002% 3932/81:020263-10S02.00/0 c’ 1981 Pergamon Press Lid
Neump.~ycholoyio, Vol. 19, No. 2, pp 263-272, 1981 Prmted in Great Bntain.
PERCEPTION
AND EXPRESSION OF EMOTION HANDERS AND LEFT-HANDERS
IN RIGHT-
WENDY HELLER and JERRE LEVY* Department
of Behavioral
Sciences,
University
of Chicago
(Receioed 23 JulJl 1980) Abstract-Photographs of facial composites, half the face smiling and the other half not smiling, were tachistoscopically presented to right-handed and left-handed subjects. Both handedness groups perceived faces as happier when it was the left side of the poser’s face that was smiling, and this held for both left- and right-handed posers. Additionally, right-handed, but not left-handed, subjects perceived faces as happier when the smiling half-face was in the left visual field.
RECENT studies [l-5] have suggested that emotions are expressed more intensely on the left side of the face. The relation of this asymmetry to cerebral lateralization remains unclear. CHAURASIA and GOSWAMI Cl], though finding a left-side dominance for smiling in righthanders, found a right-side dominance for smiling in left-handers. SACKEIM and colleagues [2,3] found that negative emotions were expressed more intensely on the left side of the face in a predominantly right-handed sample, but found no significant asymmetry in the expression of happiness. CAMPBELL [4] found the left side of the face to express more happiness in right-handers, and subsequently found the same asymmetry in favor of the left side of the face in left-handers [S]. However, she also found that, whereas right-handers no: only express more happiness on the left, but also more unhappiness (conforming with thk Sackeim and colleagues studies), left-handers expressed more unhappiness on the right of the face. MOSCOVITCH and OLDS [6] did not specifically investigate facial emotion, but they found that the left side of the face in right-handers produced more facia! expressions than the right side; they observed no consistent asymmetries in left-handers. The literature is, thus, inconsistent with respect to the types offacial emotions for which a left-side advantage is seen in right-handers and with respect to whether left-handers have a left-side, right-side, or no asymmetry in facial expression. It would not be unreasonable to suppose that the control of emotional expression might selectively depend on one hemisphere since there is clear evidence to suggest that the hemispheres are differentiated with respect to the discrimination of facial emotion. BUCHTEL et al. [7] found a left visual field (LVF) superiority in right-handers for perceiving facial emotion. CAMPBELL [4] found that subjects judged faces as happier when a smile was in the LVF and a neutral expression in the right visual field (RVF), as compared to faces when the smile was in the RVF and the neutral expression in the LVF. SAFER [S] also found a LVF advantage for judging whether two facial emotions were the same or not, an advantage that was enhanced when subjects were instructed to empathize with, rather than to verbalize, the facial emotion. If the right hemisphere is differentially competent in right-handers to *Requests for reprmts III. 60637, U.S.A.
should
be sent to Jerre Levy, 5848 S. University
Ave., Umbcrsity
of Chicago.
Chicago,
264
WENDY HFLLFK and JEKK~ LEVY
discriminate facial emotion, it also might be differentially competent to produce facial emotion. Similarly, if the nonverbal hemisphere of left-handers is selectively advantaged in the perception of facial emotion, the specializations underlying that advantage might yield better control of facial emotion. To further investigate the perception and expression of emotion. we obtained judgments from highly restricted subgroups of left- and right-handed subjects of facial emotion in similarly restricted subgroups of left- and right-handed posers that allowed us to ascertain whether subjects had a lateralized advantage for perceiving emotion and whether posers displayed an asymmetry in the expression of emotion.
Our subjects consisted of 12 right-handed males with a normal hand posture. no first-degree left-handed relntlves. and right-eye dominance for sighting, and 12 left-handed males with a normal hand posture, at lcast one tirst-degree left-handed relative,and left-eye dominance for slghtmg. We had 5 right-handed posers with the same characterlstlcs as right-handed subjects and 4 left-handed posers with the same characteristics as left-handed subjects. We had hoped that by restricting our subject and poser samples to males havmg congruent eye hand-familial handedness lateralities, all with normal hand posture. within-group variation would be reduced and between-group differences maximized. Stimuli Following the method of CAMPaC1.L [4], each poser was photographed in a nonsmiling (neutral) and in a smilmg pose. Smiles were elIcIted from posers by telling them to smile, bq makmglokes. and by smiling at the poser (which generally ehcited spontaneous, reactive smiles). The smllmp poser was photographed at the moment when the photographer felt that a natural. reactive smile was present. Two prints were made of the nonsmilme and smiling photographs of each poser, one printed m normal orientation (left side of poser’s face on right. right side of pohcr’s face on left) acd one printed m mirror-reversed orlcntatlon (left side of poser’s face on left, right side of poser’s face on right). The four photographs of each poser were carefully cut down the middle and composites wcrc made from the pair (smiling and nonsmiling) of normally orlented prints and another set of composites was made from the pair (smiling and nonsmiling) of mirror-rcverscd prints. In each composite photograph, half the Pace was smiling and the orhcr half of the face w;is nonsmIling. In one member of each pair, the smiling half-face was on the left of the photograph (the right half of the poser’s fact for normally oriented prints and the left half of the poser’s face for mirror-reversed prints), and m the other member of the pair,the smdmg half-face was on the right ofthe photograph (the left halfof the poser’s fdce for normally orlcntcd prints and the right half ofthe poser’s fact for mirror-reversed prints). Figures I and 2 show examples of the rcsullant composite pictures. The four composites from each poser were aIrbrushed and rephotographed to remo\t‘ an) obvious discrepancies at the mldline. For easeofexposition. the different types ofcomposites will bc gicen the followmg dcslgnatlons: upper-cast letters will be used to deslgnate which side ofthephoto(/rop/l [left (L) or right IR)] contamc the vmlmg half-face, and Iowcrcase letters wilt be used to designate which half of the pc~r’.> /~cc, [left (I) or right (r)] 1s smiling. Thus. the RI composite has the smile on the right of the photograph and I( IS the left side of the poser’5 face that IS smlllng. the Lr composite has ihesmile on theleft side ofthephotograph and 11~srheright halfof theposer‘sfacc that ~ssn~lmg. The RI and Lr composities are derived from the normally oricntcd prints. The Rr composite has the smile on the right of the photograph and it is the right side ofthc poser’s Face that is smiling: the LI composite has the smlr on the left sldc of the photograph and it IS theteft halfof the poser’s fact that IS smlhng. The Rr and LI composites are dcriccd from the mirror-reversed prmts.
Each poser provided two pairs ofstimuh. the RI and Lr pa,r and the Rr and Li p;ilr. Thus. ulth 9 posers. there were a total of 18 pairs of stimuli, halfderlved from normally orlentcd prints and half derived from mirror-rcvcrted prints. On each tachistoscopic trial, a pair of stirnull was shown, the first member of the pair dlbplaycd for I SOmscc in midlme,foHowed by a 200 msec blank field with center fixation X. and then the othrl- member of the pair for I50 msec in midhne. On each trial, subjects were asked to specify which member of the pair looked happler. Each of the 18 pairs of stimuli was shown twice to each subject for a total of 36 trials per subject. On one trial for each pair, the RI or Rr composite was shown first. and on the other trial for that pair. the Lr or Lt composite was shown first so that possible effects of order of composites H IthIn ;I trial were controlled
PERCEPTION
AND
Normal
EXPRESSION
OF EMOTION
Orientation
a
b Mirror
Orientation
d FIG. 1. The four composite faces of poser JR, a right-handed poser in whom the right side of the face was judged as expressing more happiness. Thus, b was judged happier than a 75% of the time, but d was judged happier than c only 54% of the time.
265
266
WENDY HELLER and JERRE LEVY
Normal
Orientation
b
a
Mirror
Orientation
FIG. 2. The four composite faces of poser KC, a left-handed poser in whom the left side of the face was judged as expressing more happiness. Thus, b was judged happier than a only 20.5 7; of the time, but d was judged happier than c 87% of the time.
PERCEPTION
AND
FXPRESSION
267
OF FMOTION
RESULTS Perceptual
bias
in subjects
In order to examine whether subjects were biased to judge faces happier when the smile appeared in the LVF, the proportion of responses in which Lr + Ll composites were called happier was determined for each subject (hereafter designated as “LVF responses”). These proportions are shown in Table 1. Table
I. Proportion
Subjects
I 2 3 4
5 6
7 8 9 IO II I2 Mean SE
of trials on which subjects judged faces happier when the smde was in the LVF
Proportion of responses to smiles in the LVF* Left-handed subjects Right-handed subjects Orientation Orientation Normal Mirror MW” Normal Muor Mea”
0.500 0.444 0.500 0.61 I 0.778 0.278 0.444 0.61 I 0.722 0.556 0.278 0.500 O.Sl8 0.044
0.889 0.61 I 0.556 0.833 0.833 0.778 0.444 0.722
0.583 0.556 0.556 0.S56 0.833 0.444 0.500 0.722 0.778 0.667 0.36 1 0.61 1
0.61 I 0.833 0.278 0.444 0.722 0.389 0.500 0.61 I 0.444 0.167 0.333 0.333
0.556 0.667 0.444 0.556 0.6 I I 0.61 I 0.61 I 0.500 0.667 0.500
0.676 0.040
0.598 0.040
0.430 0.054
0.634 0.047
0.667 0.667 0.61 I
0.500
I’horm.\,,rr =0.75
1
0.889
I.000
0.750 0.9 17 0.417 0.556 0.333 0.472 0.556
061 I 0.528 0.333 0.500 0417 0.533 0 049
I’\,,“,. \,,rr =0.x73
*Proportions are based on I8 trlala subject oruzntatlon. and represent the proportion of trials on which subjects judged ;I composite happIer &hen the smile has m the LVF. Means and standard errors of means are given at the bottom of each column.
For right-handed subjects, the mean proportion of LVF responses was 0.598 with a standard error of 0.040, which exceeds 50”;, at t ( 11) = 2.450, P = 0.0 18. 1-tailed. conforming with findings of others that there is a LVF advantage for discriminating facial emotion. Nine out of 12 dextrals gave a majority of their judgments to Lr + LI composites, 1judged LVF and RVF smiles to look happier an equal number of times, and 2 had a bias for the RVF. Among left-handers, there was no significant difference in LVF and RVF judgments. the mean proportion of LVF responses being 0.533 with a standard error of 0.049, which does not significantly differ from 50%. The failure to observe an overall bias for left-handed subjects in favor of one visual field or the other conforms with neurological findings that lefthanders are heterogeneous in their patterns of lateralization, and provides support for the hypothesis that the LVF advantage for the discrimination of facial emotion in right-handers derives from cerebral asymmetry of function. As may be seen in Table 1, however. there were large individual differences even among right-handers in the direction and/or degree of asymmetry of judgments. If the betweensubject differences within the two handedness groups are not merely random error, but represent reliable individual-difference characteristics. than the proportions of LVF
268
WENDY HFLLER and JERRE LEVY
responses given by subjects to faces in normal orientation should be highly correlated with the proportions given to faces in mirror orientation. For right-handers, it was found that the correlation in frequencies of LVF responses for the two conditions was 0.751 [t (10) =3.597, P-cO.005, l-tailed]; for left-handers, the correlation was 0.873 [t (10) = 5.660, P
oftrials on which subjects gaye”happier”judgments when smiles were produced of posers’ faces, independent of the visual field in whtch the smile appeared
SuhJects
Proportton of responses Right-handed subjects Rtght-handed Left-handed posers posers
to left side of posers’ faces* Left-handed subjects Rtght-handed Left-handed posers posers
I2
0.550 0.550 0.650 0.400 0.600 0.700 0.500 0.550 0.500 0.600 0.550 0.550
0.625 0.688 0.438 0 500 0.500 0.625 0.625 0.688 0.625 0.625 0.625 0.688
0.550 0.600 0.600 0.550 0.600 0.550 0.450 0.450 0.600 0.700 0.700 0.550
0.750 0.562 0.688 0.688 0.625 0.625 0.688 0.562 0.562 0.625 0.625 0.625
Mean SE
0.558 0.022
0.604 0.024
0.575 0.02 1
0.636 0.016
8 9
IO II
*Proporttons for right-handed posers are based on 20 triak’subject; proporttons for left-handed posers are based on 16 trtals,subject. The yalues represent the proportion of trials on which subjects judged a composite happier when the smile was produced hq the left side of posers’ faces, regardless of whether tbir was seen tn the LVF or tn the RVF. Means and standard errors of means are giycn at the bottom of each column.
hq theleft half
PERCEPTION
AND EXPRESSION
OF
EMOTION
269
judged faces happier when it was the left half that was smiling, and this held for both left- and right-handed posers. Our data, therefore, support those of CAMPBELL [4, 51, but contrast with those of CHAURASIA and GOSWAMI [l] with respect to left-handers. CHAURASIA and GOSWAMI, however, reported that they examined “lateral movements of the angles of (the) mouth” and recorded “the dominant side of spontaneous smile”. Their index of smiling dominance was, therefore, confined to movements and position of the mouth, while Campbell’s index and ours did not specify to judges which facial features were to be used for assessing comparative happiness. As may be apparent from examining the composites from Poser KC (Fig. 2), a left-hander judged to express more happiness on the left side of his face, the happier look of composite d, compared to b, or of composite a, compared to c, may have little to do with the mouth angle. If the lower part of the face is covered, it seems to us that, on the smiling half-faces, the eyes convey more happiness in d and u than in b and c. MOSCOVITCH and OLDS [6]did not record relative happiness of the two sides of the face, but rather the frequency of expressive movements made with the two sides of the face, so their results cannot be compared to ours. SACKEIM,GUR and SAUCY [2] and SACKEIMand GUR [3] did not find that happiness was judged as more intense when the left side of the face produced the emotion. In their procedure, left-left composites and right-right composites that were perfectly symmetric were compared; possibly, judgments of subjects are significantly affected by whether they must compare two asymmetric composites successively presented at tachistoscopic speeds vs a comparison of two simultaneous symmetric composites presented in free vision at long exposures. However, CAMPBELL [4] only used the asymmetriccomposite paradigm with tachistoscope in her first study with right-handers, in addition to the symmetric-composite, booklet technique,getting similar results with both procedures. In her second study [S], where left-handers were examined, only the symmetric-composite technique used by Sackeim and colleagues was employed. In pilot studies with the symmetric-composite, booklet technique, we observed no asymmetries ofjudgment in favour of either the left or right side of faces. We have no explanation for the various discrepancies, but would suggest that the weight ofevidence supports the view that the left side of the face in both left- and right-handers asymmetrically expresses happiness. From the means and standard errors displayed at the bottom of Table 2, it may be seen that both left- and right-handed subjects, in responding to both left- and right-handed posers, gave judgments in favor of the left side of posers’ faces significantly more than 50% of the time at PcO.025 or better (df= 11 for all four comparisons). In view of the great variability seen among subjects in the degree and direction of perceptual bias in perceiving happiness, it was of interest to examine possible variations among posers in degree and direction of expressing happiness. Figure 3 shows the proportion of responses for smiling left half-faces (Rl+ Ll composites) for right- and lefthanded subjects for each of the posers separately. First, it may be seen that responses of rightand left-handed subjects are highly similar to each of the posers [r =0.892, t (7) = 5.221. P
270
WENDY HELLER and JERRE LEVY
Posers FIG. 3. Proportions of responses in which the left side of the poser’s face was judged as expressing more happiness. shown for each poser indwidually and for right-handed (RH) and left-handed (LH) subjects. There were four trials per subject per poser; the proportion of these in which the left side of the face was judged happier was calculated for each subject and the mean and standard error were computed for each group of subjects for each poser. The handedness of posers. rgh-handed (RH) or left-handed (LH), is designated below each poser’s initials. The error bars arc standard errors of mean proportions.
face, and this is significant for posers MS [t (23)=2.112, PcO.05, 2-tailed], JW [t (23)=4.950, P
bias and facial asymmetry
In order to assess whether the effects of perceptual bias and facial asymmetries were additive or not in their effect on subjects' judgments, the difference in proportions of LVF responses for composites in mirror and normal orientation (reflecting facial asymmetries) was correlated with the sum of the proportions of LVF responses under the two orientation conditions (reflecting perceptual bias). Nonadditivity of the two effects would be revealed by a significant non-zero correlation between these measures. It will be noted that the covariance between the difference in two scores and their sum is equal to the difference in variance between component scores. The component score with the
PERCEPTION
AND EXPRESSION
OF
EMOTION
271
higher mean (where the two factors presumed to control responses are both operating in the same direction) will have the greater variance if the underlying factors have a positive covariance since this will increase the spread in scores. The component score with the lower mean (where the two factors presumed to control responses are operating in opposite directions) will have the high variance if the underlying factors have a negative covariance. This can be made more concrete by considering the judgments of right-handers for normally oriented faces. For these stimuli, the perceptual bias effect tends to produce responses for smiles in the LVF, but the facial asymmetry effect tends to produce responses for smiles in the RVF. If the two effects are negatively correlated, then subjects with a strong bias for the LVF will only have a weak tendency to attend to actual facial asymmetries, and will display a strong LVF bias. Subjects with little perceptual bias will be strongly sensitive to facial asymmetries and will, therefore, display a strong RVF bias. Scores will be spread out by the negative correlation between factors for the condition in which the factors operate in opposing directions. For right-handers, there was a nonsignificant correlation of - 0.124, and for left-handers, there was a nonsignificant correlation of -0.288, showing that perceptual bias and facial asymmetries were almost purely additive in their effects on subjects’judgments. To the extent that a nonadditive interaction was present, subjects with a stronger perceptual bias reacted less to actual facial asymmetries. DISCUSSION ‘Our data show, in conformance with findings of others, that right-handers as a group have a LVF advantage for the discrimination of facial emotion. Left-handed subjects were found to display no overall advantage for either visual field, though individual left-handers manifested reliable advantages for one field or the other. Correlational analyses demonstrated that differences between subjects within both handedness groups were not random, but reflected stable characteristics of subjects with respect to both direction and degree of perceptual bias. We suggest that hemispheric specialization for the discrimination of facial emotion is more variable in right-handers, and perhaps in left-handers, than is lateralization for the cognitive aspects of verbal and nonverbal processes. Though a majority of right-handers appear to be specialized for the perception of emotion in the nonverbal hemisphere, this does not apply to all, and even in those cases where it does apply, there is a significant variation in the magnitude of the LVF advantage. Confirming CAMPBELL [4, 51, we found that both left- and right-handed posers express more happiness on the left side of the face. Further, the expressive asymmetry is equally perceived by left- and right-handed subjects, who are highly similar in their responses to individual posers. Not all posers showed more expression on the left side of the face, and they varied, also, in the degree of facial asymmetry for emotional expression. Two of the four lefthanded posers manifested the left-side asymmetry, but two did not. Among right-handed posers, four of five had the left-side asymmetry, two significantly so, and one had a significant right-side asymmetry. The variation among posers in asymmetry of facial expression is, thus, similar to the variation among subjects in asymmetry of perception. Whether the control of facial expression is asymmetrically dependent on the same hemisphere that is specialized for the perception of facial emotion in unknown. It is at least conceivable that, in any given individual, though cognitive lateralization may be a poor predictor of which side of the face is dominant in expressing emotion, facial asymmetry in emotional expressivity may be strongly correlated with asymmetry in the perception of
272
WENDY HELLER and JERRE LEVY
emotion. In future studies, it will be of interest to examine asymmetry of facial emotion in posers whose perceptual asymmetries for the discrimination of facial emotion is assessed. When our study was initiated, we had nai’vely supposed that if subjects and posers were highly homogeneous in characteristics associated with the lateralization of cognitive functions, there was a good probability that they would also be homogeneous in the lateralization of emotional functions. Our data, both with respect to the perception of facial emotion and with respect to the expression of facial emotion, belie our supposition. Given the reliable variations among both subjects and posers, not only for left-handers, but also for right-handers, we believe that any conclusions regarding the relationships among perceptual asymmetries in the discrimination of emotion, facial asymmetries in the expression of emotion, and cerebral asymmetry for cognitive processes would be premature. AckJlo~~/rdyrmrnt.s~~Support by grant BNS75-23061 from the National Science Foundation and by a grant from the Spencer Foundation is gratefully acknowledged. We thank J. KAISEK for his help in stimulus preparation and data collection. We are especially happy to express our gratitude to Dr. RUTH CAhwetr.r. who generously sent a preprint of her work with right-handers prior to its publication. and whose innovative studies stimulated the research reported here. We should also like to thank two anonymous reviewers for their valuable suggestions and for bringing to our attention references of which we had been unaware.
REFERENCES I. CHAURASIA, B. D. and Goswnw~, H. K. Functional asymmetry in the face. Artu A~trrotniccr 91, 154 160, 1975. 2. SACKFIM, H. A., GUR, R. C. and SAUCY. M. C. Emotions are expressed more Intensely on the left side of the face. Scierlcc, 202, 434436. 1978. 3. SA(.KEIM. H. A. and CUR, R. C. Lateral asymmetry in intensity of emotional expression. Neurop”!‘c’hol”(licr)~.s~~/~~/~~~;~~ 16, 473.481. 1978. 4. CAMPBELL.,R. Asymmetries in interpreting and expressing a posed facial expression. Corrti.~ 14, 327-342, 1978. 5. CA>~vnt.t.L, R. Left-hander’s smiles: asymmetries in the projection of a posed facial expression. Corrr.~ 15, 571 579. 1978. 6. MOXOVITH. M. and OLDS, J. AsymmetrIes m emotional, facial expressions and their possible relation to hemispheric specialization. Paper presented at the I~rrr,~urior~rr/ N~,urop.s~~ho/og~ Socirty. Holland. June 1979. 7. BL CHTEL. H.. CAMpAru. F.. Dt- RISI~, C. and ROTA, R. Hemispheric difference in the discrimination reaction time to facial expresslons. Irul. J. P.s~d~o/. 5, 159 169. 1978. 8. SAFFR. M. A. Sex and hemisphere differences in access to codes for processing emotional expressions and faces. J. C.Yp. Psych/. : &,I.. in press. Resume
: On
et
.S des
d’une
grapes
lorsque
c’btait sujets
les
En outre, le
plus champ
des
souriante de
que
tachistoscopiquement
gauchers
h&niface
deux
ges
pr6sentS
a
sujets
les
sujets le
et
de
gauche
photographi6s droitiers
heureux
qui
mais
non
de visages
souriait
l’h&niface
non
droitiers
composes souriante.
comme plus et
eux-&mes les
sujets
visages
h6miface
les
soient
lorsque
visuel
l’autre
percevaient
c&6
2 des
photographies
ceci
Btait
droitiers
gauchers
retron+ ou
percevaient
souriante
Btait
Les
heureux gauchers. les
projet&
visadans
gauche.
Zusammenfassung Zusammengesetzte
Fotografien,
HLlfte ein nicht lxchelndes und linkshtidigen Gesichter
Versuchspersonen
als gliicklicher
llchelte.
Zus?itzllch
personen
Gesichter
linken
Gesichtsfeld
die eine Hllfte
Gesicht
wahr,
nahmen
war.
dargeboten.
wahr,
die andere
tachistoskopisch Beide
wenn die linke Site
rechtshtidlge,
als gliicklicher
ein lfichelndes,
zeigend,wurden
rechtshtidigen
Htidigkeitsgruppen
des exponierten
aber nicht linkshsndige wenn die Mchelnde
nahmen
Gesichtes Versuchs-
GesichtshWfte
im