. 17
( 115 .)


Chapter 3 Social Perception: Understanding Other People 75

simple rules of causality. The individual, or perceiver, operates as a kind of “naïve sci-
attribution The process
entist,” applying a set of rudimentary scienti¬c rules (Heider, 1958). Attribution theo-
of assigning causes of
ries are an attempt to discover exactly what those rules are.
behavior, both your own and
Heider made a distinction between internal attribution, assigning causality to
that of others.
something about the person, and external attribution, assigning causality to something
about the situation. He believed that decisions about whether an observed behavior internal attribution
The process of assigning the
has an internal (personal) or external (situational) source emerge from our attempt to
cause of behavior to some
analyze why others act as they do (causal analysis). Internal sources involve things
internal characteristic rather
about the individual”character, personality, motives, dispositions, beliefs, and so
than to outside forces.
on. External sources involve things about the situation”other people, various envi-
external attribution
ronmental stimuli, social pressure, coercion, and so on. Heider (1944, 1958) exam-
The process of assigning
ined questions about the role of internal and external sources as perceived causes of
the cause of behavior to
behavior. His work de¬ned the basic questions that future attribution theorists would some situation or event
confront. Heider (1958) observed that perceivers are less sensitive to situational (exter- outside a person™s control
nal) factors than to the behavior of the individual they are observing or with whom rather than to some internal
they are interacting (the actor). We turn now to the two theories that built directly on
Heiderʼs work.

Correspondent Inference Theory
Assigning causes for behavior also means assigning responsibility. Of course, it is pos-
sible to believe that someone caused something to happen yet not consider the individual
responsible for that action. A 5-year-old who is left in an automobile with the engine
running, gets behind the wheel, and steers the car through the frozen food section of
Joeʼs convenience store caused the event but certainly is not responsible for it, psycho-
logically or legally.
Nevertheless, social perceivers have a strong tendency to assign responsibility
to the individual who has done the deed”the actor. Letʼs say your brakes fail, you
are unable to stop at a red light, and you plow into the side of another car. Are you
responsible for those impersonal brakes failing to stop your car? Well, it depends,
doesnʼt it? Under what circumstances would you be held responsible, and when would
you not?
How do observers make such inferences? What sources of information do people
use when they decide someone is responsible for an action? In 1965, Edward Jones
and Keith Davis proposed what they called correspondent inference theory to explain correspondent inference
An inference that occurs
the processes used in making internal attributions about others, particularly when the
when we conclude that a
observed behavior is ambiguous”that is, when the perceiver is not sure how to inter-
person™s overt behavior is
pret the actorʼs behavior. We make a correspondent inference when we conclude that a
caused by or corresponds
personʼs overt behavior is caused by or corresponds to the personʼs internal character- to the person™s internal
istics or beliefs. We might believe, for example, that a person who is asked by others characteristics or beliefs.
to write an essay in favor of a tax increase really believes that taxes should be raised
(Jones & Harris, 1967). There is a tendency not to take into account the fact that the
essay was determined by someone else, not the essayist. What factors in¬‚uence us to
make correspondent inferences? According to correspondent inference theory, two
major factors lead us to make a correspondent inference:

1. We perceive that the person freely chose the behavior.
2. We perceive that the person intended to do what he or she did.
Social Psychology

Early in the Persian Gulf War of 1991, several U.S.-coalition aircraft were shot
down over Iraq. A few days later, some captured pilots appeared in front of cameras and
denounced the war against Iraq. From the images, we could see that it was likely the
pilots had been beaten. Consequently, it was obvious that they did not freely choose to
say what they did. Under these conditions, we do not make a correspondent inference.
We assume that the behavior tells us little or nothing about the true feelings of the person.
Statements from prisoners or hostages always are regarded with skepticism for this
reason. The perception that someone has been coerced to do or say something makes an
internal attribution less likely. The second factor contributing to an internal attribution is
intent. If we conclude that a personʼs behavior was intentional rather than accidental, we
are likely to make an internal attribution for that behavior. To say that a person intended
to do something suggests that the individual wanted the behavior in question to occur.
To say that someone did not intend an action, or did not realize what the consequences
would be, is to suggest that the actor is less responsible for the outcome.

Covariation Theory
Whereas correspondent inference theory focuses on the process of making internal attri-
butions, covariation theory, proposed by Harold Kelley (1967, 1971), looks at external
attributions”how we make sense of a situation, the factors beyond the person that may be
causing the behavior in question (Jones, 1990). The attribution possibilities that covaria-
tion theory lays out are similar to those that correspondent inference theory proposes.
What is referred to as an internal attribution in correspondent inference theory is referred
to as a person attribution in covariation theory. What is called an external attribution in
correspondent inference theory is called a situational attribution in covariation theory.
Like Heider, Kelley (1967, 1971) viewed the attribution process as an attempt to
apply some rudimentary scienti¬c principles to causal analysis. In correspondent infer-
ence theory, in contrast, the perceiver is seen as a moral or legal judge of the actor.
Perceivers look at intent and choice, the same factors that judges and jurors look at when
assigning responsibility. Kelleyʼs perceiver is more a scientist: just the facts, maʼam.
covariation principle The According to Kelley, the basic rule applied to causal analysis is the covariation
rule that if a response is principle, which states that if a response is present when a situation (person, object,
present when a situation
event) is present and absent when that same situation is absent, then that situation is the
(person, object, or event) is
cause of the response (Kelley, 1971). In other words, people decide that the most likely
present and absent when
cause of any behavior is the factor that covaries”occurs at the same time”most often
that same situation is absent,
with the appearance of that behavior.
the situation is presumed to
be the cause of the response. As an example, letʼs say your friend Keisha saw the hit movie Crash and raved about
it. You are trying to decide whether you would like it too and whether you should go see
it. The questions you have to answer are, What is the cause of Keishaʼs reaction? Why did
she like this movie? Is it something about the movie? Or is it something about Keisha?
In order to make an attribution in this case, you need information, and there are
three sources or kinds of relevant information available to us:

1. Consensus information
2. Distinctiveness information
3. Consistency information
Consensus information tells us about how other people reacted to the same event or
situation. You might ask, How did my other friends like Crash? How are the reviews?
How did other people in general react to this stimulus or situation? If you ¬nd high
Chapter 3 Social Perception: Understanding Other People 77

consensus”everybody liked it”well, then, it is probably a good movie. In causal attri-
bution terms, it is the movie that caused Keishaʼs behavior. High consensus leads to a
situational attribution.
Now, what if Keisha liked the movie but nobody else did? Then it must be Keisha
and not the movie: Keisha always has strange tastes in movies. Low consensus leads
to a person attribution (nobody but Keisha liked it, so it must be Keisha).
The second source or kind of data we use to make attributions is distinctiveness
information. Whereas consensus information deals with what other people think, dis-
tinctiveness information concerns the situation in which the behavior occurred: We ask
if there is something unique or distinctive about the situation that could have caused
the behavior. If the behavior occurs when there is nothing distinctive or unusual about
the situation (low distinctiveness), then we make a person attribution: If Keisha likes
all movies, then we have low distinctiveness: Thereʼs nothing special about Crash”it
must be Keisha. If there is something distinctive about the situation, then we make a
situational attribution. If this is the only movie Keisha has ever liked, we have high dis-
tinctiveness and there must be something special about the movie. Low distinctiveness
leads us to a person attribution; high distinctiveness leads us to a situational attribution.
If the situation is unique”very high distinctiveness”then the behavior probably was
caused by the situation and not by something about the person. The combination of high
consensus and high distinctiveness always leads to a situational attribution. The combi-
nation of low consensus and low distinctiveness always leads to a person attribution.
The third source or kind of input is consistency information, which con¬rms whether
the action occurs over time and situations (Chen, Yates, & McGinnies, 1988). We ask,
Is this a one-time behavior (low consistency), or is it repeated over time (high consis-
tency)? In other words, is this behavior stable or unstable? Consistency is a factor that
correspondent inference theory fails to take into account.
What do we learn from knowing how people act over time? If, for example, the next
time we see Keisha, she again raves about Crash, we would have evidence of consis-
tency over time (Jones, 1990). We would have less con¬dence in her original evaluation
of the movie if she told us she now thought the movie wasnʼt very good (low consis-
tency). We might think that perhaps Keisha was just in a good mood that night and that
her mood affected her evaluation of the movie. Consistency has to do with whether the
behavior is a reliable indicator of its cause.
The three sources of information used in making attributions are shown in
Figures 3.3 and 3.4. Figure 3.3 shows the combination of information”high consensus,
high consistency, and high distinctiveness”that leads us to make a situational attribu-
tion. Go see the movie: Everybody likes it (high consensus); Keisha, who likes few, if
any, movies, likes it as well (high distinctiveness of this movie); and Keisha has always
liked it (high consistency of behavior).
Figure 3.4 shows the combination of information”low consensus, high consistency,
and low distinctiveness”that leads us to a person attribution. None of our friends likes
the movie (low consensus); Keisha likes the movie, but she likes all movies, even The
Thing That Ate Newark (low distinctiveness); and Keisha has always liked this movie cognitive miser The idea
(high consistency). Maybe we ought to watch TV tonight. suggesting that because
Not surprisingly, research on covariation theory shows that people prefer to make humans have a limited
capacity to understand
personal rather than situational attributions (McArthur, 1972). This conforms with the
information, we deal only
(correspondence) bias we found in correspondence inference theory and highlights
with small amounts of social
again the tendency toward overemphasizing the person in causal analysis. It also ¬ts
information and prefer
with our tendency to be cognitive misers and take the easy route to making causal the least effortful means of
attributions. processing it.
Social Psychology

Figure 3.3 Information
mix leading to a situational

Dual-Process Models
We have emphasized that people are cognitive misers, using the least effortful strategy
available. But they are not cognitive fools. We know that although impression forma-
tion is mainly automatic, sometimes it is not. People tend to make attributions in an
automatic way, but there are times when they need to make careful and reasoned attri-
butions (Chaiken & Trope, 1999).
Trope (1986) proposed a theory of attribution that speci¬cally considers when
people make effortful and reasoned analyses of the causes of behavior. Trope assumed,
as have other theorists, that the ¬rst step in our attributional appraisal is an automatic
categorization of the observed behavior, followed by more careful and deliberate infer-
ences about the person (Trope, Cohen, & Al¬eri, 1991).
The ¬rst step, in which the behavior is identi¬ed, often happens quickly, automati-
cally, and with little thought. The attribution made at this ¬rst step, however, may be
adjusted in the second step. During this second step, you may check the situation to
see if the target was controlled by something external to him. If “something made him
do it,” then you might hold him less (internally) responsible for the behavior. In such
instances, an inferential adjustment is made (Trope et al., 1991).
What information does the perceiver use to make these attributions? Trope plau-
sibly argued that perceivers look at the behavior, the situation in which the behavior
occurs, and prior information about the actor. Our knowledge about situations helps
us understand behavior even when we know nothing about the person. When someone
cries at a wedding, we make a different inference about the cause of that behavior than
we would if the person cried at a wake. Our prior knowledge about the person may lead
us to adjust our initial impression of the personʼs behavior.
A somewhat different model was developed by Gilbert (1989, 1991) and his col-
leagues. In¬‚uenced by Tropeʼs two-step model, they proposed a model with three distinct
stages. The ¬rst stage is the familiar automatic categorization of the behavior (that action
Chapter 3 Social Perception: Understanding Other People 79

Figure 3.4 Information
mix leading to a person

was aggressive); the second is characterization of the behavior (George is an aggressive
guy); and the third, correction, consists of adjusting that attribution based on situational
factors (George was provoked needlessly). Gilbert essentially divided Tropeʼs ¬rst step,
the identi¬cation process, into two parts: categorization and characterization. The third
step is the same as Tropeʼs inferential-adjustment second step.
For example, if you say “Good to see you” to your boss, the statement may be cat-
egorized as friendly, and the speaker may be characterized as someone who likes the other
person; ¬nally, this last inference may be corrected because the statement is directed at
someone with power over the speaker (Gilbert, McNulty, Guiliano, & Benson, 1992).
The correction is based on the inference that you had better be friendly to your boss.
Gilbert suggests that categorization is an automatic process; characterization is not quite
automatic but is relatively effortless, requiring little attention; but correction is a more
cognitively demanding (controlled and effortful) process (Gilbert & Krull, 1988). Of
course, we need to have the cognitive resources available to make these corrections. If
we become overloaded or distracted, then we are not able to make these effortful cor-
rections, and our default response is to make internal and dispositional attributions and
to disregard situational information (Gilbert & Hixon, 1991; Trope & Al¬eri, 1997).

Intentionality and Attributions
Malle (2006) has ¬lled some gaps in our understanding of how individuals make attri-
butions by considering the relationship between intentionality (did the individual intend
to do what she actually did?) and judgments about the causes of a behavior. Judging
intent has many implications for our sense of what de¬nes blame and morality. The
offender who cries, “I didnʼt know the gun was loaded,” however falsely, is making a
claim on our understanding of intentionality and blame. If I thought the gun was not
loaded, I could not have meant to kill the victim, and hence, I am blameless, or should
be held blameless legally, if not morally.
Social Psychology

Malle asked, What constitutes ordinary folksʼ notions of what is an “intentional”
action? The responses to Malleʼs question revealed four factors: desire, belief, inten-
tion, and awareness. Desire refers to a hope for a particular outcome; belief was de¬ned
as thoughts about what would happen before the act actually took place; intention
meant that the action was meant to occur; and awareness was de¬ned as “awareness
of the act while the person was performing it” (Malle, 2006, p. 6). Further research,
however, showed that there was a ¬fth component of ordinary notions of intention-
ality. We judge whether the person actually has the skill or ability to do what was
desired. Thus, if I am a lousy tennis player, which I am, and I serve several aces in a
row, it is clear that while I desired to do so, observers, knowing my skill level, will be
unlikely to conclude that I intended to serve so well. Note here: There is a difference


. 17
( 115 .)