INFORMATION TO USERS

This manuscript has been reproduced from the microfilm master. UMI films
the text directly from the original or copy submitted. Thus, some thesis and
dissertation copies are in typewriter face, while others may be from any type of
computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality illustrations
and photographs, print bleedthrough, substandard margins, and improper
alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete manuscript
and there are missing pages, these will be noted.

Also, if unauthorized

copyright material had to be removed, a note will indicate the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and continuing
from left to right in equal sections with small overlaps.
Photographs included in the original manuscript have been reproduced
xerographically in this copy.

Higher quality 6” x 9" black and white

photographic prints are available for any photographs or illustrations appearing
in this copy for an additional charge. Contact UMI directly to order.

ProQuest Information and Learning
300 North Zeeb Road, Ann Artxar, Ml 48106-1346 USA
800-521-0600

UMI

NOTE TO USERS

This reproduction is the best copy available.

UMI

DEVELOPMENTAL DIFFERENCES IN THE ABILITY
TO DECODE FACIAL EXPRESSION OF PAIN
by
Kathleen Susan Deyo
B.Sc., McMaster University, 1980
M.Sc. University of Waterloo, 1984

THESIS SUBMITTED IN PARTIAL FULFILMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
in
PSYCHOLOGY

© Kathleen Susan Deyo, 1999
THE UNIVERSITY OF NORTHERN BRITISH COLUMBIA
JUNE 1999
All rights reserved. This work may be
reproduced in whole or in part, by photocopy or
other means, without the permission of the author.

1 ^ 1

National Library
of Canada

Bibliothèque nationale
du Canada

Acquisitions and
Bibliographic Sen/ices

Acquisitions et
services bibliographiques

395 WaKngton Straat
Ottawa ON K1A0N4
Canada

395. rue WaKngton
Ottawa ON K1A0N4
Canada
YaurHm Voerw flW rene»

O urKm N o m r4fé im e »

The author has granted a non­
exclusive licence allowing the
National Library of Canada to
reproduce, loan, distribute or sell
copies of this thesis in microform,
paper or electronic formats.

L’auteur a accordé une licence non
exclusive permettant à la
Bibliothèque nationale du Canada de
reproduire, prêter, distribuer ou
vendre des copies de cette thèse sous
la forme de microfîche/fOm, de
reproduction sur papier ou sur format
électronique.

The author retains ownership of the
copyright in this thesis. Neither the
thesis nor substantial extracts from it
may be printed or otherwise
reproduced without the author’s
permission.

L’auteur conserve la propriété du
droit d’auteur qui protège cette thèse.
Ni la thèse ni des extraits substantiels
de celle-ci ne doivent être imprimés
ou autrement reproduits sans son
autorisation.

0-612-62471-4

CanadS

Approval

Name: Kathleen Susan Deyo
Degree: Master of Science
Thesis Title: DEVELOPMENTAL DIFFERENCES IN THE ABILITY TO DECODE
FACIAL EXPRESSION OF PAIN
Examining Committee:

Chair Donald Munton, Professor
International Studies

Supervisor: Kenneth Prkachin, Professor
Psychology

M ern |)ër\^erp ^eàb m o n t, Associate Professor
Psyqhology '

fember: Judith^kÿpa^ptfAssistant Professor
Edl

External Examir

Date Approved:

f 4-^ / 'y f

Ml
ara Robinson
Prince George Regional Hospital

ABSTRACT
The primary purpose of this thesis was to investigate age-related differences in
the ability to recognize facial expression of pain. A secondary purpose was to
examine the relationship between decoding performance and selected
personality variables. A final exploratory purpose was to determine which facial
cues were most predictive of participants' pain ratings. Previous research has
indicated that the ability to decode facial expressions of emotion improves with
increasing age throughout childhood. Little research has investigated variables
that influence the recognition of pain expressions, and none have examined age
differences. Thirty-three young adults and 102 children, in three age groups
(six, nine and 12 years-of-age), viewed a videotape containing 90 two-second
excerpts depicting clinical pain. Based on previous facial measurements, the
excerpts fell into three intensity ranges - no, low and moderate to high pain. The
participants’ pain ratings were converted to sensitivity indices using signal
detection theory. There was a linear increase in sensitivity with increasing age
for children; however, the oldest children did not differ from the adults. As an
additional way to evaluate observers' judgments, participants' ratings were
correlated with patients' self-reported pain and facial actions. These results also
indicated that participants’ decoding performance improved with increasing age
Measures of empathy and self-perception were not systematically related to any
dependent measures. The importance of different facial actions to pain ratings
was examined using hierarchical regression analyses. These analyses

u

indicated that fewer facial actions were predictive of pain ratings for the younger
children in comparison to the other participants. Overall, these results imply that
sensitivity to pain expression is largely developed by late childhood. The
implications of this study were considered in terms of development, decoding
methodology and the pain communication model.

Ill

Table Of Contents
Abstract

ii

Table of Contents

iv

List of Tables

vi

List of Figures

ix

Acknowledgements

X

Introduction

2
5

Facial expression of emotion; General considerations
Encoding and decoding of universal facial
expressions of emotion
Facial expression of pain
Encoding
Decoding
Measurement of pain with children
Present study
Hypotheses

8
22
22
24
30
32
33

Method

Participants
Materials
Procedure

34
34
40

Results

Demographic information
Missing ratings
Sensitivity and bias
Correlations between participant ratings and
patient self-report and facial actions
Experience of and exposure to pain
Self-perception and empathy analyses
Comparison of two pain rating scales
Action unit importance

44
45
45
58
60
61
65
68

Summary of results
Judgment accuracy
Sensitivity
Bias
Personality variables
Gender
Signal detection theory analyses for pain judgment

72
72
73
74
77
79
84
86

Discussion

IV

Methodological issues
Facial actions
Future directions
Summary and Conclusions

88
88
90
96
98

References
Appendix A

Pain-related Facial Action Units

106

Appendix B

Questionnaires

108

Appendix C

Pain Descriptors

122

Appendix D

Supplementary Tables

124

List of Tables
Table 1 ;

Mean ages of four age groups

44

Table 2;

Mean SES of participant's family

45

Table 3;

Mean sensitivity and bias

49

Table 4:

Mean bias scores for four age groups

51

Table 5;

Correlations between SES and sensitivity and bias

52

Table 6:

Intercorrelations among sensitivity, bias and correlation
measures for the total sample

53

Table 7:

Intercorrelations among sensitivity, bias and correlation
measures for the young children

54

Table 8:

Intercorrelations among sensitivity, bias and correlation
measures for the middle children

55

Table 9:

Intercorrelations among sensitivity, bias and correlation
measures for the older children

56

Table 10:

Intercorrelations among sensitivity, bias and correlation
measures for the adults

57

Table 11:

Mean
correlations between participants' ratings
patient self-report and facial actions by gender

and

58

Table 12:

Mean
correlations between participants’ ratings
patient self-report and facial actions

and

59

Table 13:

Mean Interpersonal Reactivity Index subscale scores for
children

62

Table 14:

Mean self-perception subscale scores for children

63

Table 15:

Correlations among the difference between participants'
and patients' ratings and sensitivity and bias measures

66

Table 16:

Correlations among the difference between participants'
and patients' ratings and Interpersonal Reactivity Index
scales

66

VI

Table 17:

Mean correlations for self-report, facial action and 2 rating
scales

67

Table 18:

T-tests to compare mean correlations for adults

67

Table 19:

Regression
analysis for
action
units
participants’ ratings for the total sample

predicting

69

Table 20:

Regression
analysis for
action
units
participants’ ratings for the young children

predicting

69

Table 21:

Regression
analysis for
action
units
participants’ ratings for the middle children

predicting

70

Table 22:

Regression
analysis for
action
units
participants’ ratings for the older children

predicting

70

Table 23:

Regression analysis for action
participants’ ratings for the adults

units

predicting

70

vu

Appendix D Tables
Table D1:

Correlations among college students' self-perception
subscales and sensitivity, bias and participants’
correlations with patients’ self-report and facial actions

125

Table 02:

Correlations among adults’ IRI subscales and sensitivity,
bias and participants’ correlations with patients’ self-report
and facial actions

126

Table 03:

Correlations among young children’s self-perception
subscales and
sensitivity,
bias and participants’
correlations with patients’ self-report and facial actions

126

Table 04:

Correlations among young children’s IRI subscales and
sensitivity, bias and participants’ correlations with patients’
self-report and facial actions

127

Table 05:

Correlations among middle children’s self-perception
subscales and
sensitivity,
bias and participants’
correlations with patients’ self-report and facial actions

127

Table 06: Correlations among middle children’s IRI subscales and
sensitivity, bias and participants’ correlations with patients’
self-report and facial actions

128

Table 07:

Correlations among older
children’s self-perception
subscales and
sensitivity,
bias and participants’
correlations with patients’ self-report and facial actions

128

Table 08: Correlations among older children’s IRI subscales
and
sensitivity, bias and participants’ correlations with patients’
self-report and facial actions

129

Table 09: Correlations among the difference between participants'
and patients' ratings and self-perception subscales

129

V lll

List of Figures
Figure 1. Pain Communication Model

4

Figure 2. Sensitivity as a function of age.

50

IX

Acknowledgements
I must express my appreciation to the many people who have contributed
to the completion to this thesis. First, I would like to thank my supervisor, Ken
Prkachin, for his unique ability to provide guidance and allow autonomy
throughout this endeavour. I appreciate his respect for my multiple roles (and
associated responsibilities) and my compulsion to take more courses; which
undeniably delayed progress. Secondly, I would like to extend my gratitude to
my other two committee members. Sherry Beaumont and Judith Lapadat, for
their support, insightful comments, constructive feedback and attention to detail.
I am thankful for all the time and effort everyone on tfie committee devoted.
The videotape for this project was constructed by Reiko Graham. Her
editing expertise is gratefully acknowledged. Recruiting participants was
facilitated with the assistance of the staff at Highglen Elementary School and the
Science Alliance and Y.M.C.A. Day Camps. I thank all of the children and
families who volunteered to participate and allowed me into their homes to
collect data.
Finally, I need to express my sincerest appreciation to my friends and
family for their support and encouragement. My friends have been willing to
listen and share their ideas. I thank my daughter Elizabeth for the inspiration for
this thesis. According to the words on the yellowed comic strip on our fridge “no accomplishment is without sacrifice” (Lynn Johnston). It would not have
been possible without the love and patience of my family.

INTRODUCTION
The primary purpose of this thesis was to investigate developmental
differences in the perception of facial expression of pain using a signal detection
theory (SDT) approach. The SOT rationale and methodology will be described
in detail in subsequent chapters. The introductory section highlights research
findings about the development of the ability to decode facial expressions of
emotion in general and the influence of social variables on decoding
performance focusing on previous methods employed. This section will be
followed by a description of the prototypic facial expression of pain and a brief
discussion of the relevant literature. Then, the results of studies that have
examined characteristics of adult observers that influence their judgments of
pain expressions is presented. Hypotheses at>out the development of the ability
to decode facial expression of pain were based on the facial expression of
emotions literature. The method section describes, in detail, the procedures
used to examine sensitivity to pain expressions. The results initially focus on the
signal detection analyses and will then address a number of ancillary analyses.
The implications of the study are discussed in terms of development,
methodology and the pain expression model.
The ability to recognize facial expressions of pain has important clinical
and social implications (Prkachin & Craig, 1995). For the clinician, facial
expressions are often important in the diagnosis of a medical problem and
evaluation of treatment (Manne, Jacobsen, & Redd, 1992). The importance of

clinicians' abilities to recognize facial displays of pain is especially true for
infants, those with handicaps that preclude accurate verbal reporting and
patients for whom distorting a verbal report is a concern (Bieri. Reeve,
Champion, Addicoat, & Ziegler, 1990; Craig, Hyde, & Patrick, 1991). In the
social environment, expressing pain or distress may serve as a warning of threat
or may solicit helping behaviour on the part of the observer (Prkachin & Craig,
1995). Research is beginning to demonstrate that, at least among adults, there
are individual differences in the way that people interpret others' facial
expressions of pain (e.g., Prkachin & Craig, 1995; von Baeyer, Johnson &
McMillan, 1984). Nevertheless, to date, no study has addressed differences
between adults and children of various ages in their detection and interpretation
of pain expressions.
Prkachin and Craig (1995) have proposed a model of the nonverbal
communication of pain that will form the basic theoretical framework for the
present study. Their model integrates a general model of nonverbal
communication (Rosenthal, 1982) with Ekman's (1977) neurocultural model of
emotion and recent data on pain expression. In general, nonverbal
communication depends on two aspects of performance that influence the
effectiveness of a communication (Zaidel & Mehrabian, 1969). Encoding refers
to the process that occurs when a person converts his or her emotion into a
facial expression. Good encoders are able to emit clearly discriminable cues to
their emotions in their facial expressions. In contrast, decoding is the process of

discriminating different facial cues. A competent decoder is able to discriminate
different feelings from a vanety of cues.
According to the pain model (Prkachin & Craig, 1995), the three
processes that may occur during an episode of pain are experiential (A),
encoding, (B) and decoding (0 ) (Figure 1). The experience begins when the

Figure 1. Pain Communication Model
Experience (A )-----------------------Encoding (8 )------

> Decoding (C)

Extnnsic
Factors

Gain

Noaceotion

Painful
Stimulus

Function

Response

ÎÎ
Intnnsic
Factors

Display
Rules

person is exposed to a stimulus that exceeds the pain threshold. The person
then may encode this experience in a facial expression that is transmitted to the
social environment. This expression is then decoded by observers. Decoding

refers to both the detection and interpretation of the facial expression of pain. At
each stage of the model, there are a number of variables that may affect the pain
display or the interpretation of it.
Although there are a large number of variables that affect the initial
experience of pain, these will not be addressed here. The present discussion
will be limited to the encoding and decoding of pain expressions. Furthermore,
encoding will be discussed insofar as it is necessary to demonstrate the
existence of a "universal" or prototypical pain expression. Then, variables that
are related to the accurate judgment of pain expressions will be addressed.
Facial expression of emotion: General considerations
First of all, it should be made clear that pain is not being interpreted as an
emotion. Nevertheless, referring to the emotion expression research is
appropriate because, like pain, emotional expressions have an interpersonal
function; to convey feelings (Levenson, 1994; cited in Levenson, 1996).
According to Schweder, (1994) pain is a nonemotional feeling that is "associated
with perceptible facial icons” (p. 39).
There is ample evidence to show that nonverbal cues (e.g., gestures,
facial expressions) are critical to effective interpersonal communication. They
transmit information to the social world that may emphasize, complement or
contradict verbal messages (Zimbardo & Leippe, 1991; Bugental, Kaswan, &
Love, 1970; Riggio, 1992). In fact, Philippot, Feldman, and McGee (1992)
suggest that facial expression may be more effective than verbal cues in

^ revealing an individual's internal state. Presumably, this suggestion is because
nonverbal behaviour is assumed to be spontaneous and under less voluntary
1 control than verbal t)ehaviour (Egan, 1986; Levenson, 1994).
At this point, it is also important to address the concept of universality of
j
^ facial expressions of emotion and how universality relates to the development of
I encoding ability. If facial expressions of emotion are "universal," then encoding
I of them should occur naturally given the appropriate stimuli and maturity of the

I

I neural and muscular control mechanisms. Therefore, a discussion of "universal"

1 and "encoding ability” will be provided below.
■1

i

There is considerable debate in the literature (see Russell, 1995, for a
review) about the degree of universality of facial expressions of emotion.
Nevertheless, there is agreement that there is at least "minimal universality”

] (Russell, 1995). According to this position, across cultures, humans produce the

iJ same facial expressions in similar situations (actual or imagined), and they use
i
^ the same emotion labels (or equivalent translations for speakers of different
j languages) when asked to decode facial expressions in photographs. Ekman
j
I (1973; cited in Russell 1995) claims that the universal facial expressions are

I

I "happiness, sadness, anger, fear, disgust, and surprise” (p. 380).
à
Most researchers also agree that cultural or social display rules will
influence what different people will express (e.g., Russell, 1995). According to
this position, humans may be predisposed to reveal their emotions in
I
I recognizable facial expressions, but there is significant voluntary control over

I

; the actual expression that is displayed. Display rules are unwritten guidelines
i that govern how a person is expected to reveal one's emotion. For example, in
I our culture, display rules dictate that males and females should control their
j expression of anger or sadness in different ways. The use of display rules
; demonstrates that expression of emotion is, to some degree, under voluntary
control. Rinn (1991 ) indicates that spontaneous or involuntary facial
I expressions and voluntary expressions are under different types of neurological

I control; which would suggest that encoding ability could be different for
involuntary and voluntary expressions.
The majority of the research concerning age-related changes in encoding,
to be described below, involves the posing of expressions of emotion rather thari
the spontaneous expression that accompanies an emotion. The ability to pose
nonemotional expressions (i.e., facial expressions of emotion that are not
associated with the experience of emotion) leads to questions concerning the
role of either verbal ability or knowledge of emotion vocabulary in the
performance of the encoding tac!;s Is a good encoder able to control facial
musculature better than a poor encoder? Or, does a good encoder have a
better understanding of emotion words and expectations about how to
demonstrate that awareness?
There is one additional point to be aware of when considering the facial
expression research. It appears that encoding and decoding accuracy are often
confounded. For example, encoding accuracy is often operationally defined as a

function of the degree of consensus among the observers (i.e., decoding
;agreement). As will be shown below, in a typical encoding study, children will be
asked to pose the emotion that is appropriate for a given situation. Then, their
I accuracy at encoding the expression is assessed by the judgments of adult (e.g.,

university students) observers. Therefore, the decoders’ judgments determine
encoding accuracy.
Encoding and decoding of universal facial expressions of emotion
Given the significance of facial expressions of emotion to interpersonal
communication, it is necessary to determine how the ability to encode and
decode them develops. Developmental changes in the decoding of universal
expressions of emotion will be emphasized as they are relevant to the
hypotheses and design of this study. In addition, the methods typically used to
assess encoding and decoding performance will be described because they
differ from the measures employed in this study. Specifically, most facial
expression research involves the discrimination among different expressions,
whereas in this study, the participants’ task was to discriminate between different
intensities of the same expression (i.e., a pain expression). A SDT approach
was employed in this study for two reasons. First of all, in general, the SDT
method allows for the calculation of independent sensitivity and bias indices
(See, Warm, Dember & Howe, 1997). According to Ellermeier (1997), the
willingness to report pain (i.e., bias) affects pain ratings and therefore should be
distinguished from the sensory or sensitivity factor. Secondly, the SDT method

8

has been used previously by Prkachin (1992b; Prkachin & Craig. 1985) to
examine observers’ sensitivity to expressions of pain resulting from exposure to
1
i different intensities of painful stimulation.
I

With respect to facial expression of emotions, both encoding (i.e., often of

I posed expressions) and decoding improve during childhood until adult
I
I performance is achieved. Adults, in general, perform at nearly perfect levels
I and sometimes, but not always, there are sex differences in performance. For

I
I example, Zuckerman, Lipets. Koivumaki & Rosenthal (1975) found that women
I were slightly better encoders and significantly better decoders than men for the
I
I six universal facial expressions of emotion.
I

One of the earliest studies examined children’s production and

3

I discrimination of eight facial expressions (Odom & Lemond. 1972). The

Ii materials for this study were a total of 32 black and white photographs with each
of the eight expressions (i.e., fear, anger, distress, shame, disgust, surprise, joy
I and interest) represented four times. Children in kindergarten and grade five

I

participated in one of two discrimination and one of two production tasks. In the
matching-discrimination task, the child was to select a photograph, from a series

I of four provided at one time, to match the way the person in the standard photo
felt. For the situation-discrimination task the child was told that the people in the
four photographs being shown felt a different way. The experimenter read a
situation, and the child was to choose which photograph corresponded to it. For

' Encoding accuracy is the percentage of facial expressions that are correctly identified by
observers, whereas decoding accuracy is the correct identification of a universal facial
expression or proportion of agreement among observers.

the production tasks, the children were told to make faces as well as they could
with their faces in a frame so that their picture could be taken. In the imitationproduction task, the children were shown two photographs of the same
expression and were instructed to make a face like those shown. In the
situation-production task, the experimenter read situations and the children were
asked to make a face like they would feel if they were in the specific situation.
Results demonstrated that grade five children were more accurate on both
encoding and decoding tasks than kindergarten children. There were no
significant effects of gender on either task. Discrimination was more accurate
than production indicating that children were better at recognizing than posing
facial expressions. Furthermore, because the older children made errors, the
authors suggested that maximum sensitivity had not been attained by the 10year-old children.
Similarly, Profyt and Whissell (1991) read stories to children aged 4 - 6
years and asked them to pose how the child in each of the stories would feel.
The children's poses were videotaped. One week later, the tapes were shown to
the child, another child and to adults. Overall, encoding accuracy (i.e.,
recognizability of the expressions produced) increased with increasing age.
Performance for girls and boys was the same with the exception that girls were
better at encoding the fear expression.
With the exception of one study to be described later (Kolb, Wilson, &
Taylor, 1992), most research suggests that differences in encoding and

10

&

I decoding abilities are a function of social influences or social skill. For example,

I
1 Tucker and Riggio (1988) examined the relationship between social skills in
I
encoding of posed and spontaneous facial expressions by adults. The

spontaneous expressions were videotaped unobtrusively while subjects viewed
I slides that were chosen to arouse three emotions (disgust, happiness and

1 sadness). The participants were informed of the videotaping. They were then
given cards with an emotion and a neutral message to recite and were instructed
to pose the given emotion. The Social Skills Inventory, which measures four
I aspects of social skill, was administered following the videotaping. Groups of

I

I three judges rated the segments for which emotion a participant was expressing.
I
Encoding accuracy was defined as the percentage of judges who correctly

I

I identified which emotion the participant was expressing. Adults with higher
I social skills scores were superior encoders of posed expressions in comparison

I to those with lower scores. Spontaneous expressions were less related to social
I skills than were the posed expressions.
I
In contrast to the previously discussed study which included adults,

!

I Feldman, White, and Lobato (1982) conducted two studies that investigated the
relationship between social skill and encoding and decoding abilities of children.
In the first study, boys and girls ranging from five to 12 years-of-age were
administered a role-taking task as a measure of social skill. Then, each child's
nonverbal encoding skill was evaluated. They were instructed to sample two
drinks (sweetened and unsweetened) and try to convince (i.e., fool) an

11

interviewer that they either liked or disliked both drinks. The children were being
videotaped during their interviews. Untrained observers rated the children in
segments as truthful or deceptive. A deception ability score, the measure of
encoding ability, was the percentage of observers who identified as truthful a
child who was being deceptive. Role-taking skill was significantly correlated with
encoding ability and this was more pronounced for girls than for boys. Unlike
other studies, encoding ability did not increase with age. However, role-taking
did improve with age.
In their second study, Feldman et al. (1982) examined encoding and
decoding ability of normal and institutionalized, "emotionally distuited"
adolescents. Seventeen specific social competencies were rated by teachers or
counsellors as the indicator of social skill. Participants viewed three two-minute
videotapes that were intended to elicit positive, negative and neutral emotional
responses. The adolescents were videotaped while they viewed the tapes.
Encoding was defined as undergraduate judges’ ratings of the participants' facial
expressions on a 5-point unpleasant-pleasant scale. During the decoding phase
of the study, the participants rated the facial expressions of ten undergraduates
viewing the same tapes. Results indicated that normal adolescents were better
encoders and decoders than "emotionally distuited" adolescents. Furthermore,
social skill was correlated with encoding but not with decoding.
In Feldman’s earlier research (i.e., Feldman, et al., 1982), social
competencies were rated by teachers or counsellors. In his later work, he and

12

his colleagues have used the social competence scale of the Child Behavior
Checklist (CBCL; Achenbach & Edelbrock, 1982). For example. Custrini and
Feldman (1989) investigated the relationship between social competence, as
measured by the CBCL, and encoding and decoding performance of 9 - 1 2 year
olds. Naturally occurring expressions, elicited during viewing of videos, were
used for the encoding task. Encoding accuracy was determined by
undergraduate judges. The children's decoding task was to view videotapes of
university students’ reactions to videos and to indicate which of five emotional
expressions was being displayed. The videotaped segments had been
previously rated by undergraduate Judges. A child’s response was considered to
be correct if it agreed with the students’ judgments. Decoding was more
accurate than encoding for all five emotion categories assessed. An important
finding was the significant interaction between social competence and gender.
Boys, in both high and low social competence groups, performed with the same
degree of accuracy. However, there was a difference in the performance of
girls. High social competence girls performed better than boys, whereas low
social competence girls were worse than the boys. It should be noted, however,
that there were only three girls in the low social competence group.
According to Feldman, Philippot and Custrini (1991), there is no definition
of social competence that is acceptable to most researchers or theoreticians.
Many terms are considered synonymous, and many aspects of social behaviour
have been studied. Feldman and his colleagues (e.g., Feldman et al., 1991 )

13

assume that various social skills underlie competence (e.g., the ability to
manage impressions, communication behaviours such as patterns of eye contact
or voice intonation). Therefore, they consider social competence as a
multidimensional domain. From their perspective, "both decoding and encoding
skills can be viewed as manifestations of social competence” (p. 331).
The remaining studies to discuss will focus on decoding. Examining
infants' recognition of facial expressions requires methods quite different from
those employed with older children and adults. For example, looking times are
commonly used to assess either recognition or memory. It has been found that
four-month-old infants’ looking times are longer for happy expressions than for
angry or neutral expressions (Labarbera, Izard, Vietze & Parisi, 1976; cited in
Gosselin, 1995). Furthermore, by 5 months-of-age, infants are able to
discriminate happy and sad facial expressions as measured by a
habituation/dishabituation method of assessing infant memory (Walker-Andrews
& Lennon, 1991 ). In the habituation phase of this procedure, infants were
presented slides of one facial expression repeatedly until looking time
decreased to a criterion amount of initial looking time. The infants were then
presented with a new slide that either matched or mismatched the expression on
the previous slides. Looking times were longer for mismatched expressions
suggesting that the infants recognized that the mismatched expression was
different. Nelson and de Haan (1996) investigated whether 7-month-old infants'
brain activity was influenced by the type of emotion expressed in a face.

14

Electroencephalogram (EEG) recordings were taken while Infants were
presented slides of either happy or fearful faces or fearful or angry expressions.
They demonstrated that the infants were able to discriminate happy from fearful
expressions but not angry from fearful expressions. Therefore, the ability to
discriminate different facial expressions is present early in life. Gosselin (1995)
suggests that this discrimination of facial expressions of emotion is the
foundation of social competence that appears early in development.
It appears that decoding ability continues to develop throughout
childhood. For example, Missaghi-Lakshman and Whissell (1991 ) had children
in grades 2, 4, and 7 draw faces for tfie six universal expressions. Two weeks
later, the faces were decoded by adults and the child who drew the faces.
Decoding accuracy improved with increasing grade. There were no differences
between boys and girls. Similarly, Tremblay, Kirouac, and Dore (1987) observed
that children's decoding ability improved with age. Their methodology was
different in that children were presented with photographs of the universal facial
expressions displayed by adults and children of both sexes. They observed that
decoding performance was not affected by either the age of the model nor the
sex of the observer.
Other researchers have demonstrated that decoding ability is related to
level of intelligence. Xeromeritou (1992) investigated the decoding ability of
educable mentally retarded and nonretarded controls. A short story was read to
the children. The children identified a picture for how the person in the story felt.

15

I The story was read again, and the children then produced the word for how the
■i

person felt. When these groups were matched for verbal mental age, those with

I the older mental age were more accurate than children with a younger mental
I age. Simon, Rosen, Grossman & Pratowski, (1995) observed that facial
: expression recognition was positively correlated with IQ for adults with mild to
moderate mental retardation and was unrelated to measures of social skill. The
facial recognition task was to identify which person, from a group of six
photographs, exhibited a specific emotion. These results suggest that facial
expression recognition is a cognitive process or is dependent on veibal ability as
it is assessed by IQ tests.
Gosselin (1995) stated that although decoding is known to improve with
development, little is known about how these improvements occur. He designed
his study to investigate the types of decoding errors children committed. In his
study, short stories were read to children of two age groups ( 5 - 6 year olds, and
7 - 8 year olds). They then selected the photo that best matched how the
character in the story felt. There were no age differences in decoding happy
expressions. However, the older children were more accurate for the remaining
five expressions. Using knowledge of Facial Action Coding System (FACS;
Ekman & Friesen, 1978) for the various expressions, he was able to conclude
that improvements were due to a reduction in certain types of errors. In order to
explain the change of errors, a very basic understanding of FACS is required.
FACS is an objective, anatomically based system to describe the actions of the

16

I face. There are 44 codable facial action units (AU). Each of the universal facial
j expressions Is characterized by a combination of AUs. There is some overlap of

i
i AUs across emotion expressions. That is, some emotional expressions contain
j the same action units (AU) as other expressions. For example, brow lowering is
present in sadness, anger and fear, and raising of the upper lip occurs with tx>th
anger and disgust. Gosselin (1995) observed that with an increase in age,
children became more sensitive to combinations of facial components that
convey different emotions.
Consistent with the encoding research presented previously, a number of
studies suggest that social variables influence decoding performance. For
example, normal boys with unhappily married parents have been found to be
significantly worse at decoding five facial expressions of emotion than were girls
or boys with happily married parents (Shortt, Bush, McCabe, Gottman & Katz,
1994). in order to demonstrate the importance of social skills on nonverbal
behaviour, Feldman et al., (1991 ) cited a number of studies that demonstrated
that a variety of "distuibed" individuals (i.e., delinquents, abused children,
psychiatric patients), did not decode photographs of facial expressions as
accurately as those without disorders.
The facial expression recognition abilities of children with a specific
disorder, Attention-Deficit Hyperactivity Disorder (ADHD), were investigated by
Singh et al (1998). Children with ADHD were read two-sentence stories that
included the target emotion word and asked to identify, from a selection of six

17

photographs, the one that showed the emotion described. ADHD children were
observed to have deficits in the ability to recognize the six universal emotion
expressions, in comparison to norms for non-ADHD children (Singh, et al.,
1998). It was suggested that ADHD children have social skill deficits that could
be attributed to differences in social interactions in comparison to non-ADHD
children.
Children's interactions with their peers can be assessed with peer ratings
of specific behaviours or by sociometric methods. According to Feldman et al.
(1991 ) sociometric measures are the most common method of assessing social
competence^. They reviewed research that demonstrated that sociometric status
was associated with decoding accuracy in the expected direction. For example,
Walden and Field (1990) observed that sociometric preference scores were
significantly related to facial expression discrimination scores. Discrimination
was defined as the preschoolers’ ability to match a standard face with one from a
group of five.
Similarly, Philippot and Feldman (1990) examined the relationships
among age, social competence and decoding of facial expressions by
preschoolers. The task was to watch a videotaped scenario, with the main
character’s face blacked out, and then select an emotion face (happy, sad, or

^Although they do not state how sociometric methods are used to draw conclusions about social
competence, it is possible to speculate on the relationship between social competence and
sociometric status. One might assume that relatively popular children are more skilled in the
types of behaviours necessary to establish and maintain positive interpersonal relationships.
Conversely, disliked children may be likely to engage in social trehavlouis that are otyectionable
to their peers.

1

afraid) that was most appropriate. High social competence children were better

1 decoders independent of age, sex or emotion. There was a sex by age
■!
1 interaction. Girls improved their performance between three and four years,

I whereas boys improved between four and five years. That is, boys were
delayed by one year relative to girls. Although, this study was presented as one
involving decoding, the children's task was not to decode a given facial
expression, but rather to understand the scenario and select the facial
expression that was most suitable for the given situation.
Beck and Feldman (1989) also suggested that decoding ability is related
to social skills. Furthermore, they proposed that ability differences are primarily
a result of learning processes. Therefore, they systematically attempted to
increase competence in decoding of emotional expressions in children in grades
five to seven. Improved competence was accomplished by providing feedback
to one group and not providing it to the other. There were no sex differences of
decoding, but there was a significant effect of feedback. Boys and girls in the
feedback condition improved their accuracy at decoding different emotional
expressions.
Two possible explanations for the relationship between social
competence and decoding ability have been proposed. Children who are poor
at discriminating nonverbal cues may miss important information about others
that impairs their social competence. Conversely, those with poor social skills
may not have the same opportunities, due to more limited social interactions, to

19

j acquire the ability to differentiate facial expressions. However, it is also possible
j

that both skills are due to some third factor such as a concern for others. For

J example, Eisenberg et al. (1996) suggest that empathy or sympathy is related to
I

social functioning.

i

In contrast to the above studies which propose that social variables are
the primary influence on decoding ability, Kolb et al. (1992) suggest that
improvements with age may reflect maturation of the frontal lot)es. In their
photograph-matching and cartoon-matching tasks, they observed improvement
until approximately 14 years-of-age. This improvement was not gradual; it
increased at about eight years-of-age and then again around 13 or 14 years.
Adult patients with frontal lobe lesions performed the tasks with the same degree
of accuracy as eight- to 13-year-olds. Because of the similarity in decoding
accuracy for these two groups, they suggested that the development of decoding
ability parallels frontal lobe maturity.
There are a few points that need to t)e considered in the evaluation of the
emotion expression literature before summarizing it. First of all, there is a large
body of literature that involves some aspect of development and emotional
expression. Earlier literature (e.g., Odom & Lemond, 1972) focused on age
differences in producing and discriminating expressions. Then, research
examined encoding and decoding performance as a function of age in
combination with other variables (e.g., social skill; Custrini & Feldman, 1989).
More recently, there has been an emphasis on the effects of context on

20

I encoding and decoding of emotions (e.g., Russell, 1997). Therefore, the
I selection of studies described above was a representative sample of the
I

literature that addressed the effect of age on decoding performance. The
second comment concerns the fact that most tasks in the research are very
language-laden. In order to perform well on the typical tasks, one must have

I

I good verbal skills or have a good understanding of emotion vocabulary.

I However, proficient nonvert>al communication should not require a languagebased test to assess it. Therefore, the verbal nature of the tasks may confound
encoding and decoding performance. Finally, most studies use deliberate or
posed expressions rather than spontaneous emotional expressions. There is
evidence to suggest that these two types of expressions may have different
characteristics (e.g., symmetry or latency; Hager & Ekman, 1997) and that
different neurological systems control voluntary and involuntary expressions
(Rinn, 1991 ). Furthermore, Rinn (1991 ) states that both systems tend to act
simultaneously but with varying degrees of influence on the expression
displayed. Display rules, which are learned, govern voluntary expressions. In
sum, it would appear that the facial expression literature is somewhat difficult to
interpret due to a variety of potential confounds (e.g., verbal tasks, the role of
learning in posed expressions).
Despite the foregoing criticisms, the literature on the development of
encoding and decoding of facial expressions suggests the following conclusions.
First of all, both processes improve with increasing age. However, it is possible

21

I that different mechanisms may account for improvements in encoding and
I decoding. It has not yet been established at what age adult levels of

ii

i performance are achieved. Few studies have been conducted with children

j older than ten years of age. Gender differences are not always observed, but
I

when they are, girls tend to be more accurate than boys. Finally, more socially

Î

i skilled individuals appear to be more accurate in their recognition of facial
I expressions of emotion.

Facial Expression of Pain
Encoding
A basic assumption underlying the preceding work is that there are
universal facial expressions of emotion. Studies employ prototypical emotion
expressions as the standard forjudging decoding accuracy. The primary
purpose of the present study was to examine developmental differences in the
recognition of facial expressions of pain. In order to make judgments about the
decoding of pain expression, there is also the assumption that there is a
universal or prototypic facial expression of pain. The following section reviews
evidence that analogous to the facial expressions of emotion, there is also a
prototypic facial expression of pain.
A small number of North American research groups have used FACS
(Ekman & Friesen, 1978) to describe the facial actions of individuals in pain. Of
the 44 possible facial actions only a fraction of them have been found to be
consistently related to acute pain. Prkachin and Craig and their colleagues

22

(e.g., Prkachin & Mercer, 1989; Craig et al., 1991; Poole & Craig. 1992;
Prkachin, Berzins & Mercer, 1994) have described a cluster of facial actions that
are correlated with subjects’ self-report of pain. These actions include tightening
of the orbit (by raising the cheeks; FACS action units (AU 6 and 7), lowering the
brow (AU4), wrinkling of the nose (AU9) or raising the upper lip (AU 10). The
eyes may close (AU43). Movements around the mouth are common, but are not
reliably associated with specific painful stimuli or pain reports. The action units,
associated muscles and facial movements are described in Appendix A.
Prkachin (1992a) has described this pattern of facial actions as prototypical.
LeResche (1982) used 16 candid photographs to determine which facial
actions occurred most frequently during pain. Her results also indicated that
brow lowering, orbit tightening and eye closure were very common. In
comparison to the videotapes of pain induced during clinical examinations, she
observed less nose wrinkling and upper lip raising, whereas she observed a
greater degree of mouth opening (AUs 25, 26, and 27) and lip stretching (AU20)
Further support for the notion that the components of the expression are
universal are the findings on pain expressions in infants. Infants, premature and
full-term, exhibit a pattern of facial actions that is very similar to that observed in
adults. Even very premature infants exhibit the pattern of actions albeit less
vigourously than older infants. Specifically these actions are - brow bulge, eye
squeeze, naso-labial furrow and open mouth (Grunau, Johnston, & Craig, 1990;
Johnston, Stevens, Craig & Grunau, 1993; Craig, Hadjistavropoulos, Grunau, &

23

Whitfield, 1994; Stevens. Johnston, & Horton, 1994; Johnston. Stevens, Yang &
Horton, 1995).
It is evident that people have some ability to voluntarily control their
display of pain. In conditions where there is experimental (Prkachin, 1992b) or
clinical pain (Poole & Craig. 1992; Craig et al.. 1991; Hadjistavropoulos. Craig.
Hadjistavropoulos. and Poole, 1996; Galin & Thom. 1993), participants have
been able to mask and exaggerate their pain displays. FACS coding of the
relevant action units demonstrated that there was less facial activity of pain in
masking conditions and greater activity in the exaggerate conditions. Although
there were subtle cues to pain intensity available, for the most part, it was
possible to exert control over facial expressions of pain (Rinn. 1991).
Decoding
Adult observers are able to discriminate different intensities of pain from
the relevant facial actions. That is. observers' ratings of pain are significantly
accounted for by the FACS coded actions In one of the earliest studies on
observers’ reactions. Patrick. Craig and Prkachin (1986) investigated the
relationship between a model's pain tolerance, subjective pain reports and
observers' ratings. In the first phase of the study, three groups of participants
were videotaped while they received electric shocks of increasing intensity with
a confederate present. The confederate modeled either tolerance or intolerance
to the shocks or remained seated but was not a coparticipant. Following the
ascending series of shocks, a random series was presented varying intensity

24

level according to discomfort ratings from the previous series. In the second
phase, untrained observers rated the level of discomfort they believed the
participants were experiencing. Participants exposed to pain tolerant models
reported a similar degree of pain as those exposed to an intolerant model even
though they experienced more intense shock. Despite the comparable ratings of
discomfort, their facial expressions displayed more activity associated with pain.
Furthermore, the observers rated the subjects with the tolerant model as
experiencing more pain which was consistent with the greater intensity of shock
received.
in contrast to the study of laboratory pain described above, Prkachin et al.
(1994) employed videotapes of clinical pain. Five judges were shown a
videotape of patients with shoulder pain undergoing an examination of the
shoulder by a physiotherapist. Patients and observers rated the patients' pain
on the same scales. Facial expressions were measured using FACS (Ekman &
Friesen, 1978). Their results indicated that observers' pain ratings were
correlated with patients’ pain ratings for severe pain but not when the pain was
submaximal. Despite the variable degree of correlation between observers’ and
patients’ ratings, the facial actions were consistently correlated with the patients'
pain ratings. These results suggest that although the information necessary to
draw inferences about the person's state was available to the observers, they
did not make appropriate use of more subtle cues.

25

As indicated previously, facial expressions of pain are, to a certain
degree, under voluntary control. Given that Craig et al. (1991 ) found that
genuine pain expressions could be discriminated from suppressed and faked
displays on the basis of FACS coded expressions, Poole and Craig (1992)
examined how observers would rate dissimulated (i.e., masked or exaggerated)
pain. Would they rate pain intensity according to the facial display or would they
be able to identify the actual pain experienced? Results indicated that
observers rated more pain in the genuine, suppressed and faked conditions than
in the baseline (no pain) condition. The genuine condition differed significantly
from both dissimulated conditions. Specifically, they rated more pain for the
exaggerated faces and less pain was attributed to the suppressed faces relative
to the genuine condition. However, the patients were not totally successful in
deceiving the observers. At least some indication of pain was evident in their
facial displays. For example, for the suppression condition, observers gave
higher ratings of pain than in the baseline condition. The results suggest that
observers’ pain ratings were related to facial activity even when dissimulated.
There were no gender differences in the amount of pain observers perceived in
the various conditions.
In another study involving dissimulated pain, Hadjistavropoulos et al.
(1996) asked participants to differentiate among no pain, genuine pain, masked
pain and exaggerated pain of patients with low back pain. The videotapes of
these four conditions were taken at the patients' clinic. In the pain conditions.

26

patients were Instructed to lift their legs off the table and to genuinely express,
suppress or exaggerate the pain they were experiencing. The order of pain
conditions was countert)alanced. Participants were informed that, for each
patient, there were the four conditions. Their task was to classify the segments
into one of the categories. They then rated the intensity of the patient’s pain.
No pain and exaggerated pain were most accurately classified by observers.
However, masked and genuine pain were only correctly identified at a chance
I level. Therefore, it appears that even though patients were attempting to hide
I pain, certain cues were still present that were expressed. On the other hand,

I extreme pain expressions were more readily identified as such. Some facial

0

|| cues (e.g., brow raising and lip comer pull) were more likely to occur with
exaggerated faces and observers made systematic use of them to classify the
I video excerpts.
Prkachin and Craig (1995) suggest in their model that a "gain function”
exists that affects an observer's likelihood of reporting pain (Figure 1). Some
observers would report pain with minimal evidence whereas others would
require considerable evidence before they would decide that someone is in pain.
Overall, Prkachin (e.g., Prkachin et al., 1994; Prkachin et al., 1983) has found
I that observers demonstrate an underestimation bias. That is, observers tend to
report that a person is experiencing less pain than he or she reports.
It has been suggested that experience with pain sufferers may influence
the ratings by observers and therefore may affect the underestimation bias.

27

Subjects who lived with a family member suffering from chronic pain had a
diminished underestimation bias (Prkachin, Solomon, Hwang & Mercer, 1995).
On the other hand, nurses (von Baeyer, at al., 1984) and physiotherapy and
occupational therapy students (Prkachin, et al., 1995) had an exaggerated
underestimation bias. Taken together, these results imply that experience per
se is not reliably related to bias. Rather, the nature of the observers’ experience
affects their bias. There is evidence that, in general, attributions for actors and
observers are more similar if they like or identify with each other (Regan &
Totten, 1975). Differences in attributions may explain why the family member
would be more likely to provide ratings closer to the patients’ than unfamiliar
observers would. It has also been suggested that clinicians "distance"
themselves from the suffering of their patients and this may result in ignoring or
minimizing evidence of their patients’ distress (Prkachin, et al., 1995).
Hadjistavropoulos et al. (1996) suggested that research should address
the possibility that "there may be some individual difference variable that would
make some judges better decoders than others” (p. 257). Nurturance has been
found to be related to expressions of pain (Von Baeyer et al., 1984). Intuitively,
it seems logical that empathy, a similar construct, would be related to sensitivity
to pain expressions. Solomon (1995) used a multidimensional measure of
empathy, the Interpersonal Reactivity Index (IRI; Davis, 1983), to determine
which types of empathy were related to students’ ratings of shoulder pain
patients’ ratings. Of the four types of empathy, fantasy was significantly related

28

to observers’ ratings and empathie concern was marginally significant (p = .07).
Fantasy (F) is the tendency to identify with fictitious characters and empathie
concern (EC) is the tendency to experience feelings of warmth, compassion and
concern for others undergoing negative experiences.
Eisenberg et al. (1996) have used a simplified version of the IRI, adapted
for children, in their study of the relationships among empathy/sympathy, social
functioning and physiological responses to a videotape of a child in distress. In
addition, they observed that children high in sympathy were also rated as more
socially competent by their peers. In earlier work, Eisenberg and her colleagues
have found that sympathy is related to helpfulness (Fabes, Eisenberg &
Eisenbud, 1993), parental characteristics and coping behaviours (Eisenberg et
al., 1991 ) and physiological responding (Fabes et al., 1993; Eisenberg, et al.,
1989).
An early study by Zahn-Waxler, Radke-Yarrow and King (1979) examined
the relationship between maternal behaviour and children's reactions to another
child’s distress. They found that children whose mothers used affective
explanations when their 1 % - 2 % year old children caused or witnessed
distress were significantly more likely to react prosocially (e.g. help or hug the
other child). Distress included laehaviours other than facial expression. Mothers
who were rated as higher in empathie caregiving had children who were more
likely to behave altruistically, and their actions were often accompanied by
concerned emotional expressions. The authors concluded that maternal

29

disciplinary techniques lay down the foundation for responsiveness to feelings of
others. Although this study was not a decoding study per se, it does
demonstrate that the ability to perceive distress may be present early in life and
also that the early social environment influences sensitivity to others.
Chun, Turner and Romano (1993) discovered that children who lived with
a parent with a chronic pain problem were rated by teachers as being less
socially competent than children of healthy controls. Because social
competence was related to pain exposure, it may also be related to sensitivity to
pain expressions.
There have been no studies of the ability of children to decode facial
expression of people in pain. However, Shih and von Baeyer (1994) ascertained
that preschoolers were able to make gross distinctions between drawings of
faces in pain.

Measurement of Pain with Children
The accurate assessment of pain in adults can be challenging, and
probably even more so for children. Pain is a multidimensional experience;
therefore, it should be assessed multidimensionally (Abu-Saad, 1994). The
assessment of it must “conform to the communication capabilities of the suffering
person, whether infant, verbal child, effectual adult, or incompetent adult"
(Anand & Craig, 1996, p. 5). Given the foregoing statements, it is apparent that
self-report measures, despite being the “gold standard" of assessment, should
be supplemented with additional measures.

30

For the most part, knowledge about a child’s pain is based on observation
of the face and behaviour (Abu-Saad, 1994). However, self-report measures are
also employed. The participants in this study were required to provide ratings of
others' pain. In order to rate pain in others, it would seem logical that, at the
very least, one should be able to provide a self-report of pain. Therefore, a brief
description of self-report procedures will be presented.
There are three ways children have provided self- report of pain. The first
is to give a verbal report of their experience. Studies by Abu-Saad (1994) and
Wilkie et al. (1990) indicate that normal children over the age of about seven or
eight are able to use different descriptors to characterize their pain. Although
self-report is the gold standard’ for adults, the limited vert)al ability of younger
children has precluded its use with them. Visual analogue scales (VAS) have
been used with some success with children for rating the intensity of their pain
(Abu-Saad, 1994). However. Bien, et al. (1990) assert that these are not
appropriate for the preoperational child for a variety of reasons (e.g., limited
ability to understand that a line represents pain intensity). Therefore, they
developed the Faces Pain Scale for children’s self-report of pain severity. This
7-point scale includes line drawings, derived from children’s drawings, of faces
displaying “no pain" to “the most pain possible." The child is required to select
the face that best illustrates a pain experience. Manne et al. (1992) used an
approach similar to Bieri et al. (1990) except that the children used a 5-point
scale with associated faces. Both studies indicated that this type of instrument

31

could be used with children as young as three years of age. Unfortunately, a
faces scale would be inappropriate in the present study because a face' should
not be used to rate a facial expression. It would not be clear wtiether a face on
the scale was being used to rate the intensity of pain or to match the patients’
face. Therefore, because of the limitations of the VAS and faces scales it was
necessary to use a verbal scale for rating pain intensity. Although a 7-point
scale is more sensitive to variations in intensity, it was expected that 5-year-old
children might have difficulties with such fine discriminations.

Present Study
The purpose of this study was to investigate age-related differences in the
sensitivity to facial expressions of pain because, to date, there have been no
reported studies that included children as observers of others' facial expressions
of pain. In addition, response bias was examined to determine whether there
were age-related differences in the willingness to report pain in others. Children
of three different age groups were compared to young adults. Measures of self­
perception and empathy were included as covariates to examine whether these
individual difference characteristics influenced either sensitivity or bias in pain
ratings. Sensitivity and response bias were addressed with a signal detection
paradigm. Prkachin and his colleagues (Prkachin, Currie, & Craig, 1983;
Prkachin & Craig, 1985; Prkachin, 1992) have demonstrated the utility of signal
detection methods to analyze between group differences in sensitivity to pain
expressions and response bias.

32

Hypotheses
Based on the literature reviewed, the following hypotheses and issues
were examined.
1. If decoding pain expression is analogous to discriminating among
different facial expressions, then decoding ability should increase with
increasing age. However, an age effect might not be the same for sensitivity
and response bias in judging pain expressions.
2. Participants scoring higher on social acceptance and empathy will be
more accurate than those scoring lower in social acceptance.
3. Decoding studies do not reliably demonstrate differences between
males and females. However, where there are differences, females are found to
be more accurate than males. Therefore, if there are differences between males
and females In this study, the latter will be more sensitive.
4. Finally, there will be a comparison among age groups of the most
influential facial cues used when judging pain. That is, do children and adults
rely on the same facial cues when they rate pain? This question regarding the
influence of specific facial actions for participants of different ages is exploratory
rather than one based on previous research. Therefore, there are no a priori
hypotheses regarding the relative importance of facial cues for participants of
different ages.

33

METHOD
Participants
A total of 134 children and adults from four age groups participated in this
study. The three age groups of children were five- and six-year-olds (n = 33),
eight- and nine- year-olds (n = 35), and eleven- and twelve-year-olds (n = 33).
For convenience, these will be referred to as young, middle and older children,
respectively. The number of boys and girls in each of the groups were as
follows: young boys (n = 16), young girls (n = 17), middle boys (n = 17), middle
girls (n = 18), older boys (n = 16) and older girls (n = 17). The group of young
adults had 17 men and 16 women.
The children were solicited from a variety of sources; some were selected
from an available school, some from summer day camps, others were by referral
and the remainder from personal contacts. Therefore, it was a sample of
convenience. The young adults were university students recruited from
introductory psychology classes at UNBC, and they received course credit for
their participation.
Materials
Participants viewed videotaped segments of the faces of patients
undergoing an assessment of a shoulder injury by a physiotherapist at a sport
medicine clinic. The assessment included ten range of motion and two
accessory tests. The range of motion tests included active and passive
abduction, flexion and internal and external rotation movements as well as two

34

additional passive movements (strong and weak "lock and quadrant” tests). For
the active tests, the patients moved their arms into position. In contrast, the
patients had their arms moved by the therapist during the passive and quadrant
tests. See Prkachin and Mercer (1989) for a more detailed description of the
tests. The patients were standing, facing the camera, when the four active tests
were performed. On the videotape, the patients' heads and shoulders were
visible. For the remaining tests, the patients were lying on an examining table.
The camera angle was from above and slightly to the side of the face. The
shoulders were not usually visible for these tests.
The stimuli were carefully selected from among the entire series collected
by Prkachin and Mercer (1989) to have certain properties. The intent was to
construct a test tape displaying a series of facial expressions that could be
specified, by independent criteria, as conveying no pain, some pain and a lot of
pain. To do this, excerpts were selected to meet each category based on the
measures of pain expression available from the studies of Prkachin and Mercer
(1989) and Prkachin, Berzins and Mercer (1996). Each excerpt had been coded
for the intensity (on a 5-point scale) and duration of four specific movements
which have been shown to covary with pain: 1) brow lowering. 2) orbit
tightening, 3) levator activity and 4) eye closure. A summary score, consisting of
the sum of the products of the intensity and duration of each action was
available. This score varied between 0 and 120. A videotape was constructed
that included 30 excerpts which, according to the foregoing criteria, depicted no

35

pain activity (a score of 0 on pain actions), minimal pain activity (a score ranging
between 1 0 - 10.0), and moderate to strong pain (a score over 10). Therefore,
there were a total of 90 target stimuli. In addition to having the pain index scores
for the facial actions, the actual pain ratings by the patients were available for
comparison purposes.
The stimuli were presented in a fixed randomized order varying the sex of
the patient and the pain intensity of the facial expression. These target stimuli
were presented for two seconds followed by a five second interstimulus interval.
The trial number appeared on the screen before the presentation of the face.
The first 30 seconds of the tape were blank, then there were 12 practice trials.
These trials were followed by another 30 seconds of blank tape prior to the
presentation of the 90 target trials. The total time of the tape was less than 13
minutes.
All participants viewed the tape on the same portable Genexxa TV/video
machine. The participants were instructed to sit at a comfortable distance from
the screen. In almost all cases the distance chosen was approximately 1 meter.
One six-year-old girl, who said that she wanted to get up very close to see the
faces better, chose to sit closer. One young man said that he moved back so
that he could see the whole image on the screen better.
Participants also completed two questionnaires. The first was an
empathy questionnaire and the second was a self-perception scale. There were
a number of criteria employed in the selection of the instruments used for the

36

assessment of empathy and social skill. The first was that a similar instrument
be available for the wide range of ages observed in this project (Bryant, 1982).
That is, it was deemed desirable to have parallel forms for adults and children to
make across-age comparisons. In addition, the questionnaires should have
acceptable psychometric properties. Thirdly, either the adult or children’s
questionnaire should have a demonstrated relationship to facial expression
decoding ability in previous research.
In the case of empathy, domains of the IRI have been found to be related
to decoding accuracy of facial expression of pain (Solomon, 1995). The IRI for
adults also has excellent psychometric properties (Davis, 1983). Versions of the
children’s IRI have been employed in a variety of studies by Fabes and
Eisenberg and their colleagues (e.g. Eisenberg, et al., 1989). Furthermore, the
IRI measures different dimensions of empathy. It is probable that these
dimensions are not equally associated with pain judgments. Therefore, the
adults’ and children’s versions of the IRI were selected to assess empathy.
As indicated previously, social competence is operationalized in many
different ways. Therefore, selecting a suitable instrument was somewhat more
difficult. Social competence has been assessed with the CBCL (Achenbach &
Edelbrock, 1983) in facial expression decoding research by Feldman and his
colleagues (e.g., Feldman & Custrini, 1989). The CBCL does not have a parallel
form for adults. Therefore, an instrument that has been found to be correlated
with the CBCL or has been found to be related to other measures of social skill

37

was sought. Harter's Self-Perception Profiles for children (SPPC) and college
students (SPPCS). which measure a variety of dimensions of self-concept such
as scholastic competence and global self-worth, were selected for the following
reasons.
Despite the fact that none of the profiles have been employed in facial
expression research, there is sufficient justification for the use of them to assess
aspects of social competence via the social acceptance domain. Pope and
Ward (1997) administered the CBCL and SPPC to children with craniofacial
anomalies. Social competence was defined as the composite of the social
acceptance score and the CBCL social competence scale score Low perceived
social acceptance on the SPPC is observed in more depressed than non­
depressed students (e.g., Heath, & Weiner. 1996) and depressed children are
more likely to be rated as lower in social competence as measured by the CBCL
(Renouf, Kovacs, & Mukerji, 1997)
The social acceptance domain of the SPPC has been found to be related
to other aspects of social behaviour as well. For example. tx)th Boivin and
Hymel (1997) and Austin and Joseph (1996) have found victimization to be
associated with lower levels of social acceptance. Withdrawal and loneliness in
the former study and bullying in the latter study, were also associated with low
perceived acceptance.
Hymel, Bowker, and Woody (1993) administered a different self-concept
instrument (Self-Description Questionnaire; Marsh, Smith, & Barnes, 1983; cited

38

in Hymel, Bowker. and Woody, 1993) and compared the results with peer
ratings. Overall, they claimed that average children, as defined by sociometric
nominations, were fairly accurate in their self-perceptions. Therefore, the peer
relations domain of self-concept corresponded to peer rated social competence.
Therefore, to assess empathy, the Interpersonal Reactivity Index (IRI;
Davis, 1983) was administered to adults and the children's version of the IRI
developed by Fabes et al. (1992) was completed by children. Harter’s SelfPerception Profile for Children (SPPC; Harter, 1984) and Harter's SelfPerception Profile for College Students (SPPCS; Neeman & Harter, 1986) were
administered with children and adults respectively. The children's IRI (Fabes, et
al., 1992) and the SPPC (Harter, 1984) had been standardized with children in
grade three by their respective developers. See Appendix 8 for the
questionnaires.
The SPPC has 5 dimensions of self-concept (physical appearance,
athletic competence, behavioural conduct, scholastic competence and social
acceptance) as well as a measure of global self-worth. In the first version of the
scale, the social scale was referred to as social competence. Harter (1984) has
since reported that it measures a child's perception of his or her acceptance by
peers. The SPPCS is composed of 12 dimensions plus the global self-worth
measure. The categories on the children's version are broad whereas for the
adults, categories are more refined. Therefore, the dimensions are similar but
not exactly the same.

39

In contrast, the IRI scales are more directly comparable across ages. The
adult version has four subscales, with 7 items per scale. These are; Empathie
concern (EC), perspective taking (PT). personal distress (PD) and fantasy (F).
For the children’s IRI, the F scale was omitted and there were fewer items per
scale. The original scale contained 4 items per scale resulting in a 12-item test.
Fabes (1997) has recommended the use of a 10-item scale. The shorter scale is
obtained by omitting two of the three negatively worded items (#4 & #8). The
final version has 3 items for both EC and PD whereas PT still has 4 items.
There is one negatively worded PT item remaining on the questionnaire. The
children completed the 12 item version and analyses were conducted for both 12
and 10 Item scales.
Procedure
Children. All of the children’s sessions were conducted during the
summer. Eighty children had their sessions in their own homes. For these
children, consent was obtained at that time. The sessions for ten children were
held in a familiar home - either a friend’s, a relative’s or a family daycare. In
these cases, the parents agreed, in telephone conversations, to their children’s
involvement and allowed the adult present to provide written consent on the
parents’ behalf. The remainder of the children participated at their day camp.
The day camps were located at two elementary schools. For these cases, a
consent form signed by a parent had been returned prior to the experimental
session.

40

The TV was set up in a convenient location in the home or in an area of
the school free from distractions. Some demographic information was obtained
from the parents and/or children. This information included the child’s birthdate,
parents' education and occupation, the child's school and programme
(approximately one third of the children did not attend their neightx>urhood
schools), and whether the child or a primary caregiver experiences recurrent or
chronic pain. For most of day camp children, information about parents'
occupations and education were missing.
Parents were allowed to be present for the entire session if either they or
their children preferred. Most parents chose to watch at least some of the tape
with their children. They were informed that because one of the purposes of the
study was to examine children's ratings of pain in others, they should avoid
letting their child know how they would rate the patients' faces.
A brief description of the procedure was then given. The children were
instructed to watch each face on the video, decide if the patient hurt during the
movement, and then give their response. It was explained that there was no
audio on the tape so they had to watch the faces closely to judge if the person
was experiencing any pain with the movement. The five point rating scale was
explained and shown (0= no pain, 1= maybe pain or can’t tell,

2

-

a

little pain, 3

= moderate or a middle amount of pain and 4 = a lot of pain). A laminated sheet
of paper with the rating scale was placed in front of the participants. Pilot testing

41

indicated that children as young as 5% years of age were able to use the 5 point
scale when the researcher recorded the responses.
The practice trials were presented to assure that the participants watched
and were able to rate the faces. For the one child who said he had difficulty with
this task, a second opportunity to view the practice trials was given. The
children viewed each target stimulus and then gave their response aloud. That
is, pain ratings were recorded during the interstimulus interval by the researcher.
The trial number on the screen was used to assure correspondence between the
video and response sheet.
The two questionnaires were administered when the tape was finished.
With one exception, the first questionnaire completed was the children's version
of the IRI. The second questionnaire was Harter’s SPPC. In all cases, the
oldest children completed their questionnaires independently. Only rarely did an
older child need assistance with a word. For all of the youngest children, the
instructions and items were read for the children. They provided their answers
aloud or, more commonly, by pointing to their choices. Most of the children in
the middle group chose to record their answers themselves with the items being
read to them. However, some preferred to do their questionnaires in private.
Adults. University students were contacted to arrange a mutually
convenient time. When students arrived at the testing room they were given a
description of the study as one comparing children and adults in their ratings of
facial expressions of pain. Demographic information was then obtained. The

42

procedure for the adults was the same as above with the following exception.
After they viewed the tape and rated the pain on the 5-point scale, they watched
the tape a second time rating the excerpts using the scale that was originally
used by the patients in the study from which the excerpts were selected. The
scale of affective pain intensity is a ratio-scaled verbal descriptor instrument to
assess clinical pain (Heft, Gracely, Oubner & McGrath, 1980). See Appendix C
for this rating scale. Students then completed the IRI (Davis, 1983), and Harter’s
SPPCS (Neeman & Harter, 1986).

43

RESULTS
Demographic Information
The age and sex composition for all of the groups is in Table 1. There
were no significant differences in the ages of males and females. The majority
of the children (n = 64) attended their neighbourhood schools. Thirty-three
children attended alternate programmes provided by the school district; seven
were registered in French immersion and 26 were enrolled in the Montesson
programme. The remaining four children attended private Christian schools.
Table 1
Mean Aoes. in vears. for four aoe orouos
Aoe Group
Young Children
Middle Children
Older Children
Adults

Males
6.21
8.90
11.84
21.11

Females
6.27
8.96
11.95
23.11

Total
6.24
8.93
11.90
22.08

To determine socioeconomic status (SES), the table by Blishen, Carroll,
and Moore (1987) was used. This table is based on very precise job
classification and income information. For the most part, the participants or their
parents provided very general job descriptions or classifications. Therefore, the
most general classification in a category was used for rating SES unless more
precise employment or academic information had been given. Family SES was
taken as the highest rated occupation in a family. For the large majority of
university students, family SES was based on a parent's classification. Overall,
this sample was atx)ve average (mean = 42.74; Blishen, et al., 1987) in SES
(see Table 2). A 4 (age) X 2 (sex) ANOVA was computed to determine if there

44

were age or sex differences in family SES. A significant main effect of age was
found, F (3,1 23 ) = 2.95, g = .04. The Tukey-HSD post-hoc analysis indicated
Table 2
Mean SES of participant’s family
Aoe Group
SES
Young Children
57.37,
Middle Children
54.66^,
Older Children
55.55^,
Adults__________________ 47.23b
bnumbers with the same subscripts are
not significantly different at g = .05.
that the adults' family SES was significantly lower than that of the families of the
young children (Table 2).
Participants' Missed Retinas
A 4 (age) X 2 (sex) ANOVA was calculated to determine the effects of age
and sex on the number of stimuli participants missed. The mean number of
stimuli missed decreased with increasing age. F (3,126) = 13.39, g = .001, eta
squared = .24. The Tukey HSD test indicated that only the youngest children
differed from the adults in the number of stimuli not rated. The missing data
were not uniformly distributed among the children in the young and middle
groups. In all further analyses with participants’ ratings, the missing values were
replaced with the average rating for that stimulus.
Sensitivitv and Bias
The stimuli presented to participants can be conceived as representing
three intensities of a signal (none, weak, strong) presented in the context of
ambiguity. The rating scale used by participants can be conceived not only as

45

an indication of apparent pain intensity, but also as a confidence scale of
increasing certainty (i.e., one is unlikely to use the "strong pain” category unless
one is confident that some pain was displayed). Data obtained in this manner
can be analyzed using the methods of signal detection theory (McNichol, 1972)
to characterize a person’s performance in two ways. A measure of the
participant’s sensitivity to variations in the intensity of the stimulus can be
calculated. Various indices of sensitivity are available, depending on the
assumptions made about the task; however, they are all based on evaluating the
observer’s ability to discriminate between categories of stimuli. When the
reference category against which responses to higher intensity stimuli contains
none of the stimulus in question the measure obtained is conventionally
interpreted as representing the observer's ability to detect the higher stimulus.
In this study, the reference category included faces not displaying pain-related
facial actions. A second measure, theoretically independent of the first, can also
be calculated. This measure represents the observer’s "bias,” "response
criterion” or tendency to be liberal or conservative in making ratings. Again,
various indices are available to estimate these parameters, depending on the
nature of the task.
To calculate indices of sensitivity and response bias, trials on which the
patient was not making any pain-related facial actions were considered "noise”.
Trials on which the pain expression index varied from 1.0 to 10 were considered

46

mild signals, whereas, trials on which the pain index exceeded 10 were
considered strong.
The measure of sensitivity employed in the present study was PfAl
(McNicol, 1972). P(A), for "proportion of area." was selected because it is a
nonparametric measure of sensitivity that is ideally suited to the analysis of
signal detection experiments involving the use of a rating scale. It is an estimate
of the area of a unit square lying underneath a Receiver-OperatingCharacteristic (ROC) curve formed when "hit" probabilities are plotted on the
ordinate and "false alarm” probabilities on the abscissa. For each rating scale
category, a "hit” is considered to be the use of that category to describe a
"signal”, whereas a "false alarm” is the use of that category to describe "noise”.
For all rating scale categories, except the most lax (i.e., "no pain”), hit
probabilities, when calculated in this manner, will exceed false alarm
probabilities if the stimuli t>eing judged are distinguishable. It is conventional to
accumulate the hit and false-alarm probabilities from the most stringent (in this
case "4”) to the most lax (in this case "0") rating scale category so that in
practice the hit and false alarm probabilities ultimately sum to 1.0. Each rating
scale category then becomes a point on the ROC curve. The area under that
curve can then be estimated by application of the trapezoidal rule, yielding the
measure PfA). When calculated this way, PfA) can vary from 0 to 1.0. A value
of 0.5 represents chance performance. This reflects either an inability to
distinguish two categories of stimuli or no sensitivity.

47

Two sensitivity measures were calculated in this manner for each
participant. The first estimated participants’ sensitivity to expression of mild pain
and was calculated by considering mild pain expression to be signal and no pain
expression noise. This measure will be referred to as PfA)m or alternatively, the
sensitivity to or detectability of mild pain. The second estimated sensitivity to the
expression of strong pain and was calculated by considering strong pain
expression to be signals. This measure will be referred to as PfA)s or,
alternatively, the detectability of strong pain.
The measure of response bias,

was based on a nonparametric

technique recommended by Donaldson (1992), based on application of the
following formula to the data at each rating scale category:
B"d = [(1 - P (h it)) X (1 - p(FA)) - (p (h it) X p(FA))] /

[(1 - p(hit)) X (1 - p(FA>) + (p(hit) X p(FA))]
Measures of B"o at each rating category were then averaged to yield a single
measure of bias.
P(A)m and P(A)s were entered as repeated measures into a 2 (sex) X 4
(age) X 2 (pain intensity) repeated measures ANOVA (see Table 3 for means).
Results indicated that there was a significant within-subject pain intensity effect,
F (1, 126) = 291.91, B = .001, eta squared = .70. That is, the expression of
strong pain was easier to detect than that of mild pain. There was also a
significant main effect of age, F (3 ,1 26 ) = 85.22, b = 001, eta squared = .67,
indicating a developmental increase in sensitivity to detecting pain. These age

48

and intensity effects are shown in Figure 2. Post-hoc analyses using the TukeyHSD technique indicated that the young, middle and older children were
significantly different from each other on both P(A)m and P(A)s; however, the
older children were not significantly different from the adults. There were no
significant interactions nor was there a main effect of sex.
Table 3
Mean sensitivitv and bias
Aaearouo
Young
Male
Female
Middle
Male
Female
Older
Male
Female
Adult
Male
Female

N

Sensitivitv
Mild Strono

Bias
Mild Strona

16
17

0.58
0.62

0.65
0.70

0.39
0.44

0.34
0.35

17
18

0.71
0.74

0.83
0.82

0.54
0.48

0.38
0.37

16
17

0.81
0.84

0.89
0.90

0.60
0.46

0.49
0.35

17
16

0.87
0.85

0.93
0.93

0.43
0.42

0.31
0.26

Bias reflects the tendency to be consen/ative or liberal in signal detection
(Donaldson, 1992). Values above 0 indicate a conservative bias, whereas
values below 0 reflect a tendency to be liberal. A person with conservative bias
tends not to report a signal under conditions of uncertainty, whereas, as liberal
bias indicates a greater tendency to report signals. As can be seen from Tables
3 and 4, the participants in this study were conservative in their ratings.

49

B"o for both mild and strong pain intensities were also entered as
repeated measures into a 2 (sex) X 4 (age) X 2 (pain intensity) repeated
measures ANOVA. There were no significant between-subjects effects (age, sex

Figure 2. Sem itivity as a function of age

0.9

0.8

0.7

0.6

0.6 -1—

Young

Middle

Older

Adult

Ag* group

or age X sex interaction). However, there was a main effect of intensity on bias,
F (1,126) = 172.84, g = .001, eta squared = .58, as well as an age by intensity
interaction, F (3, 126) = 3.02, g = .03, eta squared = .07. There was a greater
degree of conservative bias for the mild intensity level. It also appears, from
observation of the means (Table 4), that the middle and older children tended

50

Table 4
Mean bias scores for four aoe groups
AaearouD
N
Bias
Mild
Strong
Young Children
33
.42
.34
Middle Children 35
.51
.38
Old Children
33
.53
.42
Adults
33
.43
.29

to be more conservative, especially at the mild level. However, when separate 4
(age) X 2 (sex) ANOVAs were conducted for each intensity level, no significant
effects resulted.
Recall that there was a main effect of age on sensitivity, and there was a
significant difference in SES across age groups. In order to ascertain whether
SES differences could account for the obsen/ed differences in sensitivity,
correlations between them were computed for each group. As can be seen from
Table 5, there were significant negative correlations between SES and
sensitivity for both mild and strong pain for only the oldest children. That is,
lower SES was associated with higher degrees of sensitivity to pain cues. In
contrast, it was the adults who were from significantly lower SES families.
Therefore, differences in SES across ages could not explain age- related
differences in sensitivity. Nevertheless, other demographic differences may be
able to account for differences in sensitivity. Specifically, the academic and
occupational backgrounds of the participants’ parents differed for the children
and adults. The mothers and fathers of the children were more likely to have
university degrees, 42% and 37%, respectively than the mothers (12% ) and

51

Table 5
Correlations between SES and sensitivitv and
bias with level of significance fo^ in parentheses
Aaearouo

Young Children
Middle Children
Older Children
Adults

Sensitivitv
Mild
-.10
(.58)
.05
(8 0 )
-.36
(.04)
.20
(.28)

Strona
.10
(.59)
-.01
(.99)
-.42
(0 1 )
.24
(2 0 )

Bias
Mild
-.02
(.92)
.15
(.38)
-.03
(.89)
.15
(.42)

Strona
-.05
(.77)
.23
(.20)
.03
(8 5 )
.19
(.31)

fathers (15% ) of the adult students. In addition, almost half (n = 47) of the
children’s mothers and 15 of the fathers were employed in education, health,
counselling or social work positions. In contrast. 8 mothers and 1 father of
adults were employed in similar occupations. The percentage of parents with
college diplomas or certification did not differ between groups for the mothers
and father (between 27% and 30%). Therefore, occupational or educational
differences between the parents may contribute to sensitivity.
As noted previously, bias and sensitivity are theoretically independent.
Correlations among these measures were computed for the total sample (Table
6) as well as for the four age groups (Tables 7 through 10). In all cases,
sensitivity to mild signals was highly correlated with sensitivity to strong cues (r
ranged from .58 to .90. g = .001 for all correlations). The mild and strong bias
intercorrelations were even greater (r = .91 to .98. g = .001 ). In contrast, the

52

bias and sensitivity indices were not significantly correlated for any of the age
groups. For the total sample, however, bias for the mild signal was significantly
correlated with sensitivity to the strong signal (r = .18, g = .04). Overall, the
above findings support the position that bias was empirically independent of
sensitivity.

Table 6
Intercorrelations among sensitivitv. bias and correlation measures for the total
sample with significance level foi in parentheses
Sensitivitv

mild
Sensitivity
mild
strong

strona

Correlations with
participant ratinos

Bias

rniM

strong

patient
self-reoort

patient facial
actions

1.00
.90
(.001)

1.00

mild

.11
(2 1 )

.18
(.04)

1.00

strong

.09
(.33)

.060
(.50)

.95
(.001)

1.00

.72
(.001)

.85
(.001)

.13
(1 4 )

.001
(.99)

1.00

.71
(.001)

.88
C001)

.28
(.001)

.16
(0 7 1 )

.87
(.001)

.

Bias

Participant
ratings
self-report

facial actions

1.00

53

Table 7
Intercorrelations among sensitivity, bias and correlation measures for vouna
children with significance level fo) in parentheses
Sensitivitv

mild

strona

Bias

mild

Correlations with
oartidoant ratinos

strono

‘

Sensitivity
mild

oatient
self-reoort

oatient fadal
actions

1.00

strong

.74
(.001)

1.00

mild

.31
(.08)

.21
(.25)

1.00

strong

.29
(.10)

.08
(.67)

.98
(0 0 1 )

1.00

.46
(.01)

.79
(.001)

.15
(.42)

-.01
(.99)

1.00

.48
(.01)

.86
(.001)

.21
(:?4)

.08
(.65)

.84
(.001)

Bias

Participant
ratings
self-report

facial actions

1.00

54

Table 8
Intercorrelations among sensitivity, bias and correlation measures for middle
children with significance level fol in parentheses
Sensitivitv

mild
Sensitivity
mild

strona

Bias

mild

strona

Correlatians with
oarticioant ratinas
oatient
self-reoort

oatient facial
actions

1.00

strong

.83
(.001)

1.00

mild

.07
(.70)

.28
(.11)

1.00

strong

.05
(.76)

.15
(.41)

.95
(.001)

1.00

.33
(.05)

.45
(.01)

.10
(.55)

.07
(.71)

1.00

.43
(.01)

.71
(.001)

.54
(.001)

.45
- (0 1 )

.72
(.001)

Bias

Participant
ratings
self-report

facial actions

1.00

55

Table 9
Intercorrelations among sensitivity, bias and correlation measures for old
children with significance level ( o ) in parentheses
Sensitivitv

mild

strono

Bias

mild

Correlations witti
oarticioant ratings

strona

oatient
self-reoort

oatient facial
actions

Sensitivity
mild

1.00

strong

.58
(.001)

1.00

mild

-.16
(.37)

.20
(.28)

1.00

strong

-.01
(.97)

.07
(.71)

.92
(.001)

1.00

.24
(.19)

.51
(.002)

.21
(.25)

.05
(.68)

1.00

.23
(.19)

.76
(.001)

.47
(.01)

.33
(0 6 )

.500
(.01)

Bias

Participant
ratings
self-report

facial actions

1.00

56

Table 10
Intercorrelations among sensitivity, bias and correlation measures for adults with
sionificance level foi in parentheses
Sensitivitv

mild

strong

Bias

mild

strona

Correlations with
participant ratinas
patient
self-report

oatient facial
actions

Sensitivity
mild

1.00

strong

.88
( 001 )

1.00

mild

.04
(.82)

.18
(.33)

1.00

strong

.19
(.29)

.18
(.32)

.91
(.001)

1.00

.39
(.02)

.47
(.01)

-.12
(.51)

-.23
(.19).

1.00

.64
(.001)

.72
(.001)

.18
(.31)

.23
(.20)

.52
(.01)

Bias

Participant ratings
self-report

facial actions

1.00

57

Correlations between participant ratings and oatient self report and facial actions
As another means of evaluating the sensitivity of observers to variations
in patients’ pain, the correlations between the participants' ratings and both the
patients' self-report of pain (SR) and their facial actions (FACS) index were
calculated. A separate correlation coefficient was computed over all 90 target
trials for each participant. The means of these correlations for each group are
presented in Table 11. A 4 (age) X 2 (sex) MANOVA was computed for the selfreport and facial action correlations. Similar to the results for sensitivity, a main

Table 11
Mean correlations between participants* ratinas and oatient
self-reoort and facial actions
Aaearouo

N

Correlations with oarticioant
ratinas
Patient Self
Reoort

Patient Facial
Actions

Young
0.18
Male
16
0.23
Female
17
0.26
0.29
Middle
0.40
Male
17
0.44
Female
0.41
0.44
18
Older
Male
0.46
16
0.50
0.45
17
0.48
Female
Adult
0.47
Male
17
0.50
0.49
Female
16
0.49
all 2 < .05
note: the correlation between patients' self-report and facial
actions = 0.61.

58

effect of age was observed for both self-report F (3,1 2 6 ) = 53.80, e = .001 and
facial actions F (3. 126) = 37.38, g = .001. Tukey-HSD post-hoc analyses
indicated that for the correlations between participant ratings and both SR and
FACS, the youngest group of children differed from the other three groups
(Table 12). The adults and older children also differed significantly from the
middle children on SR correlations. Also note that the participants' ratings were
more strongly correlated with the FACS scores than they were for the SR, t (133)
= 14.78, g = .001.

Table 12
Mean correlations between participants' ratings and patients'
self-reoort and facial actions
Aoeoroup

N

Correlations with oarticioant
ratinas

Patient
Patient
Self Reoort
Facial Actions
.22
.26
Young Children
33
.40.
.44.
Middle Children
35
.49.
Older Children
33
45w,
Adults
48b
50.
33
a. b. numbers with the same subscripts in a column are not
significantly different a tg = .05.

The SR and FACS correlations were then correlated with sensitivity and
bias to ascertain the nature of relationship among all of these dependent

I

measures. These computations were for the total sample and for the four age

I groups. For the most part, these correlations were significantly related to both

59

sensitivity measures (Tables 6 through 10). Specifically, with one exception
(i.e., older children for mild signals), both sensitivity measures were strongly
correlated with the FACS correlations. Given that facial actions were rated, it is
not surprising that the measure of sensitivity to facial actions is strongly
associated. Similarly, the SR correlations were significantly related to sensitivity
to both mild and strong pain cues for all participants with the exception of the
older children rating the mild cues.
Bias was not significantly related to self-report for any groups (Tables 6 to
10). For facial action correlations, bias to mild signals was strongly and

I
I

positively related for the middle and old children and the total sample. Only for
i the middle children was there a significant relationship for the strong signal. The
i older children and total sample correlations approached significance (g = .06
; and B = .07, respectively). In all cases, the significant correlations between bias

I

I and participant correlations were positive; which indicates that as the

jI association between participant ratings and patient scores increased so did the

I degree of conservativeness of the observer. Conversely, participants who were
I
I less conservative were found to have pain ratings that were less related to the

I patients' facial action scores.
Experience of and Exposure to oain
Participants were asked if they or a significant other experienced any
chronic or recurrent pain problem. In the case of children, parents usually
provided the answers. Neither experience of nor exposure to pain were

60

significantly correlated with the other dependent measures computed. However,
there was a trend for reporting chronic pain in a significant other to be related to
the correlation between participants’ and patients’ ratings, r = .15, p = .08.
A 4 (age) X 2 (sex) MANOVA was computed with experience of and
exposure to pain as the dependent variables. A significant main effect of age
was observed for both experience F (3 ,120) = 2.81, p - .04, eta squared = .07,
and exposure to pain F (3 ,120) = 4.24, g = .01. eta squared = 1 0 . Post-hoc
analyses indicated that adults had been exposed to more pain than either the
young or middle children.
Self-oerceotion and emoathv analvses
All children in this study completed the full 12-item version of the IRI.
Analyses were conducted for the scales with 12 items as well as with all of the
negatively worded items removed. See Table 13 for the means for each age
group. A 3 (age) X 2 (sex) MANOVA was computed for empathie concern (EC),
perspective taking (PT), and personal distress (PO) with 3 and 4 items each.
The analyses indicated that there were no significant age or sex differences on
any of the scales. Because there were no systematic differences between ages
or sexes for either 3 or 4 item scales, all further discussion of the children's IRI
will be for the 10-item test.
The Self-perception scale data were entered into a 3 (age) X 2 (sex)
MANOVA. There was a significant effect of sex on global self-worth, with boys

61

Table 13
Mean IRI subscale scores for children
Males

Females

Total

EC (4 items)
Young
11.50
11.65
11.58
Middle
12.53
12.33
12.43
Older
11.50
13.24
12.39
EC (3 items)*
Young
10.31
10.24
10.18
Middle
10.82:
10.65
10.50
Older
9.43
11.24
10.36
PT (4 items)*
10.67
Young
10.00
11.29
Middle
11.94
11.06
11.49
Older
10.31
12.71
11.55
PT (3 items)
Young
7.75
9.06
8.42
Middle
9.17
9.59
8.78
Older
8.81
7.88
9.71
PD (4 items)
Young
10.88
11.88
11.39
Middle
11.88
11.33
11.60
Older
11.31
11.53
11.42
PD (3 items)*
Young
7.75
8.47
8.12
8.64
Middle
8.33
8.49
Older
7.63
8.24
7.94
note: there are no significant agegroup. sex or agegroup X sex
interactions for any of the scales
* scales that are retained on 10-item version

having higher scores (Table 14). Significant age differences were observed for
physical appearance, athletic competence and global self-worth. In all cases,
the youngest children had significantly higher scores than the oldest children;
however, their scores were not significantly different from the scores of the
middle group.

62

For physical appearance and GSW the middle children's scores did not differ
significantly from either the older or younger group. Only for athletic
competence did the oldest children have lower scores. Therefore, the young
children did not respond in a manner that was systematically different from the
middle group of children.

Table 14
Mean Self-perception subscale scores for children

Physical Appearance
Young
Middle
Older
Athletic Competence
Young
Middle
Older
Behavioural Conduct
Young
Middle
Older
Scholastic Competence
Young
Middle
Older
Social Acceptance
Young
Middle
Older
Global Self-Worth*
Young
Middle
Older

Males

Females

Tot^l

21.38
21.23
19.68

21.41
19.61
18.05

21.39.
20.40
8.85b

18.88
20.47
16.50

19.18
17.28
15.88

19.03,
18.83.
16.18b

18.75
17.76
18.06

19.76
19.78
18.59

19.27
18.80
18.33

18.88
19.18
17.94

19.12
18.67
17.53

19.00
18.91
17.73

19.25
18.12
18.56

19.47
18.94
18.59

19.36
18.54
18.58

22.56
21.53
19.56

20.71
19.11
18.88

21.61.
20.29
19.21b

b numbers with different subscripts, within a subscale,
are significantly different a tp = .05.

a.

63

It is important to note that both of the children's questionnaires have been
validated only on children as young as third grade In this study, these
questionnaires were administered to the youngest age group despite the lack of
validation. Nevertheless, as noted above, there were no significant main effects
of age on any of the subscales of the IRI and the six-year-old children did not
respond differently from the nine-year-olds on the SPPC. In addition, tfie young
children’s scores were not different from the oldest children on three of the six
subscales of the SPPC. Therefore, for the most part, the interview format
employed for young children yielded comparable outcomes across different
ages. Furthermore, the differences between the youngest and oldest children
may not be attributable to the manner in which the instrument was administered
but, may reflect differences in the children’s self-concept.
The subscale scores of both questionnaires were correlated with all
dependent measures for all participants. The results for adults and children are
presented in Appendix D, Tables 1 to 8. In general, there were no systematic
patterns across age groups. In fact, in many cases, even the direction of
correlation differed between ages for the children. With the possible exception
of PD, which was found to be related to the dependent measures, it is
conceivable that the occasional significant effects are due to chance. Even in
the case of PD, the results do not fall into a consistent pattern.

64

A comparison of two pain rating scales
The adults completed the rating task twice. The first time, they used the
same five-point scale as the children used when they viewed the tape. The
second time they used the affective rating scale used by the patients during the
physiotherapy session. The scale of affective pain intensity is a ratio-scaled
instrument; numerical values are associated with each of the verbal descriptors.
The data from the affective scale were dealt with in a number of different
ways. First of all, an average difference or discrepancy score for each
participant was calculated by subtracting the participants' rating on the affective
scale from the patients' self-report value for each of the 90 stimuli. A single­
sample t-test was conducted to determine whether the participants' ratings were
different from the patients'. The results indicated that the patients' ratings were
significantly higher than the participants', t (32) = 5.31, g - .001. Therefore, the
adults in this study exhibited an underestimation bias, relative to the patients.
The difference score was then correlated with the bias and sensitivity
indices. Bias for both stimulus intensities was strongly associated with the
difference but sensitivity was not (Table 15). The SP and IRI subscale scores
were also correlated with the difference between participant and patient ratings.
As can be observed in Table 16, PD was positively correlated with discrepancy.
That is, as personal distress scores increase, the greater is the degree of
underestimation. Notice that this parallels the PD data for bias (Table 10). The
results for the correlations between SP subscales and difference scores are also

65

Table 15
and oatients' ratinos and sensitivitv and bias measures foi

oarticioant - oatient difference
Sensitivity
mild
strong

.10 (.59)
.11 (.53)

mild
strong

.54 (.001)
.49(01)

Bias

Table 16
Correlations amono the difference between oarticioants’
and oatients' ratinos and IRI scales to)

Empathie Concern
Perspective Taking
Personal Distress
Fantasy

oarticioant - oatient difference
-.29 (.10)
-.14 (.43)
.42 (.02)
12(51)

comparable to the findings for bias (Appendix D, Table 1). This consistency
suggests that bias and discrepancy are measuring similar constructs.
The final way the affective scores were analyzed was to correlate them
with the other five-point rating scale as well as with the patients' SR and FACS
values. Thus, there were two measures of participants' pain ratings, a patient’s
self-report of pain and a measure of pain-related facial action. All of these
measures were then intercorrelated. The mean values for these correlations are
presented in Table 17. Single-sample t-tests were conducted on different pairs
of means (Table 18). The following results will refer to comparison pairs as

66

indicated on Table 18. The participants' two pain ratings were the most highly
correlated measures (r = .68). This correlation was significantly greater than any
other correlation calculated (comparisons 1 and 2). Both of the participants'
ratings scales were more highly correlated with patients' facial actions than with

Table 17
Mean correlations for self-reoort. facial action and 2 rating scales

5-point scale
affective scale
self-report
facial actions

5-point
scale
1.00
.68
.48
.50

affective
scale

patient’s
self-report

patient’s
facial actions

1.00
.42
.45

1.00
.61

1.00

Table 18
T-Tests to compare mean correlations for adults (mean

r )

Comparison pair
1. 5-point & affective (.68) vs 5-point & SR (.48)
2. 5-point & affective (.68) vs 5-jx)int & FACS (.50)

t-value
10.36
-10.03

p
.001
.001

3. 5-point & SR (.48) vs 5-point & FACS (.50)
4. affective & SR (.42) vs affective & FACS (.45)

1.66
2.32

.11
.03

5. 5-point & FACS (.50) vs SR & FACS (.61 )
6. affective & FACS (.45) vs SR & FACS (.61)

-15.06
7.76

.001
.001

7. 5-point & SR (.48) vs affective & SR (.42)

3.73

.001

8. 5-point & FACS (.50) vs affective & FACS (.45)
1.95 .06
note: df = 32; FACS = patients' facial action score on pain index;
SR = patients' self-report of pain; 5-point = participants' categorical
rating scale; affective = participants' 17 point scale (i.e., the same scale
as patients’ SR)

67

the patients' SR but was significant only for the affective scale (comparisons 3
and 4). That is. the ratings participants made, with either rating format, were
more highly correlated with the FACSs than with the patients' subjective report.
The correlation between the patients' facial actions and self-report of pain (r =
.61 ) was greater than the FACS correlations with either the five-point (r = .50) or
the affective scale (r = .45) (comparisons 5 and 6). The five-point rating scale
was significantly more correlated with patient SR (r = .48) than was the affective
scale (r = .42). t = 3.73. g = .001. The comparison of rating scales with FACS
approached significance; the 5-point scale ratings were more related to the
FACS score than were the affective scale ratings (comparison 8).
Action unit importance
The final, exploratory purpose of this study was to examine whether or not
children of different ages and adults use the same facial cues to rate others'
pain. The index used to categorize pain into three intensity levels included
measurements of brow lowering (AU4), orbital contraction (AUorb = AU6 + AU7),
levator contraction (AUlev = AU9 + AU10) and eye closing (AU43). This index
did not include mouth opening. Previous research has indicated that mouth
opening may occur during intense pain (LeResche, 1982) and may be a salient
cue to pain intensity. Therefore, measurements of opening of the mouth were
included in the following analyses.
To determine the relative importance of the different facial actions to pain
ratings hierarchical regression analyses were computed. In all cases, the first

68

step was to remove the variance accounted for by the participants and the
patients. The second step was to enter the five facial actions together as
predictors of participant ratings. The regression analyses (Tables 19 through
23) display only the second step of the analyses. Therefore, the R* reported are
the

change for the second step.
When all stimuli were combined, a pattern emerged. Note that brow

lowering and mouth opening were most influential predictors of participants'
ratings and that levator contraction was always negatively related to participants’

Table 19
Regression analvsis for action units predicting participants'
ratings for the total sample
Facial Action
S EB
B
brow lowering
.075
.003
orbital contraction
.014
.001
levator contraction
-.016
.004
mouth opening
.577
.020
eye closing
.014
.002
R^ = .24, F (5, 12053) = 760.83, g < 001

a
.292
.100
-.041
.246
.047

t
27.82
10.08
-3.86
29.25
4.83

B
.001
.001
.001
.001
.001

Table 20
Regression analvsis for action units predicting participants’
ratings for the vouno children
Facial Action
brow lowering
orbital contraction
levator contraction
mouth opening
eye closing

B

SEB

.058
.001
-.006
.441
.006

.006
.003
.010
.047
.006

a
.211
.003
-.014
.175
.022

J

9.11
.12
-.61
9.42
1.06

B
.001
.91
.54
.001
.29

69

Table 21
Regression analvsis for action units predicting participants'
ratinos for the middle children.
Facial Action
a
SEJB
.078
brow lowering
.005
orbital contraction
.016
.003
levator contraction
-.020
.008
mouth opening
.595
.038
.016
.004
eye closing
.27. F(5, 3142) = 235.20, g < .001

a
.305
.115
-.051
.256
.066

t

15.14
6.03
-2.50
15.84
3.54

.001
.001
.01
.001
.001

Table 22
Regression analvsis for action units predicting participants*
ratinos for the older children
Variable
SÊ-B
B
brow lowering
.086
.005
orbital contraction
.018
.002
levator contraction
-.020
.008
mouth opening
.652
.035
.007
.004
eye closing
E = 34, F(5, 2962) = 310.00, g < .001

a
.353
.141
-.051
.294
.029

t

17.94
7.62
-2.58
18.66
1.59

B
.001
.001
.01
.001
.11

Table 23
Regression analvsis for action units predicting participants’
ratinos for the adults
Variable
B
Si_B
.078
brow lowering
.005
orbital contraction
.020
.002
levator contraction
-.020
.008
.619
mouth opening
.036
.017
.004
eye closing
E = 32, F(5, 2962) = 282.62, g < .001

a
.313
.152
-.050
.271
.071

t
B
15.68 .001
8.08 .001
-2.48 .01
16.97 .001
3.87 .001

70

ratings. For the total sample, adults and middle children, all action units
significantly predicted ratings. The older children's ratings were predicted by all
actions except eye closure. The young children’s ratings were only predicted by
brow lowering and mouth opening.

71

DISCUSSION
In the first part of this section, the results directly associated with the
hypotheses will be summarized briefly and in the order they were initially
presented. Then, there will be a discussion of the use of the term "accuracy".
Accuracy can be interpreted in different ways depending on the type of data
collected. A more detailed discussion of the results will follow. Finally, more
general topics that may bear on the observations will be discussed.
Summary of results
The first hypothesis predicted that decoding ability would increase with
age. In terms of discrimination, it was expected that sensitivity would improve
with age. However, a specific effe^ of age on bias was not predicted. The
results of this study did support the hypothesis that sensitivity would increase
with age. Sensitivity increased throughout childhood, and the adults' sensitivity
appeared to be higher than the oldest children, although the difference was not
statistically significant. With respect to bias, overall, the participants in this
study demonstrated a conservative bias. The effect of age on response bias,
however, was less clear. Nine- and twelve-year-old children were more
conservatively biased than the other two age groups at the mild pain intensity
level.
It was hypothesized that participants scoring higher on social acceptance
and empathy would be more accurate than those scoring lower on these
variables. There were no systematic relationships between any of the
personality measures and sensitivity or bias. However, the personal distress

72

dimension of empathy was related to bias in the opposite direction for older
children and adults.
Previous research on the effects of gender on decoding ability has t)een
inconclusive. Studies have either found no gender effects (e.g., Tremblay et al.,
1987) or they have found that females exhibit superior performance (Philippot &
Feldman, 1990). In general, the results of this study were in the predicted
direction, but none were significant.
Finally, there was a comparison among age groups of the most influential
facial cues used when judging facial displays of pain. There were no specific
predictions proposed regarding the relative importance of facial cues for
participants of different ages. Brow lowering and opening the mouth were the
most influential cues for all ages of participants. These were the only significant
facial actions predictive of young children’s pain ratings. Adults and middle
children relied on all cues to rate pain. Eye closure was not a significant
predictor for older children. With the exception of the youngest children, levator
contraction was a significant negative predictor of pain ratings.
Judgment accuracv
Typical decoding studies analyze data in terms of accuracy. In this study,
accuracy of observers’ ratings was of interest. Therefore, the first part of this
section will deal with the different ways in which the term "accuracy" can be
interpreted. Accuracy is defined in typical recognition experiments as the
agreement (proportion or frequency) among observers or with the experimenter’s

73

judgment of what is the correct emotion for the given situation (e.g., Kirouac &
Dore, 1985; Walden & Field, 1990). The definition of accuracy is comparable in
some pain studies. For example, Galin and Thom (1993) and
Hadjisdtovropouios et al. (1996) defined accuracy as the frequency of correct
labeling of video excerpts according to categories (e.g., no pain, genuine, or
dissimulated).
in this study, accuracy can be interpreted in three ways. In SOT, P(A)
(i.e. discriminability or sensitivity) is a measure of accuracy in interpreting the
presence of a signal (hit) relative to false alarms. However, accuracy can also
be the degree of relationship (i.e., correlation) between observers’ pain ratings
and patients’ ratings of pain experienced. Finally, accuracy could also be
reflected in the discrepancy between patients’ self-reports and observers’
ratings. Given that the term accuracy’ has three possible interpretations, the
use of it is ambiguous. Therefore, in the subsequent sections, the three will be
referred to by the specific labels employed in the results section: P(A) or
sensitivity, patient-participant correlations and discrepancy between ratings.
Sensitivitv
Possibly the most important and clearest finding was that sensitivity to
facial expression of pain did increase with increasing age, with the adults
approaching maximum sensitivity for the strong signal. This finding is consistent
with literature on the recognition of facial expression of emotions reviewed in the
first chapter (e.g., Gosselin, 1995; Kirouac & Dore, 1985). Note, however, that

74

the experimental tasks are different. In the previous research, participants
discriminated among different emotional expressions, whereas in this study,
participants discriminated different intensities of the same type of expression.
The significant improvement in sensitivity with age during childhood could
be due to a variety of factors. It is possible that with increasing age. children
become more sensitive to pain cues simply as a function of having more
experience with or exposure to others in pain. Although the results of this study
indicated that adults were more likely than young children to report having a
family member with a chronic or recurrent pain problem, there was no increasing
tendency to report this with age during childhood. In all likelihood the question
was not specific enough to determine whether exposure to pain is age-related.
A second possible reason for improved performance between six and
twelve years of age may be due to an increased ability to discriminate the most
important facial cues. Gosselin (1995) observed that between the ages of five
and eight, children became better at making finer discriminations between
different facial expressions. The ability to make finer discriminations with age is
also consistent with the results of the regression analyses in this study which
indicated that younger children did not use all of the cues available to them to
make their ratings.
A related possibility is that pain decoding may parallel cognitive
development. However, there were a few six-year-olds who were more sensitive
than the least sensitive adults and old children. It is unlikely that those young

75

children display other cognitive skills equivalent to children in grade six or
introductory psychology students.
It has been suggested that the ability to discriminate facial expressions
coincides with development of the frontal lobes. That is, decoding performance
will continue to improve until the frontal lobes mature at approximately 14 years
of age (Kolb et al.. 1992). Therefore, the improvement throughout childhood
would be consistent with this prediction.
The results of this study demonstrate that sensitivity to facial expression
of pain improves during childhood. However, from this study, it is not possible to
establish which of the above explanations is most viable nor is it possible to
disentangle the relative contribution of experiential, cognitive and maturational
variables to the development of sensitivity.
The observed lack of a significant difference in sensitivity between older
children and adults has two viable explanations. First of all, it is possible that
children reach adult levels of pain discrimination by late childhood. It should be
noted here that adult performance was not perfect, but it was very high. As
indicated previously, there is some debate about whether adult decoding ability
is attained prior to adolescence or whether improvement continues beyond that
age. Kolb et al. (1992) suggested that recognition performance should improve
until it reaches the maximum level at approximately 14 years of age.
The second reason for no significant difference between the older
children and adults concerns demographic differences between the groups.

76

SES differed among groups with only the youngest children being significantly
different from the adults; but. SES was significantly correlated with sensitivity
only for older children. Therefore, SES data did not parallel the differences
observed in sensitivity. Furthermore, SES was not consistently related to any of
the other dependent variables. Although SES, per se, could not account for the
differences in sensitivity t)etween adults and children, the adults' parents did
differ from the children’s parents in two major ways. First of all, the children had
a higher proportion of their parents in health, education and counselling
occupations. In addition, a higher percentage of the children's parents had
obtained a university degree. Therefore, it is possible that specific educational
or occupational differences could have contributed to relationships between age
and sensitivity. That is, the adults might have been significantly different in
sensitivity if they had similar family backgrounds. In a decoding study with
adults, Kirouac and Dore (1985) observed that significant education interactions
reflected a "complex interplay” of factors that influence judgment accuracy.
Bias
Overall, participants' ratings were biased in the conservative direction. A
conservative bias indicates that participants were unlikely to report pain unless
they were sure that there was some degree of pain present. On the other hand,
a liberal bias would have indicated that participants were inclined to report pain
if uncertain. Given that it was a clinic situation, and the participants knew that all

77

of the patients had injured shoulders, one might expect a report of pain when
uncertain rather than the opposite, at least among adults.
There were no main effects of age or gender on bias. However, there
was an interaction between stimulus intensity level and age. The middle and old
children were more biased at the mild level. When the stimuli were mild, these
children were less likely to report that the patient was experiencing pain. It is
difficult to speculate about why children of these ages would be different from
younger children and adults.
The adults underestimated pain when using the same affective scale as
the patients. This observers' underestimation bias of pain has been reported in
the past (e.g., Prkachin, Berzins & Mercer, 1994). Although this has been
referred to as an underestimation bias in the literature, it is not bias in the signal
detection tradition. An underestimation bias reflects a difference between raters'
judgments, whereas in SDT, bias requires the calculation of hit and false alarm
probabilities.
The bias measures, based on the 5-point categorical scale, were most
correlated with the discrepancy between patients’ self-report and the
participants' ratings on the same affective scale. Although the adults
underestimated the patients' pain, it was the discrepancy in ratings that was
most correlated with response bias. Although bias, as computed in this study is
different than bias when evaluated as a measure of discrepancy, there appears
to be some similarity in the concepts. For example, a comparable pattern of

78

results occurred for the correlation between the personality measures and
discrepancy or bias. In fact, depending on whether or not a researcher is
interested in accuracy of ratings or response tendency in the signal detection
tradition, either method of reporting bias could be appropriate.
Personality variables
The focus of this study was on the perception of pain in others;
personality variables were considered supplementary or ancillary. Nevertheless,
as demonstrated in the Introduction, there is evidence in the literature that
justifies the expectation that measures of personality or personal characteristics
would be related to nonverbal decoding ability.
To assess social competence and empathy with comparable measures for
such disparate age groups is difficult. Other techniques (e.g. sociometrics, role
playing) were simply unavailable due to financial, time and seasonal constraints.
The additional time necessary to collect such social skill data was not warranted
given the secondary nature of this information. Furthermore, most sessions
were conducted in children's homes during summer vacation. Therefore, in
order to keep the intrusion to a minimum, two fairly brief questionnaires were
administered.
With few exceptions, the personality variables examined in this study
were not significantly related to the participants’ perception of facial expression
of pain. The lack of interrelationships is especially conspicuous for the
sensitivity measures and for the correlations between participant ratings and

79

patient SR and FACS which were most directly associated with age. It is
interesting that the few significant effects involving personality measures tended
to be associated with bias. The foregoing results taken together suggest that
bias and sensitivity reflect different aspects of stimulus judgments. This
independence of bias and sensitivity will be discussed further in the signal
detection section below.
The observation that few self-perception or empathy domains were
systematically associated with the dependent measures warrants discussion.
There were few domain scores that were related to dependent variables across
different age groups, especially among children. Even by excluding data from
the young children (given that it is debatable that their questionnaire data be
included), there were not many consistencies in findings involving self-reported
personal characteristics between the other two groups. This general lack of
consistency does not lend itself to easy interpretation. It is, of course, possible
that children of these ages are very different from each other. But, other
explanations of the general lack of patterned responses exist. Perhaps the
children’s sample was so homogeneous that there was little variability in SPPC
and IRI scores. It is also possible that the small number of significant
correlations is due to chance.
Emoathv. For the adults in this study, only one dimension of empathy
was correlated significantly with only one dependent measure. Specifically, PD
was positively correlated with bias The more empathie distress an adult

80

reported, the less likely the observer was to report pain (i.e., conservative bias).
The results differed for each age group of children. The PD scores for youngest
children were positively associated with the correlations between their pain
ratings and the patients’. The oldest children exhibited the greatest degree of
relationship between empathy domains and bias: All of the empathy scores
were negatively correlated with bias. The more empathy reported, the less
conservative was the response tendency. Given that the oldest children and
!

I

adults were most similar on the dependent variables, it is curious that the PD

!

I

relationship with bias is in the opposite direction for the oldest children and
adults. Is it possible that adults with high PD scores have developed a
defensive reaction to another person in pain?
In contrast to other studies of empathy (e.g., Bryant, 1982), there were no

!

i

I

significant age or gender differences on any of the dimensions assessed by the
IRI. The IRI is a multidimensional instrument that measures a general selfreported response style when others are upset or need help. The affective and

i

cognitive responses may be different as well. That is, one can understand what
emotion another person may be experiencing without vicariously experiencing it

j

oneself. The use of emotion labels is a cognitive skill that is highly dependent

\

on verbal ability. Furthermore, the ability to vicariously experience another's

3

state may be independent of such a cognitive skill, as suggested by Davis
(1983). Therefore, the fact that empathy is usually operationalized by way of
verbal methods may be a confound. Many researchers have indicated that the

81

results of a paper and pencil test may not be accurate predictors of behaviour
that is presumably dependent on the construct assessed. Consequently, there
is a need for convergent validity of SR data. Eisenberg and her colleagues
(e.g., Fabes, Eisenberg & Eisenbud, 1993) have begun to employ physiological
measures (e.g., heart rate) to assess the correspondence between bodily
responses to described or viewed situations and empathy/sympathy
questionnaire results. According to Bryant (1982), a more thorough
understanding of empathy will be gained by employing multiple measures:
questionnaire data, physiological responses, facial expressions and behavioural
responses such as helping.
Social Competence. In this study, social acceptance was not associated
with any dependent measure of decoding performance. There are many
possible explanations for this finding. The first and simplest is that social
acceptance, as measured by the Self-Perception Profiles, may not be related to
sensitivity to pain expressions.
As described in the introduction, Feldman and his colleagues found that
social competence, as measured by the CBCL, was related to decoding ability.
Social competence may be related to decoding accuracy in typical tasks but this
task was not typical nor were the dependent measures. For example, P(A) is
considered to be a measure of accuracy but is not equivalent to the proportion of
correctly identified faces in an emotion decoding study. Perhaps social
competence is related to discriminating among universal facial expressions

82

when choosing among them, but may not be related to discriminating differing
intensities of the same emotional expression.
Social competence is assessed in a variety of ways (Bemdt & Burgy,
1995) that depend on different assumptions about it: sociometrics (e.g.. Walden
& Field, 1990; Rosenblum & Olson. 1997). ratings by peers (Parke et al.. 1997).
teachers (e.g.. Rosenblum & Olson. 1997) or parents; and behavioural
observation (e.g., Rosenblum & Olson. 1997). Obviously, social competence as
a social skill construct is either poorly or inconsistently defined.
The SPPC was chosen for this study because it has been observed that
social acceptance, on the SPPC. is related to other indicators of social
competence (e.g.. sociometric status; Hymel, et al.. 199). Furthermore. Pope
and Ward (1997) included social acceptance as part of a composite measure of
social competence.
In all likelihood social acceptance and social competence are related or
overlap but not synonymous. Therefore, it may not have been appropriate to
use social acceptance as an indicator of social competence.
Self-oerceotion domains. The social acceptance dimension of the self­
perception profiles was the reason the Self-Perception Profiles were selected.
The complete profiles were administered. To administer only the items of
interest could possibly have invalidated the questionnaire. Although the social
acceptance scale was of primary interest, analyses were conducted for all
domains. The self-perception domains were, for the most part, not significantly

83

associated with sensitivity or participant-patient correlations. However, some of
them were correlated with bias. Given that few correlations were significant and
nothing in the literature indicated they would be associated with decoding, they
will not be addressed further.
Gender
With one exception (i.e., GSW), there were no gender effects observed in
this study. Although most of the findings were in the expected direction, none
reached significance. This result is consistent with Missaghi-Lakshman and
Whissell (1991) who found no differences between boys and girls in decoding
and with Tremblay et al, (1987) who also reported no gender effects. But, it is in
contrast to the findings of Shortt et al. (1994) who did observe effects of gender
on recognition ability. Phillipot and Feldman (1989) had found an age by gender
interaction with boys' decoding performance delayed relative to girls. Therefore,
previous research could be considered inconclusive.
There are a few possible explanations for the lack of effect of gender.
Males and females may not differ in their perception of facial expression of pain.
It is also possible that the nature of the task in this study was sufficiently
different that the results of other decoding research are not relevant. For
example, in most other decoding research, stories or situations are read with
characters described or videos displayed. The participants then give an emotion
label or choose one from those provided. The signal detection method used
here provides the type of expression and the participant indicates the intensity

84

(from none to a lot) of one type of expression. Therefore, males and females
may respond differently on the typical decoding task than on a signal detection
task.
It is also possible that males and females may have responded differently
in past research because of the effects of the experimental situation or materials.
Gender effects in past research may be attributable to differences in reliance on
or use of the experimental context itself. For example, there may be cues in the
instructions, descriptions or videos used that differ in salience for boys and girls.
If experimental information Is more salient to one group relative to another, then
the ratings of that information may differ. The fact that the context for all
excerpts used in this study was the same (i.e., a clinic) may have eliminated any
effects.
In other decoding research, the time to complete the task is usually longer
than the 13 minutes required in this study. Girls may find decoding tasks more
interesting and may be able to attend to these tasks for longer periods of time.
As support for this possibility, only one woman and no girls commented that the
task in this study was boring. A few of the youngest girts asked if they could see
more faces. In contrast, quite a few boys, almost all in the 9-year-old group, said
It was boring and had more trouble paying attention. Even though it may have
been less interesting to some of the boys, they were able to attend to the whole
tape. If this phenomenon existed in a more demanding or time-consuming study.

85

it is possible poorer performance could be attributed to lack of interest or
attention rather than poorer discrimination ability.
There are two related additional possible reasons for the lack of gender
effects. The first is that the children's sample was biased in that a high
percentage of the children's parents had university degrees. It is possible that
more highly educated people socialize their children to be more androgynous.
Secondly, parents now may be raising their children differently than when some
of the earliest work was conducted. Consequently, boys' and girls' socialization
experiences may be more similar.
Signal Detection Theory Analyses for Pain Judgment
Ellermeier (1997) indicates that signal detection methodology is wellsuited to studies investigating differences in pain judgment. Using this
approach, he reanalyzed data that had demonstrated differences between men
and women in their ratings of pain. He hypothesized that the apparent
differences were not due to a difference in sensation, rather, they reflected a
difference in willingness to report pain (i.e., a response bias). Therefore, he
suggested treating pain ratings (i.e., category ratings) as confidence ratings
rather than as sensation magnitude. Two parameters could then be derived.
The first would reflect sensory discrimination (i.e., sensitivity) and the other
would reflect how the participants used the response categories (i.e., bias). His
reanalysis confirmed that men and women were equally "sensitive" but there
were sex differences in bias with different stimulus intensities. Kemperman et al.

86

(1997) also indicated that signal detection theory can be useful to distinguish
sensory and psychological factors in the report of pain by different patient
groups. If signal detection methods are suitable for the analysis of pain rating
data for the person experiencing pain then, they should also be appropriate to
analyze observers’ ratings.
See. et al.. (1997) evaluated various measures of bias to determine which
is the best under certain conditions. Given their analysis criteria (e.g..
independence of bias and sensitivity, effect of nonperceptual manipulations,
etc.). they recommend that, for nonparametic data. Bo is the most effective bias
measure. Furthermore, sensitivity and bias measures should be paired
depending on their categorization as parametric or nonparametric. Therefore. A
or P(A) is the appropriate (i.e., nonparametric) measure of sensitivity to
correspond to Bq. P(A) should be affected by signal salience, a perceptual
manipulation whereas Bo should be more influenced by nonperceptual factors
(e.g., personality variables).
In this study, mild and strong intensity pain expressions were the
differences in the salience of the signal. The fact that there was a significant
effect of intensity of facial expression, demonstrates that the faces were
perceived as different.
Also according to signal detection theory, sensitivity should not be
affected by nonperceptual manipulations or variables. It is logical, therefore.

87

that personality variables were essentially unrelated to the perceptual measure
but were more likely to be associated with response bias.
Methodolooical Issues
Selection of dependent measures. The above discussion demonstrates
the suitability of signal detection methods for pain judgment studies.
Nevertheless, additional or alternate measures would be appropriate depending
on the purposes of a study. In this study, P(A) and participant-patient
correlations appeared to reflect similar aspects of expression judgment. A
significant effect of age was observed for both. Similarly, they had parallel
patterns of relationships to the empathy and self-perception dimensions. As
suggested previously, bias and discrepancy also appeared to reflect similar
aspects of response tendency.
Selection of rating scale. From the results with adults, it is suggested that
a 5-point scale be employed for most purposes. In comparison to the affective
rating scale, the 5-point scale was more closely correlated with patients' self
reported pain and somewhat more correlated with facial actions. Furthermore,
the 5-point scale was highly correlated with the participants’ ratings on the
affective scale. On the other hand, the use of the affective scale would be more
suitable if discrepancy scores were of primary interest.
Facial actions
Prkachin and Mercer (1989) created the pain index used in this study
from a combination of 4 movements of the middle and upper face. Mouth activity

88

was excluded from the index because it was not reliably associated with selfreports of pain. Other investigators have found that a variety of facial actions
around the mouth may occur during pain. For example, LeResche (1982) very
clearly found that mouth activity was an aspect of the pain expressions she
coded. It is possible that the photographs employed in her study were taken in
situations that elicited a more intense pain response or emotions or motivational
states in addition to pain. Despite the fact that the patients with the shoulder
injuries reported intolerable pain for some movements, it is unlikely that the
therapist-induced pain would be as severe as childbirth or leg surgery without
anesthesia. Therefore, indices of facial pain in clinic settings may not include all
actions that could be present during more intense pain. In support of this
position is the observation that exaggerated pain expressions also tend to
include facial actions involving the mouth (Galin & Thom, 1993)
In this study mouth opening and brow lowering were the facial cues most
predictive of pain ratings by observers. In general, the youngest children used
fewer facial actions than participants of other ages to make judgments of pain.
This finding is analogous to Gosselin’s (1995) results that indicated that older
children made better use of all of the facial actions present in an emotional
display. Younger children may either attend to fewer facial cues or use fewer
cues reliably.
This study did not ask subjects to discriminate pain from other facial
expressions. The prototypical pain expression shares some facial actions with

89

other prototypical emotion expressions (e.g.. brow lowering occurs with both
pain and sadness^; Gosselin, 1995). Gosselin (1995) claimed that age-related
differences in performance can be attributed to improvements in discriminating
the critical action units associated with an emotion. Sullivan, Kilpatrick and
MacDonald (1995) state that in the course of normal daily interactions, children
are exposed to facial expressions that may include components of mixed
emotions. That is, expressions often are not pure or do not match the prototype
of an emotional expression. Together these lead to the suggestion that it would
be informative to examine the ability to discriminate pain expressions from other
facial expressions. From that, it would be possible to determine which facial
action units, alone or in combination, were critical for making a judgment of pain.
Future Directions
Future directions will be addressed from three main perspectives:
developmental and methodological topics, and the pain communication model of
Prkachin and Craig (1995).
Developmental Issues. The present study demonstrated that the age of
the observer was an important factor in accuracy of pain judgments. Accuracy
here refers to both sensitivity to differing intensities of pain expression as well as
the degree of relationship between participant and patient pain ratings.
It is recommended that a wider age range of observers be studied to
further clarify the relationship between development and accuracy. The

^ There were occasions, with the youngest children, that they provided an emotion label rather
than indicate a degree of pain. One girl in particular said that most of the pain faces were sad
but it didn't look like they hurt during the movement.
90

responses of adolescents or children younger than kindergarten could be
investigated. For example, preschoolers’ sensitivity could be assessed with a
simpler format (e.g., Did it hurt? yes/no or no, a little, a lot). Although this study
demonstrated an age-related change in sensitivity to pain expressions, it was
only possible to speculate as to why this occurred. Was it due to more
experience with pain or with a variety of other facial expressions, general
cognitive development or more specific neurological development? In order to
examine the relationship between the development of pain expression decoding
ability and any of these variables, future research will need to be precise in
specifying or measuring the predicted correlates.
Older adults may perform differently than university students.
Development continues throughout the lifespan. Therefore, adults of different
ages or those with different experiences may interpret pain displays in diverse
ways.
The children’s sample was fairly homogeneous. Therefore, a more
heterogeneous sample of children may yield different results (i.e., more variable
SES, parents’ education or occupation). Alternatively, a sample could be
selected on the basis of specific criteria. For example, children with high versus
low scores on a variable of interest could be chosen. The variable of interest
could be one that would be expected to be related to experiencing or being
exposed to pain (e.g., sport participation or health status). It is also possible that

91

social skills variables are related to pain expression decoding and different
measures could be employed.
The oldest children and adults had PD and bias correlations that were in
the opposite direction. An investigation of adolescents could further examine the
relationship between these variables. The first suggestion is to examine
adolescents’ responses on both the adults' and children's versions of the IRI.
Although the children’s items were designed to parallel the adults’, differences
could be attributed to the items on the scale. It is conceivable that adolescents
and adults do respond differently to others in distress. Adolescents may not
have very well-established strategies for responding to distress. It is also
possible that it is only the self-report responses that are different not the actual
response to them. This would suggest that measures of convergent validity
(e.g., physiological variables such as heart rate) would be useful to assess ttie
relationship between empathie distress and physiological response when
evaluating decoding ability.
Some of the action units present in pain expressions are also observed in
other facial expressions of negative emotions. Camras, Sullivan and Michel
(1993) indicated that the pain-distress expression is very similar to the anger
expression. Similarly, components of pain are observed in sadness and disgust.
Therefore, in contrast to the present study in which the intensity of one
expression was discriminated, a future study could examine the discriminability
of pain from other expressions. First of all, it was assumed when undertaking

92

this study that findings from the general emotion expression literature applied to
pain expressions. This assumption may have been faulty. Pain decoding may
develop differently than it does for other expressions. Secondly, discriminating
between expressions may be different than discriminating between different
intensities of the same expression. By examining pain in comparison to other
expressions, one would gain a clearer understanding of the critical action units
observers use when making judgments.
Methodological Issues. The youngest children performed at significantly
better than chance and the adults approached maximum sensitivity. Although
pilot testing indicated that five year olds were able to use the 5-point rating scale
it may have been too precise for them. Perhaps a 3 - or 4-point scale would have
yielded higher sensitivity scores. In addition, future research should employ a
Faces scale, as described in the introduction, to determine if children could use
it to rate others’ pain or whether they would treat it as a matching task.
Future research could further investigate the relationships among the
different dependent measures employed and determine under which conditions
they would be most appropriate. Although discrepancy and Bo are significantly
correlated, the latter is a better measure of response bias than discrepancy.
However, response bias may be interpreted differently depending on the position
of a point relative to the ROC curve. In some circumstances, the average
difference between patients' and observers’ scores may be the preferred
measure. For two reasons, it is probably easier for most readers to understand.

93

It is a simple mathematical function (i.e.. subtraction) and it is expressed in units
of the ratio scale that directly correspond to the ratings made by both observers
and patients.
Empathy could be investigated from another perspective, or in
combination with the IRI. For example, a physiological measure such as skin
conductance, heart rate or facial EMG could be employed during the
presentation of excerpts (McHugo & Smith, 1996). The use of physiological
measures would be especially important as a way to validate the PD scale
because it is intended to address self-reported distress. The use of a
physiological measure would permit an examination of whether or not the self
report of distress is related to physiological arousal.
Although social acceptance was not related to any of the dependent
variables, that does not mean that social competence or some other social skill
is not. Facial expression encoding and decoding have been found to be related
to a variety of social variables. Also, facial expressions are important for
interpersonal communication. Furthermore, the expression of pain or distress
could be a more potent cue to act for observers (e.g., to protect oneself from a
threat, to help relieve another's suffering). Therefore, future research should
employ a different measure of social skill (e.g.. sociometric status) to better
understand how social variables could influence the sensitivity to facial
expressions of pain.

94

The pain communication model. This study focused primarily on the
decoding (C) component of the model (Prkachin & Craig, 1995); therefore, the
majority of comments t)elow will be limited to this aspect. Research on this
model is beginning to address which characteristics of the observer and of the
person being observed influence pain judgments. As indicated above, the age
of the observer was an important factor in determining sensitivity to differing
intensities of pain expression. Observers' age was also associated with the
correlations among participant ratings and patient self-report of pain and patient
facial actions.
Matheson (1997) found that her participants were "predisposed" to see
more pain in the faces of elderly patients than in younger patients. This study
focused on the age of the observer rather than age of the patient. Future
studies could examine differences in patients and observers of a wider age
range. For example, children may rate the pain of other children differently than
that of adults. Conversely, older adults, who may have witnessed or
experienced a greater variety of painful situations may also differ in their
responses. Their differences could be reflected in terms of sensitivity or
response bias.
It is likely that people with diverse pain experiences decode pain
expressions differently; which leads to the suggestion that participants for future
research could be selected on the basis of their experience of or exposure to
pain. The videotape for this study had patients with shoulder injuries. It is

95

possible that an observer with a similar injury history may respond differently
than one without such experience. This could extend to other painful conditions
as well; such as arthritis, headaches, back pain, cancer.
Hadjistavropoulos et al. (1996) suggested that future research should
address the possibility that "there may be some individual difference variable
that would make some judges better decoders than others” (p.257). Although
the results of the present study suggest that the PD domain of empathy may be
associated with response bias independent of accuracy, there were no selfperception or empathy variables that were systematically related to sensitivity.
Nevertheless, other individual difference variables (e.g.. alexithymia, depression,
anxiety, etc.) may be predictive of decoding performance.
Summary and Conclusions
In general, the results of this study demonstrate that the ability to
recognize facial expression of pain, and more specifically, to discriminate
different Intensities of painful expression, improves with increasing age. Despite
the fact that the method employed in this study to assess decoding performance
differed from that typically used in facial expression of emotion research, the
results are consistent with previous research concerning the effect of age on the
ability to recognize different emotional expressions. Given that the SDT
approach used in this study yielded parallel findings in the general facial
expression literature, this suggests that SDT methods could be more widely
employed in facial expression research.

96

The results of this study also highlight the importance of using a variety of
dependent measures to evaluate observers’ performance. Specifically,
judgment accuracy can be assessed by different means, and therefore, can be
interpreted differently. For example, although sensitivity, correspondence
between raters and discrepancy can all be measures of accuracy, they afford
different interpretations of observers’ performance.
The videotape created for this study is ideal for examining characteristics
of observers that influence pain ratings. The tape could t>e employed in a
variety of settings with many different participant populations. In addition, it is
fairly brief and yet, with 90 excerpts of three intensity levels, allows the collection
of a wealth of data for analysis.
Finally, it should t>e apparent that the signal detection approach
employed in this study is especially well-suited to the study of pain. As
previously indicated in this paper, this method is useful to examine the
independent effects of sensitivity (i.e., pain intensity) and bias (i.e., the
willingness to report pain) for the person experiencing pain. The present study
demonstrated the utility of this approach to evaluate observers’ judgments. It is
documented in the literature that patients and research participants are able to
suppress or exaggerate pain expressions and also that observers may expect
dissimulation (e.g., Poole & Craig, 1992). Therefore, the SDT approach would
allow investigators to dissociate intensity and bias when examining any
participant’s pain ratings.

97

REFERENCES
Abu-Saad, H.H. (1994). Pain in children: Developing a programme of
research. Disability and Rehabilitation. 16. 45-50.
Achenbach, T.M. & Edelbrock, O.S. (1982). Manual for the child behavior
checklist and child behavior profile. Burlington, CT: Department of Child
Psychiatry, University of Vermont.
Anand, K.J.S. & Craig, K.D. (1996). New perspectives on the definition of
pain. Pain. 67. 3 - 6 .
Austin, S., & Joseph, S. (1996). Assessment of bully/victim problems in 8
to 11 year-olds. British Journal of Educational Psvcholoov 66. 447-456.
Beck, L. & Feldman, R.S. (1989). Enhancing children's decoding of facial
expression. Journal of Nonveibal Behavior. 13 269-278.
Bemdt, T J. & Burgy, L. (1996) Social self-concept. In B A. Braken, et al.
Handbook of self-conceot: Developmental, social, and clinical considerations
New York, NY: John W iley & Sons.
Bieri, D., Reeve, R.A., Champion, G.D., Addicoat, L., & Ziegler, J.B.
(1990). The Faces Pain Scale for the self-assessment of the severity of pain
experienced by children: Development, initial validation, and preliminary
investigation for ratio scale properties. Pain. 41. 139-150.
Blishen, B.R., Carroll, W.K., & Moore, C. (1987). The 1981 socioeconomic
index for occupations in Canada. Canadian Review of Sociology and
Anthropoloov. 24. 465-488.
Boivin, M., & Hymel, S. (1997). Peer experiences and social self­
perceptions: A sequential model. Developmental Psvcholoov. 33 135- 145.
Bryant, B.K. (1982). An index of empathy for children and adolescents.
Child Development. 53. 413-425.
Bugental, D.E. Kaswan, J.W., & Love, L.R. (1970). Perception of
contradictory meanings conveyed by verbal and nonverbal channels. Journal of
Personality and Social Psvcholoov. 16. 647- 655.
Camras, L A., Sullivan, J., & Michel, G. (1993). Do infants express
discrete emotions? Adult judgments of facial, vocal, and body actions. Journal
of Nonverbal Behavior. 17 171-186.

98

Chun, D Y., Turner, J.A., & Romano, J.M. (1993). Children of chronic pain
patients: Risk factors for maladjustment. Pain. 52. 311-317.
Craig, K.D., Hadjistavropoulos, H.D., Grunau, R.V.E. & Whitfield, M.F.
(1994) A comparison of two measures of facial activity during pain in the
newborn child. Journal of Pediatric Psvcholoov. 19 305-318.
Craig, K.D., Hyde, S.A. & Patrick, C J. ( 1991). Genuine, suppressed, and
fàked facial behavior during exacerbation of chronic low back pain. Pain. 46.
161-172.
Custrini, R.J. & Feldman, R.S. (1989). Children's social competence and
nonverbal encoding and decoding of emotions. Journal of Clinical Child
Psvcholoov. 18. 336-342.
Davis, M.H. (1983). Measuring individual differences in empathy:
Evidence for a multidimensional approach. Journal of Personalitv and Social
Psvcholoov. 44. 113-126.
Donaldson, W. (1992). Measuring recognition memory. Journal of
Experimental Psvcholoov: General. 121. 275-277.
Egan, G. (1986). The skilled helper A svstematic approach to effective
helping (3"* ed.). Pacific Grove, CA: Brooks/Cole.
Eisenberg, N., Fabes, R.A., Muiphy, 8., Karbon, M., Smith, M., & Maszk,
P. (1996). The relations of children’s dispositional empathy-related responding
to their emotionality, regulation, and social functioning. Developmental
Psvcholoov. 3 2 .195-209.
Eisenberg, N., Fabes, R.A., Schaller, M , Carlo, G , & Miller, P A. (1991).
The relations of parental characteristics and practices to children’s vicarious
emotional responding. Child Development. 62 1393-1408.
Eisenberg, N., Fabes, R.A., Schaller, M. & Miller, P A. (1989). Sympathy
and personal distress: Development, gender differences, and interrelations of
indexes. In N. Eisenberg (Ed ), Emoathv and related emotional responses. San
Francisco, CA: Jossey-Bass.
Ekman, P. (1977). Biological and cultural contribution to body and facial
movement. In J. Blacking (Ed ), The anthropoloov of the bodv. London:
Academic.
Ekman, P., & Friesen, W. (1978). Facial action coding svstem. Palo Alto,
CA: Consulting Psychologists Press.

99

Ellermeier, W. (1997). On separating pain from the willingness to report it.
Behavioral and Brain Sciences. 20. 448-449.
Fabes. R.A. (1997) Personal communication.
Fabes, R.A., Eisenberg, N., & Eisenbud, L (1993). Behavior and
physiological correlates of children's reactions to others in distress.
Developmental Psvcholoov. 29. 655-663.
Feldman, R.S., White, J .B , & Lobato, D. (1982). Social Skills and
nonverbal behavior. In R.S. Feldman (Ed ), Development of nonveital behavior
in children. New York: Springer-Verlag.
Feldman, R.S., Philippot, P., & Custrini, R.J. (1991). Social competence
and nonverbal behavior. In R.S. Feldman and B. Rime (Eds ). Fundamentals of
nonverbal behavior. New York: Cambridge University Press.
Galin, K E., & Thom, B E. (1993). Unmasking Pain: Detection of
deception in facial expressions. Journal of Social and Clinical Psychology 12.
182-197.
Gosselin, P. (1995). Le développement de la reconnaissance des
expression faciales des emotions chez l'enfant. Canadian Journal of Behavioural
Science. 27. 107-119.
Grunau, R.V.E., Johnson, C.C., & Craig, K.D., (1990). Neonatal facial and
cry responses to invasive and non-invasive procedures. Pain. 42. 295-305.
Hadjistavropoulos, H.D. Craig, K.D., Hadjistavropoulos, T., & Poole
(1996). Subjective judgments of deception in pain expression: Accuracy and
errors. Pain. 65. 251-258.
Hager, J.C., & Ekman, P. (1997). The assymetry of facial actions is
inconsistent with models of hemispheric specialization. In P. Ekman & E.L.
Rosenberg (Eds ). W hat the face reveals: Basic and applied studies of
spontaneous expression usina the Facial Action Codings Svstem fFACSI. New
York: Oxford University Press.
Harter, S. M9851 Manual for the Self-Perceotion Profile for Children.
Denver, CO: University of Denver.
Heath, N.L., & Wiener, J. (1996). Depression and nonacademic self­
perceptions in children with and without learning disabilities. Learning Disability
Quarterly. 19. 34-44.

100

Heft, M.W., Gracely, R.H., Dubner, R., & McGrath (1980). A validation
model for verbal descriptor scaling of human clinical pain. Pain. 9. 363-373.
Hymel. S., Bowker, A., & Woody, E. (1993). Aggressive versus withdrawn
unpopular children: Variations in peer, teacher and self-perceptions in multiple
domains. Child Development. 64. 879- 896.
Johnston, C.C., Stevens, B.J., Craig, K D , & Grunau, R.V.E., (1993).
Developmental changes in pain expression in premature, full-term, two- and
four-month-old infants. Pain. 52. 201-208.
Johnston, C.C., Stevens, B.J., Yang, F., & Horton, L. (1995). Differential
response to pain by very premature neonates. Pain. 61. 471 -479.
Kemperman, I., Russ, M.J., Clark, W.C., Kakuma, T., Zanine, E., &
Harrison, K (1997). Psvchiatrv Research. 7 0 .175-183.
Kirouac, G. & Dore, F.Y. (1985). Accuracy of the judgment of facial
expression of emotions as a function of sex and level of education. Journal of
Nonverbal Behavior. 9. 3-7.
Kolb, B., Wilson, B , & Taylor, L. (1992). Developmental changes in the
recognition and comprehension of facial expression: Implications for frontal lot)e
function. Brain and Cognition 20. 74-84.
LeResche, L. (1982). Facial expression in pain: A study of candid
photographs. Jounal of Nonverbal Behavior. 7. 46-56.
Levensen, (1994). Emotional Control: Variations and consequences. In
P. Ekman & R. J. Davidson (Eds ). The nature of emotion: Fundamental
questions. New York, NY: Oxford University Press.
Manne, S.L. Jacobsen, P.B., & Redd, W.H. (1992). Assessment of acute
pediatric pain: Do child self-report, parent ratings, and nurse ratings measure
the same phenomenon? Pain. 48. 45-52.
Matheson, D. H. (1997). The painful truth: Interpretation of facial
expressions of pain in older adults. Journal of Nonveibal Behavior. 21. 223-238.
McHugo, G.J., & Smith, C A (1996). The power effaces: A review of
John T. Lanzetta’s research on facial expression and emotion. Motivation and
Emotion. 20. 85-120.

101

Missaghi-Lakshman, M. & Whissell, C. (1991). Children’s understanding
of facial expression of emotion: II. Drawing of emotion-faces. Perceptual and
Motor Skills. 72. 1228-1230.
Neeman, J. & Harter, S. (1986). Manual for the Self-Perception Profile for
College Students. Denver, CO: University of Denver.
Nelson C.A., & de Haan, M. (1996). Neural correlates of infants’ visual
responsiveness to facial expression of emotion. Developmental Psvchobioloov.
29, 577-595.
Odom, R. D., & Lemond, C M. (1972). Developmental differences in the
perception and production of facial expressions. Child Development. 43. 359369.
Parke, R.D., O’Neil. R. Spitzer, S., Isley, S. Welsh, M., Wang. S.. Lee, J.,
Strand, C., Cupp. R. (1997). A longitudinal assessment of sociometric stability
and the behavioral correlates of children’s social acceptance. Merrill-Palmer
Quarterlv. 43. 635-662.
Patrick. C.J.. Craig. K.D.. & Prkachin. K.M. (1986). Observer judgments of
acute pain: Facial action determinants. Journal of Persianalitv and Social
Psvcholoov. 50. 1291-1298.
Philippot, P. & Feldman. R.S. (1990). Age and social competence in
preschoolers’ decoding of facial expression. British Journal of Social
Psvcholoov. 29. 43 - 54.
Philippot. P.. Feldman. R.S., & McGee, G. (1992) Nonverbal behavioral
skills in an educational context: Typical and atypical populations. In R.S.
Feldman (Ed.), Applications of nonverbal behavioral theories and research.
Hillsdale, NJ: Lawrence Erlbaum Associates.
Poole, G.D. & Craig. K.D. (1992). Judgments of genuine, suppressed, and
faked facial expressions of pain. Journal of Personalitv and Social Psvcholoov.
63. 797-805.
Pope, A.W., & Ward, J. (1997). Factors associated wiht peer social
competence in preadolescents with craniofacial anomalies. Journal of Pediatric
Psvcholoov. 22 455-469.
Profyt, L. & Whissell, C. (1991). Children’s understanding of facial
expression of emotion: I. Voluntary creation of emotion-faces. Perceptual and
Motor Skills. 73. 199-202.

102

Prkachin. K.M. (1992a). The consistency of fadai expressions of pain: A
comparison across modalities. Pain. 51. 297-306.
Prkachin, K.M. (1992b). Dissociating spontaneous and deliberate
expressions of pain: Signal detection analyses. Pain. 51. 297- 306.
Prkachin, K.M., Berzins, S., & Mercer, S.R. (1994). Encoding and
decoding of pain expressions: a judgement study. Pain. 58. 253-259.
Prkachin, K.M. & Craig, K.D. (1985). Influencing nonverbal expressions of
pain: Signal detection analyses. Pain. 21. 399-409.
Prkachin, K.M. & Craig, K. D. (1995). Expressing pain: The
communication and interpretation of facial pain signals. Journal of Nonvertial
Behavior. 19. 191-205.
Prkachin, K.M., Currie, N.A., & Craig, K.D. (1983). Judging nonverbal
expressions of pain. Canadian Journal of Behaviour Science. 15. 409-421.
Prkachin, K.M. & Mercer, S.R. (1989). Pain expression in patients with
shoulder pathology: Validity, properties and relationship to sickness impact.
Pain. 39. 257-265.
Prkachin, K.M., Solomon, P S , Hwang, T. & Mercer, S.R. (1995) Does
experience affect iudoements of oain behaviour? Evidence from relatives of
pain patients and health-care providers. Unpublished manuscript.
Regan, D.T. & Totten, J. (1975). Empathy and attribution: Turning
observers into actors. Journal of Personalitv and Social Psvcholoov. 32. 850856.
Renouf, A .G ., Kovacs, M., & Mukerji, P (1997). Relationship of
depressive, conduct, and comorbid disorders and social functioning in
childhood. Journal of the American Adademv of Child and Adolescent
Psvchiatrv. 36. 998-1004.
Riggio (1992). Social interaction skills and nonverbal behavior. In R.S.
Feldman (Ed.), Applications of nonvert)al t)ehavioral theories and research.
Hillsdale, NJ: Lawrence Erlbaum Associates.
Rinn, W .E. (1991). The neuropsychology of facial expression. In R.S.
Feldman & B. Rime, Fundamentals of Nonvertaal Behavior NY: Cambridge
University Press.

103

Rosenblum, K.L., & Olson, S.L. (1997). Assessment of peer neglect In the
preschool years: A short-term longitudinal study. Journal of Clinical Child
Psvcholoov. 26. 424-432.
Rosenthal, R. (1982) Conducting judgment studies. In K. Scherer and P.
Ekman (Eds ), Handbook of methods in nonverbal behavior research. New York:
Cambridge University Press.
Russell, J.A. (1995). Facial expressions of emotion: What lies beyond
minimal universality? Psychological Bulletin. 118. 379-391.
Russell, J.A (1997). Reading emotions from and into faces: Resurrecting
a dimensional-contextual perspective. In J.A Russell & J.M. Femandez-Ools
(Eds ), The psvcholoov of facial expression. New York: Cambridge University
Press.
See, J.E., Warm, J.S., Demaber, W .N., & Howe, S. R. (1997). Vigilance
and signal detection theory: An empirical evaluation of five measures of
response bias. Human Factors. 39 14-29.
Shih, A R . & von Baeyer, C.L. (1994) Preschool children's sériation of
pain faces and happy faces in the affective facial scale. Psvcholooical Reports.
74. 659-665.
Shortt, J.W., Bush, L.K , McCabe, J.L.R. Gottman, J.M., & Katz, L.F.
(1994) Children's physiological responses while producing facial expressions of
emotions. Merrill-Palmer Quarterlv. 40. 40-59.
Shweder, R A. (1994). "You're not sick, you're just in love”: Emotion as
an interpretative system. In P. Ekman & R. J. Davidson, The nature of emotion:
Fundamental questions. New York, NY: Oxford University Press.
Singh, S.D., Ellis, C.R., Winton, A.S.W., Singh, N. N., Leung, J.P.
Oswald, D P. (1998). Recognition of facial expessions of emotion by children
with attention-deficit hyperactivity disorder. Behavior Modification. 2 2 .128-142.
Solomon, P.E. (1995). Enhancing sensitivitv to oain expressions.
Unpublished Doctoral Dissertation, University of Waterloo.
Stevens, B.J., Johnston, C O., & Horton, L. (1994). Factors that influence
the behavioral pain responses of premature infants. Pain. 59. 101-109.
Sullivan, L A , Kirkpatrick, S.W., & MacDonald. (1995). Interpretations of
facial expressions of emotion by sexually abused and non-abused girls. Journal
of Child Sexual Abuse. 4 45-61.

104

Tremblay, C , Kirouac. G , & Dore, F (1987). The récognition of adults’
and children’s facial expression of emotion. Journal of Psvcholoov. 121. 341350.
Tucker, J.S., & Riggio, R.E. (1988). The role of social skills in encoding
posed and spontaneous facial expressions. Journal of Nonvert)al Behavior. 12.
87-97.
Von Baeyer, C L , Johnson, M.E., & McMillan, J.J. (1984). Consequences
of nonverbal expression of pain; Patient distress and observer concern. Social
Science & Medicine. 19 1319-1324.
Walden. T.A., & Field, T.M. (1990). Preschool children’s social
competence and production and discrimination of affective expressions. British
Journal of Developmental Psvcholoov. 8. 65-76.
Walker-Andrews, A S & Lennon, E. (1991). Infants’ discrimination of
vocal expressions; Contributions of auditory and visual information. Infant
Behavior and Development. 14. 131 -142.
Wilkie, D.J., Holzemer, W.L., Tester, M.D., Ward, J A., Paul, S.M., &
Savedra, M. C. (1990). Measuring pain quality; Validity and reliability of
children’s and adolescents’ pain language. Pain. 4 1 .151-159.
Xeromeritou, A. (1992). The ability to encode facial and emotional
expressions by educable mentally retarded and nonretarded children. The
Journal of Psvcholoov. 126. 571-584.
Zahn-Waxler,C., Raddke-Yarrow, M., & King, R.A. (1979). Childrearing
and children’s prosocial initiations toward victims of distress. Child Development.
50. 319-330.
Zaidel, S.F. & Mehrabian, A. (1969). The ability to communicate and infer
positive and negative attitudes facially and vocally. Journal of Experimental
Research in Personalitv. 3. 233-241.
Zimbardo, P.O. & Leippe, M. (1991). The psvcholoov of attitude change
and social influence. New York, NY; McGraw-Hill.
Zuckerman, M., Lipets, M.S., Koivumka, J H , & Rosenthal, R. (1975).
Encoding and decoding nonverbal cues of emotion. Journal of Personalitv and
Social Psvcholoov. 32. 1068-1076.

105

Appendix A
Pain-related Facial Action Units

106

Appendix A
Facial Action Units

Action Unit

Descnotion

Muscular Basis

AU4

Brow Lowerer

Corrugator Procerus

AU6

Cheek Raiser

Orbicularis oculi,
outer portion

AU7

Lid Raiser

Orbicularis oculi,
inner portion

AU9

Nose Wnnkler

Levator labii superioris,
alaque nasi

AU10

Upper Lip Raiser

Levator labii superioris

AU43

Eyes Close

Orbicularis oculi

107

Appendix B
Questionnaires

108

Children’s Thoughts & Feelings Questionnaire (IRI)
The following statements are about your thoughts and feelings in a variety of
situations. For each item, indicate how well it describes you by choosing the
appropriate number on the scale. When you have decided on your answer,
circle the number that best describes you. READ EACH ITEM CAREFULLY
BEFORE RESPONDING. Answer as honestly and accurately as you can.
ANSWER SCALE
1
N otât
all like
me.

2
A little
bit like
me.

3
Kind of
like me.

4
A lot like
me.

5
Always
like me.

1. I often feel sorry for people who don't have thethings I have.

1

2

3

4 5

2.

1

2

3

4 5

3. I have a hard time understanding why other people
do the things they do.

1

2

3

4 5

4. Sometimes I don't feel very sorry for other people when
they are having problems.

1

2

3

4 5

5. I sometimes feel like I don't know what to do when
someone gets real upset.

1

2

3

4 5

When I see someone being picked on, I feel kind of sorry for them. 1 2

3

4 5

6.

I get scared when I see an accident.

7. I sometimes try to understand my friends better by imagining
how they think about things.

1

2

3

4 5

8.

1

2

3

4 5

9. When I see someone who is very sad or upset,
I feel like I want to go somewhere else.

1

2

3

4 5

10. When I'm mad at someone, I don't waste much time trying to
understand how they feel.

1

2

3

4 5

11. I often feel sorry for other children who are sad or in trouble.

1 2

3

12. I try to understand how other kids feel before I decide what
to say to them.

1

3

When I see someone get hurt, I usually remaincalm.

2

4

109

5

4 5

Intmpersonal Reactivity index
The following statements inquire about your thoughts and feelings in a variety of
situations. For each item, indicate how well it describes you by choosing the
appropriate letter on the scale at the top of the page; A, B. C, D or E. When
you have decided on your answer, fill in the letter on the answer sheet next to
the item number. READ EACH ITEM CAREFULLY BEFORE RESPONDING.
Answer as honestly as you can. Thank you.
ANSWER SCALE:
A
DOES NOT
DESCRIBE
ME
VERY WELL
Answer

B
DESCRIBES
ME
VERY WELL
Item
1. I daydream and fantasize, with some regularity, about
things that might happen to me.
2. I often have tender, concerned feelings for people less
fortunate than me.
3. I sometimes find it difficult to see things from the "other
guy’s” point of view.
4. Sometimes I don't feel very sorry for other people when
they are having problems.
5. I really get involved with the feelings of the characters in
a novel.
6. In an emergency situations, I feel apprehensive and ill-atease
7. I am usually objective when I watch a movie or play, and
don’t often get completely caught up in it.
8. I try to look at everybody’s side of a disagreement before
I make a decision.
9. When I see someone being taken advantage of, I feel
kind of protective towards them.

110

10. I sometimes feel helpless when I am in the middle of a
very emotional situation.
11.

I sometimes try to understand my friends better by
imagining how things look from their perspective.

12. Becoming extremely involved in a good t>ook or movie is
somewhat rare for me.
13. When I see someone get hurt I tend to remain calm.
14. Other people's misfortunes do not usually disturb me a
great deal.
15. If I'm sure I'm right about something, I don't waste much
time listening to other people's arguments.
16. After seeing a play or movie, I have felt as though I were
one of the characters.
17.

Being in a tense emotional situation scares me.

18. When I see someone being treated unfairly, I sometimes
don’t feel very much pity for them.
19. I am usually pretty effective in dealing with emergencies.
20. I am often quite touched by things that I see happen.
21. I believe that there are two sides to every question and
try to look at them both.
22. I would describe myself as a pretty soft-hearted person.
23. When I watch a good movie, I can very easily put myself
in the place of a leading character.
24. I tend to lose control during emergencies.
25. When I'm upset at someone, I usually try to “put myself
in his shoes” for a while.
26.

When I am reading an interesting story or novel, I
imagine how I would feel if the events in the story were
happening to me.

Ill

27.

When I see someone who badly needs help in an
emergency. I go to pieces.

28. Before criticizing somebody, I try to imagine how I would
feel if I were in their place.

112

What I am Like
(Self-Perception Profile for Children)
Name;

_________________

Age:________

Boy or Girl

Sample Sentence

Really
True
forma

Sort of
True
forma

□

Some kids would
rather play outdoors
in their spare time

□

Some kids feel that
they are very good
at their school work

□

Some kids find it
hard io make
friends

□

Some kids do very
we// at all kinds of
sports

□

Some kids are
happy with the v#y
they look

□

Some kids often do
not like the way
they behave

□

Some kids often get
mad at themselves

a.

□

1-

2.

3.

4.

□

□

□

□

□

□
7.

□

□

□

□

Some kids feel like
they are yusf as
smarf as other kids
their age
Some kids have
a for of friends

Sort of
True

Really
True

forma

forma

□

□

□

□

□

□

But

Others dbnY feel
that they are very
good when it
comes to sports.

□

□

But

Other kids are not
happy with the way
they look.

□

□

But

Other kids usually
like the way they
behave.

□

□

But

Other kids are
pretty pfoased with
themselves.

□

□

But

Other kids arent
so sure and
worNforifthey are
as smart.

0

□

But

Other kids don’t
have very many
friends.

□

□

But

But

But

Other kids would
rather watch T V.

Other kids worry
about whether they
can do the school
work assigned to
them
For other kids it's
pretty easy.

113

9.

10.

11.

12.

13.

14.
15.

16.

17.

18.

19.

Really
True

Sort of
True

forme

forme

□

□

□

□

□

□
□

□

□

□
□

□

Some kids vMSh
they could be a lot
better at sports

□

Some kids are
happy with their
height and weight

□

Some kids usually
do the righf thing

□

Some kids don’t
like the way they
are leading their life

□

Some kids are
pretty slow in
finishing their
school work

□
□

Some kids are kind
of hard to like
Some kids think
they could do weff
at just about any
new outdoor activity
they haven't tried
before

□

Some kids wish
their body was
ditferent

□

Some kids usually
acf the way they
know they are
supposed to

□

Some kids are
happy with
themselves most of
the time.

□

Some kids often
/brgef what they
leam

Sort of
True

Really
True

forme

forme

But

Other kids feel
they are good
enough at sports.

□

□

But

Other kids wish
their height or
weight were
différent.

□

□

But

Other kids often
donYdothe right
thing.

□

□

But

Other kids do like
the way they are
leading their life.

□

□

But

Other kids can do
their school work
quickly.

□

□

□

□

□

□

But

Other kids like
their body the way
it is.

□

□

But

Other kids often
don't act the way
they are supposed
to .

□

□

But

Other kids are
often not happy
with themselves.

□

□

But

Other kids can
remember things
easKy.

□

□

But
But

Other kids are
really easy to like.
Other kids are
afraid they might
nof do well at
outdoor things they
haven't ever tried.

114

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

Really
True

Sort of
True

for m*

(b rm

□

□

□

□

□

□

□

□

□

□

□

□

Some kids are
always doing things
with a k}f of kids

□

Some kids feel that
they are better than
others their age at
sports

□

Some kids wish
their physical
appearance was
(tUferent

□

Some kids usually
getinürouMe
Iwcause of things
they do

□

Some kids # e the
kind of person they
are

□

Some kids do very
we# at their
classwork

□

Some kids wish that
more kids liked
them

□

In games and
sports some kids
usually watch
instead of play

□

Some kids wish
something about
their face or hair
looked different

□

Some kids do
things they know
they shouldn’t do

□

Some kids are very
happy being the

way they are

Sort of
True

Really
True

forme

forme

But

Other kids usually
do things by
themselves.

□

□

But

Other kids don’t
feel they can play
as well.

□

□

But

Other kids #ke
their physical
appearance the
way it is.

□

□

But

Other kids usually
dbnY do things that
get them in trouble

□

□

But

Other kids often
wish they were
someone else.

□

□

But

Other kids dbnY do
very well at their
classwork.

□

□

But

Other kids feel that
most kids do like
them.

□

□

But

Other kids play
rather than just
watch.

□

□

But

Other kids like
their face and hair
the way the are.

□

□

But

Other kids hartUy
ever do things they
know they
shouldn't do

□

□

□

□

But

Other kids wish
they were different.

115

Really
True
for me
31.

32-

33.

34.

35.

36.

□

□

□

□

□

□

Sort of
True

Sort Of
True

Really
True

form#

form#

□

□

□

□

But

Other kids are
good at new
games right away.

□

□

But

Other kids think
that they are not
very attractive or
good looking.

□

□

But

Other kids wisti
they would be
Underto others.

□

□

But

Other kids think
the way they do
things is fine.

□

□

(brnrw

□

Some kids have
trouble figuring out
the answers in
school

□

Some kids are
popular with others
their age

□

Some kids dbnY do
well at new outdoor
games

□

Some kids think
that they are
attractive or good
looking

□

Some kids are
usually very Mridto
others

□

Some kids aren’t
very happy with the
way they do a lot of
things

But

But

Other kids almost
always can figure
out the answers.
Other kids are not
very popular.

116

What I am Like
(Self-Perception Profile for College Students)
_____________________

Name:

Age:________

Male or Female

The following are statements which allow students to describe themselves. There are no right or wrong answai
since students differ markedly. Please read the entire sentence across. Frst decide which one of the two parts <
each statement best describes you; then go to that side of the statement and check whether that is just sort oftni
for you or reefy true for you. You will just check ONE of the four twxes for each statement Think about what yo
are like in the university environment as you read and answer each one.
Really
True
for me

Sort of
True
forme

Sort of
True
for me
Some students like the
kind of person they are
Some students are nof
proud of the work they do
on their job

□

=

«

:

□

□

□

□

□
□
□
□

» □ □
3

Other students are very
proud of the work they do on
their job.
Other students do not feel so
confident

Some students are not
satisfied with their social
skills

BUT

Other students think their
social skills are just fine.

Some students are not
happy with the way they
look

BUT

Other students are happy
with the way they look.

Some students like the
way they act when they
are around their parents

BUT

Some students get kind of
lonely because they dont
really have a close friend
to share things with
Some students feel like
they are just as smart or
smarter than other
students

□

Some students often
question the morality of
their behaviour

□ □

Some students feel that
people they like
romantically will be
attracted to them

- □

BUT

Other students wish that they
were different

BUT

Some students feel
confident that they are
mastering their
coursework

4.

BUT

BUT

Other students wish they
acted differently around thair
parents.
Other students dont usually
gat too lonely because they
do have a close friend to
share things with.

Really
True
forme

□ □
□ □
□ □
□
□
□
□

□
□
□
□

BUT

Other students wonder if they
are as smart

□ □

BUT

Other students feel their
behaviour is usually moral.

□ □
□ □

BUT

Other students worry about
whether people they like
romantically will be attracted
to them.

117

Really
True
for me

Sort of
True
forme

Sort of
True
for me

□ □

Wfien some students do
something sort of stupid
that later eppaars very
funny, they find it hard to
laugh at themselves.
Some students feel they
are just as creative or even
more so than other
students
Some students feel they
could do well at just about
any new athletic activity
they haven't tried before
Some students are often
disappointed with
themselves

15.
16.
17.

18.

19.

20 .

21 .

22 .

23 .

24 .

□
□
□
□
□
□
□
□
□
□

□
□
□
□

BUT

BUT

BUT

BUT

Really
True
forme

When other students do
something sort of stupid that
later appears very funny, they
can easily laugh at
themselves.

□ □

Other students wonder if they
are as creative.

□ □

Other students are afraid
they migfit not do well at
athletic activities they haven't
ever tried.

□ □

Other students are usually
quite pleased with
themselves.

□
□
□
□
□
□
□
□
□
□
□

Some students feel they
are very good at their job

BUT

Other students worry about
whether they can do their job.

Some students do very
well at their studies

BUT

Other students don't do very
well at their studies.

Some students find it hard
to make new friends

BUT

Other students are able to
make new friends easily.

Some students are happy
with their height and
weight

BUT

Other students wish their
height or weight was
diffierent

□

Some students find it hard
to act naturally when they
are around their parents

BUT

Other students find it easy to
act naturally around their
parents.

□

Some students are able to
make close friends they
can really trust.

BUT

Other students find it hard to
make dose friends they can
really trust.

□
□

Some students do not feel
they are very mentally able

BUT

Some students usually do
what is morally right

BUT

Other students sometimes
don't do what they know is
morally right.

□

Some students find it hard
to establish romantic
relationships

BUT

Other students don’t have
difficulty establishing
romantic relationships.

□

Some students don't mind
being kidded by their
friends

BUT

Other students feel that they
are very mentally able.

Other students are bothered
when friends kid them

118

□
□
□
□
□

□
□
□
□
□
□

Really
True
for me

26
27

Sort of
True
for me

BUT

Other students feel they are
very creative and inventive.

Some students dont feel
they are very athletic

BUT

Other students do feel they
are athletic.

Some students usually like
themselves as a person

BUT

Other students often don't
like themselves as a person.

□ □
.
□
□
. □
□

Some students worry that
they are not as creative or
inventive as other people

□

Some students feel
confident about their ability
to do a new job

BUT

Other students worry about
whether they can do a new
job they haven't tried before.

□

Some students have
trouble figuring out
homework assignments

BUT

Other students rarely have
trouble with their homework
assignments.

□
□
□

Some students like the
way they interact with
other people

BUT

Other students wish their
interactions with other people
were different

23 .

□

29 .

□

30 .

□

31 .

□

32 .

□

33 .

□
.
□
.
□
.
□

34 .

35

36

37

Some students feel
comfortable being
themselves around their
parents
Some students don't have
a close friend they can
share their personal
thoughts and feelings with

BUT

Other students like their body
the way it is.

□
□
□
□
□
□
□
□

BUT

Other students do have a
friend who is close enough
for them to share thoughts
that are really personal

□ □
□

□

Other students wonder if they
are as bright

□

Some students would like
to be a iMtter person
morally

BUT

Other students think they are
quite moral.

□

Some students have the
ability to develop romantic
relationships

BUT

□

Some students have a
hard time laughing at the
ridiculous or silly things
they do

Some students feel they
are better than others at
sports

□
□
□
□
□
□
□
□

Other students have difficulty
tieing themselves around
their parents

BUT

Some students do not feel
that they are very inventive

Really
True
forme

BUT

Some students feel they
are just as bright or
brighter than most people

□ □
.
□ □

38 .
39

□ □

Some students wish their
body was different

Sort of
True
for me

Other students do not find it
easy to develop romantic
relationships.

BUT

Other students find it easy to
laugh at themselves.

BUT

Other students feel that they
are very inventive.

BUT

Other students dont feel they
can play as well.

□
□ □
□ □
□ □
□ □
□ □
119

Really
True
for me

Sort of
True
forme

40

□ □
.
□ □
. □
□
. □
□
. □
□
. □
□
. □
□
□ □
. □
□
. □
□
50.
□
□

Some students really like
the way they are leading
their lives

BUT

Other students often don't
like they like the way they are
leading their lives.

41

Some students are not
satisfied with the way they
do their job

BUT

Other students are quite
satisfied with the way they do
their job.

42

Some students sometimes
do not feel intellectually
competent at their studies

BUT

Other students usually do
feel intellectually competent
at their studies.

43

Some students feel that
they are socially accepted
by other people

BUT

Other students wish more
people accepted them.

44

Some students like their
physical appearance the
way it is

BUT

Other students do not like
their physical appearance.

45

Some students find that
they are unable to get
along with their parents

BUT

Other students get along with
their parents quite well.
Other students find it hard to
make really close friends.

□
□
□

Some students can really
laugh at certain things
they do

46

Some students are able to
make really dose friends

BUT

47

Some students would
really rather t)e different

BUT

48

49

51 .

□

52.

□

53 .

□

Some students question
whether they are very
intelligent
Some students live up to
their own moral standards
Some students worry that
when they like someone
romantically, that person
won't like them back

Some students feel they
have a lot of original ideas
Some students don't do
well at activities requiring
physical skill

Other students are very
happy being the way they
are.

Sort of
True
for me

Really
True
for me

□
□
□
□
□
□
□
□

□
□
□
□
□
□
□
□

Other students feel they are
intelligent

□

BUT

Other students have trouble
living up to their moral
standards.

□ □

BUT

Other students feel that when
they are romantically
interested in someone, that
person will like them back.

□ □

Other students have a hard
time laughing at themselves.

□ □
□ □
□ □

BUT

BUT

BUT

Other students question
whether their ideas are very
original.

BUT

Other students are good at
activities requiring physical
skill.

120

□

Some students are often
dissatisfied witfi
'

BUT

Otfier students are usually
satisfied witft ttiemseives.

ttiemseives

'

121

Appendix C
Pain Descriptors

122

Pain Descriptors
A - Undetectable
B - Detectable, not unpleasant
C - Slightly Unpleasant
D - Slightly Annoying
E - Annoying
F - Unpleasant
G - Slightly Distressing
H - Slightly Miserable
I - Very Annoying
J - Distressing
K - Very Unpleasant
L - Miserable
M - Very Distressing
N - Slightly Intolerable
O - Very Miserable
P - Intolerable
Q - Very Intolerable

123

Appendix D
Supplementary Tables

124

Table D1
Correlations among coHeoe students* setf-perception subscales and sensitivitv.
bias and parüdpants* correlations wHh patients' self-repoft and facial actions
with level of significance (o) in parentheses
Correlations with
oartidoant ratinos

Bias

Sensitivitv

-.11
(.53)

oatient
self-reoort
-.50
(.80)

oatient facial
actions
.01
(.97)

-.34
(.05)

-.35
(.05)

.06
(.75)

-.10
(.58)

-.09
(.62)

-.11
(.53)

-.09
(.60)

.05
(.76)

.01
(.97)

-.16
(.38)

-.08
(6 7)

-.12
(.52)

-.24
(.18)

-.10
(.58)

-.19)
(.29)

Humor

-.02
(-91)

.04
(.84)

-.06
(.76)

-.09
(.64)

-.26
(.15)

-.08
(6 5 )

Intellectual
Competence

-.14
(.44)

-.15
(.39)

-.20
(.26)

-.25
(.16)

.07
(.72)

-.16
(.38)

Job
Competence

.06
(.75)

.05
(.77)

-.42
(.02)

-.45
(.01)

.19
(.29)

.08
(.67)

Morality

-.09
(.63)

-.06
(.76)

-.22
(.21)

-.34
(.05)

.14
(.42)

-.07
(.68)

Behaviour with
Parents

-.01
(.97)

-.09
(6 1)

-.25
(.16)

-.28
(.11)

.01
(.77)

-.18
(.33)

Romantic
Relationships

.18
(.32)

.13
(4 7 )

-.42
(.01)

-3 8
(.03)

.11
(.54)

.02
(.91)

Scholastic
Competence

-.01
(.97)

-.03
(8 7 )

-.27
(1 3)

-.30
(.09)

.07
(.71)

.01
(.94)

Social
Acceptance

.08
(.67)

.05
(.77)

-.24
(.18)

-.29
(10)

-.02
(.92)

-.10
(.58)

Global
Self-Worth

-.08
(.66)

-.07
(6 9)

-.24
(.17)

-.32
(.07)

.06
(.76)

-.08
(.67)

mild

strono

mild

strona

Physical
Appearance

.001
(.999)

-.026
(.89)

-.13
(.46)

Athletic
Competence

.14
(.45)

.11
(.55)

Creativity

-.09
(.60)

Relationships
with Friends

125

Table D2
Correlations among adults' IRI subscales and sensitivitv. bias and oartidDants*
correlations with patients* self-reoort and facial actions
Sensitivitv

Bias

mild

strono

mild

strona

Empathie Concern

-.24
(.18)

-.06
(.74)

-.03
(.86)

Perspective Taking

-.32
(.07)

-.22
(.22)

Personal Distress

-.17
(.35)

Fantasy

-.03
(.87)

Correlations with
oartidoant retinas

-.09
(.61)

oatient
self-reoort
-.10
(.57)

oatient facial
actions
-.07
(.72)

-.13
(48)

-.12
(.53)

-.10
(.58)

-.17
(.34)

-.03
(8 5)

.38
(.03)

34
(.06)

-.001
(.99)

.04
(.82)

.001
(.999)

-.03
(.89)

.06
(.76)

-.12
(.51)

.05
(.80)

Table D3
Correlations among vouna children's self-perceotion subscales and sensitivity.
bias and participants' correlations with patients' self-reoort and facial actions foi
Sensitivitv

Bias

Correlations with
oartidDant ratings

.42
(01)

oatient
self-reoort
-.10
(.58)

oatient fadal '
actions
.001
(.997)

-.19
(.30)

-.22
(.23)

.26
(.14)

.31
(.08)

.18
(32)

.36
(.04)

.30
(.09)

.33
(.06)

.31
(.08)

.09
(.62)

.05
(.77)

.09
(.62)

.09
(.61)

.15
(41)

.11
(.55)

Sodal
Acceptance

-.10
(.57)

-.17
(34)

-.10
(.58)

-.08
(.66)

-.12
(.51)

-.11
(.53)

Global
Self-Worth

-.13
(.48)

-.19
(.30)

-.01
(.94)

.02
19.11

-.07
(.69)

-.20
(.28)

mild

Strong

mild

strong

Physical
Appearance

.07
(.72)

-.03
(.88)

.38
(.03)

Athletic
Competence

.29
(.10)

.37
(.04)

Behavioural
Conduct

-.04
(.85)

Scholastic
Competence

126

Table D4
Correlations among vouna children’s IRI subscales and sensitivitv. bias and
participants' correlations with patients* self-reoort and fadal actions fot

Sensitivitv

Bias

Correlations with
participant ratinos

mild

strona

mild

strona

patient
patient facial
self-reoort
actions
.03
.01
(.86)
(96)

Empathie Concern
EC

.04
(.81)

.09
(.63)

.01
(.96)

-.01
(.98)

Perspective Taking
PT

.12
(.51)

.07
(.69)

-.14
(.43)

-.16
(.38)

-.03
(.86)

.04
(.83)

Personal Distress
PD

.08
(.65)

.25

.30
(.09)

.22
(.23)

.34
(.05)

.35
(.05)

Table D5
Correlations among middle children's self-oerceotion subscales and sensitivitv.
bias and participants' correlations with patients' self-reoort and facial actions (o)
Sensitivitv

Bias

Correlations with
participant ratinos

.04
(82)

patient
self-report
-.30
(.08)

patient facial
actions
-.22
(.20)

.06
(.72)

.03
(.87)

-.27
(.12)

-.23
(.18)

.16
(.37)

.23
(.18)

.21
(.22)

-.09
(.61)

.07
(.68)

.18
(.31)

.05
(.79)

.001
(.998)

.03
(.86)

-.04
(.81)

-.01
(.98)

Social
Acceptance

.04
(.83)

-.07
(.69)

-.13
(45)

-.10
(.56)

-.06
(.98)

-.11
(.51)

Global
Self-Worth

.01
(.94)

-.03
(.85)

.24
(-17)

.22
(.21)

-.32
(.06)

-.158
(.366)

mild

strona

mild

strona

Physical
Appearance

-.05
(.79)

-.11
(.54)

.03
(.86)

Athletic
Competence

-.24
(.17)

-.21
(.23)

Behavioural
Conduct

.18
(.32)

Scholastic
Competence

127

Table 06
Correlations among middle children’s IRI subscales and sensitivitv. bias and
participants' correlations with patients' self-reoort and facial actions to)

Sensitivitv

Bias

mild

strona

mild

strong

Empathie Concern
EC

.06
(-75)

-.02
(.90)

-.25
(.14)

Perspective Taking
PT

.07
(.68)

.23
(.19)

Personal Oistress
PO

-.02
(.93)

-.02
(.92)

Correlations with
oartidoant ratinos

-.35
(.04)

oatient
self-reoort
-.05
(.79)

oatient fadal
actions
-.15
(.39)

.06
(.74)

.01
(.98)

.15
(38)

.33
(.05)

-.07
(.70)

-.06

.15
(.39)

-.09
(.60)

Table 07
Correlations among old children's self-oerceotion subscales and sensitivitv. bias
and participants' correlations with patients' self-reoort and facial actions
( o )

Sensitivitv

Bias

mild

strono

mild

strona

Physical
Appearance

-.22
(.22)

-.11
(.53)

-.04
(.81)

Athletic
Competence

-.22
(.23)

-.22
(.21)

Behavioural
Conduct

.23
(.20)

Scholastic
Competence

Correlations with
oartidoant ratinos

.11
(54)

oatient
self-reoort
.07
(.68)

oatient fadal
actions
-.04
(.83)

-.32
(.07)

-.38
(.03)

.01
(.98)

-.15
(.41)

-.10
(.58)

-.31
(.08)

-.21
(.25)

-.03
(.89)

-.13
(47)

.24
(.18)

-.001
(.996)

-.25
(.16)

-.17
(.33)

-10
(.59)

r.09
(.64)

Sodal
Acceptance

.16
(.38)

.09
(.64)

-.22
(.23)

-.17
(.35)

.07
(.71)

.11
(.55)

Global
Self-Worth

.11
(.54)

-.04
(.82)

-.17
(.35)

-.17
(.35)

.12
(.52)

-.05
(.79)

128

Table D8
Correlations among old children’s IRI subscales and sensitivitv. bias and
participants' correlations with patients' self-reoort and facial actions

Sensitivitv

Bias

Correlations with
oartidoant ratinos
oatient
patient fadal
self-reoort
actions
-.30
-.08
(.09)
(.68)

mild

strong

mild

strona

Empathie Concern
EC

.08
(.65)

-.11
(.55)

-.32
(.07)

-2 5
(.17)

Perspective Taking
PT

.13
(.49)

-.12
(.52)

-.41
(02)

-.43
(.01)

-11
(.53)

-.29
(.11)

Personal Distress
PD

.08
(.65)

.08
(.65)

-.48
(.01)

-.54
JLQ01)

.13
( 46)

.002
(-99)

Table D9
Correlations among the difference between
participants' and patients' ratings and
self-perception scales fp)
oartidoant - oatient
difference
Physical
Appearance
Athletic
Competence
Creativity
Friendship
Humor
Intellectual Ability
Job Competence
Morality
Behaviour with
Parents
Romantic
Relationships
Scholastic
Competence
Sodal Acceptance
Global Self-Worth

-.16 (.38)
-.25 (.17)
-.11 (.54)
-.34 (.05)
.05 (.78)
-.16 (.38)
-.47 (.01)
-.24 (18)
-.32 (.07)
-.37 (.03)
-.30 (.10)
-.24 (18)
-.31 (.08)

129