Content uploaded by Marina Chernikova
Author content
All content in this area was uploaded by Marina Chernikova on Apr 12, 2018
Content may be subject to copyright.
*
University of Maryland, College Park
** University of Rome – La Sapienza
Corresponding author: ???
Original Papers
Polish Psychological Bulletin
2018, vol. 49(1) 31–39
DOI - 10.1515/ppb-2018-00xx
Gordon Allport once wrote that “the concept of
attitude...is the most distinctive and indispensable concept
in contemporary American social psychology” (Allport,
1935, p. 798), and the immense amount of research
on attitudes since then suggests he may have been right
(e.g., Ajzen & Fishbein, 1975; Albarracin, Johnson,
& Zanna, 2005; Bentler & Speckart, 1979; Eagly &
Chaiken, 1993; Kelman, 1958; Petty & Cacioppo, 1986;
Triandis, 1971). The term “attitude” yields almost half
a million results in a PSYCInfo search, much more
than other popular search terms such as “decision”,
“cognition”, or even “goals” (each of which yields less
than a quarter of a million results). Presumably, one
reason for the tremendous popularity of attitudes is that
they are commonly believed – by both laypeople and
researchers alike – to predict behavior. But is this belief
justified?
Attitudes as predictors of behavior
Attitudes are typically defined as either positive or
negative evaluations of some object (e.g., Bem, 1970;
Cacioppo, Harkins, & Petty, 1981; Eagly & Chaiken, 1993;
Fazio, 2007; Fishbein & Ajzen, 1975; Petty, Wegener, &
Fabrigar, 1997; Thurstone, 1931). Attitudes were viewed as
important predictors of behavior in classic psychological
research: for instance, Cohen (1960) saw attitudes as
“determinants of how a person will actually behave in
his daily affairs” (pp. 137–138). Likewise, Allport (1929)
concluded that an attitude reflected a “disposition to act”
(p. 221). Petty and Wegener (1998) wrote that “attitudes
[are] important because of the fundamental role that
individuals’ attitudes...play in the critical choices people
make regarding their own health and security as well as
those of their families, friends, and nations” (p. 3230).
Given the assumption that attitudes, in general, were
Arie W. Kruglanski*
Conrad Baldner**
Marina Chernikova*
Calogero Lo Destro**
Antonio Pierro**
A new perspective on the attitude-behavior relation:
The essential function of goals
Abstract: In theory, knowing an individual’s attitude about a topic should allow us to predict his or her behavior.
However, in a classic study, Wicker (1969) came to the surprising conclusion that attitudes and behaviors are only
weakly related. We present a new theoretical perspective that describes the conditions necessary for an attitude to be
translated into a behavior. More specifically, we propose that an attitude (i.e., liking of an end state) is not sufficient
to cause behavior. Rather, that liking must first become a desire, which will only occur if an individual likes a potential
future state more than the present state. The desire must subsequently be transformed into a goal, which will only occur
if the desire is perceived as attainable. The goal must then become a focal goal (i.e., be momentarily dominant in an
individual’s goal system). Lastly, in order for a particular behavior to be enacted, it must be selected as a means that
serves the focal goal. We offer empirical evidence for our theory and describe how it goes beyond previous models of
attitude-behavior relations, such as the Theory of Planned Behavior (Fishbein & Ajzen, 2010) and the MODE model
(Fazio, 1990).
Keywords: attitude, behavior, liking, want, desire, means, goal
32 Arie W. Kruglanski, Conrad Baldner, Marina Chernikova, Calogero Lo Destro, Antonio Pierro
important to the prediction of behavior, researchers
subsequently set out to find the specific kinds of attitudes
that predict behavior, as well as potential moderators of the
relationship between attitudes and behaviors (e.g., Ajzen,
1985, 2012, 2015; Fabrigar, Petty, Smith, & Crites, 2006;
Fazio, Chen, McDonel, & Sherman, 1982; Fazio, Powell, &
Williams, 1989; Fazio & Williams, 1986; Fishbein & Ajzen,
1975, 2010). Although this research has extended into the
present day, not everyone is convinced that attitudes play
a strong role in the prediction of behaviors. Indeed, based
on an extensive review of the literature up to that point,
Wicker (1969) concluded that attitudes had little or no
association with behavior. Many researchers have since
responded to Wicker’s claim, but first, it is useful to look at
the kind of evidence that he presented.
Wicker cited a meta-analysis conducted by Vroom
(1964), which examined fifteen studies on the relationship
between employee’s attitudes toward their work and
employer-rated job performance (i.e., behavior) in a variety
of fields. The total correlation between attitudes and behavior
was only .14 – and even this was positively skewed by
a disproportionately high correlation in one study, in which
the validity of the results was questioned. Likewise, Dean
(1958) studied factory workers’ attitudes towards labor
unions and their participation in union meetings. Roughly
80% of participants did not attend any meetings, even though
91% of non-attendees had favorable attitudes towards the
union. Interestingly, only 37% of non-attendees (vs. 69%
of attendees) indicated that work conditions had improved
since the union arrived. This could indicate that behavior is
not merely dependent on attitudes, but also on the perceived
utility of the behavior – a point we will return to later.
However, Wicker’s conclusion about the practicality
of using attitudes to predict behaviors extended beyond the
workplace. Wicker also cited Linn’s (1965) research on
attitudes and behaviors towards minority groups, conducted
during a time when the American civil rights movement was
in full swing. Linn asked white women whether they would
be hypothetically interested in being photographed with
a black man; this photograph would be widely distributed.
Later, the participants were asked, ostensibly as a part of
a racial integration program, if they would consent to be
photographed with a black man. 59% of the sample showed
a discrepancy between their attitude and behavior towards
being photographed; this discrepancy was strongest amongst
women who had previously shown more positive attitudes.
It was on the basis of these and many similar studies
that Wicker concluded that attitudes had, at best, a weak
relationship with behaviors. If true, this would be a grave
blow to research on attitudes, as well as to behavioral
interventions that rely on changing underlying attitudes.
However, subsequent researchers have challenged
Wicker’s conclusions and set out to uncover if, and when,
attitudes are useful predictors of behavior. This was largely
undertaken by shifting how researchers conceptualize
attitudes. That shift can be seen in the research on attitude
strength (e.g., Krosnick & Petty, 1995), or the notion that
only strongly held attitudes should lead to behavior, as
well as research on attitudes toward specific behaviors
(e.g., Fishbein & Ajzen, 2010), or the notion that behaviors
are only predicted by specific attitudes towards those
behaviors. In what follows, we explain how our theory
addresses the question of when and whether attitudes
lead to behavior, then elaborate on what our theory has
to offer that was missed by prior formulations such as
the aforementioned attitude strength and attitudes toward
specific behaviors frameworks.
The “rocky road” from attitudes to behavior
We propose that there is a rocky stretch of theoretical
road between attitudes and behaviors (Kruglanski et al.,
2015). A general attitude must first become a desire – that
is, liking of an attitude object must transform into wanting
(desire) for that object. This desire must then be joined by an
attainability, or an expectancy that the desired object can be
reached. Some above-threshold level of both desirability and
attainability is necessary for a goal to be formed (Kruglanski
et al., 1996; Kruglanski, Chernikova, Rosenzweig, &
Kopetz, 2014; Oettingen et al., 2001). Even once a goal
has been formed, however, it will not necessarily be active
in any given moment (Bargh, 1990; Bargh et al., 2001;
Fishbach & Ferguson, 2007; Stroebe, van Koningsbruggen,
Papies, & Aarts, 2013). For instance, some individuals
may have the general goal of eating healthy, but that goal
will likely not be active when they are in class listening to
a lecture. In addition, people can simultaneously have many
different goals, so a goal must be sufficiently dominant in
a given context in order to affect behavior (otherwise it will
be overridden by other, more pressing, concerns). Finally,
even when a goal is dominant, this does not mean that
a specific goal-related behavior will necessarily be pursued.
In order for a behavior to be enacted, it must first be selected
as the preferred means to the currently dominant goal. In
summary, in order for behavior to occur, multiple conditions
must be met: (1) an attitude must become a desire; (2) the
desire must be joined with a sufficient level of attainability
to form a goal; (3) the goal must be activated; (4) the goal
must be sufficiently dominant; and (5) a preferred means
to the goal (i.e., a behavior) must be selected. We expand
upon each of these conditions in the following sections (see
Figure 1 for a general overview of the theory).
Figure 1. The “rocky road” from attitude (i.e., relative
liking) to behavior
Goal 1
RL 1 A 1
Goal 2
A 2W 2RL 2
W 1
Goal 3
A 3W 3RL 3
Behavior 2
Behavior 1
Behavior 3
Behavior 4
Final
Behavior
Focal
Goal
RL = Relative Liking; W = Wanting; A = Attainability.
33
New perspective on the attitude-behavior relation
From attitudes to desire
In spite of the fact that the starting point of our theory
is a general attitude toward some object, we argue that
although such attitudes are relevant to behavior, they are
not useful predictors of behavior in and of themselves. In
order for an attitude to have an effect on behavior, it must
first be transformed into a desire (or wanting). Desire is
a product of relative liking, defined as one’s greater (or
lesser) liking of a future state, compared the current state.
Thus, in line with many prior theories of motivation, we
posit that wanting stems from a discrepancy between
liking for the present and liking for some future state
(cf. Carver & Scheier, 1982, 1998; Custers & Aarts, 2005,
2007; Higgins, 1987; Miller, Galanter, & Pribram, 1960;
Oettingen, Pak, & Schnetter, 2001; Wiener, 1948). Relative
liking can be divided into relative promotive liking (i.e.,
the future state is liked more than the current state) and
relative preventive liking (i.e., the current state is liked
more than the future state). If one is hungry, then a future
state in which she eats a favorite food is more positive that
her current hungry state (i.e., she has a relative promotive
liking towards the future state of having eaten her favorite
food). If she has just eaten a large meal, then the current
state in which she is satiated is more positive than a future
state in which she would be unbearably full (i.e., she has
a relative preventive liking towards refraining from eating
more). In each case, it is not mere liking, but relative liking
that contributes to behavior.
The distinction between liking and wanting may seem
strange at first glance, as surely individuals who like an
object would also want that object. Despite this seemingly
obvious connection, prior research and theorizing (e.g.,
Lewin, 1972) suggests that they are actually distinct
constructs. Berridge and his colleagues have argued that
liking is associated with the neurotransmitter GABA,
whereas wanting is associated with dopamine (Berridge,
2004; Berridge & Robinson, 1995, 2003; Berridge,
Robinson, & Aldridge, 2009; Pecina, Cagniard, Berridge,
Aldridge, & Zhuang, 2003; Wyvell & Berridge, 2000).
Thus, even though wanting and liking are sometimes used
synonymously in everyday parlance, they are mediated
by different neurotransmitters and, as a result, can appear
separately. Indeed, the difference between liking and
wanting can help explain the behavior of drug addicts
who use to avoid withdrawal symptoms: these individuals
often have the desire (or want) to use drugs even though
they do not receive positive feedback from their drug of
choice (Berridge, 2004). We propose that this distinction
is also present in more mundane matters. For instance,
as described earlier, one can have a positive attitude (i.e.,
liking) for her favorite food, but would not have the desire
(i.e., want) for it if she is already full. An individual may
have a positive attitude towards the benefits won by labor
unions (e.g., better wages) but he may not want more
of these benefits if his employer is facing bankruptcy.
Individuals can have many attitudes that do not become
desires, and it is precisely these attitudes that are unrelated
to behavior.
Goal formation
Although desire brings us closer to behavior, desire
alone is not sufficient. Individuals can have many desires,
many or even most of which are not attainable. For
instance, an individual could desire to become president of
her country. In our terminology, she has a positive attitude
towards being president, which has been translated into
a relative promotive liking towards the future state of being
president, relative to the current state of not being president.
However, she will only form the goal of becoming
president if she perceives this desire to be attainable.
Although desire is more important for goal formation
than attainability (i.e., an individual will not form a goal
to do something that is highly attainable but undesirable;
Kruglanski, Chernikova, Rosenzweig, & Kopetz, 2014),
an above-threshold level of both desire and attainability
is necessary for a goal to be created (Atkinson, 1964;
Kruglanski et al., 1996; Kruglanski et al., 2014; Oettingen
et al., 2001; Vroom, 1964).
Past attitude research has either ignored the role of
attainability or held it constant in a way that stripped it
of its power. For instance, Fazio, Powell, and Williams
(1989) examined the roles of attitude strength and attitude
accessibility in a study of product preference. Participants
rated a series of 100 products (e.g., peanuts, candy bars,
soda) on a scale of “like” or “dislike”; their response
latency was assessed. Participants then rated the same
products on a 7-point scale in order to assess the strength
of their attitudes. Afterwards, participants were presented
with a subset of 10 products, and were allowed to take
five home. The products that they selected, as well as the
order in which they selected them, were predicted by both
the strength and accessibility of their attitudes. However,
this design rendered attainability irrelevant, because each
product was equally attainable. Even if they did not pick
a particular product, participants could have easily procured
these products for themselves. We argue that participants
would have been less likely to select products that they
liked when those products were less (vs. more) attainable.
From goal formation to goal activation and dominance
Even though individuals who desire a specific object
and perceive this desire as attainable may form a goal, the
mere presence of a goal is necessary but not sufficient for
behavior to occur. In order to cause behavior, a goal needs
to be active in a given situation; a dormant goal that is not
of current concern is unlikely to lead to action (Eitam &
Higgins, 2010). Even when a goal is currently active, an
individual can have many active goals, some of which
are likely mutually exclusive (Kruglanski et al., 2002).
Thus, a goal must become sufficiently dominant in a given
situation before it can influence behavior, so that it is not
overridden by other, more pressing concerns. To give an
example, if an individual has the desire to eat her favorite
meal and perceives this as attainable, then she forms the
goal to eat this meal. The goal might become activated
when she passes her favorite restaurant in the street. At the
same time, though, she has other, conflicting goals. For
example, she has the goals to save money, lose weight, and
34 Arie W. Kruglanski, Conrad Baldner, Marina Chernikova, Calogero Lo Destro, Antonio Pierro
shop in a nearby clothing store, and not all of these goals
can be pursued simultaneously. One of these goals will
become dominant – the focal goal – and it is this goal that
will be pursued via a particular behavior (e.g., eating at the
restaurant).
From goal dominance to behavior
Once a goal has become sufficiently dominant, an
individual can choose from a range of behaviors that he
or she perceives as instrumental for attaining that goal.
Importantly, however, instrumentality is not the only
consideration in means selection. Individuals often have
other active goals in addition to their focal goal, and some
behaviors can serve as means for more than one goal. Thus,
a second criterion for means selection is multifinality, or
the extent to which a means can serve more than one goal
(Kopetz, Faber, Fishbach, & Kruglanski, 2011; Kruglanski
et al., 2002; Kruglanski, Kopetz et al., 2013). For instance,
if one has the focal goal of eating her favorite meal and
the supplemental goal of saving money, she could select
a behavior that is instrumental for both, such as cooking
the meal at home instead of buying it from an expensive
restaurant. Nonetheless, there is often a tradeoff between
instrumentality and multifinality, such that the more goals
a means serves, the less it is perceived as instrumental to
any single goal (Zhang, Fishbach, & Kruglanski, 2007).
Thus, the final behavior chosen to serve the goal will be
determined in part by whether the focal goal is dominant
enough to cause an individual to “forget all else” and
choose only the most instrumental means (Shah, Friedman,
& Kruglanski, 2002; Huang & Bargh, 2014), and in
part by the extent to which the person generally values
instrumentality more than multifinality (e.g., Orehek,
Mauro, Kruglanski, & van der Bles, 2012).
To summarize the theory, a person’s positive attitude
towards his favorite meal is related to the behavior of
eating that meal in the following way: his positive attitude
towards eating the meal must be viewed as more positive
than his current state of not having this meal; eating the
meal must be perceived as attainable; the resulting goal of
eating this meal must be active and dominant over other
goals; and he must perceive the chosen behavior as an
instrumental means to reach the goal of eating this meal.
As can be gleaned from this overview, there are multiple
“bridges” in between our attitudes and behaviors; if these
bridges are not crossed, then an individual’s attitude will be
unrelated to his or her behavior. Within any given sample
of participants who report their attitudes and behaviors,
we would expect that most of them would not cross all the
bridges, and consequently the correlation between their
attitudes and behaviors should typically be low, and perhaps
indistinguishable from zero (consistent with Wicker’s
original claims).
Revisiting past research
We next turn to the question of precisely how
our theory can explain the weak relationship between
attitudes and behavior that was demonstrated in past
research. Vroom (1964) found a tenuous correlation
between employee-rated attitude towards their work and
employer-rated work performance. This is to be expected,
given the many steps that we propose lie between attitudes
and behaviors. The employee’s attitude towards their
work can of course take many forms (e.g., their attitude
towards the work itself, toward the company, or toward
their wages), but even if it takes the form of an attitude
towards their specific work tasks, it will not necessarily
translate into a desire to do this work, nor is this desire
necessarily attainable in cases in which the work is difficult.
Even if employees have the goal of performing well at
work, they likely also have other goals (e.g., socializing
with colleagues, spending time with their family), which
may sometimes or even often take precedence over the
former goal. These additional factors make it unlikely
that employees’ work-related attitudes will predict their
work-related behaviors, unless their goal to perform
well is active and sufficiently dominant during the
workweek.
We next turn to Dean’s (1958) conclusion that
attitudes towards labor unions among factory workers was
a poor predictor of their attendance of union meetings.
Even if the measured attitude pertains to the benefits that
labor unions can provide, it is not necessarily the case
that these workers prefer the future state of having those
benefits over the status quo (i.e., their current state). Thus,
it is not clear that this attitude reflects a desire, which – as
described above – is necessary to cause behavior. Beyond
this, Dean also found that a majority of attendees at the
meetings, but only a minority of non-attendees, perceived
that working conditions had improved since the union
began operation. In our terms, this could indicate either
that non-attendees did not perceive that the benefits
of labor unions to be attainable, or that non-attendees
did not believe that attending labor union meetings
was instrumental for accomplishing their goals. In our
formulation, either of these perceptions would result in
a lack of attitude-related behavior.
Finally, we turn to Linn’s (1965) conclusion that the
attitudes of white women toward being photographed
with a black man were very weakly related to actually
consenting to be photographed for this purpose. The
caveats described above likely also apply here. For
instance, the attitude toward being part of this program
may not have been more positive than the current state
of not being in the program; participants may not have
seen the goal of racial integration as attainable; and the
behavior of being photographed with a black man may
not have been viewed as an instrumental means towards
this goal. In addition, given the very violent history of
racial relations in the United States, a white woman and
black man being photographed together in the early 1960s
could be a dangerous proposition for both parties. From
the point of view of the white women in Linn’s study,
the negative consequences could include at least a loss of
social status. Thus, even if participants had an active goal
to further racial integration, many participants likely also
had the goal of avoiding social punishment; this goal may
35
New perspective on the attitude-behavior relation
have been more dominant in the situation than the goal
of racial integration, which would have prevented them
from acting in accordance with their attitudes about racial
integration. We suspect that if Linn had offered participants
a third option – a behavioral means that could satisfy
both the focal goal of avoiding social punishment and the
supplementary goal of furthering racial integration – many
participants would have selected that option.
The current state of attitude research
Although we have proposed that goals play a critical
role in the prediction of behavior, previous research on
attitudes has either ignored goals or suggested they play
an unimportant role. We now turn to two major research
paradigms on attitude-behavior consistency – the attitude
strength and attitude toward the behavior frameworks –
and examine their relation to the present model. Both
perspectives differ from the present approach in several
important respects. For one, they focus on attitudes as the
main predictor of behavior, while our theory assigns that
critical role to goals. Second, both of the aforementioned
models emphasize absolute liking for an object or
behavior. On the other hand, our model focuses on the
notion of relative liking and highlights the critical role of
the discrepancy between current and future states. Third,
the prior perspectives ignore the many bridges that must
be crossed before an attitude can become a behavior. In
contrast, the current “rocky road” model elaborates upon
all of those contingencies and argues that it is necessary
to emphasize the role of goals in order to fully understand
how human behavior unfolds (Kruglanski et al., 2015).
Attitude strength
Krosnick and Petty (1995) argued that strong attitudes
have four components: resistance to change, persistence,
an impact on information processing, and an impact on
judgments. According to their theory, only strong attitudes
should resist change, be stable over time, influence how
we process information, and affect our judgments – and
behaviors. Weak attitudes should have few or none of these
components. Thus, if attitudes are only weakly related to
behaviors, one potential explanation is that the attitudes
in question were weakly held. However, subsequent
research on attitude strength concluded that the separate
components of attitude strength actually represent different
constructs, which in turn must be measured with different
methodologies (Krosnick, 1993, 1994).
Consistent with that line of reasoning, Fazio (1990,
1995) argued that attitude accessibility, or the speed
at which an attitude toward an object comes to mind, is
indicative of attitude strength. Furthermore, attitude
accessibility can be assessed through response latency, or
the amount of time participants take in order to indicate
their attitude. Theoretically, attitudes must be active – that
is, accessible – in order to have any effect on behavior.
Attitudes towards a particular object can be chronically
accessible, but can also be primed by environmental
stimuli. An individual who has a readily-accessible attitude
towards a particular object likely has a strong attitude (e.g.,
they have formed this attitude some time in the past). On
the other hand, an individual with a relatively inaccessible
attitude towards an object most likely has a weak attitude
(e.g., they may have to “create” an attitude in the moment).
The latter type of attitudes are more likely to change over
time, and thus are less likely to be related to behavior.
However, Fazio considered attitude accessibility to be just
one aspect of a larger model: the MODE model (Fazio,
1990; Fazio, & Towles-Schwen, 1999; Olson & Fazio,
2008; Schuette, & Fazio, 1995).
According to the MODE model, there are two
general types of attitude-behavior processes: spontaneous
processes, in which the behavior follows from an attitude
that is automatically activated upon perception of the
attitude object, and deliberative processes, in which
behavior follows from an attitude created in a laborious
cost-benefit analysis (Fazio, 1990). Unlike the spontaneous
process, the deliberative process is a time-consuming one
that requires motivation, time, and cognitive resources
(i.e., opportunity). In both processes, the attitude construct
is the same; the critical difference between them is that, in
the spontaneous process, strong attitudes are pre-existing
and readily available, whereas in the deliberative process,
strong attitudes are the result of an effortful process, and
are formed only if an individual has sufficient motivation
and opportunity.
Importantly, the research program on attitude strength
has neglected the critical role of goals in driving behavior.
According to our formulation, no matter how strong
or accessible the attitude, it would fail to lead to overt
behavior unless the other aspects of the model (want,
attainability, etc.) were present. In our view, although some
level of attitude strength is necessary to create wanting, it
is not sufficient for creating wanting on its own. There must
also be a discrepancy between an individual’s liking for the
present state and his or her liking for a future state in order
to create a want. A close inspection of the many studies on
attitude strength reveals that, in fact, there was often a goal
lurking in the background in each study (see Kruglanski et
al., 2015, for more details).
Attitudes towards specific behaviors
The research summarized by Wicker, as well as the
research undertaken by Krosnick, Petty, and Fazio had
at least one common component: each model studied the
relation between general attitudes and behavior. A different
approach was taken by Ajzen and Fishbein in their theory
of reasoned action (TRA; Fishbein & Ajzen, 1975; Ajzen
& Fishbein, 1980) and theory of planned behavior (TPB;
Ajzen, 1988, 1991). The underlying theme of these theories
is that while general attitudes can predict an aggregate of
relevant behaviors (Ajzen, 2012; Fishbein & Ajzen, 1974,
2010), only specific attitudes toward the behavior, and
not general attitudes toward the object, are predictive of
specific behaviors. For instance, an individual’s attitude
toward labor unions (i.e., their general attitude toward the
object) may predict an aggregate of relevant behaviors, but
will not predict attendance at union meetings or any other
36 Arie W. Kruglanski, Conrad Baldner, Marina Chernikova, Calogero Lo Destro, Antonio Pierro
specific behavior. Instead, attitudes towards the behavior
of attending union meetings should be a better predictor of
that specific behavior.
It is worthwhile to go deeper into precisely what
is proposed by TRA and TPB. Both theories focus on
behavioral intentions as the outcome of interest, rather
than behavior. This accounts for the possibility that
individuals will plan to engage in behaviors that are at
least momentarily impossible (e.g., because they lack the
skills, resources, or opportunities to enact the behavior).
The TRA postulated the existence of two predictors of
behavior: attitudes towards the behavior (which reflects
the sum of the beliefs about the potential outcomes of the
behavior) and subjective norms (which reflects both the
expectations of important others and how these important
others themselves behave). The TPB retained those two
factors and added perceived behavioral control, which
reflects the degree to which individuals believe that they
can overcome internal and/or external obstacles in order
to perform the specific behavior. Each factor in the theory
is measured in regard to a specific behavior: for instance,
in order to predict an individual’s behavioral intention
to walk her dog on Sunday evening, researchers should
measure her attitude toward walking the dog on Sunday
evenings, her subjective norms about walking the dog on
Sunday evenings, and her perceived behavioral control over
walking the dog on Sunday evenings. The TRA and TPB
have been used to predict a wide range of behaviors such as
exercise (Hausenblas, Carron, & Mack, 1997), condom use
(Carmack & Lewis-Moss, 2009), adherence to treatments
for chronic illness (Rich, Brandes, Mullan, & Hagger,
2015), and binge drinking (Smith, Coyle, Baldner, Bray,
& Geller, 2013).
Nonetheless, research within the TRA/TPB framework
has again ignored the critical role of goals in driving
behavior. According to our theory, neither the attitude
toward a behavior, nor a subjective norm, nor perceived
behavioral control would predict behavioral intentions
(or behavior) unless the goal that the behavior serves was
activated. Thus, in contrast to the TRA/TPB framework,
we suggest that general object attitudes can, in some cases,
drive behavior (assuming the rest of the contingencies
in our model are satisfied). In addition, we claim that
behavioral intentions are always goal-driven, because the
intention to perform any given behavior will be formed
only when the behavior in question serves as a means to
a currently active goal. To give an example, if an individual
likes writing (attitude toward the behavior), the person’s
friends and acquaintances praise writing (subjective norms),
and the person is able to write (perceived behavioral
control), he or she will still form the behavioral intention
to write (and engage in the behavior) only if a relevant goal
– such as publishing an article in a journal – is active and
dominant. If a different goal (such as going to a concert
with friends) is dominant instead, the person will not
form the behavioral intention to write, even if he or she
has a positive attitude about writing, subjective norms
for writing, and high perceived behavioral control about
writing.
Empirical evidence for our theory
In the preceding paragraphs, we have outlined the
many theoretical steps that lie between attitudes and
behavior, and described how prior formulations have
neglected many of these steps. Of course, it is also
important to collect empirical data that directly tests our
model. We will now describe three such studies that were
originally reported in Kruglanski et al. (2015).
The first empirical study we report pertains to the
critical role of desire in driving behavior (Kruglanski et al.,
2015, Study 2). In this study, participants at a university
dining hall who either had just eaten lunch or were about
to eat lunch were asked questions on the food that were
about to eat/had just eaten, how much they liked it, and how
much they wanted to eat it at that moment. The questions
about liking were used as a measure of attitude, whereas
the question about wanting was used as a measure of
desire. If attitude and desire are synonymous, as they are
often treated in the literature, then we would expect that
attitudes and desires would be highly related both before
and after participants had lunch. If, on the other hand,
desires but not attitudes are reduced after eating, then
we would have evidence that desire is more consistent
with relative liking than with attitude itself. Participants
who have not yet eaten, unlike participants who had just
eaten, could reasonably look forward to a state in which
they are no longer hungry. These participants could then
have an attitude towards this future state which is more
positive than the current hungry state. Consistent with our
expectations, participants who had not yet eaten had levels
of liking and wanting that were almost identical; on the
other hand, levels of wanting were significantly lower than
liking in participants who had just eaten.
The second study we report pertains to the interaction
between attitude, attainability, and accessibility on
behavioral choice (Kruglanski et al., 2015, Study 3). In this
study, students from a large Italian university participated
in an experiment on film genre preferences. Participants
reported their like or dislike of 16 film genres on
a computer program; their reaction time was also assessed
as a measure of attitude accessibility. Participants then
assessed their attitude towards each of these genres on 1–7
scales as a measure of attitude valence.
Participants were then asked to choose between
either watching a film from the thriller genre (a relatively
popular genre, according to a pre-test) or to solve math
problems. There were three experimental conditions. In
the high attainability condition, participants were told that
that if they chose to watch the film that they would have
a 70% chance of being assigned to watch the film and
a 30% chance of being assigned to solve math problems;
if they instead chose to solve math problems they had
a 100% chance of being assigned to this task. In the low
attainability condition, these probabilities were switched:
if they chose to watch the film there was a 30% chance
that would be assigned to do so and a 70% chance that they
would be assigned to solve math problems. The probability
that they would be assigned math problems if they
37
New perspective on the attitude-behavior relation
chose that task remained 100%. In the control condition,
participants were asked which they task they preferred
without reference to probabilities.
This research design allowed us to assess whether
attitude accessibility or attainability of a successful
outcome could moderate attitude valence towards that
outcome. If accessible attitudes lead to behavior, as argued
by attitude accessibility researchers (e.g., Fazio, Powell, &
Williams, 1989), then we would expect to find a significant
interaction of valence and accessibility, controlling for
attainability. If instead, as we expect, attainable desires (i.e.,
goals) must be present for behavior to occur, then we would
expect to find a significant three-way interaction between
valence, accessibility, and attainability. This result would
provide support for our theory because it would indicate
that the valence-accessibility interaction is importantly
moderated by attainability. Further, if accessibility is
viewed as a proxy for goal activation, then this result would
also lend support to our suggestion that desire, attainability,
and goal activation are all necessary steps on the path from
attitude to behavior.
In order to test our hypothesis, we first divided our
sample into control and experimental groups. In the control
group, we found evidence for a valence x accessibility
interaction, such that a strongly positive attitude towards
the thriller genre was associated with watching the film
when this positive attitude was associated with a fast
reaction time; this is consistent with the approach of Fazio
et al. (1989). In the experimental group, we were able to
examine the three-way interaction in addition to each of
the two-way interactions. We found significant effects for
both the valence x accessibility and valence x attainability
interactions as well as for the three-way interaction.
Breaking down the three-way interaction, we found that
the valence x attainability interaction was significant, such
that choosing to watch the film was most likely among
participants in the high attainability condition who had
strong and positive attitudes towards the thriller genre,
and only when these participants’ positive attitude was
associated with a fast reaction time. Thus, consistent with
our model, strongly positive attitudes, attainability, and
accessibility (which we view as a proxy for goal activation)
are important factors in behavioral prediction.
The third, and last, study we report pertains to the roles
of perceived instrumentality of the behavioral means and
goal formation in behavioral prediction (Kruglanski et al.,
2015, Study 5). Online participants completed measures
about their attitude toward the behavior of drinking alcohol
in social occasions, as well as the goal of having fun, the
instrumentality of drinking for the goal of having fun, and
the behavioral intention of drinking that night (data was
collected on Friday and Saturday evenings). Subjective
norms and perceived behavioral control with regard to
drinking alcohol were also assessed.
This research design allowed us to compare the
elements of our theory – specifically, the role of goals and
instrumental means – to the elements of the Theory of
Planned Behavior: attitude towards the behavior, subjective
norms, and perceived behavioral control (Ajzen, 2012).
If our model is correct, we would expect that goals and
instrumentality (rather than attitudes towards behavior,
subjective norms, and perceived behavioral control), would
be the most important predictors of behavioral intention.
Consistent with our expectations, we found support
for a moderated mediation model in which the effect of
attitudes towards the behavior on behavioral intention was
mediated by instrumentality; the path from instrumentality
to behavioral intention was moderated by the presence of
a goal. Subjective norms and perceived behavioral control
were not significant when attitudes towards the behavior
was included. The results from this study, as well as the
results from the other studies that we have presented,
are consistent with our model and inconsistent with rival
models. For further empirical studies on our model, and
further information about the above studies, see Kruglanski
et al. (2015).
Conclusion
To summarize, our “rocky road” model explicates the
precise conditions under which attitudes lead to behavior.
An attitude (or liking) toward an object is not sufficient
to create desire for the object in question, unless there is
a discrepancy (i.e., relative liking) between one’s current
state and a desired or undesired future state. Furthermore,
wanting in and of itself will not necessarily lead to goal
formation unless a sufficient degree of attainability is
present. In turn, goal formation will not lead to behavior
unless the goal in question is activated and dominant at
any given time. Lastly, any given behavior must be chosen
as the preferred means of goal attainment before it will
be carried out. We therefore argue that any attempts at
predicting human behavior must take an individual’s goal
constellation into account.
References
Ajzen, I. (1985). From intentions to actions: A theory of planned behavior.
In: J. Kuhl & J. Beckmann (Eds.), Action control: From cognition
to behavior (pp. 11–39). Heidelberg, Germany: Springer-Verlag.
http://dx.doi.org/10.1007/978-3-642-69746-3_2
Ajzen, I. (1988). Attitudes, personality, and behavior. Chicago, IL: Dors-
ey Press.
Ajzen, I. (1991). The theory of planned behavior. Organizational Behav-
ior and Human Decision Processes, 50, 179–211. http://dx.doi.
org/10.1016/0749-5978(91)90020-T
Ajzen, I. (2012). Attitudes and persuasion. In: K. Deaux & M. Snyder
(Eds.), The Oxford Handbook of Personality and Social Psychology
(pp. 367–393). New York, NY: Oxford University Press.
Ajzen, I. (2015). The theory of planned behaviour is alive and well, and
not ready to retire: a commentary on Sniehotta, Presseau, and Araú-
jo-Soares. Health Psychology Review, 9(2), 131–137.
Ajzen, I., & Fishbein, M. (1980). Understanding Attitudes and Predicting
Social Behavior. Eglewood Cliffs, NJ: Prentice Hall.
Albarracin, D., Johnson, B.T., & Zanna, M.P. (2005). The Handbook of
Attitudes. Mahwah, NJ: Erlbaum Publishers.
Allport, G.W. (1929). The composition of political attitudes. American
Journal of Sociology, 35, 220–238. http://dx.doi.org/10.1086/214980
Allport, G.W. (1935). Attitudes. In: C. Murchison (Ed.), A Handbook of
Social Psychology (pp. 798–844). Worcester, MA: Clark University
Press.
Atkinson, J.W. (1964). An introduction to motivation. Oxford, England:
Van Nostrand.
38 Arie W. Kruglanski, Conrad Baldner, Marina Chernikova, Calogero Lo Destro, Antonio Pierro
Bandura, A. (1993). Perceived self-effi cacy in cognitive development and
functioning. Educational Psychologist, 28, 117–148. http://dx.doi.
org/10.1207/s15326985ep2802_3
Bargh, J.A. (1990). Goal and intent: Goal-directed thought and behavior
are often unintentional. Psychological Inquiry, 1, 248–251. http://
dx.doi.org/10.1207/s15327965pb0103_14
Bargh, J.A., Gollwitzer, P.M., Lee-Chai, A., Barndollar, K., & Trötschel, R.
(2001). The automated will: Nonconscious activation and pursuit of
behavioral goals. Journal of Personality and Social Psychology, 81,
1014–1027. http://dx.doi.org/10.1037/0022-3514.81.6.1014
Bem, D.J. (1970). Beliefs, attitudes, and human affairs. Oxford, UK:
Brooks/Cole.
Bentler, P.M., & Speckart, G. (1979). Models of attitude – behavior rela-
tions. Psychological Review, 86(5), 452–464.
Berridge, K.C. (2004). Motivation concepts in behavioral neuroscience.
Physiology & Behavior, 81, 179–209. http://dx.doi.org/10.1016/j.
physbeh.2004.02.004
Berridge, K.C., & Robinson, T.E. (1995). The mind of an addicted brain:
neural sensitization of wanting versus liking. Current Directions in
Psychological Science, 4(3), 71–75.
Berridge, K.C., & Robinson, T.E. (2003). Parsing reward. Trends in Neu-
rosciences, 26, 507–513. http://dx.doi.org/10.1016/S0166-2236(03)
00233-9
Berridge, K.C., Robinson, T.E., & Aldridge, J.W. (2009). Dissecting com-
ponents of reward: ‘liking’, ‘wanting’, and learning. Current opin-
ion in pharmacology, 9(1), 65–73.
Cacioppo, J.T., Harkins, S.G., & Petty, R.E. (1981). The nature of attitudes
and cognitive responses and their relationships to behavior. In:
R.E. Petty, T.M. Ostrom, & T.C. Brock (Eds.), Cognitive responses
in persuasion (pp. 31–54). Hillsdale, NJ: Erlbaum.
Carmack, C.C., & Lewis-Moss, R.K. (2009). Examining the theory of
planned behavior applied to condom use: the effect-indicator vs.
causal-indicator models. The Journal of Primary Prevention, 30(6),
659–676.
Carver, C.S., & Scheier, M.F. (1982). Control theory: A useful con-
ceptual framework for personality-social, clinical, and health
psychology. Psychological Bulletin, 92, 111–135. http://dx.doi.
org/10.1037/0033-2909.92.1.111
Carver, C.S., & Scheier, M.F. (1998). On the self-regulation of behav-
ior. New York, NY: Cambridge University Press. http://dx.doi.org/
10.1017/CBO9781139174794
Cohen, A.R. (1960). Attitudinal consequences of induced discrepancies
between cognitions and behavior. Public Opinion Quarterly, 24,
297–318. http://dx.doi.org/10.1086/266950
Custers, R., & Aarts, H. (2005). Positive affect as implicit motivator:
On the nonconscious operation of behavioral goals. Journal of
Personality and Social Psychology, 89, 129–142. http://dx.doi.
org/10.1037/0022-3514.89.2.129
Custers, R., & Aarts, H. (2007). In search of the nonconscious sources of
goal pursuit: Accessibility and positive affective valence of the goal
state. Journal of Experimental Social Psychology, 43, 312–318.
http://dx.doi.org/10.1016/j.jesp.2006.02.005
Dean, L.R. (1958). Interaction, reported and observed: The case of one
local union. Human Organization, 17, 36–44.
Eagly, A.H., & Chaiken, S. (1993). The psychology of attitudes. Orlando,
FL: Harcourt Brace Jovanovich College Publishers.
Eitam, B., & Higgins, E.T. (2010). Motivation in mental accessibility:
Relevance of a representation (ROAR) as a new framework. Social
and Personality Psychology Compass, 4, 951–967. http://dx.doi.
org/10.1111/j.1751-9004.2010.00309.x
Fabrigar, L.R., Petty, R.E., Smith, S.M., & Crites, S.L., Jr. (2006). Under-
standing knowledge effects on attitude-behavior consistency: The
role of relevance, complexity, and amount of knowledge. Journal
of Personality and Social Psychology, 90, 556–577. http://dx.doi.
org/10.1037/0022-3514.90.4.556
Fazio, R.H. (1990). Multiple processes by which attitudes guide behavior:
The MODE model as an integrative framework. Advances in Ex-
perimental Social Psychology, 23, 75–109.
Fazio, R.H. (1995). Attitudes as object-evaluation associations: Deter-
minants, 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. (2007). Attitudes as object-evaluation associations of varying
strength. Social Cognition, 25, 603–637. http://dx.doi.org/10.1521/
soco.2007.25.5.603
Fazio, R.H., Chen, J.M., McDonel, E.C., & Sherman, S.J. (1982). Atti-
tude accessibility, attitude-behavior consistency, and the strength
of the object-evaluation association. Journal of Experimental So-
cial Psychology, 18, 339–357. http://dx.doi.org/10.1016/0022-
1031(82)90058-0
Fazio, R.H., Powell, M.C., & Williams, C.J. (1989). The role of attitude
accessibility in the attitude-to-behavior process. Journal of Con-
sumer Research, 16, 280–288. http://dx.doi.org/10.1086/209214
Fazio, R.H., & Towles-Schwen, T. (1999). The MODE model of at-
titude-behavior processes. In: Dual Process Theories in Social
Psychology, ed. S Chaiken, Y Trope, pp. 97–116. New York, NY:
Guilford.
Fazio, R.H., & Williams, C.J. (1986). Attitude accessibility as a mod-
erator of the attitude-perception and attitude-behavior relations:
An investigation of the 1984 presidential election. Journal of
Personality and Social Psychology, 51, 505–514. http://dx.doi.
org/10.1037/00223514.51.3.505
Fishbach, A., & Ferguson, M.J. (2007). The goal construct in social psy-
chology. In: A.W. Kruglanski & E.T. Higgins (Eds.), Social psy-
chology: Handbook of Basic Principles (pp. 490–515). New York,
NY: Guilford Press.
Fishbein, M., & Ajzen, I. (1974). Attitudes towards objects as predictors
of single and multiple behavioral criteria. Psychological Review,
81, 59–74. http://dx.doi.org/10.1037/h0035872
Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention, and behav-
ior: An introduction to theory and research. Reading, MA: Addison
Wesley Pub. Co.
Fishbein, M., & Ajzen, I. (2010). Predicting and changing behavior: The
reasoned action approach. New York, NY: Psychology Press.
Hausenblas, H.A., Carron, A.V., & Mack, D.E. (1997). Application of the
theories of reasoned action and planned behavior to exercise be-
havior: A meta-analysis. Journal of Sport and Exercise Psychology,
19(1), 36–51.
Higgins, E.T. (1987). Self-discrepancy: A theory relating self and affect.
Psychological Review, 94, 319–340. http://dx.doi.org/10.1037/
0033-295X.94.3.319
Huang, J.Y., & Bargh, J.A. (2014). The Selfi sh Goal: Autonomously oper-
ating motivational structures as the proximate cause of human judg-
ment and behavior. Behavioral and Brain Sciences, 37, 121–135.
http://dx.doi.org/10.1017/S0140525X13000290
Kelman, H.C. (1958). Compliance, identifi cation, and internalization
three processes of attitude change. Journal of Confl ict Resolution,
2(1), 51–60.
Köpetz, C., Faber, T., Fishbach, A., & Kruglanski, A.W. (2011). The
multifi nality constraints effect: how goal multiplicity narrows the
means set to a focal end. Journal of Personality and Social Psychol-
ogy, 100(5), 810–826.
Krosnick, J.A., Boninger, D.S., Chuang, Y.C., Berent, M.K., & Car-
not, C.G. (1993). Attitude strength: One construct or many relat-
ed constructs? Journal of Personality and Social Psychology, 65,
1132–1151. http://dx.doi.org/10.1037/0022-3514.65.6.1132
Krosnick, J.A., & Petty, R.E. (1995). Attitude strength: An overview. In:
R.E. Petty & J.A. Krosnick (Eds.), Attitude strength: Antecedents
and consequences (pp. 1–24). Mahwah, NJ: Erlbaum.
Krosnick, J.A., & Smith, W.A. (1994). Attitude strength. Encyclopedia of
Human Behavior, 1(1), 279–289.
Kruglanski, A.W. (1996). Goals as knowledge structures. In: P.M. Goll-
witzer & J.A. Bargh (Eds.), Linking cognition and motivation to
behavior (pp. 599–618). The psychology of action. New York, NY:
Guilford Press.
Kruglanski, A.W., Chernikova, M., Rosenzweig, E., & Köpetz, C. (2014).
On motivational readiness. Psychological Review, 121, 367–388.
http://dx.doi.org/10.1037/a0037013
Kruglanski, A.W., Jasko, K., Chernikova, M., Milyavsky, M., Babush, M.,
Baldner, C., & Pierro, A. (2015). The rocky road from attitudes to
behaviors: Charting the goal systemic course of actions. Psycho-
logical Review, 122(4), 598–620.
Kruglanski, A.W., Köpetz, C., Bélanger, J.J., Chun, W.Y., Orehek, E.,
& Fishbach, A. (2013). Features of multifi nality. Personality and
39
New perspective on the attitude-behavior relation
Social Psychology Review, 17, 22–39. http://dx.doi.org/10.1177/
1088868312453087
Kruglanski, A.W., Shah, J.Y., Fishbach, A., Friedman, R., Chun, W.Y., &
Sleeth-Keppler, D. (2002). A theory of goal systems. Advances in
Experimental Social Psychology, 34, 331–378.
Lewin, K. (1972). Need, Force and Valence in Psychological Fields. In:
E.P. Hollander & R.G. Hunt (Eds.), Classic Contributions to Social
Psychology. London, UK: Oxford University Press.
Linn, L.S. (1965). Verbal attitudes and overt behavior: A study of racial
discrimination. Social Forces, 44, 353–364.
Miller, G.A., Galanter, E., & Pribram, K.H. (1960). Plans and the struc-
ture of behavior. New York, NY: Henry Holt and Co. http://dx.doi.
org/10.1037/10039-000
Orehek, E., Mauro, R., Kruglanski, A.W., & van der Bles, A.M. (2012).
Prioritizing association strength versus value: the infl uence of self-
regulatory modes on means evaluation in single goal and multigoal
contexts. Journal of Personality and Social Psychology, 102(1),
22–31.
Oettingen, G., Pak, H., & Schnetter, K. (2001). Self-regulation of goal
setting: Turning free fantasies about the future into binding goals.
Journal of Personality and Social Psychology, 80, 736–753. http://
dx.doi.org/10.1037/0022-3514.80.5.736
Olson, M.A., & Fazio, R.H. (2008). Implicit and explicit measures of
attitudes: The perspective of the MODE model. In: R.E. Pet-
ty, R.H. Fazio, & P. Briñol (Eds.), Attitudes: Insights from the
new implicit measures (pp. 19–64). New York, NY: Psychology
Press.
Pecina, S., Cagniard, B., Berridge, K.C., Aldridge, J.W., & Zhuang, X.
(2003). Hyperdopaminergic mutant mice have higher “wanting” but
not “liking” for sweet rewards. Journal of Neuroscience, 23(28),
9395–9402.
Petty, R.E., & Cacioppo, J.T. (1986). Communication and persuasion:
Central and peripheral routes to attitude change. New York, NY:
Springer-Verlag.
Petty, R.E., Wegener, D.T., & Fabrigar, L.R. (1997). Attitudes and attitude
change. Annual Review of Psychology, 48, 609–647. http://dx.doi.
org/10.1146/annurev.psych.48.1.609
Petty, R.E., & Wegener, D.T. (1998). Matching versus mismatching at-
titude functions: Implications for scrutiny of persuasive messages.
Personality and Social Psychology Bulletin, 24(3), 227–240.
Rich, A., Brandes, K., Mullan, B., & Hagger, M.S. (2015). Theory of
planned behavior and adherence in chronic illness: a meta-analysis.
Journal of Behavioral Medicine, 38(4), 673–688.
Schuette, R.A., & Fazio, R.H. (1995). Attitude accessibility and motiva-
tion as determinants of biased processing: A test of the MODE mod-
el. Personality and Social Psychology Bulletin, 21(7), 704–710.
Shah, J.Y., Friedman, R., & Kruglanski, A.W. (2002). Forgetting all else:
On the antecedents and consequences of goal shielding. Journal of
Personality and Social Psychology, 83, 1261–1280. http://dx.doi.
org/10.1037/0022-3514.83.6.1261
Smith, R.C., Coyle, P.T., Baldner, C., Bray, B.C., & Geller, E.S. (2013).
A fi eld investigation of perceived behavioral control and blood al-
cohol content: A pattern-oriented approach. Addictive behaviors,
38(4), 2080–2083.
Thurstone, L.L. (1931). The measurement of social attitudes. The Jour-
nal of Abnormal and Social Psychology, 26, 249–269. http://dx.doi.
org/10.1037/h0070363
Triandis, H.C. (1971). Attitude and attitude change. New York, NY: Wiley.
Vroom, V.H. (1964). Work and motivation. New York, NY: John Wiley
& sons.
Wicker, A.W. (1969). Attitudes versus actions: The relationship of verbal
and overt behavioural responses to attitude objects. Journal of So-
cial Issues, 25, 41–78. http://dx.doi.org/10.1111/j.15404560.1969.
tb00619.x
Wiener, N. (1948). Cybernetics: Or control and communication in the ani-
mal and the machine. Cambridge, MA: MIT Press.
Wyvell, C.L., & Berridge, K.C. (2000). Intra-accumbens amphetamine
increases the conditioned incentive salience of sucrose reward: en-
hancement of reward “wanting” without enhanced “liking” or re-
sponse reinforcement. Journal of Neuroscience, 20(21), 8122–8130.
Zhang, Y., Fishbach, A., & Kruglanski, A.W. (2007). The dilution model:
how additional goals undermine the perceived instrumentality of
a shared path. Journal of Personality and Social Psychology, 92(3),
389–401.
