Hemispheric processing of emotional valence


Material Information

Hemispheric processing of emotional valence
Physical Description:
ix, 119 leaves : ill. ; 29 cm.
Froming, Karen Bronk, 1957-
Publication Date:


Subjects / Keywords:
Brain -- physiology   ( mesh )
Emotions   ( mesh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1988.
Includes bibliographical references (leaves 111-117).
Statement of Responsibility:
by Karen Bronk Froming.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001544123
notis - AHF7630
oclc - 22333872
System ID:

Full Text








To my sweet William


I would like to offer special warm, thanks to my good

friend and chair, Eileen Fennell. Her mentorship and

unfailing loyalty made my growth possible in graduate

school. I will always value her friendship.

I would like to thank my committee, Jacque, Rus, Hugh

and Ed, rather than just role models, good teachers and

thinkers, you are all friends. My hope is that as a

professional I do justice to your training.

Special thanks are extended to Carol Schramke and Kevin

Greve who helped me find and lug equipment and then showed

me how to use it. It may have seemed like a small thing,

but sharing the "trials" of a tachistoscope made it seem

less formidable. Another thank you goes to the Psychology

Department's personality area for loaning me research space.

The other thing about tachistoscopes is that they are not


To another good friend, Roger Blashfield, I don't

always find the words to let him know how I value sharing my

statistics terrors and just about everything else with him.

I am deeply indebted to him.

Finally, for keeping me to the deadline, I want to

express my appreciation to Beverly Barfield.



ACKNOWLEDGMENTS ......................................

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

ABSTRACT .............................................


1 REVIEW OF LITERATURE ........................ .

Introduction .................. ..............
Emotional Processing in Brain-Injured
Individuals ................................
Emotional Processing in Non-Neurologically
Impaired ...................................
Brain Mechanisms in Emotion ...............
Rationale and Specific Hypotheses .............
Design and Analyses ...........................

2 METHODS .......................................

Subjects ......................................
Stimuli and Apparatus .........................
Mood Induction ................................
Procedure .....................................

3 RESULTS .......................................

Preliminary Analyses ..........................
Primary Analyses .............................

4 DISCUSSION ....................................

Preliminary Analyses ..........................
Primary Analyses ............................ ..
Conclusions ...................................
















A INFORMED CONSENT ............................ 96

B SCREENING QUESTIONNAIRE ....................... 100

C DEBRIEFING FORM .............. ................ 102


E SURGENCY TASK INSTRUCTIONS .................... 108

F SCHEMATIC FOR PROCEDURES ...................... 110

REFERENCES ............. ....... ....................... 111

BIOGRAPHICAL SKETCH ................................. 117


Figures page

1. 2-way Interaction on Hits: Card
Position x Emotion (Holding Card
Position Constant) ........................ 57

2. 2-way Interaction on Hits: Card Position
x Emotion (Holding Emotion Constant) ...... 60

3. 3-way Interaction on Hits: Anxiety x
Hemiface x Emotion ........................ 62

4. 3-way Interaction on Hits: Anxiety x
Hemiface x Emotion (Holding Hemiface,
Emotion Constant) ......................... 63

5. 2-way Interaction for Reaction Time:
Emotion x Hemiface ........................ 68

6. 4-way Interaction for Reaction Time: Sex
x Anxiety x Card Position x Hemiface ...... 69

7. 3-way Interaction for Reaction Time: Sex
x Card Position x Hemiface (Collapsed
Across Anxiety Levels) ................... 70

8. 3-way Interaction on Hits: Sex x
Congruence x Card Position (Collapsed
Across Anxiety Levels and Holding Card
Position Constant) ........................ 74

Abstract of Dissertation Presented to the
Graduate School of the University of Florida
in Partial Fulfillment of the Requirements
for the Degree of Doctor of Philosophy



Karen Bronk Froming

August 1988

Chairman: Eileen B. Fennell, Ph.D.
Major Department: Clinical and Health Psychology

Chimeric faces with either incongruent or congruent

emotional facial expressions were tachistoscopically

presented at each of 3 card positions: at central fixation

such that the right and left halves of the face were

projected to the contralateral hemisphere; the entire

stimulus presented within the left visual-field (projected

to the right hemisphere); the entire stimulus presented in

the right visual-field (projected to the left hemisphere).

Male and female subjects underwent an anxiety mood induction

to determine both the singular effects of sex and anxiety on

hemispheric processing of emotions and the interactive

effects of these variables on processing. It was predicted

that both accuracy and reaction times would be better for

emotional material presented to the right hemisphere and

that anxiety would effect the performance of males and


females differently, but that overall, anxiety would serve

to activate the left hemisphere and induce a processing

load. In turn, the processing load would enhance right

hemisphere processing, thereby improving performance to

emotional stimuli projected there.

Results indicated that both accuracy and speed of

processing differ as a function of the emotional content and

as a function of the side of the face. An unexpected result

of the current study was the clear, robust effect for

preferential processing of the left hemiface over and above

a left visual-field effect. Happy facial expressions are

routinely identified quickly and accurately, particularly

for the left hemiface and regardless of the imposition of an

anxious state. However, this was not the trend for negative

or ambiguous stimuli which were affected by the presence of

an anxious mood. The hemifacial effect was influenced to a

slight degree by anxiety. Anxious mood appeared to

simultaneously alter processing of both hemifaces. The

performance for the right hemiface was reduced at the same

time as performance for the left hemiface was increased.

Anxious females seemed to exhibit greater variability in

performance; specifically, when stimuli crossed midline.

To summarize, it appears that the right hemisphere is

superior in processing emotional stimuli, but that this

effect is qualified by an attentional component which

results in preferential processing of the left hemispace.


This is particularly apparent when the subject experiences

an anxious mood. The mechanisms bearing on these results

are unclear but may be related to the asymmetric neural

systems responsible for arousal and activation as they

interact with hemispheric processing.



The study of hemispheric specialization began with

Broca's observations of his patient's language loss

following a left anterior hemorrhage. Hughlings Jackson

(cited in Ross & Mesulam, 1979) also made the observation

that while the left hemisphere may be responsible for the

verbal-linguistic aspects of language, the analogous right

hemisphere lesion resulted in a loss of the affective

components of language. Since that time, research on the

specialization of the cerebral hemispheres has indicated

that each hemisphere is differentially specialized to

process verbal or spatial material and furthermore, that

each hemisphere's information processing style is also

different (Levy, 1969; Bogen, 1977).

In addition to the left hemisphere's analytic,

sequential processing bias and the right hemisphere's

global, serial processing style, the literature on

hemispheric specialization has implicated unique

contributions of the hemispheres to the experience and

expression of emotion (Heilman, Watson, & Bowers, 1983).

Early clinical observations with neurologic patients, those


suffering from epilepsy or mental illness and those

undergoing the WADA or sodium amytal procedure, where each

hemisphere is serially anesthetized, have yielded consistent

but paradoxical results on the contributions of the left and

right hemispheres. Lesions or amytal injection of the left

hemisphere resulted in the classic depressive,

"catastrophic" reaction, while lesions or injection of the

right hemisphere were characterized by inappropriate

euphoria, jocularity, or at its extreme, denial of actual

illness (Tucker, 1981).

However, the inferences drawn from studies of

epileptics, those suffering from chronic mental illness and

normal controls, become more complex when one considers the

traditional notion that the cortex exerts a reciprocal

inhibitory effect on the contralateral hemisphere and an

inhibitory effect upon the ipsilateral subcortical

structures. The inferences which may be drawn from the

clinical and experimental data are no longer as clear. It

is evident from patient studies that 1) cortical-

subcortical discharge is relevant to emotion; and 2)

hemispheric laterality of lesion is implicated in the

experience and expression of emotion (Tucker & Williamson,

1984). It is less clear whether each hemisphere contributes

its processing bias to produce hemispheric specialization in

the valences of emotion.


Silberman and Weingartner (1986) review the literature

and emphasize that the conclusions drawn in the area thus

far, collapse data across populations (e.g. normal subjects,

neurologic patients, and psychiatric populations) when

explaining the functioning of the brain in emotional

processes. Furthermore, data gathered within these

populations are derived from a plethora of experimental

paradigms, few of which have been replicated.

The process of inference making in neuropsychology, in

general, and specifically in the area of emotion, makes

review and conclusions difficult. The first line of

analysis is clinical observation of brain-impaired

individuals. From these data, the process of hypothesis

generation occurs and experiments may be designed utilizing

normal individuals. Comparisons between the 2 groups are

difficult, if not inappropriate, because it is well known

that size, location, and type of lesion change the

behavioral manifestations of the lesion and may lead to

inappropriate tests within the normal population. While the

dialectical process between observations and experimentation

is useful and necessary, when it comes to the study of

emotion, many conclusions are drawn from the parallel data

derived from diverse populations (i.e. neurologically

impaired patients, psychiatric patients, and normal



The methodological shortcomings inherent in

neuropsychology are no less of an issue in the study of

emotion. Three domains are the subject of study: the brain-

injured, the mentally ill, and normals. In addition, it has

recently been clear that semantics have clouded the area of

study. There has yet to be a precise definition of emotion

as it differs from motivation or affect. While this may

seem a trivial issue, it is not. Furthermore, it is also

apparent that within the area of emotion which includes its

subjective, objective, and communicative aspects, it is

important to define whether one is studying the cognitive,

neurovegetative, or subjective emotional aspects. In all 3

domains of study, the results and their implications are

dependent upon precise definition.

Emotional Processing in Brain-Injured Individuals

Early in the history of brain research, observations

have been made about changed emotional behavior following

injury. Jackson (cited in Heilman, Watson, & Bowers, 1983)

noted an absence of emotional inflection or prosody

following right hemisphere lesions and was the first to

suggest that this was an analogous communication deficit to

the left hemisphere speech disruptions. Babinski (cited in

Bear, 1983) noted a different response, or rather, lack of

response in some right-sided lesions. The indifference

reaction was an often profound denial of illness or

inappropriate happiness during a time of significant


illness. Later observations were made about left hemisphere

lesions, that of depression and the "catastrophic" reaction.

Hypotheses began being made regarding the role the cerebral

hemispheres played in emotional processes.

The differential emotional reactions that arose from

cerebral injury were systematically characterized by

Gainotti (1972) who retrospectively studied 160 patient

admissions. The majority of patients with lesions within

the left hemisphere exhibited a spectrum of depressive

behavior, while those with right hemisphere lesions showed a

variety of reactions from euphoria or inappropriate laughing

to minimization of symptoms. Similar findings have been

borne out in examinations of epileptics undergoing the

sodium amytal procedure or WADA. The test is undertaken to

experimentally examine the possible outcome of surgical

removal of irritative tissue, the presumption being that

anesthetization of tissue mimics the absence of tissue. The

differing emotional reactions are the same during recovery

from the amytal injection; the left carotid injection

produces the "catastrophic" response, and the right carotid

injection is associated with denial or indifference

(Terzian, 1964; Rossi & Rosadini, 1967). However, these

findings are only suggestive because the emotional reactions

do not occur in every patient. One study has failed to

replicate this finding probably as a result of significantly


larger doses of sodium amytal which would cross the callosum

and anesthetize larger portions of the brain (Milner, 1974).

More recent evidence regarding the emotional changes

that occur during seizures and following a lengthy history

of a seizure disorder also implicate attendant emotional

changes that are dependent upon the site and side of lesion

(Blumer & Benson, 1982; Bear & Fedio, 1977). The emotional

changes are most evident with partial-complex seizures

originating in the temporal lobe and spreading into the

limbic system. Continual seizure discharge is hypothesized

to result in hyperconnection between the temporal lobes and

limbic system. The behaviors observed are opposite to the

Kluver-Bucy syndrome. The consistently agreed upon

constellation of symptoms includes decreased sexuality and

arousal, an overvaluation of ideas or beliefs, and finally,

a sense of stuckness or inflexibility (Blumer & Benson,


Because of the longstanding history of temporal lobe

epileptics being considered psychiatric patients, Bear and

Fedio (1977) undertook a retrospective analysis of the

interictal psychiatric manifestations of the disorder that

were reported in the literature. They developed an

inventory of 18 items to which the patient and the next of

kin responded. A significant finding between these groups

arose dependent upon the side of the patients' seizure

focus. The patients with left-sided foci tended toward more


realistic views about their behavior and congruent with

their relatives' perceptions, whereas those individuals with

right-sided foci tended to be ignorant of many of their

symptoms, particularly negative behaviors, and those

behaviors they were aware of were minimized when compared to

their relatives' observations.

In general, Bear and Fedio's (1977) behavioral

checklist consisted of items in 3 broad categories:

sexuality, viscosity/"stickiness" (or verbal/ideational

circumstantiality), and deepening of emotionality. The

importance of such an analysis of behavior in complex-

partial seizure patients is the discharge of these foci into

the limbic system, an area hypothesized to have strong

contributions to emotional behavior.

Another study reviewed the records of epileptics with

ictal outbursts of laughing or crying. There were

significantly more left-sided foci in cases of pathological

crying, while right-sided foci were associated with

pathological outbursts of laughing. The second portion of

the analysis involved the analysis of mood following

hemispherectomy. Consistent with the previous findings,

right hemispherectomy resulted in more instances of euphoric

mood (Sackeim, Greenberg, Weiman, Gur, Hungerbuhler, &

Geschwind, 1982).

It is also possible for patients with epilepsy to

exhibit psychotic behavior. It has long been a subject of


controversy whether the psychosis is a co-occurring

phenomenon or whether it is a by-product of the epilepsy

(Trimble, 1982). This is to some extent a moot point.

Neuropsychologists view behavior, whether psychiatric

disorders or seizure disorders, as brain-based. Flor-Henry

(1969) investigated inter-ictal psychosis and uncovered

largely schizophreniform psychosis accompanying left

hemisphere seizure foci. Cyclothymic, mood dysregulation

disorders were much more common with right hemisphere

irritative lesions. Upon closer examination of the symptoms

exhibited, one finds ideational or cognitive disturbances

such as paranoia, overvaluation of ideas, and hypergraphia

indicative of left-sided complex-partial seizures. Mood

regulation appears to be more commonly disturbed in those

people who suffer from right-sided epilepsy. The clear cut

differentiation between left- and right-sided temporal lobe

involvement and its contribution to different psychotic

disturbances has been unconfirmed by subsequent study and

may reflect classification differences between European and

American nomenclature (Trimble, 1982). However, the

observations recorded in temporal lobe, partial-complex

seizure patients point up the need to consider

cognitive/ideational factors, as well as mood and affect,

when examining the area of emotion (Silberman & Weingartner,



More recent studies have examined the expression and

understanding of emotions in brain-injured individuals.

Comparison of the effects of left hemisphere damage with

right hemisphere damage has consistently reflected poorer

performance of right hemisphere impaired patients on tasks

of vocal emotional expressiveness prosodyy), facial emotion

identification, emotional mimicry, and recognition of

emotional prosody. This line of evidence also shows

diminished arousal, both behaviorally and

psychophysiologically in right hemisphere lesion patients.

The hypoarousal is present even when attempts at stimulation

using emotion arousing stimuli are used (Morrow, Vrtunski,

Kim, & Boller, 1981).

When Hughlings Jackson observed the tonal flatness of

patients who were suffering from right hemisphere disease,

he hypothesized that the right hemisphere analog for speech

involved the affective components of language. More

recently, Ross and Mesulam (1979) have argued for the same

idea. It is intuitively appealing to suppose that the

attentional dominance responsible for orienting to novelty

might also be linked with the receptive-communicative

aspects of survival laden messages. Certainly such messages

preceded speech, are evolutionarily older, but are

nevertheless salient parts of the human capacity to

communicate. It is not surprising then that the literature

is replete with anecdotal evidence of communicative deficits


between the right brain damaged patient and spouses/family

members. Spouses frequently describe the patient as

changed; there are increased marital difficulties despite

the absence of discernible sequelae.

When comparing right vs. left brain-injured groups on

their ability to discern affective intonations within a

semantically neutral sentence, patients with right

hemisphere disease reliably chose the content over the

prosodic message despite being told to choose a visual

representation depicting the prosody (e.g. "the boy went to

the store," said in a sad voice; Heilman, Scholes, & Watson,

1975). However, the dysfunction underlying this impairment

may be a perceptual deficit or cognitive deficit, or both.

To eliminate the cognitive decision-making portion of the

task, a "same-different" perceptual discrimination task was

used in which 2 sentences were read with the same or

different affective intonations. The affective tone did not

have to be classified. The group with right hemisphere

disease still performed more poorly than either the left

hemisphere damaged group or controls (Tucker, Watson, &

Heilman, 1977). Heilman, Bowers, Speedie, and Coslett

(1983) then tried to distinguish between emotional and non-

emotional prosody by using the same groups, right and left

hemisphere impaired patients and controls. Each subject had

to listen to an audio tape of effectively intoned sentences

and to listen to environmental sounds. Those subjects who


could not perform affective discrimination could flawlessly

identify environmental sounds, such as a phone ringing,

indicating the two tasks are dissociable.

In an attempt to investigate visual affective

processing, Gardner, Ling, Flam, and Silverman (1975) had

patients match captions to cartoons with varying emotional

contents. Again, the right hemisphere patients performed

below the level of the left lesioned individuals. In fact,

these subjects tended to provide captions which relied on

their intact left hemisphere language skill and were

descriptive of the picture rather than capturing the humor

conveyed in the cartoon.

Finally, even though the right hemisphere appears to be

implicated in the expression and understanding of emotional

material, it is not clear if these deficits extend to the

subjective experience of emotion. There are clear instances

of subjective experiences or feeling states when outward

appearances are incongruent. For example, in patients with

pseudobulbar affect, the dramatic crying and grief reaction

is in opposition to the patient's report of a calm or

pleasant internal state. Likewise, those individuals who

are akinetic appear flattened but experience wide-ranging

emotion. In order to investigate the mood levels of brain

damaged individuals, investigators administered the

Minnesota Multiphasic Personality Inventory (MMPI) to

unilaterally brain injured adults who were equated for


severity of cognitive and motor deficits. Patients with

left hemisphere disease (without frank aphasic symptoms) had

significantly higher elevations on scales assessing

depressive symptoms, whereas right hemisphere patients did

not (Gasparrini, Satz, Heilman, & Coolidge, 1978). In the

analysis by Bear et al. (1977) of inter-ictal behavior of

temporal lobe epileptics with unilateral foci, self-report

of symptoms was similar in direction to the findings of

Gasparrini et al. with left-sided lesions. Those epileptics

with left hemisphere foci experienced more depressive

symptomatology and ideational difficulties, while those

people with right hemisphere disease minimized their

symptoms relative to family member assessment.

It has been a troublesome question since the initial

hypothesis of greater right hemisphere involvement in

emotion, whether emotion represents yet another function

that is best mediated by holistic/serial processing. The

spatial configuration of facial recognition and the tonal

quality of emotional vocal expression may simply be complex

pattern discrimination subserved by the right hemisphere.

Emotion, as a unique phenomenon, not just a complex

configuration of patterns, must be shown to be dissociable

from these latter elements.

Bowers, Bauer, Coslett, and Heilman (1985) attempted to

delineate the separate but relative contribution of

emotional processing to emotional facial processing apart


from the visuoperceptual skills involved. To do this they

administered a variety of facial discrimination tasks (i.e.

discrimination, pointing, and naming). They successively

administered these tasks to 3 groups of patients: left/right

hemisphere damaged patients and a group of non-

neurologically impaired patients. The examiners presented 7

subtests which varied along task demand dimensions requiring

the subject to use 1 of 2 strategies for identification.

The authors argue that making facial identity

discrimination of "same or different" involves a purely

perceptual template matching strategy. If one adds an

emotion to the discrimination, the subject could determine

the emotion was the "same or different" based on the spatial

properties of the identical faces. When the faces and

emotions are different the task becomes more complex, one

using associative strategies.

The tasks administered were as follows: Same or

different face-same emotion (simple facial discrimination

task), same or different face-different emotion (facial

discrimination across emotions), same or different emotion

discrimination-same face (emotion discrimination task), same

or different emotion-different face, name the emotion, point

to the named emotion-same face, and point to the named

emotion-different face. The results indicated the right

hemisphere impaired group performed badly relative to the

non-neurologic and left hemisphere damaged group, even on


emotion naming tasks. When facial discrimination was

covaried to partial out the effects of perceptual

discrimination, the right hemisphere damaged group was still

significantly impaired, leading the investigators to

conclude that the right hemisphere contributes to emotional

discrimination over and above the visuoperceptual

discrimination necessary for facial recognition. Whether

emotion, per se, is a distinctive ability, or whether

emotion adds to increased configural complexity was left

unanswered by this study; however, the 2 skills are clearly


Emotional Processing in Non-Neurologically Impaired

Three hypotheses for investigation have evolved from

the observations of brain-impaired individuals; first, the

right hemisphere appears to be superior in the processing of

emotional stimuli; the right hemisphere is primarily

involved in the regulation of mood and affect; and finally,

there is a reciprocal process in which the right and left

hemispheres exert their influence to the experience and

expression of different emotions. The study of normal

individuals has followed from the initial observations of

the brain-injured, in the areas of expression,

comprehension, and perception of emotions.

Several recent investigators have undertaken analyses

of facial expression. Using photographs of models posing

for different emotions, Sackeim, Gur, and Saucy (1978)


examined the composites made up of the right and left halves

of faces. The original picture was cut at its midline and

each half mirror-reversed so that the composite was

bilaterally symmetric. The left side of the face was judged

to exhibit more intense emotion. Several studies have

replicated this finding across different handedness groups.

Greater intensity of emotional expression in the left side

of the face appears to override the traditional hemispheric

asymmetries involved in motor control and may represent a

special function of the right hemisphere regardless of

handedness. Borod and Caron (1980) then established the

strength of this effect by demonstrating that the left side

of the face was more emotionally expressive in spontaneous

facial displays and not only in posed expressions. This

effect was true for right and left handers.

Tachistocopic paradigms are typically used to

understand the perceptual processes. The presentation of

stimuli may be lateralized with the dependent variables of

reaction time or accuracy being used to address the relative

left vs. right superiority, or a target is presented at

central fixation with a comparison figure presented

unilaterally. Again, the dependent measure is speed of

response or accuracy (Springer & Deutsch, 1981).

In one study, subjects were asked to view cartoon faces

expressing a range of emotions (very positive, mildly

positive, neutral, mildly negative to strongly negative). A


target face was presented laterally and then a comparison

face was presented at central fixation. Each subject had to

make a character judgment (same or different cartoon

character) and emotion judgment (same or different). The

left visual-field by emotion intensity effect was found;

however, the results left unanswered whether the effect was

simply another manifestation of a general right hemisphere

superiority for facial patterns. When differences in

character identification were covaried, there still remained

a significant difference between the fields for emotion

discrimination (Ley & Bryden, 1979). Safer (1981) utilized

a very similar procedure except that the target emotional

face was presented at central fixation (6 emotions

presented: anger, surprise, happiness, fear, sadness, and

disgust), and choices were randomly lateralized. Half the

subjects were asked if the comparison slide was the same or

different. Some subjects were asked to empathize with the

target face to make their choice, while others were to

silently label the target emotion. Overall findings were

similar to other studies, in that a greater level of

accuracy occurred for slides presented in the left visual-

field. However, when broken down by groups, the effect was

present only for those who empathized with the target face.

Subjects who labelled the emotion showed no difference

between the visual fields. Presumably the verbal task of


denoting an emotion eliminated the relative left visual-

field advantage.

Another study which examines the relative effect of an

encoding strategy on visual field effects was that of Suberi

and McKeever (1977). They asked their subjects to memorize

2 faces, either emotional faces or nonemotional faces. They

unilaterally presented these target faces amongst

distractors. Subjects who had memorized emotional targets

had a larger left-field effect, as measured by faster

reaction times, than did their counterparts who were

choosing nonemotional targets from distractors.

Superiority of the right hemisphere for specific

emotions has also been examined with variable results. It

does appear that effects are greater for extremes of

emotions (strongly happy or sad), rather than milder

expressions of emotion. Perhaps there is greater adaptive

significance to extremes, hence more orienting significance.

Reuter-Lorenz and Davidson (1981) utilized these extremes,

strongly happy, strongly sad, and neutral, then presented an

affective and neutral face to the right and left visual-

field, respectively. Subjects were told to choose the

affective face. Rather than finding an overall effect for

left visual-field and emotion, they found reaction times

were faster to the happy faces in the right visual-field,

and conversely, response times were faster to sad faces in

the left visual-field. These investigators replicated their


findings and demonstrated them, not only in right-handers,

but in inverted left-handers as well. While there are a few

studies yielding emotional valence by visual field

interactions, the findings are controversial (Bryden & Ley,


A final, quite extensive study utilizing the

tachistoscope to further delineate the role of the right

hemisphere in processing emotional valence is that of

Natale, Gur & Gur (1983). They conducted a 3 part study

systematically reviewing the effects of sex, handedness

(reflected by handwriting posture), and sensitivity/bias of

each hemisphere in processing a range of emotion. The first

portion of the study assessed a range of 6 emotions and the

relative bias each hemisphere had in processing that

emotion. The emotions were happiness, sadness, anger, fear,

surprise, and disgust. Subjects were instructed to identify

the relative positive or negative nature of the emotion on a

7 point Likert scale. Overall, negative emotional stimuli

presented in the left visual-field were rated more

negatively, but subjects rated the positive emotions more

accurately. This experiment suggested a negative processing

bias, although there were more negative emotions present to

choose from.

The next section of the study was designed to further

evaluate the sensitivity (accuracy) and bias present in the

hemispheric asymmetry. They used only the emotions judged


as most extreme, happiness and sadness. The pictures were

split at the midline and presented unilaterally each of 3

ways, as a whole face or as a composite of the 2 emotions,

Happy-Sad/Sad-Happy. Subjects were instructed to identify

faces as positive or negative. There was a significant

effect for sex, with females giving lower ratings for the

emotions than males and a significant effect for stimulus

type. The whole face emotional stimuli yielded ratings in

their respective directions (sad was rated negatively and

happy was rated positively), but interesting findings

resulted on the mixed emotional stimuli. These faces were

rated intermediately by the males and more negatively by

females. The investigators computed indices for

sensitivity, showing stimuli presented in the left visual-

field are judged more accurately and that the male subjects'

performance reflected a bias in judging affect more

positively. The varying exposure durations chosen for each

task further suggested that emotion judgments, either

positive or negative, occur at preliminary stages of

processing. However, emotion identification (happy or sad)

may occur at later stages of processing requiring longer

exposure durations. The exposure duration was then

increased and subjects were asked to identify happy vs. sad

when the composites were presented unilaterally. Results

indicated that when competing emotions are presented in the

right visual-field they are identified more frequently as


positive, whereas when stimuli are presented in the left

visual-field they are more accurately judged as either

positive or negative.

In conclusion, Natale, Gur, and Gur (1983) present data

consistent with a right hemisphere superiority in processing

emotion, but there exists a positive processing bias for

emotions presented to the left hemisphere. This bias is

more apparent for males; however, the female subjects rated

negative affect more negatively than males which may have

obfuscated the bias effect for this group. These

investigators go on to note the sex differences present in

the incidence of some affective disorders and the presence

of crying outbursts in women, while men have shown

inappropriate laughing or jocularity subsequent to cortical

lesions. Furthermore, they speculate as to the adaptive

significance of a positive or optimistic processing bias in

the presence of cross-cultural evidence of greater negative

than positive facial representations of affect, particularly

when there are many other indicators of the primacy of

negative affect.

Another experimental paradigm utilized in research on

hemispheric specialization is the dichotic listening task in

which competing auditory information is presented

simultaneously in both ears. Safer and Leventhal (1977) in

a variation on the paradigm presented short sentences that

varied affective content along the dimensions of positive,


negative or neutral tonal quality. The subjects listened to

each sentence monaurally and indicated whether they thought

the passage was positive, negative, or neutral. Those

subjects who heard the passages in their left ears judged

the passages by the tonal qualities while those subjects who

listened to the passages in their right ear made judgments

based on the propositional message conveyed. However, since

the sentences were heard monaurally, it is difficult to

interpret these results as consistent with previous dichotic

paradigms in which contralateral pathways predominate under

conditions of binaural competition. Thus, conclusions

regarding the hemispheric superiority for processing tonal

properties vs. propositional messages are more speculative.

Carmon and Nachson (1973) presented the sounds of

crying, shrieking, and laughing in each of 3 vocal

qualities, that of an adult male, adult female, and that of

a child using the dichotic listening paradigm. The vocal

sounds were paired for dichotic presentation and the subject

chose, from an array of cartoon representations, the face

that best represented the sounds they heard. In a very high

percentage of their subjects, accuracy was greater for

material presented in the left ear. There was a trend for

greatest accuracy with the sound of crying.

In an attempt to provide convergent evidence for the

right hemisphere's role in emotional processing, the

Waterloo group used dichotic listening tasks with musical


passages and effectively intoned sentences. In the first

study, the investigators composed 7 note, tonal passages in

the major and minor keys. Music written in a major key is

more often rated as effectively positive, whereas that

written in a minor key is often felt to be effectively

negative. Dichotic pairs were presented in 3 categories,

with valences the same, one in which the differences between

the competing pairs was moderate (e.g. neutral-happy), and

one in which the difference between the pairs was extreme

(e.g. happy-sad). Correct identification of the stimuli

occurred more frequently for material heard in the left ear.

Furthermore, the effect increased proportionally for the

left ear the more discrepant the competing stimuli were.

There was no effect for different affective stimuli (Bryden,

Ley, & Sugarman, 1982; cited in Bryden & Ley, 1983).

Turning to verbal material spoken in affective tones of

voice (happy, sad, angry, and neutral), investigators paired

an emotionally intoned sentence with one spoken in a neutral

tone. Subjects were told to attend to one ear and to report

the content and affective tone. Subjects also had a

multiple choice checklist to record what they heard.

Content words were chosen from the unattended sentence,

attended sentence, and synonyms for those words. Subjects

were better able to accurately identify the emotional tone

when they attended to the left ear and were more accurate

for content analysis with messages delivered to the right


ear (Ley & Bryden, 1982). This dissociation between the

content and affective analysis is strong support for the

right hemisphere's role in emotional processing.

A final related area of study is the effect of

"priming" performance of the right hemisphere with affect-

laden material. Stimulating the right hemisphere in this

manner could help to reveal whether emotion is a complex

form of pattern recognition or a unique function of that

hemisphere. To do this, one would have to perform a

lateralized task such as face recognition or a verbal,

dichotic listening task and show increased left visual-

field effects or improved left ear performance, over and

above baseline, in the presence of emotional arousal.

Results such as this were obtained in tasks devised by Ley

and Bryden (1983). Word lists with high affective loading

and high or low in imageability were memorized by subjects

after they had performed a lateralized face recognition task

and a dichotic listening task. Even though one task was

verbal in nature, the affective loading produced a greater

left ear advantage and increased the left field advantage

with facial recognition. In the dichotic task the effect

was absent for neutral words. The improvement in

performance was a direct result of the addition of emotional


Related studies involved the effects of induced emotion

or personality variables on emotional or perceptual


processing. Tucker, Antes, Stenslie, and Barnhardt (1978)

induced anxious mood and then analyzed performance on an

auditory attentional task. They exposed half of their

subjects to stressful laboratory conditions while the other

half of the subjects received a casual, reassuring

instructional set. Each subject was then asked to perform a

hemisphere specific task: a lexical decision task for the

left hemisphere and a spatial decision task for the right

hemisphere. Stressed subjects, with higher reported anxiety

on the state portion of the Speilberger State-Trait Anxiety

Questionnaire (Speilberger, 1968), had greater errors on

both the spatial and verbal task in the right-visual field.

The investigators suggest this finding represents a

processing load specifically on the left hemisphere which

subsequently alters performance. This conclusion is in

direct contradiction to the assumption that Kinsbourne made

about concurrent tasks serving to improve performance and

prime the hemisphere involved in the task (cited in Tucker &

Williamson, 1984).

To study more accurately this aspect of performance,

subjects were then asked to judge the loudness of tones

binaurally presented, to check for an attentional bias.

This section of the study had subjects divided according to

high or low trait-anxiety and each was then subjected to an

anxiety induction and questioned accordingly. Attentional

bias was significant in the high trait-anxious group for the


right ear with more loudness judgments made for that ear

even though both ears were equated for number of trials

actually having loud tones. Two interpretations are

possible; one is that high trait-anxious individuals

experienced decreased right hemisphere processing or that

there was increased left hemisphere activation. Generally,

the study has many flaws and is not thought to test

adequately either hypothesis (MacDonald & Hiscock, 1985).

In another study of the hemispheric asymmetry of

induced mood states, Tucker, Stenslie, Roth, and Shearer

(1981) had subjects undergo hypnotic inductions of happiness

and sadness. Each subject was then instructed to perform an

arithmetic task and several visualizations with imagery

ratings (for intensity of visualized image). The same

attentional bias task was used as in the preceding study.

The hypothesis was that the attentional bias might reflect a

hemispheric activation and concurrent task decrement

(following from the previous study) or enhancement of

performance if Kinsbourne's explanation of priming effects

was accurate. The researchers also examined EEG measures of

hemispheric activation. There were no differences between

moods on the arithmetic task while the shift from happy to

sad mood produced a dramatic decrease in the ability to

visualize. Concomitantly, there was little difference in

attentional capacity in the euphoric/happy condition (as

measured by change in task performance). A right ear bias


was observed in the depression condition suggesting that a

transient depressive mood may coincide with a decrement in

right hemisphere processing capacity. Analysis of the EEG

data yielded little difference in alpha activity over the

frontal regions in the happy condition. In contrast, during

the depressed condition, there was greater alpha activity in

the left frontal region and less alpha with increased

desynchronization in the right frontal region, during the

depressed condition. Interpreting desynchrony as

activation, it appears the frontal regions are mediators of

mood and that the right frontal region, in particular, is

activated in transient depression.

In a more recent study examining emotional processing

in subjects whose personality characteristics include

different baseline levels of arousal, subjects were divided

according to their "trait" scores of anxiety and how they

manifested that anxiety. The 4 groups of subjects included

those who scored high and low in anxiety and who expressed

their feelings accurately and those individuals who did not

accurately report or express their feelings of anxiety.

These latter 2 groups are labelled the defensive anxious and

the repressors or deniers. The 4 groups of subjects have

been shown to react differently on a variety of

physiological and behavioral measures. True low-anxious

individuals are the least physiologically reactive to


stressors and the repressors are the most physiologically

reactive to stress.

These 4 groups of subjects were exposed to dichotic

words tests which paired positive, negative, and neutral

monosyllabic words with each other. Subjects were

instructed to choose the word they thought they had heard

and mark it on an answer sheet. There was a significant

effect for positive affect but not for negative affect in

subjects who demonstrated a baseline right ear perceptual

advantage on neutral words (75% of all subjects demonstrated

a right ear advantage). Both repressors and true high-

anxious subjects, the most physiologically reactive, had a

greater right ear effect with emotional words, while true

low-anxious subjects had a nonsignificant decrease in this

advantage with emotion (Wexler, Warrenburg, Servis, &

Tarlatzis, 1986). The investigators proposed the following

brain changes as mediators of increased right ear advantages

on language related tasks: 1) decreased flow of material

from right to left hemisphere, no competition; 2)

facilitation of auditory receptive function with the

addition of emotion, a priming effect; 3) decrease in the

same auditory receptive ability in the right hemisphere and

4) increased sensitivity of the left hemisphere, frontal

activation to stimulus specific arousal. This latter

ability may simply mean the left hemisphere's specialized

language function, but none of the proposed reasons for the


experimental outcome are mutually exclusive. These results

do contradict the study by Tucker, Stenslie, Roth, and

Shearer (1981) but may confirm Kinsbourne's notion about

priming facilitating the function of a particular


Data from a wide variety of studies seems to suggest

the hemispheres are differentiated with respect to analyzing

emotional stimuli and regulating mood. It is observed less

frequently that each hemisphere is preferentially biased or

activated when exposed to positive emotions. Some of these

ideas are controversial. However, one thing is clear,

rarely can brain functions be dichotomized or simplified.

The hemispheres of the brain function relative to each other

and cortical structures mitigate the functioning of

subcortical structures in much the same dialectical fashion

as right and left hemispheres effect each other. It is

likely to be the case that some combination of the

aforementioned experimental conclusions have relevance to

such a necessary and basic function as emotion, but that no

one study can capture the complexity.

Brain Mechanisms in Emotion

Not all the brain mechanisms involved in the experience

and expression of emotion are known. This is further

complicated by the fact that when one speaks of emotions,

various connotations are raised as a result of the

imprecision with which "emotion" is defined within and


across disciplines. Pribram (1981) and later Heilman (1987)

note that there is a clear demarcation between primary

drives such as hunger and thirst, that they can be termed

organismic motivators or drive states. A second level

behavioral control system may be termed emotions, for though

they involve adaptive functions that may benefit survival,

they involve higher cognitive functions. Whereas drives are

the result of monitoring the internal milieu, the emotions

serve to link the internal milieu with the external

environment. The inverse is also true and may relate to how

memories are formed. Such a link between the environment

and the internal state involves systems that subserve

arousal, activation, perception, memory, meaning, and motor


Various models have been proposed depending upon the

definition one adopts for emotion. Nevertheless, it appears

that some support exists for the interactive cognitive-

arousal model as it encompasses elements of the arousal and

cognitive appraisal models. The literature in these areas

may be captured by 2 studies. Maranon (cited in Heilman,

Watson, & Bowers, 1983) induced physiological arousal in his

subjects by injecting them with adrenaline. The drugs

served to mimic the arousal in the sympathetic nervous

system when the organism experiences intense emotion.

However, The induction of arousal alone within the

peripheral nervous system did not, as James proposed (cited


in Plutchik, 1980), result in normal emotional feeling.

Next, Maranon had subjects recall an emotional memory. This

manipulation was not sufficient to produce an emotional

feeling in the normal state unless the subject was

experiencing concomitant pharmacologically induced arousal.

An earlier study by Schacter and Singer (cited in

Schacter, 1970) also provided some evidence to buttress the

cognitive-arousal interaction model by subjecting volunteers

to stressful and nonstressful situations with or without a

state of pharmacologically induced arousal. Two groups of

subjects did not experience subjective or objective evidence

of emotion: those who received the adrenaline injection

alone or those who were subjected to a stressful situation

alone. When the subject received the injection promoting

peripheral system arousal and was put into a stressful

situation, the subject often experienced intense emotion.

Although Plutchik (1980) has reviewed the shortcomings of

this study including the elimination of some subjects, use

of heartrate as a dependent measure, and the lack of

replication, variations on this experiment have pointed to

the necessity of an interaction between arousal and

cognition to produce an emotional state. However,

peripheral arousal (such as heartrate) may be a secondary

by-product of emotion and the interaction necessary may be

central arousal and cognition.


The interaction between central arousal and cognition

is necessary for any subjective experience. Arousal

attention is what provides the necessary cortical tone for

cognitive processes. The mesencephalic reticular formation,

thought to be responsible for arousal (Watson, Valenstein, &

Heilman, 1981; Heilman, Watson, & Valenstein, 1985), has

connections to the thalamic nuclei and to certain areas of

the neocortex that are also important for arousal.

Cognitive set is mediated by appropriate cortical tone so

that incoming stimulation reaching the primary sensory areas

may be decoded, analyzed in association areas, matched with

existent representations or templates, and salience

established. Once emotion has signalled the significance of

the stimuli, the accompanying visceral and peripheral

changes are brought about by the overlap between the limbic

memory systems and connections to the thalamus and

hypothalamus. The thalamic-hypothalamic outflow results in

endocrine and autonomic response, while thalamic-brainstem

reticular formation connections increase arousal.

Pribram and MacGuiness (1975) differentiated attention

into 2 components; arousal-attention which is intimately

involved in providing the aforementioned cortical tone

required for the orienting response and activation-

attention, which is called into play with sufficiently

sustained arousal which warrants action. The activation

system is thought to have 2 pathways from the brainstem,


thalamic nuclei to motor cortex and directly from the

mesencephalon to neocortex.

The importance of understanding these pathways when

integrating data about the processing of emotion in the

brain may be understood by reviewing the observed effects of

right hemisphere lesions. Individuals with right hemisphere

disease may have severely impaired ability to orient in

space, make sense of spatial cues, comprehend emotional

facial expression or understand emotional prosody.

Likewise, these individuals have impaired expressive ability

and are inadequately aroused. Patients with these

difficulties who undergo electrical stimulation show

decreased galvanic skin response on the left side, with

increased response on the right side. The same hypoarousal

appears when emotion laden stimuli are presented to the

right hemisphere (Heilman, Watson, & Bowers, 1983). The

extremes of hypoarousal occur with lesions along portions of

the pathways responsible for arousal. Lesions in the

brainstem result in coma; less extreme, but nevertheless, a

unique behavioral sequel of lesions in the pathways thought

to be responsible for arousal attention is unilateral


In summary, lesions in the right hemisphere impair both

the cognitive set and the arousal needed for emotion. Taken

together with the vast literature indicating the right

hemisphere of normals is superior in processing all types of


emotional stimuli over and above their configurational

patterns, it appears that important and adaptive

communication skills reside in the right hemisphere.

Rationale and Specific Hypotheses

The present study utilized the classic paradigm of the

tachistoscopic presentation of emotional facial material.

Pictures of emotional faces were presented both unilaterally

and centrally and may be congruent or incongruent chimeric

faces. Past research has indicated that whole emotional

facial stimuli presented unilaterally are reported more

accurately and responded to faster in the left visual-field.

Furthermore, the initial right hemisphere advantage in

processing emotions of chimeric facial material with facial

halves placed at central fixation, such that the right half

of the face was in the right visual-field, while the left

half of the face was in the left visual-field, yielded

consistently greater accuracy and faster reaction times to

the hemiface in the left visual-field.

Therefore, the present study sought to utilize chimeric

faces with congruent emotional expressions and chimeric

faces comprised of 3 emotions, happy, sad, and neutral, to

test not only the right hemisphere's superiority in

processing facial emotion, but also faces comprised of 2

competing emotions to test for the presence of valence

effects. It was thought that the forced choice of an

emotion across the visual-fields might elicit emotional


valence effects, if these existed. Subjects' performance,

based on accuracy and reaction time, could then be compared

to the standard findings in the literature.

The addition of a mood induction to the study served 2

purposes. First, the results of past attempts to understand

emotional processing in various emotional states has yielded

equivocal results. An attempt to alter mood in a within

subjects design allowed comparison of baseline processing

with the changed mood state. Secondly, various psychiatric

disorders have as one of their symptoms the state of

anxiety. There is clinical relevance to understanding the

relationship between cognitive processes and the processing

of emotional material when anxious.

Past literature has raised the question of whether

emotional processing and recognition is a more complex

spatial/pattern recognition task, rather than a unique

ability. In studies utilizing facial identity as a

covariate (Bowers, Bauer, Coslett, & Heilman, 1985), and in

those studies where an affective load is imposed on the

stimuli (Bryden & Ley, 1983), there does appear to be a

unique effect for emotion apart from the ability to

discriminate patterns.

The present study presented chimeric faces with

incongruent or congruent emotional expressions, at each of 3

card positions: central fixation, such that each half of

the face was in the right or left visual-field; the entire


stimulus in the left visual-field; and the stimulus in the

right visual-field. In each of the card position conditions

certain hypotheses were generated:

la) For incongruent chimeric faces presented at

midline, the left half of the face would be

reported more accurately.

b) The reaction times to the left side of the

faces would be faster than for the right.

c) The findings for congruent chimeric faces

would accentuate the findings for incongruent

faces, i.e. hit rate would be greater and the

reaction times would be faster.

2a) If right hemispheric superiority in facial

processing was upheld, then both halves of

the chimeric face presented in the left

visual-field would be reported equally well

(no difference in the correct responses

between the right and left halves of the


b) There would be no difference in reaction time

for either hemiface that was presented in the

right or left visual-field.

3a) Responses to facial stimuli presented in the

right visual-field would be slower and less

reliable (i.e. a decreased accuracy).


There is conflicting evidence of each hemisphere's

contribution to processing of emotional valences. Any

finding of a valence and visual-field interaction might

reflect a real difference in identifying certain emotions.

For example, more extreme emotions are reported more

accurately (Natale, Gur, & Gur, 1983). On the other hand, a

visual-field interaction with emotion might reflect the

preferential processing in a visual-field for certain

emotions. The following hypotheses were also addressed

regarding the various emotions:

4a) The emotion of happy would be reported more

accurately than sad.

b) Sad affect would be reported more accurately

than neutral.

c) The differences would be present across the

card positions of central fixation and left

or right visual-fields.

d) The differences between emotions (happy > sad

> neutral) would also be reflected in

reaction time. Reaction times would

increase, from happy to sad to neutral.

An anxious mood induction was undertaken to change both

the cognitive set of the subjects and the level of arousal.

Studies of mood inductions and emotional processing have

also had conflicting results (Silberman & Weingartner,

1986). In general, the studies reflect a tendency of the


left hemisphere to process information less efficiently and

for the right frontal region to show relatively greater

activation on EEG measures (Ahern & Schwartz, 1985). The

following hypotheses were made regarding the mood induction

procedure and its effects on task performance:

5a) The manipulation check using the Profile of

Mood States, a behavioral checklist, would

show significant increases in Anxiety/Tension

and possibly in Vigor, Anger, and Depression.

b) Differences in self-reported mood would

differ according to each subject's coping

tendency. The Repression-Sensitization Scale

was used to measure coping strategy.

c) The differences on the Repression-

Sensitization scale would indicate that

repressors, while not reporting as much mood

change would have greater self-report

evidence of arousal, whereas sensitizers,

while cognitively aware and endorsing

arousal, would endorse less self-reported

evidence of physiologic response (Weinberger,

Schwartz, & Davidson, 1979).

d) Threshold checks before and after the

anxiety induction would change. The

direction of change would be an inferred

indicator of physiologic arousal (exposure


durations would be shorter) or cognitive

disruption (exposure durations would be


6a) The mood induction would improve emotional

processing of stimuli presented in the left

visual-field (right hemisphere).

Finally, few studies have systematically examined the

presence of sex effects. The present study included males

and females in order to determine what, if any, effects

gender might have on emotional processing or the mood


7a) The performance of males on the emotional

processing task would be different than the

the performance of females (but in an unknown


Design and Analyses

The present study was a 2 (sex) x 2 (anxiety) x 3 (card

position) x 2 (hemiface) factorial design with 1-between

subjects factor (sex) and 4-within subjects factors

(anxiety, card position, hemiface, and emotion). The design

encompasses data generated on incongruent chimeric faces.

However, an analysis comparing performance on incongruent

vs. congruent chimeric faces was also carried out. The

design for this analysis was a 1-between and 4-within

subjects factorial design. However, it was not possible to

retain hemiface as another within subjects factor because of


the small number of congruent chimeric faces (9

presentations per card position); if these 9 presentations

had been further broken down according to hemiface (6

possible positions, left/right hemiface at each card

position), the likelihood of empty cells would have

diminished the meaningfulness of this factor.

For all the data generated, 2 dependent variables were

used. To analyze accuracy, a hit rate was calculated and

for efficiency of processing, a voice activated reaction

timer registered reaction time in milliseconds.

The data were analyzed using the Statistical Analysis

System (SAS) General Linear Models program (SAS, 1985).

Specifically, the analysis used Analysis of Variance (ANOVA)

with repeated measures on the 4-within subjects factors.

Significant results were further decomposed using 1-way

Analysis of Variance procedures (ANOVA) and in the case of

multiple factor interactions, dependent t-tests were used

for pairwise comparisons.



The subjects of the study were 31 male and 32 female

undergraduate volunteers from the University of Florida

Psychology Department subject pool. Each subject received

class credit for participation. The subjects met all the

following criteria to be included in the study: 1) strongly

right handed (no first degree relatives who were sinistrals;

use of right hand for selected items of the Annett

Handedness questionnaire); 2) no previous history of

neurologic or psychiatric illness; 3) no history of speech,

hearing or learning disability; 4) no history of any recent

stressful event; 5) no significant drug or alcohol use

within 72 hours of study participation; and 6) normal or

corrected to normal vision (see Appendices A, B, and C, for

the Informed Consent, screening questionnaire and debriefing


Each subject acted as their own control by undergoing a

portion of the experiment at a baseline mood level and

another portion of the experiment after being made to feel

anxious about their performance. Half the subjects received

the anxiety induction at the beginning of their performance,


while the other half received the induction at the midpoint

of the experiment. One subject was released without

completing the experiment because of clear distress.

Debriefing was carried out for this subject in the usual

fashion, as was a brief period of relaxation until mood

level was appropriate.

Stimuli and Apparatus

A Gerbrands 3-channel tachistoscope with accompanying

multi-channel millisecond timer and a Gerbrands voice

activated reaction timer with microphone were used for

presentation of stimuli and for recording time to response,

respectively. Fields 2 and 3 were used for the presentation

of the test stimuli and Field 1 was used to present both a

central fixation dot and at random intervals (trials 8, 17,

25, 33, 58, 73), a centrally positioned number. This test

number had to be reported each time it was viewed to insure

the correct position of the eyes throughout the experiment.

The visual stimuli were derived from a set of facial

photographs produced by Ekman and Friesen (1975). The set

consists of male and female models expressing each of 7

cross-culturally relevant emotions: happiness, sadness,

anger, disgust, fear, and indifference. Those rated most

intensely emotional in each of 3 consistently reported

emotions, happy, neutral, and sad, were used to compose the

final stimuli. Three independent raters judged the same 4

faces (2 male and 2 female models) as the most expressive


indicants of the 3 target emotions. Past research has

indicated that the reliability and strength of the findings

is obtained with more extreme valences of emotions

(Silberman & Weingartner, 1986). Therefore, only these 4

faces and the 3 emotions were used in the study. All facial

stimuli were altered so that extraneous features such as

hairstyle, ears, earrings, etc., would not influence the

processing of the faces apart from their emotional


One of the 4 sets of faces was used in determining

threshold exposure duration for each subject and 3 sets of

faces were used in the experiment proper. All stimuli were

constructed in the following manner. Composite facial

combinations were made for each of the 3 faces so that 9

possible combinations were available for each of the 3

models. These combinations were happy-happy, happy-

neutral, happy-sad, neutral-neutral, neutral-sad, neutral-

happy, sad-sad, sad-happy, sad-neutral.

All congruent emotional faces were composed of the

right half of the photograph (the poser's left side) and a

right half mirror-reversal of that face. The resulting

congruent chimeric face was symmetrical and used a left-left

composite expression. The reasoning for this lies in 2

experimental observations; first, that only portions of a

face act as cues to identification and secondly, when

viewed, the facial expression is asymmetric. The left side


of the face is routinely judged as more emotionally

expressive (Borod, Koff, Lorch, & Nicholas, 1986; Borod &

Caron, 1980).

Each half of the experiment consisted of 3 blocks of

composites for a total of 81 stimuli. All stimuli in each

half of the experiment were randomized and order of

presentation with the anxiety manipulation was

counterbalanced such that half the subjects received the

manipulation before the first 3 blocks and half the subjects

received the manipulation before the second 3 blocks.

Each subject underwent 1 of 2 procedures which differed

only in the order in which they received the anxiety

induction. All subjects took several questionnaires to

assess general mood and anxiety levels throughout the

experiment (The Profile of Mood States--POMS; McNair, Lorr,

& Droppleman, 1981); the second questionnaire assessed the

individual's coping style under stress (The Repression-

Sensitization Scale--R-S scale; Epstein & Fenz, 1967; see

Appendix D).

Mood Induction

The mood induction called surgency, involving subject

deception, was a task patterned after that of Hiroto and

Seligman (cited in Carver, Antonini, & Schier, 1985; see

Appendix E). Subjects were shown 5 problem sets consisting

of 10 stimulus cards which varied along 4 dimensions: shape,

color, letter, and case-type. These 4 dimensions were


further subdivided into 2 values: circle/square,

black/white, A/T, and UPPER-CASE/lower-case. Each stimulus

card was bisected and a pattern appeared in each half of the

card. The examiner instructed the subject to use hypothesis

generation strategies to deduce the predetermined answer.

They responded by guessing which side of the card (right vs.

left) contained the answer. Each subject's response

received "correct/incorrect" feedback. The feedback was

designed to "help" each subject eliminate some of the 8

values while retaining others until they knew the eventual


The anxiety-provoking deception was that feedback on

each of the 5 problems sets was predetermined and not

contingent upon the subject's response. Each subject failed

at the task and this failure was linked both with their

performance on the experimental paradigm and with the

individual's brain function.


At the outset of the experiment all subjects, regardless

of whether they received the mood induction first or second,

completed an Informed Consent for participation, a screening

questionnaire consisting of the exclusion criteria and the

Repression-Sensitization Scale. The first half of the males

and females received the anxiety induction immediately

following the preliminary questionnaires, while the

remaining subjects received the manipulation after the first


block of emotional faces had been presented on the

tachistoscope. Presumably this would account for order

effects of the manipulation on the dependent measures.

The first group of subjects then received the anxiety

manipulation (surgency task) and the Profile of Mood States

questionnaire. Threshold exposure durations were done with

the threshold stimuli using the method of ascending and

descending thresholds. Each subject started at 110

milliseconds (msec) and the timer was increased/decreased in

10 msec increments until 70% of the stimuli were correctly

identified as having 1 or 2 emotions in the face. If

threshold was obtained by decreasing in 10 msec increments,

the exposure was checked by then picking an exposure

duration below that threshold and then increasing the

exposure by 10 msec increments.

The timer, microphone, and reaction timer were then

checked for each subject and the instructions were given for

the first randomized block of 81 stimuli. As in the

threshold determination, these faces had either 1 or 2

emotions expressed on the face, but rather than responding

to this, the subjects were told to denote the emotion as

happy, sad, or indifferent (neutral was changed to

indifferent because it more easily triggered the voice onset

switch in the reaction timer). They were to indicate the 1

emotion that was immediately apparent. Even if the subject

thought they saw 2, emotions the instructions were


emphasized to denote the most salient emotion. To insure

proper eye fixation throughout the experiment, a central

fixation dot was alternately presented between target

stimuli. At randomly assigned intervals, to check the

subject's fixation, a digit would appear. Subjects were

instructed to report that number. The examiner recorded

both their response (happy, sad, neutral) and their reaction

time to response.

Following the first block of stimuli, the subjects were

told they could take a short break to visit the restroom,

get water, or to stretch. Upon their return they were

partially debriefed about the anxiety manipulation, the

expected effect upon their mood and performance. They were

told the feedback regarding their failure was bogus and the

task was designed for everyone to fail. After subjects were

reassured about their performance, they were given a second

mood questionnaire pomsS). The second randomized block of

81 stimuli was presented after the threshold was determined

for a second time.

Those subjects receiving the manipulation in the second

half of the experiment were given the Informed Consent,

screening questionnaire, R-S scale and the POMS. Thresholds

were determined as previously described and stimuli were

given. The individuals were given some mildly concerned

feedback about their performance and then the anxiety

manipulation. The POMS was given next as the manipulation


check and block 2 stimuli were given after the post-

manipulation threshold was determined. In each case the

POMS and threshold determinations were used to check the

effects of the surgency task. Following the experiment, the

subjects were fully debriefed and all questions were

answered. The entire experiment took 1.5 hours. A

schematic diagram of the procedures is provided in

Appendix F.


In order to test the various hypotheses the data were

analyzed according to how anxiety affected both accuracy and

response time. These analyses were done in two phases. In

the first phase the incongruent chimeric faces were measured

by two separate dependent variables, hits and reaction time

in milliseconds (msec). In the second phase, the impact of

congruence/incongruence in the chimeric faces was analyzed.

Two separate analyses were also performed here, with the

first using the percentage of accurate identifications as

the dependent measure and the second using reaction time as

the dependent variable.

As noted earlier, the design was a 1-between and 4-

within subjects factorial design with sex, the between

subjects variable and the following within subjects factors:

anxiety (2 levels-anxious, non-anxious) x card position (3

levels-central, left and right visual-fields) x hemiface (2

levels-right, left) x emotion (3 levels-happy, neutral,


Preliminary Analyses

The 63 male and female subjects in the present study

underwent a manipulation of their mood. The following

predicted results are relevant to the subsequent analyses:

5a) The anxiety manipulation would have its effects

reflected by increases in anxiety/tension scores.

5b) Changes in mood would also vary as a function of

coping strategy as measured by the Sensitization-

Repression Scale.

5c) Threshold checks before and after the anxiety

induction would change. The threshold change, an

inferred result of physiologic arousal or

activation would:

1) decrease, if it is an arousal phenomenon or

2) increase, if the change is a result of

cognitive disruption (or activation; Bowers &

Heilman, 1980b).

During the manipulation and subsequent check using the

Profile of Mood States pomsS), it appeared the target factor

of Tension-Anxiety (TA) on the POMS captured changed mood

for most subjects. However, a few subjects failed to

exhibit the expected response to the mood induction. While

it is only speculation, in some subjects, suspicion about

the existence of deception in psychology experiments may

have made them wary of examiner feedback; in others, lack

of involvement in the experimental procedure may have


contributed. Whatever the reason, the manipulation did not

work equally well for all subjects. Consequently, the

preliminary analyses took this observation into account by

determining if an induction effect had been achieved by

analyzing both the pre- and post- manipulation Tension-

Anxiety score and the pre- and post- Total Mood Disturbance

score (TMD). This latter score is a total of the POMS 6

factor scores (including the reverse scored Vigor factor).

Levels of self-reported anxiety may also differ

according to an individual's coping style. The literature

suggests that people who use the coping styles of repression

and sensitization, experience and express feelings of

anxiety in a different way. People who use the coping style

of repression have a tendency to under-report negative

feeling states while being highly physiologically reactive;

those people who use sensitization have a tendency to over-

report negative feelings, but are not highly physiologically

reactive. Concomitant with differing self-report,

individuals in these categories have related differences in

stimulus thresholds and autonomic responses (e. g. galvanic

skin response; Epstein & Fenz, 1967). The Repression-

Sensitization Scale (R-S) was also used to provide a

possible explanation for differences in self-reported

anxiety levels.

Each of the two analyses carried out was a 2-between

(sex, coping style) and 1-within (anxiety) subjects design


using an ANOVA procedure with score on TA and TMD as the

dependent variables. The analysis produced no significant

main effect for the anxiety manipulation on TA or on TMD.

There was also no significant difference on TA or TMD by sex

or coping style nor was there an interaction between any of

these variables.

As mentioned previously, it was observed that some

subjects responded in an unexpected manner to the mood

induction. It is not clear whether suspiciousness or

communication between subjects resulted in the deception not

working. Some subjects responded to debriefing by stating a

general lack of trust in psychology experiments. Because

the manipulation of mood was integral to the remainder of

the study, elimination of some subjects was necessary. It

was decided to calculate a difference score for all

subjects, between the anxious and non-anxious TA scores.

The distribution of difference scores was plotted and found

to have a mean of 9.62 and a standard deviation of 7.61.

These difference scores were examined for their possible

relationship to other factors. Pearson product moment

correlations between the TA difference scores and coping

style, sex and order of manipulation presentation were

computed. None of these correlations approached

significance. Thus, the difference scores appear

uncontaminated by extraneous factors. Therefore, a positive

anxiety effect for this distribution was defined as those


subjects whose difference scores fell over 1 s.d. above the

mean (i.e. > 3; a similar distribution analysis could have

been employed for the TMD scores, but since the TA and TMD

scores were significantly positively correlated, r = .74, E

< .0001, and because anxiety was the targeted emotion, the

anxiety difference score was used).

The original analysis was performed a second time on

these subjects. Forty-eight subjects, 20 males and 28

females, comprised this group. The analysis revealed a

significant main effect for the anxiety manipulation,

F(1,44) = 5.36, p < .03. Again, none of the other main

effects or interactions were significant. On the basis of

this analysis it was decided to employ these 48 subjects in

all subsequent analyses.

Finally, studies of emotional processing under

different emotional states often fail to assess convergent

evidence of the mood state. In this study, exposure

duration thresholds were determined prior to the anxiety

provoking task and after. There was a significant decrease

in threshold from anxious to non-anxious states, F(1,46) =

5.25, p < .03 (mean pre-induction threshold = 89 msec; mean

post-induction threshold = 110 msec; see summary of

significant results in Table 1).

Table 1. Summary of Significant Findings.

I. Incongruent chimeric faces: Dependent measure--hits

A. Main effects p-value
1. Hemiface p < .0001
2. Emotion p < .005
3. Card position (trend) p < .07

B. Interactions
1. Anxiety x hemiface p < .04
2. Hemiface x emotion p < .008
3. Card position x emotion p < .03
4. Anxiety x hemiface x emotion p < .05

II. Incongruent chimeric faces: Dependent measure-
reaction time

A. Main effects p-value
1. Hemiface p < .0006

B. Interactions
1. Hemiface x emotion p < .04
2. Anxiety x card position x
hemiface x sex p < .05

III. Incongruent vs. congruent chimeric faces: Dependen
measure--percent hits

A. No main effects p-value

B. Interactions
1. Anxiety x card position x
congruence x sex p < .04

IV. Incongruent vs. congruent chimeric faces: Dependen
measure--reaction time of hits

A. Main effects p-value
1. Card position p < .05

B. No interactions



Primary Analyses

There were 2 dependent variables of primary interest in

this study: accurate identification of facial emotion

(hits) and speed of processing and response (reaction time-

RT in msec). For purposes of clarification, a hit on an

incongruent face meant that 1 of 2 target emotions present

on the face was correctly identified. For congruent faces,

only 1 emotional expression was present and had to be

correctly identified. Therefore, there was a 67% likelihood

of a hit for incongruent faces while there was 33%

likelihood of a correct response for congruent faces. Each

of the dependent variables was subjected to a 1-between, 4-

within subjects ANOVA. These major analyses were first

performed on the incongruent chimeric faces and then

performance on incongruent faces was compared with congruent

faces. The analysis of the incongruent chimeric data will

be described first and is relevant to the following

predicted findings:

la) For incongruent faces presented at central

fixation, the left hemiface would have a

higher hit rate.

2a) If right hemisphere superiority in processing

is upheld, then the hit rate for both halves

of the face presented in the left visual-

field would be equal.


3a) The number of hits to each hemiface presented

in the right visual-field would be less than

hits obtained for other card positions.

4a) The hit rate for the emotion of happy would

be greater than for sad.

4b) The hit rate for sad emotion would be greater

than for neutral expressions.

4c) The differences between the emotions would be

present across the card positions.

Taking the accuracy data first, main effects were found

for hemiface, F(2,45) = 62.85, p < .0001; emotion type,

F(2,45) = 5.99, p < .005 and a nonsignificant trend was

found for card position, F(2,45) = 2.81, p < .07. Each of

these main effects is qualified by the significant higher

order interactions that were also found. A significant 2-

way interaction involved card position by emotion, F(4,184)

= 3.00, p < .03. Two other 2-way interactions involving

anxiety by hemiface and hemiface by emotion were, in turn,

qualified by a higher order 3-way interaction. The 3-way

interaction of anxiety by hemiface by emotion, F(2,92) =

3.13, p < .05, and the significant card position by emotion

interaction that was not directly involved in the more

complex interaction, were subjected to post-hoc analyses to

decompose the nature of these interactions.

Exploring the 2-way, card position by emotion

interaction first (see Figure 1), post-hoc comparisons were


done in two manners; first, to clarify differences among the

3 emotions of happy, neutral and sad (holding card position

constant) and secondly, to determine the differences for an

emotion across the card positions of central fixation, left

visual-field and right visual-field (holding emotion

constant). For each level of card position, a 1-way ANOVA

was performed with type of emotion as the independent

variable and the hit data (collapsed across all factors not

involved in this interaction) as the dependent measure. All

three of these analyses produced significant effects, e.g.

F(2,94) = 36.85, p < .0001, for the left visual-field card

position, which was the smallest effect. Each of these

significant effects was then followed up by dependent groups

two-tailed t-tests. In the pairwise comparisons of emotions

at central fixation, the comparisons for mean hits were made

between happy-neutral, happy-sad and neutral-sad (t = 11.52,

p < .0001; t = 12.18, p < .0001; t = 2.36, p < .02,

respectively). The paired comparisons for the same

emotional pairs in the left visual-field yielded 3

significant results (t = 8.55, p < .0001; t = 5.63, p <

.0001; t = -3.01, p < .004, respectively). Finally, the

right visual-field pairwise t-tests were also all

significant (t = 11.14, p < .0001; t = 6.02, p < .0001; t =

-5.08, p < .0001, respectively). To summarize, across card

Happy *** P < .0001
** p < .005
p < .05


5.0- **

C 4.5 -

M 4.0-

-3 *
Ix 3.5-

3.0 Neutral --

2.5- -


Center Left Right


Figure 1. 2-way Interaction on Hits: Card Position x
Emotion (Holding Card Position Constant).


positions, happy is reported more accurately than sad or

neutral, but sad and neutral are reported more accurately at

different card positions. At central fixation, neutral is

reported better than sad while, in either the right or left

visual-field, sad is reported more accurately than neutral.

The differences are most pronounced for the right visual-


The alternative analysis of the same interaction sheds

light on accuracy between each card position. Once again,

1-way ANOVA's were performed with card position as the

independent variable (and the data collapsed across all

factors not involved in the interaction). Each of these

analyses produced significant results, e.g. F(2,94) = 22.19,

E < .0001, for neutral emotion which was the least

significant result. The pairwise comparisons were made on

the following pairs: central-left visual-field, central-

right visual-field and left visual-field-right visual-

field. For the emotion of happy, the most easily identified

emotion, the resultant t-values were: t = 7.62, p < .0001; t

= 6.37, 2 < .0001 and t = -1.80, 2 < .08. The comparison of

neutral facial emotion at the various card positions

resulted in the following significant results: t = 2.95, p <

.005; t = 7.27, E < .0001 and t = 3.63, p < .0007,

respectively. Lastly, comparing sad emotion across space

yielded 2 significant differences, between central-left and

central-right visual-field but, no significant difference


between left and right visual-fields (t = -7.47, R < .0001;

t = -7.34, p < .0001 and t = 0.21, p < .84, respectively).

To summarize, both happy and neutral faces are more

accurately identified at central fixation. There is a non-

significant trend indicating happy is identified more

accurately in the right vs. left visual-field. However,

there is a significant disparity between the fields for

neutral emotion; neutral is more accurately identified in

the left visual-field. No differences exist between the

fields in the identification of sad expressions but, much

poorer performance is registered for central fixation (see

Figure 2).

Turning attention to the 3-way interaction (anxiety x

emotion x hemiface) found for the accuracy data, the effects

of the 3 emotions within each combination of anxiety level

and hemiface were analyzed by a 1-way ANOVA, collapsing

across the other factors not involved in the interaction.

Each of these analyses produced significant effects, e.g.

F(2,94) = 43.05, p < .0001, for the non-anxiety, right

hemiface data which was the least significant effect.

Pairwise comparisons were made between emotions in the

following order: happy-neutral, happy-sad and neutral-sad.

The finding for the right hemiface under anxious conditions

was, t = 9.65, 2 < .0001; t = 7.02, p < .0001 and t = -3.00,

p < .004, and the finding for the left hemiface under

anxious conditions was, t = 12.13,_p < .0001; t = 8.84,




Central Position
Right Visual Field
Left Visual Field


9 \

P< .08




Happy Neutral

Figure 2.


2-way Interaction on Hits: Card Position x
Emotion (Holding Emotion Constant).

*** p <
** p <
p <









E < .0001 and t = -1.71, p < .09, across the above

emotional pairs. Under non-anxious conditions paired

comparisons for the right hemiface yielded 3 significant

results: t = 8.46, p < .0001; t = 5.91, p < .0001 and t =-

2.97, p < .005, indicating that subjects' accurate response

rate to all 3 emotions are significantly different from one

another. Final paired comparisons for non-anxiety

conditions and the left hemiface resulted in the following:

t = 11.43, p < .0001; t = 9.79, p < .0001 and t = -1.09, 2 <

.28 (see Figure 3).

While the report of emotions, as measured by accuracy,

indicates differences between the emotions with accuracy

greater for the left half of the face, the preceding

analysis of the interaction did not capture the change in

each emotion across levels of anxiety. Consequently,

analyses were also carried out holding emotion and hemiface

constant and assessing the difference in each hemiface

between anxiety conditions. It was predicted that anxiety

would increase accurate identification of emotions presented

in the left visual-field. However, this interaction

reflects a more complex result. While pairwise comparison

t-tests revealed no significant differences between the

anxious and non-anxious conditions for any emotion presented

in each hemiface (see Figure 4), the basis of the anxiety

effect appears to be the slight decrease in accuracy for the

right hemiface for the neutral and sad emotions and an

o n


4 4 Q,
4 *

0 0 0
M4 CD o cM


* c
* z

S* z


o' 5
E x




o q q q
06e ic


Anx. &






I ,



1.0- ,


Figure 4.

3-way Interaction on Hits:
Emotion (Holding Hemiface,

Anxiety x Hemiface x
Emotion Constant).





increase in accuracy for these same emotions in the left

hemiface. The fact that this trend does not appear for the

happy emotion yields the 3 way interaction.

To summarize the findings thus far, the 3 emotions are

reported differently with happy being the easiest expression

to identify. Furthermore, people respond to the left side

of the face more frequently, regardless of card position.

When these effects are combined, there are significant

shifts in accurate reports as the stimuli are moved off

central position. These effects can best be summarized by

the significant decrease in the report of happy expressions

and concomitant increase in the report of sad expressions as

these expressions are moved into the left visual-field and

are processed by the right hemisphere. This change is

difficult to explain since the left hemiface even at the

central position is reportedly processed by the right

hemisphere. In addition, there are differences between the

right and left visual-fields for these 2 emotions. There is

no difference for the report of sad affect, while happy is

reported relatively better when it is presented in the right

visual-field. This may suggest a valence effect or

processing bias, although not a striking one. Finally, with

the imposition of anxiety the processing of the left and

right halves of the face change. Anxiety decreases accuracy

of report to the right hemiface simultaneous with an

increase in accuracy for the left hemiface.


The next major analysis of the data involved a 1-

between (sex) and 4-within subjects (anxiety, card position,

hemiface and emotion) ANOVA on reaction times. Reaction

time analysis is based on the assumption that hemispheric

superiority is reflected not only by accuracy, but by speed

of processing. The hypotheses relevant to these analyses


Ib) The reaction times to the hemifaces presented

in the left visual-field would be faster.

2b) There would be no differences in reaction

time between the halves of the face in either

the left or right visual-fields.

4d) The differences in responses to the 3

emotions would be reflected in reaction


This analysis resulted in a highly significant main

effect for hemiface, F(1,46) = 13.43, E < .0006. In

addition, a 2-way, hemiface by emotion interaction and a 4-

way, anxiety by card position by hemiface by sex interaction

were found (F(2,92) = 3.40, E < .04; F(2,92) = 3.13, p <

.05, for the 2- and 4-way interactions, respectively).

Since these 2 interactions do not entirely overlap, both

were subjected to post-hoc analyses.

While the hemiface main effect indicates that

processing is faster for the left hemiface, the hemiface by

emotion interaction indicates that the speed at which each


half-face is processed is to some extent a function of the

emotion displayed on that half-face. One-way ANOVA's were

conducted for each hemiface with type of emotion as the

within subject independent variable, collapsing across the

other factors not involved in the interaction. These two

analyses each produced significant results, e.g. F(2,94)=

57.99, p < .0001, for the left hemiface which was the least

significant result. Pairwise comparisons were carried out

holding hemiface constant and varying emotional pairs.

Significant differences for the right hemiface were found

for the following pairs, happy-neutral (t = 9.82, p <

.0001), happy-sad (t = 6.96, p < .0001) and neutral-sad (t =

-3.42, p < .001). Similar analyses were carried out for

differences among the emotions in the left hemiface. Two

significant differences were found for the pairs happy-

neutral and happy-sad, however, neutral-sad emotions did not

differ from one another (t = 12.49, p < .0001; t = 10.08, p

< .0001 and t = -1.52, p < .14, respectively).

While there are clearly differences in the speed at

which the various emotions are processed, when an analysis

is made holding emotion constant and comparing across the

sides of the face, regardless of emotion, the reaction time

is significantly faster to the left side of the face (see

Figure 5).

The significant 4-way reaction time interaction

involving sex, anxiety, card position and hemiface, posed


difficulties in post-hoc decomposition due to its

complexity. Appropriate pairwise comparisons were difficult

to designate until the interaction's relevance to specific

hypotheses could be graphically displayed (see Figure 6).

The complexity of the visual display did not help to clarify

the interactive effect each variable had on another. To

simplify the interaction it was decided to collapse the

interaction across anxiety conditions by creating a reaction

time difference score. To create this new variable, each

subject's reaction time in the non-anxious condition was

subtracted from the subject's reaction time in the anxious

condition. Mean difference scores were calculated by sex,

creating a 3-way interaction for card position by hemiface

by sex. These mean difference scores are graphed in Figure


Two 1-way ANOVA's for each sex, were calculated for

differences in reaction time as a function of hemiface. The

analysis for males, for the right hemiface revealed no

significant effect, while this analysis revealed a trend

toward significance for females, F(2,54) = 3.04, p < .06. A

similar analysis for the left hemiface across card position

was also carried out and revealed no significant effects for

either males or females.

At this point, post-hoc paired comparison t-tests were

carried out for females' performance for the right hemiface

to determine if mean differences in reaction times change

600 -

500 -






*** p < .0001
** p < .005
p < .05

Reaction Time Right Hemiface

Reaction Time Left Hemiface



Figure 5. 2-way Interaction for Reaction Time: Emotion x














I 0





A Anxious
NA Non-anxious


0 -

-100 -








Note: Positive numbers indicate longer reaction time for
anxious condition, therefore difference scores
between A-NA yielded a positive number.
t = -2.21, E < .04

Figure 7. 3-way Interaction for Reaction Time: Sex x Card
Position x Hemiface (Collapsed Across Anxiety


z 0

ix I


as a function of card position. The paired comparisons were

for the following card positions, center-left visual-field,

center-right visual-field and left visual-field-right

visual-field. The significant simple effect was for the

first comparison, t (46) = -2.21, p < .04. The other

comparisons were not significant. Thus, there was a

significant increase in reaction time from anxious to non-

anxious conditions when the right hemiface was shifted from

central position to the left visual-field (but for not for

the other card positions) for females (see Figure 7).

To summarize this section of the analyses, reaction

times were significantly faster to the left sides of the

faces. However, it was predicted that reaction times would

change as a function of the field the stimulus was presented

in, rather than as a function of hemiface (or position in

space). It was also predicted that the performance of males

and females would differ as a result of anxiety and this

appears to be the case. Specifically, there was a

substantial increase in reaction time for females when the

stimuli they were responding to shifted from being processed

by the left hemisphere to the right hemisphere (when the

right hemiface in the right visual-field was shifted into

the left visual-field).

The task of comparing incongruent faces with congruent

chimeric faces was the next phase of the analysis. Again,

the dependent variables of hits and reaction time were used.


However, prior to any comparative analysis, the faces had to

be equated for: frequency of presentation (the incongruent

faces were presented 27 times to each subject at each card

position, while the congruent faces were presented 9 times

to each subject at each card position) and the base rate

probability of a hit occurring on an incongruent chimeric

face (67%) as compared to a congruent chimeric face (33%).

To make the hit rates comparable for the two types of faces,

the number of correct identifications (hits) was divided by

the total number of presentations and from this percentage

the appropriate chance figure was subtracted. The

percentage of hits obtained for the 2 levels of congruence

was then used as the dependent variable in a 1-between (sex)

and 4-within subjects analysis (anxiety, card position,

level of congruence and emotions). Given the small number

of congruent faces presented, it was not possible to include

hemiface as an independent variable due to a large number of

empty cells in the analysis. The performance on congruent

chimeric faces was predicted to mirror the findings of the

incongruent faces, simply because they would be easier

stimuli to respond to.

Results of this analysis revealed only a significant 4-

way interaction involving anxiety, card position, congruence

and sex, F(2,92) = 3.25, ( < .04. The complexity of this

interaction was difficult to graphically display. It was

decided to create a difference score between anxious and


non-anxious conditions to allow for easier understanding of

the nature of the interaction. Using the difference score

as the dependent variable, the 3-way interaction of card

position by congruence by sex, was still significant F(2,92)

= 3.09, p < .05. Comparisons were carried out separately,

by sex, holding card position constant and comparing the

mean difference in hits for incongruent versus congruent

faces using 1-way ANOVA's. These analyses revealed a

significant effect for change in accuracy as a function of

congruency for females only and only in the right visual-

field card position, F(1,27) = 10.54, E < .003 (see Figure


The last analysis was conducted utilizing average

reaction time on hits as the dependent measure. The

analysis was again a 1-between (sex) and 4-within (anxiety

level, card position, congruency and emotion) ANOVA. The

only significant effect was a main effect for card position

F(2,90) = 3.34, p < .05. Paired comparison t-tests were

carried out on the following card position pairs, central

fixation-left visual-field, central fixation-right visual-

field and left-right visual-fields. The results were as

follows: t = 0.15, p < .9; t = -3.83, p < .0004 and t =-

4.31, p < .0001. These results indicate that substantial

differences exist between reaction times to congruent

chimeric faces as stimuli are shifted from central fixation

IC Incongruent chimeric
C Congruent chimeric
** p < .005








Figure 8.

3-way Interaction on Hits: Sex x Congruence x
Card Position (Collapsed Across Anxiety Levels
and Holding Card Position Constant).










to the right visual-field and between the right and left

visual fields.


The present study sought to examine the differences in

hemispheric processing of emotional chimeric faces. The

rationale for utilizing congruent faces with symmetrically

extreme representations of each emotion was to make the face

a potent example of the emotion and therefore, easier to

identify. On the other hand, incongruent faces with

competing emotional expressions might prove more difficult

to process, but in their difficulty might yield the valence

effects that are inconsistently reported in the literature.

Furthermore, if the emotional facial complexity, as

manifested by the presence of competing expressions,

elicited a valence effect, how would the emotional state of

the subject alter their processing ability?

The discussion of the data bearing on these questions

takes 2 forms. First, a general description of the complex

findings will be undertaken and secondly, an attempt will be

made to integrate the findings within a conceptual framework

of brain function. As a prelude to this discussion it must

be noted that the examination of emotion in contemporary

literature treats it as though it is an easily dissected

phenomenon. Rather, it is the culmination of biological


needs experienced and expressed on a sensory, perceptual or

autonomic level; combined with cognitive interpretation,

emotion falls under the purview of an even broader

collection of brain mechanisms. To think that summary

statements such as, "the right hemisphere is dominant for

emotion," captures the complexity of the phenomenon is, at

the very least, limited.

Preliminary Analyses

The results clearly indicate the difficulty in

uniformly manipulating subjects' mood. Even attempts at

capturing the potential differences in response to anxiety

(using the R-S scale) failed to explain why some subjects

did not become anxious. The differences in response may

have resulted from variability in the administration of the

surgency task; while not a difficult task to give, it proved

awkward for the examiner to be involved in deception. It

was also apparent that factors not directly measured, such

as self-esteem, expectancies of success, or beliefs about

psychology experiments had some impact on a subject's

performance. A third reason for the difficulty in obtaining

the desired anxiety response may have related to the various

coping strategies an individual utilizes to ward off

negative feeling states. However, the Repression-

Sensitization scores did not reflect sufficient variability

in response that would have implied different coping

mechanisms. It is unfortunate that a personality variable


such as coping tendency was either not tapped effectively by

the measure chosen or the sample did not exhibit coping

variability, since other researchers have shown this to be

an important factor in understanding the variance in

perceptual asymmetries (Wexler et al, 1986). Finally, all

self-report measures have inherent flaws based on response

bias (e.g. social desirability of answers) and the induction

of negative mood states may invoke this bias.

Nevertheless, one of the fundamental questions of the

present study necessitated culling a subgroup of

participants from the sample who did experience a shift in

mood. Interestingly, a greater number of males had to be

eliminated from the primary analyses as compared to females

(11 vs. 5). From anecdotal evidence, it appeared these

subjects became more angry with the experimenter than

concerned about their performance.

A more recent criticism of neuropsychological

methodologies investigating emotion has been that few

experiments attempt to provide multidimensional assessment

of the manipulated mood or target behavior (Bauer, 1987).

The present study did attempt to address this point by

utilizing exposure duration or thresholds as a physiological

by-product of changed emotion, in addition to self-report.

Presumably, the threshold at which a stimulus is perceived

with some degree of accuracy is a central nervous system

function. Comparing baseline thresholds with exposure


durations obtained subsequent to the anxiety induction

reflects changes in processing that have occurred as a

direct result of the change in mood. If as Heilman (1987)

asserts, 2 things are necessary for an emotional reaction,

an intact cognitive capacity or set and an appropriate level

of arousal, then a change in stimulus thresholds reflects

those changes. However, it is not clear which of these

factors it represents.

In the present study, subjects' baseline, non-anxious

thresholds were lower than in the anxious state. It took

people longer to process identical information. To be more

specific, Pribram and MacGuiness (1975) delineate several

components to attention which relate to the discussion of

discriminative stimuli such as emotion. The receptivity of

an organism to sensory stimulation is mediated by a phasic

physiologic response termed arousal. During arousal, the

cortical-reticular feedback loop acts to temporarily

decrease sensory thresholds and enhance processing. The

next phase of continued alertness is called attentiveness.

If habituation occurs, it is as a result of the

insignificance of the stimuli and no action is taken.

However, if there is some stimulus novelty or salience that

necessitates action, then sensory thresholds increase until

action on that stimulus is taken (even instances of

instructions to act, i.e. "identify the stimulus"). This

tonic physiologic response is activation. In the present


study, the resulting increase in threshold over the 2

threshold determinations suggests that anxiety has impeded

performance. The change in performance reflected in

requiring a longer exposure time to make a visual

discrimination suggests increased activation.

Primary Analyses

The primary analysis involved both accuracy and

reaction time. Both these variables capture different

dimensions of brain function and have a unique relationship

when studying emotion. For the organism to be most accurate

about a stimulus, sacrifices must be made in reaction time.

Alternatively, the necessity of quick responses (e.g. for

discriminant stimuli for approach/avoidance) sometimes

reduces accuracy. However, accuracy and reaction time are

both important to capturing the stimulus significance in

hemispheric specialization.

It appears from the present data that individuals are

more accurate in identifying emotion on the left side of the

face. This finding is across card positions (fields),

though the trend that appeared was for the information to be

reported slightly better when presented in the left visual-

field than either central fixation or the right visual-

field. However, it is a striking finding that the left

hemiface, even placed within the left visual-field has an

advantage over the right hemiface within that same field.

This contradicts the prediction that both halves of the face


presented in the field projecting to the right hemisphere

would be reported more accurately than the halves presented

at either central fixation or the right field. The possible

exception might be the left hemiface presented at central

fixation, for it too, projects to the right hemisphere.

Differences in levels of accuracy among the emotions

have been consistently reported in the literature and were

predicted in the present study (Bowers, Bauer, Coslett, &

Heilman, 1985; Thompson, 1983). The finding is also borne

out in this data. Happy affect is reported with repeatedly

greater accuracy than either sad or neutral emotion,

although sad is more clearly distinguishable from neutral.

Other investigators have noted that neutral is most often

misidentified as a negative emotion (e.g. angry or sad;

Natale, Gur, & Gur, 1983). It was not systematically

analyzed in the current experiment, but anecdotal evidence

suggests it was most often misidentified as angry. It makes

intuitive sense that if an emotional stimulus is ambiguous,

that the misidentification would be on the conservative side

for survival purposes (i.e. seen as a threat leading to

avoidance). It is interesting to note that

misidentification was not apparent during rater judgments of

the faces, only during tachistoscopic presentation

suggesting that the neutral facial expressions were

distinguishable during lengthier, foveal viewing of the

photographs. It is also curious that of the 3 emotions


presented in this study and among the array presented in

other studies (Thompson, 1983), happy expressions would

receive such preferential processing.

Turning from the main effects to a discussion of the

interactive results found for the accuracy data, the card

position by emotion interaction was addressed first. Again,

differences among the emotions were present across the card

positions. Even though it is a well-regarded finding that

all emotions are processed better in the left visual-field,

hence processed better by the right hemisphere, there are

clear differences in accuracy with happy > sad > neutral for

central fixation and surprisingly, also for the right

visual-field (left hemisphere). This suggests that one must

consider the relative strength of a hemisphere for a

specific function, much like the finding that, although the

left hemisphere is dominant for language abilities, the

right hemisphere has some rudimentary language capability

(Springer & Deutsch, 1981).

In examining the change in accurate identification for

each emotion from central fixation and across the fields

(when emotion was held constant in post-hoc comparisons; see

Figure 1), both happy and neutral expressions are readily

recognized at central fixation while sad is significantly

worse. One would expect that the 3 emotions, while having

different hit rates, would all be perceived better in the

left visual-field. This was not the case. While happy was


identified better at central fixation, there was a trend for

it to be reported better in the right, as opposed to the

left visual-field. However, the accuracy of either field

was substantially below central fixation. Similarly, there

was little difference between the fields for the report of

sad emotion, whereas the performance recorded for either

field was much better than central viewing. The greatest

disparity among emotions existed for the left visual-field.

It was predicted that the right hemisphere would be

superior in processing all emotions presented. This view is

in contrast to those investigators who hold the view that

each hemisphere is responsible for processing certain

emotions. According to this latter view, the left

hemisphere oversees positive emotions and the right

hemisphere processes negative emotion. The mechanism or

reason for such a differentiation or shared processing

ability might be that bihemispheric activation would sustain

stimulus processing efficiency (Bowers & Heilman, 1980a;

Tucker, 1981). The data from the present study (refer to

Figure 2) suggest different levels of accuracy for happy and

sad expressions as the stimuli are shifted off central

fixation to the left visual-field. For happy, there exists

a significant decrease in accuracy, while for sad there is a

significant increase in accuracy as the stimulus is moved

from the central position to the left visual-field. Then

when these emotions are presented in the right visual-field


the accuracy for happy identifications increases, while sad

remains unchanged. The most pronounced disparities among

the card positions exists for the ambiguity of neutral

facial stimuli. There is a highly significant advantage for

the left over the right visual-field in identifying neutral

emotion. If not evidence of a distinct valence effect, at

the very least, it reflects differences in processing the

various emotions and that each hemisphere may contribute to

identification. Tucker and Williamson (1984) argue that

arousal and activation systems are asymmetrically organized

in the brain and that emotion results when this reciprocal,

bottom up-top down system coincides with the differential

processing capabilities of the two cerebral hemispheres. It

is the different processing styles of the left and right

hemisphere that may contribute to emotion by half-field


Two other interactions for the accuracy data are best

discussed in the context of the more complex 3 factor

interaction of anxiety level by hemiface by emotion (see

Figures 3 and 4). It was predicted that anxiety would alter

brain function in such a way as to improve the performance

of the right hemisphere to material presented in left

visual-field. The result would be similar to the effect of

priming a hemisphere with compatible material, thereby

improving its performance. Investigators have disagreed

with this interpretation, speculating that an additional


processing load would detract from adequate performance on

hemisphere specific tasks (Tucker, 1984).

The interaction reflects the outcome of an induced mood

on brain function. Accuracy rates for emotional processing

do change indicating altered function, particularly for

negative and ambiguous stimuli. It is also the case that

the emotions are differentially processed regardless of

anxiety level. This apparently robust difference between

emotions is upheld for each hemiface, though the performance

in the left hemiface is better overall. Where anxiety

appears to have its greatest effect is on sad and neutral

emotion. Happy expressions are identified equally well in

either anxious or non-anxious states. When it comes to

negative or ambiguous stimuli, however, there is a

simultaneous shift in accuracy across the halves of the

face. With an induced anxious mood the greater left than

right hemifacial accuracy increases, while there is a

concomitant decrease in accuracy for the right hemiface (see

Figure 4).

One might argue from the threshold data that since

central nervous system arousal resulting from novel or

significant stimulation temporarily decreases sensory

thresholds that the obtained increase in thresholds with

induced anxiety reflects a changed cognitive set or it

reflects the cumulative effects of sustained arousal

resulting in activation. Activation results in the


temporary diminution of incoming sensory stimuli reflected

in increased threshold to stimulation (Gray, 1985; Watson,

Valenstein, & Heilman, 1981).

Tucker and his colleagues have shown performance

decrements on left hemisphere tasks during induced anxious

states. Similarly, in studies of various patient groups who

experience anxiety as a manifestation of their illness,

there are difficulties in processing material by the left

hemisphere (Tucker, 1984). Combining the threshold data

which implies increased activation, with the experimental

data indicating that anxiety may activate the left

hemisphere and induce a processing load resulting in

diminished performance, we can speculate about the change in

emotional processing during anxiety. For negative and

ambiguous emotional stimuli there is a selective increase in

the right hemisphere's ability to accurately process

emotions in the left hemiface, simultaneous with a decrease

for the right hemiface; therefore, it might be proposed that

anxiety induces a processing load on the left hemisphere

that slightly diminishes its processing efficiency. The

right hemisphere may, in turn, be released from the

reciprocal inhibition of the left hemisphere at the same

time increased attentional capacity is devoted to

identifying stimuli. This is all speculative particularly

in light of the marginal results obtained in the post-hoc

comparisons of the 3-way interaction when emotion is held


constant. Furthermore, it does not explain why this

difference does not exist for all emotions.

The reaction time data overlaps to some extent with the

accuracy data. Now, not only is identification more

accurate for the left hemiface, but reaction times are also

faster to the left side of the emotional face. When both

accuracy and reaction time data yield parallel results, the

effect is even more salient. It is curious to find a

possible hemispatial effect for faces that appears across

card positions. It is a further puzzle that this hemiface

effect is for the side of the face judged least emotionally

expressive (Sackeim & Gur, 1978; Borod & Caron, 1980).

It is important to define this hemispatial effect

better. Bowers and Heilman (1980b) assert that lateralized

tasks, whether tactile, visual or auditory often confound 2

factors when drawing conclusions about laterality effects.

This confound is between the side of external space a

stimulus is presented in and the "internal" space the

stimulus is presented to. External space is defined using

the body as a reference point. Therefore, right hemispace

is external space to the right of body midline, while left

hemispace is that area to the left of body midline.

Moreover, stimuli presented via tachistoscope are presented

in the left or right visual-field to the hemiretinae that

have direct connections to the contralateral hemisphere.

The fields are not synonymous with hemispace, although when


the eyes are centrally fixated, the left visual-field of a

tachistoscopic stimulus card happens to fall in the left

hemispace with respect to body midline. In the present

study, it appears that even though both hemifaces were

presented entirely within the left visual-field, so that

information is projected to the hemisphere which is

reportedly superior in processing this material, only the

left half of the face is more accurately reported. This

occurs at the other card positions as well. The unexpected

finding was that the left side of the face was also reported

better when presented in the right visual-field. It seems

to indicate a spatial effect independent of laterality

effects dictated by contralateral hemispheric pathways. The

mechanism for this is unclear.

Perhaps part of the explanation for this finding lies

in the hemispheric asymmetries present for arousal and

activation. When unilaterally brain damaged patients are

studied using manual reaction time tasks there are decreases

in reaction times on the side ipsilateral to the lesion

(Heilman & VanDenAbell, 1979). However, right hemisphere

lesions produced greater slowing bilaterally, often in

spite of smaller lesion size. In their study, Heilman and

VanDenAbell (1979) presented warning stimuli unilaterally,

via tachistoscope, to normal subjects. Warning stimuli

serve to prepare the person for responding (activation).

Stimuli presented to the right hemisphere reduced right and


left hand reaction times more than warning stimuli to the

left hemisphere reduced left hand reaction times or right

hand reaction times. The results indicated the right

hemisphere's role in activation of the individual.

In summary, the left hemiface at central fixation and

the left hemiface in the left visual-field not only project

to the hemisphere which is more adept at processing facial

information, but these stimuli are also projected to the

hemisphere most responsible for attention (arousal and

activation). Attention mediated by the right hemisphere

predominates over the left half of space. The question is

whether stimuli within the left portion of the right visual-

field also fall in this domain.

The present study randomly presented information to 3

card positions. The type of material presented "primed" the

right hemisphere for action (also shown in Suberi &

McKeever, 1977); furthermore, the mood induction seems to

have served 1 of 2 purposes: either it improved the right

hemisphere's performance by increasing some attentional

component that influenced left hemisphere activation or

anxiety specifically activates the left hemisphere and

disrupts its capacity for efficient processing. The

identification and reaction times to either hemiface

presented in the right visual-field were below the left

visual-field, but if attentional mechanisms predisposed


better processing by default, it would be to the left half

of space.

The analysis of reaction times also produced 2

significant interactions. The first, emotion by hemiface

parallels the accuracy findings, in that response time is

progressively faster from neutral, sad to happy. Ambiguous

emotion such as neutral, clearly takes the longest amount of

time to decode and even then the identification is not very

accurate (see Figure 5).

It is more difficult to describe the 4-way interaction

which was reduced by utilizing a difference score (between

subtracting reaction time performance in the non-anxious

condition from the reaction time in the anxious condition;

see Figures 6 and 7). This interaction, card position by

hemiface by sex, can best be explained by the variability in

performance by females for the right hemiface; specifically,

this difference occurs as stimuli are moved from central

fixation to the left visual-field. As a result of anxiety,

it appears that females take significantly longer to process

emotional stimuli presented in the left visual-field. This

is not the pattern for males whose higher reaction times for

the right hemiface at central fixation, progressively

decrease from the left visual-field to the right visual-


Extrapolating from these findings, if the left

hemisphere's efficiency is impeded by anxiety (as manifested


by reduced accuracy and longer reaction times) and the

effect is not a result of females responding differently to

the manipulation (there was no interaction for the

manipulation and sex) then anxiety alters the brain function

of females more than it does for males. But where within

the brain does this occur? The clue may be in which portion

of the stimuli is most effected. For females, if the right

hemiface at central fixation (projecting to the left

hemisphere) is moved across midline into the left visual-

field (projecting to the right hemisphere), reaction times

substantially increase with anxious mood. For males, the

pattern is reversed. If the notion regarding cerebral

dominance and hemispheric specialization in females is true,

namely that they are less lateralized than males (Geschwind

& Galaburda, 1985), then one would expect the differences in

processing for the left and right visual-fields to be

similar. This appears to be the case. Following from that,

there is a stronger necessity for interhemispheric

communication for males whose strongly lateralized

hemispheres need to be coordinated; consequently, there

would be less susceptibility to transient disruption. For

females, the extreme shift in reaction time is brought about

when the stimulus is shifted from the left hemisphere to the

right hemisphere. Perhaps for females, transcallosal

transmission is more easily disrupted by anxiety. However,

this still does not explain what occurs when the right