Journal of Personality and Social Psychology Copyright 1991 by the American Psychological Association. Inc.

1991, Vol. 60, No. 4, 509-517 0022-3514/91/53.00

The Trouble of Thinking: Activation and Application of Stereotypic Beliefs

Daniel T. Gilbert and J. Gregory Hixon
University of Texas at Austin

Two studies investigated the effects of cognitive busyness on the activation and application of
stereotypes. In Experiment 1, not-busy subjects who were exposed to an Asian target showed
evidence of stereotype activation, but busy subjects (who rehearsed an 8-digit number during their
exposure) did not. In Experiment 2, cognitive busyness once again inhibited the activation of
stereotypes about Asians. However, when stereotype activation was allowed to occur, busy subjects
(who performed a visual search task during their exposure) were more likely to apply these activated
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

stereotypes than were not-busy subjects. Together, these findings suggest that cognitive busyness
This document is copyrighted by the American Psychological Association or one of its allied publishers.

may decrease the likelihood that a particular stereotype will be activated but increase the likeli-
hood that an activated stereotype will be applied.

Labels are devices for saving talkative persons the trouble of think- not mean that it is the only option available." Indeed, most
ing. (Morley, 1886, p. 142) theories of social inference acknowledge the predominance of
stereotypic thinking but also award a significant role to indi-
Of all the many observations that writers have made about
viduating thought (e.g., Bodenhausen & Lichtenstein, 1987;
human prejudice and preconception, one enjoys near-perfect
Brewer, 1989; Fiske & Neuberg, 1990; see also Locksley, Hep-
consensus: A stereotype is the sluggard's best friend. As Allport
burn, & Ortiz, 1982). Under some circumstances, then, people
(1954, pp. 20-21) noted:
do think about others as unique individuals, and contemporary
We like to solve problems easily. We can do so best if we can fit theorists have debated at length about what those circum-
them rapidly into a satisfactory category and use this category as a stances might be.
means of prejudging the solution.... So long as we can get away One such circumstance falls neatly out of Allport's principle
with coarse overgeneralizations we tend to do so. Why? Well, it
takes less effort, and effort, except in the area of our most intense of cognitive economy: If a stereotype is, in fact, an energy-sav-
interests, is disagreeable. ing device, then people should be particularly prone to employ
that device when they are short on energy (e.g., Bodenhausen,
Modern theorists have raised Allport's notion of cognitive 1990; Bodenhausen & Lichtenstein, 1987; Bodenhausen &
economy to the status of a first principle. Stereotypes and pre- Wyer, 1985; Fiske & Pavelchak, 1986; Pratto & Bargh, 1991;
conceptions do not save people the trouble of thinking en pas- Rothbart, Fulero, Jensen, Howard, & Birrell, 1978). To the ex-
sant, but rather, as their cardinal function: "The task of cate- tent that individuals are cognitively busy (ije, simultaneously
gory systems is to provide maximum information with the least involved in several resource-consuming tasks), they should be
cognitive effort" (Rosch, 1978, p. 28). The ability to understand especially inclined to use stereotypes. This notion makes so
new and unique individuals in terms of old and general beliefs much sense, and has been advanced by so many sensible theo-
is certainly among the handiest tools in the social perceiver's kit rists, that it has become a hub of the conventional wisdom. The
(e.g., Andersen, Klatzky, & Murray, 1990; Hamilton, 1981; purpose of the present article is to argue that the state of affairs
Hamilton & Trolier, 1986). is not quite so simple: Cognitive busyness may indeed exacer-
Of course, the fact that people may employ such devices does bate the perceiver's tendency to use stereotypes, but it may also
not mean that they must always do so. As Fiske (1989, p. 277) abrogate precisely the same tendency.
has argued, "The idea that categorization is a natural and adap-
tive, even dominant, way of understanding other people does
Activation and Application of Stereotypes
Anyone who has ever lent a socket wrench to a forgetful
This research was supported by National Science Foundation Grant neighbor knows that a tool is useful only if one can find it.
BNS-8819836 to Daniel T. Gilbert and a National Science Foundation Stereotypes are forms of information and, as such, are thought
predoctoral fellowship to J. Gregory Hixon. to be stored in memory in a dormant state until they are acti-
We thank John Bargh, Susan Fiske, Bill Swann, and several anony- vated for use. Many theorists have assumed that the activation
mous reviewers for their insightful comments on a previous version of of a stereotype is an automatic and inevitable consequence of
this article. We also thank Traci Giuliano, Jennifer Ho, Cheryl Irwin,
and Debbi Wagner for their kind help with the preparation of stimulus
encountering the object of that stereotype. Thus, for instance,
materials. Allport (1954, p. 21) argued that "Every event has certain marks
Correspondence concerning this article should be addressed to that serve as a cue to bring the category of prejudgment into
Daniel T. Gilbert, Department of Psychology, University of Texas, action.... A person with dark brown skin will activate what-
Austin, Texas 78712. Electronic mail should be sent to ever concept of Negro is dominant in our mind." Brewer (1989),
[email protected]. Devine (1989), Dovidio, Evans, and Tyler (1986), and Fiske and

509
 510 DANIEL T. GILBERT AND J. GREGORY HIXON

Neuberg (1990) have all considered varieties of this argument, Caucasians who were native speakers ofEnglish were eligible to partici-
which (to strain the metaphor) suggests that stereotypes are pate.
tools that jump out of the toolbox when there is a job to be
done. In short, stereotypes are forms of information that must
be activated before they can be applied to perceptual or judg- Procedure
mental operations, and some theorists have argued that such Upon arrival at the laboratory, subjects were greeted by a male exper-
activation is an inevitable consequence ofexposure to the stereo- imenter who escorted them to an individual cubicle that was equipped
type object. with a video monitor and a taperecorder.The experimenter explained
There is no doubt that some mental operations require very that he was testing the hypothesis that people are capable of perform-
little effort or intent. Word meanings spring to mind when their ing two tasks simultaneously as long as the two tasks involve different
written referents are encountered (LaBerge & Samuels, 1974; cerebral hemispheres. Subjects in the busy condition were told that
Logan, 1980; Stroop, 1935); affective responses may overwhelm they would be asked to perform simultaneously a verbal and a nonver-
bal task that, according to the experimenter's ostensible hypothesis,
one with their speed and intensity (Fazio, Sanbonmatsu, Pow-
would not be mutually debilitating. Subjects in the not-busy condition
ell, & Kardes, 1986; Zajonc, 1980a); and even complex beliefs
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

were told that they had been assigned to a control condition and would
about others can be activated without one's awareness (Bargh &
This document is copyrighted by the American Psychological Association or one of its allied publishers.

therefore be performing only the verbal task.

Pietromonaco, 1982; Brewer, 1989; Devine, 1989; Lewicki, Word-fragment completion test. The experimenter explained that
1985). Nonetheless, despite the ease with which such phenom- the verbal task (which all subjects were to perform) required the subject
ena seem to occur, none of these operations is unconditionally to observe a word fragment (e.g., P ST) and then to generate its com-
automatic in that it can occur in the complete absence of inten- pletions (e.g., POST, PAST, and PEST). Subjects were told that they
tion, volition, awareness, or processing resources (Bargh, 1989). would see a videotape in which a female assistant would hold up a
Indeed, even simple perceptual operations that were once series of cards, on each of which would be printed a fragment. The
thought to be entirely automatic have been shown to have limit- subject's task was to read the fragment, generate as many completions
ing preconditions (see Kahneman & Treisman, 1984). as possible during the 15 s that the card was being displayed, and state
each of these completions aloud into a tape recorder. The word-frag-
In this article we will endorse the view that the automaticity ment completion test has been shown to be extremely sensitive to the
of stereotype activation is also conditional (Bargh, 1989), and activation of constructs that have been either recently encountered
that mere exposure to a stereotype object is therefore insuffi- (Tulving, Schacter, & Stark, 1982) or self-generated (Bassili & Smith,
cient to activate the corresponding stereotype. What, then, are 1986).
the conditions under which meeting "a person with dark brown The female assistant on the videotape displayed a total of 19 cards,
skin" will activate "whatever concept of Negro is dominant in each of which bore one fragment. Of these 19 word-completion trials, 5
our mind?" We suggest that the availability of processing re- were considered critical and 14 were considered fillers. Earlier in the
sources constitutes one essential precondition for stereotype semester, 20 Caucasian students were asked to "list all the words that
activation and that, as such, cognitively busy perceivers who come quickly to mind when you think about Asian-American stu-
encounter a stereotype object may show no evidence of stereo- dents." Of the words generated, eight concepts (i.e., words and their
close synonyms) were independently generated by at least 33% of the
type activation. Because stereotypes must be activated before
Caucasian pretest subjects. Each concept was then designated by a
they can be applied to perception and judgment, such a finding single word (e.g., smart was the word chosen to designate the concept of
would mean that, contrary to the conventional wisdom, cogni- intelligence). These eight words were then pretested in a fragment-
tively busy perceivers may occasionally be less likely than not- completion test. Three word fragments were shown to have an exceed-
busy perceivers to construe others in stereotypic terms. Experi- ingly common stereotypic completion (i.e, a completion that was gen-
ment 1 was an attempt to show that cognitive busyness can erated first by virtually all pretest subjects, e.g., QUI and
inhibit the simple activation of stereotypes, and Experiment 2 QUIET), and these three were eliminated because such word frag-
was an attempt to show that when stereotypes are activated, ments would not discriminate between subjects who had and had not
cognitive busyness can facilitate their application. experienced stereotype activation. This leftfivewords that were desig-
nated as stereotypic words. Thesefivewords met three important crite-
ria: (a) They were spontaneously generated by more than 33% of the
Experiment 1 Caucasian students who thought about Asian-American students; (b)
their fragments could easily be completed in more than one way; and (c)
Method the stereotypic completions were no more or less common or fre-
Overview quently occurring than were the other correct completions of the word
fragment (M word frequencies =53.6 and 62.0,respectively,t < 1, as
Subjects performed a word-fragment completion test while being assessed by Kucera & Francis, 1967). Thefivefragments and their
exposed to either a Caucasian or Asian female assistant. Five of the stereotypic completions were: S _ Y (SHY), S—ORT (SHORT),
word fragments could be completed with words that were stereotypi- R I _ E (RICE), POLL_E (POLITE), and N _ P (NIP).
cally associated with Asians (e.g., POLI E could be completed as Independent manipulations. Subjects saw a silent videotape in
POLITE). All subjects made as many completions as they could in 15 s. which a female assistant turned over a series of 19 cards, each of which
Half the subjects performed a resource-consuming rehearsal task dur- bore a word fragment. Half the subjects saw a videotape in which the
ing the completion test, and the remaining subjects did not. card-turning assistant was Caucasian, and theremainingsubjects saw a
videotape in which the assistant was Asian. Neither the Caucasian nor
Subjects Asian assistant spoke at any time during the videotape and, except for
the assistant's ethnicity, the two tapes were identical.
Seventy-one female students at the University of Texas participated Some subjects were made cognitively busy while they watched the
to fulfill a requirement in their introductory psychology course. Only videotape. Prior to the start of the videotape, the experimenter asked
 ACTIVATION AND APPLICATION OF STEREOTYPES 511

subjects in the busy condition to rehearse an eight-digit number while It is worth noting that busyness did not itself impair any
they watched the videotape, whereas subjects in the not-busy condition obvious aspects of task performance. Busy and not-busy sub-
were not asked to rehearse a number. This and similar rehearsal tasks jects generated an equivalent total number of correct comple-
have been used with excellent results in a number of experiments to tions across all trials (Ms =41.50 and 41.29,respectively,t < 1)
deprive subjects of processing resources (e.g., Gilbert, Pelham, & and on critical trials (Ms = 11.77 and 12.09, respectively, t < 1).
Krull, 1988, Experiment 1; Gilbert & Osborne, 1989, Experiments 1 -4; Busy and not-busy subjects generated equally common or fre-
Osborne & Gilbert, 1990, Experiments 1-3; Swann, Hixon, Stein-Se-
roussi, & Gilbert, 1990, Experiments 1 and 3). After subjects finished
quently occurring words across all trials (Ms =75.17 and 76.52,
watching the videotape, the experimenter returned to the room and, if respectively, t < 1) and on critical trials (Ms = 77.76 and 75.79,
the subject had been assigned to the busy condition, asked the subject respectively, t < 1). Finally, busy and not-busy subjects showed
to report the eight-digit number. Next, subjects completed some filler equally good recall of the assistant's race (94.1% and 91.2%
items and were then asked to report the color of the ink in which the correct recall, respectively), x2(l) < 1, and of the color in which
fragments were printed and the race of the female assistant. Finally, the word fragments were printed (58.9% and 61.8% correct re-
subjects were probed for suspicion, thoroughly debriefed, and dis- call, respectively), %20) < 1- Taken together, these results sug-
missed. gest that cognitive busyness did not prevent subjects from per-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

forming well on the completion task or from noticing the assis-

This document is copyrighted by the American Psychological Association or one of its allied publishers.

Results and Discussion tant's race, but that it did inhibit the activation of their
stereotypes about Asians.
Error Rates and Excluded Data
It is difficult to know how to interpret the errors that busy A Caveat
subjects made when reporting the eight-digit number. Large
errors may mean that busy subjects were not rehearsing the Given that busy subjects were presumably deprived of pro-
number (i.e., that they were not, in fact, busy), whereas small cessing resources, one might expect that they would perform
errors may mean that busy subjects were rehearsing the number more poorly than not-busy subjects on a variety of indices (e.g.,
and that this rehearsal was straining their resources. We faced memory for target's race, number of completions, and so on). In
this interpretative dilemma by establishing an a priori cutoff fact, they performed just as well as did not-busy subjects on
such that subjects who incorrectly reported four or more of the these indices and made few errors on the digit-rehearsal task
digits were considered to have made large errors and were ex- itself. Some authors have argued that substantial error rates on
cluded from the data set. Three of the 37 busy subjects were an overload task are necessary if one is to claim unequivocally
excluded by this criterion. Of the 32 busy subjects who re- that capacity was exceeded (e.g, Kantowitz, 1974). Thus, our
mained in the data set, 2 made small errors (i.e., they incorrectly data may be seen as suggesting that the busyness manipulation
reported one or two of the eight digits) and 30 made no errors. did not, in fact, usurp subjects' processing resources. Is it possi-
ble, for example, that busy subjects committed the number to
Analysis of Completions long-term memory prior to the experimental trials and subse-
quently made fewer stereotypic words forreasonsentirely unre-
We predicted that not-busy subjects would generate more ste- lated to resource deprivation?
reotypic completions when exposed to an Asian than a Cauca- We think not. Osborne and Gilbert (1990) showed that sub-
sian assistant, but that busy subjects would not. The number of jects who were given 20 s to memorize an eight-digit number
stereotypic completions generated by each subject was submit- responded more slowly to probes that occurred over the follow-
ted to a 2 (busyness: busy or not busy) X 2 (race: Caucasian or ing 2 min than did subjects who had not been asked to memo-
Asian) analysis of variance (ANOV\) that revealed only the pre- rize the number. This finding suggests that the typical subject
dicted Busyness X Race interaction, F(l, 64) = 4.52, p < .05. As does indeed rehearse the number (rather than merely store it in
Table 1 shows, not-busy subjects were more likely to generate long-term memory) and that thisrehearsaldoes usurp process-
stereotypic completions when exposed to an Asian than a Cau- ing resources. Nonetheless, our interpretation of the results of
casian assistant, Z(32) = 2.46, p < .05, but busy subjects were Experiment 1 would be more convincing if busy subjects had
not,f(32)<l. shown minor but ubiquitous errors on the digit-rehearsal task,
and future researchers should consider using more demanding
manipulations of load to avoid such interpretational difficul-
ties.
Table 1
1
Number of Stereotypic Completions in Experiment
Busyness Experiment 2
Assistant's race Not busy Busy
The Sufis teach that "If a pickpocket meets a holy man, he
will see only his pockets" (Dass, 1971, p. 10). The results of
Asian 3.82. 3.12,. Experiment 1 suggest that people can, in fact, be exposed to
Caucasian 3.00,, 3.24, others about whom they have stereotypic beliefs, and yet show
Difference 0.82 -0.12 no evidence of stereotype activation. Apparently, busy subjects
Note. N= 17 in each cell. Means that do not share a common sub- who encountered an Asian card turner saw only a card turner,
script differ with p < .05. Means that share a common subscript do not and not an Asian.
differ, all ft < 1 . Atfirstblush, thisfindingmay appear to contradict the doc-
 512 DANIEL T. GILBERT AND I GREGORY HIXON

trine of cognitive economy—namely, that stereotypes ease the each letter (R through U), and the delay between letters (1,000 to 3,000
task of understanding others and should therefore be most ms) were randomly determined by the roll of a die prior to the experi-
readily employed by those in greatest need of easing. In fact, the ment.
contradiction disappears if one distinguishes between the acti- The subject was instructed to use a hand-held "clicker" with analog
vation and application of information. Although "activation readout to count the number of times the letter T was immediately
increases the likelihood of a constructs being used in subse- followed by the letter U. Letters were presented for the entire duration
of the assistant's description (approximately 2.6 min), and during that
quent judgments" (Higgins, 1989, p. 78), it does not mandate time the target sequence (i.e., T-Ihen-U) occurred on six occasions.
such use, nor does it determine the precise nature of its use. It is Visual search tasks such as this have been used with excellentresultsto
possible for activated information to exert no effect on subse- deprive subjects of processing resources (e.g., Gilbert & Krull, 1988,
quent judgments or to have a variety of different effects (e.g., Experiments 1 and 2). Theremainingsubjects were assigned to be not
contrast versus assimilation; see Lombardi, Higgins, & Bargh, busy during the application phase. These subjects did not perform the
1987; Martin, Seta, & Crelia, 1990; Newman & Uleman, 1990). visual search task, and the video screen remained black as they lis-
If activation involves finding a tool in the cognitive toolbox, tened to the assistant describe a typical day in her life.
then application involves using that tool to ease the processing Thus, about one quarter of the subjects in the experiment were as-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

of information. We suggest that whereas busyness may decrease signed to the always-busy condition (they performed a digit rehearsal
This document is copyrighted by the American Psychological Association or one of its allied publishers.

the likelihood of locating the tool (activation), it may increase task during the activation phase and a visual search task during the
the likelihood of using the tool once it has been found (applica- application phase), one quarter were assigned to the never-busy condi-
tion (they performed neither the digit rehearsal task nor the visual
tion). Experiment 2 was an attempt to demonstrate these ef-
search task), one quarter were assigned to the early-busy condition
fects. (they performed a digit rehearsal task during the activation phase but
not a visual search task during the application phase), and one quarter
were assigned to the late-busy condition (they did not perform the digit
Method rehearsal task during the activation phase, but they did perform the
Overview visual search task during the application phase).1

Subjects performed a word-fragment completion test while being

exposed to an Asian or Caucasian assistant. Some subjects rehearsed Dependent Measures
an eight-digit number while they performed the test and others did not. All subjects listened to a female narrator (ostensibly the card-turning
Subjects then attempted to form an impression of the assistant as she assistant) describe a rather mundane series of events (e.g., grocery shop-
described a day in her life. Some subjects performed a visual search ping, meeting her sister's friends, going to a party, doing school work,
task while they listened to this description and others did not. Finally, and so on). After hearing the description, subjects were given 90 s to
all subjects reported their impressions of the assistant. complete their ratings of the assistant. Subjects rated the assistant on
nine trait dimensions: timid, intelligent, calm, composed, aloof, soci-
Subjects able, friendly, happy, and conversational. These trait terms were either
synonyms or antonyms of the eight stereotypic concepts that were gen-
One hundred eleven female students at the University of Texas erated by at least 33% of the pretest subjects in Experiment 1. The first
participated to fulfill a requirement in their introductory psychology five of these words were synonyms (and were thus considered typical of
course. Only Caucasian students who were native speakers of English Asian-American students) and the last four were antonyms (and were
were eligible to participate. thus considered atypical of Asian-American students). Each trait was
presented on an 11-point scale anchored at the end points with the
phrases not a very X person and a very X person, where X was replaced
Procedure by one of the nine trait adjectives.
The experiment was composed of two phases: an initial activation Next, subjects were given a 13-item recognition memory test. Of
phase (in which subjects' stereotypes about Asians either were or were these items, 5 were statements taken verbatim from the assistant's de-
not activated) and a subsequent application phase (in which subjects scription (e.g., "My sister has urged me to talk to my landlord, but I
were given an opportunity to use their stereotypes during impression don't want to"). The remaining 8 items were foils that were created by
formation). The procedures for the activation phase were identical to altering key phrases contained in the assistant's actual statements. For
those in Experiment 1. In brief, subjects performed a word-fragment example, the assistant actually said, "I live by myself in a small campus
completion test while either rehearsing or notrehearsingan eight-digit apartment," and from that statement a foil item was created that read,
number. The card-turning assistant was, once again, either Asian or "I live in a private campus dormitory." Every effort was made to create
Caucasian. foils that could be easily confused with statements that the assistant
had actually made; thus, we did not include statements that were unmis-
After the activation phase was completed, subjects were told that
they would hear an audiorecording of the card-turning assistant as she
described a typical day in her life, and that their job was to form an
1
impression of the assistant on the basis of these events. Approximately Because of an anticipated shortage of subjects in one semester, the
half the subjects were assigned to be busy during the application phase. critical conditions (always-busy and late-busy) were run prior to the
These subjects were told that while they listened to the assistant de- others (never-busy and early-busy). It is very important to note, how-
scribe a typical day in her life, one of four letters (R, S, T.orU) would ever, that our predictions involve interactive effects within these condi-
appear at a random location on the screen. The screen was divided into tions rather than main effects across them; as such, none of the pre-
a 6 X 6 invisible grid, and a letter appeared in one of the 36 sectors on a dicted effects can be accounted for solely by this breach of random
black background. The letter remained on the screen for 500 ms, disap- assignment. In addition, the obvious absence of main effects across
peared, and was followed between 1,000 and 3,000 ms later by another conditions is an excellent indicator of the equality of the previous and
letter. The sector in which each letter appeared (1 to 36), the identity of subsequently run groups.
 ACTIVATION AND APPLICATION OF STEREOTYPES 513
takable departures from the assistant's actual statements (e.g., "I killed who were not busy during the activation phase were more likely
a 79-pound lobster with avant garde poetry") and that would therefore to generate stereotypic completions when exposed to an Asian
have been easily identified as foils. Subjects never saw both an actual than a Caucasian assistant, f(48) = 3.38, p < .01. However, sub-
statement and a foil taken from the same statement. Finally, subjects jects who were busy during the activation phase generated
were asked to recall the assistant's race and the color of the ink in which equivalent numbers of stereotypic completions regardless of the
the fragments had been printed, and then were probed for suspicion,
race of the assistant to whom they were exposed, t < 1.
debriefed, and dismissed.
As in Experiment 1, busyness itself did not alter general task
Results and Discussion performance in any meaningful way. Subjects who were busy
and who were not busy during the activation phase generated
Error Rates and Excluded Data
equal numbers of correct completions across all trials (Ms =
We used the criterion used in Experiment 1 to exclude sub- 37.73 and 38.14, respectively), t < 1, and across critical trials
jects who made errors when reporting the eight-digit number. (Ms = 10.44 and 10.88, respectively), 03) = 1.13, p > .27. Busy
Eight of the 111 subjects made large errors (i.e., they incorrectly and not-busy subjects were equally adept at recalling the assis-
reported four or more of the eight digits), and these subjects tant's race (89% and 80% correct recall, respectively), x2(l) =
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

were excluded from the data set. Thirteen subjects made small 1.41, p = .24, and at recalling the color in which the fragments
This document is copyrighted by the American Psychological Association or one of its allied publishers.

errors (i.e., they incorrectly reported between one and two of the were printed (73% and 60% correct recall, respectively), x2(l) =
eight digits), and these subjects were retained in the data set. 1.88, p = .17. Finally, although busy subjects did make margin-
The remaining subjects made no errors on this task. ally more common words across all trials than did not-busy
In addition, two subjects made large errors when reporting subjects (Ms = 76.93 and 71.70, respectively), t(93) = 1.59, p =
the number of appearances of 7'-then-f/(i.e., they were incor- .11, the two groups made equally common words on the critical
rect by more than one of the six appearances), and they were trials (Ms = 75.49 and 75.64, respectively), t < 1.
excluded from the data set. Three subjects made small errors
when reporting the number of appearances of T-xhsn-U, and
these subjects were retained in the data set. The remaining Application Phase
subjects made no errors on this task. Subjects' ratings of the assistant on the nine trait dimensions
Finally, five subjects confessed to having misunderstood the were averaged to create a stereotypic perception index on which
instructions and one subject identified herself as a nonnative larger values indicated a perception of the assistant as possess-
speaker of English. These subjects were excluded from the data ing more stereotypically Asian traits (e.g, greater timidity, less
set. In summary, the data from 16 of 111 subjects were excluded sociability, and so on). Scores on this index were submitted to a 2
from all analyses. (activation: busy or not busy) X 2 (application: busy or not busy)
X 2 (race: Caucasian or Asian) ANOV\ that revealed only the
Activation Phase
predicted three-way interaction, F(l, 87) = 4.12, p < .05.
As in Experiment 1, busy and not-busy subjects were exposed As Table 3 shows, subjects who were not busy during the
to either an Asian or Caucasian assistant while they performed activation phase but who were busy during the application
a word-fragment completion test. We predicted that, as in Ex- phase (i.e, late-busy subjects) made more stereotypic ratings of
periment 1, not-busy subjects would generate more stereotypic the Asian assistant than of the Caucasian assistant, /(22) = 4.41,
completions when exposed to an Asian than a Caucasian assis- p < .001, and this was the only one of the four groups to do so
tant, but that busy subjects would not. (similar comparisons for the always-busy, never-busy, and early-
The number of stereotypic completions generated by each busy groups did not approach significance, all ts < 1). In other
subject was submitted to a 2 (activation: busy or not busy) X 2 words, busyness during the application phase increased sub-
(race: Caucasian or Asian) ANOV\ for unequal Nih&X revealed jects' tendency to view the Asian assistant in stereotypic terms,
a main effect of race, F(\, 91) = 5.36, p = .02, and the predicted but only if the corresponding stereotypes had been activated in
Activation X Race interaction, F(l, 91) = 7.48, p = .007. As the first phase. Busyness during the application phase had no
Table 2 shows, Experiment 1 was clearly replicated: Subjects discernible effect on subjects whose stereotypes were not acti-
vated earlier (i.e., always-busy subjects), and stereotype activa-
tion had no discernible effect on subjects who were not busy
Table 2 during the application phase (i.e., early-busy subjects). All sub-
Number of Stereotypic Completions in Experiment 2 jects showed superb memory for the assistant's description, al-
Busyness during activation phase
though, as might be expected, subjects who were busy during
the application phase showed somewhat poorer memory (mean
Not busy Busy d' = 2.55) than did subjects who were not (mean d' = 3.97),
*(93)= 6.55, p<. 001.2
No. of No. of
Assistant's race completions Cell n completions Celln
2
It is worth noting that of subjects who were not busy during the
Asian 3.71, 24 3.09b 23
Caucasian 2.88b 26 3.18b 22 activation phase, those who were subsequently assigned to the late-
Difference 0.83 -0.09 busy condition and those who were subsequently assigned to the never-
busy condition did not differ in terms of the number of stereotypic
Note. Means that do not share a common subscript differ with p < words produced during the word-fragment completion test. That is,
.05. Means that share a common subscript do not differ, all ps > .22. the (about to be) late-busy and the (about to be) never-busy subjects
 514 DANIEL T. GILBERT AND J. GREGORY HIXON

Table 3
Ratings of Assistant on Stereotypic Perception Index in Experiment 2
Always busy Early busy Late busy Never busy

Condition Rating n Rating n Rating n Rating n

Assistant's race
Asian 6.15 13 6.59 10 7.30 11 6.36 13
Caucasian 6.35 12 6.77 10 6.07 13 6.42 13
Difference -0.20 -0.18 1.23* -0.06

Note. Early busy subjects were busy during the activation phase but not during the application phase; late
busy subjects were busy during the application phase but not during the activation phase; always busy
subjects were busy during both phases; and never busy subjects were busy during neither phase. Larger
values indicate perceptions of the assistant as possessing more stereotypically Asian traits. Only differ-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

ences marked with an asterisk are significant with p < .05. For all other differences, t < 1. The late
This document is copyrighted by the American Psychological Association or one of its allied publishers.

busy/Asian cell (M = 7.30) differs from all other cells with p < .05, except that it differs from the early
busy/Caucasian cell (M = 6.77) with p = .11.

It is interesting to consider the possibility that the three omy suggests that stereotypes are applied in order to ease the
groups of subjects who showed no evidence of stereotype appli- burden of information processing, and thus one should not
cation may have done so for very different reasons. The early- expect to observe such application when the information-pro-
busy and always-busy subjects should not have applied their cessing task is not particularly taxing. Although we cannot de-
stereotypes because busyness during the first phase should termine which of these accounts is correct in the present case,
have kept those stereotypes from being activated. As such, we one bit of evidence is suggestive. Two groups of subjects had
would not expect (and we did not observe) stereotypic re- their stereotypes of Asian-Americans activated (i.e., the never-
sponses from these subjects. Never-busy subjects, on the other busy and late-busy subjects who were exposed to an Asian assis-
hand, should have had their stereotypes activated, and evidence tant). Of those activated subjects, those who were busy during
from the word-fragment completion test suggests that they did. the application phase (i.e., the late-busy subjects) showed a mar-
Why, then, did these subjects show no evidence of stereotype ginally reliable correlation between the degree of stereotype
application? activation and application, r(9) = .45, p = .08. Those activated
There are at least two explanations for this finding: The be- subjects who were not busy during the application phase (i.e.,
havioral suppression account suggests that stereotype applica- the never-busy subjects) showed no such correlation, r(l 1) = .05,
tion did, in fact, occur, and the individuation account suggests p = .44. If never-busy subjects strategically adjusted their rat-
that it did not. First, Devine (1989) has argued that both the ings away from the stereotypic pole of the rating scale, one
activation and the application of stereotypes are automatic, but would expect the mean of these ratings to be lowered—but one
that overt responses are not. Although there is now reason to might also expect the correlation between activation and appli-
believe that no mental process is unconditionally automatic cation to be maintained. And it is not. This one shred of evi-
(Bargh, 1989), Devine's point about the greater controllability dence, then, suggests that never-busy subjects may not have
of stereotypic behavior is an important one. Individuals may applied their activated stereotypes because they did not need to
suppress stereotypic responses either because they consider do so. Ultimately, of course, we cannot know whether the never-
such responses immoral ("It would be wrong to say that the busy subjects failed to achieve stereotypic impressions (the in-
woman is unsociable just because she's Asian") or because they dividuation account) or simply failed to announce them (the
wish to manage their impressions ("If I say the assistant is unso- behavioral suppression account). This is a perdurable dilemma
ciable, the experimenter will think I'm a bigot"). This suggests that even the most inventive students of stereotyping have been
that never-busy subjects may well have achieved stereotypic unable to resolve (see Jones & Sigall, 1971).
impressions of the assistant, but that the lack of busyness dur-
ing the application phase allowed them to adjust their re-
sponses so as not to be or appear prejudiced. General Discussion
On the other hand, it is possible that the activated stereotypes Mental experience seems inexorably to involve the admixture
of never-busy subjects did not affect their judgments because of old and new information. People rely so strongly on prior
the surfeit of cognitive resources during the application phase information to ease the burden of ongoing perception that
enabled them to individuate (rather than stereotype) the assis- some of the greatest thinkers (e.g., Kant, 1781/1965) have
tant (Fiske & Neuberg, 1990). The principle of cognitive econ- doubted whether perception could occur otherwise. As Durant
(1926, p. 272) wrote in characterizing this idealist position,
"The world as we know it is a construction, afinishedproduct,
made equally few stereotypic words when exposed to a Caucasian as- almost—one might say—a manufactured article, to which the
sistant (Ms = 2.92 and 2.85, respectively, t < 1) and equally many stereo- mind contributes as much by its moulding forms as the thing
typic words when exposed to an Asian assistant (Ms = 3.82 and 3.62, contributes by its stimuli." With few exceptions (e.g., Gibson,
respectively, t < 1). 1979), most modern psychologists have embraced this con-
 ACTIVATION AND APPLICATION OF STEREOTYPES 515

structivist perspective (Neisser, 1976) and have attempted to actions at the same time that one draws inferences about others
articulate the circumstances under which percepts and judg- (Gilbert, Krull, & Pelham, 1988). As such, social interaction
ments will be more strongly determined by stimulus properties may cause interactants to become cognitively busy and thus
or by the mind's "moulding forms." reduce the likelihood that their stereotypes about each other
As the quotation that opened this article suggests, psycholo- will be activated. A faithful churchgoer who meets a newly
gists and pundits have generally agreed that it takes more work arrived Hispanic minister may not experience activation of his
to individuate a stimulus than to construe it in terms of well- or her beliefs about Hispanics, simply because the social de-
worn knowledge (cf. Britton & Tesser, 1982). Preexisting knowl- mands of the formal encounter may usurp resources that are
edge is meant to save people "the trouble of thinking" about necessary for the activation of those concepts. Of course, this
each new stimulus they encounter, and thus theorists have been does not mean that social interaction is a panacea for prejudice.
quick to assume that circumstances that preclude individuating Our results suggest that if stereotypes are activated prior to a
thought must increase the person's reliance on preexisting resource-consuming social interaction ("Let me take you over
knowledge. To the extent that stereotypes function like other and introduce you to Father Gonzales"), then the interactants
forms of preexisting knowledge (e.g., beliefs, constructs, catego- may be especially likely to view each other in stereotypic terms.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

In short, the timing of the onset of busyness would appear to be

This document is copyrighted by the American Psychological Association or one of its allied publishers.

ries, schemas, and scripts), our findings suggest that this as-
sumption is correct—but only in part. People aw more likely to critical in determining if and when social interaction will ame-
rely on activated stereotypes when conscious deliberation be- liorate or exacerbate stereotypy.
comes difficult, but the very conditions that interfere with
conscious deliberation may also interfere with the activation of The Inevitability of Stereotypy
the stereotypes. The metaphorical irony is that just when one
needs one's tool most, it may be hardest to find. Our findings suggest that the activation of racial stereotypes
is not an unconditionally automatic consequence of exposure to
a person; rather, a perceiver must have adequate processing
Social Interaction and Stereotypy resources for such stereotypes to be activated. Thisfindingbe-
To speak of stereotypes as tools is to overlook the fact that lies strong claims such as those of Brewer (1989, pp. 5-6), who
although they may ease the burden of perception, they may also suggested:
reduce its accuracy. Stereotypic beliefs about women's roles, for The mere presentation ofa stimulus person activatescertain classi-
example, may enable one to see correctly that a woman in a fication processes . . . that occur automatically and without
dark room is threading a needle rather than tying afishinglure, conscious intent.. . . Whenever a novel social object is encoun-
but they may also cause one to mistakenly assume that her goal tered, an initial identification stage is postulated to precede any
is embroidery rather than cardiac surgery. Although stereotypes conscious, goal-driven information processing. . .[and this] pro-
cess is one of "placing" the individual social object along well-es-
are psychologically fundamental, they may also be socially per- tablished stimulus dimensions such as gender, age, and skin color.
nicious, and psychologists have long searched for ways to re-
solve this dilemma. One remedy has been to encourage people This claim has considerable intuitive appeal. How can one see a
to spend the time and effort necessary to individuate others young Asian woman and yet fail to categorize her instantly as
rather than allowing their preconceptions to dominate their such? Our studies do not deny that such rapid categorizations
judgments (see especially Fiske, 1989). The second remedy has can and often do happen; they simply deny that they must
been to increase the accuracy of the preconceptions upon happen. Our busy subjects did not experience activation of ra-
which people rely. cial stereotypes about an Asian assistant to whom they were
One oft-prescribed means of accomplishing both ends is to exposed, and although these subjects were able to recall the
foster interaction between persons and the objects of their ste- assistant's race after the experiment was over, they apparently
reotypes (e.g., Aronson, Blaney, Stephan, Sikes, & Snapp, 1978; did not categorize her in terms of race during the experiment.
Brewer & Miller, 1984). When people interact with members of And why should they have? During the activation phase, sub-
out-groups, two things may happen. First, they may gain infor- jects had no reason to form any kind of impression of the Asian
mation that increases the accuracy of their preconceptions: An assistant, whose minimal role was more akin to that of furni-
unmusical Black, a generous Jew, or a sober Irishman may, at ture than to that of an interaction partner. We suspect that, as
the very least, undermine the certainty with which people em- Brewer claims, the mere presentation of a stimulus person does
brace such racial caricatures (Taylor, 1981, pp. 102-103; Weber initiate certain classification processes; however, we do not be-
& Crocker, 1983). Second, social interaction raises the practical lieve that these classifications are inevitably along certain di-
cost of inaccurate beliefs (Swann, 1984). It is one thing to mis- mensions or that they are unaffected by the perceiver's goals. In
construe a famous Black politician or a feminist opinion leader, fact, busyness may exert its effect on stereotype activation by
and quite another to misconstrue one's dentist, student, or causing subjects to classify others only along those dimensions
daughter-in-law. In other words, people are more accountable that are directly relevant to their current information-process-
for (and thus may craft more carefully) judgments about those ing goals: When the person's race is wholly inconsequential (as it
with whom they have true and enduring commerce (Tetlock & was in our studies), busy perceivers may not have the "luxury"
Kim, 1987). of activating preexisting information that is, in fact, irrelevant
Our studies suggest a third way in which social interaction to their concerns.
may affect the use of stereotypes. Social interaction is a com- Issues such as these may be obscured by studies that do not
plex business in which one must consciously regulate one's own expose subjects to stereotype objects, but rather, to words that
 516 DANIEL T. GILBERT AND J. GREGORY HIXON

represent those objects. For example, Devine (1989, p. 5) con- Aronson, E., Blaney, N., Stephan, C, Sikes, X, & Snapp, M. (1978). The
cluded that a "stereotype is automatically activated in the pres- jigsaw classroom. Beverly Hills, CA: Sage Publications.
ence of a member (or some symbolic equivalent) of the stereo- Bargh, J. A. (1989). Conditional automaticity: Varieties of automatic
typed group," and supported this claim with data from several influence in social perception and cognition. In J. S. Uleman & J. A.
elegant experiments that demonstrate the ease with which Bargh (Eds.), Unintended thought (pp. 3-51). New York: Guilford.
words such as nigger can activate stereotypes about Blacks and Bargh, J. A., & Pietromonaco, P. (1982). Automatic information pro-
cessing and social perception: The influence of trait information
thereby affect judgments about a target whose race is not men-
presented outside of conscious awareness on impression formation.
tioned. Such studies are in keeping with social psychology's Journal ofPersonality and Social Psychology, 43, 437-449.
tradition of treating words as the symbolic equivalents of per- Bassili, J. N., & Smith, M. C. (1986). On the spontaneity of trait attribu-
sons—a tradition so widespread that some topics in social psy- tion: Converging evidence for the role of cognitive strategy. Journal
chology are denned almost entirely by subjects' reactions to of Personality and Social Psychology, 50, 239-245.
trait adjectives (e.g., person memory; see Hastie et al., 1980). Bodenhausen, G. Y(1990). Stereotypes as judgmental heuristics: Evi-
This methodological approach is quite reasonable when the dence of circadian variations in discrimination. Psychological
assumption of symbolic equivalence is true. But there may be Science, 1, 319-322.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

times when the assumption is unfounded, and thus times when Bodenhausen, G. V, & Lichtenstein, M. (1987). Social stereotypes and
This document is copyrighted by the American Psychological Association or one of its allied publishers.

the study of reactions to words may paint a misleading portrait information-processing strategies: The impact of task complexity.
of psychological processes. Journal ofPersonality and Social Psychology, 52, 871-880.
For example, it may be virtually impossible for a literate adult Bodenhausen, G. V, & Wyer, R. S. (1985). Effects of stereotypes on
decision-making and information-processing strategies. Journal of
to read the phrase Black fireman without experiencing activa-
Personality and Social Psychology, 48, 267-282.
tion of both the racial and occupational constructs (Logan, Brewer, M. B. (1989). A dual process model of impression formation.
1980; Stroop, 1935; cf. Kahneman & Treisman, 1984), and, in In R. S. Wyer & T. K. Srull (Eds.), Advances in social cognition (Vol. 1,
some sense, to do otherwise would be to fail to understand what pp. 1-36). Hillsdale, NJ: Erlbaum.
one has read. Nonetheless, it may be entirely possible for a Brewer, M. B., & Miller, N. (1984). Beyond the contact hypothesis:
literate adult to encounter a Black fireman and, given the ap- Theoretical perspectives on desegregation. In M. B. Brewer & N.
propriate information-processing goal (e.g., to find quickly Miller (Eds.), Groups in contact: The psychology ofdesegregation (pp.
someone who will enter a burning building to save a child), to 281-302). San Diego, CA: Academic Press.
construe the Blackfiremanonly in terms of his occupation (see Britton, B. K., & Tesser, A. (1982). Effects of prior knowledge on use of
Taylor, 1981). Because words and phrases contain implicit cate- cognitive capacity in three complex cognitive tasks. Journal of Verbal
gorizations of the objects they describe, they effectively demand Learning and"Verbal Behavior, 21, 421-436.
that associates of these categories be activated. As Zarate and Dass, R. (1971). Be here now. New York: Crown.
Smith (1990, p. 162) noted: "Linguistic descriptions (e.g., 'a Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and
controlled components. Journal of Personality and Social Psychol-
black person*) force a single categorization, in contrast to a real
ogy, 56, 5-18.
person, who is not only black but (perhaps) young, male, well- Dovidio, J. F, Evans, N. E., & Tyler, R. B. (1986). Racial stereotypes:
dressed, tall, speaking with a Southern accent, and so on." Our The contents of their cognitive representations. Journal of Experi-
subjects encountered someone who was an Asian and a card mental Social Psychology, 22, 22-37.
turner, and yet they showed no evidence of construing her as an Durant, W (1926). The story ofphilosophy. New "fork: Simon & Schus-
Asian. We would be very surprised to find similar results had ter.
subjects instead read the phrase Asian card turner. Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C, & Kardes, F. R.
After appraising the state of research in social cognition, Za- (1986). On the automatic activation of attitudes. Journal of Personal-
jonc (1980b, p. 192) offered this warning: "Because we cannot ity and Social Psychology, 50, 229-238.
assume a one-to-one correspondence between language and Fiske, S. T. (1989). Examining the role of intent: Toward understanding
its role in stereotyping and prejudice. In J. S. Uleman & J. A. Bargh
reality, we may not take it for granted that the same principles of (Eds.), Unintended thought (pp. 253-286). New York: Guilford.
social perception will be generated by studying words as by Fiske, S. T., & Neuberg, S. L. (1990). A continuum model of impression
studying the actual social objects for which these words stand." formation, from category-based to individuating processes: Influ-
It is in this spirit that we offer a revision of the Sufi teaching: ences of information and motivation on attention and interpreta-
When a pickpocket reads the words holy man, he will probably tion. In M. P. Zanna (Ed.), Advances in experimental social psychol-
think of a great deal more than pockets—and in so doing, he ogy (Vol. 23, pp. 1-74). San Diego, CA: Academic Press.
will reveal little about how pickpockets construe holy men in Fiske, S. T, & Pavelchak, M. (1986). Category-based versus piecemeal-
their day-to-day lives. If we really want to know how persons based affective responses: Developments in schema-triggered affect.
think about persons, we may have to introduce our subjects to In R. M. Sorrentio & E. T. Higgins (Eds), Handbook of motivation
some. and cognition (pp. 167-202). New York: Guilford.
Gibson, J. J. (1979). The ecological approach to visual perception. Bos-
ton: Houghton Mifflin.
References Gilbert, D. T, & Krull, D. S. (1988). Seeing less and knowing more: The
benefits of perceptual ignorance. Journal of Personality and Social
Allport, G. W (1954). The nature ofprejudice. New \brk: Addison Wes- Psychology, 54,193-202.
ley. Gilbert, D. T, Krull, D. S., & Pelham, B. W (1988). Of thoughts unspo-
Andersen, S. M., Klatzky, R. L., & Murray, J. (1990). Traits and social ken: Social inference and the self-regulation of behavior. Journal of
stereotypes: Efficiency differences in social information processing. Personality and Social Psychology, 55, 685-694.
Journal of Personality and Social Psychology, 59,192-201. Gilbert, D. T, & Osborne, R. E. (1989). Thinking backward: Some
 ACTIVATION AND APPLICATION OF STEREOTYPES 517
curable and incurable consequences of cognitive busyness. Journal Morley, J. (1886). Critical miscellanies (Vol. 1). London: MacMillan.
ofPersonality and Social Psychology, 57, 940-949. Neisser, U. (1976). Cognition and reality. San Francisco: Freeman.
Gilbert, D. T., Pelham, B. W, & Krull, D. S. (1988). On cognitive busy- Newman, L. S., & Uleman, J. S. (1990). Assimilation and contrast ef-
ness: When person perceivers meet persons perceived. Journal of fects in spontaneous trait inference. Personality and Social Psychol-
Personality and Social Psychology, 54, 733-740. ogy Bulletin, 16, 224-240.
Hamilton, D. L. (Ed.). (1981). Cognitive processes in stereotyping and Osborne, R. E., & Gilbert, D. T. (1990). The preoccupational hazards of
intergroup behavior. Hillsdale, NJ: Erlbaum. social life. Unpublished manuscript, University of Texas at Austin.
Hamilton, D. L., & Trolier, T. K. (1986). Stereotypes and stereotyping: Pratto, E, & Bargh, J. A. (1991). Stereotyping based on apparently
An overview of the cognitive approach. In J. F. Dovidio & S. L. individuating information: Trait and global components of sex ste-
Gaertner(Eds.), Prejudice, discrimination, and racism (pp. 127-163). reotypes under attention overload. Journal of Experimental Social
San Diego, CA: Academic Press. Psychology, 27, 26-47.
Hastie, R., Ostrom, T. M., Ebbesen, E. B., Wyer, R. S., Hamilton, D. L., Rosch, E. (1978). Principles of categorization. In E. Rosch & B. B.
& Carlston, D. E. (Eds.). (1980). Person memory: The cognitive basis of Lloyd (Eds.), Cognition and categorization (pp. 27-48). Hillsdale, NJ:
social perception. Hillsdale, NJ: Erlbaum. Erlbaum.
Higgins, E. T. (1989). Knowledge accessibility and activation: Subjectiv- Rothbart, M., Fulero, S., Jensen, C, Howard, J., & Birrell, P. (1978).
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

ity and suffering from unconscious sources. In J. S. Uleman & J. A. From individual to group impressions: Availability heuristics in ste-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Bargh (Eds.), Unintended thought (pp. 75-123). New York: Guilford. reotype formation. Journal of Experimental Social Psychology, 14,
Jones, E. E., & Sigall, H. (1971). The bogus pipeline: A new paradigm 237-255.
for measuring affect and attitude. Psychological Bulletin, 76, 349- Stroop, J. R. (1935). Studies of interference in serial verbal reactions.
364. Journal ofExperimental Psychology, 18, 643-662.
Kahneman, D., & Triesman, A. (1984). Changing views of attention Swann, W. B., Jr. (1984). Quest for accuracy in person perception: A
and automaticity. In R. Parasuraman & D. R. Davies (Eds.), Varieties matter of pragmatics. Psychological Review, 91, 451-411.
ofattention (pp. 29-61). San Diego, CA: Academic Press. Swann, W B., Jr., Hixon, J. G., Stein-Seroussi, A., & Gilbert, D. T.
(1990). The fleeting gleam of praise: Psychological processes under-
Kant, I. (1965). Critique of pure reason (N. K. Smith, Trans.). New York: lying reactions to self-relevant information. Journal of Personality
St. Martin's Press. (Original work published 1781) and Social Psychology, 59,17-26.
Kantowitz, B. H. (1974). Double stimulation. In B. H. Kantowitz (Ed.), Taylor, S. E. (1981). A categorization approach to stereotyping. In D. L.
Human information processing: Tutorials in performance and cogni- Hamilton (Ed.), Cognitive processes in stereotyping and intergroup
tion (pp. 83-132). Hillsdale, NJ: Erlbaum. behavior (pp. 87-114). Hillsdale, NJ: Erlbaum.
Kucera, H., & Francis, W. N. (1967). Computational analysis ofpresent- Tetlock, P. E., & Kim, J. I. (1987). Accountability and judgment pro-
day American English. Providence, RI: Brown University Press. cesses in a personality prediction task. Journal of Personality and
LaBerge, E. J., & Samuels, S. J. (1974). Toward a theory of automatic Social Psychology, 52, 700-709.
information processing in reading. Cognitive Psychology, 6, 293- Tulving, E., Schacter, D. L., & Stark, H. A. (1982). Priming effects in
323. word-fragment completion are independent ofrecognitionmemory.
Lewicki, P. (1985). Nonconscious biasing effects of single instances on Journal ofExperimental Psychology: Learning, Memory, and Cogni-
subsequent judgments. Journal ofPersonality and Social Psychology, tion, 8, 336-342.
48, 563-574. Weber, R., & Crocker, J. (1983). Cognitive processes in the revision of
Locksley, A., Hepburn, C, & Ortiz, V (1982). Social stereotypes and stereotypic beliefs. Journal ofPersonality and Social Psychology, 45,
judgments of individuals: An instance of the base-rate fallacy. Jour- 961-977.
nal of Experimental Psychology, 18, 23-42. Zajonc, R. B. (1980a). Thinking and feeling: Preferences need no infer-
Logan, G. D. (1980). Attention and automaticity in Stroop and priming ences. American Psychologist, 35,151-175.
tasks: Theory and data. Cognitive Psychology, 12, 523-553. Zajonc, R. B. (1980b). Cognition and social cognition: A historical
Lombardi, W. J., Higgins, E. T, & Bargh, J. A. (1987). The role of perspective. In L. Festinger (Ed.), Retrospections on social psychol-
consciousness in priming effects on categorization: Assimilation ogy(pp.180-204). New York: Oxford University Press.
versus contrast as a function ofawareness of the priming task. Person- Zarate, M. A., & Smith, E. R. (1990). Person categorization and stereo-
ality and Social Psychology Bulletin, 13, 411-429. typing. Social Cognition, 8,161-185.
Martin, L. L., Seta, J. J., & Crelia, R. A. (1990). Assimilation and con-
trast as a function of people's willingness and ability to expend effort Received February 7,1990
in forming an impression. Journal ofPersonality and Social Psychol- Revision received October 25,1990
ogy, 59, 27-37. Accepted October 30,1990 •