue
effect adequately explain the data.

The present research questions the notion that
exposure to facial expressions in slow motion facilitates
later recognition of minute rapidly-occurring facial
changes at normal speed. Were this the case, then the
slow-fast group would see nearly the same number of facial
expression in normal speed at time 2 as they did in slow
motion at time 1. This was not the case in the present
study. The lack of a training effect across speeds along
with the test ceiling in the slow motion condition may thus
have interacted to produce a significant effect for hypothe-

sis five which was in the opposite of the predicted directhmn

Limits of the Study

 

Eliciting Circumstances

In order to produce the stimulus materials, it was
necessary to videotape subjects in situations where they ex-
perienced a variety of affects. The interview conditions
were developed so that it would be possible to elicit the
eight emotions to be examined in the research. The purpose
of the interviews was to elicit through the use of individual
imagery the subjects' experience of each of these emotions,
thus allowing the subject to talk about the emotion, its
attributes and the situations which have elicited the

affect in his life.

137

The eliciting circumstances employed in this Study
seemed to produce facial expressions which were recognizable
to both subjects and raters as displays of most of the emo-
tions studied. It was found however, that positive emotions
or mild negative emotions were elicited in greater numbers
than the extremely negative emotions such as fear-terror
or shame-humiliation. The inability of the interview condi-
tions to elicit very unpleasant affects is believed to be
related to the level of trust between the subject and the
interviewer. Because of the nature of the experimental
situation, a low level of self disclosure, (especially on
the topic of unpleasant emotional states) may have caused
the low number of responses in the categories of fear-terror
and shame-humiliation.

The interview conditions were therefore found to be
effective in producing facial expressions by allowing the
subjects to produce eliciting circumstances which were rele-
vant to their own individual experiences. For very unplea-
sant emotional states such as fear and shame, however, the
low level of trust produced by the experimental situation
tended to inhibit the subjects' responses in these two

categories.

Subject Sample
The subject sample involved in the interview condie
tions consisted primarily of undergraduates in the College

of Education at Michigan State University. Subjects were

138

primarily white, and although the male-female ratio was

almost even (8 males and 7 females) the majority of seg-

ments chosen for viewing on the stimulus tape involved

female subjects (65 percent). The predominance of segments
with female subjects was especially evident in the categories
of surprise-startle and contempt-disgust. The higher number
of female segments chosen may reflect recent findings that fe-
males portray emotions more accurately and have a higher level
of facial activity than males (Buck, Miller and Caul, 1974).

All subjects participating in the study were also aware
of the videotape procedure and volunteered to participate in
the study. This willingness to participate may have indicated
a greater interest in emotions and a greater willingness to
disclose their emotions than would have been found in a more
normal population. Thus many aspects of the subject sample
in this research render it systematically nonrepresentative
of population in general.

Although the selectivity of the subject sample is a
shortcoming of the procedure, it is not a major concern in
view of the exploratory nature of the overall study. The
purpose of the present study is not to discover how widely
the phenomena tested exist in the world; the purpose is to

discover whether these phenomena in fact exist at all. It

 

is therefore important to test the hypotheses under optimal
conditions, and thus, segments were chosen which provided
the best material for the rating task. A high level of

judge agreement was therefore the sole criterion for the

139

Although the subject sample was highly selective, it
has been mentioned that there is strong evidence to suggest
that emotions are posed and judged in much the same manner
across cultures, across races, and across sexes (Ekman, 1972;
Izard, 1971; Black, 1969; Gitter, Black and Mostofsky,

1972). Thus assuming that the proper facial stimulus is
displayed by the subject, it is likely to be judged accurate-
ly regardless of these intervening variables. The highly
selective subject sample then need not exclude the possibi-
lity that similar results would be obtained if a broader

sample was employed.

Rater Sample

All raters involved in this study were participating
in graduate courses in the College of Education at Michigan
State University. Within this sample, however, several
distinct subgroups existed, the characteristics of which
may have altered the judgments made. Both the fast-fast
group and the slow-slow group were Elementary Education
teachers participating in Michigan State Extension Courses
at Jackson Community College. The fast-slow group consisted
of Masters candidates in Counseling at Michigan State, and
the slow—fast group consisted of Doctoral candidates in
College Counseling at Michigan State University. Thus the
level of previous training in the area of affect and the
motivation for learning about affective phenomena may have

differed among these three groups.

140

It is important to remember, however, that no matter
what the level of previous training, each group was trained
to identify affective facial expressions from still photo-
graphs at an average of 80 percent accuracy. This training
is the common denominator which overrides any differences in
previous level of training in the rater groups.

A measure on which the different characteristics of
the groups may have confounded the results is the frequency
of rater responses. Raters who were pursuing graduate
studies in counseling may have been more motivated to
recognize a larger number of emotions and thus may have
recorded a greater number of responses for each segment.
Since prior training was performed only on accuracy of
primary identification and not on recording the frequency
of facial expressions, the differences in rating groups may
have had an effect on the number of rater responses

recorded but not on the accuracy of the primary response.

Dependent Measures

Two dependent measures, accuracy and frequency, were
employed in this study, and these measures may have been
influenced differently by the motivation of raters. The
accuracy measure was primarily emphasized in the training
procedure, and thus raters were not specifically trained to
record more than one affect. Recording the type of primary
affect was the rater's first task, while the inclusion of

more than one category was optional. More motivated raters

141

may have recorded a substantially greater number of affects
than other less motivated raters, while the accuracy of
their primary responses might not have differed from those
of the less motivated group. The difference in the way
these two measures were presented to raters may thus have
caused the study to be affected by differing motivation in

rating groups.

Implications of the Study

 

Theoretical Implications

The primary importance of the present study is in
the fact that trained observers have been able to identify
spontaneous facial expressions recorded on videotape by
employing six emotional labels. The affects of interest-
excitement, enjoyment-joy, surprise-startle, distress-
anguish, anger-rage, and contempt-disgust are all consis-
tantly associated with specific facial patterns which can
be recorded on videotape and recognized by observers. The
findings lend further credence to the notion that facial
behavior is closely related to internal affective states
(Ekman, 1972, Izard, 1971). However, the relatively low
levels of accuracy obtained in the present study suggest
that identfiying spontaneous facial expressions from video-
tape is a more difficult task than identifying still photo-
graphs in the categories mentioned.

Although other studies (Frijda, 1953; Dusenburg and

Knawer, 1938) have shown films to be a superior presentation

 

142

mode, the dependent measure employed in these studies
did not involve the assignment of emotional labels to the
facial expressions. It is suggested here that the complex
muscular structure of the face facilitates subtle movements
and nuances of expression which are often idiosyncratic to
the subject and which have the effect of misleading or
confusing raters. The medium of videotape is more sus-
ceptible to the inclusion of these nuances, while still
photographs can screen out this extraneous information by
freezing a fraction of facial behavior. Thus although
facial expressions can be categorized accurately from video-
tapes, the task may be more complex than with still photo-
graphs due to the inclusion of information which may be
confusing or misleading to raters.

The second finding--that observers record a greater
number of emotions when viewing slow motion--is subject
to several interpretations. According to Haggard and Issacs
(1966), this effect might be due to the presence of new
facial expressions in slow motion which occur too rapidly to
be observed in normal speed. In other words the use of
slow motion changes certain aspects of the stimulus, and
this change is reflected in raters' observations. On the
other hand, it may be argued that the slow motion condition,
rather than disclosing a new set of facial expressions,
merely allows raters a longer time in which to carefully

identify the contents of the segments. In other words,

143

attributes of the tapes themselves are not altered, but the
drastically increased exposure time improves the quality of
raters' observations and discriminations. Unfortunately it
is impossible to separate lengthened exposure time from the
slow-motion condition while holding constant the visual
content of the stimulus.

A similar problem was encountered in studies of
perceptual defense and perceptual vigilance (Postman,

Bruner and McGinnies, 1949; Eriksen, 1951; McGinnies, 1949).
When an individual is exposed to value-laden stimuli at
drastically reduced exposure times, different recognition
thresholds are found. Thus the relative length of exposure
to a stimulus affects the recognition of the stimulus by
observers. Does the length of exposure influence the actual
perception of the event, or does it merely affect the recall
and response processes in the observer? Perhaps deciding on
what level the effect occurs—-in the stimulus or in the
observer-~is a moot point as long as the effect is in

fact apparent, and observers do in fact alter their ratings
when the stimulus is presented at different speeds.

The last finding of the study is interesting in light
of its theoretical implications. Observers who have already
viewed a videotape in slow motion do not record more emo-
tions when the tape is later replayed but do identify facial
expressions more accurately on the later tape. It seems
that merely one exposure to slow motion facial expression

does not sensitize observers to seeing a greater number of

144

expressions in normal speed. Perhaps as Guilford (1929)
found in overall levels of accuracy, more numerous
training trials (in this case slow motion trials) would
gradually increase the number of affects observed in normal
speed. In other words, one exposure to a slow motion video-
tape may not be sufficient to sensitize raters to a greater
number of expressions in normal speed. However, a greater
amount of exposure to slow motion would gradually sensitize
observers to a greater number of expressions.

Although observers who have viewed slow motion tapes
on the first trial do not later observe a greater number
of expressions, they do later identify facial expressions
more accurately. Thus, it is possible that the slow motion
conditions sensitize raters to the meanings of nuances of
facial expressions thus allowing them to identify these ex-
pressions more accurately later on. Since the smaller
facial muscles move more quickly than larger facial
muscles (Landis and Hunt, 1939) exposure to slow motion
may sensitize raters to these more subtle muscular move-
ments. Thus the effect of slow motion may be not to sensi-
tize raters to a greater number of facial expressions, but
by accentuating minute muscular responses, make it possible
for raters to more clearly interpret subtle facial movement,

thus improving the accuragy_of their later perceptions. The

 

training effect which occurs if slow motion is seen first
is especially striking when one considers the negative

attitudes of the same rating groups and the rater fatigue

145

effect. Although some raters were critical of the study

and tired of watching the tapes, the accuracy of their
responses nonetheless improved at later viewings. Raters
who did not see slow motion first may have been less
fatigued and had more positive feelings about the study, yet

experienced no training effect.

Clinical Implications

Many of the ideas that follow are not conclusions which
have been definitively confirmed by the data in this study
but rather have been gleaned from clinical observations
during the study and perceptions of the possible clinical
utility of this research.

The affects studied in this research seemed to be
imbued with distinctive characteristics and qualities which
emerged when trying to elicit facial expressions during the
interviews. Interest and joy, of all of the emotions
studied, were the easiest for people to experience. The
relative pleasantness and social acceptability of these
emotions doubtless contribute to their popularity. Surprise,
although it was not viewed as an unpleasant emotion, in it-
self is of such intensity that it may be associated with
fearful stimuli and loss of control on the part of the
individual. Thus it is a relatively rare emotion, but has
a distinctive facial response when elicited.

In the area of negative emotions, distress was by far

the one most frequently identified by both subjects and

146

raters. The relatively less intense nature of this emotion
coupled with a rather high level of social acceptability,
are both factors contributing to its popularity.

The emotions of anger and disgust are neither as
easily owned by individuals nor as easily identified through
facial expressions. Unlike distress, however, they are
intense emotions and the force of intensity is usually
directed toward the environment in some way. Thus both of
these emotions carry with them an aura of potency and
strength which to some extent offsets the effects of their
social undesireability.

The emotions of fear and shame have neither of the
advantages of pleasantness nor potency. Neither of these
affects are owned readily by subjects or identified easily
on the face. One possible consequence of the emotion of
fear (when experienced in a mild form) is to freeze the
facial expressions so that nothing is observable on the
face except a slight widening of the eyes and a fixation of
the gaze on the feared object. Similarly when the affect
of shame is experienced, recognition is often inhibited
because the person looks away, looks down, or physically
moves away. These effects--because of their intense, un-
pleasant nature, and the implicit admission of vulnerability
associated with them--are kept secret both from observers
and, many times, even from the one experiencing the

emotional state.

147

The second type of clinical implication derived from
this research involves the way in which affects are commu-
nicated to the world at large. Because of the complexity
of the facial structure, emotions can both be easily commu-
nicated and readily concealed. This research suggests that
emotions are in fact communicated spontaneously through the
face, and that people have some recognition of this
phenomenon as it occurs. Moreover, people can heighten
their ability to identify spontaneous emotions with some
training and exposure to the face in slow motion.

Although the face has the ability to communicate
emotions effectively, this communication is often under
conscious control and is therefore subject to alteration
and concealment. The realtively low incidence of recogniz-
able facial expressions in many hours of videotaped inter-
viewing suggests that although the face can communicate
effectively, these communications occur infrequently.

When emotions do occur, their facial displays are
sometimes shortened, masked or blurred by extraneous muscu~
lar movements. Thus, judging spontaneous emotions under
live conditions from the human face is possible under cer-
tain conditions, but it usually is a subtle and difficult
task. This task can be facilitated by providing training
in the use of emotional labels and exposing the face in
slow motion to clearly uncover more subtle nuances of

facial movement.

148

Implications for Future Research

The conclusions of this research are subject to several
limitations because of the selectivity of the subject and
rater groups. The limits of the generalizability of the study
are especially apparent when the large number of female seg-
ments is taken into account. A more balanced male-female
ratio for segments employed would enhance future research
efforts. It sould also be useful to compare the responses of
trained and untrained raters to determine whether videotapes
of spontaneous facial expressions can be recognized by indi-
viduals without prior learning.

The eliciting circumstances (the interview conditions)
could be improved in future research. Providing a more
emotion-arousing series of eliciting circumstances would
improve both the number and quality of facial expressions
elicited. This is especially important for the affects of
fear and shame, which were seldom elicited from the inter-
view procedure. The continued use of expert raters to vali-
date segments will also maximize the number of useable
segments.

The presentation of slow motion videotape of facial
expressions should be shortened to lessen the effect of
rater fatigue. This could be accomplished by speeding up
the segments slightly, using fewer segments in slow motion,
or presenting the slow motion condition in several

administrations.

149

An alteration in the dependent measure would also be
useful in future research. Type of primary emotions and
number of emotions should be recorded separately so that
the motivational factors will be approximately equal for
each of the two measures. Rather than simple listing, a
format in which the number of affects seen would be recorded
in one column and the type of primary affect in another
column should be used. This would clearly separate the
two types of responses and this would eliminate many of the
problems in the area of dependent measures.

Many other research topics would neatly contribute
to findings of the present study. A comparison of verbal
and nonverbal affect recognition utilizing the category
approach presented here and employing videotapes with and
without sound, would be an extremely useful contribution.
Moreover, the relationship between subject and rater
personality variables in posing and judging it both slow
motion and normal speed would answer many intriguing
questions.

The comparison of posed and spontaneous expressions
in slow motion and normal speed would partially answer the
question of whether posed expressions are different in
.recognizable ways from expressions which are spontaneously
elicited. Also it is possible that some affects are
:recognized more accurately in slow motion while others are

rmmre identifiable in normal speed. This area has yet to

150

be explored in any systematic way.

A possible clinical use for the findings of the
present study is to develop a training program for pro-
fessional and paraprofessional counselors which would
utilize slow motion videotapes to teach them to accurately
recognize emotions from facial expressions. The present
results suggest that if counselors are exposed to facial
expressions in slow motion they will later be able to

identify emotions in their clients with greater accuracy.

REFERENCES

REF ERENCES

Abelson, R.P., and Sermat U. Multidimensional scaling of
facial expressions. Journal of Experimental
Psychology, 63, 1962, pp. 546-554.

 

Arnold, M.B. Emotion and Personality. New York: Columbia
University Press, 1960.

 

Bartlett, E.S. A Dimensional and Typological Investigation
of the Subjective Experience of Emdtion. Ph.D.
Dissertation, Vanderbilt University, 1971.

Birdwhistle, R.C. The Kinesic Level tithe investigation of
the emotions. In P.H. Knapp (Ed.) The Expressions
of Emotions in Man, Part II, Ch. 7. New York: Inter-
national University Press, 1963.

 

Black, H. "Face and Sex Factors Influencing the Correct
and Erroneous Perception of Emotion." Proceedings
of the 77th annual APA convention, 1969, pp. 363-364.

Black, Perry. Physiological Correlates of Emotion. New
York: Academic Press, 1970.

Boring, E.G., and Titchner, E.B. A model for the demon-
stration of facial expression. American Journal of
Psychology, 34 (4), 1923, pp. 471-489.

 

Boucher, J.D. Facial displays of fear, sadness, and pain.
Percept. and Mot. Skills, 28, 1969, 239-242.

 

Bruner, J.S., and Tagiuri, R. The perception of people.
In G. Lindsey (Ed.), Handbook of Social Psychology,
Vol. 2. Reading Mass, 1954, pp. 634¥654:

jBuck, Ross; Miller, Robert E.; and Caul, William F.
Sex, personality and physiological variables in the
communication of affect via facial expression.
J. Pers. Soc. Psy. 30 (4), October 1974, pp. 587-96.

 

ZBuck, R.; Savin, V.J.; Miller, R.E.; Caul W.F. "Non-
Verbal Communication of Affect in Humans. Proceedings,
77th Annual APA convention, 1969, pp. 367-368.

151

152

Buck, R; Savin, V.J.; Miller, R.E.; Caul, W.F. Communica-
tion of affect through facial expression in humans.
J. Pers. Soc. Psy.,(23), 1972, pp. 362-371.

 

Buhler, C., and Hetzer, H. Das erste Verstandnis fur
Ausdruck in ersten Lebensjahr. Epitschrift fur
Psychologie, 107, 1928, 50-61.

 

Burns, K.L., and Beier, E.G. Significance of vocal and
visual channels in the decoding of emotional meaning.
J. Commun. 23 (1), March 1973, pp. 118-30.

 

Buzby, P. Interpretation of facial expressions. Am. J. of
Psy. 35, 1924, pp. 602-604.

Cannon, W.B. Bodily Changes in Pain,Hunger, Fear and Rage.
New York: Appleton, 1929.

Carlson, Rae, and Levy, Nissim. Studies of Jungian typology:
I. memory, social perception, and social action.
Journ. of Personality 41(4):559-576, 1973.

 

Chambers, J.L. Trait judgment of photographs and adjust-
ment of college students. J. Consult. Psy, 25 (5),
1961, pp. 433-35.

Cohen, Bertram D., and Rau, John H. The non-verbal affect
measure. Nonverbal technique for measuring affect
using facial expression photographs as stimuli.
Journ. of Consultingand Clinical Psychology 38(3);
449-451, 1972.

Coleman, J.C. Facial expressions of emotion. Psych.
Monograph, §;, 1949 (1, whole, No. 296).

 

Cuceloglu, Dogan. Facial Code in Affective Communication
Comparative Group Studies (Small Group Behavior), Vol
3(4), 1972, 395-408.

ZDarwin, Charles. The Expression of the Emotions in Man and
Animals. University of Chicago Press, Chicago and
London.

llavis, R.C. The specifity of facial expressions. Journal
of General Psychology, 10, 1934 42-58.

Imavitz, Joel, R. The Language of Emotion. New York:
Academic Press, 1969.

153

Dickey, E.C., and Knower, F.H. A note on some ethnological
differences in recognition of simulated expression of
emotion. Am. J. of Sociology, 47, 1941, 190-193.

 

Dougherty, F.E.; Bartlett, E.S.; Izard, C.E. Responses of
schizophrenics to expressions of the fundamental
emotions. J. Clin. ng. 30(3), 243-6.

 

Drag, R.M., and Shaw, M.E. Factors influencing the
communication of emotional intent by facial expression.
Psychometric Sci. 8, 1967, 137-8.

 

Duchenne, G.B. Physiology of Motion Demonstrated py Means
of Electrical Stimulation and Clinical Observation
and Applied to the Study of Paralysis and Deformities.
Translated and edited by Emanual Kaplan, M.D., J.B.
Lippincott, and Co., 1949.

Duchenne, Mechanisme de la physionomie humaine ou
Anaylse electro-physiologique de la expressiones
des passiones, 1862.

Duffy, E. An explanation of "emotional" phenomena without
the use of the concept "emotion." J. General ng.
25, 1941, 283-93.

Dunlap, K. The role of eye-muscles and mouth muscles in
the expression of the emotions. Genetic Psy.
Monograph 3, 1927, 424-435.

 

Dusenbury, P., and Knower, F.H. Experimental studies in the
symbolism of action and voice, I. A study of the
specifity of meaning in facial expression. Quarterly
J. of Speech 24, 1938, 424-435.

 

Eibl-Eibesfeldt, K. Love and Hate: A Natural
History of Behavior Patterns. Verlag, Munich: R.
Piper and Co., 1970.

Ekman, P. Body position, facial expression and nonverbal
behavior during interviews. J. Abn. and Soc. Psy.
68, 1964, 295-301.

Ekman, P. Differential communication of affect by head and
body cues. J. Pers. and Soc. Psy. 2 (5), 1965, 725-

735.

Ekman, P. "Universals and Cultural Differences in Facial
Expressions of Emotion." Nebraska Symposium on
Motivation, Vol. 19, 1971, 207-283.

154

Ekman, P. "Universals and Cultural Differences in Facial
Expression." In J.K. Cole (Ed.) Nebraska Symposium
on Motivation, 1971. Lincoln Nebraska: University
of Nebraska Press, 1972.

 

 

Ekman, P. Universal facial expressions in emotion.
Studia Psyohologica, 15 (2), 1973, 140-147.

 

Ekman, P., and Friesen W.V. Head and body cues in the
judgment of emotion: A reformulation. Percept. and
Mot. Skills 24, 1967, 711-724 (a).

 

 

Ekman, P., and Friesen, W.V. Nonverbal leakage and clues to
deception. Psychiatry 32, 1969,88-106.

 

Ekman P., and Friesen, W.V. The repetition of nonverbal
behavior--categories, origins, usage and coding.
Semiotica, l, 1969, 49-98.

 

Ekman, P., and Friesen, W.V. Constants across cultures
in the face and emotion. J. Pers. and Soc. Psy.
17 (2), 1971, 124-129.

 

Ekman, P., and Friesen, W. Detecting deception from the
body or face. J. Pers. and Soc. Psy. 29 (3), 1974
288-298.

 

Ekman P., and Friesen, W.V. Unmasking the Face: A Guide
to Recpgnizing Emotions from Facial Expressions.
Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1975.

 

Ekman, P.; Friesen, W.; and Ellsworth, Pheobe. Emotion
in the Human Face. Pergamon Press, 1972.

 

Ekman, P.; Friesen, W.; and Tomkins. Facial affect scoring
technique (FAST): A first validity study. Semiotica
3' 1971' 37-58.

 

Ekman, P.; Sorenson, E.R.; and Friesen, W.V. Pan-
cultural elements in facial displays of emotion.
Science Ap. Psy. 164, 1969, 86-88.

 

Engen, T.; Levy, N.; and Schlosberg, H. The dimensional
analysis of a new series of facial expressions. J.
Exp. Psy. 55, 1958, 454- 458.

 

Eriksen, C.A. Perceptual defense as a function of unaccept-
able needs. J. Abn. Soc. Psy. 46, 1951, 557-564.

 

155

Fair, P.C.; Schwartz, G.B.; Friedman, N.J.; Greenberg, P.S.;
Klerman, G.C.; and Gardner, E.A. Facial Expression
and Emotion: An Electromyographic Study. Unpub-
lished paper read at Eastern Psych. Assn., Philadel-
phia, Pa., April 1974.

Feleky. The expression of the emotions. Ppych. Review 21,
1914, 33-34.

 

Forrest, D.W., and Lee, S.G. Mechanisms of defense and
readiness in perception and recall. Psychological
Monographs, No. 523, Vol. 76, No. 4, 1962.

 

 

Freedman, D.G. Smiling in blind infants and the issue of
innate versus acquired. J. of Child Psychology and
Psychiatry. Vol. 5, 171-184.

 

 

Frijda, N.H. The understanding of facial expression of
emotion. Acta Psychologica, 9, 1953, 294-362.

 

Frijda, N.H. Facial expression and situational cues.
J. Abn. Soc. Psychol. 57, 1958, 149-154.

 

Frijda, N.H. and Philipszoon, E. Dimensions of recognition
of expression. J. Abn. Soc. Psyphol. 66, 1963, 45-51.

 

Frois-Wittman, J. The judgment of facial expression.
J. Exp. Psychol. 13, 1930, 113-151.

 

Fulcher, J.S. "Voluntary" facial expression in blind and
seeing children. Archives of Psychol. 38, No. 272,
1943.

 

Gates, G.S. A test for ability to interpret facial expres-
sion. Psychol. Bull. 22, 1925, 120.

 

Gitter, George,; Black, H.; and Mostofsty, D. -Race and
sex in the perception of emotionl J. of Soc. Issues.
28 (4), Fall 1972, 63-78.

 

Gladstone, W.H. A multidimensional study of facial
expressions of emotion. Australian J. of Psychol.
14, 1962, 95-100.

 

Glass, G., and Stanley, J. Statistical Methods in Education
and Psychology. Englewood Cliffs, N.J.: Prentice-
Hall, Inc., 1970.

 

Goodenough, F.C. The expression of emotions in infancy.
child Development. 2, 1931, 96-101.

 

156

Goodenough, F.C. Expression of the emotions in a blind-
deaf child. J. Abn. Soc. Psychol. 27, 1932-3, 328-
333.

 

Goodenough, F.C., and Tinker, M.A. The relative potency
of facial expression and verbal description of
studies in the judgment of emotion. J. Comp. Psyphol.
13, 1931, 365-370.

 

Goren, C.; Sarty, M.; and Wu, P. Visual following and
pattern discrimination of face-like stimuli by
newborn infants. Pediatrics 56, Oct. 1975, 544-9.

 

Gregory, W.K. Our Face from Fish to Man. New York:
Putnam, 1929.

 

Gubar, G. Recognition of Human Facial Expressions Judged
Live in a Laboratory Setting. Ph.D. Dissertation
Rutgers University, 1966, p. 313.

Guilford, J.P. An experiment in learning to read facial
expressions. J. Abn. Soc. Psychol. 24, 1929, 191-202.

 

Haggard, E.A. Professor of Psychology, Department of
Psychiatry, University of Illinois, College of
Medicine, Personal Communication, October 21, 1974.

Haggard, E., and nxmtS,Ft Micromomentary facial expressions
as indicators of ego mechanisms in psychotherapy.
In Methods of Research in Psychotherapy, Gottschalk,
L.A. and Auerbach, A.H. (Eds.). New York: Appleton
Century Croft, 1966, 154-165.

 

Hamilton, M.C. Imitation of facial expression of emotion.
J. Psychol. 80 (2), March 1972, 345-50.

 

Harrison, R.P. Pictic Analysis: Toward a Vocabulary and
Syntax for the Pictoral Code; With Research on
Facial Communication. Ph.D. Dissertation, Michigan
State University.

Hoshino, Kikuzo. The ability to identify affective
meanings of facial expressions at successive age
levels. Japanese J. of Ed. Psychol. 17 (2), 1969,
90-101.

 

Howell, R.J., and Jorgenson, E.C. Accuracy of judging
emotional behavior in a natural setting--A replication.
J. of Soc., Psychol. 31 (2), 1970, 269-270.

 

Howes, P.H., and Solomon, P.C. A note on McGuinnies'
'Emotionality and Perceptual Defense.‘ Psych. Review
57, 1950, 229-234.

 

157

Huber, E. Evolution of Facial Musculature and Facial
Expression. Baltimore: Johns Hopkins University
Press, 1931.

 

Izard, Carol E. Face of Emotion. New York: Appelton-
Century-Crofts, Merideth Corporation, 1971.

James, W. What is an emotion? Mind 9, 1884, 188-205.
Jenness, A. The effects of teaching subjects in the

recognition of facial expressions. J. Gen. Psychol.
7, 1932, 163-178.

 

Jorgenson, E.; Clay; and Howell, Robert, J. Judging
unposed emotional behavior. Psychotherapy--Theory
Res. Pract. 6 (3), 1969, 161-65.

 

 

Kanner, L. Judging emotions from facial expressions.
Psychol. Monograph 41, 1931, (3, Whole No. 186).

Kozel, Nicholas. Perception of Emotion: Race of Expressor,
Sex of Perceiver, and Mode of Presentation. Proceed-
ings, 77th Annual Convention of the APA., Vol. 77,
1969, 39-41.

Kreutzer, M.A. and Charlesworth, W.R. Infants Reactions
to Different Expressions of Emotions. Paper presented
at the Society of Research in Child Development,
Philadelphia, Pa., March, 1973.

Kwint, C. Ontoqeny of mobility of the face.
Child Development 5, 1934, 1-12.

 

Landis, C. Studies of emotional reactions 1: A preliminary
study of facial expressions. J. Exp. Psychol.
7, 1924, 325-341.

 

Landis, C. Studies of emotional reactions: II. General
behavior and facial expression. Journal Compar.
Psychol. (4), 1924, 447-509.

 

Landis, C. The interpretation of facial expression of
emotion. Journal General Psychol. 2, 1929, 59-72.

 

Landis, C. and Hunt, W.A. Magnification of time as a
research technique in the study of behavior.
Science, 85, 1937, 384-85.

Landis, C., and Hunt, W.A. The Startle Pattern. New York:
Farran, 19 39 .

 

158

Langfeld, A.S. The judgment of emotions from facial expres-
sions. J. Abn. Soc. Psychol. 13, 1918, 172-84.

 

Levitt, E.A. The relationship between abilities to express
emotional meaning vocally and facially. In J.R.
Davitz (Ed.), The Communication of Emotional Meaning
N.Y.: McGraw-Hill, 1964, 87-100.

Lindsey, D.B. The Role of reticulo-thalamocortical
systems in emotion. In Black (Ed.) Physiological
Correlates of Emotion. N.Y.: Academic Press,1970.

 

 

LoCastro, Joseph 8. Judgment of emotional communication in
the facial-vocal-verbal channels. Dissertation
Abstracts International, Vol. 33 (1_B), 1972, 443.

Lanzetta, J.T., and Kleck R. Encoding and Decoding of
facial affect in humans. J. Pers. Soc. Psy. 16 (l).
1970, 12-19.

McGinnies, E. Emotionality and perceptual defense.
Psych. Review 56, 1949, 244-251.

 

Munn, N.C. The effect of knowledge of the situation upon
judgment of emotion from facial expressions. J. Abn.
and Soc. Psy., 35, 1940, 324-338.

 

Murphy, 6.; Murphy, C.E.; and Newcomb, T.M. Experimental
Social Psychology. New York: Harper, 1937.

 

Postman, L.; Bruner, J.; and McGinnies, E. Personal values
as selective factors in Perception. J._Abn. Soc.
Psy. 43, 1948, 142—54.

 

Royal, D.C. A Multidimensional Analysis of Perception of
Emotion From Schematic Facial Expressions, Ph.D.
Dissertation, University of Michigan, 1959.

Schiffenbaur, A. Effect of observers emotional state on
judgments of emotional state of others. J. Pers. Soc.
Psy. 30 (l), 1974, 31-5.

Schlosberg. IXscale for the judgment of facial expressions.
J. Exp. PEX” 29' 1941' 497-510.

 

Schlosberg, H. Perception of facial expression in terms
of two dimensions. J. of Exp. Psychol. 44, 1952,
229-237.

Schlosberg, H. Three dimensions of emotion. Psychol.
Review, 61, 1954, 81-88.

159

Schwartz, G.B.; Fair, P.C.; Greenberg, P.S.; Friedman, M.J.;
and Klerman, G.C. Facial Electromyography in the
Assessment of Emotion. Unpublished paper read to
the Society for Psychophysiological Research. Oct.
1973. .

Schwartz, G.E.; Fair, L.; Greenberg, P.S.; Mandel, M.R.; and
Klerman, G.L. Facial Expression and Emotion: An
Electromyographic Study. Unpublished paper read at
meeting of American Somatic Society., Philadelphia,
Pa., 1974.

Schwartz, 6.; Fair, P.; Salt, P.; Mandel, M.; and Klerman,
G. Facial muscular patterning to affective imagery
in depressed and non-depressed subjects. Science,
192, April 30, 1971.

 

Secord, B.E.; Dukes, W.F. and Devan, W. Personalities in
faces: I: An experiment in social perceiving.
Gen. Psy. Monographs, 49, 1954, 231-279.

 

Secord, B.E.; Dukes, W.F. and Devan, W. Personalities in
faces III: A cross cultural comparison of physio-
gonomy and personality in faces. J. Soc. Psychol.,
43, 1956, 283-288.

 

Shannon, Anna Marian. Differences Between Depressives and
SchiZOphrenics in the Recognition of Facial Expres-
sion of Emotion (D.N.S. Dissertation). Dissertation
Abstracts International. Ann Arbor, MI., Univ. M-
Films, No. 71-27067.

Thompson, J. Development of facial expression in blind and
seeing children. Archives of Psychol., 37, 1941, No.
264.

 

Thompson, P.F., and Meltzer, C. Communication of emotional
intent by facial expression. J. Abn. Soc. Psychol.,

 

Tomkins, Sylvan, S. Affect Imagery and Consciousness Vol. I,
The Positive Affects. New York: Springer Publishing
Co., 1962.

 

Tomkins, Sylvan S. Affect Imagery and Consciousness Vol. II,
The Negative Affects. New York: Springer PubTishing
Co., 1963.

 

Tomkins, 8.8. Introduction to Affect Symposium. Proceedings,
of the 72nd Annual Convention of the American
Psychological Association, 1964.

160

Tomkins, 8.8. Introduction to Affect Symposium, APA, 1964.

Tomkins, Sylvan S., and Izard, Carol E. Affect, Cognition
and Personality Empirical Studies, New York:
Springer Publishing Co., 1965.

Tomkins, 8.5., and McCarter. What and where are the primary
affects-n-Some evidence for a theory. Perceptual
and Motor Skills, 18, 1964, 119-158. ‘*

 

 

Triandis, H.C., and Lambert, W.W. A restatement and test
of Schlosberg's theory of emotion with two kinds of
subjects from Greece. J. Abn. Soc. Psychol. 56
1958, 321-328.

Vinake, W.B. The judgment of facial expressions by three
national racial groups in Hawaii, I. Caucasian faces.
J. of Pers. 17, 1949, 407-429.

 

Vinake, W.B. The judgment of facial expressions by three
national-racial groups in Hawaii, II. Oriental faces.
J. of Soc. Psychol., 41, 1955, 184-195.

 

Watson, S.G. Judgment of emotion from facial and contextual
cue combinations. J. Pers. Soc. Psy, 24 (3).
1972, 334-42.

 

Wenger, M.A. "Emotions as Visceral Action: An Extension
of Lange's Theory." In M.L. Reyment (Ed.), Feelings
and Emotions: The Moosehart Symposium, New York:
McGraw-Hill, 1950.

 

 

Wilcox, B.M. Visual preferences of human infants for
representations of the human face. J. Exp. Child
Psychol. 7, 1969, 10-20.

 

Wilcox, B., and Clayton, F. Infant visual fixation on
motion picutres of the human face. J. Exp. Child
Psy. 6 (1), 1968, 22-32.

 

Woodworth, R.S. EbcperimentalPsychology. New York: Henry
Holt, 1938.

 

Young, Paul Thomas. Emotion in Man and Animal Its Nature
and Dynamic Basis, 2nd ed. Robert E. Krieger Pub. Co.
1973.

 

 

Zlatchin, Carl. Accuracy and inaccuracy in the recognition
of facial affects and the relationship to personality.
Diss. Abstr. Int. #74-13400.

APPENDICES

 

11111))

APPENDIX A

SIGN UP SHEET FOR PARTICIPATION IN RESEARCH

ON AFFECTIVE COMMUNICATION

APPENDIX A
SIGN UP SHEET FOR PARTICIPATION IN RESEARCH

As part of my dissertation research, I am making a series
of videotapes of people discussing their feelings and
emotions. The participants will be involved in a half-hour
interview which will be videotaped, followed by a half-
hour recall session during Which the videotape will be
reviewed and processed. In the past this has proved to be
an interesting and rewarding experience for those who wish
to participate. Times for interviews will be available
during the period of August 7 until August 15. If you are
interested, please sign up below and you will be contacted
regarding specific times. Thanks for your participation.

Dave Inman

NAME ADDRESS PHONE

 

 

 

 

 

 

 

 

 

 

 

161

APPENDIX B

COMPOSITION OF SUBJECT GROUP

162

APPENDIX B

COMPOSITION OF SUBJECT GROUP

 

 

 

 

 

 

 

 

Subject Sex Age Academic Interview*
Level
1 M 22 Grad 1
2 F 19 Jr 1
3 F 20 -- 1
4 F 30 Grad 1
5 M 21 Sr 1
6 F 20 Jr 1
7 M 25 Grad 1
8 F 20 Jr 1
9 M 28 Sr 1
10 F 47 Jr 2
11 F 21 Sr 2
13 M 38 Jr 2
14 M 26 Jr 2
15 F 34 Jr 2
16 F 31 Grad 2
17 F 31 Grad 2
18 M 27 Grad 2
* 1 = Structured
2 = Unstructured
Sex Raw Percent
Score
Male 7 .41
Female 10 .59
Total 17 1.00
Age Average = 27
Minimum = 19
Maximum = 47
Range = 28

APPENDIX C

SUBJECT RELEASE FORM FOR STUDY ON

AFFECTIVE COMMUNICATION

163

. APPENDIX c

SUBJECT RELEASE FORM FOR STUDY ON AFFECTIVE COMMUNICATION

I understand that this is a research project investi-
gating variables which are thought to be relevant to the
process of affective communication. I hereby agree to
permit films known as videotape recordings of my participa-
tion to be prepared. I understand that these films are
being produced solely for the purposes of research for this
study and that my confidentiality will be strictly maintained.

I understand that at any time during this study I am
free to discontinue my participation totally. I am also
free to disallow preparation of videotape recordings of my
responses, if,I so desire.

I have voluntarily agreed to participate in this study,
and have not been offered nor expect any remuneration for
my participation.

 

Signature of Participant

 

Date

 

Witness

APPENDIX D

SUBJECT INFORMATION FORM

164

APPENDIX D

SUBJECT INFORMATION FORM

 

 

 

 

 

Subject Number: Sex: Male‘
Female
Name: Age:
Address:
Street
City State
Phone: Major

 

 

Academic Status

Freshman
Sophomore

Junior

Senior

Graduate Student

Q.

I

APPENDIX E

SUBJECT INSTRUCTIONS:

STRUCTURED INTERVIEW CONDITION

165

APPENDIX E

SUBJECT INSTRUCTIONS: STRUCTURED INTERVIEW CONDITION

This is a study of how various emotions or affects
are experienced. We are interested in how various emotions
are identified, how they are stimulated by various situa-
tions, and how they are experienced, or felt, by the person.
You will be shown a set of photographs depicting various
facial expressions of emotion. Each photograph is
labeled with a specific emotional phrase. After the
photograph is displayed, you will be asked to describe
certain aspects of the emotion in the photograph as you
experience it. We are interested in such aspects as:
where you feel the emotion occurring in your body, what it
is like to experience that affect, and finally, what situa-
tions in your life stimulate or "trigger" that particular
affect. Following this procedure, a series of eight emo-
tional states will be explored. The emotions we are
interested in are: interest-excitement, happiness-joy, sur-
prise-startle, fear-terror, anger-rage, contempt-disgust,
distress-anguish, and shame-humiliation. Do you have any

questions?

APPENDIX G

SUBJECT INSTRUCTIONS:

RECALL INTERVIEW

167

APPENDIX G

SUBJECT INSTRUCTIONS: RECALL INTERVIEW

During the next few minutes you will be viewing a
videotape which was made of your previous conversation.
As you view the tape we would like you to stop it at points
where you see yourself as experiencing an emotion. Since
the tape will be played without sound, it will be necessary
for you to pay attention to the non-verbal indications of
emotions. In this way we hope to isolate points on the
videotape where you think your face shows that you were
feeling an emotion. At the same time you will probably
become much better at identifying how you as an individual
communicate emotions non-verbally. To identify the affects
on the videotape we will be using the following emotional
labels: interest-excitement, happiness-joy, surprise,-star-
tle, fear-terror, anger-rage, contempt-disgust, distress-

anguish, and shame-humiliation. Do you have any questions?

APPENDIX H

FORM USED TO RECORD INFORMATION FROM

RECALL INTERVIEW

168

APPENDIX H
FORM USED TO RECORD INFORMATION FROM RECALL INTERVIEW

Subject Number ‘ Tape Number

 

 

Time

urpr se-
Startle
Anger-
D stress-
An uish
Contempt-
Dis ust

S ame-
Humiliation

Excitement
Ra e

Interest-

I
U)
U)
(l)
S:
Q.
0..
(U
.11

Fear-
Terror

JO

\OCDNO‘UIohWNI-J

I—‘
O
I

11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.

 

APPENDIX I

COMPOSITION OF STIMULUS TAPE SEGMENTS

 

169

 

pHo> ma
2 up 5H
m on ma
2 mm ma
m mm ea
m qm ma
2 mm . NH
2 mm Ha
m Hm OH
H mm M ono> o~.ma
z ov h m Hoa ma pflo> o~.ma.ma
m mm m z vm ha m ooa ha
m Ha m z om ma m cm ma
m 0H v 2 pm ma m mm ma
m m m 2 mm wa m Hm ea
m m m 2 ¢m ma m cm ma
m H H z mv ma 2 mp NH
amazoqc z aw Ha m an ad
\mmmuumfio m hm 0H m we 0H
m mm m h mm m
Uwo> m~m m mm m m do w
m mm b m 5m 5 m mm b
m no 0 2 mm o 2 be m
m mg m z om m 2 «v m
m mm w m ma g 2 mm v
m Hm m m m m 2 mm m
m an m z m m 2 ma m
2 am A m m H h w a
maunmum hon unmamuwuxm
\omwumhnm \ucmamoncm 4 \ummuousH
nonadz Hmnfisz nonapz
xmm ucmfimmm anomoumo xwm ucmsmom anomwumo xom usmfivom whomwumo

 

 

 

 

mfizgwmm wads mDADZHBm .mO ZOHBHmOmZOU

H xHszmmd

170

 

 

oao> ma.ha
2 pm ma
pfio> oa.ma.¢a 2 mm ma
m mm ma 2 No ea
m cm NH m Hm NH
m on Ha m om NH
m mm 0H m mm NH
2 mm m 2 mm 0H
m Nm m m Nm m
m mg 5 m mu m
2 mm o m mp h
m ma m m mm m
m ma v m cw m
z va m 2 on v
m m N m an m
m h H z mN N
umsmmflo m ha H
\umeucoo momm\uomc<
Hmnasz umnfisz
xmm ucmsomm muommumo xmm ucmfimmm mnomoumo

 

 

 

Apmscwucoov H xHozmmmd

APPENDIX J

COMPOSITION OF STIMULUS TAPE SEGMENTS--SUMMARY

171

 

 

 

Nae mNm wmm wmm wooH unmoumm

Ne we hm Hm mH mm mHmuoa

m N 0H m m mH umsmmHouumemunoo

m m OH m N GH mommuummqm

cH N NH m H NH zmHsmq<ummmuumHo

N m o H N N wHuumumummHnmusm

m CH m 0H N NH Nonnuamssomcm

m m HH 6 m NH HamsmuHoxmnpmmumucH
ownsuosuumcs pmuouosuum m z pwo> mucoEmmm
xmm amassz uwne:z

 

 

wm¢220mllmfizm20mm mafia mDQDZHBm m0 ZOHBHWOQZOU

b

xHQmemfl

APPENDIX K

SPEED OF A SAMPLE OF SEGMENTS AND PAUSES

BETWEEN SEGMENTS CHOSEN FROM THE SLOW MOTION TAPE

172

APPENDIX K
SPEED OF A SAMPLE OF SEGMENTS AND PAUSES

BETWEEN SEGMENTS CHOSEN FROM THE SLOW MOTION TAPE

 

 

 

 

Segment Time Space
1 37 10
5 34 8
8 33 6

11 27 7
14 35 9
12

18 30 14

22 32
25 31 9
29 35 8
35 31 8
40 32 13
45 32 9
50 32 10
55 33 5
60 30 5
65 32 ll
70 33 10
76 35 6
80 35 5
85 32 18
90 40 7
95 33 5
100 33 9
32.91 8.86 Sec.

6.71 times slower
than normal speed
or 15 percent of
normal speed

APPENDIX L

RATER SAMPLE COMPOSITION

f5

173

APPENDIX L

RATER SAMPLE COMPOSITION

 

 

 

 

'1

~)

 

Raw Percent
Sex Male 16 55
Female 13 45
Total 29 100
Academic Undergrad 0
Status Master 19 66
Doctor 10 34
Total 29 100
Age 20 - 25 6 20
26 - 30 11 38
31 - 40 9 31
41 - 50 l 4
Over 50 2 7

Total 29 100

 

APPENDIX M

RATER INFORMATION FORM

Rater number:

174

APPENDIX M

RATER INFORMATION FORM

 

Name:

Sex: Male

 

Address:

Female

 

Street

I‘L‘Irflt‘ik‘: c-n'r fl

Age:

 

 

City

Phone:

State

Academic Status

Major: Freshman

 

 

Sophomore

Junior

Senior

Graduate Student

I (do, do not) want a summary of the results of this study.

APPENDIX N

RATER RELEASE OF INFORMATION FORM

I.

175

APPENDIX N

RATER RELEASE OF INFORMATION FORM

I understand that I am participating in a research
study investigating variables thought to be relevant to
the process of affective communication. I understand that
all information recorded here will be used solely for
research purposes and that my confidentiality will be
strictly maintained.

I have voluntarily agreed to participate in this
study, and have not been offereéinor expect any remunera-
tion for my participation.

I understand that at any time during this study I am
free to discontinue my participation totally.

Signature

 

Witness Date

 

it"_"""—"' ‘1

APPENDIX 0

SCHEMATIC OF FACIAL MUSCULATURE

t

176

APPENDIX 0

SCHEMATIC OF FACIAL MUSCULATURE*

 

Frontalis

Corrugator

 

 

 

 

 

Zygoma-
ticus

 

 

Masseter

 

 

Depressor

 

 

*Modified from Fair, Schwartz, Friedman, Greenberg, Klerman
and Gardner, 1974.

 

APPENDIX P

AFFECT RECOGNITION TRAINING:

CATEGORIES OF EMOTION

1!]

I.-

177

APPENDIX P

AFFECT RECOGNITION TRAINING: CATEGORIES OF EMOTION

BY

David J. Inman
1975

This part of the training program is designed to in-
troduce you to the emotional labels that we will be using
in this study. There are thousands of emotional words,
each signifying various shades of meaning, and you probably
already have your own special vocabulary for describing
emotion. However, for the purposes of this study, we will
all use the same terms so we can have a clear understanding
of what we are talking about. In this study, the following
emotional labels will be used: Interest-Excitement,
Enjoyment-Joy, Surprise-Startle, Distress-Angfiish, Anger-
Rage, and Disgust-Contempt. We will now consider each of
these affects individually.

 

 

 

1. Interest-Excitement (Int)

 

Synonyms: Curious, concentrating, attending, absorved,

 

involved, attracted, intense, fascinated.
Description: Interest-Excitement is not usually

 

considered an affect. It consists of mildly

pleasant sensations which are aroused when
one is spontaneously attending to the
enV1ronment. Most people are capable of
maintaining this affect over long periods
oftmm.

Function: Taking in information about the environment.
Facial Components: Eyebrows level but may be slightly

 

 

raiSed or lowered. Eyes open normally and
fixated. Lips may be parted and jaw
dropped slightly.

2. Enjoyment-Joy(Joy)

 

Synonyms: Happy, glad, merry, joyful, cheerful,
blissful, jubilant, gay, elated, ecstatic,
gleeful, jovial.

Description: Enjoyment-Joy is the affect which most
people prefer to experience. It is the
most pleasant of all the affects. May
be stimulated by the onset of positive
stimuli or by the cessation of painful or
unpleasant stimuli. It can vary greatly
from mild enjoyment to ecstasy.

 

 

F’”‘“""“I

Function:

 

178

Social bonding, internal positive reinforce-
ment.

Facial components: Eyebrows level or lishgtly lowered.

 

Eyes bright--may be partially closed:
wrinkles ("crow's feet") in the outer
corners. Corners of mouth lifted back and
up (exaggerated with laughing). Teeth may
be partially exposed, upper lip tensed.
Naso-labial folds (running from the nose

to the outer edge of the mouth) are evident.

3. Surprise-Startle (Sur)

 

Synonyms:
Description: Surprise is distinguished by its very

 

Astonished, amazed, shocked, flabbergasted.

3?
.l
.‘

..’-.".£:t

 

Function:

 

brief duration. It is sudden in its onset
and fades quickly. It is triggered usually
by the uneXpected; if one anticipates an
event, then one cannot be surprised. It
may also be triggered by a sudden change in
sensory stimulation such as a gunshot or
flashbulb going off.

Clearing the sensory systems for assimila-
tion of new information.

ELM. -'

Facial Components: The brows are raised so that they

 

are curved and high, making the forehead
wrinkled horizontally. The eyes are
widened so that the white is visible above
and sometimes below the pupil. The jaw
drops so that the lips are loosely parted.

4. Distress-Anguish (Dist)

 

Synonyms:

 

Sad, unhappy, miserable, hurt, dejected,
depressed, despondent, dismal, low, grieved,
suffering.

Description: Most often, the feeling of distress is

 

Function:

 

associated with a loss of some sort. Loss
of a loved one, loss of an opportunity,
loss of self—esteem, loss of health all
can cause distress. Distress is usually a
prolonged feeling, usually lasting at
least a few minutes, and sometimes for
years. It is not as acutely unpleasant as
some of the other affects, but its pro-
longed endurance may have profound conse-
quences.

Communicate deficiency to the environment
and attract helpgivers.

179

Facial Components: Inner corners of the eyebrows are
drawn up, sometimes creating vertical
wrinkles in the forehead. The skin below
the forehead is triangulated with the inner
corner up. The corners of the mouth are
down and the lip may be trembling.

 

5. Anger-Rage (Ang)

 

Synonyms: Irritated, hostile, mad, hateful, aggressive,
annoyed, cranky, cross, disagreeable,
furious, belligerant, resentful.

Description: Intense anger usually implies risk of
losing control. Consequently, anger is
diligently supressed in early socializa-
tion. Anger is likely to be triggered by
frustration, or physical threat: it may
be a reaction to a psychological hurt,
an extreme violation of values, or a
failure to fulfill expectations.

Function: To mobilize the individual's resources for
confrontation with the disturbing element
in the environment.

Facial Components: Brows are lowered and drawn
together, creating vertical lines in the
forehead. Eyes have a hard stare and a
bulging appearance. Lips are either firmly
pressed together with corners down or
drawn back in a squarish shape, baring the
teeth.

 

 

IN- {— '.. EA‘C £1

 

 

6. Contempt-Disgust (Cont)

 

Synonyms: Scornful, disdainful, skeptical, condemning,
critical, arrogant, sarcastic, spiteful,
revolted, indignant.

Description: A feeling of aversion to something
sensory--as a taste, a smell, a sight,
or a touch. Also a feeling of aversion
to more complex stimuli--ideas, things, or
people. An element of condescension
.accompanies contempt-disgust--a "looking
down on" the object as being "below" you.

Function: To "expel" the object in either a physical
sense (i.e., vomiting) or a psychological
sens by rejecting the object.

Facial Components: The brow is lowered; one or both
cheeks are raised; the nose is sometimes
wrinkled. One or both sides of the upper
lip is raised and protrudes slightly, leav-
ing the teeth exposed. At times the tongue
may be slightly extended.

 

 

 

 

180

Thus in this study, you will be learning to recognize
and label facial expressions in these six categories:

1. Interest-Excitement (Int)
2. Enjoyment-Joy (Joy)
3. Surprise-Startle (Sur)
4. Distress-Anguish (Dist)
5. Anger-Rage (Ang)
6. Contempt-Disgust (Cont)

These are not the only emotional states that exist,
but cross cultural research has shown that observers can
accurately recognize facial expressions and name them
accordingly using these labels. Please note that two
important affects have been omitted from this study:
Fear-Terror and Shame-Humiliation. Although we believe
these to be important emotional states, we were able to
elicit each of them only a few times on videotape. Thus we
do not have enough instances of the facial expressions to
provide an accurate test of recognition in this study.

The facial expressions in the training tape and in the
actual study will then only be correctly labeled in one of
the six categories discussed, and in no others.

 

 

inure-73.: “new
I a

 

' {I11

 

APPENDIX Q

ANSWER SHEET:

AFFECT RECOGNITION TRAINING

181

APPENDIX Q
ANSWER SHEET: AFFECT RECOGNITION TRAINING

In this part of the training, you will be looking at
still photographs and videotapes of people talking about
their emotions. Your task will be to watch the film or
photo, and after the presentation of the stimulus is over,
to write down the emotion or emotions you recognized.

Because the time you will have to record your answers
is very brief, you might want to use the following abbrevia-
tions for the emotional categories:

Suggested Abbreviations

 

l. Interest-Excitement £23
2. Enjoyment-Joy pr
3. Surprise-Startle Sp;
4. Distress-Anguish Dist
5. Anger-Rage Apg
6. Contempt-Disgust Cont

 

If in a particular sement or photo you see more than
one emotion, write first the most intense andppredominant
emotion followed.by the other emotion (or emotionsTf’ For
example, if in one segment you see an expression of interest
followed by a brief but intense expression of surprise,
your answer would be "surprise, interest" or in abbreviated
terms "Sur, int." Likewise if in a segment or photo you
see a blend of expressions on the person's face, try to
ascertain what the predominant emotion is, and list that one
first on the rating sheet.

 

 

 

There will be three separate parts of this training,
and each of them have a different purpose:

Part 1: Identification of still photos (posed) which
have been used in the cross cultural research. The purpose
of this part is to familiarize yourself more thoroughly with
the facial expressions and to shorten the time necessary for
recognition. (Photographs courtesy of Carrol Izard, 1975).

Part 2: Identification of posed videotapes. Purpose
here is not so much that the emotions be identified accurate-
ly as that you have a chance to get used to watching video-
taped expressions over a short exposure time.

Part 3: Rater accuracy test. Here you will be viewing
still photographs of spontaneous facial affect, and be
tested on the accuracy and reliability of the responses.

Rater #

 

182

Interest-Excitement:
Enjoyment-Joy:
Surprise-Startle:
Distress-Anguish:
Anger-Rage:
Contempt-Disgust:

E)?

U)
C.
H

U
P:
U)

(f

 

An
Cont

La

 

 

 

 

 

 

 

 

 

 

 

 

12.

 

13.

 

l4.

 

15.

 

l6.

 

17.

 

18.

19.

 

20.

 

21.

 

22.

23.

 

' l

Ill! )‘ilvill

 

 

183

24.

 

25.

 

26.

 

27.

 

28.

 

29.

 

30.

 

31.

 

F! halaisiw

 

 

32.
33

34.

 

35.

 

36.

 

37.

 

38.

 

39.

 

40.

 

41.

 

42.

 

43.

 

44.

 

45.

 

46.

 

47.

 

48.

 

49.

 

50.

 

51.

 

52.

 

 

APPENDIX R

ANSWER SHEET:

AFFECT RECOGNITION STUDY

184

APPENDIX R

ANSWER SHEET: AFFECT RECOGNITION STUDY

In this study you will be viewing videotapes of people
talking about their emotions. Your task will be to watch
the tape, and when each segment is over, to write down the
emotion or emotions that you recognized in that segment.

Since these tapes were recorded without sound, you
won't know the content of the conversation. You will be
making your judgments solely on the basis of the person's
non-verbal communication.

Because the time between segments is very brief,
and becuase you will not record your answers until the
segment is concluded, you might want to use the following
abbreviations for the emotional categories.

Suggested abbreviations:

l. Interest-Excitement: Int

 

2. Enjoyment-Joy: pr
3. Surprise-Startle: Sur
4. Distress-Anguish: Digt
5. Anger-Rage: Apg
6. Contempt-Disgust: Cont

 

If you see more than one emotion during a particular
segment, write the most intense and predominant emotion
first, and the other emotion or emotionsI'second. For
example, if, in one segment, the subject holds an expression
of mild interest followed quickly by a brief but intense
expzession of surprise, your answer would be "surprise,
interest" or in abbreviated terms, "sur, int." Likewise if
in the segment you perceive a blend of expressions on the
person's face, try to ascertain what the predominant emotion
is, and list that one first on the rating sheet.

If you have any questions about any aspect of this
procedure, please bring them up now, since after we start
the tape, we will not be able to answer questions.

 

 

 

 

_. ’__‘, _ ,

Rater #

185

 

Interest-Excitement:
Enjoyment-Joy:
Surprise-Startle:
Distress-Anguish:
Anger-Rage:
Contempt-Disgust:

Int

1?

U)
G
H

U
H
U)
('1'

 

 

 

 

 

 

 

 

 

 

 

 

11.

 

12.

 

13.

 

14.

 

15.

 

l6.

 

17.

 

18.

 

19.

 

20.

 

21.

 

22.

 

23.

 

186

24.

 

25.

 

26.

 

27.

 

28.

 

29.

 

30.

 

31.

 

‘ .
F!§....§E§L

 

 

32.
33.

34.

 

35.

 

36.

 

37.

 

38.

 

39.

 

40.

 

41.

 

42.

 

43.

 

44.

 

45.

 

46.

 

47.

 

48.

 

49.

 

50.

 

51.

 

52.

 

53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.

79.

187

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

188

80.

 

81.

 

82.

 

83.

 

84.

 

85.

 

86.

 

 

 

87.
88.

89.

 

90.

 

91.

 

92.

 

93.

 

94.

 

95.

 

96.

 

97.

 

98.

 

99.

 

100.

 

101.

 

APPENDIX S

PRELIMINARY SUMMARY m FINDINGS ACCOMPANYING QUESTIONNAIRE

IN A SEALED ENVELOPE

‘Q'V‘OJ- ‘1

189

APPENDIX S

PRELIMINARY SUMMARY OF FINDINGS ACCOMPANYING QUESTIONNAIRE
IN A SEALED ENVELOPE

Seventeen subjects were interviewed in one of two
conditions while their facial expressions were recorded on
videotape. The subjects then reviewed their own tapes and
labeled their facial expressions in one of the following
eight categories: Interest-Excitement, Enjoyment-Joy,
Surprise-Startle, Distress-Anguish, Anger-Rage, Contempt-

Disgust, Fear-Terror, and Shame-Humiliation. Few facial

expressions were labeled in the latter two categories (Fear- F
Terror and Contempt-Disgust) and these were therefore eli-
minated from the study. Moreover, during the unstruc-
tured interview condition, subjects very seldom labeled the
emotions of Anger-Rage, Contempt-Disgust, Fear-Terror, and
Shame-Humiliation. Because of the small number of segments
obrainable from the unstructured interview condition in
these categories, a test of the effect of this treatment
was not feasible and the design of the study was altered to
exclude such a test.

One hundred and one segments of 5 seconds each were
chosen from the interview tapes and edited in a random
order. A ten second pause was placed between segments to
allow raters time to record their responses. A second
stimulus tape was then made which was an exact duplicate
of the first, except that the speed of the segments was
slowed to 1/5 normal speed.

Twenty-nine graduate students in the College of Educa-
tion at Michigan State University were trained in identifying
emotions from facial expressions. Each rater viewed the
stimulus tape twice (Time 1 and Time 2), some seeing it at
different speeds. Group 1 viewed theunormal speed tape
twice; Group 2 viewed the normal followed by the slow motion
tape; Group 3 viewed slow motion followed by normal, and
Group 4 viewed the slow motion tape twice. A minimum of one
week elapsed between viewings.

It was found that spontaneous facial expressions could
be accurately identified in each of the six categories. Dif-
ferences in both number of facial expressions seen and accu-
racy of identification were related to the speed at which
the videotape was shown. When the slow motion tape was
seen at Time 1, accuracy of identification was improved
at Time 2 (p less than .05). Moreover, there was an inter-
action effect between speed of tape and Time (T1 :c T2) when
looking at the overall performance in terms of number of
affects recorded. The slow-fast group and the fast-slow
group recorded significantly more affects than the other
two groups (p less than .05).

[5. ' G‘ “PH”;

rl‘."

r
1

190

The results of viewing facial expressions in slow
motion were related to the concept of micromomentary facial
expressions (MME's). MME's are facial expressions which
occur in less than 1/5 second and which are thought to be
related to affective problem areas (Haggard and Issacs,
1964). The current results indicate that if observers
view facial expressions in slow motion at Time 1, then
they become sensitized to MME's; at Time 2 they then record
a greater number of facial expressions per segment and
identify segments with greater accuracy than they had at
Time 1. Consistent with this conclusion is the finding that
the slow-fast and fast-slow groups saw more overall facial
expressions than the fast-fast group. It is theorized that
the aversiveness of the slow-slow condition caused this
group's performance to be detrimentally effected.

These results then indicate that speed of videotape has
significant effects on both the number and the accuracy
of rater's responses to spontaneous facial expressions
of emotions.

191

APPENDIX T
POST HOC RATER QUESTIONNAIRE

Rater number

 

Group

 

Rater Performance Scale

 

Instructions: A few months ago you participated in a study
in which you identified facial expressions on videotape.

In the items that follow, please circle the number that best
corresponds to your feelings about various aspects of the
study.

 

‘: xrjsr it

After you have completed this form, return it to me in the
self—addressed envelope.

F!" . ,K

 

>1 >.

>1u4 r4 >4

H (1) mv-l

m u .H +J m

E m >, m >4 :6 E

m n .H u .4 L: m

u m 'o .u 'c a) H

u 'o .4 a .4 13 u

x o .a m -H <3 x

a: 21 E: Z 2 :3 m

3 2 l 0 1 2 3
While 1 rated the videotapes, I felt:
1. Interested 3 2 1 0 1 2 3 Bored
2. Distressed 3 2 1 0 1 2 3 Contented
3. Confident 3 2 l 0 1 2 3 Unsure
4. Comfortable 3 2 1 0 1 2 3 Uncomfortable
5. Tired 3 2 l 0 1 2 3 Energetic
6. Relaxed Tense
I thought the tapes themselves were:
7. Easy to see 3 2 1 0 l 2 3 Hard to see
8. Of poor quality 3 2 1 0 1 2 3 Of good quality
9. Pleasant‘ 3 2 1 o 1 2 3 Unpleasant
10. Hard to judge 3 2 1 0 1 2 3 East to judge
I think the entire study was
11. Executed poorly 3 2 1 0 l Executed well
12. Valuable 3 2 1 0 l 2 3 A waste of

time

APPENDIX U

DATA MATRIX FOR ACCURACY DATA:

MEAN NUMBER OF RIGHT ANSWERS IN A CATEGORY

192

r ...'.lld a?

i.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mm. 50.HI m5N. om. 55. I 00.0I 55.5m Nm.Nm 50.0m 00.5m N0.00H 0N.0m H0009
0N. mm.H mH. 0.0 55. I 00. I 00.HN 00.0N 0H.mN 00.0H 50.00 5m.mm onm
m0. 5m.mI mNH. 0m. 0.0 NH.mI 5m.mH 5m.NN 5m.Hm v5.5H m5.mm 50.Hv ummm H0009
00. Nm.mI 0H. I m5m. N0.NI 5H.NI N0.0H 55.0N 00.0m 5N.0N Nm.mm m5.0m H0008
0N. 0N. I 00. 0.0 00. I o.NI ov.HH 0v.mH 00.NH 0m.0H 0N.mN 00.0N 30Hm
mNH. I mNH.MI 0m. I m5m. N0.HI m5.HI No.5 5m.MH 00.5H 50.0 NH.mN m5.0H 000m 30H0
05. om.H 5m. m5m. I mN.H 0N0. I m5.mH m5.mN 50.0N 5m.5H 00.Nm 00.05 H0009
0.0 m5.H 0m.NI 0.0 5m. I NH.H 0m.0H mN.0H om.vH 00.0 50.5N 5m.mH son
0m5. mN. I mNm. m5m. I N0.H 5m.HI mN.m 00.0 5m.vH 50.5 mo.mN NH.NN 000m 000m
ANm I Hmv mocmEMOMumm 0H mmamno HNm + Hmv moansuomumm HHMH0>0
5v.mH 0H.5N 0m.0N 5m.mH 05.00 HH.N¢ 0m.0N 00.0N 00.0N 50.0H No.00 HH.0N H0009
m0.0H m0.¢H 50.mH 00.0 N0.0N No.0H m0.HH 00.mH NH.vH m0.0 mH.0N 05.0H 30Hm
No.5 HH.0H N0.mH No.0 50.0N 00.NN mN.m m0.HH v5.mH NH.0 50.0N 5m.0H 000m H0008
50.0 vo.mH mm.mH 05.0 50.5N m5.HN mm.0 m0.MH mN.mH 5m.0H m0.mN 00.NH H0009
00.m 00.0 00.0 0N.m 00.NH 00.HH om.m 00.0 00.5 0N.m 00.NH 00.0 30Hm
5m.m 0N.m m5.m 0m.v 50.0H m5.0H m5.m mv.0 mN.m 5m.m mN.mH 00.0 9000 30H0
00.0 Nm.NH 0N.¢H 50.0 NH.0N 00.0N 05.0 No.mH H0.vH 0m.m 5m.5N HH.0N H0909
mN.m mN.5 5m.5 m5.v NH.¢H No.0 mN.m 00.0 NH.5 m5.v m5.mH 05.0 son
m5.m 5m.v 50.0 NH.v 00.NH 05.HH 0m.v N0.¢ 00.5 m5.m N0.MH 5m.0H ummm ummm
:00 use mHo Ham >00 00H :00 00¢ 0H0 Ham >00 00H
00 md 00 me N0 H0 00 me we 00 Ne H<
Ava N coHumucmmmum um 0:00: HHmo AHmv H 00Humusmmmum pm 0:002 HHoo new 0EHH H osHB

 

 

amoumado G ZH mmm3m24 BmUHm ho meZDZ dez

D xHDzmmmd

"dead MU<MDUU¢ mom meB¢Z cfidn

APPENDIX V

DATA MATRIX FOR FREQUENCY DATA:
MEAN PERCENT OF ITEMS WITH MORE

THAN ONE ANSWER RECORDED

193

 

 

 

 

 

 

 

 

 

05N. 0HN. 000. 000. 000. 0N0. 0H0.H 050.H H0.H mmH.N mmv.H 00H.H H0009
0N0. 0N0. mHH.I 000. N00.I 00H. 000. 5mo.H 0N0. 050. 555. «H0. 30H0
00N. HOH. 00H. HNv. 00H. 05H. 0H0. 0mm. 005. mmN.H 000. 05m. 0000 H0009
00N. H00. 00H. N00. 00H. mNH. 000. 000.H V50. NN5. 500. N00. H0009
H00. 00H. 000. 05N. 000. NNO. 00N. N00. 00N. N05. 50N. 5MN. 30Hm
0HN. HON. 00H. 00m. va. 00H. 500. 050. 0m0. 050. 000. 000. 0009 30H0
0N0. N0H.I 00H.I NMN. NmH.I 000. MN5. 000. 0M5. Nmm.H 0N0. 00¢. H0009
N00.I NHH.I mHH.I 50N. 00H.I 00H. 00m. 000. 000. 50N.H 0H0. 55m. 30Hm
0N0. 050.I 0H0. mmo. 0vo.I N50. 00H. 00H. HmH. 00N. 0HH. 0HH. 000m 0000
Nm I Hm N0 + H0
00005000000 0H 000000 00c0500mu0m HHOH0>O
M00. 5N0. v05. NNO.H M00. 555. 000. 0vo.H vm0. N00.H 005. 5m5. H0009
000. 00m. 500. moo.H 0H0. 05m. 5H0. N00. N00. 0H0.H 500. 000. 30Hm
00N. mNm. 0N0. 5H0. 05N. 00H. 5N0. va. N00. 000. 00¢. 050. 0000 H0009
mmm. Nmm. 50m. 000. «On. 00N. 05m. mm5. 000. NOH.H N00. 000. H0009
50H. 0NH. 5HH. 00N. HHH. 50H. 0mH. 00N. 5HH. VHO. mmH. 00N. son
0NN. 00N. 00N. NON. 00H. 05H. va. 50v. 00m. 050. 500. 05N. 0000 30H0
000. 000. 0N0. N00. 00N. 00N. 050. 000. 0N0. 000. 5NN. 00H. H0009
HON. 05m. 00m. 505. 00m. N5N. 05N. 00N. 00N. 000. NON. 00H. 30H0
500. 5HH. 050. mNH. H00. 0N0. 000. 500. M00. 00H. 000. 000. 000m 000m
:00 0:0 0H0 000 500 00H :00 0:4 0H0 000 500 00H
N0 H
N 00H000s0m0um 00 0:00: HH0U H 00H000s00000 00 00002 HH0U 0.x N 0EH9 H 0EH9

 

 

 

 

 

 

QMQmOUmm mm3m2¢ mzo Z<mB mmoz SBHS mzmaH m0 Bzmommm Zflmz

> anzwmm¢

.0900 002000000 mom meagz «e00

APPENDIX W

DATA MATRIX FOR POST HOC RATER QUESTIONNAIRE

L

194

 

 

 

44.4 44.4 55.N 4H.4 4H.4 44.4 44.4 4N.4 N4.4 44.4 44.4 55.4 onm
5H.4 54.4 45.N 44.4 H4.4 44.4 45.4 N4.4 44.4 44.4 5H.4 44.4 0004 onm
H4.4 44.4 4N.N 44.4 44.4 44.4 44.4 54.4 5H.4 N4.4 44.4 44.4 onm
44.4 44.4 4N.4 4N.4 44.4 H4.4 44.4 44.4 44.4 44.4 N4.4 4H.4 0404 0404
NH HH 4H 4 4 5 4 4 4 4 N H
50004 0:0
mo 00H0> NuHHmso 0005 0:42 40 00004 00000 0245 H 0244

 

 

 

 

 

 

mmH<ZZOHBmmDO mmB<m Dom Bmom mom XHmfidz ¢B¢Q

3 XHDmem4

 

TATE UNIV. L IBRQRIES

| I) ll Ml III) M)

31074 5844

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MICHIGnN s
llHlWllWll'
3129