You are on page 1of 11

Journal of Personality and Social Psychology Copyright 1982 by the American Psychological Association, Inc.

1982, Vol. 43, No. 3, 503-513 0022-3514/82/4303-0503$00.75

Effects of Public and Private Self-Awareness on


Deindividuation and Aggression
Steven Prentice-Dunn and Ronald W. Rogers
University of Alabama

This study tested the proposition, derived from differential self-awareness theory
(Prentice-Dunn & Rogers, in press), that only one type of antecedent variable
traditionally associated with deindividuation (attentional cues) and a single as-
pect of self-awareness (private) are involved in the deindividuation process.
Groups of four participants were exposed to factorial combinations of attentional
cues (internal vs. external focus of attention) and accountability cues (potential
accountability to authority figures and victims) and then allowed to aggress
against a victim. As predicted, attentional cues affected private self-awareness
but not public self-awareness, whereas accountability cues altered public but not
private self-attention. External attentional cues and low accountability cues dis-
inhibited aggression relative to internal attentional cues and high accountability
cues, respectively. Exposure to external attentional cues created an internal state
of deindividuation, composed of reduced Private Self-Awareness and Altered
Experience, that mediated aggression. Results identified two major types of col-
lective aggression: One category resulted from group members' assessments of
the possibility of an authority figure's and the victim's surveillance of their at-
tacks; the other category resulted from the decreased cognitive mediation of
behavior evoked by the deindividuation process.

Antisocial aspects of crowd behavior have for the concept to be valid. Subsequent stud-
repeatedly illustrated the tenuousness with ies in the area of aggression (Prentice-Dunn
which humans adhere to norms of social pro- & Rogers, 1980; Rogers & Prentice-Dunn,
priety. Investigators as early as LeBon (1896) 1981) and socially inappropriate but non-
noted that certain group contexts insulate aggressive behavior (Diener, 1979) have pro-
individuals from feelings of social responsi- vided the necessary evidence for a subjective
bility and fear of reprisals for proscribed state that mediates deindividuated behavior.
acts. Festinger, Pepitone, and Newcomb The overriding issue in recent studies in
(1952) labeled these disinhibiting effects deindividuation has been the question of its
deindividuation. Zimbardo (1970) concep- construct validity (Diener, 1977,1979; Pren-
tualized deindividuation as a process whereby tice-Dunn & Rogers, 1980). Recent research
external input variables lessen self-aware- identifying an internal state of deindividua-
ness and concern with others' evaluation of tion that is causally related to aggression has
oneself, leading to a release of restraints confirmed the utility of the deindividuation
against disinhibited behavior. Diener (1977, construct but has failed to provide a unifying
1980) shifted the focus of attention from the interpretation of past research, which has
input variables to the mediational role of in- yielded puzzling, often inconclusive, results.
ferred internal changes in producing disin- Not infrequently, investigators have dem-
hibited actions, stating that an internal state onstrated that deindividuation manipula-
of deindividuation had to be demonstrated tions produced disinhibited acts without pro-
ducing a subjective deindividuated state.
These results are subject to the more par-
The authors gratefully acknowledge the assistance of simonious interpretation that the experimen-
Cash Spivey and John Traweek in the collection of the tal manipulations simply serve as discrimi-
data.
Requests for reprints should be sent to Steven Pren-
native cues that release aggression (Diener,
tice-Dunn, Department of Psychology, University of 1977).
Alabama, University, Alabama 35486. Recently, we (Prentice-Dunn & Rogers,
503
504 STEVEN PRENTICE-DUNN AND RONALD W. ROGERS

in press) have proposed a theory of deindi- lessened identifiability. These manipulations


viduation that attempts to rectify these past provide the potential aggressor with infor-
ambiguities by refining the deindividuation mation concerning scrutiny by authority
construct. This differential self-awareness figures, the possibility of retaliation by po-
theory of deindividuation specifies the con- tential victims, and responsibility for harm-
ditions under which deindividuation is a doing. In short, they offer information in-
likely cause of collective aggression, as well dicating the extent to which people will be
as the precursors of disinhibited but non- held accountable for their actions; hence, the
deindividuated collective aggression. The label accountability cues.
theory is based on the assumptions that (a) As Figure 1 indicates, differential self-
an individual's self-awareness has dual com- awareness theory posits that accountability
ponents—public and private and (b) the cues operate primarily on public self-atten-
variables historically posited to initiate dein- tion. Increased public self-awareness has
dividuation may be reduced to two catego- been found to increase adherence to social
ries. The major purpose of this study was to standards; that is, in the presence of others,
test the proposition, derived from differen- people tend to conform to group norms, act
tial self-awareness theory, that only one type honestly, and fulfill role expectations (e.g.,
of antecedent variable and a single aspect Buss, 1980; Klecketal., 1976), If increments
of self-attention are actually involved in the in public self-attention produce socially ap-
deindividuation process. propriate behavior, the reverse should also
Much deindividuation research has been hold true: Lowering awareness of public as-
based on the concept of self-awareness re- pects of oneself should decrease concern with
duction "as a crucial element in the deindi- social standards and reduce conformity. Sev-
viduation process. These studies (e.g., Di- eral studies investigating deindividuation
ener, 1979; Prentice-Dunn & Rogers, 1980) have found that exposure to antecedent vari-
incorporated Wicklund's (1975) undimen- ables such as anonymity and diffused re-
sional view of attention as either focused in- sponsibility decreases concern with evalua-
ward on oneself or outward toward the en- tion of oneself by others (e.g., Diener, 1976;
vironment. However, Fenigstein, Scheier, Diener, Lusk, DeFour, & Flax, 1980; Rogers
and Buss (1975) and Buss (1980) suggested & Ketchen, 1979; Zirnbardo, 1970). Thus,
that inward focus could be further divided it is possible that many prior deindividuation
into public and private aspects. Public self- manipulations have functioned to produce
awareness involves attention to oneself as a aggressive behavior, not through deindividu-
social object. Concerns about one's appear- ation, but by expectancy value theory con-
ance and the impression made in social sit- siderations (Atkinson, 1957; Lewin, Dembo,
uations fall within this domain. On the other Festinger, & Sears, 1944). Individuals ex-
hand, private self-awareness refers to a focus perience the safety of being in a group that
on personal, more covert aspects of oneself is not observed by a potential evaluative au-
such as perceptions, thoughts, and feelings. dience; thus, they are less, concerned with
We propose that it is only when antecedent social evaluation. Not expecting censure or
variables decrease private self-awareness reprisals from authority figures or potential
that the deindividuation process is evoked. victims, they behave aggressively. In short,
Numerous variables have been posited to the individuals under these circumstances
initiate the internal changes that lead to are quite aware of their actions—they sim-
deindividuated acts (see reviews by Diener, ply do not expect to suffer negative conse-
1977, and Dipboye, 1977). These antecedent quences. The accountability variables thus
variables may be reduced to two broad cat- serve as discriminative cues that signal that
egories: accountability cues and attentional disinhibited behavior will go unpunished.
cues. Hence, as Johnson and Downing (1979) sug-
Early investigators (Festinger et al., 1952; gested, anonymity-induced aggression "may
Singer, Brush, & Lublin, 1965; Zirnbardo, not require a reduction in the subjective
1970) proposed that the precursors of dein- sense of individuation but, in many in-
dividuated aggression were conditions that stances, could reflect a simple reduction in
SELF-AWARENESS AND DEINDIVIDUATION 505
ANTECEDENT CONDITIONS SUBJECTIVE STATE BEHAVIOR

Figure 1, Two categories of collective aggression: model of independent effects of alternate types of self-
awareness on aggression.

perceived negative sanctions" (p. 1537). We as usual of thoughts, moods, bodily states,
propose that antecedent conditions that lower and other internal processes. The person can-
public self-awareness produce disinhibited not retrieve standards with which to compare
actions, but not through deindividuation. and adjust his or her behavior (cf. Carver,
The second category of antecedent con- 1979; Carver & Scheier, 1981; Diener,
ditions historically advanced to induce dein- 1980) and is therefore less likely to regulate
dividuation is modifications in the person's behavior on the basis of these personal and
attentional and perceptual processes. Inves- social norms. Ordinary perceptual and emo-
tigators focusing on this class of antecedent tional processes become altered. Thus, dein-
have concentrated on fostering (a) group dividuated aggression results from reduced
cohesiveness and arousal (e.g., Diener, 1979) cognitive mediation of behavior, whereas
and (b) focus of attention away from oneself aggression due to lessened accountability is
(e.g., Diener, 1979; Diener & Kasprzyk, the product of the person's conscious weigh-
Note 1). Johnson and Downing (1979) sug- ing of the benefits of disinhibited actions
gested that such variables may have "influ- versus possible negative sanctions. Only the
ences beyond those related to identifiability" lower portion of Figure 1, then, depicts dein-
(p. 1536). Thus, when exposed to these at- dividuated aggression.
tentional cues, people immerse themselves In summary, the present study sought to
in group activity rather than focusing on assess the influences of attentional cues and
self-regulation of behavior. accountability cues on subjective deindividu-
Differential self-awareness theory asserts ation and aggression. Small groups were ex-
that attentional cues alter private self- posed to factorial combinations of the two
awareness (see Figure 1). As with studies in types of antecedent cues. Based on the dif-
public self-awareness, research on private ferential self-awareness theory of deindivid-
self-awareness has neglected to consider the uation (Prentice-Dunn & Rogers, in press),
effects of its reduction on subsequent be- we predicted that the antecedent cues would
havior. However, increased attention to pri- have independent effects on private and pub-
vate aspects of the self has been shown to lic self-awareness and aggression: Atten-
enhance resistance to attempted coercion tional cues would alter private self-focus,
(Carver, 1977) and false suggestions about whereas accountability cues would affect
bodily symptoms (Gibbons, Scheier, Carver, public self-attention. Internal attentional
& Hormuth, 1979) and to enhance mood cues and high accountability cues were ex-
(e.g., Scheier & Carver, 1977). , pected to inhibit aggression relative to ex-
Our theory proposes that a decrement in ternal attentional cues and low accountabil-
private self-awareness is the crucial media- ity cues, respectively. Exposure to external
tor in the deindividuation process. As the attentional cues was predicted to form an
antecedent situation is manipulated to direct internal state of deindividuation, whereas
attention away from oneself, private sblf-fo- exposure to accountability cues was not. Fi-
cus is reduced. The individual is not as aware nally, the use of repeated measures of sub-
506 STEVEN PRENTICE-DUNN AND RONALD W. ROGERS

jective deindividuation allowed us to assess save the experimenter from recruiting additional sub-
the existence of a possible feedback loop jects by providing the stress distractions (electric shocks
delivered via a finger electrode) themselves.
between the internal state and aggression. We explained to all subjects that they received their
Such a loop, not previously tested, has been extra credit points for simply showing up and that they
suggested by some (Diener, 1977; Prentice- could discontinue at any time. Each subject was asked
Dunn & Rogers, 1980) as the mechanism if he had any questions about or any objections to the
use of electric shock. All questions were answered and
whereby subjective deindividuation might no one declined to participate. In addition, written in-
increase aggression that would further en- formed consent was obtained. Two mild sample shocks
hance subjective deindividuation. were administered to each of the problem solvers (i.e.,
the naive subjects). The shocks were from Switches 4
(.25 mA) and 6 (.35 mA) on the aggression machine
Method and each lasted for 1 second. These samples were ad-
ministered to convince the subjects that the apparatus
Design and Subjects really worked and to give them some idea of the shocks
they would be delivering. These two shocks were de-
A 2 X 2 factorial design was employed with two be- scribed by the subjects as feeling like a "pinprick" and
tween-subjects manipulations: (a) internal attentional "a burning sensation," respectively.
cues versus external attentional cues and (b) high ac- Next, the problem solvers' distraction duties were
countability cues versus low accountability cues. Forty- detailed. Subjects were shown a room in which four
eight male students enrolled in introductory psychology seats were arranged in a semicircle around a television
courses participated in the experiment to earn extra monitor and a cabinet speaker. Shock panels were
credit. Twelve subjects were randomly assigned to each mounted on tables in front of each seat. Partial parti-
cell. tions allowed subjects to see each others' faces but
blocked any view of the shock panels other than
Apparatus their own.
Subjects were informed that, as part of the problem-
The shock apparatuses were modified Buss aggression solving process study, they would later watch and eval-
machines. The four aggression machines each contained uate video games and that subsequently their self-rat-
10 push-button switches that could be depressed to de- ings of thoughts while viewing the games would be elic-
liver "shocks" of progressively increasing intensity. Of ited. The televised program was explained as a view of
course, shocks were not actually delivered. the game as the memory subject played it in an adjacent
room. The experimenter explained that during the video
task, a green signal light on their shock panels would
Procedure be presented occasionally, indicating that a shock dis-
traction was to be administered to the memory subject.
Subjects arrived and were tested in groups of four; Each subject was then to select and press 1 of the 10
three were naive participants and one was our assistant. push-button switches: The higher the level chosen, the
This assistant's primary responsibility was to note any stronger the shock that was administered. In addition,
communication between other group members that "shock" was delivered as long as the switches were de-
might endanger independence of shock responses or self- pressed. The shock received by the memory subject was
report ratings. Fortunately, this type of communication alleged to be the average of the four intensities and
did not occur. durations selected on each trial by the four problem
The study was explained as a combination of two solvers.
experiments. The naive subjects signed up for a study Next, the experimenter emphasized that the response
entitled "Problem Solving Processes" and were tested of interest in the memory study was the memory sub-
together. The second experiment was a "Memory ject's performance on a postgame recall questionnaire
Study" for which another introductory psychology stu- (i.e., we were not concerned with how well he played
dent (actually a second experimental assistant) had vol- the game). In addition, subjects were told that any of
unteered. (After half of the experiment had been com- the 10 shock switches would be sufficient for the purpose
pleted, the assistants switched roles.) The purpose of the of the experiment. It was explained that the equipment
problem-solving study was given as the rating of thought had been designed with different shock intensities be-
processes as people performed cognitive tasks, such as cause we had not known how strong the shocks would
solving anagrams and watching and evaluating video have to be to distract someone. We explained that we
games. The memory experiment was explained as an had discovered that the different shocks all had equal
investigation of the role of stressful distractions in the effects on the memory subject's recall, so the naive sub-
memory of details. Problem-solving participants were jects could choose any intensity they wished on each
told that as the memory subject played video games, he trial. These instructions were designed to eliminate any
would attempt to remember game settings, locations of potential altruistic motivation, and they made clear that
played mistakes, and so on, for later recall on a memory use of the lowest possible intensity on every trial would
questionnaire. Subjects were informed that since both fulfill the requirements of the experiment. Use of any
studies involved the same video games and since the intensity greater than Switch 1 would only result in
memory subject had to be distracted anyway, they could additional pain to the memory subject.
SELF-AWARENESS AND DEINDIVIDUATION 507

After the distraction duties were detailed, the problem In the internal attentional-cues condition, subjects
solvers were taken to another room where they were were told throughout the experiment that the unit of
given 15 minutes (in the absence of the experimenter) interest was the individual and his unique reactions. In-
to solve 25 anagrams that formed names of popular rock structions were repeatedly given to concentrate only on
bands. Upon completion of the anagrams, the problem one's own thoughts and feelings while performing the
solvers returned to the video-task room, where their experimental tasks. The anagram task was performed
previously explained video activities were then con- in a brightly lit room at individual desks, with separate
ducted. Twenty shock distraction trials were completed. word lists and no communication or music. During the
The interval between the appearance of any two signal video task, subjects were also tested in a brightly lit
lights was initially chosen randomly, ranging from 25- room with no communication or music. To decrease sub-
60 seconds, and then'held constant across subsequent ject interest relative to the external attentional cues con-
replications. In addition, three self-report questionnaires dition, the video task involved only the Superman game,
assessing a subjective state of deindividuation were ad- taped in black-and-white. This single game was equal
ministered: (a) ah 11-item list given both before the in duration to the two video games viewed in the external
anagram task and after Shock Trial 10 and (b) a 21- condition. It contained instructions delivered in an in-
item list completed after Shock Trial 20. The latter list different voice and had comparatively less action and
included all of the items in the former list. character movement than the games of the external con-
All items were rated on 10-point Likert scales adapted dition.
from Prentice-Dunn and Rogers (1980) and Rogers and The accountability cues manipulation was intended
Prentice-Dunn (1981). Also included in each question- to affect public self-awareness by influencing the iden-
naire were items assessing public and private self-aware- tifiability of subjects' actions to the experimenter and
ness. To form a private self-awareness score, two items, the potential victim. This was accomplished through in-
based on Fenigstein et al.'s (1975) Private Self-Con- structions pertaining to the monitoring of subjects' shock
sciousness Scale, were summed for each subject. These levels and the presence or absence of numerous meetings
items were (a) "Generally, I've been very aware of of subjects with the victim. In the high accountability-
myself and (b) "Rather than thinking about myself, cues condition, the memory subject reported to the same
my mind has been concentrated on what is going on meeting room as the problem solvers before being dis-
around me" (reversal). Two additional items, based on missed for instructions elsewhere. The problem solvers
Fenigstein et al.'s (1975) Public Self-Consciousness were told that after the experiment they would again
Scale, were tallied for each subject to form a public self- meet the memory subject to discuss the experiment, in-
awareness score: (a) "I have been somewhat concerned cluding the shock levels they each used. Subjects were
about the way I've presented myself to the experi- shown how the equipment could monitor their individual
menter" and (b) "I have been concerned about what shocks. Further, they were told that the experimenter
the memory subject might think of me." was interested in the shocks that each person used and
that he would record them. Finally, the memory subject
was taken into the video-task room before the video
Experimental Manipulations game was begun and shown the arrangement of the
shock panels. As the experimenter named the shock
The first independent variable, attentional cues, was panel number for each seat, the memory subject made
designed to maximally affect private self-awareness by eye contact with the person at that particular shock
focusing attention on or away from oneself. This was panel,
accomplished through verbal instructions, the presence In the low accountability cues condition, the problem
or absence of a task fostering group cohesiveness and solvers did not see the memory subject before or during
arousal, and by the inclusion or omission of stimuli that the study and were told that they would not be meeting
maintained subject interest. In the external attentional- him after the experiment. Further, the shock monitoring
cues condition, subjects were given repeated instructions equipment was not explained. Finally, participants were
to focus attention outward and not on themselves. Fur- told that their shocks were not of primary interest to
ther, the study was described repeatedly as concerning the experimenter and therefore would not be recorded.
group problem-solving processes. Anagrams were solved
collectively with subjects' interaction encouraged in or-
der to develop group cohesiveness (cf. Marshall & Hes- Postexperimental Session
lin, 1975). Participants organized single stimulus letters
on a table in a dimly lit room. While working, subjects Following the video task, subjects completed a ques-
listened to loud rock music devoid of aggressive lyrics tionnaire that assessed suspicions about the experiment.
(e.g., songs by Jackson Browne and Bruce Springsteen), Five subjects suspected that shocks were not actually
which has been found to increase arousal (e.g., Rogers delivered; these subjects were deleted from data anal-
& Ketchen, 1979). Verbal interaction among group yses. After each experimental session, each subject was
members was also promoted during the video task. Sub- thanked and given a full debriefing that was based on
ject interest was enhanced by the use of two color video Mills's (1976) recommendations. Finally, a question-
games (Superman and Video Golf from Atari, Inc.) that naire was given to each student in a stamped envelope
were manipulated .to contain exciting play and colorful addressed to the Psychology Department's Committee
instructions describing the games. After the initial in- on Ethics, These anonymous responses were returned
structions, loud rock music was played during the video by 65% of the subjects. Respondents unanimously be-
activities. lieved that (a) the deception was warranted, (b) they
508 STEVEN PRENTICE-DUNN AND RONALD W. ROGERS

Table 1 ment cell to equate cell n.) Significant mul-


Mean Aggression and Self-Awareness Scores as tivariate effects were found for accountabil-
a Function of the Two Independent Variables ity cues, F(2, 38) = 3.41, p < .05, and
Multi-
attentional cues, F(2, 38) = 21.52, p < .0001.
variate Private Public All other multivariate effects failed to reach
Cue aggres- self- self- significance.
condition sion awareness awareness Subsequent univariate analyses of vari-
ance revealed the predicted main effect of
Accountability
High 2.66, 10.46. 10.32, accountability cues on the two items assess-
Low 3.36b 10.00, 8.00 b ing public self-awareness, F(l, 39) = 6.98,
Attentional
p < .02. As shown in Table 1, subjects in the
Internal 2.86, 12.46 a 9.14, high accountability-cues condition rated
External 3.54b 8.00b 9.18, themselves as more publicly self-aware than
did subjects in the low accountability-cues
Note. Pairs of column means that do not share a com- condition. Private self-awareness was not
mon subscript differ significantly (p < .05). significantly affected by the accountability
cues variable (p > .5).
had received a fair explanation of risks, and (c) the As expected, the attentional cues manip-
study was structured to prevent harm. All but one re- ulation significantly affected the two items
spondent agreed that they had volunteered for the study. measuring private self-awareness, F(l, 39) =
This subject failed to explain his response in the space 39.89, p < .0001. Compared to participants
provided on the questionnaire. Fortunately, this student,
along with every other subject (100% of the sample), receiving internal attentional cues, subjects
indicated that (a) the experiment should be allowed to whose attention was externally focused re-
continue and (b) they would be willing to participate ported less private self-awareness. However,
in another similar experiment. the attentional cue variable failed to influ-
ence public self-awareness (p > .9). No other
Results univariate main or interaction effects were
significant.
Aggression
Subjects' shock intensity totals and shock Internal State of Deindividuation
duration totals for Trials 1-10 and Trials Subjects' responses on the retrospective
11-20 were submitted to multivariate anal- questionnaire administered after Trial 20
ysis.1 A multivariate analysis of variance were factor analyzed for indications of a sub-
disclosed main effects for both accountabil- jective state of deindividuation. These 21
ity cues, F(2, 38) = 5.29, p < .01, and at- items included items from previous research
tentional cues, F(2, 38) = 5.00, p < .02. The isolating a subjective deindividuated state
multivariate blocks effect and all interac- (Diener, 1979; Prentice-Dunn & Rogers,
tions were nonsignificant. The multivariate 1980; Rogers & Prentice-Dunn, 1981) plus
aggression centroids are shown in Table 1. the items from Fenigstein et al.'s (1975)
Compared to subjects in the high account- Self-Consciousness Scale assessing public
ability-cues conditions, subjects receiving and private self-awareness.
low accountability cues displayed more A scree plot of eigenvalues (Cattell,
aggression. In addition, the external atten- 1966) provided by a principal-axis factor
tional-cues condition produced more aggres- analysis indicated a two-factor solution. Two
sion than did internal attentional-cues con-
dition. 1
A cell mean was added to one of the treatment cells
to equate the four cell ns; subsequently, one degree of
Public and Private Self-Awareness freedom was deleted from the total. Data analysis in the
present study not involving repeated measures did not
Subjects' public and private self-aware- include the additional cell mean. These unequal n de-
ness scores were collected after Trials 10 and signs were analyzed with the complete least squares
20 and submitted to a multivariate analysis model recommended by Overall, Spiegel, and Cohen
(1975) because this model meets the criterion of esti-
of variance (MANOVA). (As with the aggres- mating the same parameters as those in an orthogonal
sion data, a cell mean was added to one treat- design.
SELF-AWARENESS AND DEINDIVIDUATION 509

factors were then rotated using a varimax Table 2


method. A final solution was based on item Factor Loadings on the Rotated Factors
loadings of .4 or above on only one factor
Factor 1: Altered Experience
and on interpretability of the data. The re- Feeling of togetherness among group .83
sultant factor loadings are shown in Table Felt active and energetic '.83
2. (An analysis of only subjects in the ex- Time seemed to go quickly .81
ternal attentional-cues condition yielded the Felt jubilant .77
Thoughts were concentrated on the moment .76
same two factors.) One factor might be la- Felt glad .70
beled Private Self-Awareness, since it is Felt gleeful .67
composed of items pertaining to attention to Willing to volunteer again with same group .61
one's thoughts and moods. This factor con- Mind was focused on what was happening
tains elements of private self-attention found around me .59
Emotions were different from normal .53
in Prentice-Dunn and Rogers's (1980) and Thinking was somewhat altered .50
Rogers and Prentice-Dunn's (1981) Self-
Awareness factor, but it does not contain Factor 2: Private Self-Awareness
Aware of the way my mind was working .84
aspects of public self-focus found in the Self- Alert to changes in my mood .73
Awareness factor of the two previous inves- Aware of myself .63
tigations. Thus, our series of studies indi- Thought of myself a lot .63
cates that self-awareness should be differ-
Note. These two factors accounted for 30% and 16% of
entiated into public and private components. the variance, respectively. Reliabilities (coefficient al-
To complete our theory of deindividuated phas) were .89 and .73, respectively.
aggression shown in the bottom half of Fig-
ure 1, we should include the additional sub-
jective state of Altered Experience, which Rogers, 1980; Rogers & Prentice-Dunn,
is produced by attentional cues and mediates 1981) that subjective deindividuation has a
deindividuated aggression. This factor, which causal influence on aggression, path coeffi-
contained aspects of arousal, group cohe- cients were determined for the paths be-
siveness, and positive affect, is highly similar tween Altered Experience and aggression as
to the Altered Experience factor isolated by well as between Private Self-Awareness and
others (e.g., Diener, 1979; Prentice-Dunn aggression. Data used in this analysis were
& Rogers, 1980). subjects' factor scores on the two internal
To assess the effects of accountability cues state factors and a total multivariate aggres-
and attentional cues on subjective deindivid- sion score. The path coefficient from Altered
uation, a multivariate analysis of variance Experience to aggression was .30, while the
was conducted on subjects' Altered Experi- coefficient from Private Self-Awareness to
ence and Private Self-Awareness factor aggression was —.15. As such, these path
scores. This yielded a significant multivari- coefficient values were similar to those found
ate effect only for attentional cues, F(2, in the earlier studies for Altered Experience
38) = 14.70, p< .0002. Subsequent univari- (M = .28) and Self-Awareness (M = -.20;
ate analyses disclosed significant attentional- Prentice-Dunn & Rogers, 1980; Rogers &
cues effects on both Altered Experience, F( 1, Prentice-Dunn, 1981). It should be noted
39) = 15.67, p < .0006, and Private Self- that due to the small number of degrees of
Awareness, F(l, 39) = 7.60, p < .009. As freedom, only the Altered Experience path
expected, external attentional-cues subjects coefficient was statistically significant in the
had higher scores on the Altered Experience present study.
factor than the internal attentional-cues sub- Our theoretical model shown in Figure 1
jepts (Ms = .53 and —.49, respectively), but may now be examined in light of these re-
lower scores on the Private Self-Awareness sults. First, the two categories of collective
factor (Ms = —.39 and .39, respectively). aggression (accountability cues and atten-
tional cues) are independent because they
Deindividuation and Aggression were the independent variables in the ex-
perimental design. Second, the results (see
To assess the overall path model in which Table 1) demonstrated that (a) account-
we previously proposed (Prentice-Dunn & ability cues produced public but not private
510 STEVEN PRENTICE-DUNN AND RONALD W. ROGERS

self-awareness and (b) attentional cues pro- nificant. Contrary to expectations, none of
duced private but not public self-awareness. the path coefficients reached significance.
Third, the path analysis demonstrated that Thus, confirmation of the feedback loop was
the two factors of the subjective state of not obtained.
deindividuation (Altered Experience and
Private Self-Awareness) mediated deindi-
viduated aggression. (No coefficient was Discussion
computed for the path from Public Self-
Awareness to aggression because our re- This study confirmed the central propo-
search has focused on the theory of deindi- sition of differential self-awareness theory:
viduation. However, the accountability cues Traditional deindividuation antecedent vari-
main effect on both public self-awareness ables may be divided into two broad cate-
and aggression suggest that this path would gories (accountability cues and attentional
be significant.) Thus, the figure indicates two cues) that have independent effects on self-
separate networks for collective aggression: awareness and aggression. Exposure to
The top half, through public self-attention, accountability antecedents altered public
results in nondeindividuated collective ag- self attention but failed to influence private
gression; the lower half, through private self- aspects of self-focus. On the other hand, at-
attention, results in group aggression by tentional variables altered private self-
deindividuation. awareness but not public self-awareness.
To supplement the major analyses of the Private-public self-awareness theory (Buss,
present experiment, the possibility of a feed- 1980; Carver & Scheier, 1981) was ad-
back loop between subjective deindividua- vanced by the present investigation of the
tion and aggression was examined. This may effects of a reduction in self-focus on be-
be conceptually represented by bidirectional havior. Previous studies concentrated exclu-
causal paths between the internal state and sively on the examination of increased public
aggression. However, to test the causal loop or private self-attention. In addition, the
using a recursive path model, the relation- finding of no interaction between the two
ship was extended temporally to reflect the types of antecedent variables on private-
order in which the measurements were taken. public self-awareness corroborated earlier
Hence the variables were subjective dein- research (see review by Buss, 1980) that the
dividuation (Time 1), aggression (Time 1), two types of attention to self operate inde-
subjective deindividuation (Time 2), aggres- pendently of one another. Apparently, the
sion (Time 2), and subjective deindividua- decreased focus on thoughts and bodily
tion (Time 3). states entailed in a reduction of private self-
For each subject, a multivariate aggres- awareness does not automatically alter pub-
sion score, obtained from MANOVA results lic self-awareness.
on shock intensity and duration, was cal- As predicted, both accountability and at-
culated for Trials 1-10 and Trials 11-20. To tentional antecedents influenced the display
calculate deindividuation scores, seven items of physical aggression. The multivariate
were used that (a) were common to the three analysis indicated that the two types of sit-
self-report questionnaires and (b) loaded uational variables functioned independently.
clearly on either the Altered Experience or Hence, exposure to external attentional cues
Private Self-Awareness factor. Of the seven as compared to internal cues magnified
items, six were contained in the Altered Ex- aggression, whereas high accountability cues
perience factor; only one loaded on the Pri- inhibited harm-doing when compared to low
vate Self-Awareness factor. These seven accountability cues. Low accountability cues
items were summed for each of the admin- signaled subjects that any harm directed to-
istrations of the questionnaire. Thus, each ward the potential victim was likely to go
participant had two aggression scores and undetected. Conversely, cues emphasizing
three subjective deindividuation scores. high accountability made participants feel
Correlations among the variables ranged that their actions were readily identifiable
from .13 to .22, all of which were nonsig- to both the experimenter and the memory
SELF-AWARENESS AND DEINDIVIDUATION 511

subject. In fact, during debriefing, some high conscious of one's thoughts and feelings.
accountability-cue subjects expressed the This Private Self-Awareness factor con-
belief that their use of severe shocks might tained items not previously assessed in ear-
lead to acts of retribution by the victim. lier research. These items (e.g., "Aware of
Thus, accountability manipulations served myself and "Alert to changes in my mood")
simply as discriminative cues signaling referred to attention to the more covert,
whether harmful acts might be punished. unobservable aspects of self, whereas the
This is consistent with previous research il- Self-Awareness factor found in Prentice-
lustrating the impact of potential censure on Dunn & Rogers (1980) and Rogers & Pren-
aggression (see review by Baron, 1977). tice-Dunn (1981) predominantly contained
Exposure to external attentional anteced- elements of public self-focus (e.g., "Con-
ents limited cognitive mediation of behavior cerned with what experimenter thought of
by evoking the deindividuation process. This me" and "Concerned with what victim
decrement in private self-focus resulted in thought of me"). The isolation of a self-
less attention to the thoughts and behavioral awareness factor concerned exclusively with
standards necessary for behavioral regula- private self-attention was probably due to
tion (Carver & Scheier, 1981). Thus, exter- two differences between this experiment and
nal attentional cues may have decreased at- the two prior studies. First, the inclusion of
tention to behavioral norms concerning the numerous items tapping private self-focus
lenient treatment of another person who had was an obvious prerequisite for identifying
done nothing provoking. In addition, the de- such a factor. Second, the much stronger
creased ability to retrieve such standards manipulation of attentional cues (both in-
may have led to the enactment of socially ternal and external) in this study vis-a-vis
proscribed behaviors, such as aggression, past experiments may have allowed the Pri-
that are normally held in check. On the other vate Self-Awareness factor to emerge.
hand, exposure to internal attentional cues, It should be noted that the present exper-
by increasing private self-attention, facili- iment was not designed to replicate the five-
tated behavioral compliance with personal factor configuration of the internal state sug-
norms of lenient treatment (cf. Carver, gested by Rogers and Prentice-Dunn (1981).
1975). This resulted in less aggression di- Factor-analytic investigations of possible ad-
rected toward the victim. ditional facets of the state of deindividuation
The successful isolation of a subjective should employ much larger subject samples.
state of deindividuation replicated and ex- Thus, the results of MANOVA and factor
tended previous research (Diener, 1979; analysis suggest the crucial role of private
Prentice-Dunn & Rogers, 1980; Rogers & self-attention in the deindividuation process.
Prentice-Dunn, 1981) that indicates -that This is reflected in the identification of the
deindividuating antecedent variables pro- Private Self-Awareness component of sub-
duce an internal state of deindividuation. jective deindividuation, paired with the find-
The subjective state was composed of dual ing that attentional cues but not account-
orthogonal components, Altered Experience ability cues influenced subjects' scores on the
and Private Self-Awareness. internal deindividuated state. In addition,
The Altered Experience factor closely re- these results indicate the greater appropri-
sembled previous Altered Experience factors ateness of differential self-awareness theory
(Prentice-Dunn & Rogers, 1980; Rogers as an explanation of deindividuation com-
& Prentice-Dunn, 1981). Common elements pared to explanations based on reduced con-
included group cohesiveness, time distortion, cern with public self-attention; for example,
altered thinking and emotions, arousal, and impression management theory (Lindskold
positive affect. & Propst, 1980) and behavioral contagion
Whereas the Altered Experience compo- (Wheeler, 1966).
nent of subjective deindividuation may refer The initial path model evaluated in the
to the product of private self-awareness, the present study posited the subjective state of
Private Self-Awareness factor found in the deindividuation to be a causal mediator of
present study refers to the process of being aggression. This path analysis yielded path
512 STEVEN PRENTICE-DUNN AND RONALD W. ROGERS

coefficients similar to recent studies (Pren- Differential self-awareness theory sepa-


tice-Dunn & Rogers, 1980; Rogers & Pren- rates two major types of collective aggres-
tice-Dunn, 1981) that confirmed the model. sion. One category of group aggression re-
Thus, increments in Altered Experience sults from a group member's active calcu-
were shown to predict aggression, whereas lations that his or her attacks on another
reductions in Private Self-Awareness pre- person will not be subject to scrutiny or pos-
dicted harm-doing. However, due to the sible retaliation from potential victims and
small number of degrees of freedom, only authority figures. It is the product of the
the path coefficient for Altered Experience person's conscious choice between the ben-
was significant. Nevertheless, when com- efits of aggression and the possible negative
bined with the results of previous investi- consequences. On the other hand; group
gations, the available data strongly suggest aggression may result from decreased cog-
that subjective deindividuation mediated the nitive mediation of behavior. Situational
expression of aggression. cues may debilitate self-regulation of a
None of the path coefficients assessing the group member's behavior when private self-
causal feedback loop between subjective awareness is reduced by the initiation of the
deindividuation and aggression were signif- deindividuation process. Any systematic at-
icant. One problem with the analysis may tempts to prevent potentially harmful out-
account for the unexpected results. Although bursts among groups of people must consider
the factor analysis indicated a bicomponen- both types of aggression and their indepen-
tial subjective state, the deindividuation dent causes.
score on which the path analysis was based
contained too few items from one of the fac- Reference Note
tors. Of the seven items chosen a priori to 1. Diener, E., & Kasprzyk, D. Causal factors in dis-
calculate the deindividuation score, six were inhibition by deindividuation. Unpublished manu-
from the Altered Experience factor and only script, University of Illinois, 1978. (Available from
one was from the Private Self-Awareness Ed Diener, Department of Psychology, University of
factor. Thus, because Private Self-Aware- Illinois, Champaign, Illinois 61820).
ness, a crucial mediator in the deindividua-
tion process, was underrepresented in the de- References
individuation score, the feedback loop sug- Atkinson, J. W. Motivational determinants of risk-tak-
gested by Diener (1977) and Prentice-Dunn ing behavior. Psychological Review, 1957, 64, 359-
and Rogers (1980) was not adequately tested 372.
in the present experiment. Future research Baron, R. A. Human aggression. New York: Plenum,
1977.
using more items specifically tapping Private Buss, A. H. Self-consciousness and social anxiety. San
Self-Awareness is needed to evaluate such Francisco: W. H. Freeman, 1980.
a causal network. Carver, C. S. 'Physical aggression as a function of ob-
jective self-awareness and attitudes toward punish-
Conclusion ment. Journal of Experimental Social Psychology,
1975, 11, 510-519.
This test of differential self-awareness the- Carver, C. S. Self-awareness, perception of threat, and
ory suggests that the frequent linking of ac- the expression of reactance through attitude change.
Journal of Personality, 1977, 45, 501-512.
countability variables to the deindividuation
Carver, C. S. A cybernetic model of self-attention pro-
process in past research may have been in- cesses. Journal of Personality and Social Psychology,
appropriate. The present results clearly dem- 1979, 37, 1251-1281.
onstrate that although accountability cues Carver, C. S., & Scheier, M. F. Attention and self-
certainly influence the disinhibition of regulation: A control-theory approach to human be-
aggression, they do not produce deindividu- havior. New York: Springer, 1981.
ated aggression. Thus, past and future dein- Cattell, R. B. The scree test for the number of factors.
Multivariate Behavioral Research, 1966, /, 245-276.
dividuation research may be better under-
Diener, E. Effects of prior destructive behavior, anonym-
stood by reference to the alternate types of ity, and group presence on deindividuation and
self-attention evoked by experimental ma- aggression. Journal of Personality and Social Psy-
nipulations. chology, 1976, 33, 497-507.
SELF-AWARENESS AND DEINDIVIDUATION 513

Diener, E. Deindividuation: Causes and consequences. Marshall, J. E., & Heslin, R. Boys and girls together:
Social Behavior and Personality, 1977, 5, 143-155. Sexual composition and the effect of density and
Diener, E. Deindividuation, self-awareness, and disin- group size on group cohesivenss. Journal of Person-
hibition. Journal of Personality and Social Psychol- ality and Social Psychology, 1975, 31, 952-961.
ogy, 1979,57, 1160-1171, Mills, J. A. A procedure for explaining experiments in-
Diener, E. Deindividuation: The absence of self-aware- volving deception. Personality and Social Psychology
ness and self-regulation in group members. In P. Pau- Bulletin, 1976, 2, 3-13.
lus (Ed.), The psychology of group influence. Hills- Overall, J. E., Spiegel, D. K., & Cohen, J. Equivalence
dale, N.J.: Erlbaum, 1980. of orthogonal and nonorthogonal analysis of variance.
Diener, E., Lusk, R., DeFour, D., & Flax, R. Deindi- Psychological Bulletin, 1915,82, 182-186.
viduation: The effects of group size, density, number Prentice-Dunn, S., & Rogers, R. W. Effect of deindi-
of observers, and group member similarity on self- viduating situational cues and aggressive models on
consciousness. Journal of Personality and Social Psy- subjective deindividuation and aggression. Journal of
chology, 1980, 39, 449-459. Personality and Social Psychology, 1980, 39, 104-
Dipboye, R. L. Alternative approaches to deindividua- 113.
tion. Psychological Bulletin, 1977, 84, 1057-1075. Prentice-Dunn, S., & Rogers, R. W. Deindividuation
Fenigstein, A., Scheier, M. F., & Buss, A. H. Public in aggression. In R. Geen & E. Donnerstein (Eds.),
and private self-consciousness: Assessment and the- Aggression: Theoretical and empirical reviews. New
ory. Journal of Consulting and Clinical Psychology, York: Academic Press, in press.
1975, 43, 522-527. Rogers, R. W., & Ketchen, C. M. Effects of anonymity
Festinger, L., Pepitone, A., & Newcomb, T. Some con- and arousal on aggression. Journal of Psychology,
sequences of de-individuation in a group. Journal of 1979, / 02, 13-19.
Abnormal and Social Psychology, 1952,47, 382-389. Rogers, R. W., & Prentice-Dunn, S. Deindividuation
Gibbons, F. X., Scheier, M. F., Carver, C. S., & Hor- and anger-mediated interracial aggression: Unmask-
muth, S. E. Self-focused attention, suggestibility, and ing regressive racism. Journal of Personality and
the placebo effect. Journal of Experimental Social Social Psychology, 1981, 41, 63-73.
Psychology, 1979, /5, 263-274. Scheier, M. F., & Carver, C. S. Self-focused attention
Johnson, R. D., & Downing, L. L. Deindividuation and and the experience of emotion: Attraction, repulsion,
valence of cues: Effects on prosocial and antisocial elation, and depression. Journal of Personality and
behavior. Journal of Personality and Social Psy- Social Psychology, 1977, 35, 625-636.
chology, 1979, 37, 1532-1538. Singer, J. E., Brush, C., & Lublin, S. C. Some aspects
Kleck, R. E., Vaughn, R. C., Cartwright-Smith, J., of deindividuation: Identification and conformity.
Vaughn, K. B., Colby, C. Z., & Lanzetta, J. T. Effects Journal of Experimental Social Psychology, 1965,
of being observed on expressive, subjective, and phys- /, 356-378.
iological responses to painful stimuli. Journal of Per- Wheeler, L. Toward a theory of behavioral contagion.
sonality and Social Psychology, 1976, 43, 1211- Psychological Review, 1966, 73, 179-192.
1218. \ Wicklund, R. A. Objective self-awareness. In L. Ber-
LeBon, G. The crowd: A study of the popular 'mind. kowitz (Ed.), Advances in experimental social psy-
London: Ernest Benn, 1896. chology (Vol. 8). New York: Academic Press, 1975.
Lewin, K., Dembo, T., Festinger, L., & Sears, P. S. Zimbardo, P. G. The human choice: Individuation, rea-
Level of aspiration. In J. M. Hunt (Ed.), Personality son, and order versus deindividuation, impulse, and
and the behavior disorders. New York: Ronald Press, chaos. In W. J. Arnold & D, Levine (Eds.), Nebraska
1944. Symposium on Motivation (Vol. 17). Lincoln: Uni-
Lindskold, S., & Propst, L. R. Deindividuation, self- versity of Nebraska Press, 1970.
awareness, and impression management. In J. T. Te-
deschi (Ed.), Impression management theory.and
social psychological research. New York: Academic
Press, 1980. Received January 11, 1982

You might also like