You are on page 1of 13

Responses to the Negative Emotions of

Others by Autistic, Mentally Retarded,


and Normal Children

Marian D. Sigman, Connie Kasari, Jung-Hye Kwon,


and Nurit Yirmiya
University of California, Los Angeles

SIGMAN, MARIAN D.; KASARI, CONNIE; KWON, JUNG-HYE; and YIRMIYA, NURIT. Responses to the
Negative Emotions of Others by Autistic, Mentally Retarded, and Normal Children. CHILD
DEVELOPMENT, 1992, 63, 796-807. Attention, facial aFfect, and behavioral responses to adults
showing distress. Fear, and discomfort were compared for autistic, mentally retarded, and normal
children. The normal and mentally retarded children were very attentive to adults in all 3
situations. In contrast, many of the autistic children appeared to ignore or not notice the adults
showing these negative affects. As a group, the autistic children looked at the adults less and
were much more engaged in toy play than the other children during periods when an adult
pretended to be hurt. The autistic children were also less attentive to adults showing fear,
although their behavior was not difFerent From the normal children. Few oF the children in any
group showed much Facial aFFect in response to these situations. The results are discussed in
terms oF the importance of afFect in the social learning experiences of the young child.

The social relationships of children de- depictions (Braverman, Fein, Lucci, & Wa-
pend critically on attentional and affective terhouse, 1989; Hobson, 1986; Ozonoff, Pen-
interactions. Moreover, children appear to nington, & Rogers, 1989; Prior, Dahlstrom,
learn about the inanimate and animate envi- & Squires, 1990). Autistic subjects have
ronment by observing the emotional reac- been found to be less accurate in identifying
tions oF others (Gampos & Stenberg, 1981; Facial aFFect than control subjects, although
Feinman, 1982; Gunnar & Stone, 1984; Klin- group diFFerences are smaller and often not
nert, Gampos, Sorce, Emde, & Svedja, 1983; significant when groups are matched on ver-
Walden & Ogan, 1988). The importance oF bal abilities. However, even autistic individ-
understanding and communicating aFfect in uals oF normal intelligence have more diFfi-
early social interactions is highlighted when culty than normal controls in identiFying
one considers what happens to interactions afFects accurately, particularly in depictions
in which affect and attention are not inte- oF social situations (MacDonald et al., 1989;
grated, such as appears to occur with autistic OzonofF, Fennington, & Rogers, in press;
children. Clinical observations and very lim- Yirmiya, Sigman, Kasari, & Mundy, in
ited empirical data suggest that autistic chil- press).
dren do not respond to other people's afFect
in the same way as normal and mentally re- AFFective responses oF young autistic
tarded children. children in social interactions have been ob-
served. Gontrary to clinical descriptions,
InFormation about the responses oF au- empirical studies have Found Fewer overall
tistic children to the afFects oF others is lack- diFferences in facial expressions of positive
ing, despite the recognition that autistic in- aFFect between autistic and control children
dividuals universally show deviations in than anticipated (Dawson, Hill, Spencer,
their social relationships. A number of stud- Galpert, & Watson, 1990; Kasari, Sigman,
ies have investigated the capacity of older Mundy, & Yirmiya, 1990; Yirmiya, Kasari,
autistic children to identify aFfects correctly Sigman, & Mundy, 1989). On the other hand,
from photographs, drawings, and videotaped autistic children do show less positive afFect

This research was supported by grants NS25243fromNINDS and HD17662fromNICHD.


Jung-Hye Kwon is currently at Seoul University and Nurit Yirmiya is currently at the University
of Jerusalem. We are grateful to Alison Anson, Michael Espinosa, Stephanie Freeman, Margie
Greenwald, Alisa Hoffman, Nicholas Lofthouse, Alma Lopez, and Susan Toth For their contribu-
tions to this research. Send reprint requests to: Marian Sigman, Department oF Psychiatry,
68-237B NPI, UGLA Medical School, Los Angeles, CA 90024.
[Child Development, 1992, 63, 796-807. © 1992 by the Society for Research in Child Development Inc
AU rights reserved. 0009-3920/92/6304-0010$01.00]
Sigman et al. 797
in combination with attentional behaviors, ing attention was diFficult. As mentioned
such as while looking at their mothers and above, previous research has already shown
teachers, than mentally retarded and normal that autistic children inFrequently use proto-
controls (Dawson et al., 1990; Snow, Hert- declarative gestures oF showing and pointing
zig, & Shapiro, 1987). Moreover, in contrast either with Familiar caregivers (Sigman,
to normal children who combine positive af- Mundy, Sherman, & Ungerer, 1986) or with
Fect with attention to adults when engaged experimenters' (Baron-Gohen, 1989; Love-
in joint attention or protodeclarative ges- land & Landry, 1986; Mundy, Sigman, Un-
tures (Bruner & Sherwood, 1983; Jones, Gol- gerer, & Sherman, 1986). While they look as
lins, & Hong, 1991; Mundy, Kasari, & Sig- much as controls during dyadic interactions
man, in press), autistic children not only with caregivers or experimenters and during
look less at adults in response to toys or ex- protoimperative gestures, they look less at
periences that appear interesting to them the Faces oF adults showing neutral aFFect in
but, when they do, they show less positive triadic interaction situations when an inter-
aFFect than normal or mentally retarded chil- esting toy is involved. However, negative af-
dren (Kasari et al, 1990). fect may be so salient for the young child
that these group difFerences in attention may
The responses oF young autistic chil- be diminished or eliminated. One oFthe pur-
dren to the negative emotions oF others have poses oFthis study, then, was to determine
not been investigated. Young normal chil- iF autistic children would show the same
dren appear very interested in the distress amount oF attention to the negative affect of
oF others, and some children show aFfective others as normal and mentally retarded
and prosocial responses as early as the sec- children.
ond year oF life (Rheingold, Hay, & West,
1976; Zahn-Waxler & Radke-Yarrow, 1990; Three diFferent aFFects and situations
Zahn-Waxler, Radke-Yarrow, Wagner, & were used with the same sample. To sim-
Ghapman, in press). Moreover, normal chil- plify the presentation, the methods and re-
dren will look to the Face oFan adult simulat- sults are given separately For each situation.
ing Fear or disgust in response to a novel
person or toy by the end oF the first year oF Responses to the Distress of Others
liFe (Gamras & Sachs, 1991; Hornik, Ris-
enhoover, & Gunnar, 1987; Klinnert, 1984; METHOD
Klinnert, Emde, Rutterfield, & Gampos, Subjects
1986). Thus, the negative emotions oFothers Three groups oF children participated in
are oF great interest to young normal chil- all phases oFthis study. The first group con-
dren and may infiuence their emotional re- sisted oF 30 children diagnosed with autism
actions and behaviors. who were reFerred by clinicians associated
with the UGLA School oF Medicine or pri-
The aim of the present study was to in- vate practitioners. Diagnoses oFautism were
vestigate the responses oF young autistic made iF the child met two oF three difFerent
children to negative emotion shown by oth- criteria: (1) a reFerring clinician's diagnosis
ers and to contrast these reactions with those made according to DSM III-R guidelines
shown by normal children oF the same de- (DSM III-R; APA, 1987); (2) a score oF 32
velopmental level and with those shown by or above on an abbreviated version oF the
mentally retarded children oF the same de- Ghildhood Autism Rating Scale (GARS;
velopmental and chronological age. Because Schopler, Reichler, & Renner, 1986), which
most autistic children are also mentally re- was filled out From videotaped observations
tarded, their responses need to be compared oFthe child by an experimenter who was un-
with those oF nonautistic, mentally retarded aware oF the clinician's diagnosis oF the
children. One hypothesis was that the autis- child; and (3) a score oF 70 or above on the
tic children would show less aFFect and be- Autism Behavior Ghecklist (ABG; Krug, Ar-
havioral or empathetic response to the nega- ick, & Almond, 1978), which was filled out
tive affect of the adults. Given that autistic during an interview with the parent by an-
children are reported to return their moth- other experimenter.
ers' smiles less than normal controls (Daw-
son et al., 1990), it seemed likely that they The second group consisted oF30 devel-
would also show less negative afFect and be- opmentally delayed children who did not
havioral response to the negative emotions meet any oFthe criteria listed above and who
oF others. were matched with the autistic children on
chronological and mental age. Thirteen chil-
The Formulation oF a hypothesis regard- dren in this group were diagnosed with
798 Child Development
TABLE 1
SAMPLE CHARAGTEKISTICS; AUTISTIG, MENTALLY RETARDED, AND NORMALLY
DEVELOPING CHILDREN

Autistic MR Normal
Characteristics (n = 30) (n = 30) (n = 30)
Chronological age (months):
M 42.40 41.67 19.83
SD 11.22 10.71 8.21
Mental age" (months):
M 22.90 24.00 23.27
SD 10.39 9.39 10.68
Developmental quotient:
M 52.53 55.73 113.53
SD 16.14 13.83 8.73
Language age'' (months):
Expressive:
M 17.63 20.63 21.43
SD 6.68 9.58 12.07
Receptive:
M 17.37 19.90 22.50
SD 9.25 9.71 11.22
SES:=
M 3.48 3.73 4.21
SD 1.02 .87 .74
Sex: male/female 27/3 18/12 27/3
Language: English/other 29/1 26/4 30/0
Race: white/other 24/6 24/6 28/2

' As determined by a Cattell or Stanford-Binet score.


^ As determined by Reynell Language Scales scores.
" As determined by a Hollingshead score (1 = lowest to 5 = highest SES).

Down syndrome, one child had Prader- how to use a wooden pounding toy and ham-
Willie syndrome, and 16 were mentally re- mer. During this demonstration, the individ-
tarded with an unknown etiology. The nor- ual pretended to hurt herself by hitting her
mal group of 30 children was selected to finger with the hammer. For 30 sec, the indi-
match the autistic children on mental age vidual displayed Facial and vocal expres-
and gender; there were 27 males and three sions oF distress without using words. After
Females in botb tbe autistic and normal a 10-sec period of neutral aFfect, the individ-
groups. There were no significant diFfer- ual showed the child that her finger did not
ences in mean receptive or expressive lan- hurt anymore. Parents were trained by the
guage age among the groups. Other charac- experimenter to display a peak emotional ex-
teristics of the sample are shown in Table 1. pression of distress. The episodes were vid-
eotaped with a split-screen video recorder
Each child was seen in two sessions so that the upper front half oF the child was
spaced over a 1—2-week interval. Besides in view on one screen and the whole scene
the procedures described, children were including the adult was in view on the other
also evaluated with either the Gattell Devel- screen. A date-time generator displayed the
opmental Scales of the StanFord-Binet Intel- time to the tenth oF a second in the lower
ligence Test as well as the Reynell Lan- corner oF the screen. The responses of one
guage Scales. autistic and one normal child to parental dis-
tress were not recorded and therefore not
included in analyses.
Procedure
The parent and experimenter distress Three categories of behavior were
situations were administered in difFerent coded From the videotaped observations of
testing sessions, and the order was counter- each session: attention, behaviors directed
balanced across children. The individual to the toy and adult, and facial affect. Goding
was seated at 90 degrees to the leFt of the of each category was done independently by
child at a small table and showed the child two raters.
Sigman et al. 799
Attention.—The child's attention was facial aFFect independently From vocaliza-
coded every half second and categorized tion. Percent agreement calculated on a sub-
into Four mutually exclusive categories. The set (n = 10) oF the subjects was 77%, kappa
categories consisted oF: (1) looks at the adult, = .63. Total durations oF positive, neutral,
(2) looks at adult's hand, (3) looks at the toy, and negative afFects were recorded.
and (4) looks away or unFocused (e.g., scans
the room or stares into space). Total duration RESULTS
oF looks in each oF these categories was re-
corded. In addition, the latency to first look The data were analyzed with three-
at the experimenter or parent was recorded. group (autistic, mentally retarded, and nor-
Reliability oF attention calculated by count- mal) MANOVAs and Follow-up ANOVAs.
ing agreements in eacb halF second unit For Post hoc tests were computed using BonFer-
10 subjects was good, kappa = .86. roni (Dunn) tests. Preliminary analyses led
to several decisions about treatment oF the
Play behaviors.—The child's play be- data. First, eFfects related to gender were not
haviors with the toy were coded second by noteworthy, so the data For males and Fe-
second and categorized into three mutually males were combined. Second, For both the
exclusive categories: no play, passive play distress and Fear situations, the intensity oF
(holding, brieF touching), and active play. emotion showed by the adult posing the
Agreement between two raters For each play emotion was rated as low, mild, or high.
category in each period calculated on 10 Both parent and experimenter behaviors
tapes was good, kappa = .95. Latency to first were rated For tbe distress situations (kappa
touching the toy and durations oF active play = .84 on 10 tapes) and For the Fear situation
behaviors were recorded. (kappa = .70 on 15 tapes). Since there were
no significant diFFerences in the intensity oF
Behaviors toward the distressed adult.— the experimenter's or parent's demonstrated
The child's behaviors toward the distressed aFFects across the three groups, intensity of
adult were coded in two ways. First, the demonstrated afFect was not included in the
child's degree oF concern was coded using a subsequent analyses as an independent vari-
six-point rating scale: (1) shows no interest, able. Lastly, because oF variability in the du-
(2) shows a hint oF interest, (3) shows some ration oF sessions, only the first 30 sec oF
apparent interest but no clear concern, (4) each session were included in analyses.
shows one clear sign oF concern, (5) shows
more than one clear sign oF concern, and (6) Data for four measures of attention (to
shows intense aFFective involvement and/or adult, to adult's hand, to toy, and away) and
comForting bebavior. Interrater reliability on two measures of behavior (concern rating
10 tapes was 87%, kappa = .76. Second, be- and time involved in active play) were com-
havioral responses to the adult's emotional pared across groups with a MANOVA for
display oF distress were recorded. These in- parent distress and a MANOVA For experi-
cluded physical withdrawal, comForting be- menter distress. The group eFFect was sig-
haviors, soliciting comFort From parent, imi- nificant in both MANOVAs, Wilks's lambda
tating, tbrowing the toy, giving the toy to the F(12,160) = 4.41, p < .0001 and F(12,164)
distressed person, and selF-stimulation. In- = 7.64, p < .0001.
terrater agreement on these behaviors on 10 Attention to Emotional Displays
tapes was good, kappa = .74. A significant main eFFect oF group was
Facial affect.—The child's Facial ex- obtained in the duration oF attention to the
pressions were coded second by second us- adult in both situations, F(2,85) = 11.78, p
ing the Following scale: (1) very negative, (2) < .0001 For the parent and F(2,87) = 25.04, p
negative, (3) neutral, (4) positive, and (5) < .0001 For the experimenter. Post hoc tests
very positive. To minimize subjective bias indicated that autistic children looked at the
io judging Facial aFFect, coders were first adult significantly less than did mentally re-
trained on the MAX coding system, a highly tarded and normal children (see Table 2).
reliable method developed by Izard and his All the children in the normal and mentally
colleagues to categorize expressions oF aFfect retarded group looked at each adult at least
based on observations of facial movement once. Six autistic children never looked at
pattems (Izard, 1979). After coders became their parents and eight autistic children
highly reliable on all categories of MAX fa- never looked at the experimenter during the
cial expression, they were then trained on demonstration oF distress.
the current rating system. All coding was In both situations, the autistic children
conducted without sound in order to code looked to the toy much more than the other
800 Child Development
TABLE 2
MEANS FOR ATTENTION VARIABLES IN THE DISTRESS CONDITIONS

Variables Autistic MR Normal

Parent distress: (N = 29) (N = 30) (N = 29)


Number who look 23 30 29
Duration of attention:
To parent:
M 5.07** 11.10 14.05
SD (7.13) (7.89) (6.44)
To toy:
M 17.72** 9.38 11.71
SD (10.32) (8.20) (7.27)
Experimenter distress: (N = 30) (N = 30) (N = 29)
Number who look 22 30 29
Duration of attention:
To experimenter:
M 6.63** 15.88 18.57
SD (7.32) (7.34) (5.78)
To toy:
M 17.60** 7.03 5.10
SD (9.26) (6.50) (5.02)

NOTE.—Means and standard deviations in the table are given in seconds.


** Means for the autistic group are significantly different from the means for the other two
groups, p < .05.

children. There was a significant group ef- nificant demonstrator effect, with all the
fect in both parent distress, F(2,85) = 7.21, children attending more when an unfamiliar
p < .001 and experimenter distress, F(2,87) individual demonstrated distress than when
= 26.60, p < .0001. As can be seen in Table the parent demonstrated distress, F(l,85) =
2, the autistic children looked at the toy far 17.43, p < .0001 and no interaction effect.
more than the other children, particularly
when the experimenter manifested distress. Behavioral Responses:
The mean fixation time that the toy was at- Active Play and Concern
tended to was also significantly greater for Just as the autistic children looked at the
the autistic children than the other children toy more than the other children, so they
in both parent distress situations, F(2,82) = also played with the toy more. There were
8.49, p < .0004, and experimenter distress, significant group effects for toy play in both
F(2,84) = 9.46, p < .0002. There were no distress situations, F(2,85) = 4.68, p < .01
group difFerences in the amount that the for parent distress and F(2,87) = 31.77, p <
children looked away. .0001 for experimenter distress. As can be
seen in Table 3, the autistic children played
Because oF the diFFerences in attention with the toy much more than the children
to adults, latency oF initial look was com- in the other groups, particularly when the
pared across groups and individuals For experimenter showed distress. The autistic
those children who looked to the adult at children were also rated as less concerned
least once. Results indicated that tbere was in both situations, F(2,85) = 18.61, p < .0001
a significant group effect in botb conditions, for parent distress and F(2,87) = 25.01, p <
F(2,83) = 11.10, p < .0001 For parent condi- .0001 For experimenter distress.
tion, and F(2,81) = 4.73, p < .05 For experi-
menter condition. Post hoc tests revealed While the normal and mentally retarded
that autistic children looked to the adults children appeared more concerned than the
significantly later than did normal and men- autistic children, not many children in any
tally retarded children. group engaged in much comForting behav-
ior. Only three normal, seven mentally re-
In order to determine whether the chil- tarded, and six autistic children comForted
dren attended differentially depending on their parent, while no normal, six mentally
who appeared to be distressed, a 3 (group) retarded, and two autistic children com-
X 2 (demonstrator) repeated-measures Forted the experimenter. The autistic chil-
ANOVA was calculated. There was a sig- dren were somewhat more likely than the
Sigman et al. 801
TABLE 3
MEANS FOR THE BEHAVIORAL VARIABLES FOR THE DISTRESS CONDITIONS

Variables Autistic MR Normal


Parent distress:
Duration of active play:
M 13.45* 5.87 7.34
SD (12.49) (8.75) (8.47)
Empathy rating:
M 2.41** 4.23 4.14
SD (1.48) (1.30) (1.03)
Experimenter distress:
Duration of active play:
M 15.57** 2.40 1.03
SD (11.72) (5.78) (3.42)
Empathy rating:.
M 2.87** 4.73 4.73
SD (1.59) (1.08) (.69)
NOTE.—Means and standard deviations for durations are given in seconds.
* Means for autistic group are signiflcantly different from means for the mentally retarded
group, p < .05.
** Means for autistic group are significantly different from means for the other two groups,
p < .05.

children in the other groups to withdraw was seated in a child-sized chair with the
physically From the situation and were less experimenter seated slighdy in Front and to
likely to seek the parent during the experi- the child's right and the parent seated
menter distress situation. Ten normal, six slightly ahead and to the child's leFt. The
mentally retarded, and only two autistic chil- parent and experimenter Faced each other
dren approached their parent when the ex- and were seated perpendicularly to the
perimenter pretended to have hurt herselF child. A small (approximately 11 x 8 cm)
electronic robot (a "chatbox" toy manufac-
Facial Affect tured by Tomy) that was positioned inside a
Data For three affect codes (positive, curtained box at one end oF the room was
negative, and neutral) were compared in activated by remote control. The robot
three-group MANOVAs For the parent and moved Forward toward the child For 5 to 10
experimenter distress situations. There were sec, stopping approximately 50 inches From
no significant group effects in either analy- the child and in direct view oF the child.
sis. In both situations, the children showed
mostly neutral affect (20-25 sec), just a little Upon seeing the robot, both individuals
positive affect (3-6 sec), and only about 1-2 modeled expressions oF Fear For 30 sec us-
sec oF negative affect. ing Facial expressions, gestures, and vocal-
izations that were not words. The period
Responses to the Fear of Others concluded with 10 sec in which both indi-
viduals displayed neutral aFFect. In this pro-
METHOD
cedure, the experimenter displayed expres-
Subjects sions along with the parent in order to
The subjects were the same as in the maximize the likelihood that peak emotional
previous study except that the data oF one expressions would be accurately displayed
autistic child in each session were not re- and maintained For the entire test trial.
corded because oF equipment Failure. Given that our principal interest was in
Procedure the children's responses to the negative
The children's responses to Fear ex- emotional expressions oFothers, the first pre-
pressed by others were measured with a par-" sentation oF the robot was combined with
adigm developed For studying social reFer- negative emotional messages From parent
encing in normal children (Hornik & and experimenter. The impact oF the par-
Gunnar, 1988; Hornik et al., 1987; Klinnert, ent's display oF negative emotion has been
1984; Klinnert et al., 1983, 1986). The child shown to be mitigated by a previous expo-
802 Child Development
sure oF the robot accompanied by a positive (to experimenter, parent, robot, and away)
display oF emotion by the parent (Hornik et and two behavioral measures (active play
al., 1987; Walden & Ogan, 1988). Nonethe- with toy and hesitancy rating). The group
less, in this study the robot was presented to eFFect was significant. Wilks's lambda
the child again, during the second session F(12,162) = 3.98, p < .0001, as were the
and with the adults showing positive emo- group effects in all the Follow-up ANOVAs.
tions, in order to dissipate any Fear that
might have developed in the first session. Attention to Emotional Displays
In terms oF looking at the adults, the re-
The behavioral coding system was simi- sults were similar to those shown in the dis-
lar to that used with the distress situations. tress situation. The autistic children looked
Total duration and latency oF attention to ex- at botb adults Far less than the normal and
perimenter, parent, robot, and away were re- mentally retarded children (see Table 4). All
corded. Latency to first touching the robot oF the normal children, 28 oFthe 30 mentally
and duration oFpassive and active play were retarded children, and only 13 oF the 29 au-
coded in the same way as described above. tistic children looked at the adult at some
In addition, a rating oF besitancy was made point during this situation.
using a five-point scale: (1) no hesitation, ap-
proaches the robot quickly to (5) shows ex- There was a significant group x demon-
treme hesitation, no approach. Agreement strator eFFect in a 3 (group) X 2 (demonstra-
between two observers on these behaviors tor) repeated-measures ANOVA calculated
and rating scale with a subset oF 15 subjects to determine whether the children attended
was good; kappas ranged From .81 to .95. Fa- more to one individual than the other,
cial expressions were coded second by sec- F(2,86) = 3.55, p < .03. Both the normal and
ond with the same scales described above. mentally retarded children looked at the ex-
For 15 tapes, percent agreement between perimenter's Face more than the mother's
two raters was 83% for Facial affect, kappa = Face, while the autistic children hardly
.75. looked at either.
The mentally retarded children differed
RESULTS
From the normal children in that they looked
A three-group MANOVA was calculated away more; this was the only significant diF-
on the data From Four measures oF attention Ference in attention or behavior between the

TABLE 4
MEANS FOR ATTENTION AND BEHAVIORAL VARIABLES IN THE FEAR CONDITION

Autistic MR Normal
Variables (iV = 29) (N = 30) (N = 30)
Number who look to adult 13 28 30
Duration of attention:
To experimenter:
M 62** 5.90 4.62
SD (1.57) (5.85) (4.90)
To parent:
M 60** 2.42 3.45
SD (1.40) (2.69) (3.52)
To toy:
M 20.81* 14.45 17.33
SD (8.01) (7.85) (6.85)
Hesitancy rating:
M 3.03* 4.30 3.40
SD (1.64) (1.29) (1.64)
Duration of active play:
M 5.97* 1.53 2.90
SD (8.89) (4.22) (5.41)
NOTE.—Means and standard deviations for durations are given in seconds.
* Means for autistic group are signiflcantly different from means for the mentally retarded
group, p < .05.
** Means for autistic group are signiflcantly different from means for the other two groups,
p < .05.
Sigman et al. 803
normal and mentally retarded groups. The A total oF 13 autistic children looked to
mentally retarded children looked at the toy the adult in the Fear situation. Among those
less than the autistic children. There was no 13 children, 10 looked to the adult during
significant difference between the autistic the first 13 sec (the mean latency to touching
and normal children in how much attention the toy), and only two oF these children ap-
was devoted to the toy. proached the toy. Thus, autistic children
who did not look to the adult were more
Behavioral Responses: Hesitancy and likely to approach the robot than autistic
Active Play children who did look to the adult.
A significant main effect of group was
obtained For the hesitancy rating, F(2,86) = Responses to Experimenter's and Parent's
5.40, p < .006, and the total duration oF ac- Expressions of Pleasure
tive play with the toy, F(2,86) = 3.64, p < Given that the autistic children showed
.03. Autistic children were rated as signifi- so much less attention to the distress. Fear,
cantly less hesitant than the mentally re- and discomFort oF adults than the other chil-
tarded children. They played actively with dren, the question arises whether they also
the robot For about 6 sec, as contrasted with attend less to the adult showing pleasure.
about 3 sec For the normal children and 1.5 When the robot situation was readminis-
sec For the mentally retarded children, \vith tered during tbe second session witb tbe
the contrast between the autistic and men- parent and experimenter modeling pleasure,
tally retarded children being significant. the same group diFFerences were apparent in
this condition as in the Fear condition. Autis-
The number oF children who ap- tic children looked at the experimenter less
proached the toy was 17 (59%), 8 (27%), and than the children in either oF the other
15 (50%) For the autistic, mentally retarded, groups and looked at the parent less than the
and normal groups, respectively. Although mentally retarded children. The normal and
the autistic children were somewhat slower mentally retarded children may have looked
to touch the toy, the differences did not more in this session because oF carry-over
reach significance. Ten normal children, effects From the first session, so group differ-
three autistic children, and none oFthe men- ences in attention to adults' pleasure expres-
tally retarded children approached their par- sion may have been infiated.
ents during this Fear situation.
Facial Affect
As in the distress situation, the children Responses to the Discomfort
in all three groups showed predominantly of Others
neutral affect, and there were no significant METHOD
group diFFerences. The mean duration oF
neutral affect was 17.6 sec For the autistic Subjects
children, 19.7 sec For the mentally retarded The subjects were the same as dis-
children, and 16.5 sec For the normal chil- cussed above except that the experimenter
dren. The means For positive aFFect ranged discomFort situation was not administered to
From 4.5 sec For the autistic children to 7 sec one autistic and one normal child.
For the normal children. Procedure
Effects of Visual Referencing on Behavior In order to determine whether the chil-
One oF the interesting issues is the ex- dren would respond to milder Forms oF nega-
tent to which the children's looks to the tive emotion, the adult who had been play-
adult's emotional display infiuenced their ing with the child pretended to be sligbtly
behavior. It is notable that halFoF the normal ill. In one session, aFter 3 min oF interactive
children approached the toy even when the play with toys, the parent lay down on a
parents modeled Fear, so it seems that the couch, closed her eyes, and pretended to
parent's emotional expression did not have Feel some discomFort For 1 min. In the sec-
an over^vhelming effect on the children's ond session, after 3 min oF interactive play,
readiness to handle the robot. There was a the experimenter pretended to be ill For 1
significant negative correlation between the min. The parent and experimenter were in-
duration oF looking at the adult and the dura- structed to respond to the child but to re-
tion oF active play with the robot For all three main lying on the couch For the entire 1-min
groups, r's ranged From —.37 to —.44, sug- period unless the child became upset (which
gesting that the more the child attended to did not occur).
the adult's Fear expression, the less the child The child's responses were coded For
played with the robot. the occurrence oF looking at the adult, touch-
804 Child Development
ing the adult, or imitating the adult. In addi- mal children imitated parental behavior,
tion, the child's overall affect was coded as mostly by lying down near the parent. Five
(1) positive, (2) neutral, (3) negative, or (4) autistic, six mentally retarded, and 10 nor-
interest. Interest was Included because we mal children showed negative affect to pa-
expected that this might be a Frequent re- rental discomfort, and far fewer showed neg-
sponse. Ghildren's vocalizations regarding ative afFect to experimenter discomFort. Very
the situation were also recorded. On a sam- Few children commented either to their par-
ple oF nine children, there was 100% agree- ent or the experimenter about their be-
ment between two raters on noticing and havior.
touching the adult lying on the couch.
Agreement on imitating the adult was 94%, Diseussion
kappa = .77, and on aFFect rating was 94%,
kappa = .91. The results oF this study are compelling
in that in every situation autistic children
Failed to look very much at an adult showing
RESULTS some Form oF negative emotion. In contrast,
A series oF chi-squares were calculated the normal and mentally retarded children
to determine whether there were group diF- were very attentive to adults showing dis-
Ferences in the number oF children who no- tress. Fear, and discomFort. Furthermore, in
ticed, touched, or imitated the adult who the distress situation, the play oFthe normal
pretended to Feel slightly ill. There were sig- and mentally retarded children with a toy
nificant group diFFerences in the number oF seemed to be inhibited when the adult ap-
cbildren who attended to the adult. HalFthe peared hurt. Not only did the autistic chil-
autistic children did not appear to notice dren ignore or not notice the distressed
their parent's discomFort, which was true for adult, but they also played witb the toy more
only six mentally retarded and four normal and appeared less concerned than the other
children, x^(2) = 11.41, p < .003. While the children. In the Fear and discomFort situa-
group difference was also significant for at- tions, while the autistic children were less
tention to the experimenter's discomfort, the attentive to the adult than the other children,
major difference was between the mentally diFferences in behavior between the groups
reteirded and autistic children, x^(l) — 6.13, were less pronounced. During the fear situa-
p < .01, with the mentally retarded children tion, the autistic children were less hesitant
looking at the experimenter more than the and played with the robot more than the
autistic children. Twenty-two mentally re- mentally retarded children, but neither
tarded children, 12 normal children, and group was significantly difFerent From the
only 11 autistic children looked at the exper- normal children. The fact that the autistic
imenter. Fewer normal children looked at children were as hesitant in approaching
the experimenter than the parent; this differ- and playing with the robot as the normal
entiation was less clear for the mentally re- children raises the possibility that their be-
tarded and autistic children. havior may have been infiuenced by the
demonstrations of fear even if they did not
More normal children (25) touched their attend visually to these demonstrations.
parent than autistic children (11), x^(l) =
11.74, p < .001 or mentally retarded children The hypothesis that there would be
(17), x^(l) = 3.89, p < .05. On the other group differences in emotional reactions was
hand, it should be pointed out that not all not confirmed mostly because the test situa-
the autistic children were oblivious to their tions generally elicited neutral aFFect. For
parent's feigned discomfort: half looked at these young children, someone else's dis-
their parents and one-third touched their tress appeared to be interesting and, oc-
parents. The normal children were much casionally, disconcerting, but not strongly
more likely to touch their parent than the upsetting. Parental withdrawal during the
experimenter, x^(l) = 25.82, p < .001. Many discomFort situation may have been the most
fewer normal children touched the experi- disturbing oF these situations For the normal
menter than touched their parent, but the children, one-third oF whom showed some
children in the other two groups did not dif- negative affect.
ferentiate between tbe parent and the exper- With regard to the lack of attention of
imenter. the autistic children to adults showing dis-
There were no significant group effects tress, fear, and discomfort, many questions
in imitation, affect, or vocalization. Four au- remain to be answered. First, we do not
tistic, nine mentally retarded, and eight nor- know whether young autistic children also
Sigman et al. 805
Fail to look at individuals showing positive not differ in the Frequency or duration oF
emotions, although there is preliminary evi- gaze at their mother's Face (Dawson et al.,
dence From both this study and previous re- 1990). Thus, autistic children do not look
search that they are similarly inattentive. less at others' Faces in all settings, and the
Autistic children have been reported to differences in attention seem greatest when
smile less Frequently in response to their emotion is involved.
mother's smiles than normal children (Daw-
son et al., 1990). Moreover, autistic children Two possible explanations For the lack
smiled less and turned less to their caregiv- oF attention to the emotions oF others can be
ers when their caregivers praised them For suggested. First, autistic children may be in-
successFul completion oF a puzzle than con- nately less emotionally responsive to others
trol children despite showing the same as originally suggested by Kanner (1943).
amount oF pleasure at mastering the puzzle However, autistic children warmly share
beFore they were praised (Kasari, Sigman, positive affect with others in dyadic social
Baumgartner, & Stipek, in press). In Fact, games (Mundy et al., 1986; Sigman et al.,
some oF the autistic children turned away or 1986). Moreover, they show some level oF
physically withdrew when praised, behav- attachment to ceiregivers in that they demon-
iors almost never shown by the control sub- strate responses to separation and increased
jects. Thus, strong positive affect does not proximity seeking upon reunion (Rogers,
appear to attract the attention oF young autis- Ozonoff, & Maslin-Gole, 1991; Shapiro,
tic children any more than strong negative Sherman, Galamari, & Koch, 1987; Sigman
affect. & Mundy, 1989; Sigman & Ungerer, 1984).
Finally, they show nearly equivalent de-
The question can be asked whether au- grees oF emotional expressiveness in some
tistic children simply look less at all objects, social interactions as mentally retarded and
as suggested by Hermelin and O'Gonnor normal children (Yirmiya et al., 1989). Thus,
(1970). The evidence does not seem to sup- they do not seem as emotionally detached as
port this conclusion. In the distress situa- originally described.
tions in this study, the autistic children
looked longer at the toys than the other chil- The second explanation For the inFre-
dren. In our previous study oF social inter- quency oF social reFerencing is that autistic
action with an experimenter, the autistic children may lack the cognitive/affective un-
children looked at toys For an equivalent du- derpinnings For interpreting the emotions
ration oF time as the other children (Kasari shown by others. This lack oF concern with
et al., 1990). While there are times when au- others' attention and aFFect could originate
tistic children appear to show less sustained From the same source as the deficit in theory
attention to objects, this is clearly not true oF mind demonstrated in many studies oF
in all situations. older autistic children (Baron-Cohen, Les-
lie, & Frith, 1985; Perner, Frith, Leslie, &
Another hypothesis is that autistic chil- Leekman, 1989). Presumably, the normal
dren look less at Faces regardless oF emo- child uses others' Facial expressions to learn
tional expression. However, in two diFFerent about the social and inanimate world be-
interactions with their caregivers in our pre- cause the child is aware that others have in-
vious study, autistic children looked as Fre- Formation diFFerent From their own and that
quently and For equal lengths oF time at the this inFormation can be accessed. IF the au-
caregiver's Face as the other children (Sig- tistic child does not understand that other
man & Mundy, 1989; Sigman et al., 1986). people have knowledge. Feelings, or belieFs
The mentally retarded children did show that can be shared, there may be little moti-
more Face-to-Face interaction with their care- vation For checking their faces. Moreover,
givers than the autistic children, but the au- the emotions of others may be so confusing
tistic children did not differ From the nor- that the autistic child learns to ignore or
mals. Moreover, when the parent was asked withdraw from emotional signals.
to play a social game, the amount oF Face-to-
Face interaction increased in all three Because we do not understand the basis
groups, and there were no group differences. for social referencing and joint attention in
Similarly, Dawson and her collaborators vid- normal development, it is diFficult to choose
eotaped children and their mothers during a between these two explanations. Social reF-
free-play period, a more structured period, erencing may involve some capacity to con-
and a Face-to-Face interaction. In all three trast the emotional reactions oF oneselF with
situations, autistic and normal children those oF others. If this is so, the infant may
matched on receptive language ability did have some basic awareness of his or her own
806 Child Development
emotional responses and some recognition rod (Eds.), Infant social cognition (pp.
oF the tie between the emotional reactions 273-314). Hillsdale, NJ: Erlbaum.
and expressions oF others. In 1967, Stechler Camras, L. A., & Sachs, V. B. (1991). Social refer-
and Garpenter proposed that inFants develop encing and caretaker expressive behavior in a
sensory-aFFective intelligence, which uses day care setting. Infant Behavior and Devel-
internal representations that are mediated opment, 14, 27-36.
not by sensorimotor Feedback but rather by Dawson, G., Hill, D., Spencer, A., Calpert, L., &
aFFective experience. The evidence that Watson, L. (1990). Affective exchanges be-
even young inFants use the prosodic con- tween young autistic children and their moth-
trasts presented in "motherese" to gain ac- ers. Journal of Abnormal Child Psychology,
cess to the Feelings and intentions of the 18, 335-345.
caregiver supports this suggestion (Femald, Feinman, S. (1982). Social referencing in infancy.
in press). Merrill-Palmer Quarterly, 28, 445-470.
The final issue concerns the possible Fernald, A. (in press). Meaningful melodies in
consequences of limited attention to the af- mothers' speech to infants. In H. Papousek,
fect of others. A great deal of early learning U. Jurgens, & M. Papousek (Eds.), Origins
about the self and others in inFancy must de- and development of nonverbal vocal commu-
pend on observations oFthe aFFect and atten- nication: Evolutionary, comparative, and
tion oF others in a variety oF situations methodological aspects. New York: Cam-
(Mundy etal., in press; Sigman, 1989). Autis- bridge University Press.
tic individuals are severely handicapped in Gunnar, M. R., & Stone, C. (1984). The effects of
social interactions at all ages. Not all of'these positive maternal affect on infant responses to
limitations may be attributable to precursors pleasant, ambiguous, and fear-provoking toys.
in inFancy. However, the Failure to attend to Child Development, 55, 1231-1236.
the aFFect and attention oF others must limit Hermelin, B., & O'Connor, N. (1970). Psychologi-
the development oF autistic children in cal experiments with autistic children. New
many ways and must contribute to their so- York: Pergamon.
cial deficiencies. A general question that Hobson, R. P. (1986). The autistic child's appraisal
needs to be addressed is whether variations of expressions of emotion. Journal of Child
in attention to others' aFFect among normal Psychology and Psychiatry, 27, 321-342.
inFants has consequences For normal social Hornik, R., & Gnnnar, M. R. (1988). A descriptive
development. analysis of infant social referencing. Child
Development, 59, 626-634.
Hornik, R., Risenhoover, N., & Gunnar, M. (1987).
References Effects of maternal positive, neutral, and neg-
American Psychiatric Association (1987). Diag- ative affective communications on infant re-
nostic and Statistical Manual of Mental Dis- sponses to new toys. Child Development, 58,
orders (DSM III-R). Washington, DC: Ameri- 937-944.
can Psychiatric Association. Izard, G. E. (1979). The maximally discrim,inative
Baren-Cohen, S. (1989). Perceptual role-taking facial m-ovement coding system (MAX). New-
and protodeclarative pointing in autism. Brit- ark: University of Delaware.
ish Journal of Developmental Psychology, 7, Jones, S. S., Gollins, K., & Hong, H. W. (1991). An
113-127. audience effect on smile production in 10-
Baron-Cohen, S., Leslie, A. M., & Frith, U. (1985). month-old infants. Psychological Science, 2,
Does the autistic child have a "theory of 45-49.
mind"? Cognition, 21, 37-46. Kanner, L. (1943). Autistic disturbances of af-
Braverman, M., Fein, D., Lucci, D., & Water- fective contact. Nervous Child, 2, 217-250.
house, L. (1989). Affect comprehension in Kasari, C , Sigman, M., Baumgartner, P., & Stipek,
children with pervasive developmental disor- D. J. (in press). Pride and mastery in children
ders. Journal of Autism and Developmental with autism./o«rnaZ of Child Psychology and
Disorders, 19, 301-316. Psychiatry.
Bruner, J., & Sherwood, V. (1983). Thought, lan- Kasari, C , Sigman, M., Mundy, P., & Yirmiya, N.
guage, and interaction in infancy. In J. D. (1990). Affective sharing in the context ofjoint
Call, E. Calenson, & R. L. Tyson (Eds.), Fron- attention interactions of normal, autistic, and
tiers of infant psychiatry (pp. 38-52). New mentally retarded children. Autism and De-
York: Basic. velopmental Disorders, 20, 87-100.
Campos, J. J., & Stenberg, C. R. (1981). Percep- Klinnert, M. D. (1984). The regulation of infant
tion, appraisal, and emotion: The onset of so- behavior by maternal facial expression. In-
cial referencing. In M. E. Lamb & L. R. Sher- fant Behavior and Development, 7, 447-465.
Sigman et al. 807
Klinnert, M. D., Gampos, J. J., Sorce, J. F., Emde, havior in young children with autism or other
R.N., & Svedja, M. (1983). Emotions as behav- psychiatric disorders. Joumal of the Ameri-
ior regulators: Soeial referencing in infancy. can Academy of Child and Adolescent Psy-
In R. Plutchnik & H. Kellerman (Eds.), The chiatry, 30, 483-488.
emotions (Vol. 2, pp. 57-86). New York: Aca- Schopler, E., Reichler, R. J., & Renner, B. R.
demic Press. (1986). The childhood autism rating scale.
Klinnert, M. D., Emde, R. N., Butterfield, P., & New York: Irvington.
Campos, J. J. (1986). Social referencing: The Shapiro, T., Sherman, M., Galamari, G., & Koch,
infant's use of emotional signals from a D. (1987). Attachment in autism and other de-
friendly adult with mother present. Develop- velopmental disorders. Journal of the Am,eri-
mental Psychology, 22, 427-432. can Academy of Child and Adolescent Psy-
Krug, D. A., Arick, J. R., & Almond, P. J. (1978). chiatry, 26, 485-590.
Autism Screening Instrument for Educa- Sigman, M. (1989). The application of develop-
tional Planning. Portland, OR: ASIEP Educa- mental knowledge to a clinical problem: The
tional Co. study of childhood autism. In D. Gicchetti
Loveland, K. A., & Landry, S. H. (1986). Joint at- (Ed.), The emergence of a discipline: Roches-
tention and language in autism and develop- ter symposium on developmental psychopa-
mental language delay./ournaZ ofAutism and thology (Vol. 1, pp. 165-189). Hillsdale, NJ:
Developmental Disorders, 16, 335-349. Erlbaum.
MacDonald, H., Rutter, M., Howlin, P., Rios, P., Sigman, M., Mundy, P., Sherman, T., & Ungerer,
Le Conteur, A., Evered, C , & Folstein, S. J. A. (1986). Social interactions of autistic,
(1989). Recognition and expression of emo- mentally retarded and normal children with
tional cues by autistic and normal adults. their caregivers. Journa/ of Child Psychology
Journal of Child Psychology and Psychiatry, and Psychiatry, 27, 647-656.
30, 865-877. Sigman, M., & Ungerer, J. A. (1984). Attachment
Mundy, P., Kasari, G., & Sigman, M. (in press). behaviors in autistic children./ournai of Au-
Nonverbal communication, affective sharing tism and Developm,ental Disorders, 14,
and intersubjectivity. Infant Behavior and 231-244.
Development. Snow, M. E., Hertzig, M. E., & Shapiro, T. (1987).
Mundy, P., Sigman, M., Ungerer, J. A., & Sher- Expression of emotion in young autistic chil-
man, T. (1986). Defining the social deficits in dren. Journal of the American Academy of
autism: The contribution of non-verbal com- Child and Adolescent Psychiatry, 26,
munication measures. Journal of Child Psy- 836-838.
chology and Psychiatry, 27, 657-669. Stechler, G., & Garpenter, G. (1967). A viewpoint
Ozonoff, S., Pennington, B., & Rogers, S. (1989). on early affective development. In J. Hell-
Are there emotion perception deficits in your muth (Ed.), The exceptional infant (Vol. 1, pp.
autistic children? Journal of Child Psychol- 163-189). Seattle: Special Ghild Publica-
ogy and Psychiatry, 51, 343-361. tions.
Ozonoff, S., Pennington, B., & Rogers, S. (in Walden, T. A., & Ogan, T. A. (1988). The develop-
press). Executive function deficits in high- ment of social referencing. Child Develop-
functioning autistic children: Relationship to ment, 59, 1230-1240.
theory of mind. Journal of Child Psychology Yirmiya, N., Kasari, G., Sigman, M., & Mundy, P.
and Psychiatry. (1989). Facial expressions of affect in autistic,
Perner, J., Frith, U., Leslie, A. M., & Leekman, mentally retarded and normal children. Jour-
S. R. (1989). Exploration of the autistic nal of Child Psychology and Psychiatry, 30,
child's theory of mind: Knowledge, belief 725-735.
and communication. Child Development, 60, Yirmiya, N., Sigman, M., Kasari, G., & Mundy, P.
689-700. (in press). Empathy and cognition in high-
Prior, M., Dahlstrom, B., & Squires, T. (1990). Au- functioning children with autism. Child De-
tistic children's knowledge of thinking and velopment.
feeling states in other people. Journal of Zahn-Waxler, G., & Radke-Yarrow, M. (1990). The
Child Psychology and Psychiatry, 31, origins of empathic concern. Motivation and
587-601. Emotion, 14, 107-130.
Rheingold, H. R., Hay, D. F., & West, M. T. Zahn-Waxler, G., Radke-Yarrow, M., Wagner, G.,
(1976). Sharing in the second year of life. & Ghapman, M. (in press). Development
Child Development, 47, 1148-1158. of concern for others. Developmental
Rogers, S. J., Ozonoff, S., & Maslin-Gole, G. Psychology.
(1991). A comparative study of attachment be-

You might also like