Just Going Along: Nonconscious Priming
and Conformity to Social Pressure
Nicholas Epley and Thomas Gilovich
Received September 4, 1998; revised January 11, 1999; accepted January 19, 1999
This research investigates whether conformity can be elicited or suppressed by noncon-
scious priming. In Experiment 1, participants were primed for either conformity or
nonconformity using a scrambled sentences task and later placed into a conformity
situation. As predicted, participants primed with conformity expressed views that were
more similar to those of experimental confederates than did participants primed with
nonconformity. To investigate whether the inﬂuence of the primes was symmetric,
Experiment 2 included a neutral prime condition. Participants primed with conformity
again tended to conform more than those in the other two groups, but the nonconformity
primes did not induce participants to rebel against the group norm. Discussion focuses on
the asymmetry in the effectiveness of the conformity and nonconformity primes. r1999
I just went. My mind just went. And I wasn’t the only one that did it.A lot of other people did it. I just
killed . .. a lot of people were doing it, so I just followed suit. I just lost all sense of direction, of
purpose. I just started killing any kinda way I could kill. It just came. I didn’t know I had it in me.
—American soldier involved in the My Lai massacre (Bilton & Sim, 1971, p. 7)
Conformity pressure often produces a strong feeling of internal conﬂict. On one
hand, a person may feel that the opinions, attitudes, or behaviors of others are
incorrect or, even worse, immoral. On the other hand, nobody wants to be
ostracized by their friends or peers. Deciding what to do often requires consider-
able conscious deliberation, as one sizes up the costs of caving in versus the pain
This research was supported by Research Grant SBR9319558 from the National Science Founda-
tion. We are grateful to David Dunning, Dennis Reagan, and Michael Spivey for commenting on an
earlier draft of the manuscript and to Trey Billings, Kirsten Enterlin, Julie Gallagher, Andy Minor,
David Rosen, and Tyler Story for their assistance with the research. Portions of this research were
presented at the annual meeting of the American Psychological Society, 1997, Washington, DC.
Address correspondence and reprint requests to Nicholas Epley or Thomas Gilovich, Department of
Psychology, Cornell University, Uris Hall, Ithaca, NY 14853-7601. E-mail: email@example.com,
Journal of Experimental Social Psychology 35, 578–589 (1999)
Article ID jesp.1999.1390, available online at http://www.idealibrary.com on
Copyright r1999 by Academic Press
All rights of reproduction in any form reserved.
No. of Pages—12 First page no.—578 Last page no.—589
of eating lunch alone for weeks on end. Variations in numerous aspects of the
situation lead to variations in the conscious perception of these costs, which, in
turn, may inﬂuence behavior.
There are times, however, when behavior in conformity situations seems less
deliberate and less controlled. People sometimes ﬁnd themselves going along
with the group reﬂexively—without much thought and without knowing why.The
ﬁrst author, for example, recently found himself running energetically across
Manger Square in Bethlehem for the simple reason that everyone else in the
square was also running (as it turned out, because of a surprise visit by Yasser
Arafat). Similarly, the soldier involved in the My Lai massacre recalls his
seemingly automatic reaction to the behavior of his fellow soldiers. He did not
stop to consider his behavior, he just ‘‘followed suit.’’
The present research seeks to investigate a potentially important instigator of
this type of ‘‘mindless’’ or ‘‘automatic’’ conformity. Speciﬁcally, we investigate
whether priming the construct of conformity or nonconformity can lead people to
be more inclined to follow or resist the inﬂuence of others.
Automaticity in Social Cognition and Behavior
Psychologists have known for some time that many cognitive processes are
beyond conscious awareness or voluntary control (see Greenwald & Banaji, 1995,
for a review). Attitudes, for example, may be activated by the mere presence of an
attitude object (Bargh, Chaiken, Govender, & Pratto, 1992; Fazio, Sanbonmatsu,
Powell, & Kardes, 1986), and comparisons arise automatically when one is in the
presence of another person (Gilbert, Giesler, & Morris, 1995). Judgments about
ourselves and others are likewise open to inﬂuences beyond conscious control,
including the inﬂuence of information presented outside of awareness (Greenwald
& Banaji, 1995).
More recently, researchers have documented such effects not just on judgment,
but also on social behavior. In perhaps the most dramatic demonstration of this
effect, college students primed with words associated with the elderly walked
more slowly after leaving an experimental setting (Bargh, Chen, & Burrows,
1996). In another study by the same authors, white participants subliminally
presented with faces of AfricanAmericans expressed more hostility in response to
a computer failure than did those presented with white faces. Also, participants
who are exposed to words associated with hostility tend to behave in a more
hostile fashion (Carver, Ganellen, Froming, & Chambers, 1983; Herr, 1986). This
research suggests that social behavior can be activated automatically, seemingly
with no conscious thought or mediation.
We sought to expand on these ﬁndings by examining whether conformity—a
behavior of long-standing interest to social psychologists that is typically charac-
terized by conscious deliberation—can be inﬂuenced by analogous priming
manipulations. Can people be led to conform by priming them with words related
to conformity? Alternatively, can people be induced to rebel or deviate from a
group consensus by priming them with words related to nonconformity? We
conducted two studies to ﬁnd out.
Participants. Thirty-four Cornell University undergraduates volunteered to
participate for extra credit in their psychology courses.1
Materials. Participants were primed for conformity or nonconformity using a
scrambled sentences task presented in the guise of a psycholinguistics experiment
(Bargh et al., 1996; Srull & Wyer, 1979). Participants were presented with strings
of ﬁve scrambled words and asked to create a grammatically correct sentence
using four of the words in each string. For instance, the scrambled string ‘‘is kind
angry she very’’could be unscrambled to make ‘‘she is very kind’’or ‘‘she is very
Two different versions of the task, each containing 30 strings of scrambled
words, were created to prime participants either for conformity or nonconformity.
In 20 of these strings, one word was semantically related to either conformity or
nonconformity, depending on condition. To generate these priming words, 45
undergraduates in a psychology course at Cornell were asked to write down as
many words as they could think of that were related to conformity and nonconfor-
mity. Words mentioned by at least 3 participants were included in the task. Only
19 nonconformity words ﬁt this criterion, and thus one word (oppose) was
repeated. To maintain balance across the two conditions, only the 19 most
frequently mentioned conformity words were used as well, with one of the words
(respect) being repeated. The resulting words did not vary across conditions in
either length, t(36) ⫽1.23, p⬎.20, or frequency of usage (Carroll, Davies, &
Richman, 1971), t⬍1.
In the conformity version, the critical priming stimuli were adhere, agree,
comply, conform, copy, customary, emulate, follow, habitual, imitate, maintain,
mimic, obey, oblige, respect (presented twice), simulate, supportive, uniform, and
uphold. In the nonconformity version, the critical priming stimuli were challenge,
confront, counter, defy, deviate, differ, different, disagree, disobey, disrespect,
ignore, individual, independent, oppose (presented twice), opposite, rebel, refute,
reject, and unique. The four ﬁller words in each of the 20 critical strings were
unrelated to either conformity or nonconformity, as were all the words in the 10
neutral strings. The sentences were presented in a random order that was the same
for both conditions. The priming task was administered on a Macintosh LCII
computer using a program created with Hypercard.
Procedure. Each session included three confederates who were unaware of the
participant’s condition.All members of the ostensible group of four were told that
1In both Experiments 1 and 2, approximately 65% of participants were women. However, gender
had no effect on any of the dependent measures in either experiment and therefore is not discussed
580 EPLEY AND GILOVICH
the experiment concerned ‘‘psycholinguistic decision-making processes’’ and
would involve a short computer task. The experimenter told participants that the
session was actually a pilot test for an experiment to be run later in the semester.
Consequently, he was interested in receiving feedback regarding the experiment
and would be asking them a few short questions after they had ﬁnished.
They were then led to individual cubicles each equipped with a computer. From
just outside the cubicles, the experimenter gave everyone some general instruc-
tions regarding the computer program and told them to assemble in the hallway
when they were ﬁnished. The task required approximately 10 min to complete.
After several minutes, two of the confederates quietly left their cubicles and
waited in the hallway. The remaining confederate was instructed to wait in the
cubicle until the participant had ﬁnished and then to emerge 10 s later. Once all
four were in the hallway, the experimenter led everyone to another room
containing a large table and an overhead projector. The confederates always
entered ﬁrst and sat in previously assigned chairs, thereby assuring that the
participant invariably sat in the same location. After everyone was seated, the
experimenter reiterated that this was only a pilot test and that he would like some
feedback about the experiment.
The experimenter then asked two questions: ‘‘How interesting was this experi-
ment?’’ and ‘‘Overall, how much did you enjoy this experiment?’’2After each
question, two confederates immediately gave extremely positive verbal feedback
and then waited a moment to allow the participant time to respond, if he/she
desired. After a response or brief moment of silence, the experimenter told the
group that, as a social scientist, he was leery of open-ended responses and would
like a more concrete measure of their sentiments. A written version of each
question and a scale ranging from 1 (not at all) to 11 (extremely) was then
projected onto the wall. The experimenter requested that each person think about
his/her response to the question and then gestured around the room saying,
‘‘Maybe we could quickly go around the room and get your responses.’’ The
experimenter’s gesture made it clear that the confederates should respond ﬁrst and
the participant last. All confederates provided extremely positive responses
(M⫽9.67 across both questions). Given the nature of the sentence construction
task being evaluated, such responses were likely to be signiﬁcantly at variance
with the participant’s own attitudes.3After the participant provided his or her
ratings, the confederates were dismissed and the participant was debriefed.
2Participants were also asked ‘‘How difficult was this experiment?’’ However, responses to this
question were uncorrelated with the other two items in both studies (all p’s ⬎.2) and were thus
excluded from all analyses. In Experiment 1, internal reliability (Cronbach’s ␣) for the composite scale
averaging across the dependent measures rose from .68 to .81 when this item was removed. In
Experiment 2, internal reliability rose from .53 to .72. Consequently, the main conformity analyses in
both Experiment 1 and Experiment 2 are based on the two items cited in the text.
3Indeed, a separate group of control participants in Experiment 2 who were neither primed nor
exposed to any conformity pressure assigned the task a mean rating of 6.39 on the two scales.
Results and Discussion
To examine whether participants were aware of the priming manipulation, we
asked them during debrieﬁng to indicate whether, and how, the sentence construc-
tion task might have been related to the later feedback session. Nearly all
participants simply repeated the cover story in response to this probe, and none
indicated correctly how the two parts of the experiment were related. In addition,
participants were asked if they could remember any recurring themes among the
words used in the sentence construction task. Once again, no participant voiced
any awareness of the theme of conformity or nonconformity.
To create an overall conformity index, we averaged each participant’s ratings
on the two 11-point scales (␣⫽.81). As predicted, participants primed with
conformity rated the task more favorably (M⫽8.56) than did those primed with
nonconformity (M⫽7.41), separate variance t(32) ⫽2.08, p⬍.05. Stated
differently, the ratings made by participants in the conformity condition were
more similar to those voiced by the confederates (mean discrepancy ⫽1.11) than
were those made by participants in the nonconformity condition (mean discrep-
It is worth noting that although participants in the nonconformity condition
conformed less than their counterparts in the conformity condition, their mean
responses were nonetheless rather high (M⫽7.41). One would be hard pressed to
argue that participants in the nonconformity condition were particularly rebel-
lious. One ancillary ﬁnding that may help to explain these seemingly high
conformity scores is that the variance in the nonconformity group was signiﬁ-
cantly larger than that in the conformity group (Levene’sTest for Homogeneity of
Variances ⫽6.17, p⬍.05). Some of this difference doubtless results from the
ratings in the conformity group being closer to the upper limit of the scale and
therefore having less room for variability.This is unlikely to be the sole reason for
the difference, however, as the average was still rather far from the absolute
ceiling (8.56, with an upper limit of 11).
Another possible reason for this difference is that it may be difficult to prime
nonconformity without running the risk of simultaneously priming conformity.
Terms related to nonconformity are often—like nonconformity itself—coded
with reference to conformity, whereas the converse is not true for conformity. The
New Merriam–Webster Dictionary indicates that to ‘‘defy,’’ for example, is
deﬁned as ‘‘to refuse boldly to obey or to yield to’’ (Mish et al., 1989), and, of
course, ‘‘disagree’’and ‘‘disobey’’ explicitly denote the absence of agreement and
obedience, respectively. The common terms used to denote conformity, in
contrast, make no reference to the counter condition of nonconformity. To
‘‘obey,’’ for example, is deﬁned simply as ‘‘to follow the commands or guidance
of: behave obediently’’(Mish et al., 1989). Nonconformity is thus linguistically
‘‘marked’’ and conformity ‘‘unmarked,’’ and, like marked forms generally, it is
processed along with some reference to its unmarked counterpart (Clark & Clark,
1977). This suggests that the conformity primes may have been more successful—
more pure—than the nonconformity primes, which may have inadvertently also
582 EPLEY AND GILOVICH
primed conformity.This would explain both ancillary features of our data: (1) that
the ratings made by participants in the nonconformity condition, although
signiﬁcantly lower than those in the conformity condition, were still rather high in
absolute terms and (2) that the variability of the ratings in the nonconformity
condition was so large.
We conducted a second experiment to pursue this issue further. In particular, we
included a neutral priming condition to examine whether the words used to prime
nonconformity induce people to dissent as much as the words used to prime
conformity lead them to conform. If the nonconformity primes inadvertently
activate the construct of conformity for at least some (but not all) participants, the
mean responses of participants in the neutral and nonconformity groups should be
roughly comparable, with some participants in the nonconformity group rebelling
and others conforming. This should also generate greater variability in the
responses of participants in the nonconformity group compared to those in the
other two conditions. On the other hand, if the nonconformity primes successfully
activate pure nonconformity, then the responses of participants in the nonconfor-
mity condition should deviate from those of the confederates more than those of
respondents in the control group.
Experiment 2 was also designed to investigate several alternative interpreta-
tions of the ﬁndings from the ﬁrst experiment. First, the experimenter in
Experiment 1 was aware of each participant’s condition. We doubt, however, that
experimenter demand produced the observed effects because of the tightly
scripted interaction between the experimenter and participants and the more
salient demand—constant across conditions—emanating from the confederates
who were unaware of the condition. Nevertheless, to remove any doubt about this
issue we devised a procedure for Experiment 2 that would keep the experimenter
unaware of the participant’s condition.
Second, although no one indicated any awareness of the priming manipulation
during debrieﬁng, this does not guarantee a complete lack of awareness. Some
participants may have been consciously aware of the earlier primes when the
dependent measures were collected, but any such awareness may have faded by
the time of debrieﬁng. To provide a more rigorous test of awareness in Experi-
ment 2, a control condition was run in which participants were given the sentence
completion task and then, without the time-consuming and distracting procedure
of listening to the responses of the confederates, were asked whether they had
detected any recurring themes in the sentence completion materials.
Finally, control participants were asked to rate the experiment on the two
critical conformity measures, in the absence of conformity pressure. This served
to investigate the potential artifact that exposure to the conformity words may
have made the experiment more interesting and enjoyable.
Participants. The participants were 120 Cornell undergraduates who volun-
teered in exchange for extra credit in their psychology courses. Sixty participants
were run in one of three experimental conditions (conformity, nonconformity,
neutral), and 60 were run in the three corresponding control conditions.
Materials. All materials were the same as those in Experiment 1, with the
addition of the neutral priming condition. In this condition, the critical priming
words used in the conformity and nonconformity conditions were replaced by the
following: allergic, alternate, chase, excuse, imprint, meet, mobile, naive, ner-
vous, notion, orient, perplex, review, simplify, sociable, stretch, thirsty, transport,
vault, and wonder. These words did not differ from their counterparts in the other
two conditions in either length, both p’s ⬎.15, or frequency (Carroll, Davies, &
Richman, 1971), both p’s ⬎.35.
Procedure. The procedure was identical to Experiment 1 for the three experi-
mental conditions. In the control condition, participants were run in groups of one
to ﬁve and no confederates were used. After completing the computer task,
control participants were told that the experimenter would like to receive
feedback about the experiment and were handed a questionnaire containing the
two critical attitude questions.
A third item was included to measure awareness: ‘‘Many of the words included
in the sentences presented earlier on the computer were related to one another.
Can you think of any general themes or topics that most of the words in the
sentences were related to?’’After completing the questionnaire, the participants
were debriefed and dismissed.
Results and Discussion
There were three primary questions of interest in this experiment. First, what
was the direction of inﬂuence in the nonconformity condition? Were these
participants induced to rebel against the group or did the nonconformity primes
also inadvertently prime conformity? This was assessed by comparing the
behavior of participants in the nonconformity and neutral groups. Second, can the
predicted behavioral effects also be obtained in the absence of conformity
pressure? This was assessed by examining the responses of the control partici-
pants, who completed the attitude measures without conformity pressure. Finally,
were participants aware of the priming manipulations? This was likewise assessed
by examining the responses of the control participants, who were asked immedi-
ately afterward whether they had detected any recurring themes in the sentence
Conformity. The primary dependent measure was each participant’s average
response on the two 11-point scales. These data are presented in Fig. 1, which
shows that participants in the conformity condition gave higher ratings (M⫽8.73),
which therefore conformed more closely to those of the confederates than did
participants in either the nonconformity (M⫽8.03) or neutral (M⫽7.77) condi-
tions. The critical contrast between the conformity condition and the other two
584 EPLEY AND GILOVICH
conditions was signiﬁcant, separate variance t(57) ⫽2.80, p⬍.01. Examined
more closely, the responses of participants in the conformity group were signiﬁ-
cantly higher than those of participants in the control group, separate variance
t(57) ⫽3.03, p⬍.005, and marginally signiﬁcantly higher than those of
participants in the nonconformity group, separate variance t(57) ⫽1.68, p⫽.10.
The nonconformity and neutral conditions did not differ signiﬁcantly from each
other, t⬍1. This is consistent with the claim that words used to prime
nonconformity can also inadvertently prime conformity. This claim is further
supported by the ﬁnding that the variance of participants’ responses in the
nonconformity group (2.85) exceeded by a considerable margin the variance of
participants’ responses in the conformity group (0.73), Levene’s Test for Homoge-
neity of Variance ⫽3.84, p⫽.057. The variance of participants’responses in the
neutral group (1.25) was between that of the other two groups but closer to the
Behavior in the absence of pressure. Participants in the control conditions
answered the same two attitude questions as those in the experimental groups. As
expected, a one-way ANOVA on their responses revealed no signiﬁcant differ-
ences among the three priming conditions, F⬍1, indicating no priming effects in
the absence of conformity pressure (see Fig. 1). In addition, the variance in
participants’ responses in the conformity and nonconformity control groups were
almost identical (Levene’s Test for Homogeneity of Variance ⫽.36, p⬎.5).
Tests for awareness. We checked for awareness among the experimental
participants during debrieﬁng by asking them two questions. First, we asked
participants whether, and how, the sentence construction task may have been
related to the later feedback session. Once again, most participants simply relayed
the cover story, and none indicated any awareness of the true relationship.
Second, we asked participants whether they could remember any general themes
FIG. 1. Mean task ratings in experimental and control conditions.
contained in the priming task. None mentioned anything related to either
conformity or nonconformity. Thus, by the end of the experiment at least, it is
clear that experimental participants were unaware of the content of the primes to
which they had been exposed.
Results were slightly different for the control participants, who were asked to
write down any themes they had discerned among the stimulus words immedi-
ately after they had been exposed to them. In the conformity condition, one person
wrote ‘‘conformity,’’ a second wrote ‘‘obedience,’’ and a third wrote ‘‘obeying
people.’’ In the nonconformity condition, one person wrote ‘‘obedience/
rebellion’’and a second wrote ‘‘rebellion.’’ None of the other themes mentioned
were even remotely related to conformity or nonconformity. Thus, among the
control participants, there was some awareness of the priming stimuli, but it was
not pervasive (5 of 40 participants).
Does the awareness evidenced by these ﬁve participants call into question the
results observed in the experimental conditions? We think not. First, it stands to
reason that the ability to retrieve a theme running through the words contained in
the scrambled sentences task would be greatest right after the task is completed
and then fade with time. By the time participants in the experimental conditions
left their cubicles, assembled in a common room, and listened to the experiment-
er’s request for feedback, the relevant memory traces may have been inaccessible.
This would explain why none of the experimental participants in either study
reported a theme even remotely connected to conformity or nonconformity.
Indeed, it is for this reason, presumably, that past research using the scrambled
sentences paradigm has only assessed awareness of the priming stimuli among
experimental participants and invariably found none (Bargh et al., 1996; Char-
trand & Bargh, 1997; Srull & Wyer, 1979). More important, if participants were
aware of the priming stimuli and behaved in accordance with some implicit
‘‘demand,’’ we would not have obtained the results we did. In particular,
participants who were aware of being exposed to nonconformity primes and who
wished to please the experimenter would have conformed less than those in the
neutral condition. We found, in contrast, that they conformed a bit more.
This research investigated whether conformity can be inﬂuenced by manipula-
tions that are outside of conscious awareness. In both Experiments 1 and 2,
participants primed with conformity words tended to conform more in a later
situation than those primed with nonconformity words. In Experiment 2, partici-
pants primed with conformity words also conformed more than those primed with
neutral words. In addition, a series of probes revealed that experimental partici-
pants were not consciously aware of the priming manipulation.
We also found that participants primed with nonconformity were not systemati-
cally induced to rebel against the group norm. In Experiment 2, although those
primed with nonconformity again conformed less than those primed with confor-
mity, they did not conform less than participants exposed to neutral primes.
586 EPLEY AND GILOVICH
Instead, priming for nonconformity appeared, if anything, to produce slightly
more conformity behavior. It thus appears easier to induce people to mindlessly
‘‘go along’’with group pressure than to ‘‘go against.’’
Priming Conformity versus Nonconformity
There are several potential explanations for this asymmetry, and thus several
explanations for the results obtained in the nonconformity conditions of the two
experiments. First, it is typically easier for people to conform than to rebel. This
may have been particularly true in the present experiments in which deviating
from the example set by the confederates involved telling the experimenter to his
face that his experimental tasks were dull. Thus, our procedure may have simply
set a much higher ‘‘hurdle’’for expressing the impulse to rebel than the impulse to
conform. Second, because our confederates modeled conformity and not dissent,
participants in the conformity conditions may have been doubly primed—by the
words presented in the scrambled sentences task and by the confederates’
behavior. There is recent evidence that one person’s behavior can lead to
automatic mimicry by another (Chartrand & Bargh, 1999). Such an effect here
would have accentuated the effectiveness of the conformity primes and sup-
pressed the effectiveness of the nonconformity primes.
A third explanation is the one we referred to earlier: Priming nonconformity
may simultaneously prime conformity and thus lead to weak or variable effects.
As the very name implies, nonconformity tends to be understood in contrast
to—and hence in reference to—conformity. Our conformity manipulation may
thus have been relatively pure and effective, and our nonconformity manipulation
relatively impure and ineffective.
The present ﬁndings cannot deﬁnitively decide between these three interpreta-
tions. That task is left to future research. Nevertheless, although not deﬁnitive, we
believe the current data favor the latter ‘‘dual activation’’explanation because that
explanation is most congenial to the signiﬁcantly greater variability of responses
observed in the nonconformity condition. If the nonconformity primes inadver-
tently primed conformity for at least some of the participants, then one would
expect particularly variable responses among the group of participants exposed to
such primes. The priming stimuli would thus induce some participants to rebel
and others to conform.
Analogous asymmetries, of course, can be found elsewhere in the psychologi-
cal literature. One example is the marked/unmarked distinction mentioned earlier,
and another comes from the ﬁeld of visual perception, where visual traits are often
encoded in a similar prototype–deviate dichotomy. For example, tilted lines tend
to be encoded as a straight line with a certain degree of deviating tilt, whereas a
straight line is coded independently as something of a standard (Triesmann &
Gormican, 1988). The additional activation that the deviate induces (a straight
line plus tilt) explains why it seems to ‘‘pop out’’ in a display of prototypes
(straight lines), whereas a straight line does not pop out in a display of deviates.
The presence of the additional feature of the deviate (the deviating tilt) is easy to
pick up in a ﬁeld of stimuli that lack it; but the absence of such a feature is not so
easy to detect in a ﬁeld of stimuli that contain it. This, of course, is the well-known
feature positive effect (Agostinelli, Sherman, Fazio, & Hearst, 1986; Fazio,
Sherman, & Herr, 1982; Hearst, 1991).
If this interpretation is correct, then the asymmetry in the effectiveness of the
conformity and nonconformity primes was not an artifact of our particular
paradigm, but a potentially general phenomenon worthy of investigation in its
own right. This interpretation also implies that it may be particularly difficult to
activate the construct of nonconformity lexically because the marked/unmarked
distinction applies most readily to verbal primes. This does not mean, however,
that the asymmetry applies only to verbal primes. To the extent that the marked/
unmarked distinction is more than a linguistic phenomenon—that it applies to the
very constructs of conformity and nonconformity—the same asymmetry is likely
to be found for nonlexical priming as well.
Much of the recent research on priming has been concerned with whether the
behavior elicited is ‘‘automatic’’ in the sense of occurring without intention,
control, awareness, or attentional resources (for a review see Bargh, 1994). We
cannot say, of course, whether the conformity we observed in the two studies
reported here are automatic in each of these ways. That was not our goal. It is
clear, however, that our participants were not aware of the priming stimuli that
exerted an inﬂuence on their behavior, and in that sense the conformity observed
in these studies has something of an automatic quality. Thus our ﬁndings are
consistent with the proposition that although some instances of conformity may
be the product of conscious and agonizing deliberations, others may be less a
deliberate product of stimuli outside our awareness. Sometimes we ﬁnd ourselves
‘‘just going along.’’
Agostinelli, G., Sherman, S. J., Fazio, R. H., & Hearst, E. (1986). Detecting and identifying change:
Additions vs. deletions. Journal of Experimental Psychology: Human Perception and Perfor-
mance, 12, 445–454.
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.,
pp. 1–40). Hillsdale, NJ: Erlbaum.
Bargh, J. A., Chaiken, S., Govender, R., & Pratto, F. (1992). The generality of the automatic attitude
activation effect. Journal of Personality and Social Psychology,62, 893–912.
Bargh, J. A., Chen, M., & Burrows, L. (1996).Automaticity and social behavior: Direct effects of trait
construct and stereotype activation. Journal of Personality and Social Psychology, 71, 230–244.
Bilton, M., & Sim, K. (1971). Four hours in My Lai. New York: Viking Penguin.
Carver, C. S., Ganellen, R. J., Froming, W. J., & Chambers, W. (1983). Modeling: An analysis in terms
of category accessibility. Journal of Experimental Social Psychology, 19, 403–421.
Carroll, J., Davies, P., & Richman, B. (1971). The American heritage word frequency book. NewYork:
Chartrand, T. L., & Bargh, J. A. (1997). Automatic activation of impression formation and memoriza-
tion goals: Nonconscious goal priming reproduces effects of explicit task instructions. Journal of
Personality and Social Psychology, 71, 464–478.
588 EPLEY AND GILOVICH
Chartrand, T. L., & Bargh, J. A. (1999). The chameleon effect: The perception–behavior link and
social interaction. Journal of Personality and Social Psychology, 76, 893–910.
Clark, H. H., & Clark, E. V. (1977). Psychology and language. New York: Harcourt Brace Jovanovich.
Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C., & Kardes, F. R. (1986). On the automatic activation
of attitudes. Journal of Personality and Social Psychology, 50, 229–238.
Fazio, R. H., Sherman, S. J., & Herr, P. M. (1982). The feature-positive effect in the self-perception
process: Does not doing matter as much as doing? Journal of Personality and Social Psychology,
Gilbert, D. T., Giesler, R. B., & Morris, K. A. (1995). When comparisons arise. Journal of Personality
and Social Psychology, 69, 227–236.
Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Attitudes, self-esteem, &
stereotypes. Psychological Review, 102, 4–27.
Hearst, E. (1991). Psychology and nothing. American Scientist, 79, 432–443.
Herr, P. M. (1987). Consequences of priming: Judgment and behavior. Journal of Personality and
Social Psychology, 51, 1106–1115.
Mish, F. C., et al. (Eds.). (1989). The new Merriam–Webster dictionary (4th ed.). Springﬁeld, MA:
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 and
Social Psychology, 37, 1600–1672.
Triesmann, A., & Gormican, S. (1988). Feature analysis in early vision: Evidence from search
asymmetries. Psychological Review, 95, 15–48.