Empathic embarrassment

EMPATHIC EMBARRASSMENT:
REACTIONS TO THE EMBARRASSMENT OF ANOTHER

By

ROWLAND SPENCE MILLER

A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

1978

ACKNOWLEDGMENTS

The members of my supervisory committee are due my

special thanks. Drs. Franz Epting, Stephen LaTour, Larry

Severy, Marvin Shaw, and William Ware granted me their ad-

vice and criticism and were unfailingly helpful. I am

especially grateful to Dr. Barry R. Schlenker, committee

chairman, for his support and guidance throughout my graduate

training; he has been an exemplary teacher.

Debbie Curtin, Martin Fleet, Cyd Strauss, and Karen

Wollman helped me run the study and were enormously reliable

and enthusiastic. Their partnership is appreciated. I am

also thankful for my cronies Mark Leary, Don Forsyth, and

Teddi Walden, and for my friend Bonnie Miller; I'm glad they

were there.

TABLE OF CONTENTS

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

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

CHAPTER

I INTRODUCTION ...............................

The Presentation of Self...................
Conceptual Analyses of Embarrassment......
Experimental Studies of Embarrassament.....
The Nature of Empathy.......................
Experimental Studies of Empathy.............
Empathic Embarrassment: A Conceptualization
and Research Design...................

II METHOD..........................

Subjects........................
Procedure.......................

III RESULTS.........................

Actors' Responses..............
Observers' Responses..........
Actor-Observer Comparisons.....

IV DISCUSSION......................

APPENDICES..............................

A EXPERIMENTER'S SCRIPT...........

B INFORMED CONSENT FORM...........

C SOCIABILITY INVENTORY ITEMS................

D COOPERATION CONDITION INSTRUCTIONS.........

Page

ii

v

1

2
4
9
20
26

32

38

38
38

46

46
47
76

79

92

93

98

99
100

E INDEPENDENCE CONDITION INSTRUCTIONS .......

F COMPETITION CONDITION INSTRUCTIONS.........

G EMPATHY CONDITION INSTRUCTIONS............

H OBSERVATION CONDITION INSTRUCTIONS.........

I EMBARRASSMENT TASKS ........................

J CONTROL TASKS...................................

K ACTORS' QUESTIONNAIRE.......................

L OBSERVERS' QUESTIONNAIRE......................

REFERENCES... ................. .... ..................

BIOGRAPHICAL SKETCH .................................

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

EMPATHIC EMBARRASSMENT:
REACTIONS TO THE EMBARRASSMENT OF ANOTHER

By

Rowland Spence Miller

August 1978

Chairman: Barry R. Schlenker
Major Department: Psychology

Previous studies of embarrassment have focused on the

flustered reactions of actors who find themselves in embar-

rassing predicaments. However, recent studies of empathy

have implied that observers of another person's embarrass-

ment may react emotionally to the other's plight, sharing

the other's embarrassment even though the person's actions

do not reflect upon the observer. The present study examined

the reactions of observers to an actor's embarrassment, manip-

ulating the perceived link between actor and observer and

the observational set of the observer. Before being indi-

vidually assigned to actor or observer roles, same-sex pairs

of male and female subjects were induced to cooperate, com-

pete or maintain their independence before an experimental

confederate who served as their "audience." The observer

then watched the actor perform a number of embarrassing tasks

(e.g., sing the "Star Spangled Banner") under one of two

observational sets: observers were instructed either to con-

centrate on the actor's feelings or to watch his/her move-

ments carefully. Three indicators of the observer's reaction

were obtained--the observer's skin potential was recorded,

the observer's compliance with a request for help was meas-

ured, and the observer's self-report of embarrassment and

other emotions was secured. Measures of the actor's embar-

rassment and compliance were also obtained. One hundred

forty subjects participated in the experimental phase of the

study; 28 additional subjects formed an offset control group,

in which the actor performed unembarrassing tasks, to provide

a standard of comparison for the reactions of embarrassed

actors and their observers. Thus, with subject sex included

as a factor, the study was a 2 by 3 by 2 by 2 factorial design

including subjects' roles (actor or observer), the link be-

tween actor and observer (cooperation, competition, or inde-

pendence), the observer's instructional set (empathy or ob-

servation), and subject sex, with an offset control group.

It was expected that empathic observers would react

more strongly to the actors' embarrassment than would non-

empathic observers, and that observers in the cooperation

condition would be the most responsive to the actors' plight,

with competition condition observers showing the least re-

sponse. The data generally supported the first hypothesis,

although on some measures females seemed to be influenced

by the instructions to empathize more than males. Only par-

tial support was gained for the second hypothesis. In gen-

eral, observers in the independence condition reacted less

strongly than either cooperative or competitive observers;

the latter conditions, which formed links between actor and

observer, seemed to enhance the observers' reactions to the

actors' predicament.

It was also expected that empathic observers would re-

port significantly more embarrassment in response to the

actors' behavior than control observers would. This hypo-

thesis received strong support. The observers also reported

such feelings as sorriness and sympathy for the actor and

enjoyment of the observer role, but none of these responses

were as closely related to the measures of their physiologi-

cal reactivity as were their self-reports of embarrassment.

It appears that empathic observers were embarrassed for

the actors, and in the independence condition, where subjects

were virtual strangers to one another, this response seems

to have been an empathic embarrassment: The results suggest

that embarrassment is an omnibus phenomenon, influencing

both the actor in, and the observer of, an embarrassing in-

cident.

CHAPTER I

INTRODUCTION

Embarrassment is that uncomfortable state of mortifi-

cation, awkwardness, abashment, and chagrin which we all suf-

fer occasionally, but for which we usually seem ill-prepared.

Indeed, embarrassment usually seems to entail a surprising

violation of our expectations which generally leaves us feel-

ing ill-at-ease. When embarrassed, we may feel exposed,

inadequate, or self-conscious, and we may blush, tremble,

fumble, or stutter (Modigliani, 1968). We may feel ungrace-

ful and clumsy, and we may find ourselves without anything

to say, unable to meet the gaze of another person (Sattler,

1965). Embarrassment is, in short, a discomfiting experi-

ence, a "regrettable deviation from the normal state"(Goffman,

1956, p. 264).

The few analyses of embarrassment which have been prof-

fered thus far have predominantly focused on the embarrass-

ment of an actor who, through faux pas, mistake, or accident,

somehow endangers the carefully cultivated image of himself

he is trying to maintain in social interaction. For instance,

Modigliani (1968) characterizes embarrassment as an unplea-

sant sensation which is "generally precipitated by an aware-

ness that one has failed to demonstrate the demeanor considered

appropriate to a particular social interaction, and hence

that one is being perceived by others present as deficient--

as lacking certain collectively valued attributes"(p. 313).

As our review of the literature will demonstrate, this focus

on actors' embarrassment has been productive and has greatly

increased our understanding of the state. Nevertheless, this

focus has largely ignored another possible, very intriguing

class of embarrassment: empathic embarrassment, or embar-

rassment felt for another who is in an embarrassing predica-

ment even though that other's actions do not reflect upon

the observer and the observer's social image is not endangered.

This paper examines the possibility that such a state exists

and reports an experiment which studies its manifestations.

Toward that end, we will first survey the social interactional

framework from which embarrassment research has sprung and

then examine the existing analyses of embarrassment.

The Presentation of Self

In his insightful discussions of the form of social inter-

action, Goffman (1959, 1967) uses a dramaturgical analogy--

an "all the world's a stage" approach--in suggesting that

when an individual appears before others, he will usually be

motivated to control in some manner the impressions that the

others gain of him. For this reason, Goffman maintains that

a participant in any interaction "performs" for the others

present, his audience, trying to present himself in a socially

acceptable, or even desirable, fashion. The individual is

said to act out a "line," a consistent pattern of verbal and

nonverbal acts by which he expresses his view of the situa-

tion and informs his audience how he would like to be viewed.

A central component of a person's performance is his

"face," the positive social value of the attributes the per-

son claims to possess; it is the positivity or desirability

of the social image of himself that the person is trying to

maintain, and it depends on all the expressive information

available about a person at any given time--information that

the person is attempting to control. An actor tries to estab-

lish and maintain a single consistent face throughout a given

interaction, initially choosing a face to which he is enti-

tled, and skirting issues and aspects of interaction that

could endanger it. For example, a poor softball player will

either present himself as a mediocre athlete, or, in order

to be seen as a good athlete, will avoid softball games.

This ongoing maintenance of face, and the actor's attempts to

restore face once it has been lost, is termed "facework."

Thus, in any interaction, an actor claims a certain face

and attempts to present a consistent image of himself to the

other participants. Nevertheless, Goffman does not imply that

the presentation of self ordinarily involves deliberate dis-

simulation or attempts to deceive one's audience. Instead,

lines are played and faces are maintained so that interactants

may generally know what to expect from one another; the goals

of both actor and audience are to avoid disruption and to

maintain smooth, predictable interaction. Indeed, actor and

audience are thought to collaborate toward that end, over-

looking small inconsistencies in self-presentation, and allow-

ing one another the benefit of the doubt.

Often, however, despite the expertise and intentions of

the participants, an interaction does not proceed smoothly;

some mishap occurs which endangers the face of one or many

of the participants. When an interaction is disrupted for

any reason and a participant must abruptly concern himself

with the restoration of his face, embarrassment may result.

A number of authors have contributed conceptual analyses of

embarrassment to our understanding of the state; generally,

all of them characterize it as the aversive, flustered, emo-

tional state described at the beginning of this chapter, but

they differ somewhat in their discussions of its causes and

classifications.

Conceptual Analyses of Embarrassment

In his essay on the subject, Goffman (1956) treats em-

barrassment as a pervasive phenomenon which results when

unfulfilled expectations discredit one's claims to an accept-

able face, that is, when "the expressive facts at hand threaten

or discredit the assumptions a participant finds he has pro-

jected about his identity"(p. 269). Goffman suggests that

embarrassment is possible in any encounter and that it may

range from abrupt, acute discomfort to sustained unease. Its

crucial aspect is the individual's concern for the impression

he is projecting to the present audience.1 Thus, Goffman

1. Goffman also mentions an "instrumental chagrin," a
flustered discomfort an individual may suffer when he finds

considers embarrassment to be situationally specific; if

one's audience changes, an embarrassing incident should be

forgotten. Indeed, Goffman believes that embarrassment may

sometimes be so intense and so disruptive that in order to

regain one's poise one must end the encounter and flee one's

audience.

Gross and Stone (1964) define embarrassment in a man-

ner similar to Goffman, but are more detailed in their dis-

cussion of its possible causes. Embarrassment, they write,

"occurs whenever some central assumption in a transaction

has been unexpectedly and unqualifiedly discredited for at

least one participant. The result is that he is incapacitated

for continued role performance"(p. 2). Their analysis of

over 1,000 anecdotal instances of such embarrassment garnered

from colleagues and students led Gross and Stone to suggest

that embarrassing incidents are generally of three types.

First, embarrassment may result when an interactant uses an

inappropriate identity; this may occur when a) one's iden-

tity is undocumented, as then a person realizes, having com-

pleted a meal at a restaurant, that he has left his wallet

at home, b) one misplaces an identity, misnaming or forget-

ting another person, or c) one misplays an identity, allow-

ing other incompatible roles to intrude on the role one is

playing.

himself unable to perform a task vital to his long-term wel-
fare. This "chagrin" may be a feeling much like embarrassment,
but its causes are quite different; whereas embarrassment
depends on the image one maintains "before others felt to be
there at the time (p. 264)," instrumental chagrin is task-
related and may occur whether others are felt to be present or
not.

Second, a loss of poise, the control of one's self and

one's situation may be embarrassing; this may occur when one

loses command of one's a) territory, as when another person

inadvertently invades one's dressing area; b) equipment, as

when one's car stalls in the middle of a busy intersection;

c) clothes, as when one rips one's pants; or d) body, as when

one is noticeably flatulent.

Finally, when an interactant's confidence in his assump-

tions about others is shaken, the disconfirmation of his ex-

pectations may prove embarrassing. This possibility has led

to the development of performance norms that allow others

flexibility in their roles and that generally give them "the

benefit of the doubt" in order to minimize the number of

embarrassing surprises that can result from another's behavior.

In general, Gross and Stone consider the momentary break-

downs in interaction caused by embarrassing incidents to be

central to our understanding of social relationships. Like

Goffman, they suggest that actor and audience work together

to avoid embarrassment whenever possible; "provisions for

the avoidance or prevention of embarrassment, or quick re-

covery from embarrassment when it does occur are of key im-

portance to any society or social transaction, and devices to

insure the avoidance and minimization of embarrassment will

be part of every persisting social relationship" (p. 15).

In his discussion of embarrassment Weinberg (1968) ex-

tends Gross and Stone's formulation, accepting their tripar-

tite classification but suggesting that two more fundamental

dimensions also underlie any embarrassing incident. For

Weinberg, the intended or unintended nature of one's act and

the correctness or incorrectness of one's definition of the

situation are independent dimensions which together describe

four elementary forms of embarrassment (which may result in

the loss of identity, poise, or confidence detailed by Gross

and Stone).

The first form of embarrassment involves situations in

which one's intentional behavior is defined post facto as

inappropriate to the social situation. These are faux pas,

as when one arrives at a party decidedly underdressed; one

has intended to dress casually, and yet, upon arriving, one

finds that one's definition of the situation was incorrect.

A second form of embarrassment may occur when one's expecta-

tions are correct but they are breached by an unintended act.

Such an event would be an accident; one may share others'

conceptions of appropriate behavior but still disrupt the

situation by an unmeant act such as spilling coffee in one's

lap. Embarrassment may also result when one's incorrect

definition of the situation leads to an unintended act.

This third form of embarrassment is characterized by mistakes,

as when a person incorrectly believes that his zipper is zipped

and walks around all day with it open. Finally, even situa-

tions in which one's act is intended and one's definition of

the situation is correct may cause embarrassment, although

Weinberg suggests that such embarrassment is evoked only by

one's "internal" audience, not by the reactions of others.

This may occur when one must perform duties which may be per-

sonally embarrassing but which do not endanger one's face,

such as a young girl submitting to her first gynecological

examination.

At this point it should be noted that neither Weinberg

nor Gross and Stone fully explicate whose actions, whether

his own or someone else's, a person may find embarrassing.

Gross and Stone suggest that others may sometimes cause a

person's loss of identity, poise, or confidence, but they do

not elaborate, and Weinberg considers only those situations

in which a person's own actions are embarrassing to him.

However; another classificatory scheme for embarrassing in-

cidents is suggested by Sattler (1965), who explicitly takes

an actor or observer position into account.

After collecting over 3,000 instances of recalled em-

barrassment from 301 subjects, Sattler found that they could

best be organized by sorting them into one of three main

groups, situations in which the embarrassed person is a) an

agent in, b) the recipient of, or c) an observer of an em-

barrassing act. When the person (P) is an agent, his own

actions prove embarrassing to him; he may, for instance,

dress inappropriately, forget a name, be awkward or clumsy,

or make a slip of the tongue. By contrast, when P is a recip-

ient another person(O) with whom P is interacting does some-

thing which embarrasses P; 0 may criticize, praise, or tease

P, invade P's privacy, or expose P's lack of knowledge or

ability.

These categories of embarrassing incidents are plausible

and straightforward, and readily compatible with Gross and

Stone's conception. However, Sattler's third category de-

scribes embarrassing situations which are not explicitly

discussed by Weinberg or Gross and Stone. In these situations,

P is an observer of 0 and is not necessarily interacting

with him, but is embarrassed by O's actions nevertheless.

This may occur either because 0 does something which reflects

on P, as a child's improprieties reflect on his parents, or

because P is simply embarrassed for 0, sharing O's reaction

to O's embarrassing experience even though his actions are

not in any way connected with P.

This mention of apparently empathic embarrassment by

Sattler is noteworthy, and our discussion will return to it.

Now, however, we must demonstrate that, as well as being con-

ceptually analyzable into categorical schemes, embarrassment

has recognizable behavioral effects.

Experimental Studies of Embarrassment

Most of the experimental investigations of embarrassment

have been prompted by Goffman's provocative assumptions that

embarrassment is situationally specific, that interactants

are motivated to avoid it, and that facework is often needed

to recover from it. For instance, Brown (1970) and Brown

and Garland (1971; Garland & Brown, 1972) have conducted a

number of studies which show that subjects are likely to

forego tangible profits in order to avoid public embarrass-

ment. In Brown (1970), subjects performed an embarrassing

or nonembarrassing task--either sucking on a rubber pacifier

or touching with their hands a small rubber figure--and then

chose between (1) maximizing the money paid them by describ-

ing their actions to some of their classmates, or (2) accept-

ing smaller payoffs to avoid public exposure. Embarrassed

subjects sacrificed more money and retained more privacy than

did unembarrassed subjects, especially when their audience

was unaware of their costs for doing so. Moreover, a second

experiment showed that embarrassed subjects sacrifice more

money when confronted with an audience that is described as

"evaluative" than when confronted with a "nonevaluative"

audience. As Brown concluded, people apparently tend to

engage in "costly face-saving behavior"--in this case, the

avoidance of public exposure--after a potentially embarrassing

incident.

Brown and Garland (1971) extended Brown's work, conduct-

ing two studies which again investigated the sacrifice of

monetary rewards in order to preserve face. In both studies

subjects received a computerized evaluation of their singing

ability which judged them as either competent or incompetent,

and then sang before an audience; the longer they sang, the

greater their cash reward. The first study showed that "in-

competent" subjects--who reported greater embarrassment--sang

for a shorter time than did "competent" subjects. The second,

which manipulated the degree of acquaintanceship existing

between "incompetent" singers and their audiences, found that

embarrassment avoidance was greater--subjects sang for a

shorter time--before close friends than before acquaintances

or strangers. Friendship did not seem to reduce the motive

to protect one's face. A similar subsequent study by Garland

and Brown (1972) demonstrated that women tended to avoid

embarrassment to a greater extent than men and that avoidance

was greater before a judgmental expert audience, one supposed-

ly composed of "excellent" singers, than before an inexpert

audience. Thus, Brown's and Brown and Garland's work tends

to confirm one of Goffman's primary hypotheses, that embar-

rassment is an aversive state that one will avoid if possible,

even at cost to oneself.

Other studies have examined the reactions of actor and

audience to embarrassment which has already occurred, under-

mining at least one participant's face and disrupting the

smooth interactional flow. In these situations, Goffman

(1967) predicts that remedial steps will be taken by all

interactants to patch up the interaction, bypassing the dis-

ruption and attempting to restore lost face to those who need

it. The audience is expected to aid the hapless actor by

minimizing his miscue and allowing him to repair his damaged

image, and two observational studies suggest that one man-

ner in which this is accomplished is to treat an embarrassing

incident as a humorous event. Both Coser (1960) and Oleson

and Whittaker (1966) have found that an audience often re-

sponds to a person's embarrassment with laughter, and, inter-

estingly, this response helps rather than hinders the inter-

action. Laughter demonstrates that the incident is one to be

taken lightly and quickly forgotten, instead of being a seri-

ous matter that will demand future attention. Laughter "com-

bines criticism with support, acceptance with rejection"

(Coser, p. 91), informing the actor that he should be embar-

rassed but that the consequences will not be severe. Thus,

it yields "a type of correction for the person even as the

individual and situation are saved"(Oleson & Whittaker, p.388),

and, "importantly, it allows the onward flow of interaction"

(p. 388). Thus Goffman's assumptions again seem reasonable;

audiences do often seem to humorously accept an actor's im-

proprieties, glossing over incidents so that interaction can

continue.

What of the embarrassed actor? Goffman suggests that

he is motivated to redeem his public image and account for

the incident through facework, and the evidence supports

that assumption. For instance, Brown (1968) found that sub-

jects who had been exploited by an opponent in a bargaining

game, and then embarrassed by an evaluation from their audi-

ence that told them they looked foolish and weak, were much

more likely than unembarrassed subjects (who had been told

they looked good for playing fair) to sacrifice monetary

reward in order to retaliate against their opponents. Having

been humiliated, the subjects apparently ignored their own

best interests in order to reassert their strength for their

audiences.

Similarly, Modigliani (1971) found that embarrassed

subjects who had publicly failed their portion of a group

task made more image-enhancing statements to their fellow

group members than did unembarrassed subjects; they tended

to minimize their failures, derogate the task, excuse their

performances, describe other abilities, or defensively change

the subject. Thus, embarrassment usually seems to involve

some attempt by the actor to restore his deficient image to

assure those present that he is not generally as awkward or

incapable as he may have momentarily appeared.

However, another of Goffman's assumptions holds that

such actions by embarrassed actors are ordinarily directed

only at the audience which witnesses the embarrassing inci-

dent; that is, Goffman believes that embarrassment is speci-

fic to and wholly contained within a given situation and that

it should only influence one's behavior toward those who may

be aware of one's embarrassment--one's other images in other

situations should be unaffected. Apsler (1975) tested this

assumption, examining (unlike Brown [1968] and Modigliani

[1971]) the behavior of embarrassed subjects toward others

who were unaware of the embarrassing incident, and his results

provide little support for Goffman's position. Apsler embar-

rassed half of his subjects by having them perform tasks which

made them appear foolish--for example, laughing for 30 sec-

onds as if they had just heard a joke, and imitating a five-

year-old having a temper tantrum--in front of a peer obser-

ver/confederate. Other subjects performed unembarrassing

tasks such as counting silently to 50. Then, either the

observer or another confederate who ostensibly was completely

unaware of the embarrassing performances privately asked

the subject for help with a class project. Embarrassed sub-

jects complied more than did unembarrassed subjects, regard-

less of the source of the request, and embarrassed subjects

agreed to help both an observer and nonobserver of their

embarrassment equally. The results undermine Goffman's

assumption that embarrassment is situationally specific, sug-

gesting instead that embarrassment creates a general discom-

fort or concern for face that embarrassed individuals then

attempt to relieve, regardless of audience. Apsler's embar-

rassed subjects seemed to seek the image-enhancing experience

of helping someone, anyone, whether the person had witnessed

their embarrassment or not.

Apsler's results also run counter to the formulation of

embarrassment proffered by Modigliani (1968, 1971). Like

Goffman, Modigliani considers it a delimited situational

phenomenon. He stresses, for example, that a "failure in

self-presentation . does not undermine the individual's

general identity; rather it descredits a much more restricted

situational identity which he is projecting into the current

interaction' (1968, p. 315). Moreover, Modigliani suggests

that embarrassment results from the individual's belief that

others in his immediate presence perceive his situational

face to be deficient. Thus, Modigliani's position is also

weakened by Apsler's findings that remedial facework follow-

ing embarrassment is not directed only toward those who are

aware of the incident.

However, Modigliani's formulation serves to emphasize

another testable assumption of Goffman--that some external

audience must be aware of the incident for the actor to suf-

fer embarrassment. Recall that Weinberg (1968) suggests

that an internal, private form of embarrassment is possible;

his conceptual scheme allows for embarrassment even when

there is no overt deficiency in one's self-presentation. By

contrast, Modigliani's position stresses that an assumed loss

of face in the eyes of others is a necessary precondition

for the occurrence of embarrassment, and that, simply, "there

is no such thing as 'private embarrassment'" (Modigliani, 1971,

p. 16). 1

Unfortunately, the only experimental evidence concern-

ing this assumption is equivocal. Modigliani (1971) measured

what he believed to be private embarrassment in subjects who

had privately failed their portion of a group task and found

that their level of reported embarrassment was intermediate

to--and significantly different from--both that of embarrassed

subjects who had failed publicly and that of unembarrassed

subjects who had privately succeeded. The private failure

subjects reported less embarrassment than did those who had

publicly bungled their assignments, but more embarrassment

1. Goffman does suggest that embarrassment may occur
in the "imagined presence of others (1956, p. 264)," implying
that an audience may not actually have to be present for one
to suffer embarrassment. Nevertheless, Goffman holds that
in order to be embarrassed one must believe that others are
present and that one's self appears deficient to them, and
that is quite different from a "private" embarrassment which
would not depend on one's self-presentations to others.

than did those who had not failed at all. Thus, at first

glance, Modigliani's results suggest the existence of a mild

form of embarrassment which exists even in the absence of an

audience. However, the private failure subjects expected

that some of their subsequent performances would be public,

and their expectation of an audience's scrutiny may have led

to an anticipatory embarrassment which was not truly "pri-

vate" at all. Still, whatever the case, it is noteworthy

that no actual breakdown in interaction occurred to produce

the subjects' embarrassment. Instead, their mild discomfort

seemed to result from some cognitive process; as Modigliani

suggests, "private-failure subjects allowed their sense of

self-deficiency to produce an imagined sense of social dis-

approval" (p. 22).

This explicit recognition of a cognitive role in embar-

rassment by Modigliani and Weinberg and the possibility

that an audience is not necessary for embarrassment to occur

are both intriguing and intuitively plausible. Consider, for

instance, the plight of this writer to the popular advisor-

columnist Abigail Van Buren:

Dear Abby:
Recently I moved into a small apartment build-
ing with paper thin walls.
A male tenant (single) lives next door. Every-
thing he does--and I do mean EVERYTHING--can be
heard through the walls.
Late at night, and especially on weekends,
he carries on a very noisy love life.
I am not an eavesdropper. What he does is
his own business, but how do I keep his private
life from ruining my sleep and embarrassing me and
my guests?

I have met him only once, and he seems nice.
For that reason I am unable to bring myself to tell
him that I can overhear everything he does.
Is there some way I can let him know that he
is disturbing me and embarrassing me?
The Girl Next Door (1978, p. 8B)

When guests are present and her neighbor's actions intrude

upon her territory and the decorum she is trying to maintain,

the writer's loss of poise causes her to be publicly embar-

rassed. However, she is probably also embarrassed in some

manner even when she is alone and her neighbor's maneuvers

are intrusively apparent; though no audience is present, the

very fact that the situation would be embarrassing to all

participants if the neighbor knew he had been or was being

overheard may be enough to induce a private form of embar-

rassment in the writer. Indeed, many situations involve an

"if they only knew" quality in which an actor anxiously at-

tempts to keep from his audience's notice some past or pre-

sent deficiency which would surely be embarrassing if brought

to their attention, and the fear of discovery in such situa-

tions is a discomfort akin to embarrassment, if not embar-

rassment itself. The present discussion is meant to be

speculative, but it may be proper to suggest that whether or

not "private embarrassment" actually exists--and this may be

in part a definitional problem--private events often involve

a fear of embarrassment that closely relate them to "public"

embarrassments.

Two remaining studies involving embarrassment deserve

mention. First, Buck, Parke, and Buck (1970) have shown

that there are recognizable patterns of physiological arousal

which accompany embarrassment. They compared manipulated

states of fear and embarrassment in their subjects and found

that the two states could be easily distinguished. In embar-

rassed subjects who had been asked to suck on infantile ob-

jects such as a baby bottle, pacifier, and breast shield,

there was an increase in skin conductance, a deceleration

in heart rate, and an increase in subjects' looking around

the room. By contrast, fearful subjects who had been threat-

ened with shock evidenced a larger increase in skin conduc-

tance, a higher peak heart rate, and a decrease in looking

around. As well as a state of psychological discomfort,

embarrassment appears to be a distinct physiological state.

Finally, Modigliani (1968) has suggested that a trait of

embarrassability exists, being a person's general suscepti-

bility to embarrassment, and he has developed a scale to

measure it. His Embarrassability Scale consists of 26 items,

each describing a potentially embarrassing situation, and

respondents are asked to rate how embarrassed they would be

in each situation. Modigliani's results show that the Scale

is internally consistent and reliable and that it possesses

moderate predictive validity. Moreover, its correlations

with other scales suggest that high embarrassability results

when two different traits are present simultaneously: first,

high sensitivity to the evaluations of others, and second,

a tendency to believe that these evaluations are more nega-

tive than they actually are. A person possessing such traits,

says Modigliani, is particularly prone to suffer embarrass-

ment when his self-presentation is deficient.

Factor analysis of the Embarrassability Scale indicated

that it includes five distinct classes of embarrassing situ-

ations:

(1) Situations in which a person discredits his
own self-presentation through some inadvertent
foolishness or impropriety (e.g., tripping and
falling in a public place); (2) situations in which
a person finds himself unable to respond adequately
to an unexpected event which threatens to impede
the smooth flow of interaction (e.g., having atten-
tion drawn suddenly to some physical stigma of a co-
actor); (3) situations in which a person loses
control over his self-presentation through being
the center of attention without having any well
defined role (e.g., being the focal point of "Happy
Birthday to You"); (4) situations involving empathic
embarrassment wherein a person observes another
individual who is in a seemingly embarrassing pre-
dicament (e.g., watching an ineffectual comedian
on an amateur show); (5) situations in which an
individual is involved in an incident having inap-
propriate sexual connotations (e.g., walking into
a bathroom occupied by a person of the opposite
sex). (Modigliani, 1968, p. 319)

Situations like these are recognizable to all of us,

and they serve to exemplify and illustrate what we know about

an actor's embarrassment: that it is an aversive state of

both physiological and psychological arousal, characterized

by facework directed toward any available audience, which

may result from pratfalls, faux pas, accidents, mistakes,

or, in general, any incident which disconfirms the actor's

expectations of a smooth and orderly interaction.

However, our discussion still has not yet adequately

considered the embarrassment suffered by non-acting observers

of an embarrassing incident--the empathic embarrassment men-

tioned by Sattler (1965) and Modigliani (1968). Is embar-

rassment necessarily limited to the actor who blunders or

20

in some way botches his self-presentation? Or do other inter-

actants or even mere observers sometimes share his embarrass-

ment? Consideration of these questions demands an understand-

ing of the nature of empathy, and it is to this topic that

we now turn.

The Nature of Empathy

We often share in some manner the feelings of others

around us. We may wince with another when he receives a

shot, we may be saddened by the tears of a parent who is weep-

ing for a drowned child, or we may be cheered by the jubi-

lant celebration of a victorious championship team. Our re-

sponses in these situations are empathic; the others' feel-

ings are influencing our own. This does not mean that we nec-

essarily feel sorrow for the others or that we feel impelled

to help them; to the contrary, we may be motivated to avoid

those who make us sad. Moreover, another's emotion may not

instill the same emotion in us--a sadist may feel joy at

another's pain. Empathy means only that we experience an

emotion in response to another's emotion. To be specific,

Stotland (1969; Stotland, Sherman & Shaver, 1971) defines

empathy as "an observer's reacting emotionally because he

perceives that another is experiencing or is about to experi-

ence an emotion" (1969, p. 272).

There are a number of facets of this definition of em-

pathy that deserve comment. First, empathy obviously deals

with emotion. However, the quantification and measurement

of emotion has been problematic for social psychologists.

It is generally accepted that emotion is comprised of both

physiological and subjective components (Schachter, 1964),

but it is not always possible to obtain adequate measures of

both, and there may be broad individual differences in auto-

nomic responsiveness to (cf. Lacey, 1950), and cognitive

labelling of, emotion. Therefore, most studies involving

empathy have not tried to measure it directly but have instead

examined the effects of instructions to empathize on subse-

quent behavior.

Second, a distinction must be made between this concep-

tualization of empathy and that of many other researchers

(cf. Tagiuri, 1969) who are concerned with an individual's

ability to accurately assess and predict the behavior of

another person. This sensitivity to others, termed "predic-

tive empathy" by Stotland, is only a nonessential portion of

a truly empathic response. Empathy does involve an observer's

perception of another's emotions, but as Stotland et al.

(1971) argue, "whether his perception is accurate or not is

a secondary matter; what is important is that the observer

will respond to the other's experiences as he perceives them.

Thus, if there are cues which suggest a feeling, it is possi-

ble for an observer to empathize with emotions which really

do not exist"(p. 6). Predictive accuracy, then, may influ-

ence the quality and quantity of an empathic response, but

it is not empathy itself.

Third, empathy is epitomized by instances of emotional

response to the emotions of those whose experiences have no

impact on the observer's welfare. Cases in which the other's

experiences and outcomes are somehow linked to the observer's

own are of lesser interest, for the other's emotions may

serve as a conditioned signal that the observer himself is

about to be rewarded or punished. Under such conditions,

the observer's resulting emotion may not be simply empathic

but may also constitute an anticipatory reaction to the ex-

pected reward or punishment. If one's boss arrives at work

in a mean, sour mood, for instance, one's subsequent cranki-

ness may not be empathic at all, but the mere reflection of

one's expectation of a long, hard day. Experimental studies

which have stemmed from Stotland's work have generally avoided

this confounding by examining subjects' empathy with strangers.

Fourth, a distinction should also be made between "sym-

pathy" and empathy. As it is generally understood, there

seems to be little difference between sympathy and the con-

ception of empathy being developed here; for instance, the

American Heritage Dictionary (Morris, 1970) defines sympathy

as the "act of or capacity for sharing . the feelings

of another person"(p. 1303). Given this apparent similarity,

further elaboration of our definition of empathy is evidently

needed.

The work of philosopher Max Scheler is helpful in this

regard. In nis essay The Nature of Sympathy (1954) Scheler

distinguishes between "fellow-feeling 'about something';

rejoicing in his joy and commiseration with his sorrow"(p. 12)

and "true emotional identification." Fellow-feeling, suggests

Scheler,

. involves intentional reference of the feel-
ing of joy or sorrow to the other person's experi-
ence. It points this way simply qua feeling--there
is no need of any prior judgment or intimation
"that the other person is in trouble"; nor does
it arise only upon sight of the other's grief, for
it can also "envisage" such grief . (p. 13)

That is, fellow-feeling is any emotional reaction to the

perceived experiences of another--it may even produce an emo-

tion in the observer which is opposite that thought to be

felt by the other. At first glance, this is a definition

much like that we have employed for empathy. However, Scheler

also speaks of emotional identification, a sense of emotional

unity; it differs from fellow-feeling in that "here it is

not only the separate process of feeling in another that is

unconsciously taken as one's own, but his self that is iden-

tified with on's own self. Here too, the identification is

as involuntary as it is unconscious"(p. 18). This dimen-

sion of identification, the unknowing apprehension that some

similarity exists between another's attributes or situation

and one's own, may also differentiate sympathy and empathy.

For the purposes of this paper, let us equate sympathy

with Scheler's "fellow-feeling" and empathy with his "emo-

tional identification." That is, let us suggest that an

observer may sympathize with anyone; another person's joy

or pain may cause an observer to experience some emotion no

matter how great the dissimilarities between them or their

circumstances. By contrast, one may empathize only with another

who is felt to share some similarities with the observer,

some common ground which would enable the observer to feel--

24

perhaps unconsciously--that similar events could befall him.

(Indeed, as we shall see, empathy may be manipulated by vary-

ing the perceived similarity of a target person to the ob-

server.) It might be said then that a man may sympathize

with a woman who is giving birth, but he cannot empathize

with her. Of course, it may be that the difference between

sympathy and empathy is a matter of degree, and practically,

it may be impossible to distinguish between them. Still,

the conceptual clarification may prove useful in the discus-

sion ahead.

A personal note from the author which is germane to

this distinction seems justifiable here. My interest in the

notion of empathic embarrassment stems from my own reaction

to a televised film, Save the Tiger. In it, Jack Lemmon

plays a clothing manufacturer who, in one scene, suffers

hallucinations while he is addressing a convention of buyers.

His delusion causes him to see his audience as a collection

of skeletal war dead, and as a result, he is totally inca-

pacitated; his presentation degenerates into chaos. My reac-

tion to this scene was one of acute discomfort. I did not

feel sorry for the character, or pity him; instead, I was

tormented by the plight of the character. My feelings were

much like the anguish I might feel if I found that I had done

such a thing. I did not envision myself hallucinating before

an audience, but it did seem conceivable that intrinsically

similar embarrassment could befall me. Thus, I believe I

empathized with the character and that my reaction was one

of empathic embarrassment.

25

My reaction might have been much different if the char-

acter or his situation had been totally foreign to me. In

fact, if I had been unable to identify with the character,

unable to empathize with him, my reaction might not have been

one of embarrassment at all. It may be, then, that one can

only be embarrassed by another's experiences when one is able

to envision oneself in a similar situation. There may be

no such thing as "sympathetic embarrassment."

Finally, one may wonder at the origin of empathic reac-

tions. Why should the emotions of strangers--or even fic-

tional film characters (Tannenbaum & Gaer, 1965)--affect us?

A Jungian psychologist may suggest that empathy results from

the deep subconscious archetypal bonds that unite humanity,

while a defensive attribution theorist (cf. Shaver, 1970)

may argue that empathic emotions reflect our desire or fear

that similar events will happen to us. A more parsimonious

explanation, however, is that espoused by Hoffman (1975) who

suggests that empathy is a classically conditioned response:

A simple example is the child who cuts himself,
feels the pain, and cries. Later, on seeing another
child cut himself and cry, the sight of the blood,
the sound of the cry, or any other distress cue or
aspect of the situation having elements in common
with his own prior pain experience can now elicit
the unpleasant affect initially associated with that
experience. (pp. 613-614)

A learning process seems plausible, but there may be an in-

born component to empathy as well; Sagi and Hoffman (1976)

have shown that one-day-old infants respond to the sound of

another baby's crying by crying themselves! It appears that

for now, at least, the question of the origin of empathy

must remain open.

Experimental Studies of Empathy

Stotland has shown that empathy can be manipulated in

the laboratory by two different means. The first of these,

which is based on the assumption that empathy is in part a

cognitive, symbolic process, involves the type of instruc-

tions given subjects. Stotland has found that if subjects

are asked to concentrate on another person's feelings, imag-

ining how he feels or imagining how they would feel in his

place, they will empathize more than if they are merely asked

to watch the other, carefully observing his movements. The

instructional focus on the other's feelings apparently en-

hances their salience, facilitating empathic responses.

An experiment by Stotland and Sherman (reported in

Stotland, 1969) illustrates this method. In this study, a

group of subjects watched a confederate--who had ostensibly

been randomly selected from the group--receive painful, pleas-

urable, or neutral heat treatments from a diathermy machine.

The subjects were individually assigned to one of three ob-

servational-instructions groups. "Imagine-Self" subjects

were told to concentrate on how they would feel if they were

receiving the heat treatments, imagining themselves in the

confederate's position. "Imagine-Him" subjects were told

to concentrate on the confederate's feelings, imagining the

sensations he was experiencing. Finally, "Watch-Him" sub-

jects were told to carefully observe the confederate's

27

movements, watching everything he did. Stotland and Sherman

obtained measures of the subjects' palmar sweating, vasocon-

striction, and subjective feelings, and while the relation

among the measures was rather variable, their general pat-

tern indicated that subjects were more affected by the con-

federate's experience in the Imagine-Self and Imagine-Him

conditions than in the Watch-Him condition. Subjects reacted

more, physically and subjectively, when told to concentrate

on feelings rather than movements, and their perceptual set

seemed to determine the extent to which they empathized with

the confederate.

The second means of manipulating empathy is to vary the

perceived similarity of the target person to the observer,

a method exemplified in a study by Krebs (1975). In this

study, subjects attended a preliminary session to complete

a number of personality inventories. Then, upon arriving for

the experimental session, they were introduced to a confed-

erate who, according to their supposed test scores and major

interests, was either quite similar or dissimilar to them.

Electrodes were attached to measure subjects' skin conduc-

tance, vasoconstriction, and heart rate. Thereafter, subjects

observed the confederate as he performed a task, either an

innocuous cognitive and motor skill test or a roulette game

in which he alternately received a cash reward or suffered

a painful shock. Finally, on the last trial of the task,

subjects were required to determine the amount of reward or

shock the confederate could receive in a manner that amounted

to either helping themselves at a cost to the confederate

(they would get the money and he would get the shock) or

helping the confederate at a cost to themselves.

Subjects who watched the confederate perform the "high

affect" task at which he experienced pleasure and pain ex-

hibited greater psychophysiological reactions than did sub-

jects who observed the "low affect" performance, and subjects

who believed they were similar to the confederate tended to

react more strongly than those who did not. Reactions were

particularly pronounced when subjects observed a similar

confederate at the high affect task; significant changes in

all three physiological measures were obtained (although

skin conductance appeared to be the most consistent and re-

liable indicator of arousal). Moreover, subjects in the

similarity condition reported identifying with the confederate

the most and feeling the worst when he waited for shocks.

Krebs concluded that the similarity subjects empathized more

than did the dissimilarity subjects; he suggested that "the

perception of similarity increases the disposition to imagine

how one would feel in another's place and that the disposi-

tion mediates vicariously experienced emotional arousal"

(P. 1143).

Krebs' study thus demonstrated the efficacy of perceived

similarity as a manipulation of empathy. We have yet to

mention, however, his results regarding the last task trial,

which was a measure of subjects' altruism. The subjects who

reacted most empathically also behaved most altruistically;

that is, "subjects who experienced the strongest empathic

reactions toward another were most willing to help him, even

though it meant jeopardizing their own welfare"(p. 1144).

On the basis of this finding, Krebs suggested that many altru-

istic behaviors may be mediated by empathy; if another's

pleasure or pain becomes tied to one's own affect, it may

often be in one's best interest to help the other, even at

apparent cost to oneself.

Others have tied empathy to altruism, among them Aderman

and Berkowitz (1970). Although the design of this study was

less rigorous than that of Krebs, the results provided some

support for the thesis that empathic reactions can mediate

help-giving. Subjects in the study listened to a taped con-

versation under instructions to empathize with one of the

characters. In some cases, subjects concentrated on a person

in need of help, and in others, on a potential helper. Fur-

thermore, the outcome of the conversation was varied so that

no help was given, help was given but no thanks were received,

or help and thanks were exchanged. Following the tape, sub-

jects' moods were assessed and they were given an opportunity

to altruistically help the experimenter. The results showed

that subjects empathizing either with the unaided person in

need or the thanked helper were most helpful, but the moods

of those two groups of subjects were rather different; those

concerned with the unaided person were somewhat angry and

bitter while those concentrating on the thanked helper were

more happy and pleased. It appeared that the subjects were

experiencing emotional reactions similar to those they be-

lieved the characters were feeling, and that empathy was

motivating their helping behavior.

Aderman, Brehm, and Katz (1974) have also related em-

pathy to the "just world" evaluation phenomenon described by

Lerner and Simmons (1966). Noting that judgmental attention--

the "Watch-Him" condition of Stotland and Sherman (see

Stotland, 1969)--may interfere with empathy, Aderman et al.

suggested that Lerner and Simmons' instructions to their

subjects may have made them less likely to empathize, result-

ing in their derogation of an innocent victim. Aderman et

al. repeated the Lerner and Simmons procedure, having sub-

jects view a videotape of a female "victim" receiving shock

in a learning experiment. However, a third of the subjects

watched the videotape after "Imagine-Self" instructions to

empathize with the learner, another third received "Watch-

Her" instructions to merely observe her carefully, while the

remaining subjects received Lerner and Simmons' original

"watch closely" instructions. Both the "Watch-Her" and "watch

closely" instructions caused subjects to derogate the victim,

rating her lowly on a number of adjective pairs. "Imagine-

Self" subjects, however, praised the victim, rating her quite

highly, and Aderman et al. decided that "whether observers

react to an innocent victim with compassion or rejection

depends on their observational set"(p. 346). The "just

world" phenomenon can be best observed, it seems, when instruc-

tions to the evaluators tend to inhibit empathy with the victim.

Empathy with another may also influence an observer's

attributions for the other's behavior. As is well known

(see Jones and Nisbett, 1971), actors and observers tend to

differ in their perceptions of the causes of behavior; actors

generally tend to attribute their behavior to external,

situational causes, while observers tend to attribute the

same behavior to internal dispositions of the actor. However,

as might be expected from Jones and Nisbett's suggestion that

actors and observers tend to process information differently,

focusing on different aspects of the situation, empathy with

an actor can change an observer's perspective. A number of

studies have shown that empathizing observers, like the actors

themselves, tend to make situational attributions for the

actor's behavior. Regan and Totten (1975), Brehm and Aderman

(1977), and Gould and Sigall (1977) have all given subjects

either empathy or observation instructions and asked them to

watch a videotape or listen to a conversation involving a

two-person interaction, focusing on one of the characters.

Each of the studies has shown that "empathic observers pro-

vide attributions more like those typically offered by ac-

tors themselves" (Regan & Totten, p. 854) than do nonempathic

observers.

In fact, the influence of empathy on one's attributional

perspective may underlie its effects on "just world" evalu-

ations. By sharing the actor's viewpoint, empathic observers

probably tend to blame the situation, and not the actor, for

the actor's negative fate; thereby, viewing the the actor as

32

less responsible for her fate, they are probably less likely

to derogate her than are nonempathic observers.

Empathic Embarrassment: A Conceptualization and Research
Design

Thus, empathy seems manipulable in the laboratory, as

instructions to concentrate on another's feelings appear to

facilitate its emergence. Moreover, its effects appear to

be pervasive, exerting a broad influence on the reactions of

observers to other people. It seems likely, then, given the

prevalent demonstrable effects of empathy, that whenever an

actor suffers the flustered discomfort of embarrassment, oth-

ers may empathically share that embarrassment, or at least

react emotionally to it. Indeed, we have already seen that

Sattler (1965) included an "embarrassment for others" group-

ing in his categorization of embarrassing incidents--one of

his respondents reported "I am embarrassed if I see someone

else go through an embarrassing experience (p. 33)"--and that

analysis of Modigliani's (1968) Embarrassability Scale yielded

an empathic embarrassment factor. In fact, even Gross and

Stone (1964) and Goffman (1956) have mentioned phenomena

akin to empathic embarrassment, though they did not elaborate.

Gross and Stone suggested that "embarrassment is infectious.

It may spread out, incapacitating others not previously in-

capacitated" (p. 2), while Goffman noted that "when an indi-

vidual finds himself in a situation which ought to make him

blush, others present usually will blush with and for him,

though he may not have sufficient sense of shame or

appreciation of the circumstances to blush on his own ac-

count" (p. 266). The concept of empathic embarrassment is

evidently not new; nevertheless it has not yet received the

attention it deserves.

The study described herein attempts to illustrate em-

pathic embarrassment and some of its presumed parameters. It

assumes that, as Goffman's quote suggests, one may feel em-

pathic embarrassment for another individual even though he is

not embarrassed himself. Seeing that another is in a pre-

dicament that the observer would find embarrassing may actu-

ally embarrass the observer even though the actor shows no

outward indication of being embarrassed; it is possible, af-

ter all, to empathize with emotions which do not actually

exist. Furthermore, it assumes that empathic embarrassment

may be felt for strangers as well as for those with whom the

observer is related in some manner. Rephrased, this assump-

tion holds that an observer may find the actions of another

embarrassing even though the other's actions in no way reflect

upon the observer or threaten his face. One would expect a

mother to be embarrassed by the improprieties of her daughter;

the faces of "team" members are often interdependent. How-

ever, as our discussion of empathy suggested, a purely em-

pathic embarrassment may be felt for anyone with whom one

can empathize, whether the observer is interacting with

the actor or not. One's emotional reaction to the embar-

rassment of another to whom one is related or somehow linked

may well be stronger than one's reaction to a stranger's

plight, but if empathy is possible, embarrassment may still

exist.

The present study examined the reactions of observers

to an actor's embarrassment, manipulating the perceived link

between actor and observer and the observational set of the

observer. Before being individually assigned to actor or

observer roles, pairs of subjects were induced to cooperate,

compete, or maintain their independence before an experimental

confederate who served as their "audience." Cooperation was

expected to form a positive perceived link between the sub-

jects, encouraging them to consider themselves a team, at

least in the eyes of the confederate. Competition was ex-

pected to form a negative perceived link, perhaps causing

them to consider one another rivals, and independence was

expected to establish no link between them. The observer

then watched the actor perform a number of embarrassing tasks,

under one of two observational sets; observers were instructed

either to concentrate on the actor's feelings or to watch

his/her movements carefully. Three indicators of the observ-

er's reaction were obtained--the observer's skin potential,

a measure of emotional arousal (Prokasy & Raskin, 1973), was

recorded, the observer's compliance with a request for help

was measured, and the observer's self-report of embarrass-

ment was secured. Measures of the actor's embarrassment and

compliance were also obtained so that comparisons between

the reactions of actor and observer could be made. In addi-

tion, an offset control group in which the actor performed

35

unembarrassing tasks was used to provide a standard of com-

parison for the reactions of embarrassed actors and observ-

ers. Thus, with subject sex included as another factor, the

study was a 2 by 3 by 2 factorial design including subjects'

roles (actor or observer), the link between actor and observer

(cooperation, competition, or independence), the observer's

instructional set (empathy or observation), and subject sex,

with an offset control group.

Empathic embarrassment will be best illustrated if the

observers share the actors' embarrassment in the independence-

empathy cell. Indeed, although the subjects were virtual

strangers to one another, independent observers are expected

to exhibit observable emotional reactions in response to the

actors' embarrassment. Of course, it will be difficult to

determine conclusively whether an observer's reaction is

really one of shared embarrassment or rather some other emo-

tion such as pity or pleasure. Still, in one sense it mat-

ters not that we may be unable to definitively label the

observer's response--any differential responsiveness of the

observers to the actors' plight across conditions will be

instructive. Nevertheless, the battery of dependent measures

may help distinguish empathic embarrassment from other re-

sponses. For instance, observers were asked how sympathetic

they were to the actors, and how sorry they felt for them;

the greater their "sympathy," the more empathic their re-

sponses might be expected to be, and the less sorry they

felt, the less pitying their responses. Moreover, observers

36

were asked how embarrassed they felt; differences in reported

embarrassment between empathic and non-empathic observers

will suggest that the response of empathic observers is one

much like embarrassment. We will have to be cautious in

labelling as "embarrassment" the responses of the observers,

but the results of the study should be informative in any

case.

Observers in the cooperation-empathy cell should dis-

play the strongest emotional response--since they and the

actors may be considered something of a team, the actor's

embarrassment may be said to endanger their face. By con-

trast, observers who have competed with the actors might

relish the actor's embarrassment; they are expected to be

least embarrassed by the actor's predicament. Across all

conditions, the actor's reported embarrassment is expected

to be greater than that of the observer, although empathic

observers are expected to be significantly more embarrassed

than control observers watching unembarrassed actors. No

differences between the reactions of males and females are

expected, although females may tend to be more embarrassable

than males (cf. Garland & Brown, 1972).

Thus, the following predictions are made:

Hypothesis I. Subjects who are observing the actor

under instructions to empathize will exhibit more physiologi-

cal arousal, report more embarrassment, and agree to help

more than will subjects who are instructed to simply watch

the actor carefully.

Hypothesis II. In the empathy condition, observers who

have cooperated with the actor will exhibit more arousal and

helping, and report more embarrassment than will observers

who have maintained their independence, while even less re-

ported embarrassment is expected from subjects who have com-

peted with the actor. These differences between conditions

are expected to be smaller in the observation condition,

where observers should be less aroused by the actor's embar-

rassment; in fact, those who have competed with the actor

may be pleased by his/her plight.

Hypothesis III. Actors will report greater embarrassment

and will agree to help more than will observers, but empathic

observers should report significant embarrassment in response

to the actor's plight.

CHAPTER II

METHOD

Subjects

Eighty-four male and 84 female introductory psychology

college students participated in partial fulfillment of a

course requirement. Three additional subjects refused to

perform the embarrassing tasks, and they and their observers

were dismissed without penalty.

Procedure

Subjects reported to the laboratory in same-sex pairs.

There, they were greeted by the experimenter and a female

confederate, who was introduced to the subjects as an under-

graduate experimenter-trainee who was attending her first

experimental session in order to learn the procedure. The

experimenter then described the study as an investigation

of the "manner in which people form impressions about other

people" (Appendix A). It was explained that a person's first

impressions of another person were important determinants of

his subsequent evaluations of that person, and that it was

important to find out what kinds of things influenced first

impressions. In particular, the study was said to be inves-

tigating the "physiological or bodily changes that accompany

the impression formation process."

The subjects were then told that they would first complete

a short task and that a coin flip would then assign one of

them to an actor role and the other to an observer role.

Thereafter, the actor would perform a number of additional

tasks while the observer watched and listened from behind a

one-way mirror and his/her physiological reactions were re-

corded. Then the observer would be asked for his/her impres-

sions of the actor.

After the subjects gave their informed consent to this

procedure (Appendix B), they were randomly assigned to one of

three interaction conditions. All subjects then engaged in

a task loosely modelled after that of Wolosin, Sherman, and

Till (1973). They were given a "sociability inventory" which

presented them with twelve pairs of "sociable activities"

arranged in a forced-choice format (Appendix C). For instance,

one item read "If you were dating someone for the first time,

you'd probably go: a) to a movie, b) to a party." However,

the instructions accompanying these items were varied in order

to induce three different interactional sets among the sub-

jects. Those assigned to the Cooperation condition were

told that the inventory was "designed to assess how well two

individuals can anticipate and match each other's choices of

sociable activities" (Appendix D). They were instructed to

answer each question as they believed their "partner" would,

and to ignore their own personal preferences. Instead, they

were to try to match the other's responses as often as possi-

ble. Moreover, the two of them would receive a single, joint

40

score, and the more often their answers were the same, the

higher that score would be. Thus, the cooperation instruc-

tions were designed to lead subjects to think of themselves

as interdependent cooperating partners.

Subjects in the Independence condition read that the

inventory was "designed to assess how well an individual can

anticipate and match the sociable activities chosen by a

majority of his peer group" (Appendix E). They were instructed

to answer each item as they believed "most other students"

would, again disregarding their personal preferences. They

were to try to match the responses chosen most often by other

students, were told not to let the other person influence

them, and were informed they would each receive their own

individual score. Thus, the independence instructions asked

subjects to think of themselves as separate, independent

individuals.

Finally, subjects in the Competition condition were told

that the inventory ascertained how well two individuals could

"anticipate and either match or avoid each other's choices"

of activities (Appendix F). They found that the inventory

was divided into two parts and that on one half they would

play the role of a "hider," and on the other half the role

ofa "pursuer." As a hider, they were to answer the questions

as they believed their "opponent" would not; in other words,

they would be trying to "hide" from their opponent, choos-

ing answers different from his/hers. By contrast, while a

pursuer, they were to try to answer the questions as their

opponent would, as if they were attempting to "find" the

other person. Thus, on each item they were asked to either

match or avoid their opponent's answers while the opponent

was trying to avoid or match their own. In fact, there were

two different forms of competition condition instructions,

so that one subject was instructed to play the pursuer on

the first six items and the hider on the last six, while

his/her opponent was told to play the hider first, then the

pursuer. Furthermore, they were informed that on each trial

only one of them would receive any points--with points going

to the hider or pursuer depending on whether or not their

answers matched--and that, since this was a competitive task,

only one of them would win. Thus, the competition instruc-

tions led subjects to consider themselves interdependent,

competitive rivals.

The experimenter reiterated the gist of the subjects'

instructions, answered any questions, and asked them to be-

gin. In all three interaction conditions the actual amount

of interaction between the subjects was the same--each sub-

ject answered the task questions individually, circling responses

at his/her own rate, and was led to believe (in the coopera-

tion and competition conditions) that their answers would be

compared later to determine the task outcome. In no case

was feedback concerning their performances on the task actu-

ally given.

After the subjects completed the task, a coin flip de-

cided which of them was to be the actor and which the observer.

Then, after the actor was asked to sit quietly for a few

moments, the observer was taken to an adjoining room where

he/she was able to view the actor through a one-way mirror.

Beckman Ag/AgCl electrodes capable of measuring skin poten-

tial were attached to the hypothenar eminence of the subject's

right palm and to the medial aspect of the volar surface of

the right forearm--with the experimenter making a show of

instructing the confederate-apprentice in the attachment

procedure--and the subject was asked to read one of two sets

of instructions adapted from Gould and Sigall (1977). For

half the observers, those in the Empathy condition, the in-

structions read:

In a few moments, you will be watching the
"actor" perform a number of tasks. While you are
watching him, picture to yourself just how he feels.
Try to forget yourself. Concentrate on him. Your
job will be to empathize with his feelings and reac-
tions to the situation. In your mind's eye, try
to visualize how it feels to him to be performing
the tasks.(Appendix G)

The remaining observers, those in the Observation condition,

read:

In a few moments, you will be watching the
"actor" perform a number of tasks. While you are
doing so, please make careful observations of every-
thing the actor does. Observe closely all charac-
teristics of her behavior. Your job will be to
observe carefully both the frequency and pattern
of her nonverbal responses. Try to watch for hand
gestures and general shifts in body carriage. In
sum, observe her behavior as carefully as you can.
(Appendix H)

The gender of the pronouns in the instructions was varied to

correspond to the sex of the subjects. Once the observer

understood the instructions, recording of his/her skin poten-

tial by a Narco-Bio Physiograph was begun.

The experimenter and confederate then returned to the

actor's room and asked the actor to choose the tasks he/she

would perform by drawing one of four slips from an envelope.

In fact, for experimental subjects, all four slips instructed

the actor to perform four tasks shown to be embarrassing by

Apsler (1975): Actors were asked to (1) turn on a tape re-

corder and dance to the recorded music, an excerpt from an

enormously popular rock tune which lasted for 60 seconds,

(2) laugh for 30 seconds as if they had just heard a joke

(a clock was provided), (3) sing the "Star Spangled Banner"

(the words and music were printed on the back of the instruc-

tion sheet), and (4) imitate for 30 seconds a five-year-old

throwing a temper tantrum because he does not want to go to

bed (Appendix I). Actors were told not to voice remarks

about the tasks as they were performing them and were told

to complete them as rapidly as possible, moving promptly from

one to the next.

In addition to the experimental subjects, 14 males and

14 females served as subjects in an offset control group.

The 14 actors in this group performed four nonembarrassing

tasks from Apsler (1975), being asked to (1) listen, not dance,

to the music for 60 seconds, (2) count silently to 50,

(3) read for 30 seconds a book that had been left on the
table, and (4) copy as many of the words of the "Star Spangled

Banner" as they could in 60 seconds (the words were again

provided) (Appendix J). The 14 observers in the control

group interacted with the actors under independence condi-

tions and were given empathy instructions.

The experimenter then turned on a microphone which en-

abled the observer to hear the actor, and before leaving the

room, told the confederate-apprentice to go watch the pro-

ceedings with the observer. The confederate was also asked

to hand out questionnaires when the actor was finished, the

experimenter explaining that he was going downstairs to mimeo-

graph more materials. (He actually monitored the observer's

physiological records in another adjacent room, listening

to the actor over headphones.)

When the actor completed the tasks, the confederate

haltingly removed the observer's electrodes, turned off the

microphone connecting the rooms, and gave both actor and

observer their respective questionnaires (Appendices K and

L). Both questionnaires asked subjects to rate their feelings

on four eight-point bipolar adjective scales adapted from

Modigliani (1971) (e.g., poised-awkward, flustered-calm,

embarrassed-unembarrassed), and to attribute the actor's

performance to either his/her personal characteristics or

to the situation, apportioning 100 points between these fac-

tors. In addition, both questionnaires included several

19-point scales with labelled subdivisions which assessed

subjects' perceptions of the actor's embarrassment and the

explanations they used to account for the actor's behavior.

Moreover, the observer's questionnaire--in keeping with the

study's "impression formation" cover story--included the 15

bipolar adjective scales used by Lerner and Simmons (1966),

on which observers were asked to rate both the actors and

45

themselves. Finally, the observer's questionnaires contained

several more 19-point scales which asked observers for their

ratings of their own embarrassment, their sympathy and sorri-

ness for the actor, and their enjoyment of his/her actions.

The confederate waited in the larger room with the actor,

and when he/she had completed his/her shorter questionnaire,

she made this request, adapted from Apsler (1975):

Say, can I ask you a question? You're not
taking Psych 414 are you? Well, I'm taking it
this quarter and we have to do a study of our own.
I'm studying people's moods. In my study I'm hav-
ing people fill out a mood questionnaire at home
that takes about 30 minutes each day or each even-
ing for a number of consecutive days. I have some
subjects but I need more. Ideally, I need people
who can fill out the questionnaire every day for a
month, but any number of days is good, even one.
But the more days the better. Since I'm just a
student I can't give any experimental credits, but
do you think you could be a subject in my mood study?
For how many days?

When the observer completed his/her questionnaire, the

same request was made of him/her. The actor and observer

were then rejoined, the experimenter was summoned, and the

subjects were debriefed. A lively discussion often ensued,

and many subjects reported that this had been the most enter-

taining and engrossing experiment they had encountered.

CHAPTER III

RESULTS

In presenting the results, we will first briefly examine

the actors' self-reports of embarrassment to determine whether

or not the actors were indeed embarrassed by their tasks.

The observers' responses will then be compared to the experi-

mental hypotheses, and finally, the responses of actors and

observers will be compared.

Actors' Responses

The actors rated their embarrassment on two different

measures. First, they were asked to report "how you felt

while you were performing the tasks" by rating their feelings

on four eight-point bipolar adjective scales adapted from

Modigliani (1971) (i.e., at ease--self-conscious, poised--

awkward, flustered--calm, and unembarrassed--embarrassed);

following Modigliani, the mean of each subject's responses

on the four items was computed and used as an embarrassment

score. Second, the actors responded on a 19-point scale to

an item which asked, "How embarrassed were you when you were

performing the tasks?"

The responses of the actors who had performed the Em-

barrassment tasks were compared to those of the control group

actors using Dunnett's test, a multiple comparison statistic

46

which controlled the experimentwise error rate. The results,

displayed in Table 1 (with the experimental means grouped

only by interaction-type, since the manipulation of instruc-

tional set did not affect the actors), show that on both

measures the experimental actors reported significantly greater

embarrassment than the control actors did. The Embarrassment

tasks were indeed more embarrassing to the actors than were

the more innocuous Control tasks.

The actors' responses to the confederate's request for

help were also tabulated, and the experimental and control

groups compared. As Table 1 indicates, the data replicated

Apsler's (1975) results, showing that the embarrassed actors

volunteered significantly more help than did the unembarrassed

actors.

Other responses of the actors will be considered in the

section below entitled "Actor-Observer Comparisons."

Observers' Responses

Except where noted, the observers' responses to the

questionnaire were analyzed using a three-way analysis of

variance which included type of interaction (cooperation,

independence, or competition), instructional set (empathy

or observation), and subject sex as factors.

Perceptions of the actors' embarrassment. The observers

rated the embarrassment of the actor on a 19-point scale

which asked, "How embarrassed do you feel the actor was when

performing the tasks?" A main effect of instructional set,

F(1, 58) = 5.90, p < .02, a set by sex interaction,

TABLE 1

ACTORS' EMBARRASSMENT AND COMPLIANCE
CONTROL GROUP COMPARISONS

Interaction-Type

Cooperation Independence Competition

Embarrassment: Bipolar Adjective Scales

4.83 5.28 4.72

Control Mean = 3.58

Embarrassment: 19-Point Scale

11.12 12.12 11.22

Control Mean = 4.93

Compliance a

15.37 15.72 16.85

Control Mean = 10.15

Note. All means differ significantly from their respective
Controls by Dunnett's test, P < .05.

a Compliance scores refer to number of days volunteered.

f(1, 58) = 5.47, E < .03, and a triple interaction of inter-

action-type, instructional set, and subject sex, F(2, 58)

3.94, 1 < .03, were all obtained on this item. The main

effect suggested that observers given empathy instructions

considered the actors to be more embarrassed (M = 12.3) than

did those instructed to watch carefully (M = 10.1), but the

set by sex interaction showed that this effect was limited

to females. However, the triple interaction clarified the

observers' responses further, revealing that male-female

differences were confined to cooperation conditions. As

Table 2 illustrates, simple effects tests revealed a simple

interaction of set and sex within the cooperation cells,

F(1, 58) = 13.11, p < .001, and further analyses indicated

that males and females reacted quite differently in that

condition. Empathy females considered the actors more em-

barrassed than did observation females, F(1, 58) = 5.81, p < .02,

but empathy males considered the actors less embarrassed than

did their counterparts given observation instructions, F(1, 58)

= 7.35, P < .01. Moreover, the empathy males rated the ac-

tors as less embarrassed than did the empathy females, F(1, 58)

= 12.00, p < .001. Thus, the simple interaction seems to be

due, for the most part, to the unexpectedly low ratings of

the cooperation/empathy males. Indeed, as we shall see, this

pattern of responses recurs on other measures.

Otherwise, the observers' ratings of the actors' embar-

rassment generally followed a predicted pattern. Excepting

the cooperation/empathy males, empathy observers tended to

TABLE 2

OBSERVERS' RATINGS OF THE ACTORS' EMBARRASSMENT

Instructional Interaction-Type
Set
Set Cooperation Independence Competition

Males

Empathy 7.8ab 12.3c 12.5

Observation 13.8bd 73cd 11.0

Control Mean = 5.4

Females

Empathy

Observation

15.5ae

10.2
e

1j3. 5f

8.8 f

12.2

Control Mean = 7.4

Note. Means with the same single-letter subscript differ
by at least p < .05. Means with an asterisk differ
from their respective Control Means by Dunnett's
test, 2 < .05.

be more attuned to the actors' embarrassment than did their

observation counterparts, and cooperation subjects tended

to see the actors as more embarrassed than did those in other

interaction conditions. For instance, simple effects tests

revealed a simple interaction of instructional set and inter-

action-type on the males' ratings, F(2, 58) = 6.43, 1 < .005,

and tests of simple simple main effects showed that under

independence conditions empathy males perceived the actors

to be more embarrassed than did observation males, F(1, 58) =

5.15, P < .03. Also, a simple simple main effect of inter-

action-type was obtained in the observation instructional

set, F(2, 58) = 4.32, p < .02, and Duncan's multiple range

test subsequently showed that cooperation males considered

the actors more embarrassed than did independence males,

p < .05. Thus, it is clear that the manipulations of both

interaction-type and instructional set were influential,

generally causing the observers to react in the expected

manner.

Comparisons of these responses with those of observers

in the control group showed that the control means were sig-

nificantly smaller than the means of most of the experimental

cells;1 except for the observers in the independence/observa-

tion condition and the females in the competition/observation

condition, observers watching actors perform the experimental

tasks considered them more embarrassed than did observers

who were watching actors perform the less embarrassing Con-

trol tasks. The embarrassment of the experimental actors

was apparent to most of the observers.

Self-ratings of embarrassment. Observers rated their

own embarrassment on the same two measures employed by the

1. The males' and females' responses on the self-report
items were respectively compared to separate means of the
males and females in the control group. These comparisons
were less powerful than more general comparisons combining the
sexes would have been, but were necessitated by the sex
differences on these items.

actors. A multivariate analysis of variance on their re-

sponses to the four eight-point bipolar adjective scales re-

vealed a main effect of subject sex, F(4, 55) = 3.42, p < .02,

and a three-way interaction of instructional set, interaction-

type, and sex, F(8, 110) = 2.70, E < .01. Subsequent uni-

variate analyses showed these effects to be significant on

the individual items, and the means for each item are simi-

lar; thus, for ease of presentation, we shall again examine

the average of each subject's responses on the four items.

As one would expect, a univariate ANOVA on this composite

variable also disclosed the sex main effect, F(1, 58) = 6.19,

2 < .02, and the triple interaction of set, interaction, and

sex, F( 2, 58) = 5.11, P < .01.

The main effect showed that females (M = 4.4) reported

more personal embarrassment than males (M = 3.5), and as

Table 3 illustrates, this tended to be true in nearly every

cell of the design. For instance, tests of simple effects

within the interaction revealed a simple interaction of sex

and interaction-type in the observation condition, F(2, 58) =

4.49, p < .02, and further analyses showed that cooperation/

observation females reported considerably greater embarrass-

ment than did cooperation/observation males, F(1, 58) = 4.28,

p < .05. However, a simple interaction of instructional set

and subject sex, F(1, 58) = 12.03, p < .001, qualified the

main effect, demonstrating that in competition/observation

conditions males expressed more embarrassment than females,

F(1, 58) = 3.96, p < .051.

TABLE 3

OBSERVER'S AVERAGE SELF-RATINGS OF EMBARRASSMENT
ON THE BIPOLAR ADJECTIVE SCALES

Instructional Interaction-Type
Set
Set Cooperation Independence Competition

Males

Empathy 3.4 4.6 2.4bc

Observation 2.9a 3.1 4.5cd

Control Mean = 3.4

Females

Empathy 5.2 4.7 5.be

Observation 4.9af 4.1 2.5def

Control Mean = 3.7

Note. Means with the same single-letter subscript differ
by at least p < .05. Means with an asterisk differ
from their respective Control Means by Dunnett's
test, p < .05.

In fact, males and females reacted quite differently

to the competition conditions, a result which seems to have

caused the three-way interaction of the independent variables.

For example, when watching the actor after observation instruc-

tions, females reported less embarrassment in the competition

condition than in the cooperation condition, F(2, 58) = 3.17,

2 < .05 (and Duncan's multiple range test, P < .05), but

males tended to report more. Moreover, while women in the

competition condition expressed less embarrassment in the

observation set than in the empathy set, F(1, 58) = 7.44,

p < .01, competition men expressed more, F(i, 58) = 4.57,

p < .04.

Thus, the results for the competition condition are more

complex than expected. It appears that, depending on subject

sex and instructional set, a past competitive relationship

between actor and observer did not necessarily allow the

observers to remain unembarrassed by the actors' behavior.

We will return to this question, but we may find that the

competitive link between subjects, even though negative,

often caused observers to be more sympathetic to the actors'

plight than expected.

In general, the observers' instructional sets seemed to

produce the predicted effects, especially influencing the

females. Males in the competition condition reacted in an

unexpected fashion, but other observers who were watching the

actor under empathy instructions tended to be more embarrassed

than those watching under observation instructions. Moreover,

comparisons with the control group indicated that a number of

observers--most notably the independence/empathy males and all

the empathic females--reported significantly more embarrass-

ment than did control observers. Importantly, despite main-

taining their independence from the actors throughout the

experiment, both males and females in the independence/empathy

condition suffered significant embarrassment in response to

the actors' predicament.

The observers also rated their embarrassment on a 19-

point scale which asked, "Did it embarrass you to watch the

actor? How much?" A main effect of instructional set,

F(1, 58) = 4.51, p < .04, and another triple interaction of
instructional set, interaction-type, and subject sex, F(2, 58)

= 6.97, p < .002, were obtained on this item. Although the

main effect showed that observers in the empathy condition

(M = 6.9) were more embarrassed than their observation condi-

tion counterparts (M = 5.0), the interaction revealed that

this was not true in every case. A simple interaction of

instructional set and interaction-type on the females' rat-

ings, F(2, 58) = 4.73, p < .02, demonstrated that empathy

women were considerably more embarrassed than observation

women in both the cooperation, F(1, 58) = 4.80, p < .04, and

competition conditions, F(1, 58) = 11.04, p < .01, but there

were no significant differences in any other conditions

(Table 4).

A simple interaction of instructional set and subject

sex was obtained in the competition condition, F(1, 58) =

10.17, p < .01, indicating that males and females again re-

acted differently to this interaction-type. As on the bi-

polar adjective scales, empathic women tended to react sur-

prisingly strongly (in this case, reporting more embarrass-

ment than either independence/empathy women, F[2, 58] = 7.54,

TABLE 4

OBSERVERS' SELF-RATINGS OF EMBARRASSMENT

Instructional Interaction-Type
Set
Cooperation Independence Competition

Males

Empathy 3.5a 8.6 4.3b

Observation 3.6 4.6 7.2

Control Mean = 3.6

Females

Empathy 9.8ac 4.5 11.6bd

Observation 5.0, 5.8 4.3e

Control Mean = 3.9

Note. Means with the same single-letter subscript differ
by at least p < .05. Means with an asterisk differ
from their respective Control Means by Dunnett's
test, p < .05.

E < .01, and Duncan's multiple range test, P < .05 or com-

petition/empathy males, F[1, 58] = 11.04, p < .01).

Another male-female difference was indicated by a simple

interaction of sex and interaction-type in the empathy condi-

tion, F(2, 58) = 9.91, p < .001. There, female observers

reported more embarrassment than male observers in the

cooperation condition, F(1, 58) = 8.24, p < .01, as well as

in the competition condition cited above.

Comparisons with the control group showed that the empathy

females in the cooperation and competition conditions and the

males in the independence/empathy and competition/observa-

tion conditions reported significantly more embarrassment

than those subjects watching less embarrassed actors. On

this measure, then, females reported significant embarrass-

ment only after being instructed to empathize with actors

to whom they had been linked. Neither the independence/empathy

nor the cooperation/observation females who had reported

significant embarrassment on the bipolar scales differed from

the control females. This may mean either that these observ-

ers were only marginally embarrassed or that the several re-

lated adjectives were a better measure of the flusterment of

embarrassment than the single 19-point scale. By contrast

to the women, the males responded similarly to both measures,

and on both, competition/observation males reported surpris-

ingly high reactions to the actors' behavior. Also, contrary

to expectations, males in the cooperation/empathy condition

reported little embarrassment.

Thus, across the two measures of the observers' embar-

rassment, several patterns are apparent in the data. First,

though there are exceptions, the empathy instructions often

caused observers to report more embarrassment than the ob-

servation instructions did. Second, females generally ex-

pressed more embarrassment than males. Third, the differences

between the sexes were especially pronounced in the competi-

tion condition, which, contrary to expectations, did not uni-

versally minimize the observers' reactions to the actors'

behavior. Finally, in most cases, independent, empathic

observers reported significant embarrassment as a result of

the actors' embarrassment. Was this empathic embarrassment

or something else?

Self-ratings of sympathy, sorriness, and enjoyment. In

order to further define the observers' reactions, the question-

naire also asked them to rate the extent to which they felt

"sorry" for the actor, and the extent to which they were

"sympathetic" toward him/her. Another three-way interaction

of instructional set, interaction-type, and subject sex was

obtained on the sorriness item, F(2, 58) = 6.84, p < .01, and

as Table 5 makes clear, the observers' responses to this item

were similar to their self-ratings of embarrassment. Tests

of simple effects disclosed simple interactions of instruc-

tional set and sex in both the cooperation, F(1, 58) = 7.09,

p < .01, and competition conditions, F(1, 58) = 6.15, p < .02.

Further analyses showed that in these interaction conditions--

where perceived links between actor and observer were expected

to be formed--empathic females felt much sorrier for the actor

than did empathic males (F[1, 581 = 7.19, p < .01, and Eil, 58]

= 4.67, p < .04, for the cooperation and competition condi-

tions, respectively). Furthermore, while the empathic women

tended to be sorrier than the women merely watching closely

(the difference is significant in the cooperation condition,

TABLE 5

OBSERVERS' SELF-RATINGS OF SORRINESS FOR THE ACTOR

Instructional Interaction-Type
Instructional -----------^
Set Cooperation Independence Competition

Males

Empathy 3.7a 7.8 3.7b

Observation 8.6 4.3 9.0

Control Mean = 4.0

Females

Empathy 11.5ac 6.2 10.5b

Observation 5.5e 11.0 5.3

Control Mean = 5.8

Note. Means with the same single-letter subscript differ
by at least p < .05. Means with an asterisk differ
from their respective Control Means by Dunnett's
test, P < .05.

F[1, 58] = 4.22, p < .05), the empathic men tended to be less
sorry than the men told to simply observe (p < .10).

Thus, on this item males and females responded quite

differently in both the cooperation and competition condi-

tions, and there are several cells in which the observers'

responses did not match experimental predictions. As on

their ratings of their own and the actors' embarrassment,

cooperation/empathy males reacted less strongly than expected,

but competition/observation males reacted more so. The fe-

males, for their part, responded more readily under competi-

tion/empathy conditions than expected.

Two processes may be at work here. First, as mentioned

earlier, competition may have formed a stronger perceived

link between subjects than was expected, causing them to

consider themselves more of a team than predicted. This is

certainly possible, since the subjects probably believed they

appeared interdependent, but--due to the structured nature of

their interaction--only slightly competitive to their con-

federate "audience." They may not have considered themselves

true rivals, which would help explain the generous responses

of the observation males and the empathy females in the com-

petition condition. Second, males seem to have been less

responsive to the empathy instructions than females, a differ-

ence which appears most dramatically in the cooperation and

competition conditions. In those cells, where subjects were

observing a person to whom they had been linked, instructions

to empathize with the person affected females profoundly,

but hardly influenced the males.

This sex difference was also reflected in the comparison

of the experimental means to their respective controls. The

males' and females' responses differed dramatically. In the

cooperation and competition conditions, empathic females

reported significantly more sorriness than control observers,

but women in the observation condition did not, while observ-

ation men did express significant sorriness and empathy males

did not. The sexes also differed in the independence condi-

tion.

Despite this complex pattern, it should be noted that the

observers' responses to this measure were much like their

self-ratings of embarrassment on the 19-point scale; all the

groups of observers who reported significant embarrassment

on that item reported significant sorriness here. Still,

the two measures were not completely alike--some observers

who did not report embarrassment for the actors did report

sorriness for them.

Main effects of instructional set, F(1, 58) = 5.42, P < .03,

and of subject sex, F(1, 58) = 9.16, p < .01, emerged on the

observers' self-ratings of sympathy for the actor, and again

the triple interaction of instructional set, interaction-

type, and subject sex was significant, F(2, 58) = 4.48, p < .02.

The main effects showed that those told to concentrate on

the others' feelings were more sympathetic (I = 11.6) than

those instructed to watch closely (M = 8.9), and that females

were more sympathetic (M = 12.0) than males (M = 8.6). Ana-

lyses within the interaction (Table 6) showed that--as on

ratings of sorriness--simple interactions of set and sex were

significant in both cooperation, F(1, 58) = 4.22, p < .05,

and competition conditions, F(1, 58) = 4.73, p < .04. In

both those conditions, empathy women were more sympathetic

toward the actor than were observation women (1Fi, 58] = 6.42,

TABLE 6

OBSERVERS' SELF-RATINGS OF SYMPATHY FOR THE ACTOR

Instructional Interaction-Type
Set Cooperation Independence Competition

Males

Empathy 8.2a 8.5 10.5

Observation 9.2 4.7b 11.0

Control Mean = 7.4

Females

Empathy 16.5ac 11.3 14.5d

Observation 9.7c 13.6be 6.de

Control Mean = 8.3

Note. Means with the same single-letter subscript differ
by at least p < .05. Means with an asterisk differ
from their respective controls by Dunnett's test,
p < .05.

p < .01, and F[1, 58] = 8.79, p < .01 for the cooperation and

competition conditions, respectively), but the same was not

true for men. Indeed, cooperation/empathy females were sig-

nificantly more sympathetic than cooperation/empathy males,

F(1, 58) = 9.54, p < .01.

A simple interaction of sex and interaction-type within

the observation condition, F(2, 58) = 5.77, p < .01, was also

obtained on this item, and further investigation showed that,

given those instructions, the independent females were more

sympathetic than both the independent males, F(1, 58) = 9.64,

p < .01, and the competition females, F(2, 58) = 3.30, p < .05

(and Duncan's multiple range test, P < .05). In contrast to

the sympathetic responses of females in other cells, female

observers in the competition/observation cell--as on their

ratings of embarrassment and sorriness--exhibited a tendency

to be less sympathetic to the actors than were the correspond-

ing males.

Comparisons with the control means showed that the

competition males, independence females, and all of the em-

pathic females differed significantly from the controls.

Overall then, only the cooperation/empathy and competition/

empathy women, and only the competition/observation men reported

reactions significantly stronger than those of control observ-

ers on all four self-report measures. However, females in the

independence/empathy condition expressed significant embar-

rassment (on the bipolar adjectives) and sympathy for the

actors, and independent/empathy males reported significant

embarrassment and sorriness for them. The independence/em-

pathy observers seem to have been responsive to the actors'

plight.

Thus, the patterns of the observers' responses on the

four measures assessing their embarrassment, sorriness, and

sympathy are rather similar. The independent variables seem

to have had much the same effects on each item, and while

there are some differences, the data do not provide a sub-

stantial basis for differentiating any of these reactions

from the others. At this point, it does not appear that any

one of these reactions was predominant (for instance, that

the observers felt embarrassment alone, and no sorriness, for

the actors). Instead, it seems that the observers were enter-

taining a number of related reactions, all of which might be

expected to be highly correlated--feeling sorry for the ac-

tors, being sympathetic toward their plight, and being embar-

rassed by their predicament. Still, it should be emphasized

that many observers did characterize their reactions as em-

barrassment in part, and even in the independence/empathy

cell, where no perceived link between the subjects was es-

tablished by the procedure, the observers often reported sig-

nificantly more embarrassment than did control observers.

However, embarrassment, sympathy, and sorriness were

not the observers' sole reactions to the actors' embarrass-

ing behavior. The observers were also asked to rate how

enjoyable it had been to watch the actors, and though there

were no significant effects of the independent variables on

this item, the experimental observers in every condition

reported enjoying the actors' tasks more than the control

observers did (Table 7). This is perhaps surprising, since

the aversive state of embarrassment and the sobering state

of sorriness may seem to be incompatible with enjoyment.

TABLE 7

OBSERVERS' SELF-RATINGS OF ENJOYMENT;
CONTROL GROUP COMPARISONS

Instructional Interaction-Type
Set Cooperation Independence Competition

Empathy 9.7 9.3 10.0

Observation 9.3 10.2 11.0

Control Mean = 5.8

Note. All means differ significantly from Control by
Dunnett's test, P < .05.

However, there is little doubt that the embarrassing tasks

were more interesting to watch than the control tasks; several

control observers commented that they had been unsure of just

what the actor was doing and that the tasks had been fairly

boring. By contrast, experimental observers were often fasci-

nated by the actors' behavior, and were always curious as to

what would happen next. It is likely, therefore, that the

observers' ratings of enjoyment were based in part on the

widely different levels of intrinsic interest possessed by

the tasks.

It should be noted, too, that despite the embarrassment

to the persons involved, some of the actors' performances--

particularly their attempts at singing the "Star Spangled

Banner"--were rather amusing. Although the author, in the

role of the experimenter listening to the actors over head-

phones, cringed in empathy each time a subject hit an espe-

cially awful note, their actions aften prompted a wry smile.

Though embarrassing, the actors' performances were often

entertaining. Thus, it is probably true that the observers

were not pleased by the actors' embarrassment so much as that

their unusual behavior never failed to hold their attention,

and that their responses to the enjoyment item primarily reflect

this fact.

Autonomic responses--skin potential. Two measures of

the physiological reactivity of the observers were derived

from the recordings of their skin potential (which were based

on a preamplifier setting of 2 mV/cm).1 First, the number

of significant shifts in skin potential occurring within suc-

cessive 30-second time periods were coded and counted (cf.

Buck, Parke, & Buck, 1970). The observers' responses were

recorded continuously during the time the actor was perform-

ing the tasks (a period lasting from 180 to 210 seconds) and

breaking this period into 30-second segments enabled a test

for trends in the data. Changes in skin potential were scored

as significant shifts if they exceeded four millivolts within

one second, and such changes appeared on the record as siz-

able deviations from baseline levels. This measure was thus

an indication of the amount of significant physiological

reactivity over time.

1. Data for 10 subjects were unavailable due to a
variety of instances of experimenter error.

A mean reactivity score was obtained for each subject

by averaging the number of shifts occurring within the seven

30-second periods, and an analysis of variance on this meas-

ure revealed a main effect of interaction-type, F(2, 48) =

3.52, p < .04. The means indicated that cooperative observers

(M = 5.2) reacted to the actors' behavior more than did inde-

pendent (M = 2.9) or competitive (M = 3.8) observers, with the

difference between the latter two groups just missing sig-

nificance (Duncan's multiple range test, P < .05). The data

suggest, as do the various self-ratings, that the perceived

link existing between competitive actors and observers, even

though presumably negative, still prompted those observers

to react somewhat more strongly to the actors than independent

observers did. Furthermore, observers given empathy instruc-

tions seemed to react more strongly (M = 4.6) than those

receiving observation instructions (M = 3.6), as had been

predicted, but the effect was nonsignificant (p < .11).

Comparisons of the mean reactivity scores with those of

the control group showed that only the observers in the inde-

pendence/observation cell did not react to the actors sig-

nificantly more than control observers did (Table 8). This

suggests that in the independent condition, unlike the

cooperation and competition groups, there really was no

perceived link between the subjects which made the observer

reactive to the actors' plight, and that the significant

reactions of the independence/empathy observers were due

primarily to their instructions to empathize.

TABLE 8

MEAN PHYSIOLOGICAL REACTIVITY SCORES;
CONTROL GROUP COMPARISONS

Instructional Interaction-Type
Set Cooperation Independence Competition

Empathy 5.4a 3.9a 4.3a

Observation 4.9a 2.0 3.4a

Control Mean = 1.2

Note. Means with a single-letter subscript differ from
the Control Mean by Dunnett's test, p < .05.

Tests for linear, quadratic, cubic, and quartic trends

in the reactivity sums were performed by applying orthogonal

polynomials to the data and treating the resulting contrasts

as dependent variables in a multivariate analysis of vari-

ance.1 A test for an effect of measurement occasion revealed

that the observers' reactivity varied significantly over the

recording periods, F(7, 44) = 120.58, p < .001, and the lin-

ear, cubic, and quartic trends all proved to be significant

(F[ 2, 48] = 16.20, p < .001; F [2, 48] = 7.81, p < .01; and

F[ 2, 48] = 8.21, p < .01, respectively). There were no inter-

actions of the trends with the experimental conditions. It

1. This procedure avoided the stringent assumption of
a repeated measures ANOVA that the correlations between the
repeated measures be homogenous (cf. McCall & Appelbaum, 1973).

had been believed that the data would reflect a decreasing

linear trend, as the observers habitutated to the actors'

unusual behavior. However, the quartic trend seems to indi-

cate that the observers reacted anew to each of the actors'

four tasks, their skin potential jumping as each unlikely

task began.

The second physiological measure was the amplitude in

millivolts of the single largest shift in skin potential

occurring within one second during the entire observation

period. It was an indication of the size of each observer's

greatest discrete emotional response, and usually appeared

on the record as a huge, sudden jump in skin potential. No

significant effects of the independent variables were obtained

on this measure, but a trend (p < .11) much like the main

effect of interaction-type on the reactivity scores was evi-

dent; here, cooperative observers (M = 20.2) again reacted

somewhat more strongly than independence observers (M = 16.1).

Comparisons with the control group showed that, as on the

reactivity scores, independence/observation subjects were the

only observers to react no more strongly than control observ-

ers (Table 9).

Helping behavior. An analysis of variance on the ob-

servers' responses to the confederate's request for help re-

vealed only a main effect of interaction-type, F(2, 58) =

5.39, P < .01. Independent observers volunteered a consid-

erably larger number of days (M = 21.6) than did either coop-

erative (M = 12.2) or competitive (M = 13.3) observers. This

TABLE 9

SKIN POTENTIAL RESPONSE AMPLITUDES (mV);
CONTROL GROUP COMPARISONS

Instructional Interaction-Type
Set
Set Cooperation Independence Competition

Empathy 20.2a 18.2a 16.6a

Observation 20.1a 12.9a 15.2a

Control Mean = 10.1

Note. Means with a single-letter subscript differ from
the Control Mean by Dunnett's test, p < .05.

response pattern was unexpected and does not support the

relevant hypothesis, but it may be another indication that

both the cooperation and competition conditions established

links between the subjects, while the independence condition

did not; independent observers, feeling themselves to be

individuals, may have felt more pressure to help than did

cooperation or competition observers who considered them-

selves part of a team or group (cf. Latane & Darley, 1970).

A second analysis, which assessed the individual influ-

ences on the observers' compliance of the three women who

served as confederates in the study, obtained no significant

effects, showing that the observers reacted to each of the

women in much the same manner. Since all three of the women

were quite attractive, this had been expected.

Comparisons of the experimental means with that of the

control group (M = 12.8) showed that only the independence

observers (in both the empathy and observation cells) volun-

teered more help than the control observers. Thus, the ob-

servers' helping behavior did not covary with their levels

of their reported embarrassment; this result and others to be

reported shortly suggest that increased compliance is not

necessarily a concomitant of embarrassment.

Correlations. The interrelationships among the major

dependent measures were assessed by computing the Pearson

product-moment correlations between a) the observers' per-

ceptions of the actors' embarrassment, b) the average of

their self-ratings of embarrassment on the bipolar adjective

scales, c) their self-ratings of embarrassment, sorriness,

sympathy, and enjoyment on the 19-point scales, d) their

physiological reactivity scores, and e) their compliance

scores. The results are shown in Table 10.

The observers' perceptions of the actors' embarrassment

were highly positively correlated with their own embarrass-

ment, their feelings of sorriness and sympathy for the actor,

and to a lesser extent, with their physiological reactivity;

the more embarrassed they believed the actors to be, the more

embarrassed and sympathetic they were, and the stronger their

autonomic reactions were. In addition, as one might expect

from the patterns of the previous results, embarrassment,

sympathy, and sorriness were all highly intercorrelated, as

were the mean reactivity scores and skin potential magnitudes.

TABLE 10

PEARSON PRODUCT-MOMENT WITHIN-CELL CORRELATION COEFFICIENTS
AMONG THE OBSERVERS' MAJOR RESPONSES

EMB1 EMB2 SOR SYM ENJ AVG MAG COM

Actor's Emb (ACT) .364* .314* .337* .366* .163 .235 .204 .214
Observer's emb.
Adjectives (EMB1) 1.000 .535* .536* .341* -.002 .261* .204 .061

19-pt. scale (EMB2) 1.000 .664" .406* .172 .105 .043 -.066

Sorriness (SOR) 1.000 .590* .195 .027 .060 .176

Sympathy (SYM) 1.000 .106 .045 .040 .106

Enjoyableness (ENJ) 1.000 .208 .057 .145
Skin potential
Mean reactivity (AVG) 1.000 .661* .075

Magnitude (MAG) 1.000 .153

Compliance (COM) 1.000

Note. Coefficients with an asterisk are significantly different from zero by at least
S< .05.

Importantly, however, the observers' self-ratings of

embarrassment on the adjective scales were significantly

correlated with the physiological measures, while their self-

ratings of sorriness and sympathy were not. Moreover, their

reports of embarrassment were more highly correlated with

their mean physiological reactivity than their sorriness was,

t(60) = 2.29, p < .05; the correlation coefficients are .261

and .027, respectively. Thus, it appears that the emotional

arousal which accompanied their observation of embarrassed

others was more closely related to the state of awkward fluster

they labelled "embarrassment" than to the state they described

as sorriness. If one accepts Schachter's (1964) two-factor

theory of emotion--which suggests that the experience of an

emotion depends on the presence of bodily arousal and the cog-

nitive labelling of that arousal--one could conclude that the

emotion the observers were generally experiencing was more

embarrassment than sorriness.

Neither the self-ratings of enjoyment nor the measure

of compliance correlated significantly with any of the other

variables. It seems, then, that neither is closely related

to embarrassment--enjoyment, because embarrassment is gen-

erally an aversive state, and compliance because increased

helping is probably not an intrinsic effect of the state it-

self. As Apsler (1975) has shown, compliance may be a likely

means for reducing embarrassment, a facework tactic for im-

proving one's image; still, embarrassment and compliance do

not always covary.

Evaluations of the actor. In keeping with the study's

cover story, observers rated both the actors and themselves

on the 15 bipolar adjective scales employed by Lerner and

Simmons (1966) and Aderman, Brehm, and Katz (1974). Scores

expressing the observers' evaluations of the actors relative

to their self-evaluations were obtained on each item by sub-

tracting from their self-ratings their ratings of the actors;

a resulting positive score indicated that they evaluated them-

selves more positively than the actors. An analysis of vari-

ance on the sum of these "relative derogation scores" (cf.

Lerner & Simmons) disclosed no significant effects, though

several means lay in expected directions. Empathy observers,

for instance, seemed to derogate the actors (M = 5.8) less

than those in the observation condition did (M = 8.8). In ad-

dition, cooperation observers (M = 4.1) seemed to derogate the

actors less than independence observers did (M = 10.3), with

the ratings of competition subjects falling intermediately

(M = 7.3). However, a multivariate analysis of variance on

the subjects' self-minus-other scores on each item also failed

to reveal any significant effects, as did analyses of each

subject's factor scores on the four meaningful factors which

emerged from the adjective scales. Thus, the data do not rep-

licate the results of Aderman et al., who showed that instruc-

tions to empathize with a "victim" could eliminate the "just

world" evaluation effect described by Lerner and Simmons. In

this study, instructions to empathize with an' embarrassed actor

had no significant effect on the observers' evaluations of

the person.

It seemed possible that any effect of the instructional

sets had been mitigated by the observers' perception that the

actors had chosen to perform their particular tasks, and thus,

that they had brought their embarrassment upon themselves.

However, further investigation revealed that there was no

relation (r = .044) between the derogation sums and the ob-

servers' ratings on a 19-point scale of the extent of the

actors' choice of tasks. Nonetheless, there was a nearly

significant negative correlation (r = -.191, p < .09) between

the derogation sums and the observers' ratings on a 19-point

scale of the extent to which they identified with the actors;

to some degree, at least, the greater the observers' identi-

fication with the actors, the kinder their evaluations of

them. The instructional sets had no significant effect, but

the process described by Aderman et al. (1974) may have been

occurring to a slight extent in this study as well.

Analyses of variance also revealed that the independent

variables did not affect the observers' responses to 19-point

scales assessing their identification with, their similarity

to, and their liking for the actors. In addition, none of

the experimental groups differed from the control group on

these items, although all the experimental observers displayed

a tendency to consider themselves less similar to the actors

than did the control observers; this may indicate a desire

of the observers to distance themselves from--to emphasize

their dissimilarity to--the embarrassed actors (cf. Novak

& Lerner, 1968; Taylor & Mettee, 1971).

Actor-Observer Comparisons

Questionnaire items answered by both actors and observ-

ers were examined with analyses of variance which included

subject role (actor or observer), interaction-type (coopera-

tion, independence, or competition), and subject sex as fac-

tors. Instructional set was excluded from the analyses be-

cause it could not influence the actors, who were unaware of

it. (Separate analyses of the observers' responses using

instructional set as a factor revealed no significant effects

not already discussed.)

Self-ratings of embarrassment. Main effects of subject

role, F(1, 127) = 57.07, p < .001, and subject sex, F(1, 127)

= 14.14, p < .001, were obtained on the subjects' mean self-

ratings of embarrassment on the bipolar adjective scales.

As expected, actors (M = 5.9) were more embarrassed than

observers (M = 3.9), and, as we have seen before, females

(M = 5.4) generally reported more embarrassment than males

(M= 4.5).

Attributions for the actors' behavior. The subjects

were asked to explain the actors' behavior by dividing 100

points between the actors' "personal characteristics" and

the "characteristics of the situation," rating the importance

of each as a determinant of the actors' performances. Since

the two ratings were not independent, necessarily totaling

100 points, a single analysis was performed on the personal

characteristics measure. Although actors tended to attribute

slightly less importance to their personal characteristics

than observers did, no significant effects were obtained, and

a subsequent analysis on the observers' ratings alone revealed

no effect of instructional set.

Thus, these results fail to replicate both a classic

actor-observer difference in attributional perspective (Jones

& Nisbett, 1971), and several studies which have shown that

empathy instructions can reverse that perspective (Brehm &

Aderman, 1977; Gould & Sigall, 1977; and Regan & Totten,

1975). It is possible that there was something unique to

this experimental situation which prevented these effects

from occurring, but given the weight of the evidence, it is

more likely that the questionnaire item was a faulty measure.

The item concisely asked for importance ratings of personal

and situational characteristics, but--unlike the measures

used in previous studies--did not give examples of each (e.g.,

the actors' personalities, the nature of the tasks, the ex-

perimenter's requests). Such examples were purposely removed

in order to minimize any demand characteristics associated

with the item. As a result, however, the measure was prob-

ably rather ambiguous, causing each subject to apply his/her

own interpretation of what it was asking, and greatly increas-

ing the item's error variance. It may be incautious to con-

sider this data definitive.

Accounting tactics. Five items on the questionnaire

asked subjects to account for the actors' behavior by rating

the importance of various excuses and justifications for their

actions (cf. Scott & Lyman, 1968). It was expected that ac-

tors would generally attempt to justify their behavior more

than observers would, but only two items revealed main ef-

fects of subject role and both those results were contrary

to expectations. Subjects were asked, "How valuable do you

believe your (the actor's) performing the tasks was to the

research project?" and, "How important was your (the actor's)

current mood in influencing your (his/her) performance on

the tasks?" On both items the observers' ratings were higher

than the actors' (F [1, 128] = 18.78, p < .001; and F [l, 128]

= 10.04, p < .01, respectively); the actors' means were 9.7

and 10.5, respectively, and the observers', 12.1 and 12.7.

There were no significant effects on items asking subjects

to rate the extent of the actors' choice in choosing their

tasks and the importance of pressure from the experimenter

in influencing their behavior.

Helping behavior. Finally, there were no reliable actor-

observer differences in their responses to the confederate's

request. Thus, the compliance measure, though interesting in

its own right, was not an accurate indicator of embarrassment;

it failed to track the observers' embarrassment and to relia-

bly reflect actor-observer differences.

CHAPTER IV

DISCUSSION

The results of this investigation provide some support

for all of the experimental hypotheses and suggest that ob-

servers of an embarrassed person may sometimes suffer empathic

embarrassment in response to the person's plight. Hypothesis

I, which predicted that observers receiving instructions to

empathize with the actors would react more strongly and more

readily to the actors' predicament than would observers sim-

ply instructed to watch carefully, received partial support.

Main effects of the observers' instructional set were obtained

on their ratings of the actors' embarrassment, their own

embarrassment, and their sympathy for the actor, and empathic

observers generally reported more embarrassment, sympathy,

and sorriness than control observers did. However, the

triple interactions of the independent variables on these

items showed that the instructional sets influenced the fe-

males more consistently than the males. Empathic women gener-

ally reacted more vigorously than the women told to watch

closely, particularly in the cooperation and competition con-

ditions where links between the subjects were formed. By

contrast, the reactions of empathic men in these conditions

were often no different and sometimes less strong than those

of men given observation instructions.

Thus, it appears that in some cases males were less

affected by the empathy instructions than were females. This

may have occurred because the males were simply poorer empa-

thizers than the women, simply less gifted or practiced in

the art of "sharing" another's feelings. Indeed, there is

some evidence that males are less oriented toward socioemo-

tional concerns than females are (cf. Kahn, Hottes, & Davis,

1971; Vinacke, 1959). However, under independence conditions,

the empathy instructions did influence the men; there, empa-

thic males always reacted somewhat more strongly than obser-

vation males. It seems that any ineffectiveness of the empa-

thy instructions among the males was limited to the coopera-

tion and competition cells. It is possible, therefore, that

in those conditions, where subjects were told to empathize

with embarrassed, foolish actors to whom they had been linked,

the males' self-ratings reflected a tendency to dissociate

themselves from the actors lest the others' actions reflect

unfavorably on the "faces" of the observers. The combination

of the empathy instructions and the past period of coopera-

tion or competition may have made the others' actions too

threatening, causing the males to distance themselves some-

what and report that they weren't particularly embarrassed,

sympathetic, or sorry.

Nevertheless, there were no significant effects of the

independent variables on the observers' ratings of their

similarity to the actors, where one would expect such dis-

tancing to be evident, and although intriguing, this analysis

does not explain why females would not also attempt to dis-

tance themselves from the actors. Moreover, no sex differ-

ences were obtained on the physiological reactivity scores,

where a trend of instructional set revealed that empathic

observers did tend to react more strongly than observers

merely watching carefully. In fact, the similar responses

of males and females on the physiological measures--where

their responses were not under conscious control--suggest

that male-female differences on the self-report items may

have been due, in part, to the self-presentational concerns

of the subjects. Males are typically characterized as less

emotional and more stoic than females, and in reporting less

reaction than the females in the cooperation/empathy and

competition/empathy cells, the males may have been trying to

fulfill this stereotype--especially since, unlike the females,

they were filling out the questionnaire for an attractive

member of the opposite sex. In those cases where the males

might have felt the most embarrassed, they might also have

felt the most self-presentational pressure to appear cool,

calm, and collected.

Of course, any of these possible factors--the females'

greater socioemotionality, the males' desire to distance them-

selves from the actors, and the subjects' self-presentational

concerns--might have contributed to the sex differences ob-

tained on the self-report items. The precise cause of these

differences is evidently a question which awaits further

investigation. Still, with some exceptions, the instructional

sets did produce the predicted effects; in general, across

the battery of measures, observers instructed to empathize

with the actors did tend to respond more intensely to the

actors' embarrassment than did observers told to watch

carefully.

Hypothesis II, which predicted that cooperation observers

would react most strongly, and competition observers least

strongly (with these differences being most pronounced after

empathy instructions) also received only partial support.

Observers who had supposedly cooperated with the actors often

did react to the actors' behavior somewhat more vigorously

than independence subjects. This was expected, since it was

believed that cooperation would cause the subjects to con-

sider themselves a team, at least in the view of the confeder-

ate who was serving as their "audience." Contrary to expec-

tations, however, the reactions of observers in the compe-

tition condition were often larger than those of independence

observers, and sometimes as great as those of cooperation

subjects. In fact, the overall trend in the data is probably

best illustrated by the physiological measures; both of them

indicated that the responses of the cooperation subjects were

the strongest and those of the independence subjects the

weakest, with those of the competition subjects falling inter-

mediately. Thus, past competition with the actors did not

minimize the observers' reactions to their embarrassment, as

predicted. Instead, the competitive link between the sub-

jects frequently seemed to enhance the observers' responsive-

ness compared to the independence condition which maintained

the subjects' separateness.

It should be noted that subjects in the competition con-

dition were not fiercely competitive. The instructions on

their task informed them they were rivals, but the task itself

was rather ambiguous and no winner or loser was ever declared.

It is possible, then, that the competition condition left sub-

jects feeling more kindly towards one another than intended;

the interaction-type may have had the empathy-facilitating

effects of the cooperation condition, instituting an inter-

dependent relationship between the subjects, but few of the

inhibiting effects that real, overt competition might have

entailed. A more potent manipulation of competition might

have caused the observers to be less sympathetic to the actors'

plight.

In any case, the fact that a link between the subjects

tended to increase the observers' reactions to the actors'

behavior is intriguing. Two influences may be presumed to

have prompted this effect. First, the past period of inter-

dependence between the subjects may have increased the observ-

ers' interest in the actors' behavior, making their actions

more salient. Second, the link between them may have caused

the observers to fear that the actors' behavior endangered the

observers' own "faces," causing them to be embarrassed for

themselves as well as for the actors.

The independence condition seemed to be successful in

maintaining the independence of the subjects--observers in

this condition were the least physiologically reactive. More-

over, observers in the independence/observation cell were the

only ones who failed to exhibit significant arousal in response

to the actors' predicament. These observers, who were instruc-

ted to watch carefully embarrassed actors who they did not

know and with whom they had not been linked, displayed no more

physiological arousal than empathic observers watching unem-

barrassed actors. Of course, it cannot be said that the inde-

pendence subjects were truly as independent as any two ran-

domly chosen individuals. The subjects were asked not to

sign up for the study with friends, and few pairs of subjects

had been previously acquainted; still, both subjects were

students at the same university studying the same academic

subject, sometimes in the same classroom. The subjects were

all superficially similar to one another. Moreover, since

no check of the interaction manipulation was included in the

dependent measures, the effects of the manipulation must be

interpreted with caution. Nonetheless, the subjects seemed

to be as independent as could be readily accomplished, and

on balance, the independence manipulation appears to have

been successful.

Hypothesis III predicted that actors would report greater

embarrassment than observers, but that empathic observers

would still report significant amounts of embarrassment in

response to the actors' plight. This hypothesis received the

strongest support, and in terms of the study's basic intent,

it was also the most important prediction. Fundamentally,

the study attempted to ascertain whether an actor's embar-

rassment could be empathically shared by observers of that

embarrassment. The data seem to answer affirmatively.

Independent, empathic observers watching an embarrassed

actor frequently reported significantly more personal embar-

rassment than did similar observers watching an unembarrassed

actor. They also reported such reactions as sympathy and

sorriness for the actor and enjoyment of their observer-role,

but none of these reactions were as highly related to the

physiological measures of their emotional arousal as were

their self-ratings of embarrassment. The independence of

actor and observer was maintained throughout the study, so

it is unlikely that these observers felt that the others'

actions reflected upon them. In short, their reported embar-

rassment appears to be empathic embarrassment.

Naturally, an assertion that the independence/empathy

observers suffered empathic embarrassment does not deny that

they may have experienced an admixture of other emotions as

well. For instance, it is quite likely that the observers

also felt sorry for the actors; many of them reported signifi-

cant amounts of sorriness compared to the control observers,

and sorriness is a reaction one could reasonably expect from

those witnessing another person's embarrassment. However,

the observers' reports of sorriness were not related to the

changes in their skin potential, suggesting that their

sorriness was more an intellectual, cognitive response than

an emotional one.

The observers also reported that they enjoyed the ac-

tors' behavior. This result was somewhat surprising, and was

probably due more to the fact that the experimental tasks

were intrinsically interesting and amusing than to any ten-

dency of the observers to be pleased by the actors' embar-

rassment. However, the inability of the enjoyment measure

to differentiate the nuances of these responses exemplifies

the problems inherent in identifying the emotions experienced

by the observers. Of necessity, the subjects were asked to

rate their reactions on a few standard scales, and none may

have accurately described their current feelings. Thus, we

cannot positively define the observers' reactions as empathic

embarrassment; another label may better describe their re-

sponses. Still, while the present results are not truly de-

cisive, they suggest that a state akin to empathic embarrass-

ment does exist.

Comparison of the actors' and observers' responses on

the bipolar adjective scales revealed that actors did indeed

report more embarrassment than observers. This had been

expected, since it seemed unlikely that even empathic observ-

ers would be as embarrassed by the actors' behavior as the

actors themselves. However, it is apparent that the observ-

ers were empathizing with emotions which actually existed in

the actors, and these data leave open the question of whether

observers may sometimes feel embarrassed for an actor who is

unembarrassed himself. Our conceptualization of empathic

embarrassment suggests thst they may--that observers can be

embarrassed for others in embarrassing circumstances even

when those others are unaware, or tolerant of, the threats

to their "faces." Still, the question is deserving of

future research.

Furthermore, the present study did not determine whether

a "sympathetic embarrassment" is possible. It is likely

that the observers were able to identify with the actors; as

previously noted, the subjects were all somewhat similar, and

but for the flip of a coin the observers would have been in

the actors' places. Thus, the observers could probably en-

vision themselves performing the tasks, and their reactions

were likely empathic instead of sympathetic. So, it remains

to be seen whether observers would feel embarrassed for ac-

tors vastly dissimilar to themselves, actors with whom sym-

pathy, not empathy, would be a predominant response. It

would be interesting, for instance, to examine in a cross-

cultural study an observer's reaction to an actor who is ob-

viously embarrassed, but who is in a situation the observer

doesn't consider embarrassing. One might find that if the

observer cannot identify with the actor he will remain unem-

barrassed by the other's predicament; that is, only an

empathic embarrassment may be possible.

The observers' evaluations of the actors did not repli-

cate the results obtained by Aderman, Brehm, and Katz (1974)

which indicated that instructions to empathize with an actor

eliminated the "just world" evaluation effect described by

Lerner and Simmons (1966). This was initially surprising,

since the empathy instructions in the present study were

almost identical to those used by Aderman et al. However,

on a more fundamental level, there were considerable proce-

dural differences between the studies which probably caused

the differences in results. The subjects in Aderman et al.'s

study watched a videotape of a passive victim who received

shocks in a learning experiment with her arm strapped to a

chair. By contrast, observers in the present study watched

persons actively participating in their own embarrassment

after drawing a list of tasks from an envelope. The observ-

ers may have tended to fault the actors for their fate more

than Aderman et al.'s subjects did. Indeed, overall, the

observers in the present study derogated the target persons

considerably more than did the observers in the earlier

study; combining empathic and nonempathic observers, the

derogation scores were 7.1 and 1.7, respectively. The embar-

rassing tasks in the present study probably made the actors

appear foolish and silly, which Aderman et al.'s victim was

not, and whether or not the observers tended to empathize

with the actors' embarrassment, they derogated them.

The observers' compliance with the confederates' requests

for help was also somewhat surprising, failing to fulfill any

experimental predictions. Based on Apsler's (1975) results

which showed that embarrassed actors volunteered more help

than unembarrassed actors, it was expected that the observers'

compliance would track their embarrassment. It did not, how-

ever; only observers in the independence condition volunteered

more help than control observers, and there were no actor-

observer differences. The actors' data did replicate Apsler's

results, but in general, compliance did not seem to be closely

related to empathic embarrassment. As noted earlier, volun-

teering one's help may be one strategy for reducing embarrass-

ment and restoring one's "face," particularly when one is an

agent in one's own embarrassment, but empathic embarrassment

does not necessarily increase compliance.

Overall, the results suggest that embarrassment is a

rather omnibus phenomenon. Apsler (1975) has already shown

that embarrassment is not as situationally specific as Goff-

man (1956) assumed. The present results, in conjunction with

Sattler's (1965) categorical scheme, argue that embarrass-

ment may not even be limited to those persons who are agents

in, or recipients of, embarrassing actions. Embarrassment

seems to influence observers as well. Thus, the momentary

threats to an actor's "face" from which embarrassment stems

do indeed involve both actor and audience; not only will the

audience be motivated to help the actor restore the smooth

flow of interaction (cf. Goffman, 1956; Oleson & Whittaker,

1966), but, given the proper circumstances, the audience mem-

bers themselves may be personally embarrassed by the actor's

predicament. The maintenance of "face" in social interaction

seems to be such a central concern and such a precarious

task that just envisioning oneself in the place of an embar-

rassed other may cause one to suffer empathic embarrassment.

Of course, the circumstances of observation seem to be

important. One's reaction to an embarrassed actor may be

influenced by such variables as whether one's observation is

surreptitious or overt, and whether or not the actor knows

he is being watched. For instance, an actor who is being

observed secretly--and who would be embarrassed if he knew

he were being watched--might cause no embarrassment to the

observer, although both actor and observer would be embar-

rassed if the observation were overt. Indeed, a recent

study by Levine and Ranelli (1977) has manipulated such

variables. In it, observers watched a confederate actor

embarrass himself by emotionally overreacting to his failure

on a simple task. The subjects were instructed either that

the actor knew they were watching or that he didn't know (a

manipulation of their psychological visibility to the con-

federate), and were told that he could see them or that he

couldn't (a manipulation of their physical visibility). In

response to these conditions, subjects rated their comfort

while watching and their desire to participate again, items

which combined to form a "composite subjective comfort score."

Thus, the study did not attempt to assess the observers' em-

barrassment, but the results were interesting nonetheless.

Both psychological and physical visibility influenced the

subjects' comfort, suggesting that the observers disliked

"spying" on the actor or contributing to his embarrassment.

91

It is apparent, then, that the present study has inves-

tigated only a sample of the observation conditions which

might influence empathic embarrassment. Here, observers

watched same-sex actors who knew they were watching, but

who couldn't see them in return. Allowing the exchange of

facial cues or conversation between actors and observers

might alter these results, as might the use of opposite-sex

pairs of subjects, and these topics are recommended for

future research.

APPENDICES

APPENDIX A

EXPERIMENTER'S SCRIPT

Hello--what's your name? Okay, come in and take a seat

and we'll start in a few minutes. We're waiting for one more

person. (Oh, hi Karen. Come on in. I'm glad you could make

it. The subjects are here and we're ready to go.)

This is Karen, an undergraduate who is going to train

to become an experimenter. I asked her to come today so

that she could start learning the procedure, and she'll be

standing by watching things today. (Why don't you take a

seat over there?)

This is an experiment dealing with the manner in which

people form impressions about other people. Previous studies

have shown that the first impression you make on someone is

an important determinant of what they'll think of you later.

So, it's important to find out what kinds of things influence

first impressions. In particular, we're interested in the

physiological or bodily changes that accompany the impression

formation process. So, this is a study of impression forma-

tion, and what we're going to have you do is this. First,

you'll work on a short task that will help us assess your

sociability. Then, by flip of a coin one of you will be as-

signed to the role of an observer who will watch the other

person perform a number of tasks. The person who completes

MILLISECOND	CLASS.METHOD	MESSAGE
0	sobekcm_page_globals.constructor
0	sobekcm_page_globals.constructor	Application State validated or built
0	sobekcm_database.verify_item_lookup_object
0	sobekcm_page_globals.constructor	Navigation Object created from URI query string
0	sobekcm_database.verify_item_lookup_object
0	sobekcm_page_globals.display_item	Retrieving item or group information
0	sobekcm_page_globals.get_entire_collection_hierarchy	Retrieving hierarchy information
0	sobekcm_assistant.get_entire_collection_hierarchy
0	cached_data_manager.retrieve_item_aggregation
0	cached_data_manager.retrieve_item_aggregation	Found item aggregation on local cache
0	item_aggregation_builder.get_item_aggregation	Found 'all' item aggregation in cache
0	system.web.ui.page.page_load (ufdc.page_load)
0	sobekcm_page_globals.constructor.on_page_load
0	html_echo_mainwriter.add_style_references	Adding style references to HTML
0	html_echo_mainwriter.add_text_to_page	Reading the text from the file and echoing back to the output stream
84	html_echo_mainwriter.add_text_to_page	Finished reading and writing the file