Respon Kognitif

I
REPRINTED FROM
PSYCHOLOGICAL
FOUNDATIONS OF
ATTITUDES
1968
ACADEMIC PRESS INC..
NEW YORK
Cognitive Learning, Cognitive Response to Persuasion, and Attitude

Change1
ANTHONY G. GREENWALD
DEPARTMENT OF PSYCHOLOGY OHIO STATE UNIVERSITY COLUMBUS, OHIO
It is a common assumption that the effectiveness of a persuasive communication is, at least in part, a
function of the extent to which its content is learned and retained by its audience. This assumed learningpersuasion relation is based on a reasonable analogy between the persuasive communication and an
informational communication such as a classroom lecture. In the lecture, it is by defnition of the educational
situation that retention of content is taken as a measure of effectiveness. In the persuasion situation,
however, the essential criterion of effectiveness is acceptance of content. It remains an empirical question to
determine whether acceptance of a persuasive communication is related to retention of its content.
The hypothesis that acceptance of a communication is, in some part, a function of learning or retention of its
content has received explicit endorsement by a number of attitude researchers and theorists (e.g., Hovland,
Janis, & Kelley, 1953; McGuire, this volume; Miller
1'Preparation of this report and the research reported here were supported in large part by grants from the National Science Foundation
(GS-1601) and the Mershon Social Sciences Program at Ohio State University. The author is particularly indebted to Rosita Daskal Albert,
Dallas Cullen, Robert Love, and Joseph Sakumura who have participated actively in various phases of the research reported here. A
condensed version of this chapter was presented at the American Psychological Association sympo sium, Alternatives to consistency theory
in the study of attitude change, Washington, D.C., September, 1967.
& Campbell, 1959; Watts & McGuire, 1964) and has aroused no published
opposition. Indeed, this cognitive learning model of persuasion is most
reasonable. It is widely accepted that cognitions bearing on the object of an
attitude form a major component of the structure of the attitude toward that
object (see, for example, Campbell, 1947; Katz & Stotland, 1959; Krech,
Crutchfeld, & Ballachey, 1962; Rokeach, 1960; Smith, Bruner, & White, 1956).
Since the individual is not born with his cognitions, but acquires them, there
seems to be no reasonable alternative to the assumption that cognitions
bearing on attitude objects are learned. Further, the most obvious source of
such cognitions is the wealth of persuasive messages to which one is exposed
via the public communications media as well as through face-to-face
communications.
In light of the overpowering reasonableness of the persuasion-as- a-functionof-retention hypothesis, it is rather surprising how unsupporting the research
evidence is. A few studies have directly examined the relation between the
learning and attitudinal effects of persuasive communications (Insko, 1964;
Miller & Campbell, 1959; Watts & McGuire, 1964). These studies have
generally found that both communication retention and persuasion diminish
with increasing time between communication and posttest, consistent with the
hypothesis that retention is necessary for persuasion. On the other hand, these
same studies have found only weak and variable correlations between
communication retention and persuasion among subjects tested at the same
posttest interval, suggesting that the relation between retention and
persuasion is not a necessary one. In a conceptually relevant study on
impression formation, Anderson and Hubert (1963) have concluded that there
are separate memory systems for retention of a set of person-descriptive
adjectives and of the person-impression derived from them; their results thus
also suggest little or no necessary relation between a communications
retention and its effectiveness. Additionally, in a number of studies scattered
throughout the attitude literature (and reviewed in Hovland et al., 1953),
variables shown to affect opinion demonstrably such as credibility, fear
46.
COGNITIVELEARNING
ANDCOGNITIVERESPONSE
arousal, and organization of arguments have not been found to have

corresponding effects on retention of communicated arguments. Such negative
fndings on retention measures have typically been used to counter any
possible interpretation for obtained attitude change differences in terms of
unintentionally induced differences in attention to or retention of
communication content. In sum, the research evidence must be interpreted as
uncongenial to the hypothesis that persuasion is a function of retention of persuasive arguments. (In partial reviews of the relevant literature, Insko, 1967,
and McGuire, 1968, reach similar conclusions.)
It must be concluded that either (a) learning of attitude-relevant cognitions

(i.e., persuasive arguments) is unrelated to attitude formation and change; or
(b) persuasive communications can induce attitude change without necessarily
providing the cognitive content on which the attitude is based. The frst
conclusion carries the implication that learned cognitions are not fundamental
to the structure of attitudes. Rather than accept this conclusion, which runs
counter to most conceptions of attitude, it seemed worth some effort to explore
sources other than persuasive communications as possible origins of learned
attitudinal cognitions.
COGNITIVE RESPONSES TO PERSUASION
There is, of course, an important extracommunication source of cognitive

content in the persuasion situation: the cognitive reactions of the
communication recipient to incoming persuasive information. When a person
receives a communication and is faced with the decision of accepting or
rejecting the persuasion, he may be expected to attempt to relate the new
information to his existing attitudes, knowledge, feelings, etc. In the course of
doing this, he likely rehearses substantial cognitive content beyond that of the
persuasive message itself. The present hypothesis is, then, that rehearsal and
learning of cognitive responses to persuasion may provide a basis for
explaining persisting effects of communications in terms of cognitive learning.
The learning of cognitive response content may, indeed, be more fundamental
to persuasion than is the learning of communication content.
This hypothesis is not a new one. The following passages from previous
works indicate views bearing the essence of the present hypothesis.
. . . there is reason to expect that those audience members who are already opposed to the
point of view being presented may be distracted [from the content of a communication] by
"rehearsing their own arguments while the topic is being presented and will be
antagonized by the omission of the arguments on their side [Hovland, Lumsdaine, &
Sheffield, 1949, p. 201],
When exposed to [a persuasive communication], a member of the audience is assumed to
react with at least two distinct responses. He thinks of his own [opinion], and also of the
[opinion] suggested by the communicator. . . . Merely thinking about the new opinion along
with the old would not, in itself, lead to opinion change. The individual could memorize the
content of the [new opinion] while his opinion remained unchanged. Practice, which is so
important for memorizing verbal material in educational or training situations, is not
sufficient for bringing about the acceptance of a new opinion [Hovland et al., 1953, p. 11].
It was hypothesized that conformity in the communication situation will increase
attitude change to the extent to which implicit supporting responses are produced, and
decrease attitude change to the extent to which implicit interfering responses are produced
....
By supporting response is meant any implicit response made by the individual (usually a
self-verbalization), which provides arguments in favor of the overt response he makes;
which produces further motivations in the direction of the overt response; or which relates
the overt response to other stimulus situations. By interfering response is meant any
implicit response made by the individual which provides motivation against the overt
response he makes; which limits the stimulus situations to which the overt response is
applicable; or which is generally irrelevant (such as aggressive or distracting responses)
[Kelman, 1953, p. 187, 211],
Despite a number of speculations similar to the ones just cited, there has
been no direct experimental exploration of the role of cognitive responses in
persuasion, and, in fact, there is not much research that is even relevant. The
research on active participation in the communication process (e.g., Hovland et
al., 1953, Ch. 7) comes closest to being relevant. Elsewhere, isolated
experiments (e.g., Brock, 1967; Janis & Terwilliger, 1962; Kelman, 1953) have
explored dependent variables approximating the present conception of cognitive responses to persuasion. Research in which cognitive responses to
66.
COGNITIVELEARNING
persuasion are employed as independent variables in experimental persuasion

situations is particularly needed. Before proceeding to a consideration of
evidence collected in the authors laboratory, it will be useful to state the
present hypothesis with some precision.
It is proposed that the persuasion situation is usefully regarded as a complex

stimulus that evokes in the recipient a complex cognitive response. The
essential dimensions of the recipients cognitive response are, at the least, (a)
response content, i.e., degree of acceptance versus rejection of the position
advocated in the communication, and (b) intensity, or vigor, of response. The
latter dimension, as well as other possible dimensions of cognitive response,
will not be considered further in this chapter. The essential components of the
persuasion situation as a stimulus that is, as determinant of the cognitive
response content are setting, source, and communication content. An
additional major set of determinants of the cognitive response content is the
set of characteristics brought by the recipient to the persuasion situation,
including his existing repertory of attitude-relevant cognitions as well as
personality traits and group memberships.
As in many other treatments of persuasion, the cognitive response analysis

assumes that attitude change can be achieved by the modifcation, through
learning, of the recipients repertory of attituderelevant cognitions. Such
modifcation might include strengthening of existing cognitions as well as
introduction of new ones. The present emphasis on the mediating role of the
recipients own cognitive responses to persuasion may be formulated as an
assertion that cognitive modifcation of attitudes requires active (not necessarily overt) rehearsal of attitude-relevant cognitions at a time when the
attitude object or opinion issue is salient. Thus the effects of persuasive
communications might range from persuasion when the recipient rehearses
content supporting the advocated position to boomerangwhen the recipient
rehearses content opposing the advocated position.
As a consequence of the present emphasis on the recipients rehearsal of his

own responses to persuasion, it is assumed that learning of communication
content does not play an essential role in mediating the effects of persuasive
communications. The present formulation, therefore, is capable of maintaining

an analysis of persuasion effects in terms of cognitive learning while being
compatible with fndings indicating no necessary relation between
communication retention and persuasion.
It is possible to formulate the cognitive response analysis in terms of an

analogy to the classical conditioning paradigm. In this analogy, the persuasion
situation corresponds to the unconditioned stimulus in that it has a response
evocation capacity; that is, it influences the content of the recipients cognitive
response. As an analog of the unconditioned response, the cognitive response
becomes transferred to the attitude object, which is analogous to the
conditioned stimulus of the classical paradigm. While this analogy may be
decidedly useful, especially in relating the present analysis to other treatments
that have invoked the classical conditioning model (see the chapters by Staats
and Weiss in this volume), it would be inappropriate currently to regard the
model as more than a possibly suggestive analogy. A point of difficulty that
would arise if the model is taken literally, for example, would concern the
nature of the conditioned response in persuasion; it would be unnecessarily
cumbersome, at this stage, to incorporate in the present analysis an analogy to
the conditioning models assumption that conditioned responses are either
fractional components of unconditioned responses or preparatory adjustments
to unconditioned stimuli.
STUDIES OF COGNITIVE LEARNING AND ATTITUDE CHANGE
The cognitive response analysis of persuasion has guided a series of

experiments conducted at Ohio State University. While the primary focus of
this research has concerned the properties and functions of recipients
cognitive responses in persuasion situations, a secondary focus has been on the
role of communication content in persuasion. The present section is devoted
largely to the fndings of completed portions of this research program. The frst
two subsections deal with experiments in which communication content
learning and cognitive response rehearsal have been employed as manipulated
independent variables, enabling conclusions about their causal involvement in
86.
COGNITIVELEARNING
persuasion. The next two subsections are concerned with experiments in which
retention of communicated arguments and occurrence and retention of
cognitive responses have been observed as dependent variables; from these
experiments it is possible to draw conclusions about the determinants and
correlates of cognitive response content and of retention of cognitive responses
and communication content. A fnal subsection provides a brief review of
literature on variables influencing the retention of communicated persuasive
arguments.
It is to be emphasized that the research program described here is very

much in progress. At appropriate points below, the nature of planned and inprogress research will be indicated; these comments will serve, perhaps, more
to indicate the authors awareness of questions that remain to be answered
than to enlighten the reader.
ATTITUDE CHANGE AND LEARNING OF COMMUNICATION CONTENT
Previous studies of the relationship between communication retention and

opinion change have tested this relationship correla- tionally subjects
received a communication and were subsequently tested for both retention of
content and acceptance of the viewpoint of the communication. With the
assistance of Rosita Albert and Dallas Cullen, the author conducted a study
(unpublished) in which communication retention was a manipulated
independent variable, with opinion as the dependent variable. Such a design, it
may be noted, is more appropriate to drawing a conclusion about the causal
role of communication learning in attitude change than is a correlational
design.
The experiment employed three communications concerning United States

foreign aid policy. One of these consisted of 6 brief arguments favoring foreign
aid, 1 of 6 arguments opposing foreign aid, while the third was composed of all
12 of these arguments and did not draw either a favorable or unfavorable
conclusion. For each communication, a group of subjects was asked to rehearse
the individual arguments contained in it, in preparation for a subsequent

retention test; in addition to these three groups, two other groups received the
12-argument communication, and, after briefly reading it once, in entirety,
were asked to rehearse only 6 arguments that had been underlined. For one of
these groups, the 6 underlined arguments were those favoring foreign aid,
while for the other group the 6 opposing foreign aid were underlined. Subjects
were tested for opinion on the issue and then for retention of all arguments in
their communications (not just rehearsed ones) immediately following the
learning task and again, unexpectedly, after a 1-week interval.
For the group receiving the one-sided pro-foreign aid message (N= 16), an
average of 3.7 (out of 6) arguments were retained for the immediate retention
test and 3.1 for the unexpected retention test one week later. The
corresponding means for the group receiving the one-sided anti-message (N=
20) were 3.6 arguments retained immediately and 2.4 after a week. Since these
two groups differed signifcantly from each other in opinion in the directions
advocated in their communications, both immediately (F = 9.50, df= 1, 34, p < .
01) and after a week (F = 7.03, p < .02), these data were consistent with the
hypothesis that communication learning and persuasion are related. Among
subjects receiving the two-sided communications, those assigned to learn pro
arguments (N = 18) retained substantially more pro than con arguments (p < .
001, for each testing); the reverse was true, as expected, for those (N= 19)
asked to rehearse con arguments (p < .001, for each testing); and those asked
to learn both sets (N= 18) retained approximately equal numbers of each.
The effectiveness of the partial learning assignments for the two-sided

communications can be indicated by the fact that subjects retained an average
of 3.0 assigned arguments compared to 0.3 nonassigned ones on the immediate
retention test; for the delayed retention test, the corresponding means were 1.8
assigned and 0.3 nonassigned arguments. If there is, indeed, a causal relation
between argument learning and persuasion, these manipulated differences in
retention should have produced corresponding differences in opinion.However,
there were no opinion differences among the three groups receiving two-sided
communications, either immediately (F < 1, df= 2, 52) or after a week (F < 1).
106.
COGNITIVELEARNING
It must be concluded that learning of communication content is not a

sufficient condition, and perhaps not even a necessary condition, for
persuasion. That is, learned arguments supporting one side of the foreign aid
issue were ineffective when, at the time of learning, subjects were made aware
that credible opposing arguments were available even though the opposing
arguments were poorly retained. This fnding stands in contrast with those of
studies demonstrating across-cell correlations between learning and
persuasion (e.g., Insko, 1964; Miller & Campbell, 1959; Watts & McGuire,
1964). The lack of opinion differences between conditions with decided
retention differences in the present study suggests that such previously
obtained across-cell correlations should not be interpreted in terms of a causal
relation between communication retention and persuasion.
ACCEPTANCE AND RECALL OF IMPROVISED ARGUMENTS
If, as is presently supposed, attitudes change in the direction of cognitive

content rehearsed during a persuasion situation, then procedures that
manipulate the content of the recipients cognitive responses should have
persuasive effect. A traditional persuasion procedure that may be viewed as a
manipulation of cognitive response content is the improvised role-playing
procedure in which a subject is asked to deliver a persuasive message
supporting a position initially unacceptable to him. The majority of research
evidence indicates that role playing produces greater persuasion toward the
unacceptable position than does passive receipt of a persuasive communication
(cf. Insko, 1967, p. 222). Such results are quite compatible with the present
point that a communication recipients rehearsal of his own arguments may be
more important in persuasion than is his rehearsal of arguments contained in
a communication to which he is exposed. However, some commentators feel
that the currently available evidence on role playing is equivocal (see McGuire,
1966, p. 498). Because of this empirical uncertainty, the present research
program included an experiment intended to assess the effect of improvised
role playing on both acceptance and retention of arguments.
In this experiment (Greenwald & Albert, 1968), each subject improvised fve
arguments in response to an assignment to advocate either specialized (career
preparatory) or general (liberal arts) under
11
graduate education. Improvisation was obtained in response to fve neutrally

worded questions that could be answered with an argument supporting either
viewpoint. Assignment to positions was random with approximately 90 subjects
being assigned to each side. In addition to being exposed to their own
improvised arguments supporting one side of the issue, subjects carefully read
and studied for about the same amount of time they had spent improvising
a set of arguments supporting the opposite side that was actually written by
another subject in the study. Since each subjects improvisations served once as
an improvised set of arguments and once as an externally originated set (for a
different subject), this procedure served to equate quality of arguments for the
two sets over the sample of subjects, although not necessarily for each subject.
After a 20-minute irrelevant task, the subjects were tested for opinion on the
generalspecialized education issue and were then asked to recall as many
arguments as they could of those to which they had been exposed both their
own and the ones that had been improvised by another subject.
Figure 1 gives the opinion and retention results for this study. It may be
seen that subjects arrived at opinions consistent, on the average, with the
position to which they had been assigned. The opinion difference between the
two groups was statistically signifcant (F = 7.42, df= 1, 177, p< .01). A more
powerful effect was obtained with the retention measure, with subjects tending
to recall much more of their own improvised arguments than they did of those
improvised by another subject (F= 168.10, df= 1,777, p < .001). A
supplementary fnding was that subjects tended to rate their own arguments
as more original than the others (p < .001), indicating that, other things being
equal, one tends to evaluate his own thoughts more favorably than others. In
summary, these results demonstrated that the subjects rehearsal of his own
cognitions and externally originated cognitions of comparable quality tended to
result in attitude change in the direction of the content of the subjects own
cognitions. These results support the general trend of fndings in the roleplaying literature. More importantly for present purposes, they add force to
the hypothesis that the recipients rehearsal of his own cognitive responses
plays an important role in persuasion.
126.
COGNITIVELEARNING
It
noted
should
be
that
the
Greenwaldexperiment
not provide
explanation
special
of
the
recipients
improvised
arguments.
be
that
subjects
increased
retention of
own
arguments
responsible
observed
attitude
however,
Albert
does
an
of the
efficacy
own
It could
their
was
for the
effect;
SD units for mean of 90 subjects
FIG. 1. Opinion and retention as a function of position assigned for improvisation. (Bars are labeled with
the assigned position. Positive scores on the opinion measure represent favorableness to specialized
education; positive scores on the retention measure represent greater recall of arguments favoring
specialized education than of ones favoring general education.) (Greenwald & Albert, 1968.)
alternative explanations are possible in terms of
13
perception of the self as the source of arguments or in terms of subjects ability

to generate arguments particularly effective in regard to their own attitude
structure (cf. Hovland et al., 1953, pp. 233-237). A series of experiments
designed to select among these explanations will be conducted in the near
future as part of the present research program.
CONTENT AND CORRELATES OF COGNITIVE RESPONSES TO PERSUASION
A series of experiments conducted by Dallas Cullen, under the authors

supervision, explored the content of subjects thoughts
relevant to the topic of a persuasive communication that they had just

received. The communication, on the issue of general versus specialized
undergraduate education, advocated the view that college education should be
directed specifcally at career preparation (that is, should be specialized). It
was assumed that cognitive effects of the communication could be assessed by
comparing the issue-relevant thoughts of subjects who received this
communication with thoughts on the same issue expressed by control subjects
who had received a communication on a different topic.
The subjects in this series of experiments were Ohio State University

introductory psychology students who participated in classroom size groups.
Four separate experiments were conducted with only minor variations in
procedure. The basic procedure for Experiment I included an introduction to
the issue of general versus specialized education consisting mainly of a
defnition of the alternative positions. Then a communication of about 250
words containing 12 distinct arguments supporting specialized undergraduate
education was administered in printed form. Control subjects, at this time,
read a communication concerning instructional television. Next, all subjects
(including controls) were asked to collect their thoughts on the issue of
general versus specialized undergraduate education by listing thoughts that
146.
COGNITIVELEARNING
were:
. . . pertinent to forming and expressing an opinion on the issue of general versus
specialized education. These thoughts may consist of (a) information favorable to one or the
other viewpoint; (b) personal values of yours that are favorable to one or the other
viewpoint; (c) features of either viewpoint that you perceive as good; (d) features of either
viewpoint that you perceive as bad or harmful; and (e) any other thoughts you feel to be
pertinent.
In writing down these thoughts, please separate your thoughts into individual ideas to
be written down separately. An individual idea is one that, to the best of your judgment,
expresses only a single fact, value, good or bad feature, or thought.
Following this thought-listing procedure, subjects completed a brief

questionnaire of four Likert-type items measuring opinion on the generalspecialized education issue. Finally, subjects were instructed to look back at
the thoughts they had listed concerning general and specialized education and
to judge, for each individual thought, whether it was favorable to general or
specialized education and how favorable it was, on a three-point scale of
slightly favorable (1), moderately favorable (2), and very favorable (3).
Experiment II added an opinion pretest to the basic procedure; it consisted

of the four Likert-type items also used as the opinion posttest.
For Experiment III, subjects were given an additional judgment task, at the
end of the experiment, in regard to their listed thoughts. They were to assign
each thought to one of three categories: (a) those having their source in the
experimental materials (the introductory defnitions and the persuasive
communication); (b) modifcations of the experimental materials (such as
illustrations of, qualifcations of, and reactions to communicated arguments);
and (c) ideas not traceable to the experimental materials. These three
categories are to be identifed here, respectively, as (a) externally originated,
(b) recipient-modifed, and (c) recipient-generated cognitions. The aim of
Experiment III was to compare subject coding of listed-thought responses into
these three categories with independent coding by judges; no control group
was employed.
15
In the last experiment in this series, two different communication conditions

were employed. One (group IVA) employed the same communication used in
the previous experiments while the other (group IVB) was modifed by adding
material acknowledging opposing arguments (ones favoring general education)
and refuting them where possible (cf. Hovland et al., 1949, Ch. 8). Subjects in
Experiment IV also performed the additional categorization of their listed
thoughts that had been requested of subjects in Experiment III.
For Experiments I, II, and III the classifcation of thoughts into externally
originated, recipient-modifed, and recipient-generated categories was
performed by judges who were able to agree on these classifcations for 85% of
their judgments. This extent of agreement was not considered entirely
satisfactory, especially when it was found, in Experiment III, that judges
classifcations agreed with subjects classifcations for only 62% of judgments
(chance agreement would be 33 %). Since subjects were able to perform the
judging task with relative ease, it was decided that it would be most
satisfactory to use subject self-scoring for subsequent groups.
The data for the four experiments were frst examined in terms of the
quantities of thoughts as distributed among the three categories externally
originated, recipient-modifed, and recipient-generated. These data are given
in Fig. 2. Judges categorizations were used for Experiments I and II; for
Experiments III and IV, subjects own categorizations were used. The most
signifcant feature of the data summarized in Fig. 2 is that the recipientgenerated category accounted for the majority of thoughts listed by
communication- receiving subjects despite the fact that their thoughts were
tapped
Externally originated thoughts
I II ETA TZB
Recipient modified thoughts
n m HA EZ:B I n TZ
TZ
166.
COGNITIVELEARNING
Recipient generated thoughts
I1
I I n HA EB Communication receiving groups

12
Control groups
FlG. 2. Mean quantities of listed thoughts in three categories. (Numbers of subjects are as
follows: communication-receiving groups: I (48), II (48), III (68), IVA (26), IVB (28); control groups: I
(16), II (16), IV (13).) (Greenwald & Cullen, unpublished data.)
immediately after reading a communication containing a dozen distinct

relevant thoughts that could have been listed.2
Subjects judgments of the position supported by each of their listed

thoughts were used to calculate an index summarizing the thoughts content
on the general-specialized education issue. It will be recalled that each
thought was weighted according to its degree of support for the position it
supported. The index was calculated by subtracting the sum of weights for
thoughts favoring general education
2The quantities of listed thoughts in the externally originated and recipient-modifed categories
for control groups were greater than zero because control subjects could and did list or react to
material contained in introductory defnitions of the concepts of general and specialized education.
17
FlG.
3.
Mean
directional
content of listed
thoughts inRecipient modified thoughts
three categories
and
mean
posttest
attitude responses. (The extreme scores on the three category indexes are + 1.00 and 1.00; for the
posttest
attitude
responses,
the
extremes
are
+2.50
and 2.50.)
(Greenwald
&
Cullen,
unpublished
data.)
from the
those
Fbsttest attitude responses
sum
for
favoring
specialized
education,
then
dividing this
difference by
the sum
of
weights
for
all
thoughts.
This
calculation
I I M BZA EBI
H 152
was doneCommunication receiving groups
separately
Control groups
for
thoughts in
each of the three categories externally originated, recipient-modifed, and
recipient-generated. Figure 3 gives the mean values of these directional
content indexes for the various groups in the present series of experiments. For
comparison, the mean posttest opinion questionnaire responses are given at
the bottom of Fig. 3. The effects of the communications can be seen in the
generally more positive index values and posttest opinion scores for
communication groups than for controls. The
186.
COGNITIVELEARNING
signifcance levels for these comparisons, combining over Experiments I, II, and
IV (no control group was used in Experiment III) were: externally originated
category, p < .02; recipient-modifed category, p < .02; recipient-generated
category, p < .07; and posttest opinion, p < .001 (all one-tailed).
It may be noted that the communication received by subjects in group IVB

which differed from the others in that it acknowledged opposing arguments
produced the most favorable posttest opinion questionnaire responses and the
most positive responses in the recipient-generated category, while producing
the least positive responses in the externally originated category. This pattern
of fndings is suggestive of the importance of recipient-generated cognitions,
relative to externally originated ones, in persuasion.
Figure 4 gives for each group the correlations with posttest opinion for the
three category directional content indexes as well as for one based on all
thoughts combined. It may be noted that the correlations involving the
recipient-generated category index were quite high relative to those for the
externally originated category. This suggests, once more, the importance of
recipient-generated cognitions in the recipients attitude structure.3
The correlations with posttest opinion for the recipient-generated category

index and for the index based on all thoughts combined were sufficiently high
to suggest that the thought-listing procedure used in the present experiments
might be very useful as a measure of opinion. A very desirable aspect of the
thought-listing procedure is that it is applicable to virtually any attitude issue
without necessity for time consuming scaling and item selection procedures.
Research currently underway, being conducted by Cullen, is exploring the
reliability, validity, and sensitivity-to-change properties of the thought-listing
3Split-half reliability of the externally originated category index (.74) was actually slightly higher
than that for the recipient-generated category (.68), which indicates that the difference between
correlation magnitudes could not be attributed to relative unreliability of the externally originated
category index. However, it must be noted that the externally originated material was nearly
uniformly favorable to the specialized education position; thus, it is possible that range restriction
for the externally originated category index was responsible, in part, for the low-magnitude
correlations obtained between externally originated thought content and posttest opinion.
19
procedure in comparison with more traditional attitude scaling procedures.

(See experiment reported on pp. 163165 for an illustration of the use of the
thought-listing procedure as an opinion measure.)
Table 1 gives the additional correlational data that were obtained
206.
COGNITIVELEARNING
>
c
All thoughts combined
Externally originated 1hough1s
o
Recipient modified thoughts
o
2?
<p
e
o
-o
21
226.
COGNITIVELEARNING
.80
.60
.40
,20
.00
lll.l III
I Iff
Control groups
I I I KA IB
Communication receiving groups
FIG. 4. Correlations with posttest attitude for four listed thought directional content indexes.
(Greenwald & Cullen, unpublished data.)
TABLE 1
CORRELATIONS OF LISTED THOUGHT CONTENT INDEXES WITH PRETEST AND POSTTEST ATTITUDE MEASURES"
23
"Experiment II Data; N = 48.

^Pretest attitude has been partialed out of correlations with posttest attitude. c p < .01.
from Experiment II by virtue of the use of a pretest opinion measure. On

the left side of the table, it may be seen that pretest opinion was positively
correlated with all of the directional content indexes, indicating that
pretest opinion was an important determinant of the content of cognitions
rehearsed in response to the persuasion situation. On the right side of the
table, correlations between the category directional content indexes and
posttest opinion are given, with pretest opinion partialed out. Signifcant
positive correlations were obtained only for the recipient-generated
category index and for the one based on all thoughts combined again
suggesting the importance of recipient-generated cognitions in
persuasion, relative to externally originated ones.
In summary, Cullens experiments provided a variety of evidence

indicating important involvement of cognitive responses to persuasion in
attitude structure. These fndings, especially those for recipient-generated
cognitive responses, were consistent with the present hypothesis that
rehearsal of cognitive responses to persuasive communications is an
essential mediator of cognitive attitude change.
PERSUASION AND RETENTION OF COGNITIVE RESPONSES
The preceding series of experiments examined attitude-relevant

cognitions that were recorded shortly after the receipt of a persuasive
communication. Robert Loves masters thesis research (in preparation),
conducted under the authors direction, examined recipients cognitions
during a persuasive communication situation. The aim of this experiment
was chiefly to assess correlations of persuasion effectiveness with
retention of communication content and retention of cognitive response
content, observed immediately following and one week after the initial
communication situation.
24
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
Loves sample was divided into two groups, one receiving a persuasive
communication on admitting Puerto Rico as the ffty-frst state (N = 33),
the other receiving a communication advocating popular election of the
Secretary of State (N= 35); both communications were adapted with minor
modifcations from ones used by Watts and McGuire (1964). Subjects frst
received an opinion pretest (Likert- type), following which the appropriate
communication was administered in printed form. Each communication
presented three main supporting arguments in separate paragraphs:
following each paragraph, four blank lines were provided for the subject to
write a one- sentence reaction to the main point of the paragraph (the
main point was underlined in the printed communication). This procedure
served to obtain a sample of cognitive responses rehearsed during the
persuasion situation.4 After the communication, subjects opinions were
again assessed, this time using the thought-listing procedure (see p. 157),
following which subjects were given an unexpected test for recall of (a) the
three main points of the communication and (b) the reactions that had
been written following each. The same subjects were recruited for an
ostensibly different experiment, one week later, at which time the opinion
posttest and the recall test were unexpectedly readministered.
The data were analyzed for partial correlations (pretest partialed out)
of posttest opinion with (a) the content of cognitive reactions to the
communication (number of favorable minus number of unfavorable
reactions), (b) retention of the main arguments of the communication
(number of arguments recalled; maximum = 3), and (c) retention of
cognitive reactions (number of favorable minus number of unfavorable
reactions recalled). These correlational results are given in Table 2. It is
apparent that the best predictor of the effect of the communications was
the measure of content of the cognitive reactions that were written during
the communication exposure (average of 4 rs = .52); the next best
predictor was the retention index for the cognitive reactions (average of 4
rs = .30); decidedly the poorest predictor was the retention measure for
the persuasive communication itself (average of 4 r s = .03). These results
4Note that this procedure explicitly encouraged rehearsal of recipients cognitive responses
in that subjects were instructed to produce thoughts and to spend time writing their thoughts.
This removes the present experimental situation a bit from the type of communication
situation involving only a one-way transmission from source to recipient. However, this was
unavoidable in the interests of obtaining usable information about the content of recipients
cognitions during persuasion. Moreover, the situation was analogous to another important
type of persuasion situation in which source and recipient are in face-to-face confrontation, the
source typically being interrupted by the recipients reactions to the communication.
25
strongly support the present contention that cognitive responses to

persuasion are important in mediating persuasion effects while retention
of communication content is not.
Supplementary data were obtained from subjects classifcations of the

thoughts they had listed for the opinion posttests into the categories of (a)
having originated in the persuasive message, (b) having originated in
written reactions to the persuasive message, and (c) traceable to neither
of these sources. Subjects written reactions were most often represented
in their listed thoughts (mean number = 2.26),
TABLE 2
PARTIAL CORRELATIONS OF COGNITIVE REACTION AND RETENTION MEASURES WITH IMMEDIATE AND DELAYED POSTTEST
OPINION
Note: These data are from a masters thesis by Robert E. Love (in preparation). Pretest opinion
was partialed out of the correlations reported in this table.
"p < .01, one-tailed.
6
p < .05, one-tailed.
while the communication content was least often represented (mean

number= 1.21). (An average of 1.90 thoughts was traceable to neither
source.) These fndings lend further support to the conclusion that
cognitive responses to the communications were more signifcant in
providing content for cognitive attitude change than were the communications themselves.
THE COGNITIVE LEARNING PROCESS IN PERSUASION
26
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
The chief theoretical aim of this chapter has been to establish the basis
for useful analysis of attitude change as a cognitive learning process. To
do this, it was necessary to focus on cognitive responses rehearsed during
persuasion situations. Specifc learning-theoretical topics, such as the
roles of incentives, reinforcers, and conditions of practice, have been
ignored for the moment.
At present, knowledge concerning the determinants of learning of

attitude-relevant cognitions is quite limited. Certainly, much is known
theoretically about verbal learning, including learning of meaningful
material (see McGeoch & Irion, 1952; Woodworth & Schlosberg, 1954).
However, the particular variables involved in attitude-relevant learning
for example, covert rehearsal, preexisting attitudes, prior familiarity
with information, and comprehension of persuasive messages are not
well understood in learning-theoretical terms. Number of presentations of
a persuasive argument is about the only variable that is unequivocally
established as a determinant of argument retention (see, for example,
Greenwald & Sakumura, 1968; Jones & Kohler, 1958; Levine & Murphy,
1943; Waly & Cook,
1966) . Until quite recently, another widely accepted principle of

learning of persuasive arguments was that audiences would selectively
attend to and remember information consonant with their pre-existing
attitudes. Several studies had demonstrated effects of this nature (e.g.,
Jones & Kohler, 1958; Levine & Murphy, 1943). Recent attempts to
replicate this phenomenon (Greenwald & Sakumura, 1968; Waly & Cook,
1966) have met with absolutely no success, so that the phenomenon of
selective learning of attitude- consonant information must currently be
regarded as of dubious validity.
Other determinants of attitude-relevant learning that have been

implicated by empirical research are information utility (Jones &
Aneshansel, 1956) and novelty (Greenwald & Sakumura, 1968). Since
27
utility and novelty are variables known to increase attention to

persuasive information,5 it appears likely that their effects on learning of
persuasive arguments may be mediated by these attentional effects
rather than by any direct role in the learning process.
The fndings of the studies just mentioned, and of other relevant studies
not cited here, do not require the supposition that any variable other than
duration of exposure (i.e., attention) to persuasive information is a
determinant of information learning (cf. Cooper & Pantle, 1967). It seems
likely also that comprehension of information is a determinant of
retention (see Fitzgerald & Ausubel, 1963); however, minimal evidence is
available concerning this relationship. Additionally, conditions of practice,
particularly distribution of practice over time, should be expected to affect
learning of persuasive information in much the same manner that they
affect learning of other verbal material; again, little pertinent evidence is
available, although the literature concerned with primacy and recency
effects in persuasion (see Rosnows chapter in this volume, and references;
also Anderson & Hubert, 1963) may be interpreted in terms of the

conditions-of-practice variable.
The effects of rewards and punishments occurring in the persuasion

situation are certainly relevant to theoretical interpretations of cognitive
learning. However, although effects of rewards and punishments on
attitude measures have frequently been demonstrated, the processes
underlying such effects are poorly understood. Competing explanations in
terms of classical conditioning, instrumental learning, dissonance
reduction (Brehm & Cohen, 1962), and attention mechanisms (Janis &
Gilmore, 1965), all can be justifed by appeal to portions of the relevant
literature. (The reader will fnd extensive discussion of reward and
5Brock, T. C., Albert, S. M., & Becker, L. A. Familiarity, utility and supportiveness as
determinants of information receptivity, (in preparation)
28
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
punishment effects, interpreted in terms of conditioning processes, in the

chapters by Lott and Lott, Rosnow, Staats, and Weiss in this volume.)
The procedure of having subjects actively rehearse their own persuasive

arguments was found by Greenwald and Albert (1968; also summarized
earlier in this chapter) to produce substantial enhancement of argument
retention as well as noticeable self-persuasion. At the moment, it is
unknown whether these effects were due to enhanced original attention to
the improvised arguments or to other factors. Nonetheless, the focus on
learning of persuasive arguments actively rehearsed in a persuasion
situation has provided the basis for presently reasserting the importance
of cognitive learning in persuasion. Thus, the study of determinants of
persuasive-argument learninga problem area in which, to summarize
the present brief literature review, current ignorance is considerable
can be justifed not only as an interesting exercise in learning theory, but
in terms of its practical value in interpreting the basis for effective and
durable persuasion.
CONCLUSION-COGNITIVE
PROCESS,
LEARNING, AND ATTITUDE CHANGE
COGNITIVE
The present program of research set out to establish the legitimacy of a

conception of attitude change through persuasive communication as, at
least in part, a cognitive learning process. In the course of doing this, the
obtained experimental evidence repeatedly indicated that the effects of
persuasive communications are strongly mediated by the content of
attitude-relevant cognitions elicited (and thus rehearsed and learned)
during the persuasion situation. The analysis of determinants of cognitive
response content may very well require explanatory principles outside the
scope of the learning principles with which the present research was
concerned. There is, however, no lack of theories in the cognitive
integration area that might be applied to the analysis of determinants of
cognitive response content. Cognitive consistency theories, for example,
can be used to predict that cognitive responses to persuasion will be
consistent with preexisting attitude-relevant cognitions. The assimilation-
29
contrast approach (Sherif & Hovland, 1961) predicts that the individual
reacts with favorable cognitions to persuasive statements within his
latitude of acceptance and with unfavorable cognitions to statements
outside his latitude of acceptance. Reactance theory (see Brehms chapter
in this volume) predicts that unfavorable cognitive reactions will occur in
persuasive situations of a coercive nature. Brocks commodity analysis
(see his chapter in this volume) predicts favorability of cognitive reaction
to persuasive information to be a decreasing function of the perceived
availability of the information. Functional analyses of attitude change
(Katz, 1960; Sarnoff, 1960; see also Barons chapter in this volume)
provide more complex principles that might be used to predict cognitive
reactions to persuasion given knowledge about the motivational basis of
existing attitude structures.
In light of these observations, it would appear fruitful to approach the

study of persuasive communication effectiveness with a combination of
cognitive process theory and learning theory. The effects of independent
variable manipulations in persuasion situations, such as credibility,
organization of arguments, communication medium, etc., could be studied
simultaneously in terms of their effects on the content of cognitive
responses to persuasion and on learning of persuasive information.
Existing data, noted at the outset of this chapter, suggest that most of the
traditional independent variables of persuasion do not signifcantly affect
retention of persuasive information. (Their effects on attitude presumably
are mediated strictly through their effects on cognitive responses to
persuasion.) The absence of retention effects of traditional persuasion
variables may, however, only reflect the fact that attitude change
researchers have been more interested in manipulating variables that
affect acceptance of persuasive information than ones that affect attention
to and retention of persuasive information. In the combined application of
cognitive integration theory and learning theory to persuasion, cognitive
integration theory should offer an account of the processes involved in
acceptance of persuasion while learning theory should seek to account for
persistence of induced changes through learning and retention processes.
Neither of these areas of theory, when considered alone, can currently be
expected to provide a complete account of the processes by which attitudes
are formed or lastingly changed in response to persuasive
30
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
communications.
References
Anderson, N. H., & Hubert, S. Effects of concomitant verbal recall on order effects in
personality impression formation. Journal of Verbal Learning and Verbal Behavior, 1963,
2, 379-391.
Brehm, J. W., & Cohen, A. R. Explorations in cognitive dissonance. New York: Wiley, 1962.
Brock, T. C. Communication discrepancy and intent to persuade as determinants of
counterargument production. Journal of Experimental Social Psychology, 1967, 3, 296309.
Campbell, D. T. The generality of a social attitude. Unpublished doctoral dissertation,
University of California, Berkeley, 1947.
Cooper, Elaine H., & Pantle, A. J. The total-time hypothesis in verbal learning. Psychological
Bulletin, 1967, 68, 221-234.
Fitzgerald, D., & Ausubel, D. P. Cognitive versus affective factors in the learning and
retention of controversial material. Journal of Educational Psychology, 1963, 54, 73-84.
Greenwald, A. G. An amended learning model of persuasion. Paper read at American
Psychological Association, Washington, D. C., September 1967.
Greenwald, A. G., & Albert, Rosita D. Acceptance and recall of improvised arguments.
Journal of Personality and Social Psychology, 1968, 8, 31-34.
Greenwald, A. G., Albert, Rosita D., & Cullen, Dallas M. Persuasionas
a function
of
communication content learning. Unpublished manuscript, Ohio State Univ., 1968.

Greenwald, A. G., & Sakumura, J. S. Attitude and selective learning: Where are the
phenomena of yesteryear? Journal of Personality and Social Psychology, 1967, 7, 387-397.
Hovland, C. I., Janis, I. L., & Kelley, H. H. Communication and persuasion. New Haven: Yale
University Press, 1953.
Hovland, C. I., Lumsdaine, A. A., & Sheffield, F. D. Experiments on
masscommuni
cation. Princeton, N. J.: Princeton University Press, 1949.

Insko, C. A. Primacy versus recency in persuasion as a function ot the timing ot arguments
and measures. Journal of Abnormal and Social Psychology, 1964,69,381-391.
Insko, C. A. Theories of attitude change. New York: Appleton, 1967.
Janis, I. L., & Gilmore, J. B. The influence of incentive conditions on the success of role
playing in modifying attitudes. Journal of Personality and Social Psychology, 1965, 1, 17-
31
27.
Journal of fcraonaiity and Social P5>vbotogy
1987, Vol. 52. No. 5,881-489
Janis, I. L. & Terwilliger, R. F. An experimental study of psychological resistances to feararousing communications. Journal of Abnormal and Social Psychology, 1962, 65, 403410.
Jones, E. E., & Aneshansel, J. The learning and utilization of contravaluant material.
Journal of Abnormal and Social Psychology, 1956, 53, 27-33.
Jones, E. E., & Kohler, R. The effects of plausibility on the learning of controversial
statements. Journal of Abnormal and Social Psychology, 1958, 57, 315-320.Copyright 1987 by the
American Psychoiofical Asaodation. Inc.
<5022-3514/87/S00.75
The Cognitive-Affective Crossfire: When SelfConsistency Confronts Self-Enhancement

William B. Swann, Jr., John J. Griffin, Jr., Steven C. Predmore, and
Bebe Gaines
University of Texas at Austin

Self-consistency theory assumes that people want others to treat them in
a predictable manner. Selfenhancement theory contends that people want
others to treat them in a positive manner. We attempted to help reconcile
the two theories by testing the hypothesis that peoples cognitive
responses conform to self-consistency theory and their affective
responses conform to self-enhancement theory.
We presented individuals who possessed either positive or negative self-concepts with either
favorable or unfavorable social feedback. We then measured cognitive reactions to the feedback
(e.g., perceived self-descriptiveness) and affective reactions to the feedback (e.g., mood states).
Cognitive responses were primarily driven by the consistency of the feedback and affective
responses were controlled by how enhancing it was. We propose that conceptualizing cognition
and affect as partially independent mental systems helps resolve some long-standing paradoxes
regarding peoples responses to selfrelevant social feedback
.
When we undertake to cure a patient, to

free him from the symptoms of his
malady, he confronts us with a vigorous,
tenacious resistance that lasts during the
whole time of the treatment This is so
peculiar a fact that we cannot expect
much credence for it. . . . Just consider;
this patient suffers from his symptoms
and causes those about him to suffer with
him . . . and yet he struggles, in the very
interests of the malady, against one who
would help him. How improbable this

assertion must sound! (Freud, 1921, p.
248)
Improbable perhaps, yet Freuds
assertion has fared well over the years.
Self-consistency theorists, for example,
contend that much like Freuds
patients, people with negative self-
32
6.
COGNITIVE
LEARNING AND
concepts undermine opportunities to

better themselves by engaging in
cognitive and behavioral activities that
perpetuate their selfviews. Yet, as
Freud feared, such contentions have
raised a fair number of eyebrows. Selfenhancement theorists, for example,
have rejected self-consistency theory by
arguing that in fact people with
negative
self-concepts
are
highly
motivated to improve their self-views.
This raises an important question: Does
self- consistency or self-enhancement
theory offer a more compelling
characterization of human nature?
The research and preparation of this article

were supported by National Institute of
Mental Health (NIMH) Grant MH-37598 and
NIMH Research Scientist Development
Award MH-00498 to William B. Swann, Jr.,
who was a Visiting Fellow at Princeton
University during the preparation of the
article.
We are grateful to Solomon Asch, Joel
Cooper, John Dailey, Scott Dickman, Daniel
Gilbert, Edward Jones, and Nancy Hazen for
their comments on early versions of the
manuscript, to Sally Gaines for her help
during the empirical phases of the
investigation, and to John Loeh- lin for
statistical advice.
Correspondence concerning this article
should be addressed to William B. Swann, Jr.,
Department of Psychology, University of
Texas, Austin. Texas 78712.
Self-Consistency Versus SelfEnhancement
COGNITIVE
RESPONSE
are therefore motivated to preserve

their self-views, which they do by
thinking and behaving in ways that
perpetuate their conceptions of self.
Since
Leckys
(1945)
initial
statement, several theorists have
extended his formulation by identifying
several specifc cognitive and behavioral
strategies
through
which
people
stabilize their self-views (e.g., Secord &
Backman,
1965:
Swann,
1983).
Furthermore, some (e.g., Epstein, 1983;
Swann, 1983) have suggested that these
activities are mediated by a highly
general, cognitively based preference for
stimuli that are predictable, familial;
stable, and uncertainty reducing. From
this vantage point, people strive to
acquire information that confrms their
selfconceptions because their thought
processes are structured so that
confrmatory
information
seems
especially trustworthy, diagnostic, and
accurate.
Self-enhancement theory is based
loosely on various personality theories
(e.g., Homey, 1937; Rogers, 1961) and
learning theory. Its central assumption
is that people are motivated to increase
their feelings of personal worth (e.g.,
Epstein, 1973; Tessei; 1985). In
addition, self-enhancement theory (at
least in its most logically consistent and
popular form) assumes that because
people with negative self-concepts lack
self-esteem
more
than
their
counterparts, they will compensate for
their lack of self-esteem by trying to
enhance their self-views more than will
their high self-esteem counterparts
(e.g., Jones, 1973).
Self-consistency theory can be traced

to the writings of Prescott Lecky (1943).
His central assumption was that selfconceptions are critical for survival
because they enable people to predict
and control the nature of social reality Both theoretical formulations predict
(e.g., Epstein, 1973: Mead, 1934). People that people with positive self-views work
to maintain such views, albeit for

different reasons. However, the two
theories make competing predictions
regarding people with negative self-
33
views. That is, self-consistency theorists

assume that individuals with negative
self
concepts prefer negative feedback because it is predictable; seif- enhancement

theorists assume that such individuals prefer positive feedback because they want
to think well of themselves.
Given that self-consistency and self-enhancement theory make very different
predictions regarding responses of individuals who possess negative self-views, it
appears that one theory could be discounted by simply examining the relevant re sponses of such individuals. This is not so. Several decades of research have
produced mixed results, with some studies favoring self-consistency theory and
others favoring self-enhancement theory (for reviews, see Jones, 1973; Shraugei;
1975; Swann, 1985).
Shrauger (1975) attempted to bring order to this confusing state of affairs by
suggesting that some dependent variables tended to produce consistency effects
and others tended to produce enhancement effects. In particular, measures of
certain cognitive processes (e.g., recall, perceptions of the self-descriptiveness of
feedback)6 seemed to support self-consistency theory. In contrast, measures that
had a more affective flavor (e.g., pleasure or disappointment with feedback)
seemed to support the self-enhancement position. An intriguing implication of
Shraugers proposal was that people with negative self-concepts would have rather
ambivalent reactions to unfavorable feedback. Although such individuals might
value such feedback on a cognitive level, they would also fnd it affectively
abhorrent.
Although Shrauger's (1975) hypothesis was reasonably consistent with the
existing data, workers in the area were slow to accept it. One problem was that
Shrauger provided little theoretical justifcation for his notion that cognitive and
affective responses were independent. In addition, he was unable to marshal direct
empirical evidence for his hypothesis. We shall focus on these two shortcomings in
this article. We will deal frst with the empirical issues and leave the conceptual
issues for the General Discussion.
Evidence for the Cognitive-Affective Independence Hypothesis

The major problem with the evidence Shrauger (1975) cited in support of his
cognitive-affective independence hypothesis was that researchers had examined
6 After Shrauger (197 5), we use the term cognitive processes in a limited sense to refer to the
relatively analytical, controlled processes that are presumably indexed by ratings of selfdescriptiveness. As a result, when we use the term cognitive-affective crossfire, we are referring
to a conflict between the products of these analytical processes and the affective system rather
than to a conflict between the entire cognitive system and the entire affective system.
34
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
either cognitive or affective reactions; no one had examined both types of reactions
in the context of a single study. This introduced the possibility that procedural
differences other than the nature of the dependent variable could have accounted
for the conflicting results of those who examined cognitive versus affective
reactions.
Since Shraugers review, two published investigations have attempted to test his
hypothesis directly.7 Neither has offered strong support for his position. McFarlin
and Blascovich (1981) tapped cognitive responses by asking people (a) to indicate
their ability to perform a task and (b) to predict how well they would perform. The
measure of affect was problematic, however. Instead of providing participants with
feedback and measuring their affective reactions, the researchers asked them to
indicate their preferences regarding future performances.
Given the absence of a direct measure of affective reactions to feedback, the
McFarland and Blascovitch (1981) study is at best tangentially relevant to the
hypothesis that people with negative self-views value unfavorable feedback on a
cognitive level yet fnd it affectively abhorrent. Therefore, it may not be telling that
some of their results contradicted Shraugers hypothesis. Most important, contrary
to the cognitive-affective independence notion, the affective measure was as closely
associated with one of the measures of cognition (r * .62) as the two mea sures of
cognition were to one another (r = .67).
The results of a feld investigation by Moreland and Sweeney (1984) are
potentially more relevant to Shraugers hypothesis. These investigators assessed
the relation between scores on a midterm exam and students subsequent affective
states and cognitive appraisals of the exam. The fndings were complex, but it is
faidy clear that both the cognitive and affective responses supported the selfenhancement position. Contrary to Shraugers hypothesis and consistency theory,
participants with low self-esteem generally regarded positive feedback as more
self-descriptive than negative feedback.
Nevertheless, there is a good reason why Moreland and Sweeneys (1984)
measures of cognitive reactions may have failed to support self-consistency theory.
Consider that most college students possess relatively high self-esteem. Given this,
Moreland and Sweeneys procedure of identifying low-self-esteem individuals by
performing a median split may have classifed people who were in reality high in
self-esteem as low in selfesteem. Such misclassifed high-self-esteem individuals
would reject the negative feedback as being nondescriptive of self: not in the
service of self-enhancement strivings, as the authors concluded, but in the service
of self-consistency tendencies. For this reason, the pattern of cognitive responses
that Moreland and Sweeney interpreted as supportive of self-enhancement theory
may have in reality supported self-consistency theory.
7 A field investigation (N = 22) by Losco-Szpiler and Epstein (1978) also addressed this
issue, but a detailed analysis ot' the results was unavailable as of this writing.
35
In short, more than a decade has passed since Shrauger (1975) presented his
important hypothesis, and the verdict is still out. What is needed is a study in
which (a) people who possess negative (Mr positive self-views receive feedback that
is clearly consistent or inconsistent with their self views and (b) cognitive and
affective reactions are measured by instruments capable of discriminating the two.
Toward this end, we recruited individuals who scored in the upper or lower 20th
percentile of a large sample on a measure of self-esteem. We pre sented favorable
feedback to some individuals and unfavorable feedback to others. We then
measured, in counterbalanced order, cognitive and affective reactions to the
feedback.
Cognitive reactions included participants perceptions of the accuracy of the
feedback, competence of the evaluator; diagnosticity of the evaluation technique,
and attributions regarding the cause of the feedback. We used a mood measure to
tap affect because we believed that such a measure would provide us with a
relatively pure index of affect. We also measured attraction to the rater so that we
would be able to compare our fndings to those of earlier investigators who
assessed this variable. As have previous workers (e.g., Shrauger, 1975), we believed
that this measure might tap both cognitive and affective reactions because
attraction to a rater might be influenced by ones perception of that rater's
credibility as well as the mood induced by that rater.
Our major prediction was that cognitive reactions to the feedback would be
relatively independent of affective reactions. Specifcally; we anticipated that
cognitive reactions would be based on the degree to which the feedback confrmed
participants self-views, with confrmatory feedback regarded as more accurate,
diagnostic, and so on. In contrast, we expected that affective reactions would be
based on the favorability of the feedback, with favorable feedback producing more
positive mood states than unfavorable feedback.
Self-Concept
Method Participants and Measure of
Participants were 48 male and 58 female undergraduates who took part in the investigation
for credit in their introductory psychology course. Five participants were deleted because they
were suspicious of the experimental procedure.
Participants were drawn from a large sample of students who completed Helmreich, Spence,
and Stapps (1974) Texas Social Behavior Inventory (TSBI) during a pretest session at the
beginning of the semester. This scale emphasizes social self-esteem (e.g., I have no doubts
about my social competence, I am not likely to speak to people until they speak to me.'
Scores on the TSBI could range from 16 to 80; the actual range was 25 to 80. We classified
individuals who scored below the 20th percentile (51) as ncgative-seif-concept individuals and
those who scored above the 80th (66) percentile as positive-self-concept individu als.
Experimenters remained unaware of participants' TSBI scores throughout the experimental
procedure. We also measured the certainty of participants self-views. This variable had no
effects on the dependent variables and will not be discussed further.
36
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
Procedure
Cover story and speech. A female experimenter introduced the experiment as an
investigation of the accuracy of first impressions formed on the basis of nonverbal information
only. She explained that two people would be involved in the experiment, the participant and
an evaluator. The first step would be for the participant to deliver a speech. The evalu ator
would watch the participant deliver the speech through a soundproof, one-way mirror,
allegedly to prevent the evaluation from being influenced by what the participant said. The
evaluator would then awes* the participant Shortly thereafter; the participant would examine
the evaluators assessment and judge its accuracy. Comments made by participants during
debriefing revealed that they found this cover story entirely plausible.
The speech consisted of several unremarkable excerpts from Desmond Morriss novel The
Naked Ape. After giving the speech, the participant waited for 5 min while the evaluator
ostensibly prepared his or her evaluation (evaluators were always alleged to have the same sex
as the participant). The experimenter then entered with a handwritten evaluation that had been
prepared in advance. In the favorable feedback condition, the feedback asserted that the
participant was socially skilled:
From the way he (she) looked reading this speech this person seems socially self-confident
I'd say he (she) probably feels comfortable and at ease around other people he (she)
doesnt know very well.
He (she) seems to have little doubt of his (her) social competence. Thats about all I could
tell about him (her).
In the unfavorable feedback condition, the feedback was simply the negation of that used in
the favorable feedback condition:
From the way he (she) looked reading this speech this person doesnt seem real socially
self-confident Id say he (she) probably feels somewhat uncomfortable and anxious
around other people he (she) doesnt know too well. He (she) seems to have some doubts
about his (her) social competence. Thats about all I could tell about him (her).
Cognitive and affective reactions to the feedback. Immediately after reading the feedback
participants completed two series of questionnaires in counterbalanced order. One set of
questionnaires tapped their cognitive reactions to the feedback and another assessed their
affective reactions to the feedback. Three additional items measured attraction to the evaluator
We assessed four distinct cognitive reactions. Five items indexed perceived accuracy of the
feedback, five items assessed perceived competence of the evaluator, three items tapped
perceived diagnosticity of the rating technique, and two items measured participants
attributions regarding the feedback.
We assessed affective reactions with a measure of mood, Zuckerman and Lubins (1965)
Multiple Affect Adjective Checklist (maacl). This instrument is designed to measure depression,
anxiety and hostility.
37
After participants completed all measures of cognition and affect they were thoroughly
debriefed, thanked, and dismissed.
Item analysis. Items were deleted from our measures if including them in a given scale
diminished the internal consistency of that scale. According to this criterion, we deleted single
items from the perceived competence, perceived diagnosticity, and liking for the evaluator
scales. The reliability analyses reported in the Appendix indicate that all of our scales displayed
high levels of internal consistency. The precise wording of all measures included in the
analyses can also be found in the Appendix.
Observer ratings. To obtain a rough index of the veridicality of our participants self-ratings,
we had two observers watch participants give their speech and rate them on the following
bipolar trait scales: unsociable-sociable, socially confident- unconfident, socially awkwardpoised, shy-outgoing, self doubting-self assured, socially competent-incompetent, cold-warm,
nervous-at ease. Observers also attempted to guess whether participants had high or low selfesteem.
Results and Discussion

We examined the impact of the self-concept, feedback, and order of presentation
variables on participants cognitive and affective reactions. We then assessed
covariation between the cognitive and affective measures by submitting them to a
factor analysis. Finally, we examined the impact of self-concept on the ratings of
observers. The effects of sex of participant are not discussed as this variable did
not qualify any of the fndings reported here.
Impact of Feedback, Self-Concept, and Order on Cognitive Reactions

All measures of cognitive reactions were entered into 2 (feedback: favorable,
unfavorable) X 2 (self-concept: positive, negative) X 2 (order cognitive frst,
affective frst) least squares analyses of variance (ANOVAS). Out primary prediction
was that there would be an interaction between self-concept and feedTable 1
Cognitive Reaction as a Function of Self-Esteem and

Feedback Positive self-concept Negative self-concept
Favorable Unfavorable Favorable Unfavorable Cognitivefeedback

feedback
feedback
feedback
measure
{n - 22)
(n = 18)
(n = 26)
(n = 32)
Note. The higher the mean, the greater the perceived accuracy of
the feedback (range = 5-45), the greater the perceived competence
of the rater (range = 4-36), the more diagnostic the technique (range
= 2-18), and the greater the attribution to self (range = 1-9) or other
(range =
1-9).
back, such that those with positive self-concepts would regard favorable feedback
as especially self-descriptive (i.e., accurate, diagnostic, delivered by a competent
rater, reflective of self) and those with negative self-concepts would regard
unfavorable feedback as especially self-descriptive.
38
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
The means displayed in Table 1 support our predictions. The Self-Concept x

Feedback interaction was reliable for all fve cognitive measures, including
accuracy, /"(1, 90) = 75.28, p < .001, competence of the evaluator, /^l, 89) = 55.36, p
< .001, diagnosticity of the evaluation technique, F(l, 90) = 15.33. p < .001, selfattribution, F[ 1, 90) = 20.66, p < .001, and other attribution. F( 1,90) - 14.85, p < .
001,3
Simple effects analyses revealed that positive-self-concept individuals who
received favorable feedback regarded it as more accurate, F{ 1, 39) = 122.60, p < .
001, the rater as more competent, /"(I, 38) = 93.77. p < .001, and the technique as
more diagnostic, F{\. 38) = 21.56, p < .001. These individuals were also inclined to
attribute the favorable feedback to themselves, F(l, 39) - 21.56, p < .001, and not to
characteristics of the evaluator, F{ 1, 39) = 5.48. p < .03. Negative-sejf-concept
individuals displayed precisely the opposite tendency. That is, negative-selfconcept individuals who received unfavorable feedback regarded it as particularly
accurate, i*U, 56) * 12.77, p < .001. and the evaluator as particularly competent,
F(l, 56) = 6.26, p < .02, and they were not inclined to attribute the feedback to
characteristics of the evaluator, F(l. 56) = 14.46, p < .001. Relative to positive-selfconcept individuals, these individuals also displayed nonreliable tendencies to
regard the technique as more diagnostic and to attribute the feedback to themselves.
These data support the notion that peoples cognitive reactions to feedback are
driven by a concern with the consistency of the feedback with their selfconceptions. Closer examination indicated that cognitive reactions were also
influenced by the sheer positivity of the feedback. That is, ignoring the self-concept
variable, there was an overall tendency for participants to believe that the
favorable feedback was more accurate, diagnostic, and so forth, than the
unfavorable feedback. The main effect of feedback was reliable for all the measures
save the measure of other attribution: accuracy, F{ 1,90) * 17.85, p < .001, competence of the evaluatoi; F( 1, 89) = 16.50, p < .001, diagnosticity of the evaluation
technique, F(i, 90) - 7.15, p < .001, and self-attribution, F( 1,90) = 11.48, p < .001.
To assess the relative importance of the consistency and favorability of the
feedback, we compared the percentage of variance accounted for by the SelfConcept x Feedback interaction versus the main effect of feedback. The
interaction effect accounted for 88%, 85%, 76%, 56%, and 74% of the systematic
variance on the measures of accuracy, competence, diagnosticity, self-attribution,
and evaluator attribution, respectively. The feedback effect accounted for only 5%,
10%, 18%, 18%, and 10% of the systematic variance on these same measures. Of
course, one must be careful in generalizing the results of this analysis, because
characteristics of our experiment or subject population may have influenced the
outcome.
Yet if cognitive responses are only sensitive to the consistency of the feedback,
why should there have been any main effect of feedback at all? One possibility is
39
that the self-views of positive as compared to negative self-concept individuals

were more closely matched to the consistent feedback. That is, despite the fact
that our negative-self-concept participants scored in the lower 20% of our pretest
sample, their average score was 46: just 2 points below the theoretical midpoint of
the scale (range = 16-80). The self-views of negative-self-concept participants were
therefore negative in a relative sense only. In contrast, the average score of our
positive-self-concept individuals was 70. Clearly, the self-views of these
individuals were positive in an absolute as well as in a relative sense.
There were no main or interactive effects of the order variable on any of the
cognitive measures.
Affective Reactions to the Feedback

> All measures of affective reactions were entered into 2 (feedback: favorable,
unfavorable) x 2 (self-concept: positive, negative) x 2 (order cognitive frst,
affective frst) least squares anovas. Our major prediction was that positive- and
negative- self-concept individuals alike would feel better after receiving favorable
feedback as compared to unfavorable feedback. This was the case. The data in
Table 2 indicate that participants who received unfavorable as compared to
favorable feedback were more depressed, F{1, 92) = 15.41,/) < .001, hostile, /"(I, 92)
= 8.8 Up<.004, anxious,^ 1,92) = ll.97,p< .001, and experienced less negative affect
overall, F{ 1,92) = 15.92, p < .001 (the last index was a composite measure
comprised of the depression, hostility, and anxiety scores). Participants were also
more attracted to the evaluator in the favorable feedback condition, F{ 1,86) =
43.86, p<. 001.
We also expected that the consistency of the feedback would not influence
affective reactions. The measures of mood supported this prediction. That is. none
of the mood measures showed an interaction between self-concept and feedback,
all Fs ns. There was, however, a reliable Self-Concept X Feedback interaction on
the attraction variable. F{ 1, 86) = 7.19. p < .01.
Table 2
3
The degrees of freedom for different dependent measures vary slightly because
participants occasionally failed to complete measures.
Affective Reaction as a Function ofSelf Esteem and Feedback

Positive self-concept Negative self-concept Favorable Unfavorable
Favorable Unfavorable
Note. The higher the mean, the more positive the overall affective
state (range = 0-89), the less depressed (range = 0-40), the less
hostile (range = 0-28), the less anxious (range = 0-21) and the greater
the attraction to the rater.
40
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
Although everyone preferred favorable evaluators to unfavorable ones, this

tendency was stronger among positive-self- concept individuals, F\l, 38) = 41.5, p
< .001, as compared with negative-self-concept individuals, F\I, 53) = 10.93, p < .
001. Even so, the interaction effect on the attraction variable accounted for only
9% of the systematic variance; in contrast the feedback effect accounted for 67% of
the systematic variance.
The analyses also revealed that participants emotional reactions to the
feedback were more polarized when the measures of affect were collected before
rather than after the measures of cognition. The means in Table 3 indicate that
order interacted with feedback on the measure of depression, F{ I, 92) = 4.90, p < .
03, hostility, F(l, 92) = 4.68, p < .04, overall negative affect, F( 1, 92) = 4.90, p < .
03, and attraction to the evaluator, ^1, 86) = 10.68, p < .002. A similar but
nonreliable patterp characterized the measure of anxiety (p < . 16).
Simple effects analyses revealed that when the affective measures occurred
frst, participants displayed a clear preference for the favorable evaluation and
evaluator on all fve measures of affect (all ps < .001). In contrast, when the
affective measures were collected second, the preference for favorable feedback
Table 3
Affective Reaction as a Function of Order and Feedback
Affective first
Cognitive first
Note. The higher the mean, the more positive the overall affective state,
the less depressed, the less hostile, the less anxious, and the greater the
attraction to the rater.
was consistently weaker. In fact, this preference was only reliable in the case of
the measure of attraction (p < .01).
Why did affective responses to the feedback vanish when they were measured
after the measures of cognition? One possibility is that the simple passage of time
diminished affective responses. Alternatively, the act of completing the cognitive
measures may have been critical. Although this is plausible, it was not that
completing the cognitive measures focused attention on the discrepancy between
affect and cognition, because even positive-self-concept individuals (for whom
there was no such discrepancy) displayed weaker affective reactions to the feedback after completing the cognitive measures.
Covariation Between the Measures of Cognition and Affect

To determine if the measures of cognition and affect were orthogonal, we
41
entered the measures of cognitive and affective reactions into a principle

components factor analysis with oblique rotation (Nie, Hull, Jenkins,
Steinbrenner. & Bent, 1975). Two factors emerged from the initial oblique
solution. The frst factor accounted for 47% of the variance and had an eigenvalue
of 4.24. The second factor accounted for 24.4% of the variance and had an
eigenvalue of 2.19. The eigenvalues for all other factors were less than 1.
The loadings for the frst two factors after rotation can be seen in Table 4. All of
the cognitive measures loaded heavily on the frst factor only, and all of the mood
measures loaded heavily on the second factor only. The sole measure that loaded
on both factors was the index of attraction to the evaluator. It is therefore not
surprising that the correlation between the frst and second factors was a modest
.24. This evidence of cognitive-affective independence is especially striking
when one considers that the cognitive and affective measures shared some method
variance because we used self-reports to tap affective states.
In summary, the factor analysis indicated that (a) the cognitive measures were
closely related to one another but were relatively independent of the affective
measures, (b) the affective measures were closely related to one another but were
relatively
Table 4
Oblique Factor Pattern Matrix After Rotation With Kaiser

Normalization
Note. A = 0.
independent of the cognitive measures, and (c) the attraction measure was a
hybrid measure that was related to both the cognitive and affective measures.
Observer Ratings
Observers were able to discriminate positive-self-concept individuals from their
counterparts. One way anovas revealed that observers rated positive-self-concept
participants as more sociable and self-confdent than negative-self-concept
individuals, i^l, 92) = 7.01 ,p < .01, A/s = 95.9 and 85.1, respectively. Observers
were also able to guess whether participants were high or low in self-esteem at an
above-chance level, /U, 92) = 5.66, p < .02.
42
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
General Discussion
Our fndings suggest that both self-consistency and self-enhancement theory
offer valuable insights into people's reactions to social feedback. For example, as
self-consistency theory suggests, participants with negative self-concepts indicated
that unfavorable feedback was more self-descriptive than favorable feedback. As
self-enhancement theory suggests, even though those with negative self-concepts
regarded unfavorable feedback to be quite accurate and self-descriptive, they were
more depressed, anxious, and hostile after they received it Our data therefore
provide strong support for Shraugers (1975) hypothesis that cognitive reactions to
social feedback conform to self- consistency theory and affective reactions conform
to self-enhancement theory.
Even so, our data raise questions regarding the assumptions underlying both
self-consistency and self-enhancement theory. One relatively minor question,
which is specifc to self-enhancement theory, concerns the fact that the affective
reactions of participants with positive and negative self-conceptions did not differ.
This is inconsistent with the strong form of self-enhancement theory, which
stipulates that relative to individuals with positive self-concepts, those with
negative self-concepts should be more pleased by positive feedback and more
displeased by negative feedback. Our data, together with the fact that there is
little defnitive support for the strong version (see Shrauger, 1975), suggest that it
may be time to opt for a weak form of self- enhancement theory in which people
with positive self-views prefer favorable feedback just as much as those with
negative self-views.
A more fundamental problem with both self-consistency and self-enhancement
theory is raised by our evidence that cognitive and affective reactions seemed
relatively independent. This fnding is problematic for both theories because both
subscribe to the assumption of psychological unity, which holds that a
superordinate cognitive system oversees all mental activity and resolves
inconsistencies between thoughts, feelings, and actions. Our fndings clearly clash
with the unity assumption in that our participants cognitive and affective
reactions seemed independent. That is, both the factor analysis and the fact that
cognitive reactions were more likely to persist over time than affective reactions
suggest that cognitive and affective responses are independent. More important
the overall pattern of data indicate that cognitive responses were based on the
subjective veridicality of the stimuli, such as the extent to which the feedback
was consistent with the persons self-views, and affective responses were based
amply on whether or not the feedback was threatening.
One implication of our fndings, then, is that both self-consistency and selfenhancement theorists should drop the unity assumption. Some have already
begun to do this. Swann (1987), for example, has suggested that the selfverifcation formulation (a variant of self-consistency theory) applies to cognitive
and behavioral responses but not affective responses. In light of the wide range of
responses that fall into these response classes, it may be necessary for theorists to
43
become even more specifc regarding the type of responses covered by their
theories.
Our evidence that cognitive and affective responses are independent raises
questions concerning why this might be the case. Recent work by dual- and
multiple-systems theorists (e.g., Epstein, 1983; Gazzaniga, 1985; Greenwald, 1982;
Izard, 1984; Tomkins, 1981; Wilson, 1985; Zajonc, 1980) may be relevant here.
This work suggests that the cognitive and affective systems are designed to
perform very different tasks. The cognitive system is presumably designed to
classify stimuli and analyze their logical properties and subjective veridicality. For
example, when social feedback is received it is frst classifed (e.g., favorable or
unfavorable to self). Then the feedback is compared to information about the self
stored in memory. If the feedback concurs with the information in memory, it is
accepted as self- descriptive; if not it is rejected.
Two characteristics of the decision process that the cognitive system uses are
especially noteworthy. First, analysis of the subjective veridicality of stimuli is
relatively time consuming because it entails searching memory and comparing the
stimulus with stored information. Second, decisions reached by the cognitive
system are only incidentally sensitive to the valence of feedback. That is, because
the cognitive system is concerned with how incoming feedback compares with
existing knowledge of self, the valence of the feedback matters only in that it determines whether it is classifed as consistent or inconsistent with the self.
In contrast, the affective system enables the organism to respond quickly to
events that pose an immediate threat to personal safety. This rather primitive
system reacts on the basis of relatively gross discriminations (i.e., threatening vs.
not threatening, favorable to self vs. unfavorable to self) and little or no analysis of
the subjective veridicality of stimuli. This system, then, trades precision for speed.
It may not perform highly sophisticated analyses of stimuli, but it reacts quickly.
The major difference between the cognitive and affective systems, then, is how
they improve the organisms chances of survival. The cognitive system achieves
this end through a systematic analysis of the subjective veridicality of stimuli; the
affective system does so by quickly recognizing threats to safety and spurring the
organism to action. To be sure, the distinction between the two systems is not
clear-cut (e.g., Epstein, 1983). For example, affective responses are dependent on
some rudimentary cognitive analyses of stimuli, enough to allow the organism to
recognize the stimuli (e.g., Lazarus, 1984; see also discussions by Bimbaum, 1981;
Mellers, 1981; Zajonc, 1980; 1984). In our opinion, however, the fact that some
interaction may occur between the cognitive and affective systems does not diminish the utility of conceptualizing them as relatively independent systems with
distinct capabilities and agendas.
An important implication of the cognitive-affective independence notion, of
44
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
course, is that it suggests the possibility that people may be caught in crossfres
between the two. In our study, for example, participants with negative self-views
who received unfavorable feedback found such feedback cognitively acceptable yet
affectively abhorrent. This prompts one to ask how people escape from such
crossfres.
One possibility is that the cognitive system resolves such crossfres by muting or
transforming the affective response. Indeed, the structure of the cognitive and
affective systems might favor such an outcome: Insofar as the affective system is
adapted for rapid decision-making processes and the cognitive system is adapted
for more reflective processes (e.g., Epstein, 1983; Zajonc. 1980), the cognitive
system should become increasingly dominant over time.
Our fndings offer some support for the notion that cognitive responses
eventually encroach upon affective responses. That is, our participants affective
responses to feedback faded over time while their cognitive responses persisted.
Other investigators have offered additional evidence of a tendency for cognitions to
modify affective experiences. For example, recent evidence indicates that when
peoples behavioral predispositions toward some target person are based largely on
affect (e.g., when they have just met a target), inducing them to think about that
target can systematically alter their subsequent behavior toward him or her (e.g.,
Millar & Tessei; 1986; Wilson, Dunn, Bybee, Hyman, & Rotondo, 1984). Similarly,
there is evidence that people manage their emotions by altering or juggling
cognitions related to those emotions (e.g., Hochschild, 1983). For example. to cope
with a drunk and unruly passenger, a flight attendant might transform his or her
anger into sympathy by supposing that the passenger is grieving the death of a
spouse. A somewhat similar strategy is used by cognitive therapists (e.g., Beck,
1967; Ellis, 1962) who often treat emotional disturbances by encouraging clients
to develop interpretations of negative events that are highly adaptive.
From this perspective, the cognitive system is remarkably facile at fashioning
ways of avoiding or eliminating cognitive- affective crossfres. Note, however, that
such improvisations are not universally effective. At times, the source of the affect
may be so powerful that no amount of cognitive gymnastics can defy it. In some
cases, this may be for the best, particularly when the ability of the cognitive
system to mute affective states might encourage people to make behavioral choices
that actually increase affective distress in the future. For example, to the extent
that people with negative self-views convince themselves that unfavorable
feedback is desirable because it is trustworthy and predictable, they may be
tempted to seek out intimates who are apt to provide them with such feedback
(e.g., Swann & Ftsher,
1986) . These intimates may then supply them with unfavorable feedback that
fuels future bouts of depression (e.g., Swann & Predmore, 1985).
45
Implications and Conclusions

Much of the history of social and personality psychology can be understood as
an unsuccessful quest for evidence of psychological unity. Part of this history,
which involves efforts to fnd unity in peoples reactions to feedback, has been
discussed in this report. But researchers interested in responses to feedback have
not been the only ones to venture into their laboratories in search of unity and
emerge with evidence of disunity. For example, disunity and lack of consistency
has been a major theme in research on the relation of attitudes to behavior (e.g.,
Fazio & Zanna, 1981; Wickei; 1969) and in treatments of the trait-situation
controversy (e.g., Magnusson & Endler, 1977). Similarly, research on emotion and
misattribution of arousal has suggested that two separate psychological systems
contribute to the experience of emotion, one that controls arousal and another that
interprets arousal (e.g., Zillman, 1983). Furthermore, investigations of peoples
introspective powers have suggested that the psychological system that explains
overt behavior has no access to the system that generates behavior (e.g., Nisbett &
Wilson, 1977; Wilson, 1985).
To be sure, the assumption of psychological unity is appealing in many ways. It
is simple, elegant, and phenomenologicaily compelling. And it is pragmatic; were
it not for the assumption of psychological unity, holding people responsible for
their actions might be a rather awkward affair. Yet our data suggest that at least
with respect to reactions to social feedback, people are not nearly as single minded
as the unity assumption would have us believe.
References
Beck, A. T. (1967). Depression: Causes and treatment. Philadelphia: University of
Pennsylvania Press.
Bimbaum, M. H. (1981). Thinking and feeling: A skeptical review American Psychologist, 37,
99-101.
Ellis, A. (1962). Reason and emotion in psychotherapy. New York: Lvle Stuart.
Epstein, S. (1973). The self-concept revisited: Or a theory of a theory.
American Psychologist, 28, 404-416.
Epstein, S. (1983). The unconcious, the preconcious, and the self-concept In J. Suls & A. G.
Greenwald (Eds.), Psychological perspectives on the self (Vol. 2, pp. 219-247). Hillsdale.
NJ: Eribaum.
Fazio, R. H., & Zanna, M. P. (1981). Direct experience and attitude- behavior consistency. In
L. Berkowitz (Ed.), .4dvances in experimental social psychology (Vol. 14, pp. 162-202).
New York: Academic Press.
46
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
Freud, S. (1921). A general introduction to psychoanalysis. New York: Boni & Liverwright.
Gazzaniga, M. S. (1985). The social brain. New York: Basic Books. Greenwald, A. G. (1982).
Is any one in charge? Personalysis versus the principle of personal unity. In J. Suis(Ed.),
Psychological perspectives on the self{Vol. 1,151-181). Hillsdale, NJ: Eribaum.
Helmreich, R., Spence, J. T., & Stapp, J. (1974). Short form of the Texas social behavior
inventory, an objective measure of self-esteem. Bulletin of the Psychonomic Society, 4,473475.
Hochshild, A. (1983). The managed heart. Berkeley, CA: University of California Press.
Homey, K. (1937). The neurotic personality of our time. New York: Norton.
Izard, C E. (1984). Emotion-cognition relationships and human development In C. E. Izard, J.
Kagan. & R. B. Zajonc (Eds.), Emotions, cognition, and behavior (pp. 17-37). New York:
Cambridge University Press.
Jones, S. C. (1973). Self and interpersonal evaluations: Esteem theories versus consistency
theories. Psychological Bulletin, 79. 185-199.
/ Lazarus, R. S. {1984). On the primacy of cognition. American Psychologist. 39, 124-129.
Lecky, P. (1945). Self-consistency: A theory of personality. New York; Island Press.
Losco-Szpilet; J. P., & Epstein, S. (1978, April). Reactions to favorable and unfavorable
evaluations in everyday life as a function of level of self-esteem. Paper presented at the
meeting of the Eastern Psychological Association, Washington, DC.
Magnusson, D., & Endlei; N. S. (1977). Personality at the crossroads: Current issues in
interactional psychology. Hillsdale, NJ: Eribaum.
McFarlin, D. B and Blascovich, J. (1981). Effects of self-esteem and performance on future
affective preferences and cognitive expectations. Journal of Personality and Social
Psychology. 40. 521-531.
Mead, G. H. (1934). Mind, selfand society. Chicago: University of Chicago Press.
Mellers, B. A. (1981). Feeling more than thinking. American Psycholcb gist. 36. 802-803.
Millar. M. G., & Tesser, A. (1986). Effects of affective and cognitive focus on the attitudebehavior relation. Journal of Personality and Social Psychology. 51. 270-276.
Moreland. R. L., & Sweeney, P. D. (1984). Self-expectancies and reactions to evaluations of
47
personal performance. Journal of Personality, 52. 156-176.

Nie, N. H., Hull, C. H., Jenkins, J. G., Steinbrenner, K.., & Bent, D. H. (1975). Statistical
package for the social sciences. New York: Mc- Graw-Hill.
Nisbett R. E., & Wilson, T. D. (1977). Telling more than we can know: Verbal reports on
mental processes. Psychological Review, 84. 231- 259.
Rogers, C. R. (1961). On becoming a person: A therapists' view of psychotherapy. Boston:
Houghton Mifflin.
Secord, P. F., & Backman, C. W. (1965). An interpersonal approach to personality. In B.
Maher (Ed.), Progress in experimental personality research (Vol. 2, pp. 91 -125) New
York: Academic Press.
Shrauger. J. S. (1975). Responses to evaluation as a function of initial self-perceptions.
Psychological Bulletin, 82, 581-596.
Swann, W. B., Jr. (1983). Self-verification: Bringing social reality into harmony with the
self. In J. Suls & A. G. Greenwald (Eds.), Social
psychological perspectives on the self {'Vol. 2, pp. 33-66). Hillsdale, NJ: Eribaum.
Swann, W. 8., Jr. (1985). The self as architect of social reality. In B. Schlenker (Ed.), The self
and social life (pp. 100-125). New York: McGraw-Hill.
Swann, W. B., Jr. (1987). Identity negotiation: Where two roads meet. Unpublished
manuscript. University of Texas at Austin.
Swann, W. B., Jr., & Fishei; D. A. (1986). Mate selection and the self- concept. Unpublished
manuscript, University of Texas at Austin.
Swann, W. B., Jr., & Predmore, S. C. (1985). Intimates as agents of social support Sources of
consolation or despair? Journal of Personality and Social Psychology, 49, 1609-1617.
Tesset; A. (1985, August). Toward a self-evaluation maintenance model of social behavior.
Paper presented at the annual convention of the American Psychological Association, Los
Angeies.
Tomkins, S. S. (1981). The quest for primary motives: Biography and autobiography of an
idea. Journal of Personality and Social Psychology. 41. 306-329.
Wicker, A. W. (1969). Attitudes versus actions: The relationship of verbal and overt behavioral
responses to attitude objects. Journal of So cial Issues. 25.41-78.
48
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
< Wilson, T. D. (1985). Strangers to ourselves: The origins and accuracy of beliefs about ones
own mental states. In J. H. Harvey & G. Weary {Eds.), Attribution in contemporary
psychology (pp. 9-36). New York: Academic Press.
Wilson, T. D Dunn, D. S., Bybee, J. A., Hyman, D. B & Rotondo, J. A. (1984). Effects of
analyzing reasons on attitude-behavior consistency. Journal of Personality and Social
Psychology, 47. 5-16.
Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. American
Psychologist, 35. 151-175.
Zajonc, R. B. (1984). On the primacy of affect. American Psychologist, 39. 117-123.
Zillman. D. (1983). Transfer of excitation in emotional behavior. In J. T Cacioppo & R. Petty
(Eds.), Social psychophysiology: A sourcebook (pp. 215-240). New York: Guilford.
Zuckerman, M., & Lubin, B. (1965). Manual for the multiple affect adjective checklist. San
Diego, CA: Educational and Industrial Testing Service.Appendix Measures of Cognitive
and Affective Reaction

s
about you by watching you give the
speech? (nothing at all-a great deal)
All responses were recorded on 9-point

Likert scales that used the anchors listed
after each item.
Perceived Competence of
Evaluator (Alpha = .94)
Perceived Accuracy of the

Feedback (Alpha = .93)
1.
How accurate do you think this

impression of you was? (extremely
accurate-inaccurate)
2. How much could a stranger leam about

you from reading this impression of you?
(nothing at all-a great deal)
3.
How much did you agree with this

impression of you? (strongly agreedstronglv disagreed)
4. How well thought out do you think the

impression of you was? (not well thought
out-extremely well thought out)
5. How much did the other subject leam
Rate where you think the person who

wrote the impression of you would fall on
the following trait scales by circling a
number.
1.
judge other peoples personalities

(extremely unable-extremely able)
2. form accurate first impressions of others

(same as 1)
3. read other people {same as 1)
4. understand what others are thinking and
feeling (same as 1)
my behavior-totally
behavior)
Evaluation of the
Psychological Review 1995,
Vol. J02, No. 1,4-27
Diagnosticity of the
Nonverbal Technique in
Forming Impressions (Alpha
= .86)
1.
How much do you think peoples

nonverbal behavior generally reveals to
others about their personalities? (nothing
at all-a great deal)
2. How much do you think an observer can

leam about another person just by
watching (not hearing) that person give a
speech? (same as 1)
Attribution to Self and to Other

1.
2.
49
result
ofmy
To what extent do you think the

impression formed of you was not a result
of your behavior, but a result of the other
subject's personal way of judging others.
(Not at all a result of his/her way of
judging others-totally a result of his/her
way ofjudging others)
Attraction to the Evaluator (Alpha

= .91)
1. How much do you think you would like
the person who wrote this impression of
you? (would not like this person at allwould like this person a great deal)
2. Describe your general reaction to the
person who wrote the impression of you.
(extremely negative~extremely positive)
To what extent do you think the

impression formed of you today was a
Received June 24. 1985 Revision
result of the behavior you displayed
received September 9. 1986
giving the speech? (Not at all a result of
Copyright 1995 by the American Psychological Association, Inc.
0033-295X/95/$3.00
Implicit Social Cognition: Attitudes, SelfEsteem, and Stereotypes

Anthony G. Greenwald
University
of
Mahzarin
R.
Banaji
Washington
Yale University
Social behavior is ordinarily treated as being under conscious (if not always
thoughtful) control.
However, considerable evidence now supports the view that social behavior often operates in
an implicit or unconscious fashion. The identifying feature of implicit cognition is that past
experience influences judgment in a fashion not introspectively known by the actor. The
present conclusion that attitudes, self-esteem, and stereotypes have important implicit
modes of operationextends both the construct validity and predictive usefulness of these
major theoretical constructs of social psychology. Methodologically, this review calls for
increased use of indirect measureswhich are imperative in studies of implicit cognition. The
theorized ordinariness of implicit stereotyping is consistent with recent fndings of
discrimination by people who explicitly disavow prejudice. The fnding that implicit cognitive
50
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
effects are often reduced by focusing judges attention on their judgment task provides a basis
for evaluating applications (such as affirmative action) aimed at reducing such unintended
discrimination
.
Long before they became central to

other areas of psychological theory,
concepts
of
cognitive
mediation
dominated the analysis of social
behavior. The constructs on which this
article
focuses
achieved
early
prominence in social psychological
theory with formulations that were
partly (attitude) or entirely (stereotype) cognitive. By the 1930s, Allport
(1935) had declared attitude to be
social psychologys most distinctive
and indispensable concept (p. 798),
Thurstone (1931; Thurstone & Chave,
1929) had developed quantitatively
sophisticated methods for attitude
measurement, and Katz and Braly
(1933, 1935) had introduced a method
that is still in use to investigate stereotypes. Self-esteem, an attitudinal
construct to which this article gives
separate treatment because of its
prominence in recent
Anthony G. Greenwald, Department of

Psychology, University of Washington;
Mahzarin R. Banaji, Department of
Psychology, Yale University.
Preparation of this report as well as
conduct of some of the research reported in
it were supported by National Science
Foundation Grants DBC-9205890 and DBC9120987 and by National Institute of
Mental Health Grant MH-41328. We thank
Icek Ajzen, John Bargh, R. Bhaskar, Irene
Blair, Robert Bornstein, Marilynn Brewer,
Robert Crowder, Leonard Doob, Russell
Fazio, Klaus Fiedler, Deborrah Fra- ble,
Daniel Gilbert, Jack Glaser, Richard
Hackman, Curtis Hardin, Roger Hughes,
John Jost, Larry Jacoby, John Kihlstrom,
Mark Klinger, David Myers, Delroy
Paulhus, Richard Petty, Alex Rothman,
Peter Sa- lovey, Eric Schuh, Norbert

Schwarz, Constantine Sedikides, Jerzy
Trzebinski, James Uleman, Wendi Walsh,
Timothy Wilson, Joanne Wood, and two
anonymous reviewers for comments on a
draft of this article, and we thank Mitzi
Johnson both for comments and for permission to use the data presented in Figure 1.
This article is dedicated to the memory of
Tom Ostrom, a dear colleague who
continues to have a profound influence on
both of the authors.
Correspondence concerning this article
should be addressed to Anthony G.
Greenwald, Department of Psychology, NI25, University of Washington, Seattle,
Washington 98195. Electronic mail may be
sent to agg@u. washington.edu.
research, also has a long-established

history (e.g., James, 1890; see overview
in Wylie, 1974, 1979).
Through much of the period since the
1930s, most social psychologists have
assumed that attitudes, and to a lesser
extent stereotypes, operate in a
conscious mode. This widespread assumption of conscious operation is most
evident in the near- universal practice
of operationalizing attitudes (including
selfesteem) and stereotypes with direct
(instructed self-report) measures. The
pervasiveness of direct measurement
for attitudes and stereotypes was
documented by Greenwald (1990) and
by Banaji and Greenwald (1994) and is
further reviewed below. In contrast,
this article describes an indirect,
unconscious, or implicit mode of
operation
for
attitudes
and
stereotypes.8
8 The terms implicit-explicit capture a set
Implicit
Social
Cognition:
Introduction and Overview
Implicit social cognition is offered as
a broad theoretical category that
integrates and reinterprets established
research fndings, guides searches for
new empirical phenomena, prompts
attention to presently underdeveloped
research methods, and suggests
applications in various practical
settings. This section summarizes the
goals of the review, starting from a
defnition of implicit social cognition.
Definition
The signature of implicit cognition is
that traces of past experience alFect
some performance, even though the
influentia
of overlapping distinctions that are

sometimes labeled as unaware-aware,
unconscious-conscious,
intuitive-analytic,
direct-indirect, procedural-declarative, and
automatic-controlled. These dichotomies
vary in the amount and nature of implied
theoretical interpretation. This article uses
the implicit-explicit pair because of that
dichotomys prominence in recent memory
research, coupled with the present intention
to connect research on attitudes, self-esteem,
and stereotypes to memory research.
51
52
6.
COGNITIVE
LEARNING AND
earlier experience is not remembered

in the usual sensethat is, it is
unavailable
to
self-report
or
introspection (cf. Graf & Schacter,
1985; Greenwald, 1990; Jacoby &
Dallas, 1981; Jacoby, Lindsay, & Toth,
1992; Jacoby & Witherspoon, 1982;
Kihlstrom, 1990; Roediger, Weldon, &
Challis, 1989; Schacter,
1987) . As an illustration of implicit
cognition, consider a result that is
readily obtained with the task of
generating complete words in response
to incomplete letter strings (word
stems or word fragments). The
completion responses are more likely to
be words from a list to which subjects
were casually exposed earlier in the
experiment than to be equally likely
words that were not presented. This
effect of prior exposure occurs despite
subjects poor ability to recall or
recognize words from the earlier list.
The word-completion task provides an
indirect measure of the effect of the
prior experience. That is, even though
the subject is not instructed to retrieve
the earlier presented material and is
presumably not trying to do so (and
may well be incapable of such
retrieval), the subjects responses
indicate a residual effect. (For further
reviews of indirect measurement in
memory research, see RichardsonKlavehn & Bjork, 1988; Roediger,
1990; Roediger & Blaxton, 1987;
Tulving& Schacter, 1990. For further
discussion of direct and indirect
measures in other contexts that involve
unconscious cognition, see Jacoby,
Lindsay, & Toth, 1992; Reingold &
Merikle, 1988.)
COGNITIVE
RESPONSE
A template for defnitions of specifc

categories of implicit cognition is: An
implicit C is the introspectively

unidentified
(or
inaccurately
identified) trace of past experience
that mediates R. In this template, C
is the label for a construct (such as
attitude), and R names the category of
responses (such as object- evaluative
judgments) assumed to be influenced
by that construct.
To the extent that implicit cognition

differs from self-reportable (conscious
or explicit) cognition, direct measures
that is, measures that presume
accurate
introspectionare
necessarily inadequate for its study.
Rather, investigations of implicit cognition require indirect measures,
which neither inform the subject of
what is being assessed nor request
self-report concerning it. The usual
justifcation for indirect measures in
social psychological research is the
empirical one of minimizing reactivity
of research situations to avoid demand
characteristics (Orne, 1962) and
sources of self-presentational artifacts
such as evaluation apprehension
(Rosenberg, 1969). When used in this
way to minimize reactivity, indirect
measures are empirically desirable but
not theoretically essential. By contrast,
in studying implicit cognition, indirect
measures are theoretically essential.
Theory: Relation to Other

Treatments of Unconscious
Aspects of Social Cognition
Much of the present review concerns
unconscious cognitive involvement in
(and especially interference with)
deliberate judgments. This focus is still
infrequent in social cognition literature, perhaps because the range of
deliberate judgments that are affected
by unconscious cognition has only
recently become apparent, as a

consequence of the explosion of
interest in implicit memory. The
existing work to which the present
treatment is closest is that of Jacoby
and colleagues (e.g., Jacoby & Dallas,
1981; Jacoby, Toth, Lindsay, & Debner,
1992; Jacoby & Witherspoon, 1982).
Jacobys group has pioneered theory
and methods to identify implicit
memory influences and has effectively
established an important role of
unconscious cognition in deliberate
judgments.
The
phenomena
investigated by Jacoby have often been
ones in which the research subject mistakenly attributes ease of perception
on
reexposure
to
a
stimulus
(perceptual
fluency)
to
some
characteristic of the stimulus other
than
an
unremembered
recent
encounter. The present analysis of
implicit social cognition uses the same
basic misattribution principle, while
focusing on the social domain to locate
influential prior experiences and
affected target judgments.
The following paragraphs review the
various ways in which unconscious
cognition has already been integrated
into social psychological theory. After
considering the three categories that
are focal to this reviewattitudes, selfesteem, and stereotypesthe focus
shifts to some related topics.
Attitudes.
Recent
work
has
established that attitudes are activated outside of conscious attention,
by showing both that activation occurs
more rapidly than can be mediated by
conscious activity (Bargh, Chaiken,
Govender, & Pratto, 1992; Fazio, Sanbonmatsu, Powell, & Kardes, 1986)
and that activation is initiated by
(subliminal) stimuli, the presence of
which is unreportable (Greenwald,
53
Klinger, & Liu, 1989). The present

analysis of implicit attitudes extends
work on automatic activation to
explain how the attitude activated by
one object can be (mis)attributed to
another. An implicit attitude can be
thought of as an existing attitude
projected onto a novel object. The
interpretation of several important
existing fndings as implicit attitude
effects substantially expands the
predictive and construct validity of
social psychologys attitude construct.
It also prompts the empirical search for
further members of the potentially
large class of implicit attitude effects.
In the domain of attitude change, two
recent theoretical analyses (Chaiken,
1987; Petty & Cacioppo, 1986) have
distinguished relatively thoughtful
(central or systematic) from relatively
thoughtless (peripheral or heuristic)
roles of cognition in persuasion. The
implicit processes conceived in the
present analysis are, in part,
subsumed by the notions of peripheral
or heuristic processing, but also involve
processes operating even further from
the range of conscious thought than
conceived in these analyses.
Self-esteem. The broad importance

of self-esteem has been recognized in
many works over the past century (e.g.,
Allport, 1937; Cooley, 1902/1964;
Epstein, 1973; James, 1890; Rogers,
1951; Rosenberg, 1979; Sherif &
Cantril, 1947). Recent reviews have
further expanded the case for
importance of the self-attitude (e.g.,
Beck, 1979; Greenwald &Pratkanis,
1984;Seligman, 1991; Steele, 1988; S.
E. Taylor & Brown, 1988; Tesser,
1988). In the course of examining
available evidence for the implicit
operation of attitudes, evidence for
implicit operation specifcally of the
attitude toward self was so prominent
54
6.
COGNITIVE
LEARNING AND
as to prompt treatment of implicit selfesteem as a distinct topic. The present

review
of
implicit
self-esteem
phenomena demonstrates the pervasiveness of projections of the self
attitude onto other objects, while also
indicating the need for a class of
measures that presently does not exist
measures of individual differences in
implicit self-esteem.
Stereotypes. Recent reviewers have

very effectively documented the
unconscious or automatic operation of
stereotypes (Banaji & Greenwald, in
press; Bargh, 1994; Devine, 1989;
Dovidio, Evans, & Tyler, 1986; Fiske,
1989; Geis, 1993; Gilbert & Hixon,
1991; Hamilton & Sherman, 1994;
Perdue & Gurt- man, 1990). This
articles use of the implicit label for
stereotypes
serves
primarily
to
emphasize the connection of the existing body of social cognition research on
stereotypes
to
recent
cognitive
psychological research on implicit
memory (cf. E. R. Smith &
Branscombe, 1988). The present
treatment of stereotypes also includes
new fndings that demonstrate this
connection, using the methods of
implicit memory research to reveal
implicit gender stereotypes.
Effortless or automatic social
cognition. Single words have been
shown to result in effortless activation
of attitudes (discussed in the next
paragraph), and behavior-describing
sentences produce spontaneous trait
inferences (Uleman, 1987; Winter &
Uleman, 1984). These and other
effortless activation effects often
contribute importantly to, but are not
synonymous with, implicit social
cognition. As will be seen below, an
implicit effect can occur when an actor
(a) notices some aspect of an automatic
COGNITIVE
RESPONSE
effect caused by one stimulus and (b)

mislabels it in a way that that
influences the judgment of either that
stimulus or some other stimulus.
Discussions of effortless aspects of social cognition can be found in Bargh
(1989), Brewer (1988), Epstein (1991),
Gilbert (1989), and Lewicki (1986).
Priming and context effects. In

common with implicit social cognition,
priming and context effects involve the
effect of prior events on the response to
a current stimulus. However, whereas
priming and context function as
designations for operationally defned
categories of effects, implicit cognition
designates a theory-defned category of
effects. Some established priming and
context effects fall well within the
boundaries of implicit social cognition.
Others should not be grouped with
implicit effects because of the subjects
likely awareness of the effect of prior
experience on behavior or judgments.
Because the existing literatures on
priming and context effects are very
large, analysis to sort implicit effects
from other effects within those
literatures is beyond the scope of this
article. Recent theoretical discussions
of priming and context effects can be
found in Higgins (1989), Schwarz
(1990), Strack (1992), Martin and
Tesser (1992), and Petty and Wegener
(1993).
Introspective access. Nisbett and
Wilson (1977a) argued persuasively
that psychologys reliance on verbal
self-report measures was unwarranted
in the face of evidence showing the
poverty of introspective access to the
causes of behavior. Wegner and
Vallacher (1977, 1981) also drew
attention to influences on social
behavior that escape introspective
notice. They presented an implicit
psychology, drawing on and extending

Bruner and Tagiuris (1954) concept of
implicit personality theory. However,
because their work preceded the past
decades wave of research on implicit
memory, Wegner and Vallacher did not
share the present articles focus on
empirical studies that use indirect
measures. More recently, in several
works, E. R. Smith (e.g., 1984, 1990,
1994) has emphasized the importance
to social cognition of non verbalized
procedural knowledge. The present
work shares Smiths emphasis on
examining
the
introspectively
inaccessible underpinnings of social
cognition. In related work, Smith (see
E. R. Smith & Branscombe, 1988) has
also noted the relevance of implicit
memory to social cognition. The
research reviewed here sheds new light
on the nature of causes that are hidden
from introspection and provides some
methods for observing them. In this
respect, research in social cognition
now appears to be taking an important
further step along the path that was
laid out by Nisbett and Wilson (1977a).
Empirical Assessment
Falsifiability. Of any newly offered
theoretical construct, it should be
asked: How does the new construct
differ from existing ones (or is it only a
new label for an existing construct)?
The preceding paragraphs show that
implicit social cognition, although
strongly rooted in existing constructs,
offers a theoretical reorganization of
phenomena that have previously been
described in other ways and, in some
cases, not previously identifed as
having an unconscious component. The
relations to prior theorization are
emphasized in this article by using
established construct termsattitude
and stereotypeas labels for two
major categories of implicit social
55
cognition.
When a new construct indeed differs
from existing ones, the new construct
should provide a basis for either (a)
predicting
previously
unobserved
empirical phenomena or (b) guiding research to show a gain in the efficiency
with which existing phenomena can be
explained. Importantly, the construct
should be well enough linked to
research
operations
that
its
predictions, especially its predictions
that differ from those afforded by existing constructs, can be disconfrmed.
This articles strategy is to identify
parallels of method and fndings
between the domains of social
cognition and implicit memory. The
ease with which such parallels are
discovered is the main evidence for
value of the implicit social cognition
constructs. This convergence-seeking
strategy provides little opportunity for
falsifcation of the general thesis that
social cognition operates in implicit
fashion. Rather, results that appear
not to ft the thesis can be set aside as
possibly
involving
inappropriate
operationalizations.
However,
as
parallels between the two domains of
phenomena
are
increasingly
established, there should be increased
confdence
in
interpretations
of
research operations for social cognition
constructs, and, consequently, results
that fail to ft with theory will
increasingly call theory into question.
Challenge
to
measurement
method.
The
present
account
supposes that individual differences in
manifestations of implicit cognitive
effects should be predicted by
individual differences in the strength
of theorized representations that
56
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
underlie those effects. Measurement of

those individual differences is beyond
the means of present assessment
technology; consequently, a large
subset of the empirical implications of
the present analysis are currently
untestable. The need for new measurement methods and these methods
relation to available methods are
discussed later in this article.
phenomena, implicit social cognition

should be confned to modern North
American
culture.
Furthermore,
although the present treatment
focuses on just the three categories of
implicit attitudes, implicit self-esteem,
and implicit stereotypes, implicit
cognition is also expected to be
involved broadly in other social
phenomena.
Application
potential.
The
empirical phenomena of implicit social
cognition
involve
introspectively
inaccessible effects of current stimulus
or prior experience variations on
judgments and decisions. As will be
shown, these effects often result in
subjects making judgments that they
would regard as nonoptimal if made
aware of the source of influence.
Furthermore, these effects are likely to
occur in situations that involve
economically and socially important
decisions, such as hiring, educational
admissions, and personnel evaluations.
Consequently, a strong test of the
empirical value of the analysis of
implicit social cognition will be its
ability to generate applications that
can
minimize
these
unwanted
intrusions
on
judgment.
This
important
application
topic
is
considered briefly near the end of this
article.
The next three sections consider

phenomena of implicit social cognition
in the categories represented by
construct designations of attitude, selfesteem, and stereotype. Following
those comes a section that considers
principles
underlying
potential
applications that seek to reduce
unintended
implicit
cognitive
intrusions
on
judgment.
Last,
problems
associated
with
measurement of implicit social cognition
are
discussed
before
general
conclusions are stated.
Generalizability. It will be obvious

that the great majority of evidence
reviewed in this article comes from
experimental studies done in late20th-century North America. As a
result, it is very likely that some of the
specifc properties of implicit attitudes,
implicit self-esteem, and implicit
stereotypes included in this review are
culture bound and time bound. At the
same time, there is no reason to
believe that, as a general class of
Implicit Attitudes
Attitudes
are
favorable
or
unfavorable dispositions toward social
objects, such as people, places, and
policies. Attempts to establish the
validity of the attitude construct have
most often sought to demonstrate
positive
correlations
between
measured attitudes and the favorableunfavorable aspect of observed behavior toward their objects. The
frequently weak correlations observed
in these attempts defne the predictive
validity
problem
for
attitudes
(documented especially by Wicker,
1969; see also Festinger, 1964, and
LaPiere, 1934). A notable accomplishment of modern research on attitudes
has been the solution of this predictive
validity problem. That is, conditions
under
which
attitudes
strongly
correlate with behavior have now been
well identifed (especially by Ajzen &

Fishbein, 1980; Fazio, 1986, 1990b;
Fazio & Zanna, 1981; Fishbein &
Ajzen, 1974; Zanna & Fazio, 1982). In
particular, this research has established that attitudes have predictive
validity in situations in which they are
strongly activated and/or when the
actor clearly perceives a link between
attitude and behavior. Myers (1990)
summarized
these
and
related
programs of research as showing that
our attitudes predict our actions . . .if,
as we act, we are conscious of our
attitudes (Myers, 1990, p. 40,
emphasis added). Similarly, in the
description of attitude-behavior relations in their recent treatise on the
attitude construct, Eagly and Chaiken
(1993, pp. 208-211) referred to the
importance of attitudes [coming] to
mind and the perceived relevance of
attitude to action.
Although the modern synthesis
achieved by the Fishbein-Aj- zen
(1974)
and
Fazio-Zanna
(1981)
research programs is now well
established, it is difficult to avoid
concluding that the attitude construct
lost scope in the process. For those
who can remember it, there might be
justifable nostalgia for an era in
which Allport (1935) was able to
proclaim that attitude was social
psychologys
most
indispensable
concept. In an undisguised effort to
restore this prominence, the present
article seeks to preserve the modem
synthesis while asserting that its
opposite is also valid (cf. McGuires
[1973] Koan 7: The opposite of a great
truth is also true); that is, attitudes of
which the actor is not conscious at the
moment of action (implicit attitudes)
are also strongly predictive of
behavior.
57
Ignored Consciousness in
Conceptual Definitions of
Attitude
The following list gives several
defnitions that have been influential
in guiding scholarly and empirical
treatments of attitudes, as indicated
by their frequent citation in other
works. Although the list may appear
dated (the most recent entry is from
1962), it nevertheless remains current.
Recent works (e.g., Eagly & Chaiken,
1993; Fazio, 1986; McGuire, 1985;
Petty & Cacioppo, 1981; Zanna &
Rempel, 1988) continue to draw on
them and remain within their
boundaries.
Attitude is the affect for or against a
psychological object. (Thurstone, 1931,
p. 261)
An attitude is a mental and neural
state of readiness, organized through
experience, exerting a directive or
dynamic
influence
upon
the
individuals response to all objects and
situations with which it is related.
(Allport, 1935, p. 810)
Attitude is . . .an implicit, driveproducing response considered socially
signifcant in the individuals society.
(Doob, 1947, p. 136)
An attitude is a predisposition to
experience, to be motivated by, and to
act toward, a class of objects in a
predictable manner. (M. B. Smith,
Bruner, & White, 1956, p. 33)
[Attitudes] are predispositions to
respond, but are distinguished from
other such states of readiness in that
they predispose toward an evaluative
response.
(Osgood,
Suci,
&
Tannenbaum, 1957, p. 189)
[An attitude is] a disposition to react
favorably or unfavorably to a class of
objects (Samoff, 1960, p. 261).
Attitudes [are] enduring systems of
positive or negative evaluations,
emotional feelings, and pro or con
58
6.
COGNITIVE
LEARNING AND
action tendencies with respect to social

objects.
(Krech,
Crutchfeld,
&
Ballachey, 1962, p. 139)
The lack of mention of consciousness

in this collection of attitude defnitions
accurately reflects a long scholarly
tradition of nonconcem with the
distinction between conscious and
unconscious operation of attitudes. At
the same time, nothing in this
scholarly tradition actively opposes
either the possibility or the importance
of unconscious operation of attitudes.
Standing starkly in the above list as
suggesting unconscious operation is
Doobs (1947) defnition, which labels
attitude as an implicit, driveproducing response. In spite of Doobs
association with a behaviorist theory
(Hull, 1943) that had no use for
conceptions of either conscious or
unconscious cognition, it is clear that
Doob did conceive attitude as
operating unconsciously (May & Doob,
1937,
p.
13).
In
a
recent
communication to the present authors
(October 27, 1992), Doob commented,
before World War II we all were
impressed by psychoanalysis in
addition to behaviorism, suggesting
that, even though it may have gone
unmentioned in many published
treatments, the idea that attitudes
operated unconsciously was quite
acceptable in the 1940s and earlier.
That conclusion is supported also by
several passing references to the
possibly
unconscious
nature
of
attitudes in Allports (1935) review
chapter.
Implied Consciousness in
Operational Definitions of Attitude
The present authors conducted a
census of studies published in all
issues of European Journal of Social
COGNITIVE
RESPONSE
Psychology, Journal of Personality and

Social
Psychology,
Journal
of
Experimental Social Psychology, and
Personality and Social Psychology Bulletin for a single year (1989). The aim
of this census was to compare the
levels of use of direct and indirect
measures of attitudes in published
research literature. Studies that
included attitude measures were
categorized in terms of whether they
measured
attitude
directly
or
indirectly. Direct measures included
self-report procedures such as multiitem Thurstone, Likert, or semantic
differential scales of the sort that are
described in texts on attitude
measurement (e.g., A. L. Edwards,
1957; Fishbein,
1967) , as well as more informal
single-item and multi-item self- report
procedures. Indirect measures are
identifable chiefly by their lack of the
defning feature of direct measures,
that is, by their not alerting the
subject to the identity of the object of
the
attitude
being
measured.
Discussions and illustrations of indirect measures can be found in
Hammond (1948), Campbell (1950),
Jahoda, Deutsch, and Cook (1951),
Sherif and Hovland (1961, p. 93ff.),
Dawes
and
Smith
(1985),
Pratkanis(1988), and Dovidio and
Fazio (1992).
The authors 1989 census included
not only studies that obviously dealt
with attitudes (as indicated by title or
abstract), but also ones that more
incidentally included measures of
evaluation toward the self, others, or
social objects, for example in studies of
person perception and in-group bias.
Each of the 47 studies found to include
an attitude measure was judged for
presence of both direct and indirect
attitude measures. All 47 (100%) used

at least one direct measure of attitude,
and only 6 (13%) used some form of
indirect measure. An important class
of attitudes discussed in this article is
self-esteem. It is not necessary to
conduct a census of publications to
assert with confdence that selfesteem is generally assessed with
direct measures (see Wylie, 1974,
1979). In summary, the observed high
level of reliance on direct measures of
attitudes indicates a widespread (even
if not widely stated) assumption that
attitudes operate primarily in a
conscious mode.
Definition of Implicit Attitude

Evidence concerning the strength of
attitude-behavior
relations
has
generally been regarded as the
primary
evidence
bearing
on
predictive validity of the attitude
construct. However, there are other
categories of studies in which strong
predictive
effects
that
involve
attitudes are routinely obtained. Examination of these now will indicate
that some strong effects of attitude
can and do occur when the actor is not
attentionally focused on the attitude.
These fndings play a central role in
justifying the concept of implicit
attitude.
Implicit attitudes are introspectively
unidentifed (or inaccurately identifed)
traces of past experience that mediate
favorable or unfavorable feeling,
thought, or action toward social
objects.9
9 Some notes on this defnition: First, it
was generated from the earlier stated
template for defnitions of implicit cognition.
Second, alternate defnitions of the attitude
response class might be substituted for the
stated one (favorable or unfavorable feeling,
thought, or action toward social objects).
59
Relation to stimulus-response theory.

To those familiar with behavior theory
of 30-50 years ago, the debt of this defnition both to Doobs (1947) analysis of
attitude as an implicit response and to
subsequent mediationist stimulusresponse theories will be obvious. The
earliest
mediationist
stimulus-response
formulations
(mature
statements of which appear in Hull,
1952; Spence, 1956) had proposed the
existence of covert stimulus-producing
responses
(called
fractional
anticipatory goal responses) as a
means of explaining fndings achieved
within Tolmans (e.g., 1959) cognitive
approach to learning theory. Dollard
and Miller (1950), Osgood (e.g., 1957),
and Mo- wrer (1960) most fully
adapted these mediationist principles
to the analysis of human social
behavior. Campbells (1963) treatise on
the attitude construct preserved some
of the implicit, mediationist character
of the later behaviorist treatments, declaring that acquired dispositions such
as attitudes [retain] residues of
experience of such a nature as to
guide, bias, or otherwise influence
later behavior (p. 97). In retrospect, it
is apparent that the conception of
covert, stimulus-producing, mediating
events did not take hold in social
psychology when mediationist theory
was dominant in learning-behavior
theory. Nevertheless, that conception is
a virtual equivalent of the view that
reappears in the present defnition of
Third, the qualifcation [introspectively]
inaccurately
identifed
includes
an
important class of cases in which a prior
experience is identifable, but its influence on
an evaluative response is not. For example, a
student may be aware of having been graded
highly in a course, but not suspect that this
experience influences responses to the
courses end-of-term course evaluation
survey.
60
6.
COGNITIVE
LEARNING AND
implicit attitude.
Direct versus indirect measures.
Experiments that demonstrate implicit
attitudes rely on indirect measures to
detect their operation. The distinction
between direct and indirect measures
depends only on the relation between
what the subject is informed about the
purpose of a measure and what the researcher chooses to infer from the
subjects response. If the subject is
advised that A is being measured, but
the researcher uses the response to
infer something about B, the direct
measure of A is an indirect measure of
B. In the case of nonreactive indirect
measures, the subject is unaware that
anything is being measured and,
accordingly, there is no direct-measure
interpretation
of
the
subjects
response. Only a small portion of the
research considered in this article
involves the use of nonreactive indirect
measures.
An implicit attitude toward B may
be indirectly indicated by a (direct)
measure of evaluation of A, when A
and B have some relation that
predisposes the implicit influence.
Possibly, the evaluative content of this
implicit attitude may disagree with
results from a direct measure of
attitude toward B; such disagreement,
referred to as a dissociation of implicit
and explicit attitudes, is especially
interesting and perhaps most dramatically indicates the value of the implicit
attitude construct. Nevertheless, the
occurrence of dissociation is not a
necessary condition for identifying an
attitude as implicit. More critical is
that the effect of an attitudinal
manipulation on an indirect measure
may be reduced, eliminated, or
reversed when subjects are made
aware
of
the
nature
of
the
COGNITIVE
RESPONSE
manipulation. Further treatment of

this
point
appears
below,
in
considering effects of attentional
manipulations on expressions of
implicit social cognition (i.e., on
indirect measures).
Implicit Attitudes: Empirical

Findings
Halo Effects
Thorndike (1920) named the halo
effect, after noticing that personality
ratings showed a tendency for positive
characteristics to be associated with
other positive characteristics more
than they should be if experience is the
only guide. Subsequently, the halo
effect came to be regarded as the
tendency for judgment of a novel
attribute (A) of a person to be
influenced by the value of an already
known, but objectively irrelevant,
attribute (B). In this case, the direct
measure of evaluation of A implicitly
expresses the attitude toward B. The
attitude toward B is implicit, in
present terms, when the subject does
not identify the attitude toward B as
the source of the evaluation of A.
In much halo effect research,
physical attractiveness plays the role of
the objectively irrelevant attribute that
influences evaluative judgment on
various other dimensions. For example,
Dion, Berscheid, and Walster (1972)
reported that attractive males and
females are judged to be kinder, more
interesting, more sociable, happier,
stronger, of better character, and more
likely to hold prestigious jobs.
Similarly, Landy and Sigall (1974)
found that essays attributed to a
female student were judged by male
students to be of higher quality when
the stimulus materials included a
photo that showed the author to be
physically attractive, rather than

unattractive. Downs and Lyons (1991)
reported that defendant attractiveness
was associated with judges levying
smaller fnes and lower bail levels in
actual misdemeanor cases (although
this relationship was weaker or absent
in felony cases).
Physical attractiveness-based halo
effects have been found to be greater
for female than male targets (Bar-Tal
& Saxe, 1976; Wallston & OLeary,
1981). As Berscheid (1985) has pointed
out, however, halo effect research is
grossly imbalanced in its predominant
use of male subjects and female targets
a strategy that may reflect both
popular belief and evidence showing
that physical attractiveness of an
opposite-sex partner is more important
for men than for women (Feingold,
1990). Alternately, the imbalance may
indicate that attractiveness is a more
potent
implicit
attitudinal
cue
component of females than males. The
physical attractiveness-based halo
effect has been replicated in subject
populations of Black Americans (Cash
& Duncan, 1984) and Japanese
(Onodera & Miura, 1990), as well as
across the life span (Adams & Crane,
1980). Eagly, Ashmore, Makhijani, and
Longo (1991) concluded from their
meta-analytic review that a
component
of
the
physical
attractiveness halo effect can be
explained by assuming that perceivers
(consciously)
expect
physically
attractive people to be socially adept;
however, Eagly et al. also reported a
sizeable halo effect for judgments on
intellectual competence dimensions
(mean effect size d = .46), for which
there is no known basis in beliefs
about abilities associated with physical
attractiveness.
61
Nisbett
and
Wilson
(1977b)
demonstrated a reversal of the
direction of the usual halo effect by
presenting to subjects a male teacher
who was coached to act in either an
interpersonally
warm
or
an
interpersonally cold fashion. Subjects
in the warm condition later judged the
teacher to be more attractive in appearance, accent, and mannerisms
than did subjects in the cold condition.
Subjects denied that the teachers
likeableness influenced the other
judgments.
Quite
the
opposite,
subjects in the cold condition
incorrectly believed that the (unvaried)
accent, mannerisms, and physical
unattractiveness of the teacher had
reduced the teachers likeableness. The
importance of this fnding derives from
the subjects inaccurate identifcation
of the effect of one attribute on
judgment of another, despite these being attributes for which we generally
assume we are capable of rendering
independent assessments (Nisbett &
Wilson, 1977b, p. 250).
Despite the focus of much halo effect
research on physical attractiveness as
the cue, the range of potential halo
effect cues is quite broad. For example,
P. R. Wilson (1968) manipulated
academic status by introducing a male
stimulus person as a student,
demonstrator, lecturer, senior lecturer,
or professor, to different groups of
observers at Cambridge University.
The observers subsequent memory
estimates of the persons height were
an increasing function of the status
variable, with the student recalled to
be shortest and the professor tallest.
Another variant of the halo effect was
demonstrated by Howard (1992), who
showed that evaluation of a product
was positively influenced by frst
encountering it in an attractively gift-
62
6.
COGNITIVE
LEARNING AND
wrapped package rather than in a

plain package. Still a different halo
effect was reported by Frank and
Gilovich (1988), in showing that black
color of sports uniforms is associated
with
both
selfperceptions
and
observers
perceptions
of
aggressiveness.
Another
variant
appeared in a controversial study by
Peters and Ceci (1982), who submitted
manuscripts to journals that had
previously published them, using
fctitious
authors
names
and
deliberately unimpressive (fctitious)
institutional
affiliations.
The
subsequent rejection of most of these
manuscripts has several possible
explanations, but the majority of the
published
commentary
that
accompanied the Peters and Ceci
article concluded that the institutional
cue was a likely cause of negative
evaluations.
As a general interpretation of halo
effects, the present analysis supposes
that the subjects learning that an
unfamiliar target person possesses
Attribute B tends to produce a diffuse
positive
or
negative
attitude
(depending on the affective value of B)
toward the target person; that attitude
is then likely to generalize to any
specifc attribute (A) that the subject is
asked to judge. The attitude toward B
is said to operate implicitly when the
subject does not notice that B is
influencing the judgment of A. The
halo effect is of great practical
signifcance, being applied thousands
of times a day just in the world of
advertising. Advertisements typically
set their products in contexts that
contain attractive other objects,
especially famous entertainers and
physically attractive models. Because
advertising audiences are aware that
the advertiser is trying to influence
COGNITIVE
RESPONSE
attitude, there may be little likelihood

of the audience misidentifying the
source of positive reactions to the
advertising message. But, as a reader
of an earlier draft of this article
reminded the authors, advertising
audiences are frequently inattentive
and, in that circumstance, may be
susceptible to implicit effects of an
extraneous attractiveness cue. A
common situation in which halo effects
are likely to be more generally potent
is in managers evaluations of workers;
job-unrelated attractiveness may routinely
affect
such
performance
evaluations (e.g., Balzer & Sul- sky,
1992; Cooper, 1981; Tsui & Barry,
1986).
Other Implicit Attitude Effects

Mere exposure. In a review that
established the relationship between
frequency of encounter and liking for a
wide variety of stimuli in a wide
variety of contexts, Zajonc (1968)
named and established the mere
exposure effect. For most of the time
since, the mere exposure effect has
been an enigma, a robust effect
without
a
generally
accepted
explanation. However, several recent
fndings now indicate that the mere
exposure effect is an implicit attitude
effect. Bornstein and DAgostino
(1992), pursuing a lead that came from
Bomsteins
(1989)
meta-analytic
review, established that mere exposure
effects are strongest when conditions
reduce subjects memory for the effectproducing exposures. Under these
circumstances, it appears that increased perceptual fluency of a
repeatedly presented stimulus (that is,
increased ease of its identifcation on
re-exposure) is misattributed to liking,
yielding a positive evaluation of the
stimulus.10 Some other results that
10 The mere exposure effect does not share
indicate the same type of implicit effect

are the effect of ease of comprehension
on judgments of a statements validity
(Gilbert, 199111; Sherman, Mackie, &
Driscoll, 1990) and the effect of
repeated encounters with a statement
on its judged truth (Arkes, Boehm, &
Xu, 1991; Begg, Armour, & Kerr, 1985;
Hasher, Goldstein, & Top- pino, 1977;
Hawkins & Hoch, 1992).
Subliminal attitude conditioning.
Several researchers have sought to
induce positive or negative attitudes to
neutral stimuli by presenting a very
briefly flashed pleasant or unpleasant
stimulus just before presenting each
clearly visible neutral stimulus.
Subsequent affective judgments of the
initially neutral stimuli provide a test
of whether the affective value of the
flashed stimulus transfers to the
neutral stimulus. Although reliable
procedures for obtaining this result are
not yet well described, a few
researchers have reported successful
efforts (K. Edwards, 1990; Krosnick,
Betz,
Jussim,
&
Lynn,
1992;
Niedenthal, 1990; Murphy & Zajonc,
1993). This procedure has occasionally
been
described
as
attitude
conditioning;
however,
the
conditioning
label
may
be
unwarranted because of both the small
a major feature of the other implicit attitude
effects reviewed here, namely, evaluation of
one object displaced onto another. Even
though recent research has greatly advanced
understanding of the mere exposure effect,
the basis for its evaluative character remains
obscure.
11 Gilbert (1991) does not interpret the
effect of comprehension on judged validity to
be due to a misattribution of familiarity to
validity but, rather, to a direct effect of
comprehension.
Establishment
of
an
empirical test that can distinguish between
these interpretations may be difficult.
63
number of trials used and the reversal

of the usual neutral-stimulus-frst
ordering of conditioning procedures.
The result may better be understood as
an implicit attitude effect, in which an
attitude evoked by the frst (briefly
flashed) stimulus is mistakenly attributed to the second stimulus.
Instant attitudes? An instant
attitude can be defned as a nearimmediate liking or dislike for a novel
object on frst encounter with it (see
Fiske, 1982, for a similar idea captured
by the term schema-triggered affect).
Although little studied in research,
instant attitudes may be very common.
For example, it is commonat least for
the present authorsto experience a
sense of almost immediate sympathy
or antipathy for characters in drama,
fction, or news reports, and also with
individual performers or teams in
athletic competitions. In some cases,
an explicit basis for the instant
attitude may seem obvious to its
possessor: there is apparent self-other
similarity, or an athletic team is from
ones home town or university, or a
fctional character is depicted in
blatantly admirable or despicable
terms. In other cases, the instant
attitude may have no introspectively
accessible basis and, therefore, likely
merits identifcation as an implicit
attitude.
Context effects in survey research.
Schwarz and Clore (1983) reported
that quality-of-life judgments by
telephone-interviewed subjects were
greater for persons called on sunny
days than for those who were called on
rainy days. In a related study,
Schwarz, Strack, and Mai (1991) asked
groups of survey respondents about
both their happiness with marriage
and their happiness with life as a
64
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
whole. When the marital question preceded the quality of life question,
expressed life satisfaction increased for
respondents who reported happy
marriages, but decreased for those who
reported unhappy marriages, as compared with the respondents who
answered the life-as-a-whole question
frst. These effects may represent
implicit influences of peripheral
information on evaluative judgments,
but may also be explainable as
consequences
of
introspectively
available knowledge. One means of
supporting the interpretation of implicit influence is to show that drawing
attention to the influential cue reduces
the effect, which is precisely what was
shown by Schwarz and Clore (1983);
when subjects were asked to describe
the weather early in the interview, the
effect of weather on the later qualityof-life question was eliminated.
Myers & Ridl, 1979; S. E. Taylor &

Brown, 1988). Consequently, an
expectable form of implicit attitude
effect is that novel objects that are
invested with an association to self
should be positively evaluated. As in
the case of the halo effect, it is
assumed that the judgment of any
previously unevaluated attribute (A) of
the object is influenced by the (usually
positive) attitude associated with self
(Attribute B). The resulting positive
evaluation of Attribute A is an implicit
effect in the same sense that the halo
effect isit is assumed to occur
without the subjects awareness of the
influence and despite the objective
irrelevance of the self-association to
the subjects judgment task.
As just illustrated, a means of

pressing the argument that the effects
described in this section merit an
implicit designation is to show that
they weaken or disappear when
subjects are made aware of the implicit
attitudinal stimulus. The case for that
conclusion is developed further in the
section
below,
Attention
as
a
Moderator of Implicit Cognition.
Three categories of effects lend

themselves to interpretation in terms
of implicit self-esteem. Experimental
implicit self-esteem effects compare
evaluations of novel stimuli that are
arbitrarily associated with self or not;
greater liking for self-associated
objects is interpretable as an implicit
self-esteem effect. Naturally mediated
implicit self-esteem effects are similar,
except that the association with self is
made by means of object attributes
that have a preexisting association to
self. In a category that can be labeled
second-order
implicit
self-esteem
effects, the subject arrives at a
judgment that has an inferential link
to
self-esteem;
the
self-esteemconsistent nature of this judgment can
be
regarded
as
an
implicit
manifestation of selfesteem.
Implicit Self-Esteem
Studies done in the last few decades
have established that a majority of
almost any group of research subjects
reports favorable judgments when
asked to provide self-evaluations,
including the self-evaluative responses
to items on inventory measures of selfesteem. This very reliable result
provides the basis for concluding that
most people have a positive attitude
toward self (Banaji & Prentice, 1994;
Baumeister, 1982; Greenwald, 1980;
Implicit
self-esteem
is
the
introspectively unidentifed (or inaccurately identifed) effect of the selfattitude on evaluation of self-associated
and self-dissociated objects.
Experimental
Effects
Implicit
Self-Esteem
Role playing in persuasion. In

experimental demonstrations of roleplaying effects in persuasion, subjects
are induced to present arguments
supporting an arbitrarily assigned
controversial position. Compared with
control subjects who are exposed to
similar arguments from an external
source, role-playing subjects tend to
credit greater validity to the novel
arguments or to the position that
those arguments support (e.g., Janis &
King, 1954). This effect occurs both
when subjects are induced to play a
creative role in producing the
arguments (e.g., Greenwald & Albert,
1968; King & Janis, 1956) and when
subjects merely judge the validity of
novel issue-relevant arguments after
accepting
the
role
assignment
(Greenwald, 1969). The interpretation
of this result as an implicit self-esteem
effect is based on assuming, frst, that
the subjects acceptance of the roleplaying assignment creates a link of
the assigned position to self (this is a
psychological unit formation, in the
sense of Heider, 1944) and, second,
that positivity toward self generalizes
to a positive evaluation of the assigned
position.
Mere ownership. In a procedure
introduced by Feys (1991; see also
Beggan, 1992), subjects frst leam to
discriminate four computer-displayed
graphic icons (which represent the
subject in a computerized game) from
four others that represent the subjects
opponent (the computer). When
subjects subsequently judge all eight
patterns for aesthetic attractiveness,
the self-associated patterns receive
higher ratings. The mere ownership
label for this effect comes from an
explanation suggested by Nuttin
(1985, 1987) for the name letter effect
(discussed later in this section). A
65
similar result that likely also merits

an implicit selfesteem interpretation is
the instant endowment effect described in several experiments by
Kahneman, Knetsch, and Thaler
(1990). Those experiments established
that the value attached to such objects
as mugs, pens, and chocolate bars increased sharply as soon as the subject
was given the object.
Minimal group effect. In-group bias
is the tendency to judge members of
ones own group (in-group) more
favorably than comparable persons
who are members of another group
(outgroup). (This itself is a fnding that
could be included in the list, below, of
naturally mediated implicit selfesteem
effects.)
Compelling
demonstrations of a type frst
produced by Tajfel
(1970) establish that in-group bias
occurs even when care is taken to
assure that in-group and out-group
members are objectively similar in
every respect other than group
membershipfor example, by allowing
subjects to observe the random
assignment of persons, themselves
included, to groups within the
experimental situation. In this form,
in-group bias is identifed as the
minimal
group
effect.
Recent
discussions of in-group bias and
minimal group effects can be found in
Brewer (1979), Gaertner, Mann,
Dovidio, Murrell, and Pomare (1990),
Mes- sick and Mackie (1989), Mullen,
Brown, and Smith (1992), Os- trom
and Sedikides (1992), Tajfel and
Turner (1986), and Wilder and
Shapiro (1984). The minimal group
effect indicates that shared group
membership provides a sufficient link
to self to permit implicit operation of
self-esteem. Further evidence that
66
6.
COGNITIVE
LEARNING AND
minimal in-group cues may suffice as a

basis for transfer of positive affect
comes from a study by Perdue,
Dovidio, Gurt- man, and Tyler (1990),
who reported that in-group pronouns
(us and we) served as affectively
positive primes in a subliminal
activation procedure.
Naturally Mediated Implicit Self-Esteem

Effects
Similarity-attraction.
Perceived
similarity of opinions with another
person has been demonstrated to be a
powerful determinant of attraction
(see Byrne, 1969, for an overview of
the series of studies that initially
demonstrated this phenomenon). In
typical
similarity-attraction
experiments, the subject makes
judgments about a stranger (stimulus
person) after having seen a sample of
the strangers responses to a series of
opinion questions. Surreptitiously, the
strangers opinion responses have been
constructed so as to defne varying
levels of similarity to the subjects own
previously recorded opinion responses.
The measure of attraction consists of
summed responses to two judgment
items, one indicating liking and the
other a judgment of how enjoyable it
would be to work together with the
stranger.
In
a
representative
experiment by Byrne (1962), 0-7
opinion responses by the stimulus
person were manipulated to agree
with those of the subject. Attraction
was directly and strongly influenced by
the number of these agreements. The
interpretation of similarity-attraction
fndings as implicit self-esteem effects
depends on the assumption that
increasing numbers of opinions shared
with the stranger constitute increasing
strength of the link of the stranger to
self.12
12 There has recently been some debate
COGNITIVE
RESPONSE
Unlike experimental implicit selfesteem effects, for the similarityattraction effect, the link of the novel
stimulus object (stranger) to self is
mediated by attributes of this object
that have a preexisting association to
self. An interesting variant of the
similarity-attraction effect was created
by Finch and Cial- dini (1989), who led
subjects to believe that they either
shared or did not share a birthday
(same day and month) with a
notorious
historical
character,
Rasputin. Subjects were more lenient
in judging the deeds of Rasputin when
the shared birthdate created a link to
self. Subsequently, Prentice and Miller
(1992) used Finch and Cialdinis
shared birthday technique to increase
bonds between the subject and
another participant in a prisoners
dilemma
negotiating
situation.
Subjects who believed that they
shared a birthdate with the other
player cooperated signifcantly more
than did subjects who were not
provided this (false) information.
Cognitive responses to persuasion.
Novel persuasive arguments tend to
be accepted as valid to the extent that
their conclusions agree with ones
existing opinions. This effect was established in the late 1960s (e.g.,
Cullen, 1968; Greenwald,
1968) . A strong demonstration of the
over the relative contribution of similarity
versus dissimilarity cues to similarityattraction fndings (Byrne, Clore, &
Smeaton, 1986; Rosenbaum, 1986). The
opposing views in this debate ft equally with
an interpretation in terms of implicit selfesteem, but differ in their assumptions about
the relative importance of dissimilarityinduced negative attitudes and similarityinduced positive attitudes.
effect appeared in a study by Lord,

Ross, and Lepper (1979), in which
students in favor of or opposed to
capital punishment examined two sets
of evidence, one set supporting capital
punishment
and
one
opposing.
Students judgments of the relative
convincingness of the two sets of
evidence were strongly shaped by their
preexisting opinions. Each opinion
group found that the evidence, on balance, favored their own position. This
is due to the combinations of
tendencies to be attracted to the
arguments supporting ones own
position and to be repelled by those
supporting the opposing position. It is
therefore apparent that the cognitive
response effect is a close relative of the
similarity-attraction effect. Instead of
opinions
being
credited
to
a
hypothetical stranger (the similarityattraction effect), they are included in
a persuasive communication. Instead
of the effect being greater liking or
disliking for the stranger with
agreeable or disagreeable opinions, it
is
greater
liking
for
the
communication. The tendency to reject
a communication that does not
support ones own views is shown
especially strongly in the form of the
hostile media phenomenon (Vallone,
Ross, & Lepper, 1985), in which each
of two politically opposed groups
interpreted the same (presumably
unbiased) news media report of a
terrorist attack as objectionable
because of its lack of support for their
own position.
Postdecisional spreading of choice
alternatives.
In
this
procedure,
subjects are asked to rate several
members of a product category (such
as music recordings), then to choose to
receive one of two that were initially
rated close to one another, and fnally
67
to rate all of the alternatives again. In

comparison with control subjects who
make no choice, choice subjects show
an increase in rated attractiveness of
the chosen alternative relative to the
nonchosen one (Festinger & Walster,
1964; Steele, Spencer, & Lynch, 1992).
This, again, is a pattern of increased
positivity toward a stimulus that has
a link with self. In this case the link to
self is established by the subjects
choosing to receive the object. 13 A
powerful variant of this effect was
demonstrated by Langer (1975), who
sold lottery tickets to office workers for
$1 and later approached them to sell
the ticket to another buyer. Langer
varied whether, on the initial
purchase, subjects had been allowed to
choose their own ticket or were merely
handed
13 Postdecisional spreading of choice

alternatives is well known as a prediction of
cognitive dissonance theory. Interpretations
of cognitive dissonance reduction as often
occurring in the service of maintaining a
positive self-image (e.g., Aronson, 1968,
1992), or in the service of self- affirmation
(Steele, 1988), or in the service of
maintaining self-esteem (Deutsch, Krauss, &
Rosenau, 1962; Greenwald & Ronis, 1978)
potentially place the entire class of
dissonance-reducing cognitive changes into
the category of implicit self-esteem effects.
68
6.
COGNITIVE
LEARNING AND
8
COGNITIVE
RESPONSE
ATTRACTIVENESS RATING (9-polnt scale)
one. The dependent variable was the

price the subject was willing to accept
to sell the ticket. In the choice
condition, the average price asked for
the ticket was $8.67, compared with $
1.96 in the no-choice condition. Here,
a link of the item to self, however
created, increased monetary value
above the $ 1 purchase price (cf. the
mere ownership effect, above), but the
Figure 1. An implicit self-esteem effect (the initial letter effect). Subjects judged letters in
increase was much greater with the their names to be more attractive than those not in their names, with most of this effect
stronger link produced by active choice. being explained by liking for frst and last initial letters. (Data provided by 597 subjects,
with means based on a minimum of 13 observations; from M. M. S. Johnson, 1986.)
Liking for name letters. When

subjects are asked to choose a
preferred letter from each of several
pairs consisting of one letter from their
name and one not (with subjects not
being alerted to this aspect of the pairs
composition), they tend reliably to
prefer letters from their name (Nuttin,
1985). Of several possible explanations
for the effect, the most successful has
been one based on ego attachment; that
is, the preference for letters in ones
name reflects a positive attitude to self
(Hoorens, 1990; M. M. S. Johnson,
1986; Nuttin, 1985). Using a letterattrac- tiveness rating task to test this
name letter effect, M. M. S. Johnson
(1986)
obtained
the
previously
unpublished data shown in Figure 1.
These results reveal a statistically
signifcant (but weak) name letter
effect and a considerably stronger
initial letter effect, that is, enhanced
liking for letters that constitute the initials of ones name. Because initial
letters are linked more strongly to self
than are other name letters, M. M. S.
Johnsons fnding strengthens the
interpretation of the name letter effect
as a manifestation of implicit selfesteem.14
14 It might be supposed that greater
frequency of exposure to initial letters than
other name letters can explain why the name
Second-Order
Effects
Implicit
Self-Esteem
Self-positivity in judgment. There

have
been
many
research
demonstrations of judgment biases
that work to cast a positive light on
the self. These fndings of selfpositivity biases are not reviewed here
because earlier works (cited in the
next paragraph) have documented
several of them thoroughly. Self-positivity biases have special signifcance
for the present analysis because they
do more than express self-esteem
implicitly; they also provide support
for self-esteem and are, therefore,
second- order implicit self-esteem
effects.15
letter effect is strongest for initials. However,
this interpretation is implausible both
because empirical exposure frequency effects
asymptote at much lower frequencies than
the huge frequencies characteristic of letters
of the alphabet, and any frequency
advantage for initials over other name letters
must be quite small when computed as a
percentage difference in total exposures to
the letters.
15 Some of the naturally mediated implicit
self-esteem effects described previously also
likely serve to support self-esteem, but even
less directly than do those considered in this
section. For example, by helping to assure
The
tendencies
to
accept
responsibility for desired outcomes
while fnding external causes for
undesired
outcomes
and,
more
generally, to construct judgments and
revise memory in a fashion consistent
with a positive self-image were
identifed by Greenwald (1980) as
symptoms of one of a trio of cognitive
biases (egocentricity, beneffectance,
and cognitive conservatism) of the self.
These
biases,
which
share
characteristics with the operation of a
totalitarian
societys
propaganda
apparatus, were described as adaptive,
functioning to preserve the integrity of
the self as a knowledge organization.
S. E. Taylor and Brown
(1988) more
recently
reviewed
evidence for self-positive illusions,
strongly documenting the case for the
adaptive functions of these biases.
And, on the basis of their analyses of
cognitive processes of depressive
patients, Beck (1979), Scheier and
Carver (1992), and Seligman (1991)
have indicated both the prominence
and the adaptiveness of self-positivity
in normal (nondepressive) cognitive
functioning.
One example of self-positivity in
judgments warrants mention here
because of its special relevance to this
articles focus (in a section to follow) on
stereotypes and prejudices. This is
Crosbys (1984) fnding that members
of disadvantaged groups, even though
viewing other members of their own
group as targets of discrimination,
tend not to see themselves as having
that ones acquaintances will express
agreeable views, the similarity-attraction
effect protects against exposure to opinion
disagreements that might undermine
confdence in the wisdom of ones opinions.
69
been so victimized. This phenomenon,

which can be linked to just-world
illusions (Lemer, 1980), has been
confrmed in various samples in the
United States (Crosby, Pufall, Snyder,
OConnell,
&
Whalen,
1989),
Francophones in Quebec (Gui- mond &
Dube-Simard, 1983), and Haitian and
Indian immigrants in Canada (D. M.
Taylor, Wright, Moghaddam, & Lalonde, 1990).
Implicit affiliation and rejection.
Festingers (1954) theory of social
comparison
was
based
on
a
supposition that people compare
themselves with others in order to gain
an accurate assessment of their
opinions and abilities (see also Suls &
Wills, 1991). In more recent variations
of the theory, especially Tes- sers selfevaluation
maintenance
theory
(Tesser, 1988; Tesser & Campbell,
1983) and Willss (1981) analysis of
downward comparison, the potential
for social comparisons to occur in the
service of self-enhancement has been
emphasized. This modern view of
social comparison describes selfesteem
as
influenced
by
the
interrelationship of perceived relation
of self to other (person or group) and
evaluation of other. High self-esteem
accompanies either association to
positively valued others or dissociation
from negatively valued others.
This
general
principle
that
affiliations and rejections are in the
service of self-esteem is supported by
the previously described (naturally
mediated)
implicit
self-esteem
phenomena of similarity-attraction, ingroup bias, and biased cognitive response to persuasion, all of which
involve effects of variations of
perceived relation to another (person
or communication) on attraction. It
70
6.
COGNITIVE
LEARNING AND
equally encompasses several wellestablished phenomena in which

perceived relation to other is the
dependent variable, rather than a
manipulated independent variable.
Several studies have varied the others
good or bad fortune and observed the
consequences for both perceived
association with and liking for the
other. Basking in reflected glory
(Cialdini et al.,
1976) is
perceiving
a
positive
association with (and presumably
admiring) a fortunate or successful
other, whereas results characterized as
showing downward comparison (Suls,
1991; Wills, 1981) involve distancing
oneself from and being critical of a
disadvantaged,
unsuccessful,
or
unfortunate other.
Phenomena of implicit affiliation
and rejection also reveal second-order
implicit self-esteem, because either
association with an attractive other or
distancing from an unattractive other
should bolster self-esteem. Support for
this interpretation comes from studies
in which tendencies to bask in
reflected glory or to engage in selfenhancing downward comparison were
increased by procedures, such as
failing at an ego-involving task or
learning
that
one
possesses
undesirable personality traits, that
temporarily lowered subjects selfesteem (e.g., Cialdini & Richardson,
1980; Crocker, Thompson, McGraw, &
In- german, 1987). Further support
comes from studies in which subjects
are shown to be biased toward
embracing,
as
self-descriptive,
whichever of two opposing traits
(extraversion or introversion) is
presented as being predictive of
personal success (Kunda & Sanitioso,
1989; Sanitioso, Kunda, & Fong, 1990).
COGNITIVE
RESPONSE
Displaced self-esteem. Displaced selfesteem (which has not previously been

identifed as a distinct class of
fndings) is a relative of well-known
reciprocity and ingratiation effects
that involve reciprocal exchange of
gifts, favors, or praise (Jones, 1964;
Regan, 1971; see Cialdini, 1993, for an
overview
of
reciprocity
effects).
Displaced self-esteem starts with an
act of received praise, but thereafter
deviates from reciprocal exchange
phenomena in that the subsequent
return praise is perceived as an
independent judgment that has no
relation to the received praise.
Displaced self-esteem appears to have
been central to a well-known fnding by
Aronson and Linder (1965). In that
experiment, a stimulus person who
switched from criticizing to praising
the subject was liked better than one
who continuously praised the subject,
even though the latter gave more total
praise. Aronson and Linder proposed
that the switch from criticism to praise
prompted greater liking because it
made
the
praiser
seem
more
discriminating
in
interpersonal
judgments. The present (related)
interpretation is that the switch from
criticism to praise provided a more
potent boost to the subjects selfesteem than did the continuous praise,
with the return liking then being a
second-order implicit self-esteem effect;
that is, it further supported the
subjects self-esteem by boosting the
credentials of the person who offered
praise.
In
as-yet-unpublished
studies
conducted at the University of
Washington, a strong displaced selfesteem effect has been identifed in a
natural setting that routinely calls for
two parties to take turns evaluating
each otherthe end-of-course student
ratings of instructors (Greenwald,

1992b). In analyses both within and
across course sections, higher grades
of student by instructor have been
strongly associated with higher ratings
of instructor by student. The positive
evaluation of an instructor by a
student who expects to receive a high
grade appears to provide support,
implicitly, for the students selfesteem.
Alternative
Interpretations of
Implicit SelfEsteem Effects
Each of the implicit self-esteem
effects just reviewed is well established
in multiple studies. Furthermore,
several of these fndings meet
established
criteria
of
being
statistically strong effects (that is, they
correspond to product-moment correlations of at least .50, or to treatment
mean differences on a dependent
measure of at least 0.8 standard
deviations; Cohen, 1988). Also, these
effects come from broadly diverse
areas of research, including attitude
change (role playing and cognitive
response effects), decision making
(postdecision
spreading
of
alternatives), group process (minimal
group effect), interpersonal attraction
(similarity-attraction,
implicit
affiliation or rejection, and displaced
self-esteem effects), personality (selfpositivity biases), and social cognition
(name or initial letter and mere
ownership effects). Not surprisingly,
this diverse collection of fndings has a
correspondingly diverse collection of
existing theoretical interpretations,
including reinforcement theory (similarity-attraction), cognitive balance or
consistency theory (cognitive response
and similarity-attraction), cognitive
71
dissonance theory (role playing and

postdecision spreading of choice
alternatives), and social comparison
theory (downward comparison).
Interpreting various fndings in
terms of implicit self-esteem does not
require abandoning their other,
existing theoretical interpretations. In
some
cases,
the
existing
interpretations
describe
plausible
mechanisms by which implicit selfesteem may operate.16 However, the
implicit self-esteem interpretation
goes beyond other interpretations in
predicting that the various effects
should be moderated by self-esteem.
That is, there should be stronger
implicit manifestations of self-esteem
for subjects who are higher in selfesteem.17
Unfortunately, the hypothesized
moderating role of self-esteem cannot
16 As one example, Heiders (1958)
balance theory analyzes the triad composed
of (a) a perceiver and (b) a positively
evaluated other who is associatively linked to
(c) a previously neutral object. Substituting
self into the role of other yields the implicit
self-esteem effect (liking for a self-associated
object) as the result of Heiders theorized
tendency toward cognitive balance (with all
links positive) in this triad.
17 This prediction of the moderating role
of self-esteem is in some cases quite
compatible with existing theories. For
example, it is readily translated into the
terms of both balance theory and cognitive
dissonance theory. Nevertheless, with only
occasional
exceptions
(e.g.,
Aronson&Mettee,
1968;
Crocker
etal.,
1987;Steeleetal., 1992; Suls, 1991), those
theories have not guided searches for
evidence to evaluate possible relationships of
their predicted effects to individual
differences in selfesteem.
72
6.
COGNITIVE
LEARNING AND
be tested confdently with existing selfesteem measures. The problem is that

standard self-esteem inventories (e.g.,
Helmreich, Stapp, & Ervin, 1974;
Rosenberg, 1965) use a direct
measurement strategy that assesses
an
introspectively
accessible
representation of self-esteem. The
hypothesized moderating role of selfesteem in implicit self-esteem effects is
expected to be effectively testable only
with the aid of a procedure for
(indirectly) measuring implicit selfesteem. Such a measure does not now
exist." Below, the state of available
technology for indirect measurement
of implicit social-cognitive constructs
is discussed.
The Nature ofEvidence for

Implicit Attitudes and Implicit
Self-Esteem
In the absence of individualdifference measures for implicit
attitudes, supporting evidence for that
construct has focused on situations in
which it can be assumed that subject
samples
are
approximately
homogeneous in the content of their
implicit attitudes. For example, the
interpretation of physical attractiveness halo effects as implicit attitude
effects requires the assumption that
physical attractiveness is positively
evaluated by a large majority of most
research samples. Similarly, empirical
demonstrations of implicit self-esteem
require the assumption that positive
self-regard is widespread within
research samples. In this respect,
research on implicit social cognition is
on a similar footing with research on
other forms of implicit cognition. In
implicit memory research, for example,
it is assumed that exposure to a given
set of experimental materials produces
an approximately uniform effect across
subjects in establishing traces that can
COGNITIVE
RESPONSE
later influence performance on indirect

measures.
Implicit Stereotypes
A stereotype is a socially shared set
of beliefs about traits that are
characteristic of members of a social
category. Whereas an attitude implies
a consistent evaluative response to its
object, a stereotype may encompass
beliefs with widely diverging evaluative
implications.
For
example,
the
stereotype of members of a certain
group
(e.g.,
cheerleaders)
may
simultaneously include the traits of
being physically attractive (positive)
and
unintelligent
(negative).
Stereotypes guide judgment and action
to the extent that a person acts toward
another as if the other possesses traits
included in the stereotype.
As was the case for attitudes,
scholarly defnitions of stereotypes
have generally not specifed their
conscious or unconscious operation.
This can be seen, as it was for the
attitude construct, by examining a list
of influential defnitions.
A stereotype is a fxed impression,
which conforms very little to the fact it
pretends to represent, and results from
our defning frst and observing second.
(Katz & Braly, 1935, p. 181)
A stereotype is an exaggerated belief
associated with a category. (Allport,
1954, p. 191)
A categorical response, i.e., membership
is sufficient to evoke the judgment that
the stimulus person possesses all the
attributes belonging to that category.
(Secord, 1959, p. 309)
An ethnic stereotype is a generalization
made about an ethnic group, concerning
a trait attribution, which is considered
to be unjustifed by an observer.
(Brigham, 1971, p. 13)
A set of beliefs about the personal

attributes of a group of people.
(Ashmore & Del Boca, 1981, p. 16)
In stereotyping, the individual: (1)
categorizes other individuals, usually
on the basis of highly visible
characteristics such as sex or race; (2)
attributes a set of characteristics to all
members of that category; and (3)
attributes that set of characteristics to
any individual member of that category.
(Snyder, 1981, p. 415)
A cognitive structure that contains the
perceivers knowledge, beliefs, and
expectancies about some human group.
(Hamilton & Trolier, 1986, p. 133)
As in the case of attitudes, it is useful

to examine the extent to which
stereotypes operate implicitly, outside
of conscious cognition.
Implicit
stereotypes
are
the
introspectively unidentifed (or inaccurately identifed) traces of past
experience that mediate attributions of
qualities to members of a social
category.
Although research on stereotypes

has often used direct or explicit
measures (see Judd & Park, 1993),
there are also substantial research
programs on stereotypes that use
indirect measuresones in which a
stigmatizing feature with which a stereotype is associated (e.g., weight,
race, or gender) is peripheral to the
respondents judgment task (e.g.,
Darley & Gross, 1983) or in which the
purpose of investigation is otherwise
disguised (e.g., Hamilton & Gifford,
1976). Crosby, Bromley, and Saxe
(1980) were able to locate enough
research using indirect measures of
prejudicial stereotypes to conclude that
anti-Black sentiments are much more
prevalent among White Americans
than the survey data [i.e., direct or
explicit measures of stereotypes] lead
one to expect. Although the use of
indirect measures in these studies
often reflects the researchers intent to
73
avoid intrusion of unwanted demand or

impression-management artifacts
(which would plausibly suppress
accurate expressions of conscious
stereotypes), some of the research that
is summarized just below was designed
specifcally to investigate unconscious
operation of stereotypes. These studies
suggest that stereotypes are often
expressed implicitly in the behavior of
persons who explicitly disavow the
stereotype. The next two sections focus
attention on race and gender
stereotypes because these, having been
much more heavily investigated than
other stereotypes, have provided the
most persuasive evidence for implicit
stereotyping.
Implicit Race Stereotyping

Numerous fndings have established
automatic operation of stereotypies.
Gaertner and McLaughlin (1983)
presented subjects with pairs of letter
strings, requesting a yes judgment if
both were words, and no otherwise.
Using speed of yes responses to
measure
strength
of
existing
associations between the two words in
a pair, they found that White subjects
responded reliably faster to whitepositive word pairs than to black-positive pairs (e.g., white-smart vs. blacksmart). This difference did not emerge
on judgments of negative traits (e.g.,
white-lazy vs. black-lazy). These
results occurred similarly for subjects
who scored high and for ones who
scored low on a direct (i.e., standard
self-report) measure of race prejudice.
In a related study, Dovidio et al. (1986)
used the procedure of presenting a
prime (black or white) followed by a
target (a positive or negative trait) and
asking subjects to judge if the target
trait could ever be true or was
always false of the prime category.
Again, subjects responded reliably
faster to positive traits that followed
74
6.
COGNITIVE
LEARNING AND
the prime white than black, and in this

study they also responded faster to
negative traits that followed the prime
black than white. These results were
interpreted by Gaertner and Dovidio
(1986) as evidence for aversive racism,
which they defned as a conflict
between
feelings
and
beliefs
associated with a sincerely egalitarian
value system and unacknowledged
negative feelings and beliefs about
Blacks (p. 62).
Automatic operation of stereotypes
provides the basis for implicit
stereotyping. Devine (1989) reported
that, after being sub- liminally
exposed to a series of words, 80% of
which were stereotypically associated
with Black Americans (e.g., poor, jazz,
slavery, Harlem, and busing), White
subjects, in an ostensibly unrelated
second task, judged a race-unspecifed
male target to be more hostile than did
subjects for whom only 20% of the
words had the stereotype association.
Again, as was the case in the Gaertner
and McLaughlin (1983) study, this
result occurred equally for subjects
who scored high and low on a direct
measure
of
prejudice.
Because
Devines subjects may have imagined a
White target on the second task, it is
possible that this result was due to
some consequence of the priming
procedure other than its activation of a
Black stereotype (for example, it might
have activated hostility). Nevertheless,
Devines
study
pioneered
in
identifying an implicit social cognition
effect and in suggesting the role of
automatic (unconscious) processes in
stereotyping and prejudice. Gilbert
and Hixon (1991) showed that a racial
stereotype, presumably activated by
including an Asian woman in a
videotaped sequence seen by subjects,
influenced subsequent word-fragment
completions (a type of measure used
COGNITIVE
RESPONSE
frequently
in
implicit
memory
research). In the context of considering
phenomena of implicit social cognition,
this fnding, along with Devines,
might be well described as revealing
implicit racism.
Implicit Gender Stereotyping

Some of the methods used in studies of
automatic components of ethnic and
race stereotypes have begun to appear
in studies of gender stereotypes (e.g.,
Jamieson & Zanna, 1989; Klinger &'
Beall, 1992; Paulhus, Martin, &
Murphy, 1992). Also, a well-established
line of research based on the fnding
that essays were judged more favorably
when attributed to authors with male
rather than female names (Goldberg,
1968) can be interpreted as indirectly
assessing a gender stereotype that
associates men with greater
achievement than women.18 The
repeated fnding of lower ratings of the
same achievement
18 In a meta-analytic review of this

research, Swim, Borgida, Maru- yama, and
Myers (1989) concluded that evidence for this
indirect form of gender stereotyping was, on
average, weak. However, examination of the
portions of their meta-analysis that
corresponded most closely to Goldbergs
(1968) original procedures does support the
conclusion that Goldbergs fnding is well
established (cf. Banaji & Greenwald, 1994).
s
when credited to women rather than to
men might also be interpreted as reflecting
greater attitude positivity toward men.
However, because of several fndings that
show greater attitudinal positivity toward
women than men (Eagly & Mladinic, 1989;
Eagly, Mladinic, & Otto, 1991; also
unpublished fndings obtained by the
present
authors),
the
stereotype
interpretation appears preferable.
Stereotypical Gender Bias in Fame

Judgments
The stereotypic association of gender
with achievement provided a basis for the
present authors to adapt methods of implicit memory research to investigate
implicit stereotypes. Jacoby, Kelley, Brown,
and Jasechko (1989; see also Jacoby &
Kelley, 1987) demonstrated the operation of
implicit memory in fame judgments. In
their research, on Day 1 of a two-session
experiment subjects read a list that
contained names of both famous and
nonfamous people. On Day 2, 24 hr later,
the same subjects were presented with a list
containing old (previously seen on Day 1)
and new (unseen on Day 1) nonfamous
names, interspersed with old and new
famous names. Subjects judged each name
on the Day 2 list in response to the
question: Is this person famous? (to be
answered yes or no). Jacoby et al. hypothesized that, although episodic memory (i.e.,
explicit, conscious recollection) for the
nonfamous names would fade over the 24hr delay, residual perceived familiarity due
to the prior exposure could lead to
erroneous judgments of fame. That is,
subjects might misattribute the familiarity
of a name to fame and, as a consequence,
would judge more old (than new) nonfamous
names as famous. As they predicted, Jacoby
et al. (1989) found a higher false-alarm rate
for old nonfamous names than for new ones.
This effect, described by Jacoby et al. (1989)
as making nonfamous names become
famous overnight. indicates a

unconscious influence of memory.
75
potent
Adapting the Jacoby et al. (1989)

procedure, Banaji and Greenwald (in press)
used equal numbers of male and female
names (Jacoby et al., 1989, had used a large
majority of male names) and varied the
gender of nonfamous names by attaching a
female or male frst name to a common last
name (e.g., Peter Walker or Susan Walker).
Data from each subject's judgments for each
of the four within-subject conditions (old
male, new male, old female, and new
female) were reduced to a hit rate
(proportion of famous names correctly
judged famous) and a false-alarm rate
(proportion of nonfamous names mistakenly
judged famous). Signal-detection analysis
(Green
&
Swets,
1966)
permitted
decomposition of the hit and false-alarm
data into measures of sensitivity (d ') to the
stimulus variable (name fame in this case),
and threshold or criterion ( 0 ) for assigning
the judgment. The analyses used these two
derived measures, replacing f t with its
logarithm, which is better suited for
statistical analyses because of the greater
approximation
to
normality
of
its
distribution.
The main fndings are graphed in Figure
2. As previously mentioned, hit rates
(correct identifcations of famous names)
were greater for male than female names;
false alarms were greater for old (previously
seen) nonfamous names than for new ones;
and the boost in false-alarm rates for old
names was greater for male than female
names. The signal-detection analysis
indicated signifcantly greater sensitivity to
the fame varia-
76
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
20
g ioo
0.
0
5
tion { d ' ) for new (i.e., not previously seen)

male names than for any other category.
However, the strongest results from the
signal-detection analysis were on criterion
(indexed by log beta). The criterion for fame
judgments was lower (i.e., names were more
readily judged famous) for old than new
names and for male than female names. In
addition to these two main effects on
criterion, there was a signifcant interaction
effect such that the familiarization
procedure (presentation on Day 1) had a
greater effect of lowering criterion for male
than for female names.
These results clearly show that an
ambiguous stimulusa name that elicits a
sense of familiarityis more likely to produce a mistaken judgment of fame if it is
male rather than female. This adaptation of
an implicit memory experimental procedure
provides clear evidence for implicit gender
stereotypes that associate male gender, more
than female gender, with achievement.
Because names serve as flexible vehicles for
communicating social categories (especially
ethnic ones), and because judgments other
than the fame judgment can be used as
indirect measures of traits other than
achievement, the implicit memory procedure
is potentially adaptable to examine a broad
variety of stereotypies (cf. Klinger & Beall,
1992). By using
a
0.
75
0
<5
<
o>
50
to
25
male femalemate
old names
female
newnames
0
4.0
0
B
3.0 male female
male
female
3.0
re
<5
J
3
O
)
2.0
1.
0
0.
Figure 2. An implicit
stereotyping effect. (Data from Banaji & Greenwald, in
0
press, averaged over four experiments, total N = 186; procedure described in

text.) Signal-detection analysis indicated that subjects were somewhat more
sensitive (d') to male than female fame for new names, which were presented
for the frst time in Session 2. However, the most substantial result was that
subjects had a lower criterion (/3) for judging fame for male than female names,
especially for old names, which had been increased in familiarity by being
presented in Session 1. (Hit and false-alarm rates are, respectively, the tall and
the short bars in the top panel.)
large set of names, such procedures are protected against the threat that
stereotyping effects are due to idiosyncratic properties of the particular names
selected as representative of larger categories (see Kasof, 1993).
Stereotypical Gender Bias in Judgments of Dependence and Aggression

Higgins, Rholes, and Jones (1977; also Srull & Wyer, 1979) demonstrated that
the presentation of trait-category information in one context can influence
judgments of an ambiguously described target person in an unrelated context.
Banaji, Hardin, and Rothman (1993) used a variant of Higgins et al.s procedure to
extend the concept of implicit stereotyping. On the basis of documented
stereotypes that link males to the trait of aggressiveness and females to the trait
of dependence (see Basow, 1986; Broverman, Vogel, Broverman, Clarkson, &
77
Rosen- krantz, 1972), Banaji et al. (1993) asked whether activating one of these
traits would differentially influence judgments about male and female target
persons. The prediction was that the activated trait might influence judgments
only for the category (male or female) for which the trait was stereotypically
appropriate, thereby demonstrating implicit operation of the stereotype.
In Banaji et al.s (1993) dependence experiment, subjects who were exposed to
primes that described dependent behaviors judged a female target (Donna) as
more dependent than did subjects who rated the same target after exposure to
neutral primes. However, subjects exposed to exactly the same dependence primes
judged a male target (Donald) as less dependent than did subjects who rated the
target after exposure to neutral primes. In the aggression experiment, subjects
who were exposed to primes that described aggressive behaviors judged the male
target as more aggressive than subjects who rated the same target after exposure
to neutral primes. When judging a female target, previous exposure to the same
aggression primes did not affect judgment. A third experiment replicated this
pattern of data for dependence ratings (i.e., a more extreme judgment of a female
target after exposure to dependence than neutral primes, but no such difference
for a male target) and also showed no relationship between such judgments and
explicit memory for the primes.
These experiments demonstrated that a gender stereotype moderates, and may
be a necessary precondition of, the Higgins et al. (1977) trait-priming effect. Like
the previous demonstration of gender bias in fame judgments, this result involves
implicit social cognition; it occurs without the subject being consciously aware of
an influence of recent experience (name exposure and trait exposure, respectively).
At the same time, these effects reveal gender stereotypes because they occur selectively when the information content of recent experience stereotypically fts the
social category (gender) of the target of judgment. The lack of subject sex effects in
both procedures indicates that the stereotypes they reflect are culturally shared
among both men and women, rather than being stereotypes of an out-group by an
in-group (cf. Jost & Banaji, 1994).
Attention as a Moderator of Implicit Cognition
Decisions that affect peoplefor example, personnel evaluations and

admissions decisionscomprise a large and very important class of situations in
which implicit cognition can intrude on deliberate judgment, with the result of
producing unintended discrimination. The decision maker who intends to
maintain a nonprejudicial course in these judgments may have little basis for
knowing whether or how a specifc cue is implicitly intruding on judgment.
Consequently, decision makers could usefully be acquainted with general
strategies for reducing implicit cognitive effects, even when they are ignorant of
the precise form that those effects may take.
78
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
Existing research indicates an important role for attention in general strategies

for reducing undesired implicit influences on judgment. The general principle is
that attentional focus attenuates weak automatic influences on judgment. This
principle can be observed in operation in settings in which the weak automatic
influence produces both desired and undesired effects.
Effect of Attention on Weak Automatic Influences That Produce Desired

Responses
In perception research, Wyatt and Campbell (1951; see also Bruner & Potter,
1964) demonstrated that asking subjects to generate hypotheses about the identity
of a blurred picture reduced their ability to identify the picture as it was gradually
brought into focus. T. D. Wilson and Schooler (1991; see also T. D. Wilson, Dunn,
Kraft, & Lisle, 1989) reported that subjects who were asked to introspect on their
reactions to several brands of jelly subsequently rank ordered those brands in a
fashion more discrepant from expert rankings than did subjects not given the
opportunity to introspect.
In research on memory, Schooler and Engstler-Schooler (1990) found that, after
attempting to verbally describe the memory image of a previously seen face,
subjects showed impaired ability to choose that face from an array that included
several foils. Graf and Mandler (1984; see also Overson & Man- dler, 1987) found
that deliberate attempts to complete word stems with words recently seen (but
poorly attended) resulted in fewer correct responses than a comparison condition
in which subjects simply attempted to provide completions. Research on
hypermnesia has demonstrated that subjects will sometimes show improved recall
after relaxing their efforts to retrieve, compared with sustained retrieval efforts
(Erdelyi & Kleinbard, 1978; Madigan, 1976). In other words, when memory traces
are weak, active effort to retrieve (using direct measures) may interfere with
retrieval compared with more relaxed efforts that approximate indirect
measurement procedures.
For all of these results, it is plausible that subjects attentional efforts disrupted
the influence of weak cueseither perceptual cues or memory tracesthat might
otherwise have guided judgment. Marcel (1983) observed that subliminal semantic
activation was more evident among subjects who took a relaxed approach to their
experimental task than among those who tried effortfully to extract information
from stimuli that were degraded by backward masking. Mandler (in press)
concluded that the conscious/deliberate attempt to retrieve . . . sublim- inally
presented material. . . seems to interfere with access. In other words, for
detecting effects of weak stimuli, direct measures (which focus attention on task
stimuli) can be less sensitive than indirect measures.
Effect of Attention on Automatic Influences That Produce Undesired Responses

Distraction increases implicit effects. In the social cognition domain, the
following several fndings have demonstrated that decreased attention, due to
79
distraction or time pressure, results in increased implicit effects of cues that are
peripheral to the subjects task. Kruglanski and Freund (1983) found that imposing time pressure on a judgment task (thereby reducing attentional resources
available for the task) increased the level of ethnic stereotyping in
subjectsjudgments. Similar fndings involving gender stereotyping were obtained
by Jamieson and Zanna (1989) and Pratto and Bargh (1991). Gilbert and Hixon
(1991) used a distracting memory load to reduce attentional resources available
for a word-completion task that was used as an indicator of ethnic stereotyping
and found increased evidence of stereotyping on the indirect measure. (Although
Gilbert and Hixon reported, as just described, that stereotype expression was
increased by distraction, they also reported that stereotype activation was reduced
by distraction.) Paulhus and Levitt (1987) placed affect-arousing words adjacent to
trait words that were being judged as self-descriptive or not and found that this
procedure increased self-positive responding (endorsement of positive traits or
rejection of negative ones). One interpretation of Paulhus and Levitts fnding is
that the affect-arousing words diverted attentional resources from the judgment
task, allowing implicit self-esteem to be a stronger influence on judgment than in
the control (nonarousing adjacent words) condition.
Attention to source of implicit effect reduces the effect. Several studies have
demonstrated that when attention is focused on the source of an implicit effect
that interferes with judgment, that interference is reduced or eliminated (and
sometimes even reversed). For example, consistent with the expectation of a
physical attractiveness halo effect, it has been found that subjects award lighter
sentences to attractive criminal defendants (Efran, 1974;Sigall&Ostrove, 1975).
However, when the crime was related to attractiveness (a swindle), Sigall and
Ostrove observed a reversal that was presumably related to subjects increased
focus on the defendants attractiveness and consideration of its relation to the
crime, that is, physically attractive defendants received harsher sentences.
Schwarz and Clores (1983) fnding that quality-of-life judgments were affected by
weather was eliminated when subjects attention was focused on the irrelevant
cue by frst asking them to describe the weather. These studies show that a cue
that redirects attention to the source of a possible implicit effect can produce a
reduction or reversal of that implicit effect. Research by Fiske and Neuberg (1990;
Fiske, 1993; Neuberg, 1989) has shown similarly that increased attention (in their
case, induced by giving subjects a goal of forming accurate impressions) increases
the use of individuating information, rather than category-based stereotypes, in
judging stimulus persons.
Recall of implicit cue decreases implicit effect. In the memory domain, implicit
cognitive effects have been found either to be weakened or reversed for subjects
who could recall the stimuli that ordinarily produce those effects. Jacoby et al.
(1989) showed that the false fame effect was reduced when the initial list of
nonfamous names was well enough attended so that subjects would recognize
nonfamous names as having been encountered earlier in the experiment. E. R.
Smith, Stewart, and Buttram (1992) found that facilitative effects of repeatedly
judging the same stimulus were larger when the stimulus was not recognized as
having been judged previously (7 days earlier) than when it was recognized. In a
80
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
review of much research showing that mere exposure to visual objects increases
their judged attractiveness, Bornstein (1992) observed that this mere exposure
effect was increased by procedures that decreased memorability of prior
exposures. That observation was subsequently bolstered by an experimental test
in which memorability of exposures was reduced by using very brief presentations
(Bornstein & DAgostino, 1992). Lombardi, Higgins, and Bargh (1987; see also
Strack, Schwarz, Bless, Kiibler, & Wanke, 1993) reported that effects of a priming
procedure on trait inferences from an ambiguous description were reversed among
subjects who could recall some of the priming stimuli. Relat- edly, Martin, Seta,
and Crelia (1990) found that this reversal of priming for a recallable priming
manipulation was itself undone when subjects had a distracting task concurrent
with the test of the priming effect, indicating that the reversal observed by
Lombardi et al. (1987) depended on the availability of attentional resources.
Exceptions. It is clear from existing literature that attending to or recalling a
cue does not invariably reduce its cognitive impact (see Martin & Achee, 1992;
Petty & Wegener, 1993; Schwarz & Bless, 1992; Strack, 1992). In an unpublished
replication of Johnsons initial letter effect (see Figure 1) performed by the present
authors, subjects were asked to provide their names in advance, and it was
mentioned that the letters they were about to rate included ones that were in their
names. The authors expected that drawing attention to this cue would reduce its
effect, but it did not; rather, the initial letter effect was undiminished. In
retrospect, it was plausible that the effect was not altered because subjects had no
reason to suppress this particular implicit self-esteem effect. Presumably, one
should even expect augmentation of an implicit effect if the direction of the effect is
consistent with a conscious motive. For example, consider a hypothetical variant
on the initial letter effect experiment in which students are asked to rate
attractiveness of various colors. If, in advance of the rating, the experimenter
points out that the university colors are among those being rated, those colors
might be rated as more attractive than in a control condition in which attention is
not drawn to the association.
Conclusion. The fndings summarized in this section share the point that
conscious attentional effort can weaken the influence of a current or previous cue
on performance. However, the method of weakening likely depends on whether, at
the time of the performance measure, the cue is or is not clearly cog nized. When
attentional effort is directed to a weak stimulus or memory trace, the reduced
effectiveness of that cue is likely due to conscious strategies overriding and
interfering with automatic cognitive effects. This interpretation is very
reminiscent of a principle derived from learning-behavior theories of the 1940s
and 1950sthat increases in motivation or drive amplify stronger habits relative
to weaker ones (e.g., Spence, 1956). On the other hand, when the cue in question is
cognized clearly, reduction of its implicit effect likely occurs because (and only to
the extent that) the judge can anticipate and compensate for the events possible
influence. In this case, reversals of potential implicit effects may occur when the
judge overcompensates for the influence effect, perhaps because the judge
overestimates its magnitude or is overly zealous in seeking to avoid any appearance of having been influenced.
81
Appraisal of Strategies for Avoiding Unintended

Discrimination
Over the last several decades, concern about the prevalence of social
discrimination in employment, education, and other public settings has led to the
proposal and implementation of antidiscrimination strategies that fall into three
categories: Blinding denies a decision maker access to potentially biasing
information. Exactly the opposite of blinding, consciousness raising encourages
the decision maker to have heightened awareness of potential cues that could
elicit discrimination. The third strategy, affirmative action, differs from the other
two in having a deliberate compensatory component: An attribute that is known to
be responsible for adverse discrimination is treated instead as if it were a positive
qualifcation for the decision in question. The treatment here makes no attempt to
acknowledge the substantial existing research on these three strategies, instead
considering their possible effectiveness in situations in which discrimination
might be based on implicit attitudes (for example, based on halo effects or
prejudices) or implicit stereotypes.
Blinding. In principle, blinding appears to qualify as a foolproof method of
avoiding unintended discrimination. If a stigmatizing attribute is unavailable to a
decision maker, it would seem that it could not possibly influence judgment. The
value of blinding appears to be confrmed by experimental tests that permit clean
manipulation of presence versus absence of a potentially stigmatizing attribute,
independent of all other stimulus variation (see, e.g., the review by Eagly,
Makhijani, & Klon- sky, 1992). However, because natural possession versus
nonpossession of almost any socially stigmatizing attribute tends to be correlated
with possession of other characteristics that are not (and often cannot be) removed
by blinding, effective blinding of the sort producible in experiments is often not
achievable in practice.
Consciousness raising. Research on the role of attention in weakening the
effects of implicit cognition (reviewed above) supports consciousness raising as a
strategy for avoiding unintended discrimination. That is, when a decision maker is
aware of the source and nature of a bias in judgment, that bias may effectively be
anticipated and avoided. Consciousness raising may also have some value in
attenuating implicit bias when the source of implicit bias is not properly
identifed, as suggested by fndings that attentional effort reduces effects of weak
cues.
Affirmative action. The controversiality of affirmative action is well captured by
its frequently being described as reverse discrimination. To discriminate in favor
of a disadvantaged group may seem a defect even to those who desire only not to
discriminate against them. At the same time, research on implicit social cognition
provides several bases for concluding that unintended discrimination can be
avoided only by deliberately applying compensatory strategies. Affirmative action
strategies may therefore be deemed suitable not only as compensation to a
stigmatized group for past explicit discrimination by others who intended to
82
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
discriminate against them, but also as compensation for past, present, and likely
future implicit discrimination by persons who have no intent to discriminate. That
is, in addition to (or instead of) their interpretation as reverse discrimination,
affirmative action strategies might be understood as strategies for reversal of
discrimination.
The contrast of affirmative action with blinding can be considered in the context
of a case reported by Allmendinger and Hackman (1993). American symphony
orchestras have a long tradition of predominant male membership, and women instrumentalists have historically not found equal opportunity in obtaining
experience that can allow them to compete effectively with men. Being alert to
possible discrimination, orchestras now routinely have candidates for vacant
positions perform from behind a partition, removing all cues other than the sound
of the performance. Unfortunately, if the performance reveals (as it should) any
benefts accrued from (differential) past experience, then men will maintain
relative success, even in hypothetical cases of equal aptitude and equal ability to
beneft from experience. Disadvantages that are inherited from past discrimination are not undone by blinding.
Measuring Individual Differences in Implicit Social Cognition

Implicit social-cognitive effects have been demonstrated most clearly in
experimental studies in which a group of subjects is uniformly exposed to cues
that influence their subsequent responses on indirect measures; the implicit effect
is sought in comparisons between averaged performances of groups exposed to
different cues. Although such experimental designs effectively demonstrate some
basic properties of implicit social cognition, they do not allow assessment of
individual differences. The many existing individual-difference measures of attitudes, self-esteem, and stereotypes primarily assess introspectively accessible
self-knowledge (and, sometimes, deliberately managed self-presentations). To
measure individual differences in introspectively inaccessible implicit social
cognition, sensitive indirect measures are needed. Two categories of such indirect
measures have received substantial development effort.
Judgment Latency Measures

Uses of judgment latencies to provide indirect measures of social cognition have
been based on the contributions of Dond- ers (1868/1969), Sternberg (1969), and
Posner (1978). Representative uses appear in the work of Bargh (1982), Devine
(1989), Dovidio and Fazio (1992), Niedenthal (1992), Perdue and Gurtman (1990),
and E. R. Smith et al. (1992). Because judgment latencies tend to show
substantial within-person variability, obtaining measures with adequate
reliability requires averaging the subjects response latencies to large numbers of
similar stimuli. Consequently, latency-based indirect measures have been used
chiefly for hypothesis tests that compare measures averaged over groups of
subjects. Recent treatments of the use of latency measures to assess individual
differences in automatic operation of attitudes can be found in works by Bargh et
al. (1992), Roskos-Ewoldsen and Fazio (1992), and Fazio (1990a, 1993).
83
Projective Measures
A very different approach to indirect measurement, based on research of Murray
(1943) and McClelland, Atkinson, Clark, and Lowell (1953), asks a respondent to
generate stories in response to ambiguous photographs or drawings, or to generate
descriptions of what is seen in abstract stimuli (for example, inkblots). The use of
projective measures for implicit motives, and their comparison with measures of
explicit motives, was the subject of an article by McClelland, Koestner, and
Weinberger
(1989) , who concluded that projective and direct measures assess different
constructs. A similar conclusion was reached in a meta-analytic review of
achievement motivation measures by Spangler (1992), who also reported that
projective measures of achievement motivation had greater predictive validity
than did parallel questionnaire (direct) measures.
Other Indirect Measures

There has been a continuing series of well-justifed calls for development of
indirect measures for use in social psychological research (e.g., Campbell, 1950;
Hammond, 1948; Jahodaetal., 1951; Webb, Campbell, Schwartz, Sechrest, &
Grove, 1981). Those calls have not gone unheeded, and there is consequently a
substantial history of social psychological uses of indirect measures, especially in
the domain of prejudice research. For example, indirect measures used by Word,
Zanna, and Cooper (1974) showed that White interviewers maintained greater
physical distance, demonstrated less eye contact, and administered shorter
interviews when interacting with Black (as opposed to White) interviewees.
Milgram, Mann, and Harter (1965) showed that stamped envelopes left in various
public places (lost letters) were more likely to be mailed (by the pass- ersby who
found them) the more favorable were public attitudes toward the addressee
organizations. Gaertner and Bickman
(1971) developed an effective telephone version of the Milgram et al. lost-letter
technique, showing that it plausibly assessed prejudice toward Blacks. Porter,
Geis, and Jennings-Walstedt (1983) showed that greater use of head-of-table
seating as an indicator of leadership for male than female stimulus persons
provided an indirect measure of gender stereotyping. J. D. Johnson, Jackson, and
Gatto (in press) similarly reported an indirect measure of race stereotyping in the
form of greater influence by damaging inadmissible evidence for Black than White
defendants in a simulated trial. Campbell, Kruskal, and Wallace (1966) observed
spontaneous seating aggregation by race in a large classroom, suggesting that
spatial proximity can serve as an indirect indicator of racial attitudes. Greenwald
and Schuh (in press) developed a citation analog of the Campbell et al. (1966)
aggregation measure, showing that scientifc citations aggregate along lines of
author ethnicity (Jewish and non- Jewish).
In summary of existing efforts at indirect measurement of implicit social
cognition: Research on latency decomposition, projective tests, and miscellaneous
84
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
other procedures indicate that indirect measurement of individual differences in

implicit social cognition is possible. At the same time, such measurement has not
yet been achieved in the efficient form needed to make research investigation of
individual differences in implicit social cognition a routine undertaking. When
such measures do become available, there should follow the rapid development of a
new industry of research on implicit cognitive aspects of personality and social
behavior.
Conclusion
Much social cognition occurs in an implicit mode. This conclusion comes from a
reinterpretation of many fndings that indicate the importance of implicit
operation of attitudes, and of the self-esteem attitude in particular, and also from
existing and new evidence for the implicit operation of stereotypes. By adding this
conception of the implicit mode to existing knowledge of the explicit mode of
operation of social psychologys basic constructs, the scope of those constructs is
extended substantially. In addition, many possibilities for application in decisionmaking settings are suggested by interpreting social judgment in terms of an
interaction of implicit and explicit social cognition.
Implicit social cognition overlaps with several concepts that were signifcant in
works of previous generations of psychologists. Psychoanalytic theorys concept of
cathexis contained some of the sense of implicit attitude, and its concept of ego
defense similarly captured at least part of the present notion of implicit selfesteem. Partly under the influence of psychoanalytic theory, in the 1930s and
1940s, attitudes were regarded as capable of unconscious operation. The
authoritarian personality concept (Adorno, Frenkel-Brunswik, Levinson, &
Sanford,
1950) extended the psychoanalytic approach to include social phenomena of
prejudice and stereotyping. At a time when the influence of psychoanalytic theory
in academic psychology was declining, its conceptions of unconscious phenomena
that related to implicit social cognition were being imported into behavior theory
(Dollard & Miller, 1950; Doob, 1947; Osgood, 1957). The New Look in Perception of
the 1950s focused on several phenomena that are interpretable as implicit social
cognition. The developing cognitive approach to these phenomena can be seen in
Bruners (1957) introduction of the concept of perceptual readiness. Still later, the
New Look approach was tied together with psychoanalytic theoretical influences in
a cognitive-psychological account by Erdelyi (1974, 1985).
Importantly, the psychoanalytic, behaviorist, and cognitive treatments just
mentioned all lacked an essential ingredient, that is, they lacked reliable
laboratory models of their focal phenomena that could support efficient testing and
development of theory. The missing ingredient is now available, as cognitive
psychologists have succeeded in producing several varieties of unconscious
cognition reliably in the laboratory (see overviews by Greenwald, 1992a;
Kihlstrom, 1987), and investigations of implicit social cognition are well underway
(see Bornstein & Pittman, 1992; Uleman & Bargh, 1989). The methods of research
85
on implicit memory, in particular, are applicable to the implicit attitude, selfesteem, and stereotype phenomena reviewed in this article. Perhaps the most
signifcant remaining challenge is to adapt these methods for efficient assessment
of individual differences in implicit social cognition.
References
Adams, G. R., & Crane, P. (1980). An assessment of parents and teachers expectations of
preschool childrens social preferences for attractive or unattractive children and adults.
Child Development, 51, 224- 231.
Adorno, T. W., Frenkel-Brunswik, E., Levinson, D. J., & Sanford, R. N. (1950). The
authoritarian personality. New York: Harper.
Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior.
Englewood Cliffs, NJ: Prentice-Hall.
Allmendinger, J. M., & Hackman, J. R. (1993). The more, the better? On the inclusion of
women in professional organizations. Report No. 5, Cross-National Study of Symphony
Orchestras, Harvard University.
Allport, G. W. (1935). Attitudes. In C. Murchison (Ed.), A handbook of social psychology (pp.
798-844). Worcester, MA: Clark University Press.
Allport, G. W. (1937). Personality: A psychological interpretation. New York: Holt, Rinehart, &
Winston.
Allport, G. W. (1954). The nature of prejudice. Cambridge, MA: Addi- son-Wesley.
Arkes, J. R., Boehm, L. E., & Xu, G. (1991). Determinants of judged validity. Journal of
Experimental Social Psychology, 27, 576-605.
Aronson, E. (1968). Dissonance theory: Progress and problems. In R. P. Abelson, E. Aronson,
W. J. McGuire, T. M. Newcomb, M. J. Rosenberg, & P. H. Tannenbuam (Eds.), Theories of
cognitive consistency: A sourcebook (pp. 5-27). Chicago: Rand McNally.
Aronson, E. (1992). The return of the repressed: Dissonance theory makes a comeback.
Psychological Inquiry, 3, 303-311.
Aronson, E., & Linder, D. E. (1965). Gain and loss of esteem as deter minants of interpersonal
attractiveness. Journal of Experimental Social Psychology, 1, 156-171.
Aronson, E., & Mettee, D. (1968). Dishonest behavior as a function o f differential levels of
induced self-esteem. Journal of Personality and Social Psychology, 9, 121-127.
Ashmore, R. D., & Del Boca, F. K. (1981). Conceptual approaches to stereotypes and
stereotyping. In D. L. Hamilton (Ed.), Cognitive processes in stereotyping and intergroup
behavior (pp. 1-36). Hillsdale, NJ: Eribaum.
Balzer, W. K., & Sulsky, L. M. (1992). Halo and performance appraisal research: A critical
86
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
examination. Journal of Applied Psychology, 77, 975-985.

Banaji, M. R., & Greenwald, A. G. (1994). Implicit stereotyping and unconscious prejudice. In
M. P. Zanna & J. M. Olson (Eds.), The psychology ofprejudice, The Ontario Symposium (Vol.
7, pp. 55-76). Hillsdale, NJ: Eribaum.
Banaji, M. R., & Greenwald, A. G. (in press). Implicit stereotyping in judgments of fame.
Journal of Personality and Social Psychology.
Banaji, M. R., Hardin, C., & Rothman, A. J. (1993). Implicit stereotyp ing in person judgment.
Journal of Personality and Social Psychology, 65, 272-281.
Banaji, M. R., & Prentice, D. A. (1994). The self in social contexts. Annual Review of
Psychology, 45, 297-332.
Bargh, J. A. (1982). Attention and automaticity in the processing of selfrelevant
information. Journal of Personality and Social Psychology, 43, 425-436.
Bargh, J. A. (1989). Conditional automaticity: Varieties of automatic influence in social
perception and cognition. In J. S. Uleman & J. A. Bargh (Eds.), Unintended thought (pp. 351). New York: Guilford Press.
Bargh, J. A. (1994). The four horsemen of automaticity: Awareness, intention, efficiency, and
control in social cognition. In R. S. Wyer, Jr., & T. K. Srull (Eds.), Handbook of social
cognition (2nd ed., Vol. 1, pp. 1-40). Hillsdale, NJ: Eribaum.
Bargh, J. A., Chaiken, S., Govender, R., & Pratto, F. (1992). The gener ality of the automatic
activation effect. Journal of Personality and Social Psychology, 62, 893-912.
Bar-Tal, D., & Saxe, L. (1976). Physical attractiveness and its relationship to sex-role
stereotyping. Sex Roles, 2, 123-148.
Basow, S. (1986). Gender stereotypes: Traditions and alternatives. Monterey, CA: Brooks/Cole.
Baumeister, R. F. (1982). A self-presentational view of social phenomena. Psychological
Bulletin, 91, 3-26.
Beck, A. T. (1979). Cognitive therapy of depression. New York: Guilford Press.
Begg, I., Armour, V., & Kerr, T. (1985). On believing what we remember. Canadian Journal of
Behavioural Science, 17, 199-214.
Beggan, J. K. (1992). On the social nature of nonsocial perception: The mere ownership effect.
Berscheid, E. (1985). Interpersonal attraction. In G. Lindzey & E. Aronson (Eds.), The
handbook of social psychology (Vol. 2, pp. 413- 484). New York: Random House.
Bornstein, R. F. (1989). Exposure and affect: Overview and meta-analysis of research, 19681987. Psychological Bulletin, 106, 265-289.
Bornstein, R. F. (1992). Subliminal mere exposure effects. In R. F. Bornstein & T. S. Pittman
(Eds.), Perception without awareness: Cognitive, clinical, and social perspectives (pp. 191-
87
210). New York: Guilford Press.

Bornstein, R. F., & DAgostino, P. A. (1992). Stimulus recognition and the mere exposure
effect. Journal of Personality and Social Psychology, 63, 545-552.
Bornstein, R. F., & Pittman, T. S. (Eds.). (1992). Perception without awareness: Cognitive,
clinical, and social perspectives. New York: Guilford Press.
Brewer, M. (1979). Ingroup bias in the minimal intergroup situation: A cognitive-motivational
analysis. Psychological Bulletin, 86, 307-324.
Brewer, M. (1988). A dual process model of impression formation. In T. K. Srull & R. S. Wyer
(Eds.), Advances in social cognition (Vol. 1, pp. 1-36). Hillsdale, NJ: Eribaum.
Brigham, J. C. (1971). Ethnic stereotypes. Psychological Bulletin, 76, 15-33.
Broverman, I., Vogel, S. R., Broverman, D. M., Clarkson, F., & Rosen- krantz, P. S. (1972). Sex
role stereotypes: A current appraisal. Journal of Sex Roles, 28, 59-78.
Bruner, J. S. (1957). On perceptual readiness. Psychological Review, 64, 123-152.
Bruner, J. S., & Potter, M. C. (1964). Interference in visual recognition. Science, 144, 424-425.
Bruner, J. S., & Tagiuri, R. (1954). The perception of people. In G. Lindzey (Ed.), Handbook of
social psychology (Vol. 2, pp. 634-654). Reading, MA: Addison-Wesley.
Byrne, D. (1962). Response to attitude similarity-dissimilarity as a function of affiliation
need. Journal of Personality, 30, 164-177.
Byrne, D. (1969). Attitudes and attraction. In L. Berkowitz (Ed.), Advances in experimental
social psychology (Vol. 4, pp. 35-89). New York: Academic Press.
Byrne, D., Clore, G. L., & Smeaton, G. (1986). The attraction hypoth esis: Do similar attitudes
affect anything? Journal of Personality and Social Psychology, 51, 1167-1170.
Campbell, D. T. (1950). The indirect assessment of social attitudes. Psychological Bulletin, 47,
15-38.
Campbell, D. T. (1963). Social attitudes and other acquired behavioral dispositions. In S. Koch
(Ed.), Psychology: A study of a science (Vol. 6, pp. 94-172). New York: McGraw-Hill.
Campbell, D. T., Kruskal, W., & Wallace, W. (1966). Seating aggregation as an index of
attitude. Sociometry, 29, 1-15.
Cash, T. F., & Duncan, N. C. (1984). Physical attractiveness stereotyping among Black
American college students. The Journal of Social Psychology, 122, 71-77.
Chaiken, S. (1987). The heuristic model of persuasion. In M. P. Zanna,
J. M. Olson, & C. P. Herman (Eds.), Social influence: The Ontario Symposium (Vol. 5, pp. 339). Hillsdale, NJ: Eribaum.
88
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
Cialdini, R. B. (1993). Influence: Science and practice (3rd ed.). New York: Harper Collins.
Cialdini, R. B., Borden, R. J., Thorne, A., Walker, M. R., Freeman, S., & Sloan, L. R. (1976).
Basking in reflected glory: Three (football) feld studies. Journal of Personality and Social
Psychology, 34, 366- 375.
Cialdini, R. B., & Richardson, K. D. (1980). Two indirect tactics of image management:
Basking and blasting. Journal of Personality and Social Psychology, 39, 406-415.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ:
Eribaum.
Cooley, C. H. (1964). Human nature and the social order. New York: Scribners. (Original work
published 1902)
Cooper, W. H. (1981). Ubiquitous halo. Psychological Bulletin, 90,218- 244.
Crocker, J., Thompson, L. L., McGraw, K. M., & Ingerman, C. (1987). Downward comparison,
prejudice, and evaluations of others: Effects of self-esteem and threat. Journal of
Personality and Social Psychology, 52, 907-916.
Crosby, F. (1984). The denial of personal discrimination. American Behavioral Scientist, 27,
371-386.
Crosby, F., Bromley, S., & Saxe, L. (1980). Recent unobtrusive studies of Black and White
discrimination and prejudice: A literature review. Psychological Bulletin, 87, 546-563.
Crosby, F. J., Pufall, A., Snyder, R. C., OConnell, M., & Whalen, P.
(1989) . The denial of personal disadvantage among you, me, and all the other ostriches. In
M. Crawford & M. Gentry (Eds.), Gender and thought: Psychological perspectives (pp. 7999). New York: Springer- Verlag.
Cullen, D. M. (1968). Attitude measurement by cognitive sampling. Unpublished doctoral
dissertation, Ohio State University.
Darley, J. M., & Gross, P. H. (1983). A hypothesis-confrming bias in labeling effects. Journal
of Personality and Social Psychology, 44, 20- 33.
Dawes, R. M., & Smith, T. L. (1985). Attitude and opinion measurement. In G. Lindzey & E.
Aronson (Eds.), Handbook of social psychology (3rd ed., Vol. 1, pp. 509-566). New York:
Random House.
Deutsch, M., Krauss, R., & Rosenau, N. (1962). Dissonance or defensiveness? Journal of
Personality, 30, 16-28.
Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and controlled components.
Dion, K., Berscheid, E., & Walster, E. (1972). What is beautiful is good. Journal of Personality
and Social Psychology, 24, 207-213.
Dollard, J., & Miller, N. E. (1950). Personality and psychotherapy. New York: McGraw-Hill.
Donders, F. C. (1969). Over de snelheid van psychische processen. Ond- erzoekingen gedaan
89
in het Psyiologish Laboratorium der Utrechtsche Hoogeschool; 1868-1869 [On the speed of
mental processes]. (W. G. Koster, Ed. and Trans.). Acta Psychologica, 30, 412-431. (Original
work published 1868)
Doob, L. (1947). The behavior of attitudes. Psychological Review, 54, 135-156.
Dovidio, J. F., Evans, N. E., & Tyler, R. B. (1986). Racial stereotypes: The contents of their
cognitive representations. Journal of Experimental Social Psychology, 22, 22-37.
Dovidio, J. F., & Fazio, R. H. (1992). New technologies for the direct and indirect assessment
of attitudes. In J. Tanur (Ed.), Questions about questions: Meaning, memory, expression,
and social interactions in surveys (pp. 204-237). New York: Russell Sage Foundation.
Downs, A. C., & Lyons, P. M. (1991). Natural observations of the links between attractiveness
and initial legal judgments. Personality and Social Psychology Bulletin, 17, 541-547.
Eagly, A. H., Ashmore, R. D., Makhijani, M. G., & Longo, L. C. (1991). What is beautiful is
good, but. . . : A meta-analytic review of research on the physical attractiveness stereotype.
Psychological Bulletin, 110, 109-128.
Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. Fort Worth, TX: Harcourt
Brace.
Eagly, A. H., Makhijani, M. G., & Klonsky, B. G. (1992). Gender and the evaluation of leaders:
A meta-analysis. Psychological Bulletin, 111, 3-22.
Eagly, A. H., & Mladinic, A. (1989). Gender stereotypes and attitudes toward women and
men. Personality and Social Psychology Bulletin, 15, 543-558.
Eagly, A. H., Mladinic, A., & Otto, S. (1991). Are women evaluated more favorably than men?
An analysis of attitudes, beliefs, and emotions. Psychology of Women Quarterly, 15, 203-216.
Edwards, A. L. (1957). Techniques of attitude scale construction. New Ybrk: AppletonCentury-Crofts.
Edwards, K. (1990). The interplay of affect and cognition in attitude formation and change.
Efran, J. S. (1974). The effect of physical appearance on the judgment of guilt, interpersonal
attraction, and severity of recommended punishment in a simulated jury trial. Journal of
Research in Personality, 8, 45-54.
Epstein, S. (1973). The self-concept revisited: Or a theory of a theory. American Psychologist,
28, 404-416.
Epstein, S. (1991). Cognitive-experiential self-theory: An integrative theory of personality. In
R. Curtis (Ed.), The relational self: Convergences in psychoanalysis and social psychology
(pp. 111-137). New Ybrk: Guilford Press.
Erdelyi, M. H. (1974). A new look at the New Look: Perceptual defense and vigilance.
Psychological Review, 81, 1-25.
Erdelyi, M. H. (1985). Psychoanalysis: Freuds cognitive psychology. New York: W. H.
Freeman.
90
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
Erdelyi, M. H., & Kleinbard, J. (1978). Has Ebbinghaus decayed with time? The growth of
recall (hypermnesia) over days. Journal of Experimental Psychology: Human Learning and
Memory, 4, 275-289.
Fazio, R. H. (1986). How do attitudes guide behavior? In R. M. Sorren- tino & E. T. Higgins
(Eds.), Handbook of motivation and cognition: Foundations of social behavior (pp. 204-243).
New Ybrk: Guilford Press.
Fazio, R. H. (1990a). A practical guide to the use of response latency in social psychological
research. In C. Hendrick & M. S. Clark (Eds.), Research methods in personality and social
psychology (pp. 74-97). Newbury Park, CA: Sage.
Fazio, R. H. (1990b). Multiple processes by which attitudes guide behavior: The MODE model
as an integrative framework. In M. Zanna (Ed.), Advances in experimental social
psychology (Vol. 23, pp. 75- 110). San Diego: Academic Press.
Fazio, R. H. (1993). Variability in the likelihood of automatic attitude activation: Data reanalysis and commentary on the paper by Bargh, Chaiken, Govender, and Pratto. Journal
of Personality and Social Psychology, 64, 753-758.
Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C., & Kardes, F. R.
(1986). On the automatic activation of attitudes. Journal of Personality and Social
Fazio, R. H., & Zanna, M. P. (1981). Direct experience and attitude- behavior consistency. In
L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 14, pp. 161-202). New
Feingold, A. (1990). Gender differences in effects of physical attractiveness on romantic
attraction: A comparison across fve research paradigms. Journal of Personality and Social
Psychology, 59, 981 -993.
Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7, 117-140.
Festinger, L. (1964). Behavioral support for opinion change. Public Opinion Quarterly, 28,
404-417.
Festinger, L., & Walster, E. (1964). The post-decision process. In L. Festinger (Ed.), Conflict,
decision, and dissonance (pp. 100-112). Stanford, CA: Stanford University Press.
Feys, J. (1991). Briefly induced belongingness to self and preference. European Journal of
Social Psychology, 21, 547-552.
Finch, J. F., & Cialdini, R. B. (1989). Another indirect tactic of (self-) im age management:
Boosting. Personality and Social Psychology Bulletin, 15. 222-232.
Fishbein, M. (1967). A consideration of beliefs and their role in attitude measurement. In M.
Fishbein (Ed.), Readings in attitude theory and measurement (pp. 257-266). New York:
Wiley.
Fishbein, M., & Ajzen, I. (1974). Attitudes toward objects as predictors of single and multiple
behavior criteria. Psychological Review, 81, 59- 74.
91
Fiske, S. T. (1982). Schema-triggered affect: Applications to social perception. In M. S. Clark

& S. T. Fiske (Eds.), Affect and cognition: The seventeenth annual Carnegie Symposium on
Cognition (pp. 55-76). Hillsdale, NJ: Eribaum.
Fiske, S. T. (1989). Examining the role of intent: Toward understanding its role in
stereotyping and prejudice. In J. S. Uleman & J. A. Bargh (Eds.), Unintended thought (pp.
253-283). New York: Guilford Press.
Fiske, S. T. (1993). Controlling other people: The impact of power on stereotyping. American
Psychologist, 48, 621-628.
Fiske, S. T., & Neuberg, S. L. (1990). A continuum of impression formation, from categorybased to individuating processes: Influences of information and motivation on attention
and interpretation. In M. Zanna (Ed.), Advances in experimental social psychology (Vol. 23,
pp. 1-74). San Diego, CA: Academic Press.
Frank, M. G., & Gilovich, T. (1988). The dark side of self- and social perception: Black
uniforms and aggression in professional sports. Journal of Personality and Social
Gaertner, S. L., & Bickman, L. (1971). Effects of race on the elicitation of helping behavior: A
focus on liberals. In P. Katz (Ed.), Toward the elimination of racism (pp. 183-211). New
York: Pergamon Press.
Gaertner, S. L., & Dovidio, J. F. (1986). The aversive form of racism. In S. L. Gaertner & J. F.
Dovidio (Eds.), Prejudice, discrimination, and racism (pp. 61-89). New York: Academic
Press.
Gaertner, S. L., Mann, J. A., Dovidio, J. F., Murrell, A. J., & Pomare, M. (1990). How does
cooperation reduce intergroup bias? Journal of Personality and Social Psychology, 59, 692704.
Gaertner, S. L., & McLaughlin, J. P. (1983). Racial stereotypes: Associations and ascriptions
of positive and negative characteristics. Social Psychology Quarterly, 46, 23-30.
Geis, F. (1993). Self-fulflling prophecies: A social psychological view of gender. In A. E. Beall
& R. J. Sternberg (Eds.), The psychology of gender (pp. 9-54). New York: Guilford Press.
Gilbert, D. T. (1989). Thinking lightly about others: Automatic components of the social
inference process. In J. S. Uleman & J. A. Bargh (Eds.), Unintended thought (pp. 189-211).
New York: Guilford Press.
Gilbert, D. T. (1991). How mental systems believe. American Psychologist, 46, 107-119.
Gilbert, D. T., & Hixon, J. G. (1991). The trouble of thinking: Activation and application of
stereotypic beliefs. Journal of Personality and Social Psychology, 60, 509-517.
Goldberg, P. (1968). Are women prejudiced against women? Transaction, 5, 28-30.
Graf, P., & Mandler, G. (1984). Activation makes words more accessible, but not necessarily
more retrievable. Journal of Verbal Learning and Verbal Behavior, 23, 553-568.
Graf, P., & Schacter, D. (1985). Implicit and explicit memory for new associations in normal
and amnesic subjects. Journal of Experimental Psychology: Learning, Memory, and
92
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
Cognition, 11, 501-518.

Green, D. M., & Swets, J. A. (1966). Signal detection theory and psychophysics. New York:
Wiley.
Greenwald, A. G. (1968). Cognitive learning, cognitive response to persuasion, and attitude
change. In A. G. Greenwald, T. C. Brock, & T. M. Ostrom (Eds.), Psychological foundations
ofattitudes (pp. 147
170). New York: Academic Press.
Greenwald, A. G. (1969). The open-mindedness of the counterattitudinal role player.
Journal of Experimental Social Psychology, 5, 375- 388.
Greenwald, A. G. (1980). The totalitarian ego: Fabrication and revision of personal history.
American Psychologist, 35, 603-618.
Greenwald, A. G. (1990). What cognitive representations underlie social attitudes? Bulletin of
the Psychonomic Society, 28, 254-260.
Greenwald, A. G. (1992a). New Look 3: Unconscious cognition reclaimed. American
Greenwald, A. G. (1992b). Using student ratings to assess instructional quality. Unpublished
manuscript, University of Washington.
Greenwald, A. G., & Albert, R. D. (1968). Acceptance and recall of improvised arguments.
Greenwald, A. G., Klinger, M. R., & Liu, T. J. (1989). Unconscious processing of dichoptically
masked words. Memory and Cognition, 17, 35-47.
Greenwald, A. G., & Pratkanis, A. R. (1984). The self. In R. S. Wyer & T. K. Srull (Eds.),
Handbook of social cognition (Vol. 3, pp. 129-178). Hillsdale, NJ: Eribaum.
Greenwald, A. G., & Ronis, D. L. (1978). Twenty years of cognitive dissonance: Case study of
the evolution of a theory. Psychological Review, 85, 53-57.
Greenwald, A. G., & Schuh, E. S. (in press). An ethnic bias in scientifc citations. European
Journal of Social Psychology.
Guimond, S., & Dube-Simard, L. (1983). Relative deprivation theory and the Quebec
nationalist movement: The cognitive-emotion distinction and the personal-group
deprivation issue. Journal of Personality and Social Psychology, 44, 526-535.
Hamilton, D. L., & Gifford, R. K. (1976). Illusory correlation in interpersonal perception: A
cognitive basis of stereotype judgments. Journal of Experimental Social Psychology, 12,
392-407.
Hamilton, D. L., & Sherman, J. W. (1994). Stereotypes. In R. S. Wyer & T. K. Srull (Eds.),
Handbook of social cognition (2nd ed., Vol. 2, pp. 1-68). Hillsdale, NJ: Eribaum.
93
Hamilton, D. L., & Trolier, T. K. (1986). Stereotypes and stereotyping: An overview of the
cognitive approach. In S. L. Gaertner & J. F. Dovidio (Eds.), Prejudice, discrimination, and
racism (pp. 127-157). New York: Academic Press.
Hammond, K. (1948). Measuring attitudes by error-choice: An indirect method. Journal of
Abnormal and Social Psychology, 43, 38-48.
Hasher, L., Goldstein, D., & Toppino, T. (1977). Frequency and the conference of referential
validity. Journal of Verbal Learning and Verbal Behavior, 16, 107-112.
Hawkins, S. A., & Hoch, S. J. (1992). Low-involvement learning: Memory without evaluation.
Journal of Consumer Research, 19, 212-225.
Heider, F. (1944). Social perception and phenomenal causality. Psychological Review, 51, 358374.
Heider, F. (1958). Interpersonal relations. New York: Wiley.
Helmreich, R., Stapp, J., & Ervin, C. (1974). The Texas Social Behavior Inventory, an
objective measure of self-esteem. JSAS Catalog of Selected Documents in Psychology, 4, 79.
Higgins, E. T. (1989). Knowledge accessibility and activation: Subjectivity and suffering from
unconscious sources. In J. S. Uleman & J. A. Bargh (Eds.), Unintended thought (pp. 75123). New York: Guilford Press.
Higgins, E. T., Rholes, W. S., & Jones, C. R. (1977). Category accessibility and impression
formation. Journal of Experimental Social Psychology, 13, 141-154.
Hoorens, V. (1990). Nuttins affective selfparticles hypothesis and the name letter effect: A
review. Psychologica Belgica, 30, 23-48.
Howard, D. J. (1992). Gift-wrapping effects on product attitudes: A mood-biasing explanation.
Journal of Consumer Psychology, 3, 197 224.
Hull, C. L. (1943). Principles of behavior. New York: Appleton-Century- Crofts.
Hull, C. L. (1952). A behavior system. New Haven, CT: Yale University Press.
Jacoby, L. L., & Dallas, M. (1981). On the relationship between autobiographical memory and
perceptual learning. Journal of Experimental Psychology: General, 110, 306-340.
Jacoby, L. L., & Kelley, C. M. (1987). Unconscious influences of mem ory for a prior event.
Personality and Social Psychology Bulletin, 13, 314-336.
Jacoby, L. L., Kelley, C. M., Brown, J., & Jasechko, J. (1989). Becoming famous overnight:
Limits on the ability to avoid unconscious influences of the past. Journal of Personality and
Jacoby, L. L., Lindsay, D. S., & Toth, J. P. (1992). Unconscious influ ences revealed: Attention,
awareness, and control. American Psychologist, 47, 802-809.
94
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
Jacoby, L. L., Toth, J. P., Lindsay, D. S., & Debner, J. A. (1992). Lectures for a layperson:
Methods for revealing unconscious processes. In R. F. Bornstein & T. S. Pittman (Eds.),
Perception without awareness (pp. 81-120). New York: Guilford Press.
Jacoby, L. L., & Witherspoon, D. (1982). Remembering without awareness. Canadian Journal
of Psychology, 36, 300-324.
Jahoda, M., Deutsch, M., & Cook, S. (1951). Research methods in social relations: With special
reference to prejudice. Part 1: Basic processes. New York: Dryden Press.
James, W. (1890). The principles of psychology. New York: Holt.
Jamieson, D. A., & Zanna, M. P. (1989). Need for structure in attitude formation and
expression. In A. R. Pratkanis, S. J. Breckler, & A. G. Greenwald (Eds.), Altitude structure
and function (pp. 383- 406). Hillsdale, NJ: Eribaum.
Janis, I. L., & King, B. (1954). The influence of role-playing on opinion change. Journal of
Abnormal and Social Psychology, 49, 211-218.
Johnson, J. D., Jackson, L. A., & Gatto, L. (in press). Justice is still not colorblind:
Differential racial effects of exposure to inadmissible evidence. Personality and Social
Psychology Bulletin.
Johnson, M. M. S. (1986). The initial letter effect: Ego-attachment or mere exposure?
Unpublished doctoral dissertation, Ohio State University.
Jones, E. E. (1964). Ingratiation: A social psychological analysis. New York: AppletonCentury-Crofts.
Jost, J. T., & Banaji, M. R. (1994). The role of stereotyping in system- justifcation and the
production of false consciousness. British Journal of Social Psychology, 33, 1-27.
Judd, C. M., & Park, B. (1993). Defnition and assessment of accuracy in social stereotypies.
Psychological Review, 100, 109-128.
Kahneman, D., Knetsch, J. L., & Thaler, R. H. (1990). Experimental test of the endowment
effect and the coase theorem. Journal of Political Economy, 98, 1325-1347.
Kasof, J. (1993). Sex bias in the naming of stimulus persons. Psychological Bulletin, 113, 140163.
Katz, D., & Braly, K. (1933). Racial stereotypes of 100 college students. Journal of Abnormal
Katz, D., & Braly, K. (1935). Racial prejudice and racial stereotypes. Journal of Abnormal
Kihlstrom, J. F. (1987). The cognitive unconscious. Science, 237, 1445 1452.
Kihlstrom, J. F. (1990). The psychological unconscious. In L. A. Pervin (Ed.), Handbook of
personality: Theory and research (pp. 445-464). New Yjrk: Guilford Press.
95
King, B. T., & Janis, I. L. (1956). Comparison of the effectiveness of improvised versus nonimprovised role playing in producing opinion changes. Human Relations, 9, 177-186.
Klinger, M. R., & Beall, P. M. (1992, May). Conscious and unconscious effects of stereotype
activation. Paper presented at the annual meetings of the Midwestern Psychological
Association, Chicago.
Krech, D., Crutchfeld, R. S., & Ballachey, E. L. (1962). Individual in society. New York:
McGraw-Hill.
Krosnick, J. A., Betz, A. L., Jussim, L. J., & Lynn, A. R. (1992). Subliminal conditioning of
attitudes. Personality and Social Psychology Bulletin, 18, 152-162.
Kruglanski, A. W& Freund, T. (1983). The freezing and unfreezing of lay-inferences: Effects
on impressional primacy, ethnic stereotyping, and numerical anchoring. Journal of
Kunda, Z., & Sanitioso, R. (1989). Motivated changes in the self-concept. Journal of
Landy, D., & Sigall, H. (1974). Beauty is talent: Task evaluation as a function of the
performers physical attractiveness. Journal of Personality and Social Psychology, 29, 299304.
Langer, E. J. (1975). The illusion of control. Journal of Personality and Social Psychology, 32,
311-329.
LaPiere, R. (1934). Attitudes versus actions. Social Forces, 13, 230- 237.
Lerner, M. (1980). The belief in a just world: A fundamental delusion. New York: Plenum
Press.
Lewicki, P. (1986). Nonconscious social information processing. New York: Academic Press.
Lombardi, W. J., Higgins, E. T., & Bargh, J. A. (1987). The role of con sciousness in priming
effects on categorization: Assimilation versus contrast as a function of awareness of the
priming task. Personality and Social Psychology Bulletin, 13, 411-429.
Lord, C. G., Ross, L., & Lepper, M. R. (1979). Biased assimilation and attitude polarization:
The effects of prior theories on subsequently considered evidence. Journal of Personality
Madigan, S. (1976). Reminiscence and item recovery in free recall. Memory & Cognition, 4,
233-236.
Mandler, G. (in press). On remembering without really trying: Hyperm- nesia, incubation,
and mind-popping. In C. Umilta & M. Moscovitch (Eds.), Attention and performance XV:
Conscious and unconscious information processing. Cambridge, MA: MIT Press.
Marcel, A. (1983). Conscious and unconscious perception: An approach to the relations
between phenomenal experience and perceptual processes. Cognitive Psychology, 15, 238-
96
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
300.
Martin, L. L., & Achee, J. W. (1992). Beyond accessibility: The role of processing objectives in
judgment. In L. L. Martin & A. Tesser (Eds.), The construction of social judgments (pp. 195216). Hillsdale, NJ: Eribaum.
Martin, L. L., Seta, J. J., & Crelia, R. A. (1990). Assimilation and con trast as a function of
peoples willingness and ability to expend effort in forming an impression. Journal of
Martin, L. L., & Tesser, A. (Eds.). (1992). The construction of social judgments. Hillsdale, NJ:
Eribaum.
May, M., & Doob, L. W. (1937). Competition and cooperation. New York: Social Science
Research Council.
McClelland, D. C., Atkinson, J. W., Clark, R. A., & Lowell, E. L. (1953). The achievement
motive. New York: Appleton-Century-Crofts.
McClelland, D. C., Koestner, R., & Weinberger, J. (1989). How do self attributed and implicit
motives differ? Psychological Review, 96,690- 702.
McGuire, W. J. (1973). The yin and yang of progress in social psychology: Seven koan.
McGuire, W. J. (1985). Attitudes and attitude change. In G. Lindzey & E. Aronson (Eds.),
Handbook ofsocial psychology (3rded., Vol. 2, pp. 233-346). New York: Random House.
Messick, D. M., & Mackie, D. M. (1989). Intergroup relations. Annual Review of Psychology,
40, 45-81.
Milgram, S., Mann, L., & Harter, S. (1965). The lost-letter technique: A tool for social science
research. Public Opinion Quarterly, 29, 437- 438.
Mowrer, O. H. (1960). Learning theory and symbolic processes. New York: Wiley.
Mullen, B., Brown, R. J., & Smith, C. (1992). Ingroup bias as a function of salience,
relevance, and status: An integration. European Journal of Psychology, 22, 103-122.
Murphy, S. T., & Zajonc, R. B. (1993). Affect, cognition, and awareness: Affective priming with
optimal and suboptimal stimulus exposures. Journal of Personality and Social Psychology,
64, 723-729.
Murray, H. A. (1943). Thematic apperception test manual. Cambridge, MA: Harvard
University Press.
Myers, D. G. (1990). Social psychology (3rd ed.). New York: McGraw- Hill.
Myers, D. G., & Ridl, J. R. (1979, August). A better than average insight into pride.
Psychology Today, pp. 89, 95-96,98.
Neuberg, S. L. (1989). The goal of forming accurate impressions during social interactions:
Attenuating the impact of negative expectancies. Journal of Personality and Social
97

Niedenthal, P. M. (1990). Implicit perception of affective information. Journal of
Niedenthal, P. M. (1992). Affect and social perception: On the psychological validity of rosecolored glasses. In R. F. Bornstein & T. S. Pittman (Ed.), Perception without awareness:
Cognitive, clinical, and social perspectives (pp. 211-235). New York: Guilford Press.
Nisbett, R. E., & Wilson, T. D. (1977a). Telling more than we can know: Verba) reports on
mental processes. Psychological Review, 84, 231 - 259.
Nisbett, R. E., & Wilson, T. D. (1977b). The halo effect: Evidence for unconscious alteration of
judgments. Journal of Personality and Social Psychology, 35, 250-256.
Nuttin, J. M. (1985). Narcissism beyond Gestalt and awareness: The name letter effect.
European Journal of Social Psychology, 15, 353 361.
Nuttin, J. M. (1987). Affective consequences of mere ownership: The name letter effect in
twelve European languages. European Journal of Social Psychology, 17, 381-402.
Onodera, T., & Miura, M. (1990). Physical attractiveness and its halo effects on a partner:
Radiating beauty in Japan also? Japanese Psychological Research, 32, 148-153.
Orne, M. T. (1962). On the social psychology of the psychological experiment: With particular
reference to demand characteristics and their implications. American Psychologist, 17, 776783.
Osgood, C. E. (1957). A behavioristic analysis of perception and language as cognitive
phenomena. In Contemporary approaches to cognition (pp. 75-118). Cambridge, MA:
Harvard University Press.
Osgood, C. E., Suci, G. J., & Tannenbaum, P. H. (1957). The measurement of meaning.
Urbana, IL: University of Illinois Press.
Ostrom, T. M., & Sedikides, C. (1992). Out-group homogeneity effects in natural and minimal
groups. Psychological Bulletin, 112, 536 552.
Overson, C., & Mandler, G. (1987). Indirect word priming in connected semantic and
phonological contexts. Bulletin of the Psychonomic Society, 25, 229-232.
Paulhus, D. L., & Levitt, K. (1987). Desirable responding triggered by affect: Automatic
egotism? Journal of Personality and Social Psychology, 52, 245-259.
Paulhus, D. L., Martin, L. C., & Murphy, G. K. (1992). Some effects of arousal on sex
stereotyping. Personality and Social Psychology Bulletin, 18, 325-330.
Perdue, C. W., Dovidio, j. F., Gurtman, M. B., & Tyler, R. B. (1990). Us and them: Social
categorization and the process of intergroup bias. Journal of Personality and Social
Perdue, C. W., & Gurtman, M. B. (1990). Evidence for the automaticity of ageism. Journal of
98
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
Peters, D. P., & Ceci, S. J. (1982). Peer-review practices of psychological journals: The fate of
published articles, submitted again. Behavioral and Brain Sciences, 5, 187-255.
Petty, R. E., & Cacioppo, J. T. (1981). Attitudes and persuasion: Classic and contemporary
approaches. Dubuque, IA: Brown.
Petty, R. E., & Cacioppo, J. T. (1986). Communication and persuasion: Central and peripheral
routes to persuasion. New York: Springer- Verlag.
Petty, R. E., & Wegener, D. T. (1993). Flexible correction processes in social judgment:
Correcting for context-induced contrast. Journal of Experimental Social Psychology, 29,
137-165.
Porter, N., Geis, F. L., & Jennings-Walstedt, J. (1983). Are women invisible as leaders? Sex
Roles, 9, 1035-1049.
Posner, M. I. (1978). Chronometric explorations of mind. Hillsdale, NJ: Eribaum.
Pratkanis, A. R. (1988). The attitude heuristic and selective fact identifcation. British
Journal of Social Psychology, 27, 257-263.
Pratto, F., & Bargh, J. A. (1991). Stereotyping based on apparently individuating information:
Trait and global components of sex stereotypes under attention overload. Journal of
Prentice, D., & Miller, D. T. (1992). The psychology of ingroup attachment. Paper presented at
the conference on The Self and The Collective, Princeton University. Princeton, NJ.
Regan, D. T. (1971). Effects of a favor and liking on compliance. Journal of Experimental
Reingold, E. M., & Merikle, P. M. (1988). Using direct and indirect measures to study
perception without awareness. Perception and Psychophysics, 44, 563-575.
Richardson-KJavehn, A., & Bjork, R. A. (1988). Measures of memory. Annual Review of
Roediger, H. L. (1990). Implicit memory: Retention without awareness.
American
Roediger, H. L., & Blaxton, T. A. (1987). Effects of varying modality, surface features, and
retention interval on priming in word fragment completion. Memory and Cognition, 15,
379-388.
Roediger, H. L., Weldon, M. S., & Challis, B. H. (1989). Explaining dissociations between
measures of retention: A processing account. In H. L. Roediger & F. I. M. Craik (Eds.),
Varieties of memory and consciousness (pp. 3-42). Hillsdale, NJ: Eribaum.
Rogers, C. (1951). Client-centered therapy. Boston: Houghton-Mifflin.
Rosenbaum, M. (1986). The repulsion hypothesis: On the nondevelopment of relationships.
99
Rosenberg, M. J. (1965). When dissonance fails: On eliminating evalu ation apprehension

from attitude measurement. Journal of Personality and Social Psychology, 1, 28-42.
Rosenberg, M. J. (1969). The conditions and consequences of evalua tion apprehension. In R.
Rosenthal & R. L. Rosnow (Eds.), Artifact in behavioral research (pp. 279-349). New York:
Academic Press.
Rosenberg, M. (1979). Conceiving the self. New Ybrk: Basic Books.
Roskos-Ewoldsen, D. R., & Fazio, R. H. (1992). On the orienting value of attitudes: Attitude
accessibility as a determinant of an objects attraction of visual attention. Journal of
Sanitioso, R., Kunda, Z., & Fong, G. T. (1990). Motivated recruitment of autobiographical
memories. Journal of Personality and Social Psychology, 59, 229-241.
Sarnoff, I. (1960). Psychoanalytic theory and social attitudes. Public Opinion Quarterly, 24,
251-279.
Schacter, D. (1987). Implicit memory: History and current status. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 12, 432-444.
Scheier, M. F., & Carver, C. S. (1992). Effects of optimism on psychological and physical wellbeing: Theoretical overview and empirical update. Cognitive Therapy and Research, 16,
201-228.
Schooler, J. W., & Engstler-Schooler, T. Y. (1990). Verbal overshadowing of visual memories:
Some things are better left unsaid. Cognitive Psychology, 22, 36-71.
Schwarz, N. (1990). Feelings as information: Informational and motivational functions of
affective states. In E. T. Higgins & R. M. Sorren- tino (Eds.), Handbook of motivation and
cognition: Foundations of social behavior (Vol. 2, pp. 527-561). New Ybrk: Guilford Press.
Schwarz, N & Bless, H. (1992). Constructing reality and its alternatives: An
inclusion/exclusion model of assimilation and contrast effects in social judgment. In L. L.
Martin & A. Tesser (Eds.), The construction of social judgments (pp. 217-245). Hillsdale,
NJ: Eribaum.
Schwarz, N & Clore, G. L. (1983). Mood, misattribution, and judgments of well-being:
Informative and directive functions of affective states. Journal of Personality and Social
Schwarz, N., Strack, F., & Mai, H. P. (1991). Assimilation and contrast effects in part-whole
question sequences: A conversational logic analysis. Public Opinion Quarterly, 55, 3-23.
Secord, P. (1959). Stereotyping and favorableness in the perception of Negro faces. Journal of
Abnormal and Social Psychology, 59, 309- 315.
Seligman, M. E. P. (1991). Learned optimism. New York: Knopf.
Sherif, M., & Cantril, H. (1947). The psychology of ego-involvements. New York: Wiley.
Sherif, M., & Hovland, C. I. (1961). Social judgment: Assimilation and contrast effects in
communication and attitude change. New Haven, CT: Yale University Press.
100
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
Sherman, S. J., Mackie, D. M., & Driscoll, D. M. (1990). Priming and the differential uses of
dimensions in evaluation. Personality and Social Psychology Bulletin, 16, 405-418.
Sigall, H., & Ostrove, N. (1975). Beautiful but dangerous: Effects of offender attractiveness
and nature of the crime on juridic judgment. Journal of Personality and Social Psychology,
31, 410-414.
Smith, E. R. (1984). Model of social inference processes. Psychological Review, 91, 392-413.
Smith, E. R. (1990). Content and process specifcity in the effects of prior experiences. In T. K.
Srull & R. S. Wyer (Eds.), Advances in social cognition (Vol. 3, pp. 1-60). Hillsdale, NJ:
Eribaum.
Smith, E. R. (1994). Procedural knowledge and processing strategies in social cognition. In R.
S. Wyer & T. K. Srull (Eds.), Handbook of social cognition (2nd ed., Vol. l,pp. 99-151).
Hillsdale, NJ: Eribaum.
Smith, E. R., & Branscombe, N. R. (1988). Category accessibility as implicit memory. Journal
of Experimental Social Psychology, 24, 490-504.
Smith, E. R Stewart, T. L., & Buttram, R. T. (1992). Inferring traits from a behavior has longterm highly specifc effects. Journal of Personality and Social Psychology, 62, 753-759.
Smith, M. B., Bruner, J. S., & White, R. W. (1956). Opinions and personality. New Ybrk:
Wiley.
Snyder, M. (1981). On the self-perpetuating nature of social stereotypes. In D. Hamilton (Ed.),
Cognitive processes in stereotyping and intergroup behavior (pp. 183-212). Hillsdale, NJ:
Eribaum.
Spangler, W. D. (1992). Validity of questionnaire and TAT measures of need for achievement:
Two meta-analyses. Psychological Bulletin, 112, 140-154.
Spence, K. W. (1956). Behavior theory and conditioning. New Haven, CT: Yale University
Press.
Srull, T. K., & Wyer, R. S. (1979). The role of category accessibility in the interpretation of
information about persons: Some determinants and implications. Journal of Personality
Steele, C. M. (1988). The psychology of self-affirmation: Sustaining the integrity of the self. In
L. Berkowitz (Ed.), Advances in experimental social psychology ("Vol. 21, pp. 261-302). New
Steele, C. M., Spencer, S. J., & Lynch, M. (1992). Self-image resilience and dissonance: The
role of affirmational resources. Unpublished manuscript, Stanford University.
Sternberg, S. (1969). Memory scanning: Mental processes revealed by reaction-time
experiments. American Scientist, 57, 421-457.
Strack, F. (1992). The different routes to social judgments: Experiential versus informational
strategies. In L. L. Martin & A. Tesser (Eds.), The construction of social judgments (pp.
101
249-275). Hillsdale, NJ: Eribaum.

Strack, F., Schwarz, N., Bless, H., Kiibler, A., & Wanke, M. (1993). Awareness of the influence
as a determinant of assimilation versus contrast. European Journal of Social Psychology,
23, 53-62.
Suls, J. (1991). Similarity and self-esteem in downward comparison. In J. Suls & T. A. Wills
(Eds.), Social comparison: Contemporary theory and research (pp. 51-78). Hillsdale, NJ:
Eribaum.
Suls, J., & Wills, T. A. (1991). Social comparison: Contemporary theory and research.
Hillsdale, NJ: Eribaum.
Swim, J., Borgida, E., Maruyama, G., & Myers, D. G. (1989). Joan McKay versus John
McKay: Do gender stereotypes bias evaluation? Psychological Bulletin, 105, 409-429.
Tajfel, H. (1970). Experiments in intergroup discrimination. Scientific American, 223, 96-102.
Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In S.
Worchel & W. G. Austin (Eds.), Psychology of intergroup relations (pp. 7-24). Chicago:
Nelson-Hall.
Taylor, D. M., Wright, S. C., Moghaddam, F. M., & Lalonde, R. N.
(1990) . The personal/group discrimination discrepancy: Perceiving my group, but not
myself, to be the target of discrimination. Personality and Social Psychology Bulletin, 16,
254-262.
Taylor, S. E., & Brown, J. D. (1988). Illusion and well-being: A social psychological perspective
on mental health. Psychological Bulletin, 103, 193-210.
Tesser, A. (1988). Toward a self-evaluation maintenance model of social behavior. In L.
Berkowitz (Ed.), Advances in experimental social psychology (Vol. 21, pp. 181 -227). New
Tesser, A., & Campbell, J. (1983). Self-defnition and self-evaluation maintenance. In J. Suls
& A. G. Greenwald (Eds.), Psychological perspectives on the self (Vol. 2, pp. 1 -32). Hillsdale,
NJ: Eribaum.
Thorndike, E. L. (1920). A constant error in psychological ratings. Journal of Applied
Thurstone, L. L. (1931). The measurement of attitudes. Journal of Abnormal and Social
Thurstone, L. L., & Chave, E. J. (1929). The measurement of attitude. Chicago: University of
Chicago Press.
Tolman, E. C. (1959). Principles of purposive behavior. In S. Koch (Ed.),
Psychology: A study of a science (Vol. 2, pp. 92-157). New York: McGraw-Hill.
102
6.
COGNITIVE
LEARNING AND
COGNITIVE
RESPONSE
Tsui, A. S., & Barry, B. (1986). Interpersonal affect and rating errors. Academy of
Management Journal, 29, 586-599.
Tulving, E., & Schacter, D. (1990). Priming and human memory systems. Science, 267, 301306.
Uleman, J. S. (1987). Consciousness and control: The case of spontaneous trait inferences.
Personality and Social Psychology Bulletin, 13, 337-354.
Uleman, J. S., & Bargh, J. A. (Eds.). (1989). Unintended thought. New York: Guilford Press.
Vallone, R. P., Ross, L., & Lepper, M. R. (1985). The hostile media phenomenon: Biased
perception and perceptions of media bias in coverage of the Beirut Massacre. Journal of
Wallston, B., & OLeary, V. (1981). Sex makes a difference: Differential perceptions of men
and women. Review of Personality and Social Psychology 2, 9-41.
Webb, E. J., Campbell, D. T., Schwartz, R. D., Sechrest, L. B., & Grove, J. B. (1981).
Nonreactive measures in the social sciences. Boston: Houghton-Mifflin.
Wegner, D. M., & Vallacher, R. R. (1977). Implicit psychology. New York: Oxford.
Wegner, D. M., & Vallacher, R. R. (1981). Common-sense psychology. In J. P. Forgas (Ed.),
Social cognition: Perspectives on everyday understanding ( p p . 225-246). London: Academic
Press.
Wicker, A. W. (1969). Attitudes vs. actions: The relationship of verbal and overt behavioral
responses to attitude objects. Journal of Social Issues, 25, 41-78.
Wilder, D. A., & Shapiro, P. N. (1984). Role of out-group cues in determining social identity.
Journal of Social and Personality Psychology, 47, 342-348.
Wills, T. A. (1981). Downward comparison principles in social psychology. Psychological
Bulletin, 90, 245-271.
Wilson, P. R. (1968). Perceptual distortion of height as a function of ascribed academic status.
Journal of Social Psychology, 74, 97-102.
Wilson, T. D., Dunn, D. S., Kraft, D., & Lisle, D. J. (1989). Introspec tion, attitude change,
attitude-behavior consistency: The disruptive effects of explaining why we feel the way we
do. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 22,287-343). New
Wilson, T. D., & Schooler, J. W. (1991). Thinking too much: Introspec tion can reduce the
quality of preferences and decisions. Journal of Personality and Social Psychology, 60, 181
-192.
Winter, L., & Uleman, J. S. (1984). When are social judgments made? Evidence for the
spontaneousness of trait inferences. Journal of Personality and Social Psychology, 47, 237252.
103
Wood, J. V., Taylor, K. L., Michela, J. L & Gaus, V. (1993). Measuring self-esteem indirectly:
Personality correlates and responses to threats to self-esteem. Unpublished manuscript,
University of Waterloo, Ontario, Canada.
Word, C. O., Zanna, M. P., & Cooper, J. (1974). The nonverbal mediation of self-fulflling
prophecies in interracial interaction. Journal of Experimental Social Psychology, 10, 109120.
Wyatt, D. F., & Campbell, D. T. (1951). On the liability of stereotype or hypothesis. Journal of
Abnormal and Social Psychology, 46,496-500.
Wylie, R. C. (1974). The self-concept (Vol. 1). Lincoln: University of Nebraska Press.
Wylie, R. C. (1979). The self-concept (Vol. 2). Lincoln: University of Nebraska Press.
Zajonc, R. B. (1968). Attitudinal effects of mere exposure. Journal of Personality and Social
Psychology, 9 (No. 2, Pt. 2).
Zanna, M. P., & Fazio, R. H. (1982). The attitude-behavior relation: Moving toward a third
generation of research. In M. P. Zanna, E. T. Higgins, & C. P. Herman (Eds.), Consistency
in social behavior: The Ontario Symposium (Vol. 2, pp. 283-301). Hillsdale, NJ: Eribaum.
Zanna, M. P., & Rempel, J. K. (1988). Attitudes: A new look at an old concept. In D. Bar-Tal &
A. W. Kruglanski (Eds.), The social psychology of knowledge ( p p . 315-334). Cambridge,
England: Cambridge University Press.
Received July 20, 1993 Revision received December 7,

1993
Accepted April 12, 1994
MEMBERS OF UNDERREPRESENTED GROUPS: REVIEWERS FOR

JOURNAL MANUSCRIPTS WANTED
If you are interested in reviewing manuscripts for APA journals, the APA
Publications and Communications Board would like to invite your participation.
Manuscript reviewers are vital to the publication process. As a reviewer, you would
gain valuable experience in publishing. The P&C Board is particularly interested
in encouraging members of underrepresented groups to participate more in this
process.
If you are interested in reviewing manuscripts, please write to Leslie Cameron at
the address below. Please note the following important points:
To be selected as a reviewer, you must have published articles in peerreviewed journals. The experience of publishing provides a reviewer with the
basis for preparing a thorough, objective review.
104
6.
COGNITIVE
LEARNING AND
.Journal of Personality and Social Psychology 1979,
COGNITIVE
RESPONSE
Vol. 37, No. 12, 2181-2199
To select the appropriate reviewers for each manuscript, the editor needs
detailed information. Please include with your letter your vita. In your letter,
please identify which APA journal you are interested in and describe your
area of expertise. Be as specifc as possible. For example, "social psychology" is
not sufficientyou would need to specify "social cognition" or "attitude
change" as well.
Reviewing a manuscript takes time. If you are selected to review a

manuscript, be prepared to invest the necessary time to evaluate the
manuscript thoroughly.
Write to Leslie Cameron, Journals Office, American Psychological Association, 750

First Street, NE, Washington, DC 20002-4242
Attitudes and Cognitive Response:

An Electrophysiological Approach
John T. Cacioppo
University of Iowa
Columbia
Richard E. Petty
University of Missouri
Two experiments employed electrophysiological procedures for assessing the

covert information-processing activity of message recipients. In Experiment
1.
24 subjects expected to hear discrepant communications and were requested to
collect their thoughts following each forewarning. As discrepancy increased,
anticipatory counterargumentation increased, whereas production of favorable
thoughts and agreement decreased. In addition, following forewarnings oral
muscle, cardiac, and respiratory activity increased, whereas nonoral muscle activity remained constant and quiescent. In Experiment 2, 60 subjects anticipated
and heard a proattitudinal, a counterattitudinal, or a neutral communication. They
evaluated more positively and generated more favorable thoughts and fewer
counterarguments to the proattitudinal than to the counterattitudinal advocacy,
but rated similarly the neutral and proattitudinal advocacies. As in Experiment 1,
incipient oral muscle activity increased following the forewarning of an involving
counterattitudinal advocacy; it also increased for all conditions during the message.
Patterns of subtle facial muscle changes reflected the affective nature of the
cognitive responding before and during the message. These results provide
evidence that electrophysiological assessments offer objective, concurrent, and
independent measures of cognitive response in persuasion, and support the notion
of recipients as active information processors when topic involvement is high.
process
the
information
Recent work in attitude change has persons
emphasized the manner in which contained in persuasive messages.
Copyright 1979 by the American Psychological Association, Inc. 0022-3514/7^/3712-2181 $00.75
2181
Investigators have studied how such

variables as source credibility (Cook,
1969; Gillig & Greenwald, 1974),
distraction (Petty, Wells, & Brock,
1976), message repetition (Cacioppo &
Petty, 1979), message comprehensibility
(Eagly, 1974), forewarning of perThe
authors
thank
Anthony
G.
Greenwald, Curt A. Sandman, and the action
editor for comments concerning this
research; Charles W. Snyder for technical
advice and assistance; Keith Richenbacher,
Debra Horowski, and June Chen for
assistance in Experiment 1; and Leo
Quintanar, Kevin Smith, and J. P. Holbrook
for assistance in Experiment 2. This
material is based on work supported in part
by
>IationalSdenc^^ouijj^jjjji^ailtiiiSaZilSffL
Requests for reprints should be sent to John
T. Cacioppo, Department of Psychology,
University of Iowa, Iowa City, Iowa 52242, or
to Richard E. Petty, Department of
Psychology,
University
of
Missouri,
Columbia, Missouri 65211.
suasive intent (Petty & Cacioppo,

1979a) and of topic and position (Petty
& Cacioppo,
105
, number of arguments employed (Calder, Insko, & Yandell, 1974), issue

involvement (Petty & Cacioppo, 1979b),
heart rate (Cacioppo, 1979), group
discussion (Burn- stein & Vinokur,
1977), and so forth affect the profle of
cognitions (e.g., counterarguments,
favorable thoughts, neutral thoughts),
and attitude change. Theoretical
interest in the influence on persuasion
of a persons idiosyncratic cognitive
responses to an advocacy is certainly
not new (cf. Hovland, Lumsdaine, &
Sheffield, 1949), but the present level of
research activity in the area marks a
shift in emphasis toward this approach
(cf. Petty, Ostrom, & Brock, in press).
The reinvigorated interest in these
covert thought processes stems in part
from the apparent inability of classical
learning theories to provide
parsimonious accounts of observed attitude changes (Greenwald, 1968; Petty,
1977)
,
1977)
Sandman, & Walker, 1978; Edwards & Alsip,
1969)
; and (f) observing the cognitive and
attitudinal effects of exogenously manipulated
heart rate (Cacioppo, 1979). The cardiac response generally covaries with the cognitive
requirements or difficulty of the task, particularly when the task requirements are substantial. Although considerable controversy still
exists in psychophysiology regarding the
neurophysiological and/or biological processes
controlling the heart rate responses observed in
the aforementioned studies, the empirical link
between the cardiac response and cognitive
activity seems well established. Whether the
cardiac acceleration observed during the
performance of cognitive tasks is initiated by a
metabolic control center (cf. Obrist et al.,
1970)
or by a modulating negative feedback
system (cf. Lacey, 1967) is unimportant here.
Central to the present study is the fnding that
the cardiac response reflects considerable
changes in cognitive activity, even when these
changes are spontaneous and self-induced
(Schwartz, 1971; Schwartz & Higgins, 1971).
Methodological Considerations
Several methodological safeguards were instituted here to assure that the electrophysiological measures we observed reflected covert
information processing. First, a measure of
general (nonlinguistic) somatic activity was
obtained in addition to the measures of. oral
EMG and cardiac activity. To the extent that the
activation is specifc to the speech muscle fbers
and heart rate when processing an external
event, we are more confdent that we have
tapped covert processing rather than irrelevant
(i.e., unreliable) movements, postural shifts, and
so forth. Second, the level of heart rate and EMG
activity observed while anticipating and
processing the stimulus was compared with
prestimulus measures to assess whether or not
electrophysiological response actually occurred
(McGuigan, 1970). And third, Miller and Baron

(1973) have suggested that observations of oral
EMG (and cardiac) activity may not differentiate
the cognitive elaboration of the advocacy from
the covert rehearsal of the arguments
constituting the persuasive message. One of our
aims was to determine the existence of cognitiveresponse processes in persuasion; hence, we
obtained recordings of the electrophysiological
measures while subjects were anticipating
highly
involving
and
counterattitudinal
advocacies as well as while subjects were
processing them. A good deal of evidence
documenting that antidpating a discrepant and
involving
message
evokes
issue-relevant
cognitive responding has accrued using the
thought-listing procedure (Cacioppo, Petty, &
Snyder, 1979; Petty & Cacioppo, 1977; Cialdini &
Petty, in press).
Experiment 1
The aim of Experiment 1 was primarily
methodological. We sought to test the utility of
an electrophysiological assessment of cognitive
response. To do so, we forewarned individuals
that they would hear a (counterattitudinal)
message and we asked them to collect their
thoughts about the issue. Immediately
preceding our forewarning, we collected basal
measures of physiological response, and we
continued to record these meaures while the
individuals were anticipating the discrepant
message (i.e., when they presumably were
following instructions and were generating
cognitive responses concerning the advocacy). We
expected oral EMG and cardiac activity, but not
general somatic activity, to be heightened during
the collect thoughts interval, compared to basal
levels.
Confrmation of these expectations would
illustrate the applicability and utility of electrophysiological techniques for studying cognitive response in persuasion. Obtaining these
results would not, of course, provide evidence
that cognitive responses in persuasion are
generated naturally, since we asked the individuals to collect their thoughts following the
forewarning. A second experiment addressed this
latter issue.
ELECTROPHYSIOLOGY AND PERSUASION
2185
Table 1.
Procedure
distinguish the types of affective response

evoked, though differentiation between affectand non-affect-laden thought sequences using
heart rate has been observed (Schwartz,
1971) . The discrepancy of the impending advocacy was manipulated in Experiment 1 to
obtain gradations of affect-laden processing to
assess its physiological effects. Additionally, the
thought-listing procedure was used to explore
the relationships among the cognitive (e.g.,
favorable
thoughts,
counterarguments,
neutral/irrelevant thoughts) and electrophysiological (e.g., oral EMG, heart rate) responses.
Method
Subjects and Design
When subjects arrived at the laboratory, they were

placed in a sound-attenuated room and were seated in
a comfortable chair. Electrodes were attached for
measuring oral (orbicularis orislips; digastricus
chin; platysmathroat) EMG activity, nonspeech
(trapeziusback) EMG activity, heart rate, breathing
rate, and cephalic pulse amplitude. A 5-minute
adaptation period preceded the experimental trials.
Subjects were told that in about 40 minutes they
would hear several different messages having direct
consequences for undergraduates, and that before the
presentation of the messages we should like to obtain
their comments on and evaluations cf the position (i.e.,
advocacy) to be advanced in each message. The subjects
were asked to sit quietly and collect their thoughts for
the minute following each announcement; then, at the
experimenters signal, subjects were asked to list
everything about which they had been thinking
(subjects were given 3 minutes to do so), to rate their
agreement with the upcoming advocacy, and to
complete several ancillary measures (i.e., felt effort,
involvement, distraction, and responsibility). The
nature of these forms is described in detail in Petty and
Cacioppo (1977). The forewarning, collect thoughts, and
thought-listing intervals were repeated six times to
cover six different topics, each separated by a variable
intertrial interval (ITI) ranging from 90 to 120 seconds.
Each ITI was initiated when the experimenter
requested the subjects to please sit quietly for the next
minute or so. The fnal 60 seconds of each ITI served
as the baseline measure for the subsequent 60-second
collect thoughts interval.
Three replications of eight male undergraduates

were conducted. Subjects were tested individually, with
Chin muscle activity. Two Grass ESS cup electrodes
assignment to replication determined randomly. A 3 X
3 X 2 X 2 mixed design was employed in which the flled with Grass EKG Sol were placed on the midline of
three replications served as a between- subjects factor, the chin; the frst was placed 1.8 cm above the point of
and levels of communication discrepancy (low, the chin and the second was
moderate, and high), two different topics within each
level of discrepancy, and interval during which
electrophysiological
measures
were
recorded
(prewarning baseline and postwarning collect
thoughts interval) served as within-subjects factors.
A interval
second aim
Experiment
1 was
toanalyses
explore of
the effects of affect- and nonaffect-laden cognitive
(The
factorofwas
relevant only
to the
responding
on electrophysiological
response patterns. Neither oral EMG (cf. Garrity, 1977;
the electrophysiological
measures.)
McGuigan, 1978) nor cardiac activity (Cacioppo & Sandman, 1978; Harris, Katkin, Lick, &
Habberfeld, 1976) appear toTable 1
Experimental
Materials Stimuli
A separate audiotape was prepared for each of the

three replications. Each tape contained the experimental instructions and six announcements regarding the source of, topic of, and position to be
advanced in an upcoming message. (In fact, however,
the messages were never presented.) Each level of
discrepancy for each topic appeared in one replication.
The tapes (i.e., replications) differed in the order of the
topics, which was determined randomly for each
replication. The experimental stimuli are displayed in
placed 1.8 cm below the point of the chin. Following

Cacioppo et al. (1978), chin muscle activity was
calculated for the 60-sec intervals of interest using the
following formula:
Activity index = ----------------
iH
where st is the scale value of a particular amplitude of

EMG activity (larger amplitude EMG activity was
assigned larger scale values; seals values ranged from
0, for deflections 2 mm or less, to S, for deflections
exceeding 2 cm); U is the total horizontal length (i.e.,
time) of a particular scale value of EMG activity
measured in millimeters; and n is the number of
distinct instances of a particular amplitude of EMG
activity. The data were quantifed in this manner for
the 60-sec baseline and collect thoughts intervals.
Lip muscle activity. Two Grass ESS cup electrodes
flled with Grass EKG Sol were placed 1 mm below the
bottom lip; each was placed .5 cm in from the ends of
the mouth. Lip muscle activity was calculated using
the activity index described above.
Throat muscle activity. Two Grass ESS cup electrodes flled with Grass EKG Sol were placed off
midline of throat; the frst was placed approximately
1.0
cm to the right of and level with the
midpoint of the throat, and the second was placed
approximately 1.0 cm to the left and 1.0 cm above the
midpoint. Throat muscle activity was calculated using
the activity index described above.
Back muscle activity. Subjects were asked to place
their fngertips on their collarbone while the electrodes
were secured. Two Grass ESS cup electrodes flled with
Grass EKG Sol were placed over the trapezius muscle
group. The frst was placed
4.0
cm outward from the midline of the line
passing between the frst thoracic and seventh cervical
vertebrae, and the second was placed halfway between
the spine and the head of humerus (near the point of
the shoulder). Back muscle activity was calculated
using the activity index described above.
Heart rate. Grass ESS cup electrodes flled with EKG
Sol were placed over the lower left rib cage and the
right collar bone. The signal was amplifed by a Grass
wide-band AC preamplifer. Heart rate was calculated
by counting the number of beats that occurred in the 1minute intervals of interest.
Breathing rate. The respirometer was a sliding piston
(consisting of a photocell and a small light) mounted on
an elastic band and placed around the subjects chest
(Shmavonian, Miller, & Cohen, 1968). Breathing rate
was calculated by counting the number of cycles (to the
nearest half cycle) occurring in the intervals of interest.
Cephalic pulse amplitude.19 A photoplethysmograph

was placed over the supraorbital notch (above the
eyebrow), providing a relative measure of blood volume
in the supraorbital artery. The photoplethysmograph
was comprised of three light-emitting diodes (LED),
radiating 100 /tW of light output at 660 nanometers.
The LEDs were spaced at 120 on a radius of 6.3 mm
around a high-speed photoconductor, and resistance
changes were recorded with a Grass oscillograph.
Cephalic pulse amplitude was calculated for the
intervals of interest using the following formula:
Amplitude index =
pulse amplitude in mm
X total mV/cm calibration
standard sensitivity in mV/cm
where total mV/cm was the sensitivity of the preamp
settings for a given subject, and the standard sensitivity was 1 mV/cm.
Data reduction. Physiological processes were monitored during the 1 minute preceding and following
each of the six forewarnings. Persons scoring the
electrophysiological data were unaware of the experimental hypotheses and of the treatments 'with" which
the data were associated.
The Grass Model 7 polygraph used in the experiment
was equipped with three preamplifers capable of
measuring electromyographic activity. Since four EMG
measures were of interest ip the experiment, a random
procedure was used to determine which three of the
four EMG measures would be recorded for each subject
within a replication. However, all electrode placements
were prepared on each subject, and the subject was
unaware of the dummy electrode placements.
Difference scores relative to the prewarning levels were
computed for each electrophysiological measure.- '
19 A photoplethysmograph was placed over the
supraorbital artery just above the eye. The amount of
light reflected back onto the photoplethysmograph is
inversely proportional to the amount of blood between
the photoplethysmograph and the supraorbital notch
(over which it was placed). Brain blood perfusion was
assumed to be indirectly related to the relative volume
changes of the pulse wave that traveled to the brain via
the supraorbital artery. Pulse rate was not considered
to be important because of the venous return
mechanism, which drains the blood from a given body
area at approximately the same rate as the blood is
delivered to that area (Wallace & Wallace, 1968). The
validity of this measure, however, has not yet been
frmly established. The plethysmograph used in these
studies was developed by Robert Isenhart. More
information about the plethysmograph is provided in
Sandman, McCanne, Kaiser, and Diamond (1977).

The subjects classifed their cognitive response in a
manner described by Petty and Cacioppo (1977); After
listing their thoughts, subjects were instructed to place
a plus (+) next to those thoughts that were in favor of
the advocacy, a minus () next to those thoughts
opposed to the advocacy, and a zero (0) next to those
thoughts that were either neutral toward or irrelevant
to the advocacy. Frequency counts served as measures
of cognitive response.
2185
These tests revealed a violation of the

hotdv assumption only for the measure of
neutral thoughts, X 2 (2) = 8.11. Neutral thought
production was unaffected by discrepancy even
with a positively biased F ratio.
127).
The means for all cognitive response, agreement, and ancillary measures are summarized
in Table 2. The analyses indicated that increasing the discrepancy between the subjects
Measurements of the seven electrophysiological dependent measures (chin, lip, throat,

and back EMG activity, heart rate, breathing
rate, and cephalic pulse amplitude) Table
were 2obtained the minute preceding and following
each
Mean Responses
to Thought Listing and
forewarning. Parametric analyses Questionnaire
of the Measures for Low, Moderate, and High
Levels nonof Communication Discrepancy: Experiment 1
questionnaire data were conducted while
parametric analyses of the electrophysiological
data were conducted, since the latter data were
not distributed normally (cf. Schwartz, Fair,
Salt, Mandel, & Klerman, 1976a, 1976b).
Cognitive Response, Agreement,
and Ancillary Measures
We expected topic-relevant cognitive responding and agreement to be affected in a
particular manner by discrepancy, but we had no
particular expectations regarding neutral
thought production and the ancillary measures.
Hence, we set the experimentwise error rate at .
10 and distributed this protection unequally
across the tests (i.e., .05
for counterargumentation, favorable thought production,
and agreement; .05 for the remaining fve
measures). We used Bonferroni-adjusted critical
values for all tests, conducted two- tailed tests
throughout, allocated 99% of the alpha
associated with the contrasts for the frst set of
variables into the tail corresponding to the
predicted direction of the effect (saving 1% as
acknowledgment that opposite rather than
predicted results sometimes obtain), and
allocated 50% of the alpha associated with the
tests for the second set of variables into each
tail.20 Further, the Huynh and Feldt (1970) test
for the homogeneity of treatment-difference
variance (hotdv) was conducted for each
dependent measure (cf. Harris, 1975, pp. 12520 Required for statistical signifcance by these adjustments were ps < .017 for counterarguing, favorable
thoughts, and agreement, and ps < .01 for the
remaining measures.
Note. Entries for cognitive response measures indicate

the mean frequency obtained in thought listings.
Entries for questionnaire items are mean response to
11-point scale items in which higher numbers indicated
more agreement, effort, involvement, distraction, and
responsibility. Twenty-four subjects received two
forewarnings at each level of discrepancy.
initial positions and the advocated position led

to more anticipatory counterargumentation,
F(2, 42) ~ 8.80, p < .01; production of fewer
anticipatory tavorable thoughts, F( 2, 42) = 3.70,
p < .017; and greater felt effort in preparing
cognitively for the message, F(2, 42) = 5.33, p < .
01. One additional test approached signifcance:
Agreement tended to decrease as discrepancy
increased, F ( 2 , 42) = 3.57, p < .019. No other
effect or interaction was statistically signifcant.
This pattern of results is similar to that
obtained in prior research on cognitive response
and agreement as a function of communication
discrepancy (Brock, 1967; Cacioppo, 1977).
Electrophysiological Measures
We hypothesized that the activity of the oral
muscles and heart rate would increase
Median change in EMG activity (mm)
0.10 0.00 0.10 0.20
Back
Figure 1. Median change from baseline for lip, chin,

throat, and back electromyographic (EMG) activity
following a forewarning about an impending counterattitudinal advocacy.
during the collect thoughts interval (relative to

basal levels), since we expected cognitive
response processes to be reflected electrophysiologically. Strong support for the hypothesis
was obtained (see Figure 1). Oral EMG activity
was elevated signifcantly after forewarnings of
involving counterattitudinal communications (by
the Wilcoxon Test: lip, p < .001; chin, p < .001;
throat, p < .10). Also evident in Figure 1, general
somatic activity, as measured by back EMG
activity, was not altered by the anticipation of
the counterattitudinal message (p > .25). The
Wilcoxon Test for changes from baseline revealed
that heart rate ( M d n = 2.00 bpm, p < .02) and
breathing rate { M d n = 1.0 cycle/min, p < .01)
increased following the forewarning as well,
whereas' cephalic pulse amplitude was left
unchanged ( M d n = 0.00). No other comparisons
were signifcant statistically.9
Correlational Analyses
The affective intensity of the cognitive re-
sponses, which was varied by increasing communication discrepancy, did not affect the
electrophysiological activity monitored in this
study. Canonical correlations between the
cognitive responses (i.e., counterarguments,
favorable thoughts, neutral/irrelevant thoughts)
and the relevant electrophysiological measures
(i.e., lip, chin, throat, and cardiac activity) were
computed within each level of discrepancy to
determine whether the predicted relationship
between cognitive and electrophysiological
activity existed and to explore whether any
association existed between the affective nature
of the covert processing and electrophysiological
activity. The correlations were respectable (r = .
47 for low discrepancy conditions; r .44 for
moderate discrepancy conditions; and r .64 for
high discrepancy conditions). When a canonical
correlation was calculated, collapsing across the
levels of discrepancy, a coefficient of .42 was
obtained.
Furthermore, the electrophysiological specificity obtained in this research is in striking
contrast to the massive and diffuse arousal
associated with the fght-or-flight reaction of
extreme emotional states (Cannon, 1927) and
misattribution phenomena (Schachter, 1964; see
also, Rhodewalt & Comer, 1979).
The calculation of within-cells correlations
among cognitive responses and agreement revealed that anticipatory counterargumentation
correlated negatively with agreement (r = .67,
p < .01), favorable thoughts (r =
.67,< .01), and neutral thoughts (r =
.45, p < .01). Favorable thoughts correlated

positively with agreement (r = .72, p < . 01)

/ Within-cell correlations further revealed that
unfavorable thoughts were positively correlated
with ratings of effort (r = .42, p < .05),
involvement (r = .43, p < .05), and responsibility
(r = .29, p < .05), and correlated negatively with
ratings of distraction (r = .29, p< .05). None of
the electrophysiological measures was correlated
significantly with any ancillary measure.
In sum, electrophysiological and thoughtlisting measures indicated that topic-relevant

cognitive responding accompanied the anticipation of an involving counterattitudinal
communication. Inspection of Figure 1 supports
the notion that cognitive response processes in
persuasion can be measured concurrently and
without the subjects doing anything overt in
particular (e.g., listing thoughts). Of course, no
evidence was provided in Experiment 1
concerning the natural existence or elicitation of
cognitive response in persuasion, since the
subjects were aware that they were to list their
thoughts; in fact, they had been instructed to
collect
their
thoughts
following
each
forewarning. We were also unable to distinguish
electrophysiologically the affective natures of
the recipients covert preparations for the
advocacy.
We conducted a second experiment to address
these issues.
Experiment 2
The frst issue, that of providing electrophysiological evidence concerning the cognitive
responses elicited naturally in persuasion, was
addressed easily in the design of Experiment 2.
Rather than asking the subjects to collect or list
thoughts,
we
simply
monitored
the
electrophysiological activity displayed during the
anticipation and presentation of a single
advocacy. Subjects had no notion that they would
subsequently be asked to list their thoughts.
The second issue, developing an electrophysiological measure capable of differentiating
the affective nature of cognitive responding
(should it exist in persuasion), proved to be more
2189
difficult. Previous studies of attitudes and bodily

reactions have employed three basic research
strategies: Procedures have been employed (a) to
tap the physiological processes indicating covert
information processing (e.g., Experiment 1;
Cacioppo, 1979); (b) to measure an evaluative
reaction by monitoring a classically conditioned
physiological response (e.g., Tognacci & Cook,
1975); and (c) to assess the naturally occurring
physiological indicators of affective states (e.g.,
Cooper, 1959; Hess, 1965). Studies of attitudes
applying the third research strategy have often
used measures of pupillary response or
electrodermal activity (galvanic skin responses).
These studies have provided some evidence that
attitudes, if extreme, may be measurable. Even
in these instances, however, the polarity of the
attitude (i.e., positive or negative) has not been
distinguishable (Cacioppo & Sandman, in press;
Mueller, 1970).
Recent work on the neuromuscular substrates
of emotion and depression offered us a potential
solution. Darwin (1965/1872) frst documented
the specifcity and reliability of facial muscle
patterning in the expression of emotions (cf.
Cacioppo & Petty, in press-a). More recently,
Schwartz and his colleagues (Schwartz, 1975;
Schwartz et al., 1976a, 1976b) have found that
generating imagery or attempting to experience
through fantasy the emotional states of
happiness, sadness, and anger leads to
distinctive patterns of EMG activation of the
face. Moreover, these patterns go unnoticed by
subjects as well as observers (see also Izard,
1971). We reasoned that by monitoring these
facial (i.e., corrugator, zygomatic, depressor
anguli- oris, and mentalis) muscles during the
anticipation and presentation of advocacies, we
would be able to distinguish favorable from
unfavorable (i.e., counterargument) cognitive
responses emitted by a recipient. In Experiment
1 and in previous research, we have found oral
EMG to distinguish the extent rather than the
affectivity of covert processing (e.g., Cacioppo &
Petty, in press-b, in press- c). Hence, we
considered the measure of mentalis EMG
activity, which taps the electrical activity of the
muscle fbers between and including the lower
lip and chin, as a measure of the extent rather
than emotionality
McGuigan,
1978)
of
processing
(see
also
. Again, heart rate was recorded.

Method
Subjects and Design

Sixty male undergraduates were led to believe they
were evaluating the sound quality of taped radio
editorials that had been produced by the students in a
sound-engineering course. Electrodes were placed on
each subjects body, and subjects were tested
individually in a darkened, sound-attenuated room to
reduce external distractions from the task. Forty-eight
subjects were forewarned about and heard either a
proattitudinal or a counterattitudinal advocacy on one
of two topics (alcoholic beverages or visitation hours).
Twelve additional subjects were forewarned only that
they would hear a taped communication, and they
heard a message about an obscure news event. Subjects
in this group served in an external control (neutral
advocacy) condition. The assignment of subjects to
condition again was determined randomly.21
Materials
The topics of alcoholic beverages and visitation hours
were selected because initial pilot testing revealed
existing university regulations regarding them to be
highly involving and counterattitudinal. Forewarnings
and messages were constructed that advocated the
adoption of either stricter (counterattitudinal) or more
lenient (proattitudinal) regulations regarding these
issues. The neutral message concerned a small
archeological fnd and was obtained from a past issue of
a national news magazine.
Procedure
When subjects arrived at the laboratory, they were
told that their task was to evaluate the sound quality of
a taped radio editorial, that electrodes would be
attached, and that during the study we would be
recording the involuntary bodily responses that
accompany listening to a communication. Subjects
were instructed to refrain from unnecessary
movements, to breathe normally, and to keep their eyes
closed throughout the study. After adapting to the
laboratory, subjects again heard these instructions and
were told that the study would begin shortly. At this
point, a computer-controlled procedurewhich involved
21 An additional factor included in the design was
whether subjects were informed that the communication had implications locally or not. Manipulation
checks revealed that our manipulation of this factor
failed here, so this factor is not discussed further.
(a) a 60-sec prewarning (baseline) interval, (b) a 15-sec

forewarning, (c) a 60-sec postwarning-premessage
interval, and (d) a 120-sec messagewas initiated.
After listening to the tape, the subjects read the
following:
Because your own opinion about the position
advocated on the tape may influence the way you rate
the quality of the tape, we would like to obtain a
measure of how you feel about the views proposed by
the speaker on each scale below.
The subjects responded to four 'J-point semantic
differentials; their responses were summed to obtain a
measure of their attitude toward the advocacy. In the
same manner as in Experiment 1, subjects were
instructed to list everything about which they had
thought during the message (subjects were given 3
minutes). Afterwards, subjects rated their listed
thoughts as favorable (+), unfavorable (), or
neutral/irrelevant (0) toward the message. Subjects
then rated on 11-point Likert-type scales their felt
involvement, effort, and distraction, the personal
relevance of the message, the sound quality of the tape,
and the speakers rate of delivery and enthusiasm.
Heart rate. Grass ESS cup electrodes flled with
Grass EC3 paste were placed over the lower left rib and
the right collar bone. The signal was amplifed by a
Narco Biosystem Physiograph AC preamplifer. The
output was displayed on a Narco Biosystem
Physiograph 6 and was transmitted on-line to a PDP-8I
laboratory computer for analysis.
Facial muscle activity. Grass ESS cup electrodes
flled with Grass EC3 paste were placed adjacent to
each other in pairs with interelectrode resistance reduced to less than 10,000 ohms. The four muscles over
which the pairs of electrodes were placed were the
corrugator (just above the eyebrow), zygomatic (upper
cheek), depressor anguli oris (lower cheek), and
mentalis (between lip and chin) on the left side of the
face (cf. Schwartz et al., 1976a). Since surface
electrodes were used, recordings of EMG activity were
obtained from these and surrounding muscle groups.
Each EMG measure was amplifed by a Narco
Biosystem Physiograph AC preamplifer, individually
rectifed and summed by an EMG integrator with a
time constant of .2 sec. The average integrated EMG
was displayed on the physiograph with a full-scale pen
deflection of 40 mm (1 mm = 7.5 #iV). The integrated
EMG also was transmitted on-line to the PDP-8I
laboratory computer, sampled 10 times per second, and
recorded. The computer was programmed to eliminate
from its recordings any obvious movement artifact or
overt (e.g., visually detectable) facial expression.
Data reduction. The cognitive data were scored in
the same manner as in Experiment 1. Physiological
responses were monitored during the minute preceding
the forewarning through the completion of the 120-sec

message. The data for each measure were averaged for
each subject and each interval. Difference scores
relative to the prewarning (basal) levels were
calculated for each measure and interval, and
nonparametric analyses were employed, since the data
were not distributed normally. The presentation here of
the results for facial EMG activity is similar in format
to the presentation by Schwartz et al. (1976a, 1976b) to
facilitate comparisons.

Our purposes in this study were to determine
if cognitive responses were generated naturally
in persuasion settings and, if so, whether or not
the affective nature of these responses could be
assessed electrophysiologically. A multivariate
analysis of variance of the 11 questionnaire
measures was conducted frst to determine the
general effects of the experimental factors. As
expected, the effect of position was highly
signifcant, based on Wilks lambda, F(ll, 30) =
3.63, p < .01, and the effect of topic was not
signifcant, F(ll, 30) = 1.74, p > .11. Hence, all
analyses reported below are collapsed across the
topic factor.
Table 3
Mean Cognitive and Attitudinal Responses as a
Function of the Affectivity of the Advocacy:
Experiment 2
2189
thoughts, ^(1, 40) = 6.58, <.02, and fewer

counterarguments, F(l, 40) = 6.50, p < .02, than
did the counterattitudinal advocacy.
The Dunnett test for comparisons involving
an external control mean (Kirk, 1968) was
employed to determine the relative effects of the
proattitudinal and counterattitudinal communications relative to the neutral communication on cognitive and attitudinal responding.
These comparisons revealed that the neutral
communication differed from the counterattitudinal communication in evaluation and in the
number of neutral/irrelevant thoughts elicited
(ps < .05); in addition, these communications
differed marginally in the number of
counterarguments elicited (<.10). On the other
hand,
the
neutral
and
proattitudinal
communications were evaluated and were
thought about similarly (see Table 3). Evidently
the subjects enjoyed hearing our neutral
message about an archeological dig.
Analyses of the ancillary measures and ratings of tape quality failed to produce any signifcant effects.
Electrophysiological Measures
All tests of the signifcance of changes from
prewarning baselines for the electrophysiological
measures were conducted using two- tailed
Mann-Whitney tests.
* The mean differs from the corresponding neutral

mean at the .05 level by Dunnetts test.
Cognitive Response, Attitude,

and Ancillary Measures
Subjects anticipated and listened to a proattitudinal advocacy, a counterattitudinal advocacy, or a neutral communication, and rated
their evaluation of and thoughts about the taped
presentation. As is evident from an inspection of
the means in Table 3 for these measures, the
proattitudinal advocacy was evaluated more
positively, /'(I,40) =
30.37, p < .001, and elicited more favorable
Does cognitive responding occur naturally?

The analyses of covert oral (mentalis) EMG
activity, which was the most sensitive measure
of covert processing in Experiment 1, indicated
that it was elevated during the postwarningpremessage interval for the counter attitudinal
condition (Mdn .24 jxV, p < .03). This fnding
replicates that of Experiment 1, but here was
obtained without explicit requests for subjects to
collect their thoughts. Interestingly, oral EMG
activity was not altered signifcantly during this
interval in the proattitudinal or neutral
conditions (ps > .15). Still, the presentation of a
message, whether it was proattitudinal, counteratiitudinal, or neutral, led to increased oral
EMG activity (Mdn = .43 uV, 1.21 p-V, and 1.26
respectively, ps < .03). Also, as in

Experiment 1, the affective nature of the covert
processing was not distinguishable by oral EMG;
the mentalis activity did not differ as a function
of position. The analyses of heart rate indicated
that this measure was less sensitive than oral
EMG activity, as differences from basal levels
were evident only during the presentation of the
communications (counterattitudinal Mdn = .73
bpm; proattitudinal Mdn = .36 bpm; neutral
Mdn 2.68 bpm; ps < .05). Finally, changes in
heart rate during the communication were
greater for the neutral than for the counterattitudinal (p < .03) and the proattitudinal (p < .
04) conditions. The cause of this difference is not
immediately apparent, but the news story about
the archeological fnd may have required more
thought to comprehend or elaborate, producing
greater accelerations of heart rate.
/x\\
Is the emotional tone of this cognitive activity

distinguishable? We next sought to determine if
the affective nature of the cognitive responses
was distinguishable by the pattern of facial
EMG activity. Schwartz and his associates have
demonstrated that pleasant states (e.g.,
happiness) lead to less corrugator and more
zygomatic and depressor EMG activity than do
unpleasant states (e.g., sadness, anger), with
corrugator EMG activity providing the most
discriminating measure (Schwartz, Fair, Salt,
Mandel, Mieske, & Klerman, 1978). Figure 2
displays the median change from baseline for
these measures as a function of position and
interval in the present study.
Several fndings are evident immediately
upon inspection of Figure 2. First, only corrugator activity was altered during the an-
nouncement of the forewarning (upper panel),

with greater and equal activation relative to
baseline appearing in all conditions (ps < .01).
Less evident in the upper panel of Figure 2 is
the marginally signifcant tendency for
zygomatic activity to discriminate between the
counterattitudinal and neutral forewarnings (p
< .06).
During the postwarning-premessage interval
(middle panel of Figure. 2), corrugator activity
remained elevated from basal levels (ps < .02),
though a signifcant decrease from the
forewarning level was displayed in the
proattitudinal condition (p < .01). Betweengroup comparisons yielded a nonsignifcant
difference in corrugator activity between the
proattitudinal and counterattitudinal conditions
(/> < .11), with the direction of the difference
that which would be expected from Schwartz
and his colleagues research. That is, corrugator
activity was higher when anticipating the
counter than proattitudinal advocacy. Finally,
the activity in the zygomatic muscle region in
the neutral condition was enhanced relative to
basal levels (p < .03) and distinguished this
condition from the group anticipating a
counterattitudinal advocacy (/ < .01). This too is
consistent with the previous studies of emotional
fantasy and imagery (cf. Schwartz, 1975). No
other differences were statistically signifcant.
The presentation of the messages (lower panel of
Figure 2) resulted in elevated corrugator EMG
activity relative to basal (ps < .01) and
postwarning-premessage levels (p

<0
s4
>
.5
40
3.5
3.
0
56
'l.O
0.
5
0
2
5
III
2.0
1.
5
..Median Change
1.0
for Zygomaticus
0.
5and Depressor
0HV
1.2
Median Change for Corrugator
HV
80
7.
5
7.0
'6.6
0
0
55
5.
a
^
4.5
0
40
Pro- Counterattitudinal attitudinal
Figure 2. Median change from baseline for corrugatory (C), zygomaticus (Z), and depressor (D) electromyographic
activity during the forewarning (top panel), postwarning-premessage (middle panel), and message (bottom panel)
intervals. (The data are displayed separately for subjects who anticipated and heard the neutral [ n 1 2 ] ,
proattitudinal i n = 24], and counterattitudinal [ = 24] advocacies.)
2189
< .01); in addition, it was elevated marginally

during this interval compared to forewarning
levels (s<.06). The zygomatic activity continued
to differentiate the affectivity of the covert
processing: Zygomatic EMG activity during the
counterattitudinal message was lower than that
displayed during baseline (p < .05), the
proattitudinal message (<.01), and the neutral
message (p < .05). Furthermore, the zygomatic
activity during the proattitudinal message was
signifcantly greater than its basal level and
marginally greater than its forewarning (p < .08)
and postwarning-premessage levels (p < .08).
Similarly, depressor EMG activity was enhanced
marginally during the proattitudinal message
compared to its postwarning- premessage level

(p < .06) and compared to the counterattitudinal
message (p < .10), effects also similar to those
found by Schwartz et al. (1976a, 1976b). No
other
comparisons
approached
statistical
signifcance.
In sum, oral EMG activity increased from
baseline after the forewarning of an impending
and involving counterattitudinal communication,
even though we did not request subjects to collect
their thoughts and subjects were unaware that
they would be asked to lis
their thoughts, heard only one forewarning

and message, and had no reason to suspect the
purpose of our measurements. Furthermore,
anticipating
and
hearing
proattitudinal,
counterattitudinal, and neutral communications
led to distinctive and predictable patterns of
facial EMG activity during the stimulus
sequence. Specifcally, less corrugator and more
zygomatic activity was observed in the
proattitudinal and neutral conditions than in the
counterattitudinal condition. The activity of the
depressor muscle region, for the most part, was
unaffected by the affectivity or interval of the
communication sequence. The differences that
were observed, however, were as predicted:
Depressor activity tended to be greater during
the proattitudinal and neutral communications
than
during
the
counterattitudinal
communication. Finally, although the patterns of
facial EMG activity were similar for the neutral
and proattitudinal conditions, there was a high
degree of similarity between these conditions in
the cognitive and evaluative responses as well.
Correlational Analyses
Canonical correlations between the cognitive
responses (i.e., counterarguments, favorable
thoughts, neutral thoughts) and electrophysiological scores (i.e., heart rate, mentalis,
corrugator, zygomatic, and depression EMG
activity) were calculated, once using the
physiological responses for the postwarningpremessage interval, and once using the responses for the message interval. Since thought
listings were obtained immediately following the
message interval, the canonical correlation
between thoughts and the bodily responses from
this interval should show the strongest
association; but physiological responses from
both intervals should correlate somewhat with
the listed thoughts. Correlations were .30 for the
postwarning-premessage interval and .4.6 for the
message interval, X3(15) = 7.20 and 16.20,
respectively, ns. Hence, the covariation of
cognitive and electrophysiological response was
weak by these indices, but the latter index was
stronger than the former, as expected.
2197
Calculations of within-cell correlations among

the cognitive response data revealed that, as in
Experiment 1, counterargumentation correlated
negatively with the attitude toward the
communication (r = .30, p < .05) and favorable
thoughts (r = .41, p < .05), whereas favorable
thoughts correlated positively with attitude (r = .
45, p < .05).
General Discussion
Theory and research in persuasion have
focused recently on the covert idiosyncratic
responses of individuals. In this article, we have
reported two experiments describing the theory
for and development of electrophysiological
procedures for assessing this covert cognitive
activity. Moreover, the evidence obtained
suggests strongly that cognitive response
processes are evoked naturally, at least when the
advocacy is involving and counterattitudinal.
In Experiment 1, we found that subjects who
had been asked to collect their thoughts about an
upcoming
discrepant
message
exhibited
increased oral muscle, cardiac, and respiratory
activity, whereas nonoral somatic activity
remained constant and quiescent. These results
are in accord with research in cardiovascular
psychophysiology (e.g., Lacey et al., 1963) and
with the literature on the electromyographic
concomitants of thought (Jacobsen, 1973;
McGuigan, 1978; Sokolov,
1972) . Further, these results demonstrate that
cognitive responses in persuasion settings are
measurable concurrently and reliably, without
asking the subject to respond overtly during the
measurement and without the subjects
awareness of the purpose or focus of the
(electrophysiological) measurement instruments.
A second experiment was conducted to provide
answers to two important questions: (a) Do
subjects engage normally in active, covert
processing when anticipating and hearing
persuasive appeals? (b) If so, can the emotional
tone of this cognitive activity be assessed
electrophysiologically?
Using
an
elec-
tromyographic technique to measure subtle

responses of the facial muscles, we found that
the forewarning of a counterattitudinal, but not
a proattitudinal or neutral, communication led
to elevated oral EMG activity. Furthermore, the
anticipation
and
presentation
of
counterattitudinal, proattitudinal, and neutral
communications led to active though covert
processing activity, the affective nature of which
was revealed in the concomitant patterning of
facial EMG activity.22
Cognitive Encoding or Elaboration?

It might be argued that the increased oral
muscle activity indexed silent rehearsals rather
than cognitive elaborations of the message
arguments (cf. Miller & Baron, 1973). Listening
to prose does cause a slight increase in oral
muscle activity (McGuigan & Bailey, 1969a).
Hence, comprehending and rehearsing the
message arguments probably contributed to the
elevated oral muscle activity exhibited during
the message presentation. However, the
anticipation of a counterattitudinal advocacy led
to elevated speech muscle activity when there
were no message arguments to rehearse. One
might argue that subjects were rehearsing the
forewarning and potential message arguments,
stopping to generate counterarguments only
when asked by the experimenter at the end of
the study to list everything about which you
thought. Besides the absence of parsimony, this
explanation fails to account for the subtle facial
expressions indicating affect-laden cognitive
responding that was displayed while subjects
awaited the message. More persuasively, this
explanation cannot account for the failure of
subjects who were expecting a proattitudinal
message to display signifcant increases in oral
muscle activity during the postwarningpremessage interval.
22Love (1972) attempted to detect subtle changes in
facial expression by videotaping the shoulders and face
of subjects as they listened to an advocacy. Raters then
scored the nonverbal cues emitted by these subjects.
This measure proved to be insensitive to the
experimental manipulations. The electrophysiological
approach illustrated here has the advantage of being
sensitive to subtle changes in responding by recipients.
Cognitive, Attitudinal, and

Electrophysiological Response
According to the cognitive response analysis,
a forewarning of an upcoming and involving
discrepant communication elicits preparatory
cognitive
activity
(e.g.,
anticipatory
counterargumentation) in an effort to rally ones
cognitive defenses (Cialdini et al., 1976;
McGuire & Papageorgis, 1962; Petty & Cacioppo, 1977). Interestingly, there are no studies
in the literature that report the effects of
anticipating a proattitudinal message that
collected thought listings. We believe that
anticipating a counterattitudinal as compared to
a proattitudinal message generally results in
deeper processing (i.e., more extensive cognitive
preparationCacioppo & Petty,
1979) . This difference may be attributable to
the relative importance in defending from attack
ones attitudes and beliefs (e.g., avoiding
cognitive inconsistency), or to the relative
thought that was devoted previously to (or
scripts developed for) proattitudinal rather than
counterattitudinal positions. The present study
does not consider which of these interpretations
is most plausible.
It should be noted that the observation of
greater oral EMG activity being obtained when
anticipating a counterattitudinal rather than
proattitudinal or neutral communication does
not imply that oral EMG activity is a measure of
counterargumentation. Indeed, the results of
both experiments suggest it is not. As mentioned
above, we believe that the cognitive preparation
for a counterattitudinal rather than a
proattitudinal or neutral message elicits more
extensive processing, which may result in the
generation of counterarguments, favorable
thoughts, and/or neutral thoughts. Accordingly,
the activation of the oral muscles reflects this
difference in the extent of cognitive elaboration
rather than the affectivity of the processing.
Evidence that oral EMG reflects this depth of
processing has been found in a study in which
subjects viewed a word and identifed whether it
was printed in uppercase letters (shallow
processing) or was self-descriptive (deep
processing Cacioppo & Petty, in press-b, in
press-c). We found that deeper processing

resulted in elevated speech EMG activity.
2197
Brock, T. C. Communication discrepancy and intent to

persuade as determinants of counterargument
production. Journal of Experimental Social Psychology, 1967, 3, 269-309.
Finally, a word might be said about the

Burnstein, E., & Vinokur, A. Persuasive argumentation and
sensitivity and specifcity of the electrophysio- social comparison as determinants of attitude
logical measures employed. The procedures we polarization. Journal of Experimental Social
employed for scoring these responses eliminated Psychology, 1977, 13, 315-332.
obvious
movement
artifacts
and
facial
Cacioppo, J. T. Heart rate, cognitive response, and
expressions (the number of the edits did not
persuasion. Unpublished doctoral dissertation, The
differ across the experimental conditions) ;
Ohio State University, 1977.
hence, we intentionally confned ourselves to the
study of covert electrophysiological responses. Cacioppo, J. T. The effects of exogenous changes in
heart rate on the facilitation of thought and reNevertheless, in this research, oral EMG activity
sistance to persuasion. Journal of Personality and
has been a more sensitive measure of covert
Social Psychology, 1979, 37, 487-496.
information processing than heart rate (see also
McGuigan, 1978). Heart rate and speech muscle Cacioppo, J. T., Glass, C. R., & Merluzzi, T. V. Selfactivity increased following the forewarning of
an upcoming discrepant message, but the change
in heart rate in Experiment 2 was not signifcant
statistically ; heart rate did increase
signifcantly during the presentation of the
communications,
although
again,
the
electromyographic measures proved more
sensitive. Similarly, the various measures of
facial EMG activity were not equally sensitive to
affect-laden processing. Schwartz and his
colleagues (1976a, 1976b, 1978) have found, as
we here have found, that corrugator EMG
activity best distinguishes subtle affective states.
These instances of differing sensitivity, or
response discoi dance, illustrate a point made by
the Laceys (e.g., 1959, 1967) regarding the specifcity of autonomic and somatic activation.
Moreover, this remarkable specifcity contrasts
sharply with Cannons (1927) theory of
physiological arousal and emotion, and provides
unique evidence for cognitive response processes
in persuasion.
References
Baron, R. S., Baron, P. H., & Miller, N. The relationship
between distraction and persuasion. Psychological
Bulletin, 1973 , 80, 310-323.
Blatt, S. J. Patterns of cardiac arousal during complex
mental activity. Journal of Abnormal Social
Psychology, 1961, 63, 272-282.
Blaylock, B. Some antecedents of directional fractionation: Effects on intake-rejection, verbalization
requirements, and threat of shock on heart rate and
skin conductance. Psychophysiology, 1972, 9, 40-52.
statements and self-evaluations: A cognitiveresponse analysis of heterosocial anxiety. Cognitive

Therapy and Research, 1979, 3, 249-262.
Cacioppo, J. T., Harkins, S. G., & Petty, R. E. The

nature of attitudes and cognitive responses and their
relationships to behavior. In R. E. Petty, T. M.
Ostrom, & T. C. Brock (Eds.), Cognitive responses in
persuasion: A text in attitude change. Hillsdale, N.J.:
Eribaum, in press.
Cacioppo, J. T., & Petty, R. E. Effects of message
repetition and position on cognitive response, recall,
and persuasion. Journal of Personality and Social
Psychology, 1979, 37, 97-109.
Cacioppo, J. T., & Petty, R. E. Neurontuscular circuits
in affect-laden information processing. The
Pavlovian Journal of Biological Science, in press, (a)
Cacioppo, J. T., & Petty, R. E. Electromyographic
specifcity during covert information processing.
Psychophysiology, in press, (b)
Cacioppo, J. T., & Petty, R. E. Lip and nonpreferred
forearm EMG activity as a function of orienting task.
Biological Psychology, in press, (c)
Cacioppo, J. T., Petty, R. E., & Snyder, C. W. Cognitive
and affective response as a function of relative
hemispheric involvement. The International Journal
of Neurosciences, 1979, 9 , 81-89.
Cacioppo, J. T., & Sandman, C. A. Physiological
differentiation of sensory and cognitive tasks as a
function of warning, processing demands, and reported unpleasantness. Biological Psychology, 1978,
6, 181-192.
Cacioppo, J. T., & Sandman, C. A. Psychophysiolog- ical
functioning, cognitive responding, and attitudes. In
R. E. Petty, T. M. Ostrom, & T. C. Brock (Eds.),
Cognitive responses in persuasion. Hillsdale, X.J.:
Eribaum, in press.
Cacioppo, J. T., Sandman, C. A., & Walker, B. B. The
effects of operant heart rate conditioning on
cognitive elaboration and attitude change. Psychophysiology, 1978, IS, 330-338.
Calder, B. J., Insko, C. A., & Yandell, B. The relation of
cognitive and memorial processes to persuasion in a
simulated jury trial. Journal of Applied Social
Psychology, 1974, 4, 62-93.
Cannon, W. B. The James-Lange theory of emotions: A
critical examination and an alternative theory.
American Journal of Psychology, 1927, 39, 106124.Cialdini, R. B., Levy, A., Herman, C. P.,
Kozlowski, L. T., & Petty, R. E. Elastic shifts of
opinion: Determinants of direction and durability.
Journal of Personality and Social Psychology, 1976,
34, 663-672.
Cialdini, R. B., & Petty, R. E. Anticipatory opinion
effects. In R. E. Petty, T. M. Ostrom, & T. C. Brock
(Eds.), Cognitive responses in persuasion. Hillsdale,
N.J.: Eribaum, in press.
Cook, T. D. Competence, counterarguing, and attitude
change. Journal of Personality, 1969, 37, 342- 3S8.
Cooper, J. B. Emotion in prejudice. Science, 1959, 130,
314-318.
Craik, F. I. M., & Tulving, E. Depth of processing and
the retention of words in episodic memory. Journal of
Experimental Psychology:General,
1975, 104, 263-294.
Cullen, D. M. Attitude measurement by cognitive
sampling. Unpublished doctoral dissertation, Ohio
State University, 1968.
Darwin, C. The expression of the emotion in man and
animals. Chicago: University of Chicago Press, 1965.
(Originally published, 1872.)
Eagly, A. H. Comprehensibility of persuasive arguments
as a determinant of opinion change. Journal of
Personality and Social Psychology, 1974, 29, 758-773.
Edwards, D. C., & Alsip, J. E. Stimulus detection
during periods of high and low heart rate. Psychophysiology, 1969, S, 431-434.
Elliott, R. The motivational signifcance of heart rate.
In P. A. Obrist, A. H. Black, J. Brener, & L. V. DiCara
(Eds.), Cardiovascular psychophysiology. Chicago:
Aldine, 1974.
Festinger, L., & Maccoby, N. On resistance to persuasive communications. Journal of Abnormal and
Garrity, L. I. Electromyography: A review of the current

status of subvocal speech research. Memory &
Cognition, 1977, 5, 615-622.
Gillig, P. M., & Greenwald, A. G. Is it time to lay the
sleeper effect to rest? Journal of Personality and
Goldwater, B. C. Psychological signifcance of pupillary
movements. Psychological Bulletin, 1972, 77, 340-355.
Greenfeld, N. S., & Sternbach, R. A. (Eds.). Handbook
of psychophysiology. New York: Holt, Rinehart &
Winston, Inc., 1972.
Greenwald, A. G. Cognitive learning, cognitive response
to persuasion, and attitude change. In A. G.
Greenwald, T. C. Brock, & T. M. Ostrom (Eds.),
Psychological foundation of attitudes. New York:
Academic Press, 1968.
Hardyck, C. D., & Petrinovich, L. F. Treatment of
subvocal speech during reading. Journal of Reading,
1969, 12, 361-368, 419-422.
Harris, R. J. A primer of multivariate statistics. New
York: Academic Press, 1975.
Harris, V. A., Katkin, E. S., Lick, S. R., & Habber- feld,
T. Paced respiration as a technique for the
modifcation of autonomic response to stress. Psychophysiology, 1976, 13, 386-391.
Hess, E. H. Attitude and pupil size. Scientific American, 1965, 212, 46-54.
Hovland, C. I., Lumsdaine, A. A., & Sheffield, F. D.
Experiments on mass communication. Princeton,
N.J.: Princeton University Press, 1949.
Huynh, H., & Feldt, L. S. Conditions under which mean
square ratios in repeated measurements designs
have exact F-distributions. Journal of the American
Statistical Association, 1970, 65, 1582- 1589.
Insko, C. A., Turnbull, W., & Yandell, B. Facilitative
and inhibiting effects of distraction on attitude
change. Sociometry, 1974, 37, 508-528.
Izard, C. E. The face of emotion. New York: Ap- pletonCentury-Crofts, 1971.
Jacobsen, E. Electrophysiology of mental activities and
introduction to the psychological process of thinking.
In F. J. McGuigan, & R. A. Schoonover (Eds.), The
psychophysiology of thinking: Studies of covert
processes. New York: Academic Press, 1973.
Janis, I. L., & Mann, L. Decision making: A psychological analysis of conflict, choice, and commitment. New York: Free Press, 1977.
Kahneman, D. Attention and effort. Englewood Cliffs,

N.J.: Prentice-Hall, 1973.
Keating, J. P., & Brock, T. C. Acceptance of persuasion
and the inhibition of counterargumentation under
various distraction tasks. Joumal 'of Experimental
Kirk, R. E. Experimental design: Procedures for the
behavioral sciences. Belmont, Calif.: Brooks/Cole,
1968.
Lacey, J. I. Psychophysiological approaches to the
evaluation of psychotherapeutic process and outcome.
In E. A. Rubinstein & M. B. Parloff (Eds.), Research
in psychotherapy. Washington, D.C.: American
Psychological Association, 1959.
Lacey, J. I. Somatic response patterning and stress:
Some revisions of activation theory. In M. H. Appley
& R. Trumbull (Eds.), Psychological stress: Issues in
research. New York: Appleton, 1967.
Lacey, J. I., Kagan, J., Lacey, B., & Moss, H. A. The
visceral level: Situational determinants and
behavioral correlates of autonomic response patterns. In P. H. Knapp (Ed.), Expression of the
emotions in man. New York: International Universities Press, 1963.
Langer, E., Blank, A., & Chanowitz, B. The mindlessness of ostensibly thoughtful action: The role of
placebic information in interpersonal interaction.
36, 635-642.
Lorens, S. A., Jr., & Darrow, C. W. Eye movements,
EEG, GSR, and EKG during mental multiplication.
Electroencephalography
and
Clinical
Neurophysiology, 1962, 14, 739-746.
Love, R. A videotape technique for the measurement
of cognitive response to persuasion. Unpublished
doctoral dissertation, Ohio State University, 1972.
McGuigan, F. J. Subvocal speech during silent reading.
(Accession No. ED 015115). Bethesda, Md.: ERIC
Document Reproduction Service, 1967.
McGuigan, F. J. Covert oral behavior during the silent
performance of language tasks. Psychological
Bulletin, 1970, 74, 309-326.
McGuigan, F. J. Conditioning of covert behavior some
problems and some hopes. In F. J. McGuigan & D. B.
Lumsden (Eds.), Contemporary approaches to
conditioning and learning. Washington, D.C.:
Winston, 1973. (a)
McGuigan, F. J. Electrical measurement of covert
processes as an explication ot higher mental events."
In F. J. McGuigan & R. A. Schoonover (Eds.), The
processes. New York: Academic Press, 1973. (b)
2197
McGuigan, F. J. Cognitive psychophysiology. Englewood

Cliffs, N.J.: Prentice-Hall, 1978.
McGuigan, F. j'., & Bailey, S. C. Covert response
patterns during the processing of language stimuli.
Interamerican Journal of Psychology, 1969, 3, 289299. (a)
McGuigan, F. T., & Bailey, S. C. Longitudinal study of
covert oral behavior during silent reading. Perceptual
and Motor Skills, 1969, 28, 170. (b)
McGuigan, F. J., Culver, V. I., & Kendler, T. S. Covert
behavior as a direct electromyographic measure of
mediating responses. Conditional Reflex, 1971, 6,
145-152.
McGuigan, F. J., Keller, P., & Stanton, E. Covert
language responses during silent reading. Journal of
Educational Psychology, 1964, 55, 339-343.
McGuigan, F. J., & Rodier, W. I., III. Effects of auditory
stimulation on covert oral behavior during silent
reading. Journal of Experimental Psychology, 1968,
76, 649-655.
McGuigan, F. J., & Schoonover, R. A. (Eds.). The
processes. New York: Academic Press, 1973.
McGuire, W. J., & Papageorgis, D. Effectiveness of
forewarning in developing resistance to persuasion.
Public Opinion Quarterly, 1962, 26, 24-34.
Miller, N., & Baron, R. S. On measuring counterarguing. Journal for the Theory of Social Behavior,
1973 , 3, 101-118.
Mueller, D. J. Physiological techniques of attitude
measurement. In G. F. Summers (Ed.), Attitude
measurement. Chicago: Rand McNally, 1970.
Obrist, P. A., Webb, R. A., Sutterer, J. R., & Howard, J.
L. The cardiac-somatic relationship:
Some
reformulations. Psychophysiology, 1970, 6, 569- 587.
Petty, R. E. A cognitive response analysis of the
temporal persistance of attitude changes induced by
persuasive communications. Unpublished doctoral
dissertation, Ohio State University, 1977.
Petty, R. E., & Cacioppo, J. T. Forewarning, cognitive
responding, and resistance to persuasion. Journal of
Petty, R. E., & Cacioppo, J. T. Effects of forewarning of
persuasive intent and involvement on cognitive
responses and persuasion. Personality and Social
Psychology Bulletin, 1979, 5, 173-176. (a)
Petty, R. E., & Cacioppo, J. T. Issue involvement can
increase or decrease persuasion by enhancing
message-relevant cognitive responses. Journal of
Personality and Social Psychology, 1979, 37, 19151926. (b)
depression: An electromyographic study. Psychosomatic Medicine, 1976, 38, 337-347. (a)
Petty, R. E., Ostrom, T. M., & Brock, T. C. (Eds.).

Cognitive responses in persuasion. Hillsdale, N.J.:
Eribaum, in press.
Schwartz, G. E., Fair, P. L., Salt, P., Mandel, M. R., &

Klerman, G. L. Facial muscle patterning to affective
imagery in depressed and nondepressed subjects.
Science, 1976, 192, 489-491. (b)
Petty, R. E Wells, G. L., & Brock, T. C. Distraction can

enhance or reduce yielding to propaganda: Thought
disruption versus effort justifcation. Journal of
Posner, M. I. Psychobiology of attention. In M. S.
Gazzaniga & C. Blakemore (Eds.), ffandbook of
psychobiology. New York. Academic Press, 1975.
Rhodewalt, F., & Comer, R. Induced-compliance attitude change: Once more with feeling. Journal of
Experimental Social Psychology, 1979, 15, 35-47.
Rogers, T. B., Kuiper, N. A., & Kirker, W. S. Selfreference and the encoding of personal information.
35, 677-688.
Romer, D. Distraction, counterarguing, and the internalization of attitude change. European Journal
of Social Psychology, 1979, 9, 1-17.
Sandman, C. A. McCanne, T. R., Kaiser, D. N., &
Diamond, B. Heart rate and cardiac phase influences
on visual perception. Journal of Comparative and
Physiological Psychology, 1977, 91, 189-202.
Schwartz, G. E., Fair, P. L., Mandel, M. R., Salt, P.,

Mieske, M., & Klerman, G. L. Facial electromyography in the assessment of improvement in depression. Psychosomatic Medicine, 1978, 40, 355-360.
Schwartz, G. E., & Higgins, J. D. Cardiac activity
preparatory to overt and covert behavior. Science,
1971, 173, 1144-1146.
Shmavonian, B. M., Miller, L. H., & Cohen, S. I.
Differences among age and sex groups in electrodermal conditioning. Psychophysiology, 1968, 5, 119131.
Sokolov, A. N. Speech-motor afferentiation and the
problem of brain mechanisms. Soviet Psychology,
1967, 6, 3IS.
Sokolov, A. N. Studies of the Speech mechanisms of
thinking. In M. Cole & I. Maltzman (Eds.), A
handbook of contemporary Soviet psychology. New
York: Basic Books, 1969.
Sokolov, A. N. Inner speech and thought. New York:
Plenum Press, 1972.
'
Schachter, S. The interaction of cognitive and physiological determinants of emotional state. In L.

Berkowitz (Ed.), Advances in experimental social
psychology. New York: Academic Press, 1964.
Tognacci, L. N., & Cook, S. W. Conditioned autonomic

responses as bidirectional indicators of racial
attitude. Journal of Personality and Social Psychology, 1975, 31, 137-144.
Schwartz, G. E. Cardiac responses to self-induced

thoughts. Psychophysiology, 1971, 8, 462-467.
Tursky, B., Schwartz, G. E., & Crider, A. Differential

patterns of heart rate and skin resistance during a
digit-transformation task. Journal of Experimental
Psychology, 1970, S3, 451-457.
Schwartz, G. E. Biofeedback, self-regulation, and the

patterning of physiological processes. American
Scientist, 1975, 63, 314-324.
Schwartz, G. E., Fair, P. L., Salt, P., Mandel, M. R., &
Klerman, G. L. Facial expression and imagery in
Wallace, S., & Wallace, J. D. Hemodensitometry in the

evaluation of cerebral circulation. Radiology, 1968,
90, 454-465.
Received October 25, 1978

1
' This unavailability of measures of implicit self-esteem may be rectifed by efforts currently underway
(Wood, Taylor, Michela, & Gaus, 1993).

5
We are indebted to Richard Harris for providing a computer program with which to test the hotdv assumption.
6
Subjects were asked to count to fve aloud and to move and tense slightly in their chair before the completion of
the experiment to assess the validity of the EMG electrode placements. Trapezius EMG activity increased during
body tensing and movements. Lip and chin EMG activity increased during overt oral behavior (counting aloud),
whereas the throat EMG placement proved to be a relatively insensitive measure.
7
Analysis of covariance procedures were employed to explore some of the possible causal sequences of cognitive
responding and attitude change. It should be noted, however, that these procedures do not prove that a particular
causal model is operating. These analyses were conducted here to assess if the reduced agreement found with
increasing discrepancy possibly resulted from counterargumentation. Previously we have found that topic-relevant
thinking appears to mediate the subsequent agreement with the advocacy (Cacioppo & Petty, 1979; Petty &
Cacioppo, 1977), rather than vice versa. We expected the same results from the present study. (See Insko et al.,
1974, for an extended rationale for the use of analyses of covariance [axcovaI with attitude and cognitive response
2197
measures.)
A series of analyses of covariance were performed in which agreement served as the criterion and each cognitive
response (one per analysis) served as the covariate. A second series of covariance analyses was performed in which
agreement served as the covariate and each cognitive response served as the criterion. The results of these
analyses revealed only that the covariate, counterarguments, reduced the overall F ratio for the discrepancy factor
for agreement: the F statistic for agreement was reduced from a significant 3.57 to a nonsignifcant 1.70 ( d f 2 ,
41). The analysis of covariance with counterarguments as the criterion and agreement as the covariate did not
eliminate the signifcant F ratio for counterarguments, ancova F ( 2 , 41) = 5.53. These analyses are consistent with
the notion that increasing discrepancy increased counterargumentation, which then reduced agreement.
9
Electroencephalographic measures were obtained also, the results of which are to be reported else where, since
they were collected to address a different issue. Suffice it to say that enough electrodes were attached to subjects to
make the cover story concerning the measurement of involuntary processes seem entirely plausible to them.

Respon Kognitif

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Respon Kognitif

Încărcat de

Drepturi de autor:

Formate disponibile

I

ACADEMIC PRESS INC..

Cognitive Learning, Cognitive Response to Persuasion, and Attitude

DEPARTMENT OF PSYCHOLOGY OHIO STATE UNIVERSITY COLUMBUS, OHIO

arousal, and organization of arguments have not been found to have

It must be concluded that either (a) learning of attitude-relevant cognitions

There is, of course, an important extracommunication source of cognitive

to persuasion than is the learning of communication content.

persuasion are employed as independent variables in experimental persuasion

It is proposed that the persuasion situation is usefully regarded as a complex

As in many other treatments of persuasion, the cognitive response analysis

As a consequence of the present emphasis on the recipients rehearsal of his

communications. The present formulation, therefore, is capable of maintaining

It is possible to formulate the cognitive response analysis in terms of an

STUDIES OF COGNITIVE LEARNING AND ATTITUDE CHANGE

The cognitive response analysis of persuasion has guided a series of

It is to be emphasized that the research program described here is very

Previous studies of the relationship between communication retention and

The experiment employed three communications concerning United States

the individual arguments contained in it, in preparation for a subsequent

The effectiveness of the partial learning assignments for the two-sided

It must be concluded that learning of communication content is not a

If, as is presently supposed, attitudes change in the direction of cognitive

graduate education. Improvisation was obtained in response to fve neutrally

alternative explanations are possible in terms of

perception of the self as the source of arguments or in terms of subjects ability

A series of experiments conducted by Dallas Cullen, under the authors

relevant to the topic of a persuasive communication that they had just

The subjects in this series of experiments were Ohio State University

Following this thought-listing procedure, subjects completed a brief

Experiment II added an opinion pretest to the basic procedure; it consisted

In the last experiment in this series, two different communication conditions

Recipient modified thoughts

Recipient generated thoughts

I I n HA EB Communication receiving groups

immediately after reading a communication containing a dozen distinct

Subjects judgments of the position supported by each of their listed

It may be noted that the communication received by subjects in group IVB

The correlations with posttest opinion for the recipient-generated category

procedure in comparison with more traditional attitude scaling procedures.

Table 1 gives the additional correlational data that were obtained

"Experiment II Data; N = 48.

from Experiment II by virtue of the use of a pretest opinion measure. On

In summary, Cullens experiments provided a variety of evidence

The preceding series of experiments examined attitude-relevant

strongly support the present contention that cognitive responses to

Supplementary data were obtained from subjects classifcations of the

while the communication content was least often represented (mean

At present, knowledge concerning the determinants of learning of

1966) . Until quite recently, another widely accepted principle of

Other determinants of attitude-relevant learning that have been

utility and novelty are variables known to increase attention to

also Anderson & Hubert, 1963) may be interpreted in terms of the

The effects of rewards and punishments occurring in the persuasion

punishment effects, interpreted in terms of conditioning processes, in the

The procedure of having subjects actively rehearse their own persuasive

The present program of research set out to establish the legitimacy of a

In light of these observations, it would appear fruitful to approach the

communication content learning. Unpublished manuscript, Ohio State Univ., 1968.

cation. Princeton, N. J.: Princeton University Press, 1949.

The Cognitive-Affective Crossfire: When SelfConsistency Confronts Self-Enhancement

University of Texas at Austin

When we undertake to cure a patient, to