| MadSci Network: Medicine |
A few articles for your review on nail biting and what constitutes "habits"
I hopw you are able to further research on this most interesting topic.
Thank you for your question.
Blurry Words and Fuzzy Deeds
The Attribution of Obscure Behavior
Daniel T. Gilbert
University of Texas at Austin
Shawn E. McNulty
University of Texas at Austin
Traci A. Giuliano
University of California, Los Angeles
J. Eric Benson
Department of Community and Regional Planning University of Texas at Austin
ABSTRACT
When people attempt to infer the existence of traits from another's
behavior, they categorize the behavior, characterize the actor in trait
terms, and then correct that inference with information about situational
constraints. The 1st 2 stages require fewer attentional resources than does
the 3rd. However, when behavior is obscure (i.e., difficult to categorize
because its features are not easily apprehended), the 1st stage should
consume resources on which the 3rd stage depends, and undercorrected
inferences should result. In 2 experiments, behavior was made obscure by
distorting its visual or acoustical parameters. Although the obscure
behaviors could logically have been attributed to the constraining
situations in which they occurred, Ss who observed such behaviors were
especially unlikely to correct their trait characterizations of the actors.
----------------------------------------------------------------------------
----
This research was supported by National Science Foundation Grant BNS-8819836
to Daniel T. Gilbert.
We thank Olga Colon, Valerie Edwards, Greg Hixon, Josh Holohan, Cheryl
Irwin, Britt Simpson, Janet Spence, Al Swinkels, and Maggie Walsh for their
generous help with the production and execution of these experiments. We
also thank Eliot Smith, Bill Swann, Yaacov Trope, and several anonymous
reviewers for their helpful comments on a previous version of this article.
Correspondence may be addressed to Daniel T. Gilbert, Department of
Psychology, University of Texas, Austin, Texas, 78712.
Electronic mail may be sent to pshu666@utxvm.bitnet
Received: August 20, 1990
Revised: March 11, 1991
Accepted: June 11, 1991
----------------------------------------------------------------------------
----
The study of behavioral perception must occasionally strike the layperson as
an exercise in the obvious. After all, most of us have the experience of
effortlessly reading the inner states of others from their overt behavior:
Our interpretations of a knit brow, a coy smile, or an angry rebuke are
drawn so quickly and easily that "typically they are not experienced as
interpretations at all" ( Heider, 1958 , p. 82). Although the identification
of action is constrained by the action's context (a condemned prisoner's
nervous laughter rarely means "I am having a dandy time"), by the action's
effects on the environment (a boxer's kidney punch rarely means "I love
you"), and by the actor's dispositions (the compliments of a notorious
ingratiator rarely mean anything at all), these constraints are usually less
than complete and thus leave ample room for interpretation. Just what a
person is doing can, under many circumstances, be a matter of debate ( Ryle,
1949 ; Taylor & Crocker, 1981 ; Vallacher & Wegner, 1985 ; cf. Baron, 1988 ;
McArthur & Baron, 1983 ).
A variety of factors can alter the identification of a behavior, and Trope
(1986) has developed an elegant model of how these alterations influence
subsequent attributional work. In essence, the model argues that the
identity one assigns to an action can govern the inferences one draws about
the actor: Weeping will afford different inferences about an actor when it
occurs at a wedding than at a funeral, and prior knowledge of these formal
occasions may even change the perceived extremity of the emotional display.
In short, the answer to "Why?" is conditioned by the answer to "What?" The
trait-attribution model of Gilbert, Pelham, and Krull (1988) incorporates
Trope's (1986) insight by postulating three conceptually distinct stages in
the process of trait inference: Categorization of the behavior,
characterization of the actor in trait terms, and correction of the trait
inference with information about facilitative or inhibitory situational
forces (see also Quattrone, 1982 ). Thus, the utterance "Nice to see you"
may be categorized as an affable statement, the speaker may be characterized
as having a genuine affinity for the recipient of the remark, and, finally,
this inference may be corrected for the fact that the recipient is a
powerful superior whose presence demands polite affability.
The key feature of this model is that the three stages are assumed to vary
in the amount of conscious attention they require. Specifically, the
correction stage is presumed to require more attentional resources than
either the categorization or characterization stages and should thus be more
easily impaired by competing cognitive demands. A variety of experiments
have substantiated this prediction. For example, subjects who saw a woman
behave anxiously under anxiety-provoking circumstances categorized the
behavior as manifest anxiety, characterized the woman as dispositionally
anxious, and then corrected this inference by taking into account the
constraining circumstances of her behavior. However, when subjects were
asked to perform simultaneously an additional resource-consuming task, they
were much less likely to correct their characterizations. Although these
"cognitively busy" subjects noticed and remembered the anxiety-provoking
circumstances, they failed to use that information to institute correction
and persisted in considering the woman dispositionally anxious ( Gilbert,
Pelham, & Krull, 1988 ; see also Gilbert, Krull, & Pelham, 1988 ; Gilbert &
Osborne, 1989 ).
According to Trope's (1986) model, the categorization of an action may
influence subsequent attributions by changing the information on which those
attributions are predicated (e.g., perceived behavioral extremity). The
Gilbert, Pelham, and Krull (1988) model suggests an additional route by
which behavioral categorization may influence attribution. In previous tests
of this model, extraneous tasks (e.g., digit rehearsal, visual scanning, and
so on) were used to deplete perceivers' overall processing resources and, as
the model predicted, this depletion led to failure at the fragile correction
stage. The same logic suggests that if the relatively effortless
categorization stage was somehow made more effortful, then correction could
be similarly disrupted. In other words, obscure behavior (i.e., behavior
that is difficult to categorize because its key features are difficult to
apprehend) may pose an interpretive challenge that requires attention, and
if attentional resources are used to meet this challenge, they may be
unavailable for correction. Perceivers who are forced to deliberate about
what an action is may have fewer attentional resources with which to ponder
what the action means .
Categorization, then, may influence subsequent attributions not only by
altering the information on which these attributions are based (as Trope's
[1986] model predicts and research has shown) but also by siphoning off
attentional resources that are necessary for the completion of attributional
processing. Two experiments tested hypothesis this.
Experiment 1
Method Overview
Subjects watched a silent videotape of an anxious-looking female target who
was having a discussion with a stranger. Half the subjects learned that the
target was discussing an anxiety-provoking topic (her sexual fantasies), and
the remaining subjects learned that the target was discussing a mundane
topic (her ideal vacation). Half the subjects in each of these conditions
watched a normal videotape (easy categorization), and the remaining subjects
watched a visually degraded videotape in which the target's behavior was
obscured (difficult categorization). After viewing the videotape, subjects
estimated the level of the target's trait anxiety. (Most of the procedures
and measures in this experiment were used previously by Gilbert, Pelham, &
Krull, 1988 , and Gilbert & Osborne, 1989 .)
Subjects
Subjects were 34 female students at the University of Texas at Austin who
participated to fulfill a requirement in their introductory psychology
course. All subjects had normal, near normal, or corrected-to-normal vision.
Only native speakers of English were eligible to participate.
Instructions
On arrival at the laboratory, subjects were greeted by a male experimenter
who gave them a brief oral introduction to the experiment, provided them
with complete written instructions, and then escorted each subject to a
cubicle (equipped with video monitor) where she remained for the duration of
the experiment.
The written instructions explained that subjects would be asked to watch a
short clip from a videotape of a getting-acquainted conversation that had
ostensibly taken place earlier in the year. This conversation was alleged to
have been part of a project on the role of discussion topics in friendship
formation. Subjects were told that two unacquainted female students had been
assigned to discuss 1 of 10 different topics for about 5 min and that
subjects would be seeing a short (approximately 20 s) clip from that
conversation. It was explained that during the getting-acquainted
conversation the camera had been positioned behind one of the discussants,
and thus only one of the discussants (the target) would be visible on the
videotape.
Manipulation of Situational Constraints
Subjects were shown a list of the 10 potential discussion topics (dual
career couples, scary movies, drug testing, health and nutrition, sexual
fantasies, ideal vacations, humiliating accidents, student housing, women in
art, and interracial dating). Subjects were told that, to protect the
privacy of the discussants, the videotape would be shown without any sound,
but that they would be able to tell which of the 10 topics was being
discussed because the topic would appear in subtitles at the bottom of the
screen.
All subjects were shown a video clip in which the target appeared extremely
anxious and uneasy. 1 Half the subjects were randomly assigned to the
anxious topic condition. In this condition, the subtitle indicated that the
target was discussing an anxiety-provoking topic (i.e., her sexual
fantasies). The remaining subjects were assigned to the mundane topic
condition. In this condition subjects saw precisely the same behavior that
was seen by subjects in the anxious topic condition. However, in this
condition, the subtitle indicated that the target was discussing a mundane
topic (i.e., her ideal vacation). In the anxious topic condition, then, the
target's behavioral anxiety could logically be attributed to the nature of
the topic she was discussing and thus was not necessarily indicative of
dispositional anxiety; in the mundane topic condition, however, the same
behavior could not logically have been caused by the nature of the
discussion topic and thus was an excellent indicator of dispositional
anxiety.
Manipulation of Behavioral Obscurity
Half the subjects were randomly assigned to the normal behavior condition.
Subjects in this condition were shown a high-quality videotape that was free
of visual noise. The remaining subjects were assigned to the obscure
behavior condition. Subjects in this condition were shown a videotape that
had been visually degraded by misadjusting the tracking mechanisms on two
videocassette recorders and then using those machines to make successive
copies of the master tape. The final fourth generation tape had a great deal
of visual noise (streaking, blurring, color dropout, and snow), which
interrupted the flow of the action and obscured some of its details. It is
important to note that the subtitle was inserted only after the videotape
had been visually degraded; as such, the subtitles on both the normal and
degraded tapes were equally legible.
Dependent Measures
Subjects rated the target's dispositional anxiety on four 13-point bipolar
scales that were anchored at the endpoints with the phrases (a) is probably
comfortable (uncomfortable) in social situations, (b) is a calm (nervous)
sort of person, (c) is generally relaxed (anxious) with people, and (d) is
(not) easily embarrassed. Each scale was preceded by the phrase, "When I
think about how the woman in the film actually is in her day to day life, I
think she..."
Procedure
Before the experiment began, subjects were shown a silent, 20-s, visually
normal practice tape of a woman (not the target) sitting in a room, and they
were then given an opportunity to rate that woman on the trait-anxiety
measures. This enabled subjects to become familiar with both the procedure
and the dependent measures. After the practice tape was over, the
experimenter started the experimental tape. Subjects watched a 20-s clip of
an anxious-looking woman (the target) who was apparently engaged in
conversation with an off-screen partner. When the tape ended, subjects were
allowed up to 30 s to rate the target on the four trait-anxiety measures. At
the end of the experiment all subjects were asked to recall the discussion
topic. Finally, subjects were debriefed and dismissed.
Results and Discussion
We expected that subjects who viewed obscure behavior would make
correspondent inferences about the target and that subjects who viewed
normal behavior would not. Ratings on the four trait-anxiety scales were
averaged to create a trait-anxiety index. The omission of the embarrassment
item improved the reliability of this index (coefficient a changed from .79
to .84), and thus the item was deleted. The three-item index was submitted
to a 2 (topic: anxiety-provoking or mundane) * 2 (behavior quality: normal
or obscure) analysis of variance (ANOVA), which revealed a main effect of
topic, F (1, 30) = 6.95, p < .01, and a main effect of behavior quality, F
(1, 30) = 7.69, p < .01. These main effects were qualified by the predicted
Topic * Behavior Quality interaction, F (1, 30) = 6.30, p < .05. As Table 1
shows, subjects who saw normal behavior considered the anxious-looking
target to be more trait-anxious when she was discussing a mundane rather
than an anxiety-provoking topic. In contrast, subjects who saw obscure
behavior considered the target to be equally trait-anxious regardless of the
topic she discussed. In short, subjects who should have found the target's
behavior easy to categorize showed evidence of having used the discussion
topics to correct their characterizations of the target, whereas subjects
who should have found the behavior difficult to categorize did not.
Our own viewing of the degraded videotape left us confident that the
target's behavioral anxiety was obscure but, with a bit of effort,
detectable. Still, one might wonder whether subjects who saw that videotape
could actually detect the anxious behavior at all. The magnitude of the
ratings leaves little doubt that subjects did ultimately observe
considerable state anxiety. Subjects who saw a degraded videotape rated the
target as very anxious (i.e., well above the midpoint of the scale) and, in
fact, made ratings that were indistinguishable from the ratings of subjects
who saw a normal videotape that was subtitled with a mundane topic. In
addition, all subjects recalled the discussion topic that the target had
ostensibly been discussing, indicating that they had noticed and remembered
this information, too.
Experiment 2
The results of Experiment 1 suggest that when an action cannot be
categorized unthinkingly, perceivers will devote attentional resources to
the task and thus have fewer resources with which to correct their
characterizations of the actor. As provocative as this initial finding may
be, it is far from conclusive. First, because the target's behavior was held
constant and the presence of situational constraints was manipulated, we
predicted no difference between the attributions of the two groups of
subjects who viewed obscure behavior. There are, of course, always several
reasons why two experimental conditions might yield equivalent data, and a
bolder procedure entails manipulating the nature of the target's behavior
and predicting a larger difference for subjects exposed to obscure rather
than normal versions of that behavior. In Experiment 2 we did just that.
Second, Experiment 1 provided no evidence to indicate which stage of
processing (the initial categorization of the behavior or the subsequent
attributional correction) was affected by behavioral obscurity. In the
second part of Experiment 2, we confronted this issue directly. Finally, we
hoped to show that the phenomenon demonstrated in Experiment 1 transcended
experimental particulars; as such, Experiment 2 used verbal (rather than
nonverbal) behavior and examined attributions of attitude (rather than
emotion).
Method Overview
Female subjects listened to an audio recording of a "dating game" in which a
male contestant answered "loaded questions" posed by a female questioner.
The wording of the woman's questions clearly indicated that she preferred
men who had either a traditional or modern sex role orientation, and the man
always made claims that were consistent with the woman's preference. Half
the subjects in each of these conditions listened to a normal version of the
tape, and the remaining subjects listened to an acoustically degraded
version. All subjects then attempted to estimate the male contestant's true
sex role attitudes.
Subjects
Seventy-eight female students at the University of Texas at Austin
participated to fulfill a requirement in their introductory psychology
course. Only native speakers of English were eligible to participate.
Instructions
On arrival at the laboratory, subjects were greeted by a female experimenter
who gave them a brief oral introduction to the experiment, provided them
with complete written instructions, and then escorted each subject to a
cubicle equipped with headphones that were connected to a tape player in
another room. The instructions explained that the experimenter was
interested in "the process by which people form impressions of others-in
particular, how women form impressions of men."
Subjects were told that they would listen to a brief audio recording of two
college students who were playing "the dating game." Ostensibly, this
recording had been made as part of an earlier experiment in which male
students had attempted to win a date with an attractive female student.
Subjects were told that their task was to listen to an excerpt from this
dating game and then make judgments about the male contestant's true
attitudes toward women and their roles in society. Subjects were reminded
that "male contestants are often willing to bend the truth a little bit in
order to get a date with an attractive female."
Manipulation of Target's Behavior
Subjects were randomly assigned to one of two target-behavior conditions. In
the traditional condition, subjects heard the woman ask questions in which
she expressed a strong preference for men with traditional sex role
orientations (e.g., "I like it when men come after me, and not the other way
around. Who do you think should do the asking out?"), and they then heard
the man indicate that he possessed such an orientation (e.g., "I like to be
the one to ask a girl out"). In the modern condition, the woman's questions
betrayed a clear preference for men with modern sex role orientations (e.g.,
"When I see something or someone I want, I go after it. Who do you think
should do the asking out?"), and the man then indicated that he possessed
such an orientation (e.g., "I admire a woman who's not afraid to make the
first move"). In both conditions, then, a male target made self-descriptive
claims that were entirely consistent with the constraints of the leading
questions he was being asked (see Ginzel, Jones, & Swann, 1987 ; Swann,
Giuliano, & Wegner, 1982 ).
Manipulation of Behavioral Obscurity
Half the subjects were randomly assigned to the normal behavior condition.
These subjects listened to a recording of the dating game that was played at
a fairly high volume so that the recording was extremely easy to hear. The
remaining subjects were assigned to the obscure behavior condition. These
subjects listened to a recording in which the man's constrained responses
(but not the woman's constraining questions) were acoustically degraded by
lowering the volume to a just audible level. In the normal behavior
condition, one could not help but clearly hear the man's responses, but in
the obscure behavior condition, one had to concentrate rather intensely to
hear precisely what he said. Subjects in the obscure behavior condition were
asked to forgive the fact that a short circuit in the man's microphone had
caused the poor recording of his responses.
Dependent Measures Global trait measures.
Subjects rated the male target on three 13-point bipolar scales that were
preceded by the phrase "I think the male contestant has ..." and were
anchored at the endpoints with the phrases (a) very traditional
(progressive) attitudes toward women, (b) very conservative (liberal)
attitudes about relationships, and (c) very old-fashioned (modern) attitudes
about dating. These scales were intended to assess subjects' global
impressions of the male target's sex role orientation.
Specific trait measures.
Subjects rated the male target on 10 items taken from the Marital Relations
(Male Form) Subscale of the Male-Female Relations Questionnaire (MFRQ;
Spence, Helmreich, & Sawin, 1980 ). The MFRQ requires subjects to respond to
each item on a 5-point bipolar scale that is anchored at the endpoints with
the phrases strongly disagree and strongly agree. Subjects estimated the
male target's agreement with 10 statements that were chosen for their face
validity. All of the items were phrased such that agreement indicated a
traditional sex role orientation (e.g. "I would expect to be head of the
house simply because I'm a man"). These items were intended to assess
subjects' specific impressions of the male target's sex role orientation.
Experimental checks.
Three experimental checks were made at the end of the session. First,
subjects were asked whether they had normal, slightly impaired, or highly
impaired hearing. Second, subjects were asked to report how difficult it had
been for them to hear the audiotape on a 4-point Likert-type scale that
ranged from not at all difficult to very difficult. Third, subjects were
shown two of the woman's questions and were asked to recognize the gist of
the man's response. So, for example, subjects read "When the female asked
'Who do you think should do the asking out, men or women?' the male said
that (a) men should do the asking or (b) women should do the asking."
Subjects were asked to choose one of these responses or to state that they
did not know which of these responses was correct. This measure ensured that
only subjects who had heard and understood the audiotape would be included
in the analyses.
Procedure
Subjects listened to an audiotape in which a male contestant was subtly
urged to claim either a modern or traditional sex role orientation. The
audiotape was either acoustically normal or degraded. Subjects then
completed the global trait measures (three trait scales), the specific trait
measures (10 MFRQ items), and the experimental checks. Finally, subjects
were probed for suspicion, debriefed, and dismissed. 2
Results and Discussion Experimental Checks
All subjects reported having normal or near normal hearing. However, given
routine variation in the auditory acuity of subjects, we expected that some
subjects in the obscure behavior condition might actually fail to hear the
audiotape altogether and thus respond randomly when presented with the
dependent measures. In fact, 13 subjects failed to recall the gist of the
male target's verbal responses (i.e., they failed an extremely easy two-item
recognition memory test), and these subjects were excluded from all
analyses. Although inclusion of such subjects adds random noise to the data
and thus decreases the significance levels of the effects, it has virtually
no effect on the pattern of results. Finally, a 2 (target's response:
traditional or modern) * 2 (behavior quality: normal or obscure) ANOVA
revealed that subjects in the normal behavior condition found the audiotapes
much easier to hear than did subjects in the obscure behavior condition ( M
s = 0.03 and 2.12, respectively on a 0- to 3-point scale), F (1, 61) =
271.92, p < .001.
Global Trait Measures
The three trait scales were averaged to create a global sex role orientation
index, and this index was submitted to ANOVA. The analysis revealed a main
effect of target's response, F (1, 61) = 4.06, p < .05, and the expected
Target's response * Behavior Quality interaction, F (1, 61) = 3.14, p = .08
(means are shown in the top portion of Table 2 ). Planned comparisons
revealed that subjects who heard obscure behavior made correspondent
inferences about the target, t (61) = 2.68, p < .01, whereas subjects who
heard normal behavior did not, t < 1. In other words, subjects who were
forced to devote considerable attention to categorizing the target's
behavior as traditional or modern were especially unlikely to correct their
corresponding characterizations of the target's attitude. Indeed, there was
a significant positive correlation between difficulty of categorization
(i.e., answers to the question "How difficult was it to hear the tape?") and
the tendency to draw correspondent inferences about the target, r (63) =
.20, p = .05. This correlation is especially impressive because difficulty
of categorization was only roughly assessed by self-report, and people are
known to be quite poor at reporting accurately on such mental operations
(e.g., Jacoby, Kelley, & Dywan, 1989 ; Nisbett & Wilson, 1977 ).
It is worth noting that in both this and the previous experiment, subjects
who observed normal behavior showed no evidence of correspondence bias. This
may surprise those who consider the correspondence bias an unrelenting
phenomenon, but, in fact, the bias has failed to appear in any number of
classic studies. For example, Jones, Davis, and Gergen (1961) asked subjects
to make inferences about a job applicant who claimed to have the temperament
for a desirable job, and subjects showed no evidence of correspondence bias.
Ross, Amabile, and Steinmetz's (1977) well-known experiment is often cited
as an example of correspondence bias, but a close reading shows that
quizmasters showed no such effect. Indeed, it is now clear that subjects can
be expected to show little or no bias when they are held accountable for
their inferences ( Tetlock, 1985 ), when they discuss their judgments with
others ( Wright & Wells, 1985 ), when a familiar target is highly motivated
to convey his or her true dispositions ( Fleming, Darley, Hilton, & Kojetin,
1990 ), when subjects attend primarily to the nonverbal behavior of a
"leaky" target ( Gilbert & Krull, 1988 ), when the target is distinctly
unenthusiastic ( Jones, Worchel, Goethals, & Grumet, 1971 ), and when the
dependent measures stray too far from the target's behavior ( Cantor,
Pittman, & Jones, 1982 ; Sumpton & Gregson, 1981 ). Laboratory
experimentation is, of course, meant to demonstrate possibilities and not to
provide actuarial information; thus, one simply cannot say how often the
correspondence bias emerges in day-to-day life and how often it is
attenuated by other factors. What is known is that there are a number of
variables that promote and inhibit the bias, and knowledge of these
variables is critical to an understanding of the bias itself.
Specific Trait Measures
Subjects' ratings of the target on the 10 MFRQ items were averaged to create
a specific sex role orientation index. This 10-item index was submitted to
ANOVA, which revealed only the expected Target's Response * Behavior Quality
interaction, F (1, 61) = 8.50, p < .01. As the lower portion of Table 2
shows, subjects who heard obscure behavior made correspondent inferences
about the target, t (61) = 1.96, p = .05, whereas subjects who heard normal
behavior did not. 3
Which Stage of Processing Was Affected?
The results of Experiment 2 are consistent with the hypothesis that
difficult categorization can impair subsequent attributional correction;
however, they are also consistent with the alternative hypothesis that
difficult categorization affects the categorization stage rather than the
subsequent correction stage of processing. It is possible, for example, that
behavioral obscurity simply caused subjects to perceive the target's
responses as extreme (i.e., especially liberal or conservative). As such,
subjects who observed obscure behavior would be expected to make more
extreme attributions about the target even if they did perform adequate
attributional correction, simply because their initial perceptions of the
behavior were "perceptually inflated" (see Trope, 1986 ; Trope, Cohen, &
Maoz, 1988 ).
To rule out this alternative hypothesis, we asked 54 female subjects to
participate in a replication of Experiment 2. However, this time we told the
subjects that the male target was a paid actor and that their job was simply
to determine how well he played his part. In particular, subjects were asked
to rate the target's behavior, rather than his disposition, on three global
dimensions (traditional/progressive, conservative/liberal, and
old-fashioned/modern). The results of this follow-up experiment strongly
supported our original (and not the alternative) hypothesis. Subjects
considered the obscure behavior much more difficult to hear than the normal
behavior, M s = 2.18 and 0.00, respectively; F (1, 44) = 188.88, p < .001.
Nine of the 54 subjects showed imperfect memory for the target's behavior
and were thus excluded from all analyses. (It is comforting to note that
this was precisely the same proportion as failed the memory test in
Experiment 2.) The remaining subjects' ratings on the three scales were
highly reliable (coefficient a = .88) and were thus averaged to create a
behavioral perception index. An ANOVA on this index revealed only a main
effect of target's response: Subjects thought that the modern target made
much more progressive, liberal, and modern statements than did the
traditional target, M s = 6.46 and 8.71, respectively; F (1, 41) = 14.61, p
< .001. Behavior quality had absolutely no main or interactive effects on
these judgments (both F s < 1). In addition, self-reported difficulty during
categorization was completely uncorrelated with the perceived extremity of
the behavior, r (43) = .07, ns. In short, behavioral obscurity affected
subjects' attributions about the target but did not affect their perceptions
of the target's behavior.
This finding should not be seen as undermining the validity of Trope's
(1986) model. That model suggests that when behavior is ambiguous,
information about situational constraints will be used in resolving this
ambiguity; thus, knowing that tears (ambiguous behavioral features) are
being shed at a wedding (happy situation) rather than at a funeral (sad
situation) may cause perceivers to identify the behavior as especially
joyous. Why, then, didn't knowledge of situational constraints (e.g., the
modern woman's leading questions) affect our subjects' perceptions of the
male target's responses (e.g., cause them to perceive his responses as
especially liberal)? The key, we believe, lies in the difference between
ambiguous and obscure behavior. Ambiguous behavior is behavior in which
clearly identifiable indicators (e.g., tears) may signal more than one inner
state (e.g., joy or sadness). Obscure behavior, on the other hand, is
behavior in which indicators are difficult to identify ("What's she doing
with her hands?"), but once identified ("Oh. I see. She's biting her nails")
can be reasonably interpreted to indicate only one inner state ("She must be
nervous"). In short, behavioral ambiguity and obscurity are not the same
thing, and thus Trope's model does not (in our view) make predictions that
run counter to our findings.
General Discussion
People are rarely stumped by the question "What is that fellow doing?"
Actions generally speak for themselves, and most of us are able to
categorize the behaviors of others with an ease that may occasionally belie
the intricate nature of that process. Yet, one need only miss a key phrase
in a colleague's talk, or look away from a film at a critical moment, to be
reminded of how fragile the understanding of action can be. When a momentary
distraction causes that understanding to collapse, conscious attention must
be called on to guide its reconstruction.
The present studies demonstrate some of the consequences of calling on
conscious attention in this way. When attention is devoted to the
categorization of behavior, it cannot fuel the effortful components of the
attributional analysis. As such, only the relatively effortless components
of that analysis are completed, and strong dispositional characterizations
are the result. In short, the contemplation of what may undermine the
discovery of why. Rather than reiterate the theoretical implications of this
finding, we will close by discussing some traditional questions on which it
may have some bearing.
Understanding the Unfamiliar
When individuals interact with members of other races, genders, and
nationalities, they often draw more dispositional inferences from those
behaviors than from the identical behaviors of their cohort. People think of
out-groups in simple ways and are thus overly influenced by the
unrepresentative actions of their members ( Linville & Jones, 1980 ;
Quattrone & Jones, 1980 ). In addition, unfamiliarity with the norms that
govern an out-group member's behavior may not allow us to estimate the
consensus that such actions enjoy. We may assume, for example, that an
Israeli acquaintance is particularly bold without realizing that close
physical proximity is a Middle Eastern custom and, as such, reveals little
about the acquaintance as a unique individual.
Our research suggests an additional mechanism by which such effects may be
produced. Despite the universality of spontaneous emotional expressions
Ekman, 1971 ), some attitudes and emotions are displayed by different
cultures in different ways. A Brit may affect a stoic smile when unhappy, a
Japanese may nod despite strenuous disagreement, and even members of
familiar subcultures may use words in ways that confound their usual meaning
("That saxophone player is really bad "). Even when an outsider knows these
display rules, it is likely that their application requires considerable
thought ("Let's see, if hot is cool and bad is good, then radical must mean
..."). To the extent that out-groups attach different behavioral signs to
the thing being signified, then passing from the former to the latter may be
something of a labor. One reason why we draw dispositional inferences about
out-group members, then, may be that we work hard to understand what they
are doing and are thus unable to give much thought to why they are doing it.
Categorization by Actors and Observers
Jones and Nisbett (1972) suggested that there is "a pervasive tendency for
actors to attribute their actions to situational requirements, whereas
observers tend to attribute the same actions to stable personal
dispositions" (p. 80, italics added). Despite some notable exceptions
Monson & Snyder, 1977 ), this simple proposition has proved enormously
reliable and robust. One explanation of the effect is that actors and
observers assign different identities to the actor's behavior. Actors often
have "inside information" that may enable them to construe their behavior in
ways that an observer might not. Whereas an observer may think of eating
meat loaf as an expression of hunger, the actor may know that eating Aunt
Shirley's meat loaf is an act of duty and sacrifice. As such, actors and
observers may occasionally make different attributions because they are, in
fact, making attributions for different behaviors (cf. Kahneman & Miller,
1986 ).
But even when actors and observers do agree about the identity of an action,
it seems reasonable to suggest that actors will generally find it easier to
categorize their own behavior than will observers. Knowledge of present
intent and past behavior may enable the actor to make sense of actions that
would initially strike an uninformed observer as quite anomalous. The person
who places a rubber chicken on a friend's grave certainly knows that this
was the favorite prop of the departed comedian, but an uninformed observer
will have to work particularly hard to understand an offering of a silly
toy. Our experiments suggest that the payoff for the observer's hard work
may be an undercorrected dispositional inference about the actor. The
divergent attributions of actors and observers may be, at least in part, a
consequence of the relative difficulty each has in determining just what
behavior it is that needs attributing.
Hidden Meaning of Action
Experimental psychologists have long considered clinical judgment to be
something of a slow-moving target. In addition to having undue confidence in
their judgments ( Oskamp, 1965 ) and perceiving relations where none exist
Chapman & Chapman, 1967 ), psychodynamic therapists have also been accused
of showing a profound bias toward dispositional explanations of human
behavior. As Nisbett and Ross (1980, p. 244) remarked, "Freud risked
elevating the fundamental attribution error to the status of a scientific
principle."
Although it is difficult to gauge the pervasiveness of this dispositional
bias, one line of reasoning suggests that the bias is a natural consequence
of the clinical task. The goal of most insight therapies is to identify a
person's behavior in ways that the person might not. A client may claim that
she is visiting her mother every Tuesday and taking a class in quantum
mechanics, but the therapist's job is to discover the transcendental
connection that allows both actions to be similarly construed. Clinicians
must certainly exert considerable effort to uncover the single category to
which apparently disparate behaviors belong; it is no meager insight to
realize that frequent visits with a critical parent and enrollment in a
demanding seminar may both be subtle means of undermining one's sense of
self-worth.
Our research suggests that because such categorizations are difficult to
achieve, clinicians may be able to devote less thought to the situational
antecedents of the behaviors that they have categorized. Ironically, it may
be the clinician's shrewd categorization of action that casts a somewhat
dispositional pale on the analysis of its causes. Our studies suggest that
observers should be inclined to draw dispositional inferences for those
actions whose "hidden meanings" challenge discovery. In fact, the more
deeply obscured the meaning of behavior, the more likely observers should be
to embrace a dispositional explanation for it-a speculation that brings
whole schools of psychotherapy to mind.
Conclusion
A popular bit of graffiti has it that "time is nature's way of keeping
everything from happening at once." In a similar vein, it might be said that
consciousness is nature's way of keeping people from thinking all their
thoughts at the same time. Indeed, the most fundamental limit of
consciousness is that it takes but a single object, and thus thinking about
one thing necessarily precludes thinking about another. Drawing inferences
from human behavior is no exception to this rule: To the extent that we stop
to ponder one aspect of a person's behavior, we may be prevented from
considering another aspect, and our ultimate understanding of the person may
therefore be incomplete. The present experiments suggest that when what
becomes a tough question, social perceivers may never ask why.
References
Baron, R. M. (1988). An ecological framework for establishing a dualmode
theory of social knowing.(In D. Bar-Tal & A. W. Kruglanski (Eds.), The
social psychology of knowledge (pp. 48-82). Cambridge, England: Cambridge
University Press.)
Cantor, N., Pittman, T. S. & Jones, E. E. (1982). Choice and attitude
attribution: The influence of constraint information on attributions across
levels of generality. Social Cognition, 1, 1-20.
Chapman, L. J. & Chapman, J. P. (1967). Genesis of popular but erroneous
diagnostic observations. Journal of Abnormal Psychology, 72, 193-204.
Ekman, P. (1971). Universals and cultural differences in facial expressions
of emotion.(In J. K. Cole (Ed.), Nebraska Symposium on Motivation (Vol. 19,
pp. 207-283). Lincoln, NE: University of Nebraska Press.)
Fleming, J. H., Darley, J. M., Hilton, J. L. & Kojetin, B. A. (1990).
Multiple audience problem: A strategic communication perspective on social
perception. Journal of Personality and Social Psychology, 58, 593-609.
Gilbert, D. T. & Krull, D. S. (1988). Seeing less and knowing more: The
benefits of perceptual ignorance. Journal of Personality and Social
Psychology, 54, 193-202.
Gilbert, D. T., Krull, D. S. & Pelham, B. W. (1988). Of thoughts unspoken:
Social inference and the self-regulation of behavior. Journal of Personality
and Social Psychology, 55, 685-694.
Gilbert, D. T. & Osborne, R. E. (1989). Thinking backward: Some curable and
incurable consequences of cognitive busyness. Journal of Personality and
Social Psychology, 57, 940-949.
Gilbert, D. T., Pelham, B. W. & Krull, D. S. (1988). On cognitive busyness:
When person perceivers meet persons perceived. Journal of Personality and
Social Psychology, 54, 733-740.
Ginzel, L. E., Jones, E. E. & Swann, W. B. (1987). How naive is the naive
attributor?: Discounting and augmentation in attitude attribution. Social
Cognition, 5, 108-130.
Heider, F. (1958). The psychology of interpersonal relations. (New York:
Wiley)
Jacoby, L. L., Kelley, C. M. & Dywan, J. (1989). Memory attributions.(In H.
L. Roediger & F. I. M. Craik (Eds.), Varieties of memory and consciousness:
Essays in honor of Endel Tulving (pp. 391-422). Hillsdale, NJ: Erlbaum.)
Jones, E. E., Davis, K. E. & Gergen, K. J. (1961). Role playing variations
and their informational value for person perception. Journal of Abnormal and
Social Psychology, 63, 302-310.
Jones, E. E. & Nisbett, R. E. (1972). The actor and the observer: Divergent
perceptions of the causes of behavior.(In E. E. Jones, D. E. Kanouse, H. H.
Kelley, R. E. Nisbett, S. Valins, & B. Weiner (Eds.), Attribution:
Perceiving the causes of behavior (pp. 79-94). Morristown, NJ: General
Learning Press.)
Jones, E. E., Worchel, S., Goethals, G. R. & Grumet, J. F. (1971). Prior
expectancy and behavioral extremity as determinants of attitude attribution.
Journal of Experimental Social Psychology, 7, 59-80.
Kahneman, D. & Miller, D. T. (1986). Norm theory: Comparing reality to its
alternatives. Psychological Review, 93, 136-153.
Linville, P. W. & Jones, E. E. (1980). Polarized appraisals of out-group
members. Journal of Personality and Social Psychology, 38, 689-703.
McArthur, L. Z. & Baron, R. M. (1983). Toward an ecological theory of social
perception. Psychological Review, 90, 215-238.
Monson, T. C. & Snyder, M. (1977). Actors, observers, and the attribution
process: Toward a reconceptualization. Journal of Experimental Social
Psychology, 13, 89-111.
Nisbett, R. E. & Ross, L. (1980). Human inference: Strategies and
shortcomings of social judgment. (Englewood Cliffs, NJ: Prentice-Hall)
Nisbett, R. E. & Wilson, T. D. (1977). Telling more than we can know: Verbal
reports on mental processes. Psychological Review, 84, 231-259.
Oskamp, S. (1965). Overconfidence in case-study judgments. Journal of
Consulting Psychology, 29, 261-265.
Quattrone, G. A. (1982). Overattribution and unit formation: When behavior
engulfs the person. Journal of Personality and Social Psychology, 42,
593-607.
Quattrone, G. A. & Jones, E. E. (1980). The perception of variability within
in-groups and out-groups: Implications for the law of small numbers. Journal
of Personality and Social Psychology, 38, 141-152.
Ross, L., Amabile, T. M. & Steinmetz, J. L. (1977). Social roles, social
control, and biases in social-perception processes. Journal of Personality
and Social Psychology, 35, 485-494.
Ryle, G. (1949). The concept of mind. (New York: Penguin Books)
Spence, J. T., Helmreich, R. L. & Sawin, L. L. (1980). The male-female
questionnaire: A self-report inventory of sex-role behaviors and preferences
and its relationship to masculine and feminine personality traits, sex-role
attitudes, and other measures. JSAS: Catolog of Selected Documents in
Psychology, 10, -87.
Sumpton, R. & Gregson, M. (1981). The fundamental attribution error: An
investigation of the sensitivity to role-conferred advantages in
self-presentation. British Journal of Social Psychology, 20, 7-11.
Swann, W. B., Giuliano, T. & Wegner, D. M. (1982). Where leading questions
can lead: The power of conjecture in social interaction. Journal of
Personality and Social Psychology, 42, 1025-1035.
Taylor, S. E. & Crocker, J. (1981). Schematic bases of social information
processing.(In E. T. Higgins, C. P. Herman, & M. P. Zanna (Eds.), Social
cognition: The Ontario Symposium (Vol. 1, pp. 89-134). Hillsdale, NJ:
Erlbaum.)
Tetlock, P. E. (1985). Accountability: The neglected social context of
judgment and choice.(In L. L. Cummings & B. M. Straw (Eds.), Research in
organizational behavior (Vol. 7, pp. 297-332). Greenwich, CT: JAI Press.)
Trope, Y. (1986). Identification and inferential processes in dispositional
attribution. Psychological Review, 93, 239-257.
Trope, Y., Cohen, O. & Maoz, Y. (1988). The perceptual and inferential
effects of situational inducements on dispositional attributions. Journal of
Personality and Social Psychology, 55, 165-177.
Vallacher, R. R. & Wegner, D. M. (1985). A theory of action identification.
(Hillsdale, NJ: Erlbaum)
Wright, E. F. & Wells, G. L. (1985). Does group discussion attenuate the
dispositional bias? Journal of Applied Social Psychology, 15, 531-546.
1
Two different actresses were used as targets. Approximately half the
subjects in each condition saw one actress, and the remaining subjects saw
the other. The identity of the actress had no effect on any of the analyses
to be reported and is not discussed further.
2
It may be useful to keep in mind that subjects in this and the previous
experiment observed behavior that was dynamic and ongoing. In such
situations, a subject may be thought of as executing the
categorize-characterize-correct sequence with each new quantum of behavior.
When the categorization stage takes too much capacity (i.e., time or
attention), the subject should not be able to execute the correction stage
before she has to go back and execute the categorization stage again for the
next behavioral quantum.
3
In fact, subjects who heard normal behavior showed an unexpected reversal, t
(61) = 2.10, p < .05. This reversal was due primarily to ratings on two
items that described emotional displays: (a) I think I should be emotionally
stronger and tougher than my wife and (b) I expect my wife to be more
emotionally dependent on me than I am on her. These items (and no others)
mapped directly on to one of the modern target's responses and thus caused
subjects to infer that the modern target had a traditional attitude toward
emotional displays. Indeed, an ANOVA performed on an eight-item index that
excluded these two MFRQ items revealed only the expected Target's Response *
Behavior Quality interaction, F (1, 61) = 8.25, p < .01. Most important, the
unexpected reversal was no longer reliable, t (61) = 1.57, p > .12.
Habits as Knowledge Structures
Automaticity in Goal-Directed Behavior
Henk Aarts
Department of Psychology and Language Eindhoven University of Technology
Ap Dijksterhuis
Department of Social Psychology University of Nijmegen
ABSTRACT
This study tested the idea of habits as a form of goal-directed automatic
behavior. Expanding on the idea that habits are mentally represented as
associations between goals and actions, it was proposed that goals are
capable of activating the habitual action. More specific, when habits are
established (e.g., frequent cycling to the university), the very activation
of the goal to act (e.g., having to attend lectures at the university)
automatically evokes the habitual response (e.g., bicycle). Indeed, it was
tested and confirmed that, when behavior is habitual, behavioral responses
are activated automatically. In addition, the results of 3 experiments
indicated that (a) the automaticity in habits is conditional on the presence
of an active goal (cf. goal-dependent automaticity; J. A. Bargh, 1989 ),
supporting the idea that habits are mentally represented as goal-action
links, and (b) the formation of implementation intentions (i.e., the
creation of a strong mental link between a goal and action) may simulate
goal-directed automaticity in habits.
----------------------------------------------------------------------------
This work was facilitated by a J. F. Schouten Institute for User System
Interaction grant and a Royal Netherlands Academy of Arts and Sciences
fellowship.
We thank John Bargh, Peter Gollwitzer, Sander Koole, Rene Lion, Edwin Poppe,
Peep Stalmeier, and Marcel Zeelenberg for their helpful comments on earlier
versions of this article. We are grateful to Ankie Bosch, Erica Derijcke,
Peter De Vries, Alderina Fokkema, Christel Van Vught, and Ronald Willemsen
for their help in collecting the data.
Correspondence may be addressed to Henk Aarts, Department of Psychology and
Language, DG 0.23, Eindhoven University of Technology, Eindhoven, the
Netherlands, 5600 MB.
Electronic mail may be sent to h.a.g.aarts@tm.tue.nl
Received: November 24, 1997
Revised: April 23, 1999
Accepted: May 14, 1999
----------------------------------------------------------------------------
----
The majority of people's actions are executed on a routine basis. The better
part of the behavioral repertoire is frequently exhibited in the same
physical and social environment and has taken on a habitual character (e.g.,
James, 1890 ; Ouellette & Wood, 1998 ; Triandis, 1980 ). These habits are
extremely useful in that they enable one to perform one's actions in a
mindless, automatic fashion. James (1890) emphasized the importance of
habits: "We must make automatic and habitual, as early as possible, as many
useful actions as we can" (p. 122). James's reasoning was straightforward.
The more actions one can delegate to the unconscious, the more room there is
to do things that necessarily require conscious processing. Writing an
article would be a more difficult affair if typing (and driving, and taking
a shower, and even brushing one's teeth) required conscious planning.
Despite a large contemporary literature on automatic processes in social and
nonsocial cognition and behavior (e.g., Bargh, 1989 , 1997 ; Hasher & Zacks,
1979 ; Norman & Shallice, 1986 ; Smith & Lerner, 1986 ), the available
research on the concept of habit is largely confined to studies in which the
role of past behavior was investigated in the context of attitude-behavior
models. An important contribution in this field was made by Bentler and
Speckart (1979) , who investigated students' consumption of alcohol and
marijuana. These authors suggested that such actions become habitual over
time and, importantly, that they can be instigated without mediation of
attitudes or intentions (or "products of reasoning" in general). Indeed, the
results of their study clearly showed that a measure of habit (obtained by
self-reported frequency of behavior in the previous 2 weeks) does predict
future behavior directly, indicating that such behavior is initiated
automatically, that is, without deliberation and thought.
The work of Bentler and Speckart (1979) has been replicated by many other
investigators in a wide variety of behavioral domains. Ouellette and Wood's
(1998) meta-analysis of studies on habits showed that the direct influence
of past behavior on future behavior was most pronounced for behaviors
executed frequently and consistently in a stable context. Behaviors carried
out less often were more accurately predicted by consciously formed
intentions toward the behavior. This pattern of results indeed confirms the
assumption that when a behavior has been performed many times in the past,
future behavior becomes increasingly under control of an automatized
process, whereas a behavior executed less frequently is (still) guided by
evaluative interpretations and considerations (as expressed, for instance,
in the theory of reasoned action; Fishbein & Ajzen, 1975 ). In summary,
there is ample evidence indicating that habitual behavior (with habitual
defined as a function of relative frequency of past performance) is
automatic. It is determined by past behavior and not mediated by attitudes,
intentions, or other concepts referring to more deliberate or conscious
processes (see also Aarts, Verplanken, & van Knippenberg, 1998 ; Triandis,
1980 ).
The direct influence of past behavior performed frequently on future
behavior also underscores the behaviorists' maxim that behavior is largely
influenced by habit (e.g., Hull, 1943 ; Skinner, 1938 ; Watson, 1914 ).
Actually, the direct relation between past and future action shows that
people simply do things as they did them before. Because the concept of
habit is strongly rooted in behaviorist approaches to learning theory, it
was assumed for a long time that mental (or cognitive) processes do not
mediate the automatic activation of habitual responses to environmental
stimuli. In contemporary research, however, it is often argued that
cognition does play a role in the direct control of environmental cues over
behavior (e.g., Bargh & Chartrand, 1999 ; Bargh & Gollwitzer, 1994 ; Norman
& Shallice, 1986 ; Ronis, Yates, & Kirscht, 1989 ). In line with this
research, our main hypothesis is that habits are mentally represented and
that they can be activated automatically.
Specifically, we conceive of habits as a form of goal-directed automatic
behavior (cf. goal-dependent automaticity; Bargh, 1989 ). Habits are
represented as links between a goal and actions that are instrumental in
attaining this goal. The strength of such links is dependent on frequent
co-activation of the goal and the relevant actions in the past. The more
often the activation of a goal leads to the performance of the same action
under the same circumstances, the stronger the habit (i.e., the link between
goal and action) will become. Our purpose was to investigate some key
assumptions derived from this conceptualization. First, we hypothesized that
habitual behavior can be activated automatically. Second, we assumed that
this automatic activation does not occur for behavior that is not habitual
(not represented as a goal-action link). Third, we assumed that the
automatic activation of a habitual action is goal dependent. That is,
actions are automatically activated provided that the relevant goal is
activated in the first place.
Development and Representation of Habits
Goals refer to desired, or anticipated, outcomes or end states ( Austin &
Vancouver, 1996 ; Gollwitzer & Moskowitz, 1996 ; Locke & Latham, 1990 ).
These goals can be the consequence of physiological needs, such as thirst
and hunger, as well as various other "needs" or motives, such as restoring
personal hygiene, making friends, acquiring knowledge, or becoming a
professor (e.g., Geen, 1995 ; Mook, 1996 ). When goals are not pursued
often, actions that one can deploy to attain desired outcomes are not
specified in these goals. Instead, goal activation guides organisms to
select an action or to form a plan to perform a certain action (see also
Carver & Scheier, 1981 ; Cohen, 1996 ; Srull & Wyer, 1986 ), but it does not
lead to the immediate instigation of an action. Hence, when pursuing
relatively unfamiliar goals, people are likely to ponder the possible
actions they can use to achieve the goal before they engage in an action. 1
When goals are pursued regularly, however, the need to pay conscious
attention to details dwindles (e.g., Anderson, 1982 ; Fitts & Posner, 1967 ;
Newell & Rosenbloom, 1981 ). When people select the same actions more often
and when these actions lead to goal achievement in a satisfactory manner,
the actions become mentally linked to the goal. That is, selecting and
performing the same goal-directed behavior frequently and consistently leads
to associations between the goal and the instrumental actions (i.e., to the
formation of a habit). As a result, activation of these goals spreads
automatically to the associated actions (cf. Anderson, 1993 ; M‰ntyl‰,
1993 ). The exhibition of habits, then, is the result of the automatic and
immediate activation of the habitual action on the instigation of a goal.
These goals are critical. In the case of habits, the instigation of the goal
to act is necessary to activate the associated actions automatically. Many
well-practiced or skilled actions, such as typing, driving a car, and riding
a bicycle, are usually qualified as automatic or habitual, but they require
the activation of a goal. For instance, an undergraduate student needs a
goal to use a bicycle (e.g., traveling to attend lectures at the university)
to activate subsequent behavioral steps such as "going to the garage to get
the bike" and "turning left at the statue of the hairy dromedary." In a
sense, habits can be seen as hierarchical mental representations in which
activation of a goal leads to activation of a number of associated behaviors
lower in the hierarchy. The proposal of such a hierarchical structure of
action is consistent with others who have previously proposed such a
representation ( Carver & Scheier, 1981 ; Gallistel, 1985 ; Miller,
Galanter, & Pribram, 1960 ; Mischel, 1973 ; Powers, 1973 ; Vallacher &
Wegner, 1987 ; see also Schank & Abelson, 1977 ).
It should be noted that our perspective on the development of (the cognitive
representation of) habits is also partly based on the recent work of Bargh
(1990 ; see also Bargh & Gollwitzer, 1994 ). He has suggested that when the
same choices are frequently pursued and implemented in a given situation (or
as the result of a given goal), an association between the mental
representation of that situation and the representation of the goal-directed
action will emerge. Frequent coactivation of a particular situation and a
particular behavioral decision increases the strength and accessibility of
that association. Hence, frequent and consistent performance of a
goal-directed action in a specific situation facilitates the ease of
activating the mental representation of this behavior (and hence the
resulting action itself) by the situation. Similar principles have also been
proposed and empirically established for the activation of other mental
representations, such as attitudes and stereotypes ( Devine, 1989 ; Fazio,
Sanbonmatsu, Powell, & Kardes, 1986 ; for an overview, see Higgins, 1996 ).
Activation of Goal-Directed Action
That the environment is indeed capable of activating goal-directed behavior
automatically has been established in the domain of motivations. In a test
of their automotive model, Bargh and colleagues ( Bargh, 1997 ; Bargh,
Gollwitzer, Chai, & Barndollar, 1999 ) showed that goals (and the associated
actions) can be elicited directly by the environment. Activation in this
case occurs without the person's awareness. They found that participants who
were primed with achievement or affiliation goals behaved in accordance with
the primed goal (solving either many or a few word puzzles in the presence
of a confederate who appeared to be not skilled in the task). In other
words, the situation elicited the relevant behavior as dictated by the
primed goal.
Support for the idea that the activation of actions depends on the
underlying mental representation and that this representation differs as a
function of one's personal history was obtained by Bargh, Raymond, Pryor,
and Strack (1995) . They established that power (as a situational feature)
and sexuality are mentally associated, but only for those with sexual
harassment tendencies. Hence, this research shows not only that the
environment can have a direct impact on goal-directed behavior but also that
this effect is dependent on whether goal and action are associated: The
environment affects goal-directed behavior automatically only if goals are
associated with the environment (i.e., when this goal was pursued earlier in
the same situations). Our conceptualization of habit is comparable with this
perspective. The environment can activate goal-directed behavior
automatically, but only when this behavior is habitual, that is, only when
the behavior is associated with the activated goal.
In summary, this evidence supports the idea that goal-directed behavior is
mentally represented and can be automatic. As we have argued, our
conceptualization of habits is based on these ideas. We conceive of habits
as a form of goal-directed automatic behavior. Habits are represented as
associations between goals and actions that allow the instigation of
automatic behavior on the activation of these goals by the environment. The
degree of "habitualness" of behavior is argued to be a consequence of the
frequency with which these goal-directed actions have been performed in
similar situations in the past.
In the present article, we report three experiments that were designed to
test the key assumptions about habits formulated earlier. In these
experiments, we studied cycling behavior among Dutch college students as an
example of habitual behavior. Our main purpose was to investigate whether,
and under what circumstances, bicycle use is capable of being directly
activated by travel goals such as "going to attend lectures at the
university" and "going shopping at the city center mall."
Experiment 1
In the first experiment, we tested the hypothesis pertaining to the key
assumption that habits can be seen as mental associations between travel
goals and (transportation) action and, hence, that these goals can activate
habitual transport behavior automatically. That is, we hypothesized that
habitual action is activated automatically on the instigation of a goal and
that such actions are not activated among people for which the behavior is
not habitual. A secondary aim was to show that goals can exert their
influence on habitual responses when people are not aware of the relation
between an earlier-primed goal and the habitual response.
In this experiment, habitual and nonhabitual bicycle users were primed or
not primed with travel goals (e.g., having to attend lectures) and then
asked to respond to the word bicycle after being presented with locations
(e.g., university) that corresponded to the earlier-activated travel goal.
Response latencies on the location-bicycle links served as the dependent
variable. We assumed that habitual bicycle users would show enhanced
accessibility and thus respond faster to the word bicycle than nonhabitual
bicycle users but only after being activated with the goal to travel. We did
not expect habitual and nonhabitual bicycle users to differ in their speed
of responding when the goal to travel was not activated.
Method Participants and design.
Fifty-four students at the University of Eindhoven participated in the
experiment, receiving 5 Dutch guilders (approximately $3) in return. Because
the experiment focused on bicycle use for short trips, only university
students were recruited who lived in or around Eindhoven and who owned a
bicycle. However, these participants varied in the frequency with which they
used their bicycles, which is crucial for obtaining different levels of
bicycle habit strength (i.e., nonhabitual vs. habitual). Participants were
randomly assigned to one of the two experimental conditions: a goal priming
condition and a no goal priming control condition.
Selection of materials.
Initially, a pilot study was conducted to obtain travel goals and
corresponding locations for which bicycle use constituted a realistic
transport mode option for all participants. Forty-two University of
Eindhoven students were presented with 60 different locations inside and
outside the city and were asked to assess the usefulness of four travel
modes (bicycle, bus, walking, and train) to these locations. We obtained 5
locations (e.g., a shopping area, called "heuvelgalerie," a sports center,
and a popular night-life area, called "stratumseind") for which a bicycle
constituted a realistic option for all students (percentages mentioning use
of a bicycle for travel to these destinations were nearly all 100%). Thus,
these locations represent travel destinations for which a bicycle is a
realistic travel mode. We subsequently asked the students to mention the
main reason to travel to each of the 5 locations. For each location, we
selected the most frequently mentioned reason, which provided us with
descriptions of five travel goals (e.g., shopping at the city center mall).
These five travel goals and corresponding travel destination-bicycle pairs
composed the targets of interest in Experiment 1 (as well as in the
subsequent experiments). In addition, 5 locations were obtained involving
trips for which a bicycle did not represent an option (e.g., Maastricht, a
city approximately 90 km from Eindhoven). Furthermore, for each of the other
three travel modes (walking, bus, and train), 5 location-option combinations
and 5 location-no option combinations were obtained. The latter 35
location-travel mode units (i.e., all except the first 5) served as filler
trials in the association task (described later).
Experimental task and procedure.
Participants worked in separate cubicles. Computers were used to run the
experiments and the computer program provided all instructions. As a cover
story, participants were told that they would take part in a study conducted
by the Department of Psychology and Language. Moreover, they were told that
the study consisted of three separate tasks designed by different department
research teams. In reality, the first task served as the manipulation phase
for goal priming, and the second task was designed to study the effects of
habit and goal priming on the speed of responding to the target bicycle
trips. Habit strength was assessed in the third task.
In the first task, announced as the "media and information use inventory
task," we were allegedly interested in several aspects of communication and
language (e.g., participants were asked to estimate how many hours a week
they watch television, whether they make use of video text, and so on). As
part of this task, half of the participants learned that they had to read
five different sentences that designated students' activities in daily life
(goal priming condition). Participants were told that we were interested in
how long it takes individuals to read each sentence. This information was
allegedly helpful for the purpose of designing new communication systems.
Participants were instructed to press a button after they had carefully read
the sentence. The five sentences actually described the five different
travel goals (e.g., going shopping at the city center mall) that
corresponded to the five travel locations (e.g., heuvelgalerie) obtained in
the pilot study and used later in the travel mode association task. It is
important to note that we did not provide the travel locations per se;
rather, we provided only the reason to go to each location (e.g., going
shopping). Each description thus consisted of one short sentence
representing a goal to travel to a destination with a certain transport
mode. The sentences were presented in random order in the center of the
computer screen. The other half of the participants (those in the control
condition) did not read these sentences as part of the inventory task and
thus started with the second task without being exposed to (or primed with)
the travel goals. 2
The second task was announced as an association task allegedly designed to
study relations between all sorts of locations and travel behavior.
Participants were told that 40 different location words would appear briefly
on the screen followed by a mode of transport. Their task was to indicate,
as quickly and as accurately as possible, whether the presented mode would
constitute a realistic means of transport for the previously presented
location. Furthermore, they were told that 200 different locations were
stored in the memory of the computer and that the computer would first
randomly select the 40 locations. An hourglass was displayed on the computer
screen, simulating the selection procedure. By using this (fake) selection
procedure, we hoped to further reduce any perceived connection between the
goal priming task and the association task. The 40 location-transport mode
trials were presented in random order and preceded by eight practice trials.
Thus, the five target location-bicycle trials were embedded in the filler
trials. This relatively large number of fillers was incorporated to create a
genuine multiple response situation, that is, to ensure that participants
had to respond to different locations with different travel modes.
An experimental trial consisted of the following sequence of events: (a)
presentation of a row of asterisks (i.e., fixation point) for 500 ms, (b)
presentation of the location word for 200 ms, (c) presentation of a row of
asterisks (i.e., postmask) for 100 ms, and (d) presentation of a travel
mode. Thus, the stimulus onset asynchrony (SOA; i.e., the time between
presentation of the location and the travel mode) was set at 300 ms. This
time interval was assumed to be too short to allow participants to form
expectancies and to implement strategic processes, and hence we measured
automatic responses this way (cf. Neely, 1991 ). The travel mode word
remained on the screen until the participant responded. Everything appeared
at the same location on the screen. Responses were collected from the PC's
keyboard. Participants pressed a key labeled yes or no. To obtain maximum
speed during the task, participants were instructed to keep their fingers
above the keys throughout the task. Response latencies were measured from
the onset of the travel mode, but participants had to respond within 3 s. If
participants completed a trial within the allotted time, the message "pay
attention" was presented for 2 s on the screen, indicating that the computer
would initiate the next trial. However, if no response was given after 3 s,
the message "please, respond faster" was presented, followed by the
announcement of the next trial. The dependent variable was the response
latency across the five target location-bicycle pairs.
With the third task, introduced as the leisure time inventory, we measured
habit strength. Participants' estimates of frequency of bicycle use in the
recent past for different trips were used to obtain information on their
bicycle habit strength. This operationalization coincided with Hull's (1943)
early work on habit formation in which he proposed that, as the number of
repeated pairings between a situation (e.g., travel location) and a response
(e.g., travel mode) increases, so does the strength of that association or
habit. Specifically, participants were presented with a sample of 10 travel
locations; for each location, they were asked to count the number of times
they had traveled there with their bike in the previous 2 weeks. The
presented locations consisted of the five target destinations and five other
locations situated near the target locations. Next, we averaged the
frequency estimates of bicycle use across the 10 destinations and, on the
basis of a median split, categorized participants as nonhabitual or habitual
in regard to bicycle use. Because this measurement procedure could serve as
a prime for locations and bicycle, we decided to measure habit strength at
the end of the experimental session, that is, after the goal priming and
association tasks (see Bargh & Chartrand, 1999 , on the subject of unwanted
effects of priming). Furthermore, as a means of attenuating possible
influences of the previous tasks on the estimates, participants were
explicitly instructed to be as accurate as possible in their recall ( Aarts
& Dijksterhuis, 1999 ; cf. Thompson, Roman, Moskowitz, Chaiken, & Bargh,
1994 ).
After the measurement of habit strength, participants were thoroughly
debriefed. The debriefing indicated that participants were unaware of the
hypotheses under investigation. Moreover, they did not perceive a link
between the two tasks and, therefore, perceived no connections between the
travel locations used in the goal priming and association tasks. Therefore,
we could conclude that we succeeded in creating two ostensibly unrelated
tasks.
Results and Discussion
All participants completed the five target trials within the allotted time.
Only latencies concerning yes responses across the five location-bicycle
pairs were included in the analyses (99.3% of all responses, a percentage
that might be expected on the basis of our pilot study). We computed the
average response latency across the five target location-bicycle trials for
each participant. Response latencies were subjected to a 2 (goal priming:
present vs. absent) * 2 (habit strength: nonhabitual vs. habitual)
between-subjects analysis of variance (ANOVA). The means for each cell in
the design are displayed in Table 1 .
As Table 1 shows, habitual participants' responses were slightly faster than
nonhabitual participants' responses, F (1, 50) = 3.89, p < .06. The main
effect of goal priming was nonsignificant, F (1, 50) = 0.10. More important,
however, the ANOVA revealed the predicted two-way interaction of habit
strength and goal priming, F (1, 50) = 5.87, p < .02. Planned comparisons
showed that, in the no goal prime condition, habitual participants' response
latencies did not differ reliably from nonhabitual participants' response
latencies, F (1, 50) = 0.11, ns. However, habitual participants' response
latencies were significantly faster than responses of nonhabitual
participants in the goal priming condition, F (1, 50) = 9.21, p < .005.
Planned comparisons between the no goal prime and goal prime conditions
yielded no reliable effect in the nonhabitual group, F (1, 50) = 2.30, ns,
and a marginally significant effect in the habitual group, F (1, 50) = 3.22,
p < .08. This second effect shows that habitual participants reponded faster
after being primed with travel goals.
The results of Experiment 1 supported our predictions. Habitual bicycle
users who were primed with travel goals showed faster responses than
nonhabitual bicycle users. Furthermore, this effect did not appear in the
absence of goal priming. As hypothesized, the data show that activation of
travel goals is required to reveal the mental accessibility of the habitual
travel behavior. In general terms, the automaticity of habitual behaviors is
conditional on the presence of a goal (cf. goal-dependent automaticity;
Bargh, 1989 ).
Experiment 2
The main goal of Experiment 2 was to compare habits with conscious planning.
We have argued that habits are represented as links between goals and
actions instrumental in attaining these goals and that these links are the
result of frequent coactivation of goal and action. However, there is
another way in which strong links between goals and actions are established,
namely through the formation of implementation intentions ( Gollwitzer, 1993
, 1996 ). These intentions take the form of "I will do x whenever situation
z occurs," and hence they link an action to a goal. Such intentions are
strategically formed by individuals to promote the initiation of
goal-directed action, especially when the performance of the action has to
be postponed and alternative actions can interfere. For instance, a person
intending to drive an alternative route when going home from work (a
nonhabitual goal-directed action that is often insufficiently implemented)
may increase the chances of indeed driving this route by planning.
Our hypothesis is that habits can be simulated by implementation intentions
but that habitual behavior does not profit from planning (for a similar line
of reasoning, see Bargh & Gollwitzer, 1994 ). This idea stems from the
assumption that strategic planning leads to essentially the same goal-action
links (e.g., "I will use the bicycle when having to attend lectures at the
university") as the ones we assume to represent habits. These intentionally
formed associations between goal and action are functionally equivalent to
habitual associations, and hence such actions may be automatically activated
as well ( Gollwitzer, 1993 ). The only difference is that habits are the
result of frequent past behavior, whereas links stemming from implementation
intentions are the result of conscious (recent) planning. However, and this
is important, because habits are already backed by strong links between
goals and action, it is anticipated that only nonhabitual individuals may
benefit from planning in the sense of enhanced accessibility or faster
responses to a travel mode after goal priming. Habitual people, on the other
hand, already possess these strong links and are expected not to benefit
from planning. Thus, by comparing habits and planning, we may be able to
show similarities between more chronic (habitual) and temporary primed forms
of automatized actions ( Bargh, Bond, Lombardi, & Tota, 1986 ; Fazio et al.,
1986 ; Srull & Wyer, 1986 ).
To test these ideas, we examined the interaction between bicycle habit
strength and the formation of implementation intentions (by means of
planning travel goals) using the same association task as in Experiment 1.
All participants were given the same five travel goals. However, two
different planning procedures were designed. Participants in the
experimental condition planned to use the bicycle for travel goals, whereas
participants in the control condition were required to plan a different
action, namely to repair a flat tire of a bicycle. This was done to ensure
that we did not make the term bicycle more accessible for one group than for
the other.
Experiment 2 served two further purposes. First, we tried to replicate the
effect of habit on associative strength between goals and actions under
conditions of goal activation by manipulating goals in a different manner.
In this experiment, participants learned that they actually had to travel on
finishing the experiment. Second, we included mediator variables to rule out
alternative explanations for the observed effects of habit and planning. For
instance, the two planning conditions could yield differences in strength of
travel goal activation or perceived feasibility of using a bicycle. As a
result, the effect of planning on response times could be attributable to
variances in these variables. Also, it can still be argued that the effects
of bicycle habit are guided by evaluative rather than goal-directed
automatic processes and, thus, may be mediated by perceived desirability of
(or attitudes toward) bicycle use. For the present purpose, three potential
mediators seemed relevant to test for mediator effects: strength of travel
goal instigation, perceived feasibility of bicycle use, and desirability of
bicycle use.
Method Participants and design.
University students were recruited who lived in or around the city of
Eindhoven and who owned a bicycle. Fifty-three undergraduates were randomly
assigned to one of the two experimental conditions: a related planning
condition and an unrelated planning condition (control). Habit strength
(nonhabitual vs. habitual) was measured. Participants received 5 Dutch
guilders (approximately $3) in return for taking part.
Experimental task and procedure.
Participants were told that they would take part in research conducted by
the Department of Psychology and Language and that three separate tasks
designed by different research teams had to be performed. The experiment was
run on computers. The computer program provided all instructions.
Participants worked in separate cubicles and were provided with three
consecutive tasks: a planning task, an association task, and the habit
measure.
In the first task, participants learned that the study involved the relation
between language and planning in daily life and that they would be requested
to plan the steps required to perform a certain task. Initially, all
participants were told that one of five different travel goals should be
personally attained by use of a bicycle after the experiment and that they
had to report on their experiences of the attained goal. Subsequently, they
were instructed to read the descriptions of five travel goals (see
Experiment 1). Participants learned that the decision as to which of the
goals they had to attain would be revealed to them after the experimental
session. Furthermore, to stress the importance of the task, we informed part
icipants that the study was designed to test whether the language used in
the planning task is affected when individuals do not know in advance which
goal they have to attain. At this point, the instructions for the two
conditions began to differ.
Half of the participants were then told to imagine having a flat tire that
had to be repaired. These participants were asked to plan the subgoals
required to repair the flat tire. This condition was referred to as the
unrelated planning condition and could be treated as a control condition,
because participants were requested to plan activities not directly related
to the attainment of the travel goals. They were handed a booklet containing
separate sheets listing five major subgoals of repairing a flat tire (e.g.,
searching for the repair kit and looking for a spot to fix the flat tire).
This was done to keep the working load and procedure similar to the related
planning condition (as described subsequently). Moreover, as a means of
ensuring that participants perceived the task as realistic, they were told
that the five subgoals emerged from the planning activities of students in a
previous study. For each subgoal, participants were requested to write down
when (time of the day), where (locating the spot to attain the subgoal), and
how (procedure) they would accomplish it.
In the experimental condition (referred to as the related planning
condition), participants were asked to plan the five travel goals. They were
also provided with a booklet, but this time the booklet contained the
assignment to plan the three steps of the five travel goals on separate
sheets. For each travel goal, participants were requested to write down when
(time of the day), where (locating the district of the destination, e.g., in
the city center), and how (the route of travel) they would accomplish it. It
should be noted that we asked participants to write down the district and
not the location itself. All participants were given 6 min to complete the
planning task.
After the planning task, as part of a larger questionnaire, participants
responded to the following three questions:
How likely do you believe it is that you will use the bicycle for one of the
five travel goals?
To what extent do you believe that, on average, the bicycle is a feasible
mode to use for the five travel goals?
To what extent do you believe that, on average, the bicycle is a desirable
mode to use for the five travel goals?
The first item served as a check of the strength of the travel goals, and
the last two items captured the perceived feasibility of and desirability of
(attitide toward) using the bicycle for the five travel goals. All items
were accompanied by unipolar 9-point response scales ranging from not at all
(1) to very much (9).
After completion of the planning task and questionnaire, participants
learned the same association task as in Experiment 1. Moreover, we used the
same cover story and the same fake selection procedure as in Experiment 1 to
further minimize the perceived connection between the target locations used
in the planning and association tasks. The dependent variable was the
response latency across the five target location-bicycle pairs.
After completing the association task, participants reported their frequency
estimates of bicycle use across 10 travel destinations (see Experiment 1).
On the basis of a median split, they were categorized as nonhabitual or
habitual in regard to bicycle use.
Debriefing indicated that participants had no idea about the true nature of
the experiment. First, all participants were unaware of the hypotheses under
investigation. Furthermore, although most participants expressed a belief
that the tasks were dealing with the same subject (i.e., the tasks focused
on travel behavior), none of them had actually noticed that some of the
locations in the association task were related to the travel goals in the
planning procedure. In summary, none of the participants indicated suspicion
as to the actual relation between the tasks. Not surprisingly, some
participants spontaneously asked which trip they were supposed to make,
revealing that we succeeded in the instigation of actual travel goals. Of
course, we informed all participants that the travel goals were provided
only to test our hypotheses, but we added that they were free in making
whatever trip they wanted to make.
Results and Discussion
All participants completed the five target trials within the allotted time.
Only latencies concerning yes responses across the five location-bicycle
pairs were included in the analyses (99.25% of all responses). We computed
the average response latency across the five target bike-location trials for
each participant. Response latencies were subjected to a 2 (planning:
unrelated vs. related) * 2 (habit strength: nonhabitual vs. habitual)
between-subjects ANOVA. The means for each cell in the design are displayed
in Table 2 .
As can be seen in Table 2 , habitual participants' responses were faster
than nonhabitual participants' responses, F (1, 49) = 4.66, p < .04, thereby
replicating the effect of habit strength on response latency in the goal
activation condition of Experiment 1. The main effect of planning was also
highly significant, F (1, 49) = 9.04, p < .005. Participants' response
latencies in the related planning condition were faster than participants'
responses in the unrelated planning condition. More important, the ANOVA
revealed the predicted two-way interaction of habit strength and planning, F
(1, 49) = 5.56, p < .03. Planned comparisons revealed that habitual
participants' response latencies were not affected by type of planning, F
(1, 49) = 0.02, ns. However, nonhabitual participants' response latencies
were significantly faster in the related planning condition than in the
unrelated planning condition, F (1, 49) = 13.95, p < .003. These latter
results show that planning (or formation of implementation intention)
facilitated the speed of nonhabitual participants' responses, whereas this
was not the case for habitual participants.
Potential mediators of the observed effects.
Three variables were measured after completion of the planning task. With
these measures, we wanted to rule out potential mediators. Specifically, we
measured strength of goal activation and perceived feasibility and
desirability of bicycle use. We first examined whether there were any
significant effects of planning and habit on the three mediators, and we
subsequently performed a 2 (planning: unrelated vs. related) * 2 (bicycle
choice habit strength: nonhabitual vs. habitual) between-subjects analysis
of covariance (ANCOVA) with the potential mediators as covariates.
Strength of travel goal activation.
Participants in all conditions believed to the same extent that they had to
use the bicycle for one of the five travel goals ( M = 7.55, SD = 1.88).
Nonhabitual participants did not differ significantly from habitual
participants, F < 1, and participants in the unrelated planning condition
did not differ from participants in the related planning condition, F < 1.
The interaction of habit strength and planning was also nonsignificant, F <
1.52. Note that the mean was well above the midpoint of the 9-point scale,
suggesting that, on average, participants took the manipulation of goal
instigation seriously.
Perceived feasibility of bicycle use.
There were no differences between conditions with respect to participants'
conviction that the bicycle was a feasible option to use for the travel
goals ( M = 8.36, SD = 0.86), as indicated by nonsignificant effects of
habit strength, planning, and their interaction (all F s < 1).
Perceived desirability of bicycle use.
Participants in the different conditions judged the bicycle equally
desirable ( M = 7.42, SD = 1.67). The main effects of habit strength and
planning and their interaction were nonsignificant (all F s < 1).
An ANCOVA with the three mediator variables as covariates yielded the same
pattern of significant results for habit, planning, and their interaction: F
(1, 46) = 5.54, p < .03; F (1, 46) = 9.63, p < .005; and F (1, 46) = 5.52, p
< .03, respectively (these effects were much the same as the ones resulting
from the original ANOVA's). Taken together, then, these analyses indicate
that the observed pattern of results is attributable neither to the
"magnitude" of goal activation nor to the perceived feasibility and
perceived desirability of using the bicycle as a mode of transport for the
five travel goals.
The first conclusion to be drawn is that, in Experiment 2, we replicated the
results of Experiment 1. After activation of a travel goal, habitual bicycle
users responded faster to bicycle trips than nonhabitual bicycle users.
Moreover, we obtained a reliable effect of planning. When travel goals were
furnished with related implementation intentions, participants responded
much faster than those who formed unrelated implementation intentions. The
latter group was much slower in linking the bicycle to the travel
destinations. As hypothesized, this effect was present only among
nonhabitual bicycle users. Habitual bicycle users did not benefit from
planning, which supports the idea that these participants already possessed
strong associations between travel goals and transport behavior. The
observed effects were not mediated by judgmental or strategic processes, as
indicated by the mediator analyses.
Together, the results of Experiments 1 and 2 suggest that habits are
mentally represented as associations between goals and actions. In addition,
we demonstrated that the links between goal and action can be simulated,
namely when the transport behavior is linked (by planning) to the travel
goal.
Experiment 3
It should be noted that, in Experiments 1 and 2, the strength of the
goal-action link was assessed with a task in which participants were
requested to associate a transport mode option with a briefly presented
travel location that corresponded to the earlier-activated travel goal.
However, although we assume that the inclusion of locations does not
influence the hypothesized process, the purpose of Experiment 3 was to
replicate the major finding of Experiments 1 and 2 by experimentally
manipulating the presence versus absence of the locations words.
Specifically, we sought to test whether the speed of responding to bicycle
use is enhanced as a function of the locations. Our conceptualization
entails that these locations are irrelevant and that it is the activated
goal that is responsible for the habitual response. However, one may argue
that the presented locations themselves can enhance the accessibility of the
habitual response. Experiment 3 was designed to rule out this possibility.
To assess the influence of the presence versus absence of location words, we
used a procedure largely similar to the sequential priming paradigm ( Fazio
et al., 1986 ; Neely, 1977 ). In the present procedure, participants were
briefly presented with a prime word and required to indicate whether a
subsequent word was a verb or not. In other words, we wanted participants to
produce a response of the form "word x is an action" after being briefly
presented with another word. Among these prime words, we included the five
location words as well as five neutral words. All of these words were
immediately followed by the word cycling (SOA: 300 ms). Goal activation was
experimentally manipulated, and habit strength was measured by self-reported
frequency of past bicycle use. If the results of Experiments 1 and 2 are not
the product of the presence of location words, as we hypothesize, habitual
bicycle users should respond faster to the word cycling than nonhabitual
bicycle users, irrespective of whether it was preceded by a location (but,
of course, dependent on goal priming). Hence, we predicted an interaction
between habit strength and goal activation. Conversely, if presentation of
location words does affect the responses, a three-way interaction effect
among goal activation, habit strength, and the presence versus absence of
location words should emerge.
Method Participants and design.
University students were recruited who lived in or around the city of
Eindhoven and who owned a bicycle. Eighty-nine undergraduates participated
in the experiment, receiving 7.5 Dutch guilders (approximately $4.50) in
return. They were randomly assigned to experimental conditions. Habit
strength (nonhabitual vs. habitual) was measured.
Experimental task and procedure.
Participants were informed that they would take part in research conducted
by the Department of Psychology and Language and that three separate tasks
designed by different research teams had to be performed. The experiment was
run on computers. The computer program provided all instructions.
Participants worked in separate cubicles and were provided with three
consecutive tasks: a goal activation task, a verb verification task, and the
habit measure.
As part of the introduction to the experimental session, participants
learned that one part of the study involved actual performance of behavior
and that they would be requested to perform a certain task. Specifically,
participants were told that one of five tasks should be executed at home
after the experiment and that they had to report on their experiences later.
Participants learned that the task they had to perform would be announced
after the experimental session. Subsequently, they were instructed to
carefully read the descriptions of the five respective tasks and to press a
button after they had read each description. The descriptions were presented
in random order in the center of the computer screen. Half of the
participants received five tasks unrelated to travel behavior (e.g., making
a telephone call or watching a movie on TV). This condition, labeled the
unrelated goal condition, could be treated as a control condition. In the
experimental condition (referred to as the related goal condition),
participants were exposed to the five travel goals used in the previous two
experiments.
After activation of the goal, participants were confronted with the verb
verification task, allegedly designed to study people's capacity to detect
different types of words. Participants were informed that there would be two
words presented one after the other on the screen and that they were to
press the yes or no button to indicate, as quickly and accurately as
possible, whether the second word was a verb or not. For the sake of
clarity, participants were told that the verbs designated mundane activities
people perform and that the nonverbs designated mundane objects. No
explanation or instructions were given regarding the prime words (cf. Bargh,
Chaiken, Govender, & Pratto, 1992 ).
An experimental trial consisted of the following sequence of events: (a)
presentation of a row of asterisks (i.e., fixation point) for 500 ms, (b)
presentation of the prime word for 200 ms, (c) presentation of a row of
asterisks (i.e., postmask) for 100 ms, and (d) presentation of a (second)
target word. The second word remained on the screen until the participant
responded. Everything appeared at the same location on the screen. Responses
were collected from the PC's keyboard. Participants pressed a key marked yes
or no. To obtain maximum speed during the task, participants were instructed
to keep their fingers above the keys throughout the task. Response latencies
were measured in milliseconds from the onset of the second word to the time
participants pressed a button. The interval between word trials was 2 s.
Participants were required to respond to two blocks of 80 word pairs. In
each block, 40 of the target words were verbs, and 40 were not verbs. Among
the 40 verbs, we presented the verb cycling five times. In one block (Block
1), the verb cycling was preceded by location words. In the other block
(Block 2), cycling was preceded by words not designating a location. Thus,
the word trials were identical for the two blocks with the exception that,
in one block, cycling was preceded by location primes (location-cycling
pairs), whereas this was not the case in the other block (no location-
cycling pairs). In each block, the other word pairs served as
fillers. Within blocks, the word trials were presented in random order and
preceded by eight practice trials. As a means of controlling for order
effects, participants were randomly assigned to one of the two order
conditions: Block 1-2 versus Block 2-1. Participants were allowed to rest
for 30 s between blocks. The dependent variable was response latency across
the five cycling trials.
After completion of the verb verification task, participants reported their
frequency estimates of bicycle use across 10 travel destinations. On the
basis of a median split, they were categorized as nonhabitual or habitual in
regard to bicycle use.
Debriefing indicated that participants had no idea about the true nature of
the experiment. First, all participants were unaware of the hypotheses under
investigation. Second, none of the participants indicated suspicion as to
the actual relation between the tasks.
Results and Discussion
Only latencies concerning yes responses across the cycling trials were
included in the analyses (98% of all responses). We computed the average
latency across the five neutral word-cycling trials and the five
location-cycling trials for each participant. Response latencies were
subjected to a 2 (goal activation: unrelated vs. related) * 2 (habit
strength: nonhabitual vs. habitual) * 2 (location prime: absent vs. present)
ANOVA. Goal activation and habit strength were between-subjects variables.
The ANOVA revealed that habitual bicycle users responded slightly faster to
cycling than the nonhabitual users, although this effect failed to reach
significance, F (1, 85) = 1.57, ns. However, the expected interaction
between goal activation and habit strength was significant, F (1, 85) =
5.35, p < .03. No other effects were reliable (including the three-way
interaction among goal activation, habit strength, and location prime, F <
1). Table 3 shows the Goal Activation * Habit Strength effect on latencies,
with the means collapsed across the two types of prime words.
To explore the nature of the Goal Activation * Habit Strength interaction,
we conducted planned comparison tests. These tests showed that, in the
unrelated goal activation condition, habitual participants' response
latencies did not differ reliably from nonhabitual participants' response
latencies, F (1, 85) = 0.58, ns. However, habitual participants' response
latencies were significantly faster than responses of nonhabitual
participants in the related goal activation condition, F (1, 85) = 6.34, p <
.02. Comparisons between the unrelated and related goal activation
conditions yielded no reliable effect in the nonhabitual group, F (1, 85) =
1.95, ns, and a marginally significant effect in the habitual group, F (1,
85) = 3.58, p < .07.
In summary, the results of Experiment 3 indicate that the automatic
activation of a habitual response is conditional on the presence of a travel
goal, thereby replicating the results of the previous experiments. This
interaction effect emerged regardless of whether travel locations were
presented or not presented, suggesting that the locations did not influence
the observed pattern of results. In addition, the fact that the locations
did not facilitate the speed of responding to the habitual travel mode
indicates that location words are not semantically related to cycling per
se. Hence, the presence of these location words in the previous experiments
cannot account for our findings.
General Discussion
In the introduction, we defined habits as goal-directed automatic behaviors.
Habits are mentally represented as associations between goals and actions.
These associations are shaped by frequent performance of actions and require
the activation of the goal to become manifest. The more frequently one
engages in a certain goal-directed behavior in similar situations, the
stronger the association becomes and, hence, the easier it is to
automatically elicit the behavior by activating the goal. These ideas were
supported in three experiments: Habitual responses were activated on the
instigation of a goal. Furthermore, these same responses were not activated
in the absence of the instigation of a goal, and they were not activated
when they were not habitual.
In Experiment 2, we tested an additional hypothesis. In this experiment, we
compared habitual goal-action associations with associations resulting from
conscious planning, and some participants formed implementation intentions
such as "I intend to perform action x whenever situation z occurs" (cf.
Gollwitzer, 1993 ; Gollwitzer & Brandst‰tter, 1997 ). These intentions are
supposed to lead to the formation of associations between goals and actions
and were hypothesized to be functionally equivalent to habitual
associations. Indeed, participants who formed implementation intentions
showed enhanced associative strength between travel goal and travel mode,
just as did participants for whom the use of this travel mode was habitual.
Our results indicate that this effect is manifest only when behavior is not
habitual. Participants who had already developed strong habits did not
benefit from planning. This makes sense because these participants already
possessed strong links between goals and behavioral responses. With the
formation of implementation intentions, people seem to be able to simulate
goal-directed automaticity in habitual behavior ( Gollwitzer, 1996 ; Orbell,
Hodgkins, & Sheeran, 1997 ). As such, these findings extend empirical
investigations on the similarities between chronic and temporary sources of
accessibility in the domain of social constructs and attitudes ( Bargh et
al., 1986 ; Fazio et al., 1986 ).
The theoretical significance of the present findings lies in the emphasis on
the cognitive mechanism that mediates the often empirically established
direct link between frequency of past behavior and later behavior (e.g.,
Aarts et al., 1998 ; Ouellette & Wood, 1998 ). Our results in regard to
automatic activation of associative links support the notion that frequently
pursued goals (or habits) are mentally represented in a way comparable to
other mental structures that are often repeatedly consulted and
automatically activated, such as stereotypes and attitudes (e.g., Bargh &
Gollwitzer, 1994 ; Devine, 1989 ; Fazio et al., 1986 ; Kruglanski, 1996 ).
Breaking Habits Through Planning
Our studies are confined to the domain of travel behavior. However, although
the present results cannot simply be generalized to all kinds of frequently
performed goal-directed behaviors, it is worthwhile to speculate on possible
ways in which planning can help to break harmful or undesirable habits
(e.g., drinking too much, eating the wrong foods, and applying social
stereotypes). After all, our results suggest that associations between goals
and actions that arise from frequent co-activation (i.e., habits) can be
simulated by planning. Is it possible to "replace" one association with a
different association?
Of course, planning can assist only on occasions in which the pursued goal
can indeed be achieved by multiple actions (either objectively or
subjectively assessed). If the habitual action is the only possible one,
there is no alternative action one can or will plan. But let us restrict
ourselves to goals that can be attained in multiple ways. As an example,
imagine a person who always uses a car to travel from home to work and
decides to use a bicycle instead. Can this person increase the probability
of performing the counterhabitual intended action by planning this action?
This is presumably dependent on the relative strength of the habitual
association and the association that is the result of planning. It is likely
that an association developed through planning can override a habitual
association if the former is stronger.
In recent treatments of action control (e.g., Norman & Shallice, 1986 ), it
has often been argued that if multiple behavioral representations are
activated, the one with the highest activation level will ultimately "win"
the fight for dominance and guide overt behavior. This means that if goal
activation leads to the activation of multiple behavioral representations
(i.e., the habitual one and the planned one), the one with the highest
activation level will guide overt behavior. The activation level of the
alternative representations, in turn, is conditional on the strength of the
associations between the activated goal and the different behavioral
representation. An interesting avenue for further research would be to
investigate how much planning is needed and what the content of the plan
should be before planned behavior can override habitual behavior. Our
results, which show that a little planning as to when, where, and how to
achieve a goal is enough to create associations that are as strong as
habitual associations (at least when the strength of these associations is
measured soon after planning), are promising in this regard.
Return of the Habit
The concept of habit has a long-established history in theory and research.
William James (1890) devoted an entire chapter to the concept of habit and
the utility of habitual behavior. Early sociologists conceived of habits as
behavioral patterns that serve as functional rules to control a society
(e.g., Durkheim, 1893 ; see also Camic, 1986 ). They used a broad definition
of habits to account for the stability of social institutions. The term
habit was also used by theorists writing on evolutionary processes, who
invoked the concept to denote the elementary behaviors of lower species. It
was in this sense that Darwin (1859) wrote of such things as the "feeding
habits of British insects." Darwin's work was related to the physiological
literature of that time, revealing an interest in the movements of
decapitated chickens, headless frogs, brainless cats, and the like. Later,
the term was used for reflex actions, which were conceived as motor
responses activated by nerve cells excited by stimuli external to the
organism ( Fearing, 1930 ).
In psychology, research on habits has long been dominated by behaviorist
approaches to learning theory, typically providing a rather "mechanistic"
account of the rise of behavioral responses (e.g., Skinner, 1938 ; Watson,
1914 ). That is, habits were merely conceptualized as automatic responses to
stimulus cues with no consideration of the intervention of mental processes.
However, this view was a bit rigid in the sense that it portrayed the actor
as a victim of habitual programs, which seems rather unlikely from a
functional point of view. As we have argued here, habits are goal directed,
and their activation is dependent on goals (see also Hull, 1931 ; Tolman,
1932 ). People automatically enter the garage and take their car or bicycle
only if there is a reason to do so (even though they may not be aware of
this reason at the time of action).
Lately, the term habit has rarely been used as an explanatory concept in the
psychology of human behavior (some exceptions to this rule can be found in
the introduction). In our view, understanding of mundane behavior and even
behavior in general can benefit greatly from the psychology of habits
Aarts, Paulussen, & Schaalma, 1997 ; Verplanken & Aarts, in press ). At
this time, many researchers consider human behavior as being guided by
reasoning. Accordingly, much effort is devoted to trying to explain various,
if not all, actions by studying the relations among attitudes, intentions,
and behavior. We believe, however, that although the emphasis on more
reason-based and deliberate processes is helpful for an understanding of
certain behaviors, it is not the only useful concept for insight into
behavior in general. Much of what people do in daily life becomes highly
automatized. In these cases, consciousness has delegated the onset and the
proceeding of behavior to the unconscious. We believe that in trying to
explain what mode of transportation people choose, what they eat, drink, and
smoke, and when and how they brush their teeth, habits will prove to be
conceptually very useful tools. We therefore hope that the present analysis
will contribute to the further development of the concept of habit.
References
Aarts, H. & Dijksterhuis, A. (1999). How often did I do it: Experienced ease
of retrieval and frequency estimates of past behavior. Acta Psychologica,
103, 77-89.
Aarts, H., Paulussen, T. & Schaalma, H. (1997). Physical exercise habit: On
the conceptualization and formation of habitual health behaviours. Health
Education Research: Theory and Practice, 12, 363-374.
Aarts, H., Verplanken, B. & van Knippenberg, A. (1998). Predicting behavior
from actions in the past: Repeated decision making or a matter of habit?
Journal of Applied Social Psychology, 28, 1356-1375.
Anderson, J. R. (1982). Acquisition of cognitive skills. Psychological
Review, 96, 369-406.
Anderson, J. R. (1993). Problem solving and learning. American Psychologist,
48, 35-44.
Austin, J. T. & Vancouver, J. B. (1996). Goal constructs in psychology:
Structure, process, and content. Psychological Bulletin, 120, 338-375.
Bargh, J. A. (1989). Conditional automaticity: Varieties of automatic
influence in social perception and cognition.(In J. S. Uleman & J. A. Bargh
(Eds.), Unintended thought (pp. 3-51). New York: Guilford Press.)
Bargh, J. A. (1990). Auto-motives: Preconscious determinants of social
interaction.(In R. M. Sorrentino & E. T. Higgins (Eds.), Handbook of
motivation and cognition (pp. 93-130). New York: Guilford Press.)
Bargh, J. A. (1997). The automaticity of everyday life.(In R. S. Wyer (Ed.),
Advances in social cognition (Vol. 10, pp. 1-61). Mahwah, NJ: Erlbaum.)
Bargh, J. A., Bond, R. N., Lombardi, W. J. & Tota, M. E. (1986). The
additive nature of chronic and temporary sources of construct accessibility.
Journal of Personality and Social Psychology, 50, 869-878.
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, 892-912.
Bargh, J. A. & Chartrand, T. L. (1999). A practical guide to priming and
automaticity research.(In H. Reis & C. Judd (Eds.), Handbook of research
methods in social psychology (871-932). New York: Cambridge University
Press.)
Bargh, J. A. & Gollwitzer, P. M. (1994). Environmental control of
goal-directed action: Automatic and strategic contingencies between
situations and behavior.(In W. D. Spaulding (Ed.), Nebraska Symposium on
Motivation (Vol. 41, pp. 71-124). Lincoln: University of Nebraska Press.)
Bargh, J. A., Gollwitzer, P. M., Chai, A. L. & Barndollar, K. (1999).
Bypassing the will: Nonconscious self-regulation through automatic goal
pursuit. (Manuscript submitted for publication)
Bargh, J. A., Raymond, P., Pryor, J. B. & Strack, F. (1995). Attractiveness
of the underlying: An automatic power -> sex association and its
consequences for sexual harassment and aggression. Journal of Personality
and Social Psychology, 68, 768-781.
Bentler, P. M. & Speckart, G. (1979). Models of attitude-behavior relations.
Psychological Review, , 452-464.
Camic, C. (1986). The matter of habit. American Journal of Sociology, 91,
1039-1087.
Carver, C. S. & Scheier, M. G. (1981). Attention and self-regulation: A
control-theory approach to human behaviors. (New York: Springer-Verlag)
Cohen, G. (1996). Memory in the real world. (London: Taylor & Francis)
Darwin, C. (1859). On the origin of species. (Cambridge, MA: Harvard
University Press)
Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and
controlled components. Journal of Personality and Social Psychology, 56,
5-18.
Durkheim, E. (1893). The division of labor in society ((G. Simpson, Trans.).
New York: Free Press)
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.
Fearing, F. (1930). Reflex action: A study in the history of physiological
psychology. (Baltimore: Williams & Wilkins)
Fishbein, M. & Ajzen, I. (1975). Belief, attitude, intention and behavior:
An introduction to theory and research. (Reading, MA: Addison-Wesley)
Fitts, P. M. & Posner, M. I. (1967). Human performance. (Monterey, CA:
Brooks/Cole)
Gallistel, C. R. (1985). Motivation, intention, and emotion: Goal-directed
behavior from a cognitive-neuroethological point of view.(In M. Frese & M.
Sabini (Eds.), Goal-directed behavior: The concept of action in psychology
(pp. 48-66). Hillsdale, NJ: Erlbaum.)
Geen, R. G. (1995). Human motivation. (Pacific Grove, CA: Brooks/Cole)
Gollwitzer, P. M. (1993). Goal achievement: The role of intentions.(In W.
Stroebe & M. Hewstone (Eds.), European review of social psychology (Vol. 4,
pp. 141-185). Chichester, England: Wiley.)
Gollwitzer, P. M. (1996). The volitional benefits of planning.(In P. M.
Gollwitzer & J. A. Bargh (Eds.), The psychology of action (pp. 287-312). New
York: Guilford Press.)
Gollwitzer, P. M. & Brandst‰tter, V. (1997). Implementation intentions and
effective goal pursuit. Journal of Personality and Social Psychology, 73,
186-199.
Gollwitzer, P. M. & Moskowitz, G. B. (1996). Goal effects on action and
cognition.(In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology:
Handbook of basic principles (pp. 361-399). New York: Guilford Press.)
Hasher, L. & Zacks, R. T. (1979). Automatic and effortful processing in
memory. Journal of Experimental Psychology: General, 108, 356-388.
Higgins, E. T. (1996). Knowledge activation: Accessibility, applicability,
and salience.(In E. T. Higgins & A. W. Kruglanski (Eds.), Social psychology:
Handbook of basic principles (pp. 133-168). New York: Guilford Press.)
Hull, C. L. (1931). Goal attraction and directing ideas conceived as habit
phenomena. Psychological Review, 38, 487-506.
Hull, C. L. (1943). Principles of behavior: An introduction to behavior
theory. (New York: Appleton-Century-Crofts)
James, W. (1890). The principles of psychology. (London: Macmillan)
Kruglanski, A. W. (1996). Goals as knowledge structures.(In P. M. Gollwitzer
& J. A. Bargh (Eds.), The psychology of action (pp. 599-618). New York:
Guilford Press.)
Locke, E. A. & Latham, G. P. (1990). A theory of goal-setting and task
performance. (Englewood Cliffs, NJ: Prentice Hall)
M‰ntyl‰, T. (1993). Priming effects in prospective memory. Memory, 1,
203-218.
Miller, G. A., Galanter, E. & Pribram, K. H. (1960). Plans and the structure
of behavior. (New York: Holt, Rinehart & Winston)
Mischel, W. (1973). Toward a cognitive social learning reconceptualization
of personality. Psychological Review, 80, 252-283.
Mook, D. G. (1996). Motivation: The organization of action. (New York:
Norton)
Neely, J. (1977). Semantic priming and retrieval from lexical memory: Roles
of inhibitionless spreading activation and limited-capacity attention.
Journal of Experimental Psychology: General, 106, 226-254.
Neely, J. (1991). Semantic priming effects in visual word recognition: A
selective review of current findings and theories.(In D. Besner & G.
Humpreys (Eds.), Basic processes in reading: Visual word recognition (pp.
264-336). Hillsdale, NJ: Erlbaum.)
Newell, A. & Rosenbloom, P. (1981). Mechanisms of skill acquisition and the
law of practice.(In J. R. Anderson (Ed.), Cognitive skills and their
acquisition (pp. 1-55). Hillsdale, NJ: Erlbaum.)
Norman, D. A. & Shallice, T. (1986). Attention and action: Willed and
automatic control of behavior.(In R. J. Davidson, G. E. Schwartz, & D.
Shapiro (Eds.), Consciousness and self-regulation: Advances in research and
theory (Vol. 4, pp. 1-18). New York: Plenum.)
Orbell, S., Hodgkins, S. & Sheeran, P. (1997). Implementation intentions and
the theory of planned behavior. Personality and Social Psychology Bulletin,
23, 945-954.
Ouellette, J. & Wood, W. (1998). Habit and intention in everyday life: The
multiple processes by which past behavior predicts future behavior.
Psychological Bulletin, 124, 54-74.
Powers, W. T. (1973). Behavior: The control of perception. (Chicago: Aldine)
Ronis, D. L., Yates, J. F. & Kirscht, J. P. (1989). Attitudes, decisions,
and habits as determinants of repeated behavior.(In A. R. Pratkanis, S. J.
Breckler, & A. G. Greenwald (Eds.), Attitude structure and function (pp.
213-239). Hillsdale, NJ: Erlbaum.)
Schank, R. C. & Abelson, R. P. (1977). Scripts, plans, goals, and
understanding. (Hillsdale, NJ: Erlbaum)
Skinner, B. F. (1938). The behavior of organisms. (New York:
Appleton-Century-Crofts)
Smith, E. R. & Lerner, M. (1986). Development of automatism of social
judgments. Journal of Personality and Social Psychology, 50, 246-259.
Srull, T. K. & Wyer, R. S. (1986). The role of chronic and temporary goals
in social information processing.(In R. M. Sorrentino & E. T. Higgins
(Eds.), Handbook of motivation and cognition (pp. 503-549). New York:
Guilford Press.)
Thompson, E. P., Roman, R. J., Moskowitz, G. B., Chaiken, S. & Bargh, J. A.
(1994). Accuracy motivation attenuates covert priming: The systematic
reprocessing of social information. Journal of Personality and Social
Psychology, 66, 474-489.
Tolman, E. C. (1932). Purposive behavior in animals and men. (New York:
Appleton-Century-Crofts)
Triandis, H. C. (1980). Values, attitudes, and interpersonal behavior.(In H.
E. Howe, Jr., & M. Page (Eds.), Nebraska Symposium on Motivation (Vol. 27,
pp. 195-259). Lincoln: University of Nebraska Press.)
Vallacher, R. R. & Wegner, D. M. (1987). What do people think they're doing?
Action identification and human behavior. Psychological Review, 94, 3-15.
Verplanken, B. & Aarts, H. (in press). Habit, attitudes, and planned
behaviour.(Is habit an empty construct or an interesting case of
goal-directed automaticity? In W. Stroebe & M. Hewstone (Eds.), European
review of social psychology. Chichester, England: Wiley.)
Watson, J. B. (1914). Behavior: An introduction to comparative behavior.
(New York: Holt)
1
In current research on the goal concept, different dimensions are postulated
on which goals may vary, such as level of abstraction, difficulty,
complexity, and temporal range (e.g., Austin & Vancouver, 1996 ; Gollwitzer
& Moskowitz, 1996 ). Of course, these goal dimensions are not necessarily
orthogonal. For the sake of argument, however, the focus here is on the
functionality of relatively short-term goals. For example, an
undergraduate's wish to travel to the university to attend lectures can be
seen as a short-term goal, one that may be functional in achieving a
long-term goal (e.g., earning money or becoming a professor).
2
One may remark that the goal priming condition and the control condition
differ on more aspects than simply goal priming (e.g., some people did spend
more time in the laboratory than others). However, such differences can have
an impact on the main effect of goal priming but not the hypothesized, and
critical, two-way interaction of habit strength and goal priming.
Try the links in the MadSci Library for more information on Medicine.