Fazio 2007

Social Cognition, Vol. 25, No. 5,2007, pp.
603-637
ATTITUDES AS OBJECT-EVALUATION
ASSOCIATIONS OF VARYING STRENGTH
Russell H. Fazio
Ohio State University
Historical developments regarding the attitude concept are reviewed, and set the
stage for consideration of a theoretical perspective that views attitude not as a hy-
pothetical construct, but as evaluative knowledge. A model of attitudes as ob-
ject-evaluation associations of varying strength is summarized, along with
research supporting the model's contention that at least some attitudes are repre-
sented in memory and activated automatically upon the individual's encountering
the attitude object. The implicationsof the theoretical perspective for a number of
recent discussions related to the attitude concept are elaborated. Among these is-
sues are the notion of attitudes as "constructions," the presumed malleability of au-
tomatically activated attitudes, correspondence between implicit and explicit
measures of attitude, and postulated dual or multiple attitudes.
For nearly 25 years now, a particular view of attitudes has formed the
underpinnings of my research program on the consequences of attitudes
for attention, categorization, judgment, and behavior. It was in 1982 that
my colleagues and I first proposed that attitudes can be viewed as ob-
ject-evaluation associations in memory (Fazio, Chen, McDonel, &
Sherman, 1982). The perspective has proven much more illuminating
(and occupied us for much longer) than we had envisioned at the time. It
has fostered examination of a wide variety of questions regarding atti-
tudes, and it continues to provide a valuable perspective for considering
new issues. In this article, I summarize the theoretical model and some
of the research findings that support it, as well as the perspective that the
Preparation of this article was supported by Grant MH38832 from the National Institute
of Mental Health.
The author thanks William Cunningham, Anna Han, Valerie Jefferis, Christopher Jones,
and Natalie Shook for their helpful comments on an earlier draft.
Please address correspondence to Russell H. Fazio, Department of Psychology, 1835
Neil Avenue, Ohio State University, Columbus, OH 43210-1287; E-mail:
[email protected].
603
604 FAZIO
model provides regardirig some recent discussions of attitudes as cor\-

structions, the presumed malleability of attitudes, correspondence
between implicit and explicit measures of attitude, and postulated dual
or multiple attitudes.
BACKGROUND
A VERY BRIEF HISTORY
Even as long ago as 1935, when Gordon AUport wrote his influential
chapter for the Handbook of Social Psychology, the term "attitude" already
had a rich history. AUport provides an intriguing historical account of
the term's early meanings. He notes that "attitude" had been used in the
arts to refer to the posture of a figure in a painting or sculpture. This con-
notation was evident in what was among the very first usages of the term
in experimental psychology—within the study of reaction time. In the
late 1800s, numerous studies revealed that participants who were men-
tally prepared to press a telegraph key upon a signal were able to re-
spond more quickly than those whose attention was focused on the
incoming signal itself. The importance of this state of preparedness,
sometimes referred to as a "task-attitude" or "mental set," was
demonstrated repeatedly in studies of perception and memory.
Beginning with these early tracings of the attitude concept, AUport
(1935) reviewed a large number of definitions that had been offered. His
analysis of their common threads and tbe debates they had inspired led
AUport to propose what is certainly the most widely known of tbe early
definitions of attitude. "An attitude is a mental and neural state of readi-
ness, organized througb experience, exerting a directive or dynamic in-
fluence upon tbe individual's response to all objects and situations witb
which it is related" (p. 810). Thus, the early references to posture and
readiness were featured in AUport's definition. However, the definition
seems to go much further when it posits that this state of readiness influ-
ences a broad array of responses, which raises an important question. If
no such influence is observable, then is there no attitude?
Therein lies what later came to prove especially controversial about
the attitude concept. AUport's definition, like many others in the litera-
ture, presupposes behavioral responses that are consistent with the atti-
tude—that is, the very definition calls for behavioral consistency.
Without evidence of such consistency, the defining criteria of an attitude
have not been achieved. In the late 1960s and early 1970s, this assump-
tion regarding behavioral correspondence began to be examined—and
seriously challenged. Wicker's (1969) influential review of the available
ATTITUDES AS OBJECT-EVALUATION ASSOCIATIONS 605
evidence suggested that verbal reports of attitudes bore Uttle relation to

subsequent behavior. Bem (1972) espoused a view of attitudes, not as
potent determinants of responses, but as epiphenomenal explanations
for past behavior. Many attitude researchers rose to the challenge and
demonstrated that sizable relations could be observed between verbal
reports of attitude and subsequent behavior under certain conditions.
This is not the time to review those findings. Many pertinent reviews are
available (e.g., Fazio & Roskos-Ewoldsen, 2005; Wallace, Paulson, Lord,
& Bond, 2005; Zanna & Fazio, 1982), aU of which clearly confirm that
attitudinal reports can sometimes prove strongly predictive of behavior.
More relevant to the present concerns is what this controversy regard-
ing the attitude-behavior relation implies for a definition of attitude.
The fact that the answer to the question, "Is there a relation between atti-
tudes and behavior?" is a resounding "sometimes" (see Zanna & Fazio,
1982 for an overview of generations of research concerning attitude-be-
havior consistency) suggests that assumptions regarding behavioral
correspondence should not be a part of one's very definition of attitude.
A definition that includes such a criterion is open to circularity; it essen-
tiaUy "defines away" the question of attitude-behavior consistency. If
behavioral correspondence is not observed, there is no attitude. Hence,
the very definition does not aUow for the possibility that a multitude of
factors may play a role in determining whether an individual's attitude
is evident behavioraUy in a given situation. Instead of prejudging the at-
titude-behavior relation, any definition of attitude should leave the
matter open to theoretical and empirical investigation (Zanna &
Rempel, 1988). This reasoning was very central to the model of attitudes
that I came to pursue and will expUcate shortly.
ATTITUDES: HYPOTHETICAL CONSTRUCTS?
There is, however, an additional sense in which a definition of attitude

that postulates influences on observable responses at its very core is
problematic. Is an "attitude" being viewed as a hypothetical construct?
If so, in what sense is it "hypothetical": hypothetical in the sense that it is
not directly observable, or in the sense of being merely a convenient con-
ceptual abstraction for describing observed covariance? Does an atti-
tude "exist" as a latent variable that is presumed to play a mechanistic
role when the individual encounters or considers the attitude object, or,
is it no more than a useful codeword for scientific discourse?
Classic discussions of the distinctions that might be drawn between
various forms of hypothetical constructs and intervening variables in-
clude MacCorquodale and Meehl (1948) and Hilgard (1958). However,
606 FAZIO
the distinction that I wish to emphasize here is most clearly articulated

by DeFleur and Westie (1963) in their influential article entitled "Atti-
tude as a Scientific Concept." They contrast two different "types of latent
process conceptions [of attitude]: (1) those which impute the empirical
existence of a hidden mechanism, and (2) those which postulate a hypo-
thetical mediating variable which is not regarded as having empirical
referents, counterparts, or existence but which is simply a construction
which serves as a convenient tool for analysis" (p. 24). To exemplify the
latter category, DeFleur and Westie point to Green's (1954) conception:
Like many psychological variables, attitude is a hypothetical or latent

variable, rather than an immediately observable variable. The concept
of attitude does not refer to any one specific act or response of an indi-
vidual, but is an abstraction from a large number of related acts or re-
sponses. For example, when we state that individual A has less a
favorable attitude toward labor organizations than individual B, we
mean that A's many different statements and actions concerning labor
organizations are consistently less favorable to labor than are B's com-
parable words and deeds. We are justified in using a comprehensive
concept like attitude when the man's related responses are consistent
(p. 335).
Conceptualized in this way, then, "attitudes" do not exist as entities

within the individual. Moreover, scientific reference to "attitude" is not
even appropriate without evidence of consistent verbal and behavioral
responding. Thus, this approach once again raises the issue of prejudg-
ing the attitude-behavior relation and "defining away" the problem of
inconsistency. To the extent that inconsistency is observed between ver-
bal reports and behavior, there is no "attitude" (or at least should not be),
not even in the mind of the scientist.
In contrast, DeFleur and Westie (1963) classify AUport's definition in
the first of the two categories they delineated. Because Allport refers to at-
titudes as "exerting a directive influence," they characterize his definition
as including "the additional idea that the individual's behavior is some-
how 'shaped,' 'guided,' or 'mediated' by some underlying process [empha-
sis in original]...some 'inner mechanism,' some unobservable 'something'
that constrains, influences, mediates, or otherwise determines that consis-
tency will appear among the individual's responses to the attitude stimu-
lus" (p. 23). As the tone implies, DeFleur and Westie are critical of the
approach, emphasizing two reasons for their skepticism. The first con-
cerns conceptual imprecision and the lack of relevant empirical evidence:
"...the latent process conception of attitude must be entertained as most tentative
[emphasis in original]. In posing such an entity it is not precisely clear, at
least at the present time, whether the advocates have named a discovery,
or discovered a name. Until the mechanisms involved in this internal la-
tent process can be described more fully, the latent variable must remain
'an unknown something'" (p. 24). The second point of skepticism con-
cerns what DeFleur and Westie called "the fallacy of expected correspon-
dence" (i.e., the presumption that both verbal reports and actions would
be affected by the same latent process). They questioned this presumption
on the basis of emerging evidence of inconsistency between
verbalizations and other forms of action toward the attitude object (e.g.,
LaPiere, 1934).
Although the precise meaning that is intended is sometimes not clear,
references to attitudes as hypothetical constructs continue to appear in
the contemporary literature. For example, Greenwald and Nosek (in
press) assert: "Attitudes, like other psychological constructs, are hypo-
thetical and unobservable." Schwarz «& Bohner (2001) are more explicit
when they characterize attitudes as a scientific reification: "Attitudes are
a hypothetical construct, invented by researchers to account for a body
of phenomena" (p. 438). In my view, and as I hope will become all the
clearer shortly, this lingering ambiguity about the sense in which the at-
titude concept might be viewed as a hypothetical construct has contrib-
uted to the very confusion that inspired this Special Issue in the first
place. To preview the argument briefly, a view of attitudes as hypotheti-
cal abstractions that the scientific observer imputes upon targets of ob-
servation invites neglect of critical questions regarding the acquisition
and use of evaluative knowledge. A functional and adaptive system for
judgment and decision making requires that individuals leam from
prior experience, that the acquired evaluative knowledge be repre-
sented in memory, and that it be activated efficiently when it may prove
useful in later situations. The science is not positioned to ask these ques-
tions if attitudes are viewed as a mere reification. Unless attitudes are
thought to have some form of existence within the individual, we fail to
acknowledge the importance of prior learning for current evaluative
judgment.
LESSONS LEARNED?
Awareness of this background regarding the history of the attitude con-

cept should sensitize any attitude theorist and researcher to certain is-
sues. An adequate conceptualization of attitudes must satisfactorily
address the matters that have proven pivotal over the last 75 years or so.
To ignore the lessons that have been (should have been) learned detracts
from the cumulative advances and progress that are so evident in the lit-
608 FAZIO
erature. First, there appears little value to defining attitudes in such a

way as to rest the concept on the fulcrum of behavioral consistency. As
Zanna and Rempel (1988) argued, definitions that presuppose atti-
tude-behavior consistency dismiss important questions that should be
open to theoretical consideration and empirical investigation. Second,
and relatedly, DeFleur and Westie's (1963) "fallacy of expected corre-
spondence" must be avoided. Any theoretical model of attitudes will
need to acknowledge that behavior sometimes corresponds to verbal re-
ports of attitude and sometimes does not. Third, unless the idea that atti-
tudes can exert influence is to be abandoned, an adequate
conceptualization of attitudes will need to embodied within a model of
the process(es) by which such influence occurs. lust as DeFleur and
Westie (1963) argued, imobservable "somethings" that remain unspeci-
fied hidden mechanisms of influence should not be tolerated. Moreover,
the specified mechanisms should contribute to our understanding of
why measures of attitude sometimes prove predictive of later behavior
and sometimes not.
ATTITUDES AS OBJECT-EVALUATION ASSOCIATIONS
With this background in mind, let me summarize the view of attitudes

that has been the focus of the research my colleagues and I have con-
ducted. The model was first proposed by Fazio et al. (1982), tested in a
variety of ways in subsequent research over the next decade, and most
systematically delineated in a review chapter by Fazio (1995). In brief,
the model views attitudes as associations between a given object and a
given summary evaluation of the object—associations that can vary in
strength and, hence, in their accessibility from memory.
Thus, attitudes are defined as summary evaluations. However, this
should not imply that attitudes are necessarily cold, belief-based judg-
ments of favorability. The term "evaluation" is used broadly to include
not only analytic assessments but also "hot" affective reactions. Like
Zanna and Rempel's (1988) formulation, the model views these
evaluative summaries as potentially stemming from beliefs, affect,
and/or behavioral information. The attitude may be based on appraisals
of the attributes that characterize the object, as in expectancy-value
frameworks. It can stem from emotional reactions that the attitude object
evokes, as in the case of conditioned emotional responses. It can be based
on one's past behaviors and experiences with the object. Or, it can also be
based on some combination of these potential sources of evaluative in-
formation. Relatedly, the attitude may be the outcome of a rather passive
associative learning process (e.g., De Houwer, Thomas, & Baeyens,
2001; Olson & Fazio, 2001). It may be a relatively simple inference from
attitudinally diagnostic information, including freely chosen behavior
(Bem, 1972), proprioceptive feedback (e.g., Strack, Martin, & Stepper,
1988), perceptual fluency (e.g., Reber, Winkiehnan, & Schwarz, 1998), or
ease of retrieval (e.g., Brinol, Petty, & Tormala, 2006). Or, the attitude
may represent the summary product of a more active process of proposi-
tional reasoning that involves careful scrutiny of the validity of informa-
tion regarding the attitude object (e.g., Deutsch & Strack, 2006;
Gawronski & Bodenhausen, 2006; Petty & Cacioppo, 1986; see also
Gawronski & Bodenhausen, this issue). The processes by which the atti-
tude was formed and its informational basis may have implications for
the resulting strength of the object-evaluation association and, hence,
the accessibility of the attitude from memory (see Fazio, 1995; Petty,
Haugtvedt, & Smith, 1995). However, regardless of how it was formed,
the attitude is defined as the summary evaluation, devoid of any
presumptions regarding behavioral consistency.
Importantly, the model of attitudes is very explicit with respect to the
matter of attitudes being viewed as hypothetical constructs invented by
the scientific community. By viewing attitudes as associations in mem-
ory, the model obviously is positing that attitudes can "exist"—that is, if
the individual has formed any such evaluative association, then the atti-
tude is represented in memory. Attitudes, then, are simply a form of
knowledge—evaluative knowledge, more precisely—and are repre-
sented in memory just as any other knowledge is represented. Just as we
associate "bread" with "butter" and "doctor" with "nurse," we can asso-
ciate "yuck" with cockroaches, or a sense of ecstatic delight with
chocolate or single-malt scotch.
Thus, according to this view, attitudes are no more hypothetical than
any other form of knowledge. We do not speak of a person's knowledge
that ants are small, but strong (i.e, the associations between ant and
small and between ant and strong) as a hypothetical construct. Obvi-
ously, there is an element of unobservability here, just as is true for atti-
tudes. As scientific observers, we have difficulty discerning whether the
individual possesses knowledge regarding the size of ants. We need to
pose a verbal question or observe a relevant behavior, such as the indi-
vidual's muscular preparation when asked to pick up a cardboard box in
which an ant has been placed. Likewise, we cannot see the individual's
attitude toward single-malt scotch directly, but we can inquire, we can
observe, and we can arrange tests assessing the individual's liking. In
both cases, we can reach the conclusion that the knowledge, evaluative
knowledge in the latter case, exists.
Another critical aspect of the model is that it highlights variability in
the strength of the object-evaluation association. Some attitudes for
610 FAZIO
some individuals are characterized by strong associations. Others are

weaker. For still others, no summary evaluation may even be available
in memory, possibly because the object falls in a domain that carries no
hedonic significance for the particular individual. So, a college basket-
ball fan will have strong evaluative associations to a wide number of
teams, coaches, and players, will develop attitudes toward new players
readily, and will experience affective reactions to certain teams' wins
and losses. On the other hand, for other individuals, evaluative knowl-
edge regarding all but the most heralded entities in the college basket-
ball scene will be absent. In many cases, no a priori evaluation will be
available in memory. Some individuals may exhibit a greater propensity
to form attitudes across a variety of domains than other individuals do,
as is apparent, for example, with respect to the individual difference
measure of the need to evaluate Qarvis & Petty, 1996). However, sub-
stantial variability across domains is to be expected, just as a function of
people's interests. Sports are hedonically relevant for some people but
not all. Politics can be rife with evaluative associations for some, but
again not for all. Some people have sophisticated evaluative knowledge
regarding cooking techniques, recipes, and spices, whereas for others
these entities are tinlikely to be associated with anything other than
indifference and may even be complete unknowns.
Thus, across people and objects, attitudes necessarily vary with re-
spect to their associative strength. Building upon a distinction high-
lighted by Converse (1970), we have referred to this variability as the
attitude-nonattitude continuum (Fazio, Sanbonmatsu, Powell, & Kardes,
1986). At the nonattitude end of the continuum is the case of the individ-
ual lacking any a priori evaluative association to the object. Due to its
novelty or its basis in a sphere of indifference for the individual, no rele-
vant attitudinal representation is available in memory. As we move
along the continuum, an evaluation is not only available but also is more
strongly associated with the attitude object. Thus, for some people, the
object-evaluation association may be such that the object is atti-
tude-evoking. The associated evaluation is activated automatically
from memory upon mere observation or mention of the attitude object.
Given that the term "automatic" has come to be used in multiple
senses (see Bargh, 1994; Moors & De Houwer, 2006; Shiffrin & Dumais,
1981 for relevant discussions), it will be useful to spend a moment expli-
cating just what is meant by "automatic" attitude activation within our
theoretical model. From the model's perspective, the key feature is
inescapability. Encoimtering the attitude object activates the associated
evaluation without the individual's intent, and does so even if the indi-
vidual is attempting to engage in some other activity. Hence, what is
most relevant to our use of the term is Shiffrin and Dumais's (1981) char-
acterization of automatic as "all processes whose initiation the subject

cannot control"—ones that occur "whenever a given set of external initi-
ating stimuli are presented, regardless of a subject's attempt to ignore or
bypass the distraction" (p. 117). When the object is encountered, the atti-
tude is activated, not necessarily to a level of awareness, although that is
possible, but to a level of accessibility that increases the likelihood that
the evaluation will influence the interpretation of subsequent
information (Bruner, 1957; Higgins, 1996).
Any such automatic activation of the attitude is viewed as playing a
critical role in the process by which an attitude may exert influence on in-
formation processing, judgment, and behavior. Indeed, from its outset,
this theoretical conceptualization of attitudes has been embodied within
a model attempting to specify the process(es) by which attitudes
"guide" behavior (for early discussions, see Fazio, Powell, Herr, 1983;
Fazio, 1986). Thus, the MODE Model (Motivation and Opportunity as
DEterminants; see Fazio, 1990; Fazio & Towles-Schwen, 1999) postu-
lates that attitudes can influence behavior in a relatively spontaneous
fashion, to the extent that they are activated automatically from memory
upon the individual's encountering the attitude object. By biasing per-
ceptions of the object in the immediate situation, the activated attitude
can prompt attitudinally consistent behavioral responses. Moreover,
this can occur without the individual necessarily engaging in any
effortful reflection regarding his or her attitude toward the object and
without any necessary awareness on the part of the individual that his or
her activated attitude has biased construal of the object.
More will be said about attitude-to-behavior processes in a later sec-
tion of this article. For the time being, what is important to note is that the
model of attitudes as object-evaluation associations of varying strength
absorbs the earlier-mentioned "lessons." It does not presuppose atti-
tude-behavior consistency as part of its very definition. Thus, it avoids
DeFleur and Westie's (1963) "fallacy of expected correspondence." The
model is also clear about the status of attitude as evaluative knowledge,
akin to any other knowledge in terms of its representation in memory.
Yet, the model specifies mechanisms by which this representation can
influence behavior and points to the strength of the object-evaluation as-
sociation in memory as a determinant of the extent to which any such in-
fluence is likely. Thus, the model provides a perspective as to why and
how attitude-behavior consistency varies.
Although the matter is somewhat of a digression, it will be useful to
highlight one additional point before concluding this section. The use of
the term object-evaluation association is not intended to commit or re-
strict the theoretical perspective to an associative network model of
memory. Of course, one can view an attitude as an evaluative tag that is
612 FAZIO
lirvked to the representation of the attitude object in memory and that has
the potential to receive some activation whenever the object itself is acti-
vated. This certainly was the dominant approach to memory at the time
the model was first proposed. However, the theoretical perspective is
also completely consistent with the connectionist approaches that have
emerged since that time. An attitude can be viewed as a pattern of activa-
tion that emerges as a consequence of the cormection weights that a sys-
tem has learned to apply to a set of input units (see also Conrey & Smith,
this issue). Indeed, various colleagues and I have considered attitudes
and attitude accessibility from the perspective of connectionist models
(Smith, Fazio, & Cejka, 1996; Fazio, Eiser, & Shook, 2004). Moreover, we
have proposed what is one of the most fully detailed connectionist mod-
els of attitude development available in the literature and have con-
ducted numerous simulations employing the network architecture
(Eiser, Fazio, Stafford, & Prescott, 2003; Eiser, Stafford, & Fazio, 2007).
Whereas an associative network approach views the attitude as a dis-
crete, symbolic unit that is stored and capable of activation, the
connectionist approach views the attitude as a pattern of activation that
is generated by the learned connection strengths between units. What is
important from the present perspective is that both approaches allow for
an evaluation to be evoked when an object is encountered and both
maintain that the likelihood of such attitude evocation is a function of
past learning. Indeed, it is the prior learning that constrains a
connectionist network to reliably produce roughly the same pattern of
activation in response to two similar sets of input iinits.
(SOME) ATTITUDES ARE REPRESENTED IN MEMORY
Whether one pursues an associative network or a connectionist ap-

proach, there exists an abundance of empirical evidence that attitudes
are represented in memory and capable of automatic activation upon
presentation of the attitude object. Briefly reviewing such evidence, be-
fore turning to a direct discussion of some issues that have arisen in the
last few years regarding the attitude concept, will be valuable. The evi-
dence that I wish to highlight has one central, and I believe compelling,
feature. During a task or in a situation in which individuals are pre-
sented with attitude objects but provided with no reason to actively con-
sider their attitudes toward the objects, their performance is affected by
their attitudes. Thus, presentation of those objects must have evoked the
attitudes despite their irrelevance to the immediate task demands.
The most well-known of such research paradigms is the evaluative
priming procedure in which attitude objects are presented as primes un-
related to the participants' primary task of identifying the cormotation of

target adjective (Fazio et al., 1986). Responding is facilitated by the pre-
sentation of a prime that is evaluatively congruent with the target adjec-
tive. This is not the time to review what is now a very extensive literature
(see Fazio, 2001; Klauer & Musch, 2003 for reviews). Instead, I will sim-
ply highlight a few findings that are especially relevant to the argument
that some preexisting attitudinal representation is evoked automatically
and that the likelihood of such activation occurring is a function of the
strength of the object-evaluation association.
Early research concerning the attitude-nonattitude continuum was
premised on the notion that, when directly queried about their attitudes,
individuals with stronger associations would be able to respond more
quickly than those with weaker or absent associations. Thus, latency of
response to a direct attitudinal query was employed as a dependent
measure assessing such attitude accessibility. T'hese experiments suc-
ceeded in identifying a number of variables that decreased the time that
individuals required to report their attitudes, including direct experi-
ence with the attitude object (Fazio et al., 1982), inference from free-cho-
sen as opposed to manded behavior (Fazio, Herr, & Olney, 1984), and
most relevantly, repeated attitudinal expression. Noting and rehearsing
the association in an early phase of the experiment enhanced the speed
with which individuals could later indicate their attitudes (Fazio et al.,
1982; Powell & Fazio, 1984). Numerous later experiments have em-
ployed both this latency measure of associative strength and the attitude
rehearsal manipulation within the evaluative priming procedure. Prim-
ing effects have been stronger for attitude objects involving stronger
object-evaluation associations (see Fazio, 2001 for a review).
Just as object-evaluation associations can be strengthened by re-
hearsal, they can be weakened by repeated nonuse. In a recent series of
experiments (Sanbonmatsu, Posavac, Vanous, Ho, & Fazio, 2007), atti-
tude objects from one of two sets that involved strong evaluative associ-
ations (e.g., holiday, disease) were presented repeatedly (e.g., 40
presentations of each object name) in a task in which the automatically
activated attitude was irrelevant. Participants were required only to re-
cite the object word aloud as it gradually emerged from a rectangular
block of masking dots. Presumably, attitudes were automatically acti-
vated upon presentation of the objects, at least early in the task, but they
were not of any use. When participants later underwent a priming pro-
cedure that included these words, as well as the set of control objects,
they displayed less evidence of attitude activation in response to the
words that had been repeatedly presented in the earlier task. Such con-
sequences of repeated nonutilization of the attitude were observed even
when the two phases of the experiment were separated by a 24-hour de-
614 FAZIO
lay. Thus, experimental manipulations that serve to strengthen, or to

weaken, object-evaluation associations have a corresponding effect on
the likelihood of automatic attitude activation.
A conceptually parallel finding regarding the moderating role of atti-
tude strength has been observed at the level of an individual difference.
Hermans, De Houwer, and Eelen (2001) compared individuals who had
scored extremely high or extremely low on Jarvis and Petty's (1996)
Need to Evaluate Scale (NES). The scale assesses the extent to which in-
dividuals report having attitudes and evaluating objects regularly. The
NES moderated the magnitude of the typical evaluative priming effect,
with the higher-NES participants displaying greater evidence of
automatic attitude activation.
Importantly, evidence that at least some attitudes are represented in
memory and can be evoked automatically extends well beyond the
evaluative priming paradigm. A variety of findings point to an influence
of attitudes on tasks during which individuals experience no need to
consider their evaluations of the presented objects. For example,
Roskos-Ewoldsen and Fazio (1992) observed effects of attitude accessi-
biUty on a variety of tasks related to visual attention. More atti-
tude-evoking objects—whether measured via latency of response to a
query or experimentally manipulated—attracted more attention. When
presented as distractors, attitude-evoking objects were more likely to be
noticed incidentally in one such experiment, and more likely to interfere
with performance of a visual search task in another.
In research concerning the construal of multiply categorizable objects.
Smith et al. (1996) demonstrated that categories made more atti-
tude-evoking by virtue of an attitude rehearsal manipulation were more
influential. Tlie more attitude-evoking categories were more likely to
govern interpretation of the multiply categorizable objects. When the ob-
jects (e.g., sunbathing, Pete Rose) were presented as cues in a cued-recall
test, they more effectively cued the categories for which object-evaluation
associations had been strengthened than categories from a control set
(e.g., cancer versus beach, and baseball player versus gambler, depend-
ing upon which had been assigned to the attitude rehearsal task versus
the control task). In one experiment, this effect was observed despite the
imposition of a delay of one fuU week between the attitude rehearsal ma-
nipulation and the cued recall test. Recent research by Ferguson, Bargh,
and Nayak (2005) has yielded related evidence regarding the impact of
automatically activated attitudes on construals of subsequently pre-
sented stimuli. Although associative strength was not included as a factor
in the design, positively and negatively valued attitude objects were se-
lected as stimuli on the basis of normative data showing them to involve
relatively strong object-evaluation associations. Subliminal presentation
of these object names influenced how subsequent ambiguous stimuli

were categorized. The ambiguous objects were more likely to be classified
as relating to a more positive category when they had been preceded
subliminally by a positively valued object.
Conceptually parallel effects have been observed with respect to pho-
tos of multiply categorizable persons. In a two-session experiment,
Fazio and Dunton (1997) first assessed participants' automatically acti-
vated racial attitudes via an evaluative priming procedure. As is typical
in such research, a full range of attitude estimates were obtained. Some
participants showed evidence of automatically activated negativity in
response to Black faces; for others, positivity was evoked; and for others,
neither occurred. In the second session, participants made similarity
judgments of pairs of stimulus persons who varied in race, gender, and
occupation. Those similarity judgments were more strongly influenced
by race for those participants for whom race was more attitude-evoking,
in either the positive or the negative direction. In other words, a
curvilinear, U-shaped relation was observed between the estimates of
automatically activated racial attitudes and categorization by race. Ap-
parently, individuals for whom race was attitude-evoking were more
likely to have their attention drawn to the target's skin color and, as a
result, to use race (as opposed to gender or occupation) more heavily as a
basis for judging similarity.
As an additional example of research demonstrating that attitudes can
be evoked and exert an influence in situations in which they are not rele-
vant to task demands, we can consider experiments that have examined
the effects of accessible attitudes on sensitivity to changes in the appear-
ance of the attitude object (Fazio, Ledbetter, & Towles-Schwen, 2000).
This work involved "head-and-shoulders" photos of undergraduates.
At the beginning of the experiment, participants were exposed to these
photos multiple times and each time expressed their attitudes (how at-
tractive they personally found the photo) or engaged in a control task
(estimating the individual's height). A subsequent task involved the
presentation of these original photos, as well as computer-generated
morphs representing varying degrees of change from the original pho-
tos. Participants were to note whether any given photo was exactly the
same as one presented earlier or different in any way. The two condi-
tions did not differ, in either error rates or latencies, when judging the
unchanged stimuli, thus establishing that the original photos were en-
coded equally well. However, the conditions did differ with respect to
the morphs, especially with respect to those that more closely resembled
the originals and, hence, were likely to evoke the previously rehearsed
attitude. The attitude rehearsal participants had relatively more diffi-
culty detecting the presence of change (either more errors or longer la-
616 FAZIO
tencies to answer correctly, across the various experiments). They also

perceived less change when they did detect it, viewing the morph as rel-
atively more likely to be a different photo of a person seen before than a
photo of a novel person never before seen. Thus, stimuli that were
changed, but remained sufficiently similar to the original to evoke the at-
titude associated with the original stimulus, were assimilated in the
direction of the attitude. This assimilation apparently contributed to an
initial sense of familiarity with the changed stimuli, making the
detection of change relatively more difficult.
Finally, research by Chartrand and Bargh (1999, as discussed in Bargh
& Chartrand, 1999) has examined the interesting possibility that differ-
ential mood states can sometimes arise, for seemingly unknow^n rea-
sons, simply as a function of repeated exposure to valenced stimuli. In
the context of an alleged visual acuity task that involved indicating as
quickly as possible the side of the screen on which flashes appeared, par-
ticipants were subliminally exposed to four nouns numerous times. The
between-subjects design involved nouns that normative data had
shown to be strongly positive (e.g., music, friends), strongly negative
(cancer, cockroach), weakly positive (parade, clown), weakly negative
(Monday, worm), or neutral (building, plant). Relative to the neutral
condition, subliminal exposure to the objects involving strong
evaluative associations, but not those involving weaker associations, in-
fluenced subsequent mood reports. Thus, repeated automatic activation
of positivity (or negativity), as a result of the exposure to objects involv-
ing strongly associated positive (or negative) evaluations, generated a
positive (or negative) mood.
I may have belabored the point, but these various findings all illustrate
an influence of attitudes, especially ones involving stronger object-eval-
uation associations, on subsequent task performance or judgments. Im-
portantly, this influence is evident even though the subsequent situation
provides no reason for actively considering those attitudes. Thus, visual
search performance, cued recall, judgments of similarity, perceptual as-
similation, and reports of current mood were all affected by the presence
(sometimes subliminal presence) of attitude-evoking objects. The find-
ings converge on one fundamental conclusion. At least some attitudes
are represented in memory, and they are activated automatically when
the attitude objects are encountered.
(SOME) ATTITUDES ARE CONSTRUCTED
As emphasized by the notion of object-evaluation associations varying

in strength and the attitude-nonattitude continuum, not all objects are
represented in memory with strongly associated evaluative linkages.

Yet, situational demands sometimes force individuals to make
evaluative judgments and decisions regarding novel entities (e.g., the
prospect of dining at a newly opened restaurant or voting for a newly
emerging candidate for public office) or alternatives within what is typi-
cally a domain of indifference (e.g., persons who are barely attentive to
college basketball during the regular season, but annually enter brackets
into an NCAA tournament pool). In such cases, attitudes need to be con-
structed. The effort and deliberation devoted to such construction will
vary as a function of just how motivated the individual might be to reach
an accurate assessment (Kruglanski, 1989; Fazio, 1990; Tetlock & Kim,
1987). An evaluation may be constructed relatively effortlessly on the
basis of sheer resemblance to entities for which one already has attitudes
represented in memory (e.g., Duckworth, Bargh, Garcia, & Chaiken,
2002; Gilovich, 1981; Shook, Fazio, & Eiser, 2007), or, the individual may
deliberate extensively about the specific attributes that characterize the
entity and their favorability (e.g., Sanbonmatsu & Fazio, 1990).
Because attitudes vary in strength, they are differentially sensitive to
information that is salient in the immediate situation. Such a theoretical
assertion is by no means novel. Indeed, Bem's (1972) impactful proposal
that "individuals come to 'know' their own attitudes, emotions, and
other internal states partially by inferring them from observation of their
own behavior and/or the circumstances in which this behavior occurs"
was accompanied by the proviso that such attitudinal inference pro-
cesses occur "to the extent that internal cues are weak, ambiguous, or
uninterpretable" (p. 2). A considerable amount of evidence has accumu-
lated in support of this reasoning. The relevant research typically as-
sumes a particular form. A manipulation makes salient either positive or
negative information related to the attitude object, and the impact of that
manipulation on subsequent attitudinal reports is compared in two
samples that differ with respect to some index of the strength of the par-
ticipants' initial attitudes. For example, Chaiken and Baldwin (1981) em-
ployed a linguistic manipulation, based on earlier research by Salancik
and Conway (1975), to make salient to individuals either pro-ecology or
anti-ecology behaviors that they had performed. Due to how the ad-
verbs "occasionally" and "frequently" were paired with various ecol-
ogy-related behaviors, participants found themselves indicating that
they had performed many pro and few anti behaviors, or vice versa. This
linguistically biased review of their past behaviors led participants to
differential views of their favorability toward environmentalism. How-
ever, it did so more for some individuals than for others. In an earlier ses-
sion, participants had been administered various measures of
attitude—a cognitively oriented measure based on an expectancy-value
618 FAZIO
framework and a more affectively oriented measure. Only individuals

who had responded relatively inconsistently across the two different
measures, suggesting that they lacked clear knowledge of their
attitudes, were influenced by the linguistically biased recall of past
behaviors.
Similarly, Hodges and Wilson (1993) examined the moderating role of
attitude accessibility on the effect that analyzing reasons has on attitudi-
nal reports. Immediately after explaining the reasons for their attitudes,
people report attitudes that are very consistent with the valence implied
by the now very salient reasons that they articulated. However, these
easily verbalizable reasons may be an unrepresentative subset and,
hence, may be somewhat discrepant with the individuals' attitudes and
with their prior reports (see Wilson, Hodges, & LaFleur, 1995). Hodges
and Wilson (1993) found that the extent to which such consequences oc-
curred varied as a function of the accessibility of individuals' attitudes.
Those who responded relatively quickly to an attitudinal query at Time
1 were less influenced by verbalizing reasons for their attitudes at a later
time than were those who responded slowly. The latter group—those
with lower attitude accessibility—displayed a stronger correlation be-
tween their Time 2 attitude reports and the favorability implied by the
reasons they articulated and, as a result, a weaker correlation between
their Time 1 and Time 2 reports. These differences occurred despite the
fact that the two groups displayed equivalence in terms of the extent to
which the reasons they generated were consistent with the Time 1 rat-
ings. Thus, even though they verbalized equivalently unrepresentative
reasons, individuals with greater attitude accessibility were less influ-
enced by these momentarily salient reasons when they were reporting
their attitudes at Time 2, and their reports displayed greater stability
across time.
Relatedly, many additional studies have found the stability of attitude
reports over time and/or resistance to counterattitudinal information to
vary as a function of attitude accessibility (e.g., Bassili, 1996; Bassili &
Fletcher, 1991; Fazio & Williams, 1986; Zanna, Fazio, & Ross, 1994).
Moderating effects of the sort reviewed in this section clearly illustrate
the importance of attitude strength. Contextually salient information
can influence the attitudinal ratings that people provide at any given
moment in time, but the extent of that influence varies. When objects are
represented in memory with relatively weakly associated (or absent)
evaluative linkages, individuals faced with a query, or some other situa-
tional demand for evaluative judgment, will need to construct an atti-
tude. Hence, they will be more affected by the currently salient
information than individuals for whom a preexisting evaluative
association is activated automatically.
PROBLEMS WITH A STRONG "ATTITUDES AS

CONSTRUCTIONS" PERSPECTIVE
In light of the research summarized in the preceding section, I have to

admit to some bewilderment regarding recent discussions portraying
all (or virtually all) attitudes as momentary constructions (e.g., Schwarz
& Bohner, 2001; Zaller & Feldman, 1992; see also Schwarz, this issue).
These formulations view attitudes as evaluative judgments that are al-
ways computed from scratch on the basis of information that is accessi-
ble at that moment, as opposed to their being represented and activated
from memory. Thus, the strong version of this perspective is contrary to
the evidence that at least some attitudes are represented in memory and
exert influence even when individuals have no reason to reflect upon,
retrieve, or construct evaluations.
The argument typically advanced as the reason to entertain a con-
structionist perspective centers on the contextual dependence of
evaluative judgments. There can be no question that attitudinal reports
can be influenced by salient contextual information. We already have
highlighted some such evidence, for example, Chaiken and Baldwin
(1981) and Hodges and Wilson (1993). However, this research demon-
strates not orUy contextual dependence, but also moderation of this
dependence as a function of attitude strength.
In my view, then, the strong constructionist perspective is contra-
dicted by existing data. In addition, as Fazio and Olson (2003a) argued, it
is characterized by an inherent logical problem. To illustrate, let us con-
sider an individual confronted with a situational demand that requires
evaluation of attitude object X. Because X is novel (e.g., a new automo-
bile model or a political candidate) and bears no obvious resemblance to
objects toward which attitudinal associations exist in memory, the indi-
vidual will need to engage in some degree of construction. Presumably,
the individual will compute an attitude toward X on the basis of what-
ever attributes of X happen to be salient at the time. Let us presume that
these include attributes A, B, and C. If these attributes are themselves
positively valued, then the individual will conclude that X is positive. If
they are predominately negative, then X will be viewed as negative.
From where, though, did the evaluations of attributes A, B, and C
emerge? Why did they take on positive or negative value? The strict con-
structionist perspective has to maintain that the attitudes toward the at-
tributes were themselves the outcome of a construction process. So,
salient attributes of A—a,, aj, and ag—inform the computation of the atti-
tude toward A. But, why does aj assume a particular valence? Obvi-
ously, the B construction process is open to a problem of infinite regress.
620 FAZIO
Ultimately, some relevant evaluation—some outcome of prior learning

history—needs to be represented in memory.
The importance of prior learning points to yet another conceptual
problem v^^ith the strong constructionist perspective. What kind of infor-
mation processing system would not learn from its computational ef-
forts? Would any functional system, once it has concluded that X is
good, lose all trace of its construction output? Would it have to
"recompute" entirely from scratch with no savings in releaming the next
time a situation demands an evaluative judgment of X? If attitudes are
viewed as evaluative knowledge, akin to any other knowledge, why
would such knowledge not accumulate over time? Associative models
of learning and memory call for the development or strengthening of an
excitatory evaluative link. Connectionist models call for a change in the
learned weights connecting the input units and the output.
Social and cognitive scientists do not question that people readily ac-
quire semantic knowledge. They learn that the probability of encounter-
ing "butter" is greater given the presence of "bread" than its absence,
that the probability of seeing a "nurse" increases in the presence of a
"doctor." Research on impression formation and social perception has
repeatedly demonstrated that perceivers develop associations between
a target person and traits that characterize the person's behavior (e.g.,
Carlston & Skowronski, 2005; Uleman, 1999). Why should evaluative
knowledge be any different?
Even concepts that at first require extensive computational resources
clearly benefit from learrung. A young child performs basic arithmetic
operations such as addition by counting. So, to determine the total num-
ber of balls that are available, the child points and counts the two red
balls, "one, two," and continues counting "three, four" while pointing to
the two blue balls. Knowing the principle of cardinality, the child
proudly announces that there are "four" balls. The answer is obviously
the outcome of a construction process. However, over time and with
practice, the child needs to rely less and less on pointing, counting, and
procedtural rules. The child comes to know that 2 + 2 = 4. The output of
the previous constructions is learned and represented in memory as the
sum of two sets of two. Again, why should evaluative knowledge be any
different?
On the basis of these arguments alone, I fail to see how a strong "atti-
tudes as constructions" perspective can be granted credence. Nor am I
alone. Even proponents seem to agree: " . . . we adopt a strong version of
a construal model, assuming, for the sake of the argument, that respon-
dents always [emphasis in original] need to compute a judgment from
scratch and can't recall their previous evaluations. As anyone who re-
members that a movie was "boring"—^but can't recall any relevant de-
ATTITUDES AS OB|ECT-EVALUATION ASSOCIATIONS 621
tails—realizes, this extreme assumption is unrealistic" (Schwarz &

Bohner, 2001, p. 444). It is this very unrealism that necessitates consider-
ation of an attitude-nonattitude continuum. Attitudes are evaluative
knowledge represented in memory, but there will be objects for which
individuals lack such knowledge and, hence, need to construct overall
evaluations on the basis of whatever relevant information they can
access from memory or whatever the situation makes available.
However, the outcomes of any such construction efforts are not for-
gotten. To the contrary, they facilitate progression from the status of a
nonattitude toward the status of a strongly associated evaluation that
can be activated automatically when the object is encountered. This atti-
tude development has functional value for the individual, enabling effi-
cient and relatively effortless appraisal of the object. Eor example,
attitude rehearsal has been shown to ease later decision making, de-
creasing the resources that are needed to reach evaluative decisions.
When asked to make rapid decisions about their preferences between
pairs of objects, individuals who had earlier been induced to develop
and rehearse attitudes toward novel objects displayed less cardiovascu-
lar reactivity than individuals who had been exposed to the objects
equally often but in the context of a task that was unrelated to attitudes
(Blascovich et al., 1993; Fazio, Blascovich, & DriscoU, 1992). Likewise,
less reactivity was exhibited when individuals made decisions involv-
ing objects for which they had earlier rehearsed attitudes than objects for
which they had not (Blascovich, et al., 1993). Activation of the rehearsed
attitudes during the pairwise preference task apparently made the deci-
sion task less demanding. Fewer resources were required to cope with
the demands of the task (see Fazio, 2000 for an extensive discussion of re-
search related to the object appraisal function of attitudes). Thus,
position along the attitude-nonattitude continuum influences the very
nature of the evaluative judgment process.
THE MODEL'S PERSPECTIVE REGARDING A FEW

CONTEMPORARY ISSUES
RESPONSES TO EXPLICIT MEASURES OE
ATTITUDE ARE VERBAL BEHAVIORS
I have tried to be careful and precise in this article about my use of the
term attitude versus individuals' reports of their attitudes. Responses to
a verbal query should not be viewed as attitudes per se. They are verbal
expressions of the attitude, subject to many forces beyond the influence
of any evaluative associations that might exist in memory. The distinc-
622 FAZIO
tion is one that attitude theorists have long recogruzed. In his classic arti-
cle "Attitudes can be measured," Thurstone (1928) used the term
"opinion" to refer to "a verbal expression of attitude" and argued that an
opinion merely "symbolizes an attitude" (p. 531). Although Thurstone
advocated using "opinions as the means for measuring attitudes," he
noted that "neither [a person's] opinions nor his overt acts constitute in
any sense an infallible guide to the subjective inclinations and prefer-
ences that constitute his attitude. Therefore, we must remain content to
use opinions, or other forms of action, merely as indices of attitude" (p.
532).
Thurstone was concerned that a person's verbal behavior may be in-
fluenced by factors in addition to the attitude. He specifically made ref-
erence to possibilities ranging from intentional misrepresentation in the
interest of avoiding outright dispute with others to less than frank ex-
pressions stemming from courtesy. However, there are many factors in
addition to self-presentation concerns that can make responses to verbal
queries poor indices of the attitude of interest. As many survey research-
ers have noted (e.g., Schuman & Presser, 1981; Schwarz & Bohner, 2001;
Tourangeau & Rasinksi, 1988), even well-intentioned responding to a
verbal measure of attitude is a complex process involving issues of ques-
tion comprehension, scale interpretation, and the use of appropriate
standards of comparison. In this sense, verbal reports of attitudes are al-
ways constructions, and those responses can be influenced by construals
and standards that bear little relation to what the survey researcher
intended.
Correspondence between Implicit and Explicit Measures. E v e n if the
"communication" between the questioner and the respondent has pro-
ceeded as intended, the verbal report may be influenced by many factors
in addition to the evaluative association represented in memory. By not
requiring direct reports, implicit measures of attitude aim to provide a
more proximal estimate of these attitudinal representations than is pos-
sible with explicit measures. This difference alone may result in an ob-
served discordance between the attitude estimates provided by implicit
and explicit measures. My colleagues and I have approached this issue
of correspondence between implicit and explicit measures from the per-
spective of the MODE Model of attitude-behavior processes (Fazio,
1990; Fazio & Towles-Schwen, 1999; Fazio & Olson, 2003b). Because re-
sponding to an explicit measure is itself a verbal behavior, it can be
viewed as related to the multiple attitude-behavior processes that form
the focus of the MODE Model. The verbal expression may stem from a
relatively spontaneous process involving the direct influence of the au-
tomatically activated attitude on the immediate perceptions of the atti-
tude object noted in the question. Alternatively, it may stem from a very
deliberative analysis of the attributes that characterize the attitude ob-

ject, provided that the individual has both the motivation and the oppor-
tunity to engage in such effortful reflection. Finally, the verbal response
may stem from what the MODE model refers to as mixed processes, in-
volving both automatic and controlled components. In such cases, the
verbal expression may be influenced not only by the relevant automati-
cally activated attitude, but also by the downstream consequences of
any number of motivational factors. Individuals may be so concerned
about accuracy that they check the implications of the activated attitude
against their knowledge regarding the attributes of the object, instead of
just "going with their guts." Sufficiently motivated individuals may
compare the activated attitude for its consistency with various internal
standards and cherished values. Of course, as Thurstone noted, they
may "spin" their attitudes so as to make their verbal responses more
palatable to the apparent audience.
According to the MODE Model, then, any attitudes activated auto-
matically by the mention of the attitude object in an explicit measure are
merely a "starting point" for the verbal behavior. Whether the verbal re-
sponse reflects the attitude will depend on the motivation and opportu-
nity factors. Verbally expressed judgments occur farther "downstream"
than the automatic activation of any relevant attitude and may be influ-
enced by motivational factors that can override the effect of the automat-
ically activated attitude. For these reasons, greater correspondence
between implicit and explicit measures is expected when the motivation
to deliberate and/or the opportunity to do so are low. When both are
high, any additional information that the motivated individual consid-
ers has the potential to produce discordance between the verbal re-
sponses and the attitude estimated by the implicit measure. This
reasoning has received empirical support in a number of investigations
concerned with race-related judgments (see Fazio & Olson, 2003b for a
review).
Neither Unconscious nor Dual Attitudes. Importantly, from the per-
spective of the MODE model, any observed discordance between im-
plicit and explicit measure does not necessitate reference to the
unconscious (e.g., Banaji, 2001) or to dual representations of attitude
(e.g., Wilson, Lindsey, & Schooler, 2000). As Fazio and Olson (2003b)
emphasized, no implicit measure speaks to the question of whether indi-
viduals are or are not aware of their attitudes. Additional evidence, well
beyond the administration of an implicit measure, is required to make
any such inference (see Olson & Fazio, 2002, for an example). In addi-
tion, no explicit measure guarantees the existence of a distinct represen-
tation in memory, independent of the attitude that is automatically
activated when the object is encountered. Responding to the explicit
624 FAZIO
measure is an exercise in verbal behavior and, hence, does not speak to

the presence or absence of preexisting representations.
ON THE PRESUMED MALLEABILITY OF

AUTOMATICALLY ACTIVATED ATTITUDES
Research involving implicit measures of attitude focused at first on

their correspondence with explicit measures and their utility as predic-
tors of judgments and behavior. However, interest in the possibility of
change in automatically activated attitudes, especially prejudice,
spawned research in which impHcit measures served as dependent
measures. The findings (see Blair, 2002 for a detailed review) revealed
what has been viewed by some as a surprising sensitivity of implicit
measures to immediate contextual factors and, hence, has raised ques-
tions about the relative stability versus malleability of automatically
activated attitudes. In my view, the case for malleability has been over-
stated. What has been observed is malleability of the scores on the im-
plicit measure. I do not regard such findings as necessarily indicative
of malleability of the attitude. Instead, the malleability may be more
precisely pinpointed as stemming from either the malleability of the
measure or the malleability of the attitude object. It is important that
the distinctions not be blurred.
Malleability of the Measure: The IAT's Sensitivity to Extrapersonal Associa-
tions. Considerable recent research has demonstrated that pre-exposure
to different kinds of information influences Implicit Association Test
(IAT, Greenwald, McGhee, & Schwartz, 1998) scores (see Blair, 2002 for a
review). Essentially, salient positive information about a given category
produces IAT scores indicative of more favorable evaluations. Such ma-
nipulations have involved exposure to movie clips that depicted Blacks
at a harmonious family event versus an argumentative, gang-related
scene (Wittenbrink, Judd, & Park, 2001), presentation of a series of either
admired Black individuals and disliked Whites or disliked Blacks and
admired Whites (Dasgupta & Greenwald, 2001), assigning participants
to interact with a Black partner who occupied a superior or subordinate
task role (Richeson & Ambady, 2003), exposure to violent and misogy-
nous rap music (Rudman & Lee, 2002), and introduction to the IAT (spe-
cifically as a procedure for assessing prejudice) by a Black versus a
White experimenter (Lowery, Hardin, & Sinclair, 2001). Yet, such find-
ings do not constitute unambiguous evidence for the inference that auto-
matically activated attitudes have been modified. The critical issue in
need of resolution is whether such contextual manipulations (a) pro-
duce an actual change in the relevant mental representation, to which
the IAT is sensitive, or (b) simply render momentarily salient

information that assists the participant with the response-mapping
problem posed by the IAT.
The findings that my colleagues and I have recently accumulated re-
garding the sensitivity of the IAT to extrapersonal associations lead me
to favor the second alternative, or at least to the belief that considerable
caution is required in drawing any inferences about attitude change on
the basis of such IAT findings. Extrapersonal associations—ones that
neither form the basis of the individual's attitude nor become activated
automatically in response to the object—can influence the IAT as it is tra-
ditionally implemented. Especially relevant is research in which both at-
titudes and extrapersonal associations were experimentally created in
the lab (Han, Olson, & Fazio, 2006). Participants formed a preference for
one novel stimulus over another, with the former being described in
ways that made it objectively far superior. They were then exposed to
the videotaped comments of two young boys who expressed opinions
that were either consistent with that preference or incredulously incon-
sistent. The results revealed that those who were given attitudinally in-
consistent extrapersonal associations were able to use that information
to facilitate coping with the incompatible mapping problem posed by
the traditional IAT. They exhibited a significant reduction in their pre-
sumed preference for the objectively superior of the two stimuli, com-
pared to those who had received the consistent extrapersonal
association. This attenuation of the IAT effect occurred despite the fact
that the participants rated the sources of the attitude-inconsistent
extrapersonal information as irrational and foolish. On the other hand,
the extrapersonal associations did not influence either a subliminal
priming measure or a version of the IAT that was "personalized" and,
hence, less susceptible to the influence of extrapersonal associations
(Olson & Fazio, 2004). These measures proved sensitive to the attitude,
regardless of the congruency of the extrapersonal information. Thus, the
various findings noted earlier regarding contextual sensitivity may
reflect the influence of momentarily salient extrapersonal associations.
As such, they may point more to the malleability of the IAT than to the
malleability of attitudes.
Malleability of the Attitude Object: Let's not forget Asch. Some of the re-
search illustrating contextual sensitivity has employed, not the IAT, but
an evaluative priming procedure. When Black faces are presented
within the background context of a church interior, they prime less
negativity than when the context is an urban street corner or a jail
(Wittenbririk et al., 2001; Barden, Maddux, Petty, & Brewer, 2004). Simi-
larly, current motivational states have been shown to influence the eval-
uation that is primed by the presentation of a related object. Water
626 FAZIO
primes more positivity when individuals are thirsty (Ferguson & Bargh,
2004), food stimuli activate more positivity when individuals are hun-
gry (Seibt, Hafner, & Deutsch, 2007), and cigarettes prime more
positivity when heavy smokers are in a state of deprivation as opposed
to having just had a cigarette (Sherman, Rose, Koch, Presson, & Chassin,
2003). Whether the context is internal or external to the individual, find-
ings of this sort need not be interpreted as evidence of the malleability of
automatically activated attitudes or of objects being associated with
multiple attitudes. Instead, they may represent varying construals of the
attitude object.
This is the very issue that Asch (1948) addressed long ago when he ar-
gued against the "doctrine of suggestion, prestige, and imitation" as the
explanation for the often-observed finding that a statement is evaluated
more positively when it is attributed to a more prestigious source. He
noted that the finding need not presuppose that "an unchanged object of
judgment undergoes change of evaluation" (p. 255) as a function of the
prestige of the individual from whom it emanates. The apparent
changes in evaluation may be due to "a change in the object of judgment
rather than in the judgment of the object" (p. 256). To use the classic ex-
ample, "rebellion" does not refer to the same entity when attributed to
Jefferson versus Lenin. Thus, the statement that "a little rebellion, now
and then, is a good thing and necessary in the political world as storms
are in the physical" is evaluated differently due to a change of meaning.
The importance of such change of meaning has been documented re-
peatedly in research concerning impression formation (e.g., Hamilton &
Zanna, 1974; Higgins & Rholes, 1976).
In parallel fashion, I would argue that water is not the same attitude
object when one is thirsty as when not, a given food is not the same object
when one is hungry versus sated, and cigarettes are not the same atti-
tude object when a smoker is in a state of craving versus not. Likewise,
situational cues can lead to varying construals of any given object. The
very same African American male wiU conjure very different interpreta-
tions depending on whether he is dressed in a three-piece suit versus
gang attire. Virtually all persons, objects, and issues are multiply
categorizable. It is how an object is construed at any moment in time that
determines attitude activation (see Barden et al., 2004; Fazio, 1986; Fazio
& Dunton, 1997; Lord & Lepper, 1999; Sinclair & Kunda, 1999; Smith et
al., 1996 for discussion of such categorization processes). Thus, it is not
surprising that the attitudes activated in a priming experiment may vary
as a function of the effect of current concerns and context on construal of
the objects.
I do not mean to imply that automatically activated attitudes are inca-
pable of change. To the contrary, I believe that some research findings
provide evidence strongly suggesting that appropriately designed so-

cial influence techniques can produce an actual change in the represen-
tation of the attitude (e.g., Kawakami, Dovidio, Moll, Hermsen, &
Russin, 2000; Olson & Fazio, 2006; Rudman, Ashmore, & Gary, 2001).
However, given the abundance of evidence regarding the relative stabil-
ity of accessible attitudes over time and their resistance to
counterinformation (see Fazio, 1995 for a review), I am skeptical that the
mere contextual salience of counterattitudinal information has any im-
pact on the attitude representation. To me, it seems far more likely that
any such empirical demonstration reflects the malleability of the im-
plicit measure or the malleability of the object than the malleability of the
attitude itself. In any case, I would urge that researchers recognize these
multiple possibilities and exercise caution in the inferences that are
drawn from the observation of change in implicit measure scores.
ATTITUDE AMBIVALENCE
It is rare for any given object not to be characterized both by some attrib-
utes that might be considered positive and by some that might be con-
sidered negative. This creates the potential for the experience of
ambivalence when judging the object or making an approach/avoid-
ance decision. To the extent that the individual lacks an accessible atti-
tude and is forced to construct an evaluation on the spot in response to a
verbal query or a need for action, the individual may experience ambiv-
alence if both the positive and the negative features of the object are sa-
lient (Newby-Glark, McGregor, & Zanna, 2002). However, the essence
of more reasoned forms of attitude development is an integration of any
such conflicting information into a summary evaluation. Once formed,
and associated with the attitude object, this summary evaluation effec-
tively resolves the ambivalence. In future similar situations, the object
will evoke the summary evaluation, instead of requiring continued con-
sideration of the attribute information. Thus, ambivalence amounts to
pre-decisional coriflict and can be obviated by the activation of a rele-
vant summary evaluation developed at an earlier point in time. Accord-
ing to the MODE model, individuals may revisit the attribute level, if
they are sufficiently motivated and have the opportunity to do so
(Fabrigar, Petty, Smith, & Crites, 2006; Sanbonmatsu & Fazio, 1990).
However, when either motivation or opportunity is lacking, the earlier
formed summary evaluation will predominate.
It is important to note that the potential for ambivalent reactions to an
attitude object relates very closely to the fact that most objects can be con-
strued in multiple ways. Ambivalence is often portrayed, as I have
628 FAZIO
framed it above, as stemming from the subordinate attribute level (e.g,

Kaplan, 1972). One is tom by the object's obvious possession of both pos-
itive and negative features (e.g., chocolate cake is both delicious and fat-
tening). However, ambivalence can also be viewed as an issue of
categorization (see Fazio, 1994 for an earlier discussion of this perspec-
tive). To which category does the object belong? This more
superordinate emphasis highlights that the object can be construed in
multiple ways. In many situations, contextual cues and current motiva-
tions may foster a particular categorization. In the context of a self-in-
dulgent dinner celebrating a significant life event, the chocolate cake is
likely to be categorized as a wonderfully tasty dessert and evoke nothing
but positivity. Shortly before or after a visit to the gym, the same cake
may be quickly relegated to the fattening category and evoke negativity.
When contextual cues are less dictating of the categorization outcome,
one construal may still be more likely to predominate over the other as a
function of the individual's own proclivities. As noted earlier while dis-
cussing the influence of attitude accessibility on categorization, the more
attitude-evoking possibility is at an advantage (Fazio & Dunton, 1997;
Smith et al., 1996.) Thus, committed and successful dieters are likely to
have developed such strongly associated negativity to fattening foods
that they are more apt to "assign" the chocolate cake to the category of
"fattening" than are individuals who have the advantage of a high me-
tabolism (see Ferguson, 2007; Fishbach, Friedman, & Kruglanski, 2003
for related research findings). Neither of these two types of individuals
is likely to experience much ambivalence. Ambivalence is more likely to
arise in situations in which, and among people for whom, potential cate-
gorizations essentially compete for attention. Viewed from this perspec-
tive of superordinate categorization, the ambivalence stems, not from
the simultaneous activation of positive and negative sentiments, but
from shifting (possibly rapidly shifting) construals of the attitude object.
When categorized in one way, the chocolate cake evokes positivity.
When categorized in another way, it activates a negative attitude. Thus,
it is the malleability of the attitude object, not the malleability of the
attitude or the possession of multiple attitudes toward the object, that
dictates the experience of ambivalence.
Such a view has important implications for the measurement of atti-
tudes and attitude ambivalence. When we as researchers ask individu-
als for their evaluations of an object, we often do so in a rather abstract
manner that lacks specific contextual cues. As a result, we may be re-
questing evaluation at a more superordinate level of categorization than
respondents typically employ for that object or set of objects. Individu-
als' awareness of the potential for multiple categorizations and their
consideration of varying evaluations may prompt them to express rat-
ings indicative of ambivalence, effectively noting that their evaluations

"depend" on the details. However, in a specific situation, any such am-
bivalence may not be experienced because sufficient cues exist to disam-
biguate the categorization and, hence, prompt activation of the attitude
associated with a single category.
SUMMARY AND CONCLUSIONS
The major message that I hope this article has succeeded in communi-
cating can be summarized in two words. Attitudes exist! They are not a
scientific reification. They are not hypothetical constructs. Attitudes
are evaluative knowledge, represented in memory in the same way as
any other form of knowledge. They summarize our prior learning with
respect to the valence of the outcomes produced by a given object. As
summary evaluations associated with the representation of an attitude
object, these attitudes can be activated from memory automatically
when the object (or a sufficiently related object) is encountered. Ac-
cording to the MODE Model, such automatically activated attitudes
serve as the "starting point" for our appraisals of the object in the im-
mediate situation. When either the motivation to deliberate further, or
the opportunity to do so, is lacking, these immediate perceptions will
influence evaluative judgments, verbal expressions, decisions, and
overt behavior in a relatively direct manner. In this way, the evaluative
knowledge that we have acquired as a result of our prior learning his-
tory—our collection of attitudes—proves to be very functional. Atti-
tudes simplify our day-to-day existence, enabling efficient appraisal
of the objects that we encounter. Although relatively thoughtless, these
appraisals promote approach behavior toward objects from which we
probabilistically will experience pleasure and avoidance of objects
likely to produce pain. Thus, attitudes form the cornerstone of a truly
functional system by which learning and memory guide behavior in a
fruitful direction.
It is important to recognize, however, that the direct "downstream"
consequences of automatically activated attitudes are not inevitable.
The automatically activated evaluation is a starting point or default
value, but its influence can be overridden. Under some circumstances
(the presence of both motivation and opportunity), individuals may
check the validity of their irrunediate appraisals against other knowl-
edge that they have regarding the attitude object, or examine its consis-
tency with motivational goals to which they aspire. Such knowledge and
motivational forces can attenuate the impact of the automatically
activated attitude or even produce an effort to correct for its influence.
630 FAZIO
Responses to explicit measures of attitudes are themselves verbal be-

haviors. They are expressions of the attitude, not the attitude per se. As
with any behavior, these verbal expressions can be influenced by not
only any attitude that is automatically activated but also downstream
motivational forces. Hence, discordance between implicit measures as-
sessing the automatically activated attitude and explicit measures can
occur, especially under conditions characterized by a motivation to de-
liberate and the opportunity to do so. The discordance is readily tinder-
standable in terms of motivated correction for the influence of the
automatically activated attitude. In and of itself, then, any such discor-
dance does not require reference to the unconscious or to dual
representations of attitudes in memory.
REFERENCES
Allport, G. W. (1935). Attitudes. In C. Murchison (Ed.), Handbook of social psychol-
ogy (pp. 798-844). Worcester, MA: Clark University Press.
Asch, S. E. (1948). The doctrine of suggestion, prestige, and imitation in social
psychology. Psychological Review, 55, 250-276.
Banaji, M. R. (2001). Implicit attitudes can be measured. In H. L. Roediger, III, J. S.
Nairne, I. Neath, & A. Surprenant (Eds.), The nature of remembering: Essays
in honor of Robert G. Crowder (pp. 117-150). Washington, DC: American
Psychological Association.
Bargh, J. A. (1994). The four horsemen of automaticity: Awareness, intention, ef-
ficiency, and control in social cognition. In R. S. Wyer & T. K. SruU (Eds.),
Handbook of social cognition (Vol. 1, pp. 1-40). Hillsdale, NJ: Erlbaum.
Bargh, J.A., & Chartrand, T.L. (1999). The unbearable automaticity of being.
American Psychologist, 54, 462-479.
Barden, J., Maddux, W. W., Petty, R. E., & Brewer, M. B. (2004). Contextual mod-
eration of racial bias: The impact of social roles on controlled and automat-
ically activated attitudes. Journal of Personality and Social Psychology, 87,
5-22.
Bassili, J. N. (1996). Meta-judgmental versus operative indexes of psychological
attributes: The case of measure of attitude strength. Journal of Personality
and Social Psychology, 71, 637-653.
Bassili, J. N., & Fletcher, J. F. (1991). Response-time measurement in survey re-
search: A method for CATI and a new look at nonattitudes. Public Opinion
Quarterly, 55, 331-346.
Bem, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.), Advances in exper-
imental social psychology (Vol. 6, pp. 1-62). New York: Academic Press.
Blair, I. (2002). The malleability of automatic stereotypes and prejudice. Personal-
ity and Social Psychology Review, 6, 242-260.
Blascovich, J., Ernst, J. M., Tomaka, J., Kelsey, R. M., Salomon, K. L., & Fazio, R. H.
(1993). Attitude accessibility as a moderator of autonomic reactivity dur-
ing decision making. Journal of Personality and Social Psychology, 64,

165-176.
Brifiol, P., Petty, R. E., & Tormala, Z. L. (2006). The malleable meaning of subjec-
tive ease. Psychological Science, 17, 200-206.
Bruner, J. S. (1957). On perceptual readiness. Psychological Review, 64,123-152.
Carlston, D. E., & Skowronski, J. (2005). Linking versus thinking: Evidence for
the different associative and attributional bases of spontaneous trait trans-
ference and spontaneous trait inference. Journal of Personality and Social
Psychology, 89,884-898.
Chaiken, S., & Baldwin, M. W. (1981). Affective-cognitive consistency and the
effect of salient behavioral information on the self-perception of attitudes.
Journal of Personality and Social Psychology, 41,1-12.
Conrey, F.R., & Smith, E.R. (2007). Attitude representation: Attitudes as patterns
in a distributed, connectionist representational system. Social Cognition, 25
(5), 718-735.
Converse, P. E. (1970). Attitudes and non-attitudes: Continuation of a dialogue.
In E. R. Tufte (Ed.), The quantitative analysis of social problems (pp. 168-189).
Reading, MA: Addison-Wesley.
Dasgupta, N., & Greenwald, A. G. (2001). On the malleability of automatic atti-
tudes: Combating automatic prejudice with images of admired and dis-
liked individuals. Journal of Personality and Social Psychology, 81,800-814.
DeFleur, M. L., & Westie, F. R. (1963). Attitude as a scientific concept. Social
Forces, 42,17-31.
De Houwer, J., Thomas, S., & Baeyens, F. (2001). Associative learning of likes and
dislikes: A review of 25 years of research on human evaluative condition-
ing. Psychological Bulletin, 127, 853-869.
Deutsch, R., & Strack, F. (2006). Duality models in social psychology: From dual
processes to interacting systems. Psychological Inquiry, 17,166-172.
Duckworth, K. L., Bargh, J. A., Garcia, M., & Chaiken, S. (2002). The automatic
evaluation of novel stimuli. Psychological Science, 13,513-519.
Eiser, J. R., Fazio, R. H., Stafford, T., & Prescott, T. J. (2003). Connectionist simula-
tion of attitude learning: Asymmetries in the acquisition of positive and
negative evaluations. Personality and Social Psychology Bulletin, 29,
1221-1235.
Eiser, J. R., Stafford, T., & Fazio, R. H. (2007). How negative attitudes resist
change: A connectionist account of the consequences of global and tar-
geted expectancies. Unpublished manuscript. University of Sheffield,
Sheffield, UK.
Fabrigar, L., Petty, R. E., Smith, S. M., & Crites, S. L. (2006). Understanding
knowledge effects on attitude-behavior consistency: The role of rele-
vance, complexity, and amount of knowledge. Journal of Personality and So-
cial Psychology, 90,556-577.
Fazio, R. H. (1986). How do attitudes guide behavior? In R. M. Sorrentino & E. T.
Higgins (Eds.), The handbook of motivation and cognition: Foundations ofsocial
behavior (pp. 204-243). New York: Guilford Press.
632 FAZIO
Fazio, R. H. (1990). Multiple processes by which attitudes guide behavior: The

MODE model as an integrative framework. In M. P. Zanna (Ed.), Advances
in experimental social psychology (Vol. 23, pp. 75-109). New York: Academic
Press.
Fazio, R. H. (1994). Attitudes in Associated Systems Theory. In R. S. Wyer (Ed.),
Advances in social cognition (Vol. VII, pp. 157-167).
Fazio, R. H. (1995). Attitudes as object-evaluation associations: Determinants,
consequences, and correlates of attitude accessibility. In R. E. Petty & J. A.
Krosnick (Eds.), Attitude strength: Antecedents and consequences (pp.
247-282). Hillsdale, NJ: Erlbaum.
Fazio, R. H. (2000). Accessible attitudes as tools for object appraisal: Their costs
and benefits. In G. Maio & J. Olson (Eds.), Why we evaluate: Functions of atti-
tudes (pp. 1-36). Mahwah, NJ: Erlbaum.
Fazio, R. H. (2001). On the automatic activation of associated evaluations: An
overview. Cognition and Emotion, 15,115-141.
Fazio, R. H., Blascovich, J., & DriscoU, D. M. (1992). On the functional value of at-
titudes: The influence of accessible attitudes upon the ease and quality of
decision making. Personality and Social Psychology Bulletin, 18,388^01.
Fazio, R. H., Chen, J., McDonel, E. C , & Sherman, S. J. (1982). Attitude accessibil-
ity, attitude-behavior consistency and the strength of the object-evalua-
tion association. Journal of Experimental Social Psychology, 18,339-357.
Fazio, R. H., & Dunton, B. C. (1997). Categorization by race: The impact of auto-
matic and controlled components of racial prejudice. Journal of Experimen-
tal Social Psychology, 33,451^70.
Fazio, R. H., Eiser, J. R., & Shook, N. J. (2004). Attitude formation through explo-
ration: Valence asymmetries.'/ourwa/ of Personality and Social Psychology, 87,
293-311.
Fazio, R. H., Herr, P. M., & Olney, T. (1984). Attitude accessibility following a
self-perception process. Journal of Personality and Social Psychology, 47,
277-286.
Fazio, R. H., Ledbetter, J. E., & Towles-Schwen, T. (2000). On the costs of accessi-
ble attitudes: Detecting that the attitude object has changed. Journal of Per-
sonality and Social Psychology, 78,197-210.
Fazio, R. H., & Olson, M. A. (2003a). Attitudes: Foundations, functions, and con-
sequences. In M. A. Hogg & J. Cooper (Eds.), The sage handbook ofsocial psy-
chology (pp. 139-160). London: Sage.
Fazio, R. H., & Olson, M. A. (2003b). Implicit measures in social cognition re-
search: Their meaning and use. Annual Review of Psychology, 54, 297-327.
Fazio, R. H., Powell, M. C , & Herr, P. M. (1983). Toward a process model of the
attitude-behavior relation: Accessing one's attitude upon mere observa-
tion of the attitude object. Journal of Personality and Social Psychology, 44,
723-735.
Fazio, R. H., & Roskos-Ewoldsen, D. R. (2005). Acting as we feel: When and how
attitudes guide behavior. In T. C. Brock & M. C. Green (Eds.), Persuasion:
Psychological insights and perspectives (pp. 41-62). Thousand Oaks, CA:

Sage.
Fazio, R. H., Sanbonmatsu, D. M., Powell, M. C, & Kardes, F. R. (1986). On the
automatic activation of attitudes. Journal of Personality and Social Psychol-
ogy, 50, 229-238.
Fazio, R. H., & Towles-Schwen, T. (1999). The MODE Model of attitude-behav-
ior processes. In S. Chaiken & Y. Trope (Eds.), Dual process theories in social
psychology (pp. 97-116). New York: Guilford Press.
Fazio, R. H., & Williams, C. J. (1986). Attitude accessibility as a moderator of the
attitude-perception and attitude-behavior relations: An investigation of
the 1984 presidential election. Journal of Personality and Social Psychology,
51,505-514.
Ferguson, M. J. (2007). On the automatic evaluation of end-states. Journal of Per-
sonality and Social Psychology, 92,596-611.
Ferguson, M. J., & Bargh, J. A. (2004). Liking is for doing: The effects of goal pur-
suit on automatic evaluation. Journal of Personality and Social Psychology, 87,
557-572.
Ferguson, M. J., Bargh, J. A., & Nayak, D. A. (2005). After-affects: How automatic
evaluations influence the interpretation of subsequent, unrelated stimuli.
Journal of Experimental Social Psychology, 41,182-191.
Fishbach, A., Friedman, R. S., & Kruglanski, A. W. (2003). Leading us not unto
temptation: Momentary allurements elicit overriding goal activation.
Journal of Personality and Social Psychology, 84,296-309.
Gawronski, B., & Bodenhausen, G. V. (2006). Associative and propositional pro-
cesses in evaluation: An integrative review of implicit and explicit attitude
change. Psychological Bulletin, 132,692-731.
Gawronski, B., & Bodenhausen, G. V. (2007). Unraveling the processes underly-
ing evaluation: Attitudes from the perspective of the APE Model. Social
Cognition, 25(5), 687-717.
Gilovich, T. (1981). Seeing the past in the present: The effect of associations to fa-
miliar events on judgments and decisions. Journal of Personality and Social
Green, B. F. (1954). Attitude measurement. In G. Lindzey (Ed.), Handbook of social
psychology (pp. 335-369). Cambridge, MA : Addison-Wesley.
Greenwald, A. G., McGhee, D., & Schwartz (1998). Measuring individual differ-
ences in cognition: The implicit association task. Journal of Personality and
Social Psychology, 74,1469-1480.
Greenwald, A. G., & Nosek, B. A. (in press). Attitudinal dissociation: What does
it mean? In R. E. Petty, R. H. Fazio, & P. Brifiol (Eds.), Attitudes: Insights
from the new implicit measures. Mahwah, NJ: Erlbaum.
Hamilton, D. L., & Zanna, M. P. (1974). Context effects in impression formation:
Changes in connotative meaning. Journal of Personality and Social Psychol-
ogy, 29,649-654.
Han, H. A., Olson, M. A., & Fazio, R. H. (2006). The influence of experimen-
634 FAZIO
tally-created extrapersonal associations on the Implicit Association Test.

Journal of Experimental Social Psychology, 42, 259-272.
Hermans, D., De Houwer, J., & Eelen, P. (2001). A time course analysis of the af-
fective priming effect. Cognition & Emotion, 15,143-165.
Higgins, E. T. (1996). Knowledge activation: Accessibility, applicability, and sa-
lience. In E. T.Higgins & A. W. Kruglanski (Eds.), Social psychology: Hand-
book of basic principles (pp. 133-168). New York: Guilford Press.
Higgins, E. T., & Rholes, W. S. (1976). Impression formation and role fulfillment:
A "Holistic Reference" approach. Journal of Experimental Social Psychology,
11, i^ll-A^b.
Hilgard, E. R. (1958). Intervening variables, hypothetical constructs, parameters,
and constants. The American Journal of Psychology, 71,238-246.
Hodges, S. D., & Wilson, T. D. (1993). Effects of analyzing reasons on attitude
change: The moderating role of attitude accessibility. Social Cognition, 11,
353-366.
Jarvis, W. B. G., & Petty, R. E. (1996). The need to evaluate. Journal of Personality
and Social Psychology, 70,172-194.
Kaplan, K. J. (1972). On the ambivalence-indifference problem in attitude theory
and measurement: A suggested modification of the semantic differential
technique. Psychological Bulletin, 77,361-372.
Kawakami, K., Dovidio, J. F., Moll, J., Hermsen, S., & Russin, A. (2000). Just say
no (to stereotyping): Effects of training on the negation of stereotypic asso-
ciations on stereotype activation. Journal of Personality and Social Psychol-
ogy, 78, 871- 888.
Klauer, K. C, & Musch, J. (2003). Affective priming: Findings and theories. In K.
C. Klauer & J. Musch (Eds.), The psychology of evaluation: Affective processes
in cognition and emotion (pp. 7-50). Mahwah, NJ: Erlbaum.
Kruglanski, A. W. (1989). Lay epistemics and human knowledge. New York: Ple-
num.
LaPiere, R. T. (1934). Attitudes vs. actions. Social Porces, 13,230-237.
Lord, C. G., & Lepper, M. R. (1999). Attitude Representation Theory. In M. P.
Zanna (Ed.), Advances in experimental social psychology (Vol. 31, pp.
265-343). San Diego: Academic Press.
Lowery, B. S., Hardin, C. D., & Sinclair, S. (2001). Social influence effects on auto-
matic racial prejudice. Journal of Personality and Social Psychology, 81,
842-^55.
MacCorquodale, K., & Meehl, E. (1948). On a distinction between hypothetical
constructs and intervening variables. Psychological Review, 55, 95-107.
Moors, A., & De Houwer, J. (2006). Automaticity: A theoretical and conceptual
analysis. Psychological Bulletin, 132,297-326.
Newby-Clark, I. R., McGregor, I., & Zarma, M. P. (2002). Thinking and caring
about cognitive inconsistency: When and for whom does attitudinal am-
bivalence feel uncomfortable? Journal of Personality and Social Psychology,
82,157-166.
Olson, M. A., & Fazio, R. H. (2001). Implicit attitude formation through classical
conditioning. Psychological Science, 12,413-417.
Olson, M. A., & Fazio, R. H. (2002). Implicit acquisition and manifestation of clas-
sically conditioned attitudes. Social Cognition, 20, 89-103.
Olson, M. A., & Fazio, R. H. (2004). Reducing the influence of extra-personal as-
sociations on the Implicit Association Test: Personalizing the IAT. Journal
of Personality and Social Psychology, 86, 653-667.
Olson, M. A., & Fazio, R. H. (2006). Reducing automatically activated racial prej-
udice through implicit evaluative conditioning. Personality and Social Psy-
chology Bulletin, 32,421-433.
Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of persua-
sion. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol.
19, pp. 123-205). San Diego: Academic Press.
Petty, R. E., Haugtvedt, C, & Smith, S. M. (1995). Elaboration as a determinant of
attitude strength: Creating attitudes that are persistent, resistant, and pre-
dictive of behavior. In R. E. Petty & J. A. Krosnick (Eds.), Attitude strength:
Antecedents and consequences (pp. 93-130). Mahwah, NJ: Erlbaum.
Powell, M. C, & Fazio, R. H. (1984). Attitude accessibility as a fimction of re-
peated attitudinal expression. Personality and Social Psychology Bulletin, 10,
139-148.
Reber, R., Wirtkielman, P., & Schwarz, N. (1998). Effects of perceptual fluency on
affective judgments. Psychological Science, 9,45^8.
Richeson, J. A., & Ambady, N. (2003). Effects of situational power on automatic
racial prejudice. Journal of Experimental Social Psychology, 39,177-183.
Roskos-Ewoldsen, D. R., & Fazio, R. H. (1992). On the orienting value of atti-
tudes: Attitude accessibility as a determinant of an object's attraction of vi-
sual attention. Journal of Personality and Social Psychology, 63,198-211.
Rudman, L. A., Ashmore, R. D., & Gary, M. L. (2001). "Unlearning" automatic bi-
ases: The malleability of implicit prejudice stereotypes. Journal of Personal-
ity and Social Psychology, 81, 856-868.
Rudman, L. A., & Lee, M. R. (2002). Implicit and explicit consequences of expo-
sure to violent and misogynous rap music. Group Processes & Intergroup Re-
lations, 5,133-150.
Salancik, G. R., & Conway, M. (1975). Attitude inferences from salient and rele-
vant cognitive content about behavior. Journal of Personality and Social Psy-
chology, 32, 829-840.
Sanbonmatsu, D. M., & Fazio, R. H. (1990). The role of attitudes in mem-
ory-based decision making. Journal of Personality and Social Psychology, 59,
614-622.
Sanbonmatsu, D. M., Posavac, S. S., Vanous, S., Ho, E. A., & Fazio, R. H. (2007).
The deautomatization of accessible attitudes. Journal of Experimental Social
Schuman, J., & Presser, S. (1981). Questions and answers in attitude surveys: Experi-
ments in question form, wording, and context. New York: Academic Press.
636 FAZIO
Schwarz, N. (2007). Attitude construction: Evaluation in context. Social Cogni-

tion, 25(5), 638-656.
Schwarz, N., & Bohner, G. (2001). The construction of attitudes. In A. Tesser & N.
Schwarz (Eds.), Blackwell handbook of social psychology: Intraindividual pro-
cesses (pp. 436-457). Maiden, MA: Blackwell.
Seibt, B., Hafner, M., & Deutsch, R. (2007). Prepared to eat: How immediate affec-
tive and motivational responses to food cues are influenced by food depri-
vation. European Journal of Social Psychology, 37,359-379.
Sherman, S. J., Rose, J. S., Koch, K., Presson, C. C , & Chassin, L. (2003). Implicit
and explicit attitudes toward cigarette smoking: The effects of context and
motivation. Journal of Social & Clinical Psychology, 22,13-39.
Shiffrin, R. M., & Dumais, S. T. (1981). The development of automatism. In J. R.
Anderson (Eds.), Cognitive skills and their acquisition (pp. 111-140).
Hillsdaie, NJ: Erlbaum.
Shook, N. J., Fazio, R. H., & Eiser, J. R. (2007). Attitude generalization: Similarity,
valence, and extremity. Journal of Experimental Social Psychology, 43,
641-647.
Sinclair, L., & Kunda, Z. (1999). Reactions to a Black professional: Motivated in-
hibition and activation of conflicting stereotypes. Journal of Personality and
Social Psychology, 77, 885-904.
Smith, E.R., Fazio, R. H., & Cejka, M. A. (1996). Accessible attitudes influence cat-
egorization of multiply categorizable objects. Journal of Personality and So-
cial Psychology, 71,888-898.
Strack, F., Martin, L. L., & Stepper, S. (1988). Inhibiting and facilitating condi-
tions of the human smile: A nonobtrusive test of the facial feedback hy-
pothesis. Journal of Personality and Social Psychology, 54, 768-777.
Tetlock, P. E., & Kim, I. (1987). Accountability and judgment processes in a per-
sonality prediction task. Journal of Personality and Social Psychology, 52,
700-709.
Thurstone, L. L. (1928). Attitudes can be measured. American Journal of Sociology,
33,529-554.
Tourangeau, R., & Rasinski, A. (1988). Cognitive processes underlying context
effects in attitude measurement. Psychological Bulletin, 103,299-314.
Uleman, J. S. (1999). Spontaneous versus intentional inferences in impression
formation. In S. Chaiken & Y. Trope (Eds.), Dual-process theories in social
psychology (pp. 141-160). New York: Guilford Press.
Wallace, D. S., Paulson, R. M., Lord, C. G., & Bond, C. F. (2005). Which behaviors
do attitudes predict? Meta-analyzing the effects of social pressure and
perceived difficulty. Review of General Psychology, 9, 214-227.
Wicker, A. W. (1969). Attitudes versus actions: The relationship of verbal and
overt behavioral responses to attitude objects. Journal of Social Issues, 25,
41-78.
Wilson, T. D., Hodges, S. D., & LaFleur, S. J. (1995). Effects of introspecting about
reasons: Inferring attitudes from accessible thoughts. Journal of Personality
and Social Psychology, 69,16-28.
Wilson, T. D., Lindsey, S., & Schooler, T. Y. (2000). A model of dual attitudes. Psy-
chological Review, 107,101-126.
Wittenbrink, B., Judd, C. M., & Park, B. (2001). Spontaneous prejudice in context:
Variability in automatically activated attitudes. Journal of Personality and
Social Psychology, 81,815-827.
Zaller, J., & Feldman, S. (1992). A simple theory of the survey response: Answer-
ing questions versus revealing preferences. American Journal of Political
Science, 36,579-616.
Zanna, M. P., & Fazio, R. H. (1982). The attitude-behavior relation: Moving to-
ward a third generation of research. In M. P. Zarma, E. T. Higgins, & C. P.
Herman (Eds.), Consistency in social behavior: The Ontario symposium (Vol. 2,
pp. 283-301). Hillsdale, NJ: Erlbaum.
Zanna, M. P., Fazio, R. H., & Ross, M. (1994). The persistence of persuasion. In R.
C. Schank & E. Langer (Eds.), Beliefs, reasoning, and decision making: Psy-
cho-logic in honor of Bob Abelson (pp. 347-362). Hillsdale, NJ: Erlbaum.
Zanna, M. P., & Rempel, J. K. (1988). Attitudes: A new look at an old concept. In
D. Bar-Tal & A. W. Kruglanski (Eds.), The social psychology ofknowledge (pp.
315-334). Cambridge, UK: Cambridge University Press.

Fazio 2007

Uploaded by

Copyright:

Available Formats

Fazio 2007

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fazio 2007

Uploaded by

Copyright:

Available Formats

Social Cognition, Vol. 25, No. 5,2007, pp.

model provides regardirig some recent discussions of attitudes as cor\-

evidence suggested that verbal reports of attitudes bore Uttle relation to

ATTITUDES: HYPOTHETICAL CONSTRUCTS?

There is, however, an additional sense in which a definition of attitude

the distinction that I wish to emphasize here is most clearly articulated

Like many psychological variables, attitude is a hypothetical or latent

Conceptualized in this way, then, "attitudes" do not exist as entities

Awareness of this background regarding the history of the attitude con-

erature. First, there appears little value to defining attitudes in such a

ATTITUDES AS OBJECT-EVALUATION ASSOCIATIONS

With this background in mind, let me summarize the view of attitudes

some individuals are characterized by strong associations. Others are

acterization of automatic as "all processes whose initiation the subject

(SOME) ATTITUDES ARE REPRESENTED IN MEMORY

Whether one pursues an associative network or a connectionist ap-

related to the participants' primary task of identifying the cormotation of

lay. Thus, experimental manipulations that serve to strengthen, or to

of these object names influenced how subsequent ambiguous stimuli

tencies to answer correctly, across the various experiments). They also

(SOME) ATTITUDES ARE CONSTRUCTED

As emphasized by the notion of object-evaluation associations varying

represented in memory with strongly associated evaluative linkages.

framework and a more affectively oriented measure. Only individuals

PROBLEMS WITH A STRONG "ATTITUDES AS

In light of the research summarized in the preceding section, I have to

Ultimately, some relevant evaluation—some outcome of prior learning

tails—realizes, this extreme assumption is unrealistic" (Schwarz &

THE MODEL'S PERSPECTIVE REGARDING A FEW

deliberative analysis of the attributes that characterize the attitude ob-

measure is an exercise in verbal behavior and, hence, does not speak to

ON THE PRESUMED MALLEABILITY OF

Research involving implicit measures of attitude focused at first on

the IAT is sensitive, or (b) simply render momentarily salient

provide evidence strongly suggesting that appropriately designed so-

framed it above, as stemming from the subordinate attribute level (e.g,

ings indicative of ambivalence, effectively noting that their evaluations

SUMMARY AND CONCLUSIONS

Responses to explicit measures of attitudes are themselves verbal be-

ing decision making. Journal of Personality and Social Psychology, 64,

Fazio, R. H. (1990). Multiple processes by which attitudes guide behavior: The

Psychological insights and perspectives (pp. 41-62). Thousand Oaks, CA:

tally-created extrapersonal associations on the Implicit Association Test.

Schwarz, N. (2007). Attitude construction: Evaluation in context. Social Cogni-

You might also like