Handbook of Psychology, Volume 5, Personality and Social Psychology

Attitude Functions 305

To overcome this problem, various techniques have been
developed. For example, the bogus pipeline procedure
(Jones & Sigall, 1971) deceives participants into believing
that the researcher can detect their true feelings about an
attitude object, after which participants are asked to report
their attitude toward the attitude object. This technique has
been shown to reduce social desirability in responses to sim-
ple self-report attitude measures (Roese & Jamieson, 1993).
Another approach involves assessing participants’ physio-
logical responses to attitude objects. Unfortunately, many
physiological measures are incapable of distinguishing posi-
tive and negative affective reactions (e.g., skin conductance,
papillary response; Petty & Cacioppo, 1983; Guglielmi,
1999). Positive and negative evaluations can be distin-
guished, however, using facial electromyography (EMG)
recordings (Cacioppo, Petty, Losch, & Kim, 1986), which
detect the relative amount of electrical activity in the muscles
that control smiling and frowning.
Two other psychophysiological techniques show consider-
able promise. One technique detects a specific pattern of elec-
trical activity in the centroparietal region of the brain
(amplitude of the late positive potential: Cacioppo, Crites, &
Gardner, 1996; Gardner, Cacioppo, Crites, & Berntson, 1994),
whereas the other examines the frequency and latency of eye
blinks for attitude objects (Ohira, Winton, & Oyama, 1998).
Future research should test whether these techniques are more
closely linked to one attitude component (e.g., affect) than to
others and whether the techniques yield support for separate
positive and negative dimensions in evaluations.

Implicit Attitudes

Another limitation of most self-report measures of attitudes
is that they assess only explicit attitudes, which are con-
sciously retrievable from memory. As discussed in Petty’s
chapter on attitude change, explicit, conscious attitudes may
differ in numerous ways from implicit, nonconscious atti-
tudes (Greenwald & Banaji, 1995; Wilson, Lindsey, &
Schooler, 2000). Thus, it is useful to measure directly the
nonconscious attitudes.
Several techniques are available to accomplish this
goal. One approach involves extracting self-report, attitude-
relevant information without relying directly on partici-
pants’ conscious determination of their attitude. For example,
researchers can calculate participants’ attitudes from their
responses to open-ended measures, even though these mea-
sures do not directly ask participants to report their at-
titudes. Other measures circumvent respondents’ inferential
processes more strongly by recording behavior that occurs
outside of participants’ conscious control. For example,

researchers can unobtrusively measure participants’ non-

verbal and verbal behaviors toward other people as an indica-

tion of liking (e.g., Fazio, Jackson, Dunton, & Williams,

1995; Word, Zanna, & Cooper, 1974). Because people have

difficulty consciously monitoring such behaviors, their be-

haviors may often reveal attitudes of which the participants

are unaware (see Dovidio, Kawakami, Johnson, Johnson, &

Howard, 1997).

The most common measures of implicit attitudes use elab-

orate priming techniques (e.g., Dovidio et al., 1997; Fazio

et al., 1995; Greenwald, McGhee, & Schwartz, 1998). For

example, Fazio et al.’s (1995) “bona fide pipeline” presents

participants with a target attitude object and asks participants

to classify subsequently presented adjectives as being good or

bad. Theoretically, positive evaluations should be activated in

memory after viewing an attitude object that evokes a positive

attitude. This priming of positive affect should cause partici-

pants to be faster at classifying positive adjectives (e.g., nice,

pleasant) than at classifying negative adjectives (e.g., disgust-

ing, repugnant). In contrast, after viewing an attitude object

that evokes a negative attitude, participants should be slower

at classifying positive adjectives than at classifying negative

adjectives. Indeed, evidence from several studies suggests that

the latency to classify positive versus negative adjectives is

affected by the prior presentation of a liked or disliked attitude

object, particularly when participants hold a strong attitude

toward the attitude object (Fazio, 1993; Fazio, Sanbonmatsu,

Powell, & Kardes, 1986; cf. Bargh, Chaiken, Govender, &

Pratto, 1992). Moreover, attitude scores can be derived from

the speed of responding to the positive versus negative adjec-

tives following the positive versus negative primes, and these

attitude scores predict attitude-relevant behavior toward the

attitude object (Fazio et al., 1995). Greenwald et al.’s (1998)

“implicit association test” similarly relies on facilitating ver-

sus inhibiting effects of evaluation on task performance. An

interesting issue is whether such measures of implicit attitudes

can be adapted to test the models of attitude content and atti-

tude dimensionality.

ATTITUDE FUNCTIONS

Although models of attitude structure are useful for describ-

ing ways in which attitudes may be represented in memory,

these models do not address attitude functions, which are the

psychological motivations that attitudes fulfill (Olson &

Zanna, 1993). Understanding the functions of attitudes

should clarify why people bother to form and maintain atti-

tudes, as well as how underlying motivations influence the

valence and structure of attitudes.