You are on page 1of 12

BRAIN

AND

LANGUAGE

7, 127-138 (1979)

Hemispheric Differences in Processing


Emotions and Faces
ROBERT

G. LEY AND M. P. BRYDEN


University of Waterloo

Visual field differences for the recognition of emotional expression were investigated using a tachistoscopic procedure. Cartoon line drawings of five adult male
characters, each with five emotional expressions ranging from extremely positive
to extremely negative, were used as stimuli. Single stimuli were presented unilaterally for 85 msec. Subjects (N = 20) were asked to compare this target face to a
subsequent centrally presented face and to decide whether the emotional expressions of the two faces, or the character represented by the two faces, were the
same or different. Significant left visual field (LVF) superiorities for both character and emotional expression recognition were found. Subsequent analyses demonstrated the independence of these effects. The LVF superiority for emotional
judgments was related to the degree of affective expression, but that for character recognition was not. The results of this experiment are consistent with experimental and clinical literature which has indicated a right hemispheric superiority
for face recognition and for processing emotional stimuli. The asymmetry for
emotion recognition is interpreted as being an expression of the right hemispheres
synthetic and integrative characterisitics, its holistic nature, and its use of imagic
associations.

Numerous studies employing a variety of techniques have demonstrated the lateralization of cognitive abilities. Despite the proliferation of
studies investigating the duality of mans brain and behavior, the purview
has been largely cognitive, and the emotional side of mind and man has
been ignored. Scant reference exists in the literature to the possible
hemispheric specialization for emotional functions. The present study
investigates the hypothesis that the human brain is laterally or .differentially specialized for processing emotional stimuli.
The data presented in this paper were drawn from an M.A. Thesis submitted by the senior
author. This work was supported by a Research Studentship from the Ontario Mental Health
Foundation to R. G. L. and by Grant A-95 from the National Research Council of Canada to
M. P. B. An earlier version of the paper was presented at the Psychonomic society meetings
in Washington, D.C., November 1977. Requests for reprints should be addressed to Professor M. P. Bryden, Department of Psychology, University of Waterloo, Waterloo, Ontario
N2L 3G 1. Canada.
127
Copynghr
@I 1979 by Academic Press, Inc.
All rights of reproduction
in any form reserved.

128

LEY AND BRYDEN

Several lines of clinical evidence point to possible hemispheric specialization for emotions. Research on split-brain patients (Sperry, Gazzaniga,
& Bogen, 1969), anosognosics (Critchley, 1953; Weinstein & Kahn, 1955),
and patients with unilateral brain lesions (Goldstein, 1939; Hecaen, 1962;
Gainotti, 1969) has revealed characteristic emotional responses accompanying stimulation or injury lateralized to one hemisphere. Terzian
(1964) has reported that intracarotid injection of sodium amytal produces
different affective behavior when injected on the left or right sides. Research on the therapeutic efficacy of electroconvulsive shock treatment
(ECT) has shown differences in the response of the two hemispheres to
ECT (Galin, 1974; Robertson & Inglis, 1977), perhaps indicative of differential hemispheric involvement in pathological depression.
In general, these reports suggest a depressive-catastrophic response
following left-hemispheric intervention, and a euphoric-maniacal response to right-hemisphere disturbance.
The clinical literature thus suggests an asymmetry of emotional expression following intervention or injury to one hemisphere. However, the
emotional behavior produced by left-hemispheric invasion may be a byproduct of the associated language disturbances and thus essentially artifactual. If this is the case, a predominantly right-hemispheric mediation
of emotional processes might be indicated. A number of studies of laterality effects in normal individuals support such an interpretation. For
example, Haggard & Parkinson (1971) asked subjects to identify both the
emotional intonation and the verbal content of sentences which were
dichotically presented in competition with a continuous babble. The
stimuli consisted of six sentences read in four emotional tones: anger,
boredom, happiness, and distress. They reported a slight left-ear advantage (LEA) in identifying the emotional tone of the sentence. Carmon and
Nachson (1973) used a dichotic procedure to investigate hemispheric
asymmetry for perception of nonverbal sounds: the cries, shrieks, and
laughter of a child and of an adult male and female. They also found a
slight but significant LEA.
Safer and Leventhal (1977) had subjects listen to taped monaural passages that had three levels of content: positive, negative, and neutral.
These passages were then read in three tones of voice: positive, negative,
and neutral. The positive tone passageswere read in an exuberant, happy,
almost laughing fashion. The negative tone passages were read in a
strong, angry sounding voice and the neutral passages were read in an
objective manner. Subjects listened to either the left or right ear. Although subjects were not instructed about which cue to use, 29 of 36
subjects who listened on the left ear used the tone of voice cues to rate the
passages. Conversely, 21 of 36 subjects who listened on the right ear used
the content cues to evaluate the passages. These results again suggest the
involvement of the right hemisphere in processing emotional stimuli.

EMOTIONS AND FACES

129

In visual experimentation, Dimond and Farrington (1977) have reported


that heart rate is greater when an unpleasant film is presented to the right
hemisphere than when it is shown to the left. At the same time, a more
humorous film elicited greater response when shown to the left hemisphere. Despite this left hemispheric effect, the authors interpret their
findings as indicating that the right hemisphere bears . . . responsibility
for emotion viewed in the broad sense (p. 259).
Using lateralized tachistoscopic presentation, Suberi and McKeever
(1977) have also reported an emotion effect involving the right hemisphere. Subjects were asked to match unilaterally presented faces to
previously memorized target faces as rapidly as possible. Reaction times
favored the left visual field more strongly for subjects who had memorized
emotional faces than for those who had memorized neutral faces. While
this study does not show an LVF superiority for the identification of
emotional expression, it does indicate that faces in general are better
identified in the LVF when affect is present in the memory set.
In contrast to the results of Suberi and McKeever (1977), a pilot study
by the present authors found evidence for an LVF superiority in the
recognition of emotional expression. Cartoon line drawings of three
human faces (a man, a woman, and a child), each depicting three different
emotional expressions (very happy, neutral, and very sad) were used as
stimuli. Subjects were asked to judge whether a unilaterally presented
face had the same or different emotional expression as a subsequent
centrally presented face. A clearly significant LVF superiority was obtained. Although one might argue that this is simply another manifestation
of the right-hemispheric superiority for face recognition (e.g., Rizzolatti,
Umilta, & Berlucchi, 1971; Patterson & Bradshaw, 1975), the argument
for a specific involvement of emotion is enhanced by the fact that the LVF
effect was much larger for the emotional stimuli than for the neutral ones.
In general, then, the studies of emotion recognition in normal subjects
suggest that there is a greater right-hemispheric involvement in such
tasks. The present study is designed to replicate this earlier work, to
extend it to a wider range of emotional stimuli, and to show that the effect
is specific to the judgment of emotions.
METHOD
Subjects The subjects were three male and 17 female undergraduate students who volunteered for the experiment and were paid for their participation. All were right-handed as
ascertained by a handedness questionnaire (Bryden, 1977), and all had normal or corrected
to normal vision. Subjects attended one experimental session which lasted approximately 80
min.
Stimuli and apparatus The stimuli were 25 cartoon line drawings of five adult male faces
(Fig. 1). Isolated discriminable features, such as hair, ears, nose, facial lines, etc. were
similar for each character, as was the overall size, shape, and general gestalt of the head.
Each character was drawn depicting five emotional expressions ranging from positive to

LEY AND BRYDEN

FIG. 1. The cartoon drawings used in the experiment. The columns show Subjects Ml
through MS. The rows represent the emotions from Extremely Positive at the top through
Extremely Negative at the bottom.

EMOTIONS AND FACES

131

negative. There was an extremely positive, mildly positive, neutral, mildly negative, and
extremely negative emotional expression for each face.
A study with five graduate students established the reliability of the discriminability of the
affective expressions. Agreement was 100%when judges were asked to sort the 25 faces into
five groups, representing extremely positive, mildly positive, neutral, mildly negative, and
extremely negative emotional expressions.
The original cartoons were reduced in size, copied, and affixed to white 17.7 x 25.5 cm
cards for tachistoscopic presentation. The test stimuli were approximately 7 cm in height
and 4.5 cm in width. Three sets of the 25 stimuli (5 characters x 5 emotions) were used. Two
of these sets were used for test material and one set for comparison items. For test stimuli,
the faces were located in two positions, either to the left or right of fixation. The stimuli were
positioned so that the center of the face was along the horizontal meridian and 4.5 cm to the
left or right of fixation. Thus, each of the faces appeared once in each visual field, making a
total of 50 stimulus cards. The comparison stimulus drawings were centrally located. A
black dot located in the center of the preexposure field served as a fixation target.
Stimuli were binocularly presented in a three-field Scientific Prototype tachistoscope
(Model GB). The viewing distance was 123.75 cm. When exposed the stimulus appeared
centered I to one side of the fixation point. The faces subtended at an angle at 2.
Procedure After being introduced to the experimental situation subjects were read instructions describing the experimental procedure. Subjects were instructed to look through the
viewer and to fixate the central point when the experimenter gave a verbal ready signal.
Throughout the experimental period subjects were reminded to fixate the central point when
given this ready signal. Following the ready signal a face was flashed on the screen (85
msec) to the right or left of the fixation point. The target face was immediately followed by
another face, the comparison stimulus, which the subject had longer to look at (I set). Pilot
work had established that the exposure durations were sufficient to guarantee judgment
accuracy to be significantly above chance. Subjects were asked to judge whether the
emotional expressions of the two faces were the same or different. For example, if both
faces were extremely positive, mildly positive, neutral, mildly negative, or extremely negative, these expressions were to be judged the same. Any other pairings were to be judged
as different. For example, a mildly negative and an extremely negative emotional expression were to be judged as different. Subjects were also asked to judge whether the
characters were the same or different, that is, whether the same man was pictured twice,
despite differences in emotional expression. The order of reporting emotion and characters
was counterbalanced across subjects.
As a preliminary the subject was shown the five characters (with neutral expressions) and
then shown the range of emotional expressions (the five emotional expressions for one
character). Ten practice trials, using geometric designs rather than faces, were administered.
The experiment consisted of three blocks of trials of 70, 80, and 50 trials, with two
intervening rest periods. Each of the 25 faces (5 characters x 5 emotions) was paired with
four other faces including itself, so that the following response possibilities existed for each
face: (I) same character, same emotion; (2)same character, different emotion; (3) different
character, same emotion; and (4) different character, different emotion. Trial order was set
so that for every 100 trials each of the 25 faces appeared as a target four times, twice in the
LVF and twice in the RVF. In addition, the emotional expressions were the same on 50% of
the trials and different on 50% of the trials. Similarly, the characters were the same on 50%
of the trials and different on 50% of the trials. Presentations were randomized and counterbalanced across the sets of 100 trials.
The experimenter manually recorded the subjects responses. Following the experimental
trials, subjects filled out the handedness questionnaire and were debriefed.

132

LEY AND BRYDEN

RESULTS AND DISCUSSION

As an initial step, separate analyses were carried out on the errors made
in emotion judgments and in character judgments. An emotion error was
defined as a mistake made in judging the similarity or difference of the two
emotional expressions. A character error was a mistake in judging the
similarity or difference of the two characters, in other words, whether the
two presentations were of the same person or not.
In both analyses, order of report (character or emotion judgments first)
was a between-Ss factor, while visual field, character, and emotional
expression were within-Ss variables. In both analyses, the main effect and
primary interactions involving order were not significant. It seems to
make little difference whether subjects made judgments on emotional
expression or character first.
Emotion Recognition

Analysis

The anova for emotion errors produced a highly significant main effect
for Visual Fields (F( 1,18) = 25.779, p < .OOl) (see Table 1). This attests to
a greater LVF accuracy in judging emotional expressions. This LVF
superiority for emotion recognition is consistent with findings in previous
studies.
The most important result, however, is the significant Visual Field X
Emotion interaction (F(4,72) = 2.519, p < .05) (see Table 1). Visual field
differences were greatest for the extreme emotional expressions. The
LVF superiority was highly significant for the extremely positive (p <
.OOl) and extremely negative (p < .OOOl)emotional expressions but not
statistically significant for the mildly positive, neutral, and mildly negative
emotional expressions. Although the main effect for Characters was significant (F(4,72) = 5.178, p < .005) there was a remarkable consistency to
the errors. The mean number of errors was identical for four out of the
five characters. The lower error rate for the fifth character accounted for
the significant main effect.
Different emotions had different error rates. The main effect of Emotions was highly significant (F(4,72) = 22,389, p < .OOl). Fewest errors
were made on the extreme emotions, both positive and negative.
The analysis of the emotion errors also revealed a significant Character
x Emotion interaction (F(l6,28) = 2.508, p < .005), indicating that the
five emotions were not equally discriminable over the five characters.
There was also an Order x Visual Field x Character x Emotion interaction (F( 16,288) = 1.992,p < .025). The interpretation of this latter interaction is not immediately evident.
Character Recognition

Analysis

Again, the main effect of Visual Fields was significant (see Table 2)
(F( 1,18) = 11.581,p < .005) attesting to an LVF superiority in recognizing

46
20

37

MN
EN

a EP, extremely positive;


b Decimals omitted.

34*
35
49

MI

EP
MP
N

Emotional
expressions

MP, mildly

30

positive;

36

39
26

38
32
46

14
41
43

46
8

M3

M2

N, neutral:

37

46
30

16
39
53

M4

Left visual field


Characters

MN,

38

44
20

33
43
49

M5

mildly

36

44
21

27
38
48

negative:

45

55
40

39
39
51

Ml

EN,

extremely

36

50
34

24
30
43

M2

negative

45

49
30

50
46
51

M3

45

48
44

32
40
63

M4

Right visual field


Characters

TABLE 1
PERCENTAGE ERRORS ON EMOTION JUDGMENTS FOR DIFFERENT FACES

44

39
39

43
49
53

MS

43

48
37

37
41
52

4
16

10
3
4

L-R
difference

5
u

el
5
z
Lc

260
31
20
29
-2
26

27

EP
MP
N
MN
25
EN

17

20

24

14
14
24
23

M3

23
21
25
29

M2

12

5
15
13
19

M4

Left visual field


Characters

33

35

31
36
25
35

M5

22

19

20
24
21
21

26

33

35
35
34

26
31
31

31
26
31
31

25
34

M3

20
19

M2

30
35

Ml

29

35
24
39

18
8
18
12

29
32

M5

10
8

M4

Right visual field


Characters

EP, extremely positive: MP, mildly positive: N, neutral; MN, mildly negative; EN, extremely negative.
b Decimals omitted.

Ml

Emotional
expressions

TABLE 2

PERCENTAGE ERRORS ON CHARACTER JUDGMENTS FOR DIFFERENT FACES

27

E
;s
z

8
-2
29
25
31

12

zi
0

3
2

L-R
difference
23
26

EMOTIONSANDFACES

135

characters. This finding is consistent with previous studies demonstrating


an LVF superiority for face recognition.
This analysis also shows a significant Visual Field x Emotion interaction (F(4,72) = 5.748,~ < ,001). This effect is not as stable as it was in the
Emotion judgments (see Table 2). However, there is a consistent LVF
superiority for emotional expressions, excepting the mildly negative expression. For this expression there is a slight, but nonsignificant RVF
superiority. Once again the largest visual field difference occurred with
the extremely negative emotional expression. This difference was highly
significant (p < .OOOl).There was also a significant LVF superiority for
the neutral emotional expression @ < .005). Unlike the emotion recognition analysis, however, the pattern is not systematically related to the
degree of emotion.
The analysis also demonstrated that the characters were differentially
recognizable, as the main effect for Characters (F(4,72) = 22.77,~ < .OOl)
was significant. Certain faces were more readily identifiable in certain
visual fields, accounting for the significant Visual Field x Character
interaction (F(4,72) = 5.108, p < .005). Although a significant Order x
Visual Field x Emotion interaction exists (F(4,72 = 2.697, p < .05), an
examination of it yields no fruitful information.
The analyses of the emotion and character errors in this experiment
have extended and given strong support to the argument for a righthemispheric involvement in emotion recognition. A significant LVF
superiority for recognition of emotions was obtained. Furthermore, this
LVF superiority was most pronounced for extreme emotional expressions, especially negative. The results would also suggest that the
LVF superiority for face recognition cannot account for all of the emotion
recognition effect, since different patterns of lateralization as a function of
emotional expression were obtained with the two types of judgment.
In order to assess the independence of character and emotion effects
further, laterality coefficients were derived for each subject for emotion
errors and for character errors. This statistic expresses the observed
degree of laterality as a percentage of the maximum possible difference
that could have occurred given that particular level of accuracy. It was
proposed by Halwes (1969) and is recommended by Bryden (1978).
A Pearson correlation coefficient was then calculated between the
laterality coefficients for the character judgments and those for the emotion judgments. This correlation was .28 (p > .lO). Although this result
lends further credence to the hypothesized independence of these effects,
the small sample size could possibly be argued as a mitigating factor.
To demonstrate the independence of these effects further, two analyses
of covariance were carried out. When the influence of the character error
variance was removed (i.e., character errors were the covariate, emotion
errors were the dependent variable) all of the significant main effects and

136

LEY AND BRYDEN

interactions of the emotion error analysis were maintained. The main


effects of Visual Field (p < .005), Characters (p <.OOl), and Emotions (p
< .OOOl)were all significant. The interactions of Visual Field x Emotion
(p < .04), Character x Emotion (JJ< .OOl), and Order x Visual Field x
Character x Emotion (p < .016) also remained significant. Note that the
significance levels of the three interactions and the main effect of Emotions were in fact increased when the effects of the character errors were
controlled.
Thus, the character error variables contributed minimally to variation
in the emotion error means. The LVF superiority for emotions is dissociable from the LVF superiority for face recognition. In short, these seem
to be independent effects.
From the above covariance analysis it would seem that face recognition
does not affect emotional expression recognition. The direction of effect
does not seem to be from faces to emotions. However, it is possible that
character recognition is dependent on processes involved in emotion
recognition.
A second analysis of covariance was performed with the covariate and
dependent variable reversed. Emotion errors were now the covariate and
character errors were the dependent variable. Again all of the significant
main effects and interactions were preserved, excepting one. The significant main effect for Visual Fields, which was previously p < .005,
disappeared (p < .21). The main effect of Characters was significant (p <
.OOOl). The interactions of Visual Field x Character 0, < .OOl), Visual
Field x Emotion @ < .OOOl),and Order x Visual Field x Emotion @ <
.03) were significant.
The loss of the main effect for Visual Fields is somewhat difficult to
explain. There are two possible explantions for this result. First, the
primary analyses suggest that the character effect is smaller than the
emotion effect. Second, the character effect may be partly due to the
emotional expression effect. This possibility has been uncontrolled in
other face recognition experiments and should be investigated further.
DISCUSSION

The present experiment indicates that the right hemisphere is more


intimately involved than the left in recognizing, responding to, and/or
processing emotional stimuli. The analyses further demonstrate that the
LVF effect for emotion is independent of face recognition and depends on
the degree of emotional expression. The assumption that the LVF
superiority is tied in some way to the degree of expressed affect is
bolstered by the fact that this relationship has appeared across sensory
modalities in both dichotic listening (Carmon & Nachson, 1973) and
tachistoscopic paradigms (Suberi & McKeever, 1977).
There are many characteristics of right hemispheric functioning that

EMOTIONS AND FACES

137

perhaps uniquely bias it for processing affective information. Semmes


(1968) has proposed that elementary functions are focally represented in
the left hemisphere and more diffusely represented in the right hemisphere. The diffuse organization of the right hemisphere would result in
proficient integration of dissimilar units and a specialization for multimodal coordination, such as visual spatial abilities. Such an organization
of the right hemisphere would be especially well suited for processing
emotional stimuli. If one believes that recognizing an emotion entails an
integration of stimulus information from a variety of sources (both sensory and motor units) and across space and time, Semmes model would
argue that the synthesis would occur in the right hemisphere. This right
hemispheric superiority for synthesizing diverse units of information is
closely allied to its often described holistic or gestalt processing capacity (Levy, 1972; Ornstein, 1972; Safer & Leventhal, 1977).
Cross-cultural research on emotions has revealed that all facial expressions can be categorized into nine discriminable, universally recognizable patterns (Izard, 1971; Ekman & Friesen, 1971). Again the ability
to recognize such gestalten could be attributed to the right hemispheric
holistic processing capacity.
A third characteristic of emotional processing can also be associated
with right hemispheric functioning. Safer and Leventhal(1977) have made
the point that emotions are often accompanied by, or associated with,
imagery. The right hemisphere is more involved than the left in processing
or generating imagic material (Davidson & Schwartz, 1976). In our
present studies, the stimuli were perhaps too benign to evoke any imagery. However, it is easy to imagine that more lifelike stimuli such as
screams, laughter, or gruesome faces could result in the subjects imaging.
In summary then, synthetic and integrative characteristics, a holistic
and gestalten nature, and imagic associations are three features of processing affective material that would differentially favor right hemispheric
mediation of the task.
REFERENCES
Bryden, M. P. 1977. Measuring handedness with questionnaires. Neuropsychologiu,
15,
617-624.
Bryden, M. P. 1978. Strategy effects in the assessment of hemispheric asymmetry. In G.
Underwood (Ed.), Srrutegies ofinformntion
processing. New York: Academic Press.
Carmon, A., & Nachson, 1. 1973. Ear asymmetry in perception of emotional non-verbal
stimuli. Acta Psychologica, 37, 351-357.
Critchley, M. 1953. The parietal lobes. London: Arnold.
Davidson, R., & Schwartz, G. 1976. Patterns of cerebral lateralization during cardiac
biofeedback versus the self-regulation of emotion: Sex differences. Psychophysiology,
13, 62-74.
Dimond, S. J., & Farrington, L. 1977. Emotional response to films shown to the right or left
hemisphere of the brain measured by heart rate. Acta Psychologica, 41, 255-260.

LEY AND BRYDEN

138

Ekman, P., & Friesen, W. 1971. Constants across cultures in the face and emotion. Journal
of Personality and Social Psychology, 17, 124-129.
Gainotti, G. 1972. Emotional behavior and hemispheric side of the lesion. Cortex, 8(l),
41-55.
Galin, D. 1974. Implications for psychiatry of left and right cerebral specialization. Archives
of General Psychiatry, 31, 572-583.
Goldstein, K. 1939. The organism: A holistic approach to biology derivedfrom pathological
data in man. New York: American Book.
Haggard, M. P., & Parkinson, A. M. 1971. Stimulus task factors as determinants of ear
advantages. Quarterly Journal of Experimental Psychology, 23, 168-177.
Halwes, T. G. 1969. Effects of dichotic fusion on the perception of speech. Supplement to
Status Report on Speech Research, Haskins Laboratories, New Haven, Conn.
Hecaen, H. & Angelergues, R. 1962. Agnosia for faces (Prosopagnosia). Archives ofNeurotogy,

7, 92-100.

Izard, C. E. 1971. The face of emotion. New York: Appleton-Century-Crofts.


Levy, J. 1972. Lateral specialization of the human brain: Behavioral manifestations and
possible evolutionary basis. In John A. Kiger (Ed.), The biology ofbehavior. Corvallis:
Oregon State Univ. Press.
Ornstein, R. 1972. The psychology of consciousness. San Francisco: Freeman.
Patterson, K., & Bradshaw, J. 1975. Differential hemispheric mediation of nonverbal visual
stimuli. Journal of Experimental Psychology: Human Perception and Performance, 1,
246-252.
Rizzolatti, G., Umilta, C., & Berlucchi, G. 1971. Opposite superiorities of the right and left
cerebral hemispheres in discriminative reaction time to physiognomical and alphabetical material. Brain, 94, 431-443.
Robertson, A. D., & Inglis, J. 1977. The effects of electroconvulsive therapy on human
learning and memory. Canadian Psychological Review, 18, 285-307.
Safer, M., & Leventhal, H. 1977. Ear differences in evaluating emotional tones of voice and
verbal content. Journal of Experimental Psychology: Human Perception and Performance, 3(l), 75-82.
Semmes, J. 1968. Hemispheric specialization, a possible clue to mechanism. Neuropsychologia,
6, I l-26.
Sperry, R. W., Gazzaniga, M., & Bogen J. 1969. Interhemispheric relationships. In P.
Vinken (Ed.), Handbook of clinical neurology. Amsterdam: North Holland. Vol. 4.
Suberi, M., & McKeever, W. F. 1977. Differential right hemispheric memory storage of
emotional and non-emotional faces. Neuropsychologia,
15, 757-768.
Terzian, H. 1964. Behavioral and EEG effects of intracarotid sodium amytal injections. Acta
12, 230-240.
Neurochirugia,
Weinstein, E. A., & Kahn, R. L. 1955. Denial of illness: Symbolic and physiological
aspects. Springfield, IL.: Charles C Thomas.

You might also like