External Control 389a concern for restoring consistency can produce what can be
described asreverse incentiveeffects (cf. Aronson, 1992;
Wicklund & Brehm, 1976). In a prototypical experimental
arrangement, subjects are induced to perform an action that
they are unlikely to enjoy (e.g., a repetitive or boring task) or
one that conflicts with an attitude they are likely to hold (e.g.,
writing an essay in support of raising tuition at their univer-
sity). At this point, varying amounts of monetary incentive
are offered for the action’s performance; some subjects are
offered a quite reasonable sum (e.g., $20), others are offered
a mere pittance (e.g., $1). Virtually all subjects agree to par-
ticipate regardless of the incentive value, so technically they
all perform a counterattitudinal task (i.e., a task that conflicts
with their attitude concerning the task).
According to Festinger, the dissonance experienced as a
result of such counterattitudinal behavior can be reduced by
changing one of the cognitive elements to make it consistent
with the other element. In this situation, the relevant cognitive
elements for subjects presumably are their feelings about the
action and their awareness they have performed the action.
Because the latter thought cannot be changed (i.e., the damage
is done), the only cognitive element open to revision is their
attitude toward the action (which conveniently had not been
assessed yet). So, the theory holds, subjects faced with this
cognitive dilemma will adjust their attitude toward the action
to make it consistent with the fact that they have engaged in the
action. Subjects who performed a boring task now consider it
interesting or important. Subjects who wrote an essay espous-
ing an unpopular position now indicate they hold that position
themselves. In effect, subjects rationalize their behavior by
indicating that it really reflected their true feelings all along.
At this point, one might assume that all subjects would
follow this scenario. But revising one’s attitude is not the
only potential means of reducing the dissonance brought on
by counterattitudinal behavior. Festinger suggested that a
person can maintain his original attitude if he or she can jus-
tify the counterattitudinal behavior with other salient and rea-
sonable cognitive elements. This is where the large versus
small reward manipulation enters the picture. A subject of-
fered a large incentive (e.g., $20) for performing the act
can use that fact to justify what he or she has done. Who
wouldn’t do something boring or even write an essay one
doesn’t believe if the price were right? The reward, in other
words, obviates the psychological need to change one’s feel-
ings about what one has done. A subject offered a token in-
centive (e.g., $1), on the other hand, cannot plausibly argue
that the reward justified engaging in the boring activity or
writing the disingenuous essay. The only recourse in this sit-
uation is to revise one’s own attitude and indicate liking for
the activity or belief in the essay’s position.
Note the upshot here: The smaller the contingent reward,
the more positive one’s resultant attitude toward the behavior;
or conversely, the larger the contingent reward, the more neg-
ative one’s attitude toward the rewarded behavior. This repre-
sents a rather stunning reversal of the conventional wisdom
regarding the use of rewards to influence people’s behavior.
To be sure, large rewards are useful—often necessary—to get
a person to perform an otherwise undesirable activity or to
express an unpopular attitude. But the effect is likely to be
transitory, lasting only as long as the reward contingency is in
place. To influence the person’s underlying thoughts and feel-
ings regarding the action, and thereby bring about a lasting
change in his or her behavioral orientation, it is best to em-
ploy the minimal amount of reward. In effect, lasting social
influence requires reconstruction within the person rather
than inducements from the outside.
Mental processes are notoriously hard to pin down objec-
tively, of course, and this fact of experimental psychology
has always been a problem for dissonance theory. Festinger
and his colleagues did not attempt to measure what they
assumed to be the salient cognitions at work in the reward
paradigm, nor have subsequent researchers fared much better
in providing definitive evidence regarding the stream of
thought presumably underlying the experience and reduction
of psychological tension. With this gaping empirical hole in
the center of the theory, it is not surprising that other theorists
soon rushed in to fill the gap with their own inferences about
the true mental processes at work. In effect, the results ob-
served in cognitive dissonance research served as something
of a Rorschach for subsequent theorists, each of whom saw
the same picture but imparted somewhat idiosyncratic inter-
pretations of its meaning. Not all interpretations have fared
well, however, and among those that have, there is sufficient
common ground to characterize (in general terms at least) a
viable alternative to the dissonance formulation.
Central to the alternative depiction of reverse incentive ef-
fects is the assumption that people’s minds are first and fore-
most interpretive devices, designed to impose coherence on
the sometimes diverse and often ambiguous elements of per-
sonal experience. In analogy to Gestalt principles of percep-
tion, cognitive processes “go beyond the information given”
(Bruner & Tagiuri, 1954) to impart higher-order meaning that
links the information in a stable and viable structure. With re-
spect to the dissonance paradigm, subjects’ cognitive playing
field is presumably populated with an abundance of salient or
otherwise relevant information. These cognitive elements in-
clude the nature of the task (the activity or essay) and the
money received, of course, but they no doubt encompass an
assortment of other thoughts and feelings as well. Thus, sub-
jects may be sensitized to their sense of personal freedom and