Psychological Bulletin

 

© 1995 by the American Psychological Association 

January 1995 Vol. 117, No. 1, 21-38 

 

For personal use only--not for distribution. 

The Correspondence Bias  

Daniel T. Gilbert

 

Department of Psychology University of Texas at Austin  

Patrick S. Malone

 

Department of Psychology University of Texas at Austin  

ABSTRACT

 

 
The correspondence bias is the tendency to draw inferences about a 
person's unique and enduring dispositions from behaviors that can be 
entirely explained by the situations in which they occur. Although this 
tendency is one of the most fundamental phenomena in social 
psycholoxgy, its causes and consequences remain poorly understood. This 
article sketches an intellectual history of the correspondence bias as an 
evolving problem in social psychology, describes 4 mechanisms (lack of 
awareness, unrealistic expectations, inflated categorizations, and 
incomplete corrections) that produce distinct forms of correspondence 
bias, and discusses how the consequences of correspondence-biased 
inferences may perpetuate such inferences.  

 

 
Portions of this article were written while Daniel T. Gilbert was a fellow at the Center for 
Advanced Study in the Behavioral Sciences. That fellowship was made possible by the 
John D. and Catherine T. MacArthur Foundation and by a Research Scientist 
Development Award from the National Institute of Mental Health (1-KO2-MH00939). 
The generous support of these institutions is gratefully acknowledged.  
We thank Josh Aronson, Brian Giesler, Tory Higgins, Lee Ross, Bill Swann, Yaacov 
Trope, and several anonymous reviewers for their helpful comments on an earlier version 
of this article. The article benefited immeasurably from the comments of the late Ned 
Jones.  
Correspondence may be addressed to Daniel T. Gilbert, Department of Psychology, 
University of Texas Austin, Mezes Hall 330, Austin, Texas, 78712.  
Electronic mail may be sent to 

gilbert@psyvax.psy.utexas.edu

 

 

Received: March 2, 1993  
Revised: May 24, 1994  
Accepted: May 30, 1994 

 

 

One will seldom go wrong if one attributes extreme actions to vanity, 
average ones to habit, and petty ones to fear. ( 

Friedrich Nietzsche, 

1886/1984, p. 59 

)  

Despite the homilies of philosophers, no one has yet found a simple formula for 
understanding others. The problem, of course, is that a person's inner self is hidden from 
view. Character, motive, belief, desire, and intention play leading roles in people's 
construal of others, and yet none of these constructs can actually be observed. As such, 
people are forced into the difficult business of inferring these intangibles from that which 
is, in fact, observable: other people's words and deeds. When one infers the invisible from 
the visible, one risks making a mistake. Three decades of research in social psychology 
have shown that many of the mistakes people make are of a kind: When people observe 
behavior, they often conclude that the person who performed the behavior was 
predisposed to do so–that the person's behavior corresponds to the person's unique 
dispositions–and they draw such conclusions even when a logical analysis suggests they 
should not.  

In this article, we describe the causes and consequences of this particular mistake, which 
we call 

the correspondence bias. 

We do not attempt a complete review of the voluminous 

literature on this topic. Rather, we first define the correspondence bias in terms of the 
person—situation distinction that is fundamental to attribution theory. Second, we offer a 
brief and selective history of the study of the correspondence bias. Third, we describe the 
sequence of events that unfolds when attributions are made and then use this description 
to taxonomize and explicate the mechanisms that cause correspondence bias. Finally, we 
describe some of the consequences that may explain why this bias persists.  

Attribution Theory's Rational Canon  

People care less about what others do than about why they do it. Two equally 
rambunctious nephews may break two equally expensive crystal vases at Aunt Sofia's 
house, but the one who did so by accident gets the reprima nd and the one who did so by 
design gets the thumbscrews. Aunts are in the business of understanding what makes 
nephews act as they do, and social psychologists are in the business of explaining how 
aunts achieve those understandings. The theories that provide these explanations are 
known as attribution theories.  

There is no shortage of attribution theories (e.g., 

Ajzen & Fishbein, 1975 

; 

Bem, 1972 

; 

Hilton & Slugoski, 1986 

; 

Jones & Davis, 1965 

; 

Kelley, 1967 

; 

Medcof, 1990 

; 

Reeder & 

Brewer, 1979 

; 

Trope, 1986 

; 

Weiner et al., 1972 

). Although these theories differ in both 

focus and detail, each is grounded in a common metaphor that construes the human skin 
as a special boundary that separates one set of "causal forces" from another. On the sunny 
side of the epidermis are the external or situational forces that press inward on the person, 
and on the meaty side are the internal or personal forces that exert pressure outward. 
Sometimes these forces press in conjunction, sometimes in opposition, and their dynamic 
interplay manifests itself as observable behavior. As such, aunts can determine the causes 
of behavior in much the same way that they determine the causes of physical movement: 
By observing the motion of an object ("The balloon rose rapidly in the morning sky") and 
then subtracting out the contribution of external forces ("A light wind nudged the balloon 
ever upward"), an observer can estimate the magnitude and direction of the internal 
forces ("The balloon must have contained helium that contributed to the speed of its 

ascent"). According to attribution theories, aunts think of nephews as they think of 
balloons: objects whose behavioral motions are partially determined by the prevailing 
winds and partially determined by the rare and noble gasses with which genes and 
experience have inflated them.  

Attribution theories suggest that the psychological world is a mirror of the physical world 
and that the two are therefore penetrated by the same logic. Ordinary people seem to 
believe that others behave as they do because of the kinds of others they are and because 
of the kinds of situations in which their behaviors unfold; thus, when a person makes an 
attribution about another, she or he attempts to determine which of these factors–the other 
person or the other person's situation–played the more significant role in shaping the 
other person's behavior. Is the basketball player a graceless shooter, or did poor lighting 
cause him to miss the free throw? Did the senator speak in favor of abortion rights 
because she truly believes in freedom of choice, or was she merely pandering to the 
desires of her liberal audience? Did the student appear sad because he is chronically 
depressed, or had he just received word of a failing grade? Each of these is a question 
about the relative contributions to behavior of situational and dispositional factors, and 
this distinction is, perhaps, the defining feature of attribution theory.  

Attribution theory's fundamental distinction leads quite naturally to its fundamental rule: 
When a behavior occurs in the presence of a sufficiently strong, facilitative force, an 
observer should not infer that the actor is predisposed to perform that behavior. Just as 
one should not conclude that a balloon that rises on a windy day is filled with helium, one 
cannot make unequivocal inferences about the abilities of an athlete, the convictions of a 
politician, or the mental health of a student when poor lighting, a roomful of opinionated 
voters, or sudden bad news may have induced their behaviors. In other words, one should 
not explain with dispositions that which has already been explained by the situation. This 
logical rule was first formalized by 

Jones and Davis (1965) 

as the 

law of noncommon 

effects 

and later extended and codified by 

Kelley (1967) 

as the 

discounting principle, 

which warns observers not to attribute an effect to any one causal agent (e.g., a 
disposition) when another plausible causal agent (e.g., a situational force) is 
simultaneously present. In other words, when people do precisely what the physical 
environment or the social situation demands, dispositional inferences are logically 
unwarranted.  

This simple rule is eminently reasonable, but, as with the interstate speed limit, someone 
seems to have neglected to tell the drivers. Although ordinary people may acknowledge 
the logical validity of the discounting principle when it is stated in the abstract, they are 
sometimes willing to abandon it in practice. People may make inferences about the 
dispositions of others even when situational forces explain the behavior quite nicely. In 
scores of experiments, subjects have violated attribution theory's logical canon by 
concluding that an actor was predisposed to certain behaviors when, in fact, those 
behaviors were demanded by the situations in which they occurred. Basketball players 
who are randomly assigned to shoot free throws in badly lighted gyms may, on average, 
be judged as less capable than players who are randomly assigned to shoot free throws on 
a well-lighted court (e.g., 

Ross, Amabile, & Steinmetz, 1977 

). Politicians who are 

randomly assigned to read prochoice speeches may, on average, be judged as more 
prochoice than politicians who are randomly assigned to read prolife speeches (e.g., 

Jones & Harris, 1967 

). Students who are randomly assigned to receive bad news may, on 

average, be judged as more chronically depressed than students who are randomly 
assigned to receive good news (e.g., 

Gilbert, Pelham, & Krull, 1988 

). And so on. 

Although this logical error has been called "as robust and reliable a phenomenon as any 
in the literature on person perception" ( 

Quattrone, 1982a, p. 376 

), after nearly 30 years 

of research there is still no single, widely accepted explanation for its occurrence.  

One might wonder how such a perdurable puzzle could have failed to yield a solution. 
We argue here that, in fact, the correspondence bias is a puzzle that has yielded too many 
solutions and that theoretical progress on this problem has been impeded by a failure to 
recognize that the correspondence bias comprises a number of distinct phenomena that 
only pose as one. Indeed, one reason why theorists disagree about 

the 

cause of the 

correspondence bias is that they are often studying different phenomena. Our goals in the 
remainder of this article are to place the correspondence bias in its historical perspective, 
sketch a generic model of attributional processes that describes four distinct causes of 
correspondence bias, and, finally, explore the inferential and interpersonal consequences 
of the bias. 

1 

 

Brief History of the Correspondence Bias  

People tend to think that others are as they act, and the intellectual roots of this tendency 
are so deep in Western  thought that any attempt to describe them without discussing 
Durkheim, Weber, Marx, and Freud cannot help but fall short. Fortunately, we have set a 
less ambitious task for ourselves in this section, namely, to describe the genesis and 
development of this idea within that small pocket of Western thought known as 
experimental social psychology. The correspondence bias has been a problem in social 
psychology since its inception and has been described by some as the central problem of 
the field. How has the study of this phenomenon progressed? We see four events as 
intellectual watersheds, and these are the publication of landmark essays by 

Lewin (1931) 

, 

Ichheiser (1949) 

, 

Jones and Harris (1967) 

, and 

Ross (1977) 

.  

Aristotelian Thinking in the 1930s 

 

In 1931, Kurt Lewin published an unusual essay in which he analyzed the philosophical 
revolution that Galileo had brought to 17th-century physics. Lewin was not interested in 
the history of physics per se; rather, he believed that the transition in physics from an 
Aristotelian to a Galilean view was a transition common to the evolution of all scientific 
thinking. In Aristotle's physics, the behavior of objects was ascribed to the individuating 
properties of those objects: Heavy things, for example, had gravity, whereas light things 
had levity, and these properties "explained" why the heavy and light things fell and rose. 
Indeed, this mode of explanation dominated physical science for centuries. According to 
Lewin, Galileo's insight was that the behavior of objects can be understood only in 
reference to the situation in which that behavior occurs. As 

Lewin (1931) 

noted:  

For Aristotelian concepts ... the vectors which determine an object's 
movements are completely determined by the object. ... The tendency of 
light bodies to go up resided in the bodies themselves; the downward 
tendency of heavy objects was seated in those objects. In modern physics, 
on the contrary, not only is the upward tendency of a lighter body derived 
from the relation of this body to its environment, but the weight itself of a 
body depends on such a relation. ... The properties and structure of the 
object involved remain important also for the Galilean theory of dynamics. 
But the situation assumes as much importance as the object. (p. 29)  

Lewin (1931) 

argued that psychology was stuck in an Aristotelian mode, and he 

challenged psychologists to become Galileans. Until psychology stopped thinking of 
behavior as the expression of dispositional properties of the person and began to think of 
it as an interaction between the person and the environment, Lewin argued, it would be 
doomed to remain in its already-prolonged infancy. In this essay, as in all his work, 
Lewin did more perhaps than any other social psychologist to show how an 
understanding of situations was critical to an understanding of human behavior. His, then, 
was the first important contribution to the study of correspondence bias, because only 
when social psychology had itself recognized the significance of situational forces would 
it be prepared to ask whether ordinary people recognized the same.  

Social Blindness in the 1940s and 1950s 

 

By the end of World War II, the sentiment that 

Lewin (1931) 

expressed had emerged 

independently among intellectuals of all stripe. Literary works, political essays, and 
psychological reports played on the common theme that human affairs are governed more 
often by accident than by intention and that people are more often the prisoners of their 
times than the captains of their destinies. This realization was not, of course, new, nor 
was it any one person's doing. But to the extent that one person served as its focus, that 
person was probably not Kurt Lewin but Adolph Hitler. Although some psychologists 
offered Aristotelian explanations for the Nazi phenomenon (e.g., the authoritarian 
personality), others took a different lesson from the Third Reich, which demonstrated that 
social situations can be fantastically powerful determinants of action. Choreographed 
legions of citizen-soldiers goose stepping to the chorus of "Sieg heil!" provided a more 
powerful testament to the malleability of human behavior than Lewin's essays or 
Skinner's pigeons ever could. In addition, the horrors of the Nazi eugenics program 
created something of an intellectual backlash among American scholars, who became 
wary of explanations that appealed to the "nature of man." Thus, as 

Keller (1992, p. 261) 

noted, "it was perhaps inevitable that, in the aftermath of the war, it would be to nurture 
that the development of human behaviour would be attributed."  

What made this insight interesting was the fact that it ran directly counter to the 
individualist tradition of Western culture (a tradition initiated by the Greeks, especially 
Aristotle, but not much in vogue before 800 BC). Americans of the industrial age were 
weaned on a pabulum that was one part Ayn Rand and one part Horatio Alger: Anyone 
could be rich, anyone could be famous, anyone could be president. In a land of boundless 

opportunity, the only constraints on one's achievements were one's own talents and 
persistence. Thus, although the culture taught that people were the screenwriters, 
directors, and stars of their own lives, careful observation seemed to teach otherwise. 
Many careful observers remarked on the incongruity between the standard doctrine and 
the hard reality, but none provided a more thoughtful, clearheaded, and detailed 
psychological analysis than did Gustav Ichheiser. Almost a decade before 

Heider (1958) 

planted the seeds of attribution theory, 

Ichheiser (1949) 

discussed the problem of 

unwarranted dispositional inference in plain and eloquent terms:  

We all have in everyday life the tendency to interpret and evaluate the 
behavior of other people in terms of specific personality characteristics 
rather than in terms of specific social situations in which those people are 
placed. ... It is hardly possible to exaggerate the importance of this type of 
social blindness in the crisis of our age. ... Many things which happened 
between the two world wars would not have happened if social blindness 
had not prevented the privileged from understanding the predicament of 
those who were living in an invisible jail. (p. 47)  

Ichheiser (1949) 

wrote about this particular form of social blindness in all its various 

guises, described several concrete psychological mechanisms that could cause it, and 
even prescribed remedies. No one was listening. Even when 

Heider (1958) 

made the 

same point 9 years later, it was dwarfed by the magnitude of his book's other 
contributions. In fact, no one seemed particularly interested in studying the phenomenon 
whose importance was "hardly possible to exaggerate" until 1967. In that year, Ned Jones 
and Victor Harris performed an experiment that, fortunately enough, did not turn out as 
they had planned.  

Observer Bias in the 1960s 

 

No decade of American history has been more overcharacterized than the 1960s. But 
surely this was a decade of change, a decade in which television brought war and 
assassination into the living rooms of ordinary people, who began to wonder whether 
they could truly shape their own destinies in the face of the powerful social forces that 
were apparently sweeping the nation. In a world that seemed to be spinning out of 
control, it was difficult to remain a faithful subscriber to the traditional verities of self-
determination and hard work. In considering other mass phenomena, such as the Great 
Society (which sought to undo the "mere accidents of situation" that distinguished the 
poor from the middle class) and the Black Power and feminist  movements (which argued 
that inhibitory situational forces could hobble entire classes of capable people), one can 
see how the situationist insights of Lewin and Ichheiser continued to penetrate the 
ordinary American experience.  

It was in these times that Ned 

Jones and Keith Davis (1965) 

published the first systematic 

model of dispositional inference. Two years later, Jones and Harris set out to test one of 
the theory's less subtle predictions: When an actor is unconstrained by the social 
situation, observers will infer dispositions from the actor's behavior; when an actor is 

entirely constrained, however, observers will make no such inference. In a now-classic 
experiment, subjects were shown essays that supported or opposed Cuba's president, 
Fidel Castro, and were told either that the essayist was free to determine which side of 
that issue he would espouse or that the essayist had been instructed by his debate coach to 
defend a particular point of view. As expected, observers inferred strong pro- and anti-
Castro attitudes when the essayist had freely chosen to defend those respective positions. 
But contrary to expectation, observers made similar (albeit much weaker) inferences 
when the essayist had been ordered by a debate coach to defend his stated position ( 

Jones & Harris, 1967 

).  

The result was too puzzling to leave alone. Here were perfectly intelligent college 
students who, when exposed to the coerced political statement of another student, seemed 
to be saying, "Well, yes, I know he was merely completing the assignment given him by 
his debate coach, but to some degree I think he personally believes what he wrote." This 
observer bias, as 

Jones and Harris (1967) 

called it, was replicated under a variety of 

circumstances that ruled out some of the more obvious artifactual explanations (e.g., that 
subjects had misunderstood the instructions, that the essays were unrealistic; see 

Jones, 

1979 

, 

1990 

; but see also 

Fein, Hilton, & Miller, 1990 

; 

A. G. Miller, Ashton, & Mishal, 

1990 

). Over the next 10 years, Jones and Harris's "attitude attribution paradigm" fueled 

an active cottage industry that produced dozens of careful replications and extensions. 
Despite the considerable research activity and its cumulative results, two things failed to 
happen. First, no one offered a convincing psychological explanation of the observer 
bias, which proved both robust and enigmatic–something of a stray puppy that no one 
could quite get rid of but whose owner no one could seem to track down. Jones and 
Harris popularized 

Heider's (1958, p. 54) 

maxim that behavior "tends to engulf the total 

field" but correctly noted that "this describes the results without really explaining them" ( 

Jones & Harris, 1967, p. 22 

). Second, experiments concentrated on the attribution of 

attitudes and remained somewhat paradigm bound. As a result, the observer bias piqued 
the interest of only a few dozen social psychologists who did research on attribution and 
social perception, and much of that research activity centered on local aspects of the 
attitude attribution paradigm itself rather than on the general phenomenon of observer 
bias.  

The first of these problems has defied simple remedy: There is not today a single, 
commonly accepted explanation of the correspondence bias. We argue later that this is 
because the bias is actually a constellation of separate phenomena that require separate 
explanations. But even if the dog's owner could not be located, social psychologists 
would soon realize that what looked like a wayward puppy was, in fact, a champion 
canine.  

The Fundamental Attribution Error in the 1970s and 1980s 

 

Social psychology is, in the broadest sense, the science of environmental influence, and a 
considerable number of its experiments seek to demonstrate what Lewin, Ichheiser, 
Heider, and Hitler all knew: A person's behavior can be predicted, in large part, from 
knowledge of the social circumstances in which it occurs. In 1977, Lee Ross offered 

social psychologists a pair of insights. First, he argued that without the observer bias, the 
business of social psychology would be a dreadful bore. If social psychologists intuitively 
recognized the true strength of situational influences, then their demonstrations would be 
mere platitudes. Indeed, what made the experiments of Festinger and Aronson and 
Schachter and Milgram so interesting was that consumers of the research could be relied 
on to underestimate the strength of the social situations that the experimenters had 
engineered and, therefore, to be surprised by the experimental results.  

Ross's (1977) 

second insight was much more important. He realized that the social 

psychologist's tendency to underestimate the power of situations (which he called the 
fundamental attribution error) was shared by social psychology's subjects and that this 
was the key to understanding their behavior in a wide range of seemingly unrelated 
experiments. 

Jones and Harris's (1967) 

subjects, for example, had failed to realize how 

motivating a debate coach could be when he ordered a debater to defend an unpopular 
position. Similarly, subjects in 

Festinger and Carlsmith's (1959) 

classic dissonance study 

had failed to realize how much pressure an experimenter could exert by politely asking 
them to tell a little white lie. 

Bierbrauer's (1979) 

subjects failed to realize how 

intimidating Milgram's experimenter could be when he donned a white lab coat and 
comma;nded one person to electrocute another. And so on. In each of these cases, 
subjects had mistaken a strong situation for a relatively weak one. They had mistaken 
highly constrained actors for lightly constrained actors and, as such, made the kinds of 
inferences about the former that one usually reserves for the latter. Ross was able to use a 
single principle to explain why subjects acted as they did and why social psychologists 
found it so interesting. Subjects and psychologists, he argued, were not the sorts of 
creatures they thought themselves to be: The determinants of their behavior were at odds 
with their theories about the determinants of their behavior, and thus they were capable of 
surprising themselves.  

Ross's (1977) 

thesis had many lasting effects. But most important among these was that it 

showed that the tendency to make unwarranted dispositional inferences was not just some 
backwater curiosity but, rather, that it constituted the very heart of the social 
psychological enterprise. In so doing, Ross unbound the phenomenon and demonstrated 
the richness of its implications.  

Causes of Correspondence Bias  

The correspondence bias has been evolving as an intellectual problem in social 
psychology for some six decades. As we have already noted, we consider 

the 

correspondence bias to be something of a misnomer inasmuch as several different 
psychological mechanisms can give rise to the same general effect (i.e., the inference of 
dispositions from situationally induced behaviors). Although these mechanisms are often 
confused and rarely distinguished in the literature, we suggest that there are, in fact, four 
distinct causes of correspondence bias: (a) lack of awareness, (b) unrealistic expectations, 
(c) inflated categorizations, and (d) incomplete corrections.  

We begin our discussion of these causes by sketching the sequence of events that occurs 
when an attribution is made (see 

Figure 1 

). If an observer is to have any hope of 

performing a "proper" attributional analysis that takes into account the role of situational 
forces, she or he must first recognize the situation in which the actor is functioning ("The 
terrorist is threatening the hostage"). Of course, observers bring to this recognition a 
general set of beliefs about how people typically behave in such situations ("Most people 
will say anything to avoid being murdered"), and these beliefs constitute expectations 
(although not necessarily conscious expectations) for the behavior of the particular actor 
("I expect the hostage to make anti-American remarks"). Next, the observer must 
perceive and categorize the particular actor's behavior ("The hostage is making anti-
American remarks"). Finally, the observer must determine whether the actor's behavior 
violates the expectations that the observer's knowledge of the situation has engendered. If 
so ("Those anti-American remarks are stronger than I expected them to be"), the observer 
will draw a dispositional inference about the actor ("I think the hostage is somewhat 
sympathetic to the terrorists' cause"). If not, the observer will refrain from drawing such 
inferences ("The hostage is only doing what anyone would do in such a situation and thus 
is not necessarily sympathetic to the terrorists' cause"). In short, only when people 
observe behavior that is more extreme than the situation leads them to expect do they 
make dispositional inferences about the actor. Although attribution theories do differ in 
their essential details and may suggest slightly different sequences and combinations of 
these steps, most theorists would probably agree that these represent the major events in 
an attributional analysis. We suggest that errors at any one of these four stages can 
produce the correspondence bias.  

Lack of Awareness of Situational Constraints 

 

To avoid the correspondence bias, an observer must realize that a situation is playing a 
causal role in an actor's behavior. But one can implicate situational forces as causes only 
when one is aware that such forces exist in the first place. If one does not know that a 
hostage is being threatened, a senator cajoled, or a basketball player hindered, then one 
cannot even begin to do the inferential work that accurate attribution requires. Two 
problems–the invisibility problem and the construal problem–may make it particularly 
difficult for observers to attain the basic information they need to complete their 
attributional tasks.  

The invisibility problem. 

 

Actors can be weighed and behaviors can be filmed, but when one tries to point to a 
situation, one often stabs empty air. Indeed, the constructs to which the word 

situation 

refers often have little or no physical manifestation: One cannot see, smell, taste, or hear 
"audience pressure," which exists only in the mind of the public speaker. When 

Skinner 

(1971) 

tried to explain why ordinary people attribute behaviors to the internal traits of 

actors rather than to the environmental stimuli that he considered the true causes of those 
behaviors, he implicated situational invisibility as the primary culprit:  

We recognize a person's dignity and worth when we give him credit for 
what he has done. The amount we give is inversely proportional to the 
conspicuousness of the [situational] causes of his behavior. If we do not 
know why a person acts as he does, we attribute his behavior to him. (p. 
55)  

Indeed, subjects in experiments must be specifically informed that the reader of an anti-
American speech was being coerced by terrorists, an anti-American audience, or a debate 
coach because there is nothing in the behavior itself that relays this information (cf. 

Baron, 1988 

; 

McArthur & Baron, 1983 

). If the observer cannot see the actor's situation 

(i.e., the gun to the head, the cheering crowd, or the coach's instructions), then the 
observer may not know about the actor's situation and thus will surely fail to take that 
situation into account when making an attribution. This is precisely what happened in a 
well-known experiment conducted by 

Ross, Amabile, and Steinmetz (1977) 

in which 

subjects were arbitrarily assigned to play the role of contestant or quizmaster in a mock 
game show. Quizmasters were allowed to generate a list of questions from their private 
store of arcane knowledge, and, as expected, contestants typically failed to answer those 
questions. Surely contestants were faced with a much more difficult task than were 
quizmasters, and surely task difficulty was a powerful determinant of their performances. 
Nonetheless, observers of the game show concluded that the quizmasters were genuinely 
brighter than the contestants. Because observers could not actually see the "invisible jail" 
in which contestants were imprisoned, their impoverished understanding of the situation 
led them to have inappropriate expectations for the contestants' behavior–expectations 
that could not help but be dashed by reality.  

We suspect that such awareness is often difficult to achieve in everyday life because 
many situational forces are temporally or spatially removed from the behavioral episodes 
they constrain. Social norms and parental threats are potent forces that physically exist 
only in the brains of the people whose behaviors they are constraining, and nothing in the 
behavioral episode itself may bring these forces to the observer's attention. Even when 
situational constraints are physically present in the behavioral episode, they may often 
escape notice because the cues that evoke behavior are often both subtle and powerful. 
For example, the power of a smile to induce a smile is just short of reflexive ( 

Hinsz & 

Tomhave, 1991 

), as is the power of a gaze to direct a gaze ( 

Milgram, Bickman, & 

Berkowitz, 1969 

; 

Walden & Ogan, 1988 

). In short, it can be difficult to attain awareness 

of the forces that are compelling an actor's behavior, and when observers lack such 
awareness they are predictably prone to correspondence bias.  

The construal problem. 

 

Attributionists often talk about situational forces or environmental constraints as though 
these terms described a clearly bounded class of virtually interchangeable phenomena. In 
fact, there are at least two very different kinds of situational constraints that pose very 
different attributional problems for the ordinary observer of behavior. 

Behavioral 

constraints 

alter an actor's behavioral options by altering the actor's capacity to enact 

those options or by altering the capacity of the environment to sustain them. Such 

constraints are entirely independent of the actor's understanding of them. For example, 
the contestant-subjects in 

Ross, Amabile, and Steinmetz's (1977) 

experiment had no 

choice but to provide incorrect answers on many trials. Regardless of what they may have 
felt, wanted, thought, or believed, the objective difficulty of the quizmasters' esoteric 
questions necessitated that contestants would perform poorly. In a sense, the range of 
behavioral options available to the contestants was narrowed by a vagary of the situation 
that was entirely independent of their understanding of the situa tion. Task difficulty (a 
constraint commonly used in attribution experiments) always affects performances by 
directly constraining the actor's behavior.  

But not all constraints affect behavior directly. 

Psychological constraints 

do not change 

an actor's behavioral options so much as they change her or his understanding of those 
options. For example, some essayists in the 

Jones and Harris (1967) 

experiment were 

ostensibly instructed to write pro-Castro essays. However, the constraint imposed by a 
debate coach's instructions is quite different from the constraint imposed by a role-
conferred advantage. Unlike a role-conferred advantage, instructions neither force the 
essayist's hand nor make an anti-Castro speech difficult to write. Rather, the debate 
coach's instructions merely alter the payoffs associated with the two behavioral options. 
When a debate coach assigns a debater to defend Castro, the option of writing a pro-
Castro speech is suddenly infused with rich rewards (e.g., the goodwill of the debate 
coach) and the option of writing an anti-Castro speech is suddenly fraught with risks 
(e.g., public humiliation). The disparity between these payoffs may be sharp, but the 
essayist is still technically free to reap either. The essayist's behavioral options are not 
altered by the debate coach's instructions; rather, the essayist's motivation to enact each 
of the behavioral options is altered. Social pressure (a constraint commo nly used in 
attribution experiments) always affects expressive behavior by changing the actor's 
beliefs and desires, which then guide the actor's behavior.  

Attributionists typically treat these two classes of constraints as though they were 
identical, and this is a real mistake. Behavioral and psychological constraints not only are 
conceptually distinct but present very different attributional problems to the observer. 
When constraints are psychological, the unbiased observer need not be aware of the 
actor's situation as it is objectively constituted (i.e., the stuff of the external world); 
rather, the observer must be aware of the situation as it is subjectively construed by the 
actor (i.e., the actor's understanding of that stuff). Even if an observer can see the 
cheering crowd in all its clamorous and colorful glory, the critical question in this case is 
whether the politician can see the crowd and, if so, whether she sees it the same way the 
observer does. The senator who gives a prochoice speech to a local chapter of the 
National Organization of Women is behaving as any politically astute observer would 
expect, and thus her behavior may not call for a dispositional explanation. But if the 
senator mistakenly believes that she is addressing a convention of Roman Catholic 
priests, then a dispositional explanation is surely warranted. When constraints are 
psychological rather than behavioral, it is not the situation as it is but the situation as the 
actor sees it that matters.  

If observers have trouble recognizing the situation as it is (the invisibility problem), then 
they may have even greater trouble recognizing the situation as the actor sees it (the 
construal problem). People seem quite willing to act on the "egocentric assumption" that 
the situation the y perceive is the situation that the actor perceives as well ( 

Dunning, 

Griffin, Milojkovic, & Ross, 1990 

; 

Griffin, Dunning, & Ross, 1990 

; 

Griffin & Ross, 

1991 

; 

Ross & Nisbett, 1991, pp. 82—89 

). This egocentric assumption seems itself to 

have two origins. First, people generally have a difficult time using their imaginations to 
put themselves in someone else's epistemic shoes. To appreciate a situation from 
another's perspective, one must be able to imagine what that situation would have looked 
like if one had precisely the knowledge that the other person had. If a homeowner shoots 
Santa Claus after mistaking him for an armed intruder, then the juror who is considering 
the homeowner's claim of self-defense must ask himself or herself, "Would I have felt in 
mortal danger had I not known that the fat guy with the toy gun was Saint Nick?" 

Keysar 

(1994) 

has shown that people have considerable difficulty trying to partial out the effects 

of their idiosyncratic knowledge when attempting to take the perspectives of others. For 
example, subjects who knew that Linda disliked the restaurant that Ely had recommended 
perceived sarcasm when Linda left a note for Ely that read, "The restaurant you 
recommended was marvelous, 

just 

marvelous." Fair enough. But subjects also thought 

that Ely would perceive sarcasm in Linda's note, even though Ely did not know about 
Linda's culinary disappointment. Apparently, subjects could not quite imagine how the 
note would have appeared to someone who lacked their special knowledge of Linda's 
dining experience (see also 

Gilbert, 1991 

; 

Keysar, 1993 

; 

Schul & Burnstein, 1985 

; 

Wilson & Brekke, 1994 

; 

Wyer & Unverzagt, 1985 

). Similarly, once people know the 

solution to a problem, they are instantly unable to appreciate how difficult the problem 
would be for someone who did not know the solution ( 

Fischoff, 1975 

; 

Jacoby, Kelley, 

& Dywan, 1989 

).  

The second reason why observers may adopt the egocentric assumption is that people 
tend philosophically toward naive realism; that is, they consider their perceptions of the 
world to be the products of lower order, sensory processes that are informationally 
encapsulated ( 

Fodor, 1983 

) and that operate in about the same manner for everyone who 

shares their biology ( 

Griffin & Ross, 1991 

; 

Jacoby, Toth, Lindsay, & Debner, 1992 

; 

Jones & Nisbett, 1972 

). They do not seem to believe (as most psychologists and 

philosophers do) that perceptions are achieved by higher order, cognitive processes and 
are thus influenced by one's idiosyncratic beliefs, attitudes, and expectations. Ordinary 
people seem to have an "old look" view of their eyes as video cameras and their 
perceptions as captured images that are projected on some sort of cinema screen in the 
theater of the mind. As such, they expect anyone who possesses the same basic video 
equipment to experience the same perceptions they do. Because percepts are actually 
interpretations rather tha n reflections of the objective world, the philosophy of naive 
realism may lead observers to mistakenly assume that the actor shares their idiosyncratic 
view of the actor's situation.  

It is painfully obvious that observers must be aware of situational constraints if they are 
to consider the role that such constraints may have played in producing an actor's 
behavior. The correspondence bias is occasionally defined as the tendency to overlook or 

ignore situational forces; as the foregoing analysis suggests, however, the failure to 
recognize the presence of a situational force is a cause of correspondence bias. It may 
even be the primary cause. But it is by no means the only cause.  

Unrealistic Expectations for Behavior 

 

The observer may see the cheering crowd and may even understand the actor's 
idiosyncratic view of that crowd. Nonetheless, if the observer is to make an accurate 
attribution about a particular speaker's beliefs, he or she must also have a reasonably 
good idea of how a crowd typically affects a speaker's performance. Are most speakers so 
cowed by the mob that they pander to their audience's hopes and fears, or do most 
speakers ignore the preferences of their listeners and stand their ground? As we have 
noted, dispositional inferences occur when the observer's expectations are dashed by the 
actor's behavior, but surely such inferences are only as good as the expectations on whose 
dashing they depend. Clearly, observers who are completely aware of the actor's situation 
may still have unrealistic expectations about how that situation should affect the actor's 
behaviors (e.g., "A true liberal would never make a conservative speech"). How accurate 
are the ordinary person's estimates of the power of particular situations to evoke 
particular behaviors?  

The real-world survey research that might answer that question has not been done, but a 
consideration of the processes by which estimates of situational power are made may 
prove informative. One way to estimate the "power" of a situation is to estimate the 
typicality of the actor's response to that situation; indeed, the languages of power and 
typicality are virtually interchangeable inasmuch as a strong situation may be defined as 
one in which "anyone would have done the same thing." Yet, without the behavioral 
equivalent of a world almanac, each observer is left to his or her own devices when trying 
to determine what anyone would have done. One useful device is the availability heuristic 
( 

Tversky & Kahneman, 1973 

), which leads observers to judge behaviors that are easily 

imagined or remembered as especially common. This means that a behavior that just 
happens to be common in the observer's corner of the world, recent in the observer's 
experience, or part of the observer's own behavioral repertoire may be seen as enjoying 
greater consensus than in fact it does. 

Ross, Greene, and House (1977) 

asked subjects to 

make behavioral decisions (e.g., to choose or refuse to wear a signboard that read 

Eat at 

Joe's 

) and found that both choosers and refusers considered their own easily imagined 

choices to be typical of the population. As such, they drew dispositional inferences about 
those who made different choices. Using one's own imagined response to a situation as a 
basis for estimating the typical response (and hence the power of the situation) is risky 
business, not only because others may behave differently, but also because one does not 
always behave as one thinks one would. For example, decades of research on cognitive 
dissonance and attitude attribution have shown that when experimenters ask subjects to 
write counterattitudinal essays, compliance rates are exceptionally high. Yet, when 

Sherman (1980) 

asked college students to predict whether they would accede to an 

experimenter's request to write such an essay, nearly three quarters predicted they would 
not.  

To the extent that easily imaginable actions (such as one's own) are thought to be typical 
actions, use of the availability heuristic can lead observers to have unrealistic 
expectations for the behavior of others (e.g., "I would never make anti-American remarks 
under those circumstances, and I bet most other people wouldn't either"). When such 
expectations are violated, unwarranted dispositional inferences may result. The second 
obstacle to accurate attribution, then, is that even when the observer is perfectly well 
aware of the actor's situation, her or his expectations for behavior in that situation may be 
unrealistic (see 

Reeder, Fletcher, & Furman, 1989 

). In short, people may incorrectly 

estimate the power of certain situations to induce certain behaviors.  

Once again, the correspondence bias is sometimes defined as the tendency to 
underestimate the power of situations, but, as the foregoing analysis suggests, this is 
merely one of several causes of the bias. Nonetheless, social psychologists have made 
much of the ordinary person's tendency to underestimate the power of situations, and thus 
it is instructive to consider two caveats. First, when people underestimate the power of 
situations, they will be prone to make unwarranted dispositional inferences about actors 
who violate the erroneous expectations that such underestimates create. However, these 
underestimates should also cause observers to fail to make dispositional inferences when 
such inferences are, in fact, warranted. For example, observers typically underestimate 
the proportion of subjects who will deliver intense shock in the Milgram obedience 
situation (see 

Bierbrauer, 1979 

). That is, they unrealistically expect defiance and thus 

make unwarranted dispositional inferences about those who obey. But consider the 
inferences that such observers should make about the small but significant group of 
subjects who do refuse to deliver the shock. From the psychologist's point of view, such 
disobedience is rather unusual, and dispositional inferences are probably warranted. The 
observer, however, will be misled by his or her erroneous expectations and will conclude 
that these disobedient subjects were only "doing what anyone would do." Such observers 
will miss the opportunity to make the dispositional inferences that the data rationally 
require. The first caveat, then, is that when observers underestimate the power of the 
situation, they will indeed make logically incorrect inferences, but those incorrect 
inferences need not be dispositional.  

The second caveat is also important. Although observers may err by underestimating the 
power of situational forces, there are also instances in which they err by overestimating 
the power of these forces. 

Lepper, Greene, and Nisbett (1973) 

, for example, showed that 

children may overestimate the influence of a reward on their decision to play with 
particular toys and that this overestimation of situational power can lead the children to 
underestimate their dispositional interest in the toy. Likewise, 

Strickland's (1958) 

subjects overestimated the extent to which a watchful supervisor was responsible for an 
employee's honest performance and thus mistakenly underestimated the employee's 
dispositional honesty. Such laboratory demonstrations are paralleled by everyday 
experience. Many people are surprised, for instance, when they read in their daily paper 
about identical twins who were separated at birth and yet grew to have common habits, 
preferences, beliefs, and aptitudes as adults. Such remarkable "coincidences" are 
surprising only when one overestimates the power of environmental factors to shape 
personality and behavior.  

Together, these two caveats suggest that erroneous estimates of situational strength need 
not be underestimates, and that even if they are, underestimates need not lead to 
correspondence bias. Some theorists claim that people routinely underestimate the power 
of situations. This is undoubtedly true of subjects in social psychology experiments, but 
what does this say about the behavior of people in general? It says that underestimation 
can occur, but it does not say how often underestimation does occur ( 

Mook, 1983 

). 

After all, social psychology experiments are purposefully constructed to contain 
situational influences whose power will be underestimated because these are precisely the 
kinds of situational influences that intrigue the social psychologist. But such experiments 
are among the poorest vehicles for obtaining actuarial information about attributions. 
Because no effort is made to select representative situations or subjects, such experiments 
cannot reveal the kinds of attributions people usually, normally, routinely, generally, or 
typically make. This is not a condemnation of the experimental method but a recognition 
of its purpose, its capacities, and its limitations. As the Japanese say of fools, 

Ki ni yotte 

uo o motomu 

("You ask an elm tree for pears").  

Inflated Categorizations of Behavior 

 

If lack of awareness and unrealistic expectations were the sole causes of correspondence 
bias, then one would expect the bias to disappear when observers were completely aware 
of the actor's situation (whether actual or psychological) and had realistic expectations for 
behavior in that situation. But this is not the case. In fact, rather than providing a 
prophylactic against correspondence bias, awareness of situational constraints may 
actually cause it.  

One way to interpret 

Figure 1 

is that attribution requires observers to perform something 

of a "matching test" in which they compare the actor's behavior with their expectations 
for that behavior and determine whether the behavior meets these expectations. But, of 
course, observers compare their expectations not with the actor's actual behavior but with 
their perceptions of the actor's behavior. Why should this make a difference? Although 
some behaviors are easily perceived or categorized (it is the rare observer who confuses a 
punch in the mouth with a peck on the cheek), others admit to multiple identities, and the 
categorization of such ambiguous behaviors may be profoundly affected by knowledge of 
the context in which they occurred. Just as a political slogan may seem more radical 
when uttered by Vladimir Lenin than by Thomas Jefferson ( 

Lorge, 1936 

), so a mother's 

tears may appear more passionate when shed at her daughter's funeral than at her 
daughter's birthday party. This Kantian notion–variously called schema -driven 
processing, fctop-down processing, or perceptual assimilation–is among psychology's 
most venerable and robust (see 

Fiske & Taylor, 1991, pp. 96—177 

).  

Trope (1986 

; 

Trope, Cohen, & Maoz, 1988) 

has shown that an observer's awareness of a 

situation can give rise to expectations for an actor's behavior that, in turn, may induce the 
perceptual assimilation of that behavior. Interestingly, this phenomenon can have rather 
paradoxical effects. For example, if a situational force (e.g., a terrorist's threat) actually 
induces a certain kind of behavior (e.g., an anti-American speech), then the observer who 
is aware of the situation and who has a realistic estimate of its power should expect 

precisely that sort of behavior. However, the very awareness that enables the observer to 
have a realistic expectation for behavior may also cause the observer to have an 
unrealistic perception of that behavior; in this case, the behavior may be seen as 
conforming more to situational demands than it actually does. The observer who properly 
estimates the power of a terrorist's threat should be prepared to hear an anti-American 
speech, but that very expectation may cause the observer to believe that she has heard a 
very anti-American speech. As the generic attributional model predicts, such an observer 
should then be struck by the mismatch between her expectations and her perception of 
"reality" and should draw a dispositional inference about the speech maker. The irony, of 
course, is that the observer's excellent knowledge of the situation has "inflated" her 
categorization of the actor's behavior, which in turn has led her to make an unwarranted 
dispositional inference about an actor whose situation she understands perfectly.  

This inflated categorization effect (first suggested by Trope) has been demonstrated 
several times. 

Snyder and Frankel (1976) 

asked subjects to watch a silent film of a young 

woman being interviewed. Some subjects were told that the woman was being asked to 
discuss politics, and others were told that she was being asked to discuss sex. Some of the 
subjects were given this informa tion about the interview topic before seeing the film, and 
some were given the information only after seeing the film. Snyder and Frankel found 
that when subjects learned about the interview topic only after seeing the filmed behavior 
(in which case perceptual assimilation of the behavior was unlikely to have occurred), 
they took into account the anxiety-provokingness of the woman's situation and concluded 
that she was less dispositionally anxious in the "sex interview" than in the "politics 
interview" condition. That is, subjects used the discounting principle (i.e., "Don't attribute 

x 

units of anxious behavior to dispositional anxiety when the person is in a situation that 

provokes precisely 

x 

units of anxious behavior"). But subjects who learned about the 

interview topic before seeing the film drew precisely the opposite conclusion. 
Apparently, subjects who expected the woman to be talking about sex saw a great deal of 
anxiety ( 

x + n 

) in her somewhat ambiguous behavior. Although these subjects also used 

the discounting principle, they used it to discount a behavior that had already been 
overinflated during categorization ([ 

x + n 

] 

− 

x 

> 0). Similar effects have been shown by 

Trope and Cohen (1989) 

and 

Trope et al. (1988) 

.  

The third type of correspondence bias, then, is caused by inflated categorizations of the 
actor's behavior. But just as the power of situations can be overestimated as well as 
underestimated, 

Trope's (1986) 

model makes it quite clear that an observer's expectations 

will not necessarily induce perceptual assimilation of the actor's behavior. Indeed, there 
are a variety of well-documented circumstances under which expectations can cause 
people to perceive a stimulus as especially different than they expect it to be. For 
example, a member of a Baptist church who expects to hear a sermon on Christian morals 
and is instead surprised by a lecture on the health benefits of bisexuality may perceive 
that speech as somewhat more liberal (and not somewhat more conservative) than it 
actually is. Similarly, there are circumstances under which expectations have little or no 
effect on one's perceptions: Expecting to see an apple may cause one to misidentify a 
pear at the far end of a dark room, but probably not an airplane from 10 paces on a sunny 
day. The conditions that promote assimilation, contrast, or neither are well explicated 

(e.g., 

Lombardi, Higgins, & Bargh, 1987 

; 

Martin, Seta, & Crelia, 1990 

), and our only 

point here is that awareness of the situations that constrain an actor's behavior may, but 
need not, engender inflated categorizations of that behavior and, ultimately, 
correspondence bias.  

Incomplete Corrections of Dispositional Inferences 

 

We have argued that dispositional inferences are the products of a mismatch between the 
observer's expectations for and perceptions of the actor's behavior. If the observer 
improperly calculates the value of either of these elements, a "false mismatch" will result, 
and correspondence bias may follow. Does this mean that when a match (rather than a 
mismatch) is detected, the observer will refrain from drawing a dispositional inference? 
The logical answer is yes, but the psychological answer is no.  

Recent research suggests that observers may draw dispositional inferences about an actor 
even when the actor's behavior matches their expectations and that they must 
subsequently undo or correct such inferences when they finally notice the match ( 

Quattrone, 1982b 

). Fine-grained analyses of attributional process suggest that, under 

many conditions, observers spontaneously draw trait inferences from behavior ( 

Lupfer, 

Clark, & Hutcherson, 1988 

; 

Uleman, 1987 

; 

Winter & Uleman, 1984 

; 

Winter, Uleman 

& Cunniff, 1985 

; cf. 

Bassili & Smith, 1986 

; 

Whitney, Waring, & Zingmark, 1992 

) and 

that they draw such inferences with remarkable efficiency ( 

D'Agostino & Fincher-

Kiefer, 1992 

; 

Newma n, 1991 

). 

Gilbert, Pelham, and Krull (1988) 

combined these 

insights into a model that suggests that when people attempt to understand others, they 
begin by inferring the presence of a corresponding disposition. Only after having done so 
do they check to see whether the actor's behavior actually matched their own expectations 
(i.e., whether the behavior was precisely what the situation required). In essence, 
observers seem to take one step forward (they draw dispositional inferences) and then 
one step backward when they must (they correct those inferences when the actor's 
behavior matches their expectations).  

Figure 2 

is an expanded version of 

Figure 1 

that includes these ideas. Frankly, such 

microanalyses of attributional process would be an esoteric concern were it not for one 
thing: The model suggests that the initial dispositional inference is relatively resource 
efficient (i.e., it does not require considerable effort or conscious attention) and that the 
subsequent correction is less so. Because the two processes differ in the amount of 
thoughtful deliberation they require, they are differentially susceptible to impairment by 
competing cognitive demands. Specifically, the initial dispositional inference is relatively 
unaffected by the other tasks in which the observer may be concurrently engaged, 
whereas the fragile correction of that inference becomes difficult or impossible. 
Observers who are not able to devote their attention to attributional work draw 
dispositional inferences about the actor but are unable to correct such inferences even 
when they notice that they are unwarranted.  

Such an effect has been demonstrated by 

Gilbert, Pelham, and Krull (1988) 

, who asked 

observers to watch a videotape of a woman who appeared quite nervous while engaging 

in a conversation with a stranger. Subjects were not allowed to hear the conversation, but 
subtitles on the screen told them what sorts of  topics the woman and her partner had been 
assigned to discuss. Sometimes the subtitles depicted an anxiety-provoking situation (the 
woman was discussing her sexual fantasies) and sometimes a mundane situation (the 
woman was discussing her hobbies). Control subjects used the discounting principle and 
took account of the woman's situation when they made inferences about her (i.e., they 
rated the woman as less dispositionally anxious when she was discussing anxiety-
provoking rather than mundane topics). But observers who were asked to rehearse a set 
of word strings did not (i.e., they rated the woman as dispositionally anxious regardless 
of the topic she was discussing). In other words, observers who rehearsed word strings 
seemed to draw dispositional inferences about the actor and then failed to correct those 
inferences with information about the topics the actor was discussing.  

One might wonder whether observers were simply too busy rehearsing the word strings 
even to read the topics. If observers did not have the situational constraint information, 
then they could hardly be expected to use it, and their tendency toward correspondence 
bias could be understood as a result of lack of awareness resulting from situational 
invisibility. To eliminate this possibility, 

Gilbert, Pelham, and Krull (1988) 

had observers 

rehearse a very special set of word strings: the discussion topics themselves. In effect, 
observers were asked to memorize the situational constraint information, and the model 
predicted that those observers who were asked to memorize this information would be the 
least likely to use it. This is just what happened. Subsequent research has shown that 
observers who engage in any one of a variety of demanding activities (e.g., visual search 
tasks, digit rehearsal, gaze fixation, or strategic self-presentation) will draw dispositional 
inferences about an actor but will fail to take the second step and correct those inferences, 
even when the actor's behavior conforms perfectly to their expectations ( 

Gilbert, Krull, 

& Pelham, 1988 

; 

Gilbert, McNulty, Giuliano, & Benson, 1992 

; 

Gilbert & Osborne, 

1989 

; 

Osborne & Gilbert, 1992 

). The fourth type of correspondence bias occurs, then, 

when observers are unable or unwilling to correct the dispositional inferences that they 
seem to draw with relative spontaneity and ease. 

2 

 

Just as there are limits on the other mechanisms that cause correspondence bias, so too 
are there limits on the incomplete correction mechanism. For example, 

Krull (1993) 

has 

suggested that initial dispositional inferences are not fully automatic inasmuch as they 
require that the observer have the goal of understanding the actor (see 

Bargh, 1989 

). 

Krull showed that observers who want to understand a situation rather than an actor do 
not initially draw dispositional inferences about the actor. Rather, they draw inferences 
about the situation ("That must be a very anxiety-provoking topic she's discussing") and 
then correct those inferences with information about the actor's dispositions ("Of course, 
that woman could just be a nut case, so maybe the topic isn't so anxiety provoking after 
all"). According to Krull, observers are capable of executing either of two information-
processing sequences: dispositional inference followed by situational correction (the D 
sequence) and situational inference followed by dispositional correction (the S sequence). 
The observer's epistemic goals determine which of these sequences is executed. 

Gilbert, 

Pelham, and Krull (1988) 

showed that when cognitively loaded observers execute the D 

sequence, their ability to correct their dispositional inferences is impaired, and they 

display correspondence bias. 

Krull (1993) 

showed that when observers execute the S 

sequence, cognitive load brings about a conceptually opposite effect. In such cases, the 
loaded observer draws situational inferences with ease, but her or his dispositional 
correction of those inferences is impaired. In short, incomplete corrections need not 
necessarily lead to correspondence bias because they are not necessarily situational 
corrections of dispositional inferences.  

The Salience Bugaboo: A Fifth Cause? 

 

The preceding sections have described four distinct causes of correspondence bias. 
Observers may draw unwarranted dispositional inferences because (a) they lack 
awareness of the actor's situation as it is objectively constituted or subjectively construed, 
(b) they have inappropriate expectations for how a person will behave in such a situation, 
(c) their awareness of the actor's situation has led to an inaccurate perception of the 
actor's behavior, or (d) they lack either the motivation or the capacity to correct the trait 
inferences they may have spontaneously and effortlessly made. But even readers who 
have only a passing familiarity with the literature on correspondence bias will have noted 
a conspicuous lapse in our discussion: We have said nothing so far about the well-known 
salience effect that is so often invoked as an explanation of correspondence bias. This 
lapse was intended. We believe that, depending on what one means by the term 

salience, 

the explanation is either redundant with other explanations or lacking in empirical 
support.  

Let us review 

Heider's (1958) 

famous maxim in its entirety: "Behavior ... has such salient 

properties that it tends to engulf the field rather than be confined to its proper position as 
a local stimulus whose interpretation requires the additional data of a surrounding field–
the situation in social perception" (p. 54). Can the language of gestalt psychology be 
peeled back to determine just what Heider was trying to say here? One interpretation is 
that Heider was merely pointing out what 

Ichheiser (1949) 

pointed out, what 

Skinner 

(1971) 

pointed out, and what many thinkers have pointed out: Behavior is easy to see, but 

situations often are not. The relative pallidness of situational constraints may sometimes 
prevent observers from attaining information about situational constraints (lack of 
awareness) or prevent them from using that knowledge when making attributions 
(incomplete corrections). In either case, it may give rise to correspondence bias. If 
salience simply refers to the general pallidness of situational information, then salience 
may exert its effects through either of two mechanisms that we have already discussed.  

Indeed, were this all that theorists meant when they invoked "the salience explanation," 
we would have no argument. But this is not all they mean. Rather, some take 

Heider's 

(1958) 

statement to mean that there is something almost magical about salient behaviors 

that causes observers to draw dispositional inferences about the actors who perform them. 
At first blush, this suggestion has the ring of reason: "Behavior is more salient than the 
situation in which it occurs, and this is why people attribute the behavior to the actor who 
is performing it." But a moment of thought reveals that  the notion is a non sequitur. In 
classical attributional terms, behavior is an effect to be explained, and dispositions and 
situations are two possible causes of that effect. Why, then, should the salience of the 

effect facilitate attribution to one of those causes? It is not, after all, the person's 
dispositions that are salient, but the person's behavior. Indeed, in no other case would 
reasonable people argue that the salience of an effect "explains" its attribution: If a 
physical symptom such as vomiting were particularly salient, this would not explain why 
physicians attribute that symptom to a virus rather than to a bacterium.  

It would seem that there is no compelling reason why the salience of behavior should 
facilitate dispositional inference, except to the extent that salient behavior obscures 
situational influences. Nonetheless, theorists argue that "what you 

attend 

to is what you 

attribute 

to" and insist that "there is no generalization coming from the Heider-inspired 

attribution literature of the 1970's that is better supported than this" ( 

Nisbett, 1987, p. 

109 

; 

Ross & Nisbett, 1991, p. 140 

). What is the basis of this ostensible support? Four 

studies are widely cited as demonstrating that the salience of an actor facilitates 
dispositional inferences by an observer; in our view, however, these studies provide 
precious little support for such a conclusion.  

There are two problems with these studies. First, each used what is now recognized as a 
notoriously flawed measure of attribution (see especially 

F. D. Miller, Smith, & Uleman, 

1981 

). Throughout the 1970s, attribution theorists commonly asked subjects to make 

attributions on a scale whose endpoints were 

situational 

and 

personal 

(or sometimes 

dispositional 

). Some investigators used a single scale, and some used separate scales and 

then analyzed difference scores, but all asked subjects to estimate the extent to which a 
given behavior was caused by "something about the person" or by "something about the 
situation." Unfortunately, all behaviors are capable of revealing something about the 
person or something about the situation. Students of S-O-R psychology know that any 
action (R) can be described in terms of the environmental factors that enabled it (S) or in 
terms of the psychological constructs that mediated it (O). The American hostage who 
denounces his country on Iranian television and the student-activist who denounces her 
country on a college campus seem on the face of it to provide clear examples of 
situationally and dispositionally caused behaviors. But the student-activist has parents, 
peers, teachers, and audience members (situational factors) who are potent sources of 
influence on her behavior, and the political hostage has thoughts, feelings, and goals 
(dispositional factors) that led him to recite his captors' dogma. Asking whether such 
behaviors were caused by situations or caused by persons is in some senses akin to asking 
whether a golf ball moved across a green because it was round or because someone 
tapped it with a putter. Both the ball's shape and the force of the stroke are reasonable 
ways to describe the origin of its motion. As such, an observer's preference for one 
description over the other may reflect little more than linguistic convention (one does not 
usually implicate the invariant roundness of a golf ball when explaining its movement), 
point of view (the golfer and the golf ball manufacturer will surely supply different 
answers to the question "Why did that ball roll so nicely?"), or even intellectual 
orientation (psychoanalysts may stress the role of roundness, and behaviorists may stress 
the power of putters; see 

Hilton & Slugoski, 1986 

; 

Kahneman & Miller, 1986 

; 

McGill, 

1989 

; 

Nisbett & Ross, 1980 

). The person—situation scale is a theoretical jambalaya that 

has time and again been shown to comprise a psychometrically unsound dimension (e.g., 
see 

Buss, 1978 

; 

Kruglanski, 1975 

; 

F. D. Miller et al., 1981 

; 

Solomon, 1978 

; 

White, 

1991 

); thus, even if the studies that provide the empirical foundation for the salience 

explanation had found the predicted differences on this scale, the interpretation of those 
findings would be unclear. From our point of view, however, the more damning problem 
is the second one: For the most part, these studies did n  

Taylor and Fiske. 

 

In the best known of the four studies, 

Taylor and Fiske (1975) 

positioned observers so 

that one participant in a conversation was more salient than the other, and they then asked 
observers to complete two kinds of measures. Observers reported their perceptions of 
each of the participants on the person—situation scale, and they also reported their 
perceptions of the interaction itself (e.g., how much a participant set the tone of the 
interaction, caused his or her partner's responses, and so on). The investigators predicted 
differences on both measures. Instead, they found that subjects' perceptions of the 
interaction were indeed influenced by their seating positions; however, when  

subjects indicated how dispositionally caused and how situationally 
caused each behavior was for each confederate ... none of the predictions 
were borne out; in fact, there were no significant effects or trends in any of 
the analyses. ... [The salient participant's] behavior was not seen as 
indicative of his dispositions, nor was his partner's behavior seen as 
situationally based. ( 

Taylor & Fiske, 1975, p. 442 

)  

Taylor and Fiske (1975) 

wondered whether the reason for this unexpected result might 

have been that subjects had been asked to attend to the conversation rather than to one of 
the participants. In a second experiment, they explicitly instructed some subjects to attend 
to one participant, and, "contrary to the hypothesis, subjects who were told to observe one 
participant in particular were no more likely to see his behaviors as dispositionally based 
than were subjects who were not told to attend to any participant in particular" ( 

Taylor & 

Fiske, 1975, p. 443 

). In an extension of this study, 

Ellis and Holmes (1982) 

found that 

directing a participant's (rather than an observer's) attention to an interaction partner had 
absolutely no effect on the participant's tendency to draw dispositional inferences about 
the partner. In short, the study that is most often cited in support of the salience effect 
found no evidence whatsoever for the contention that salient behavior facilitates 
dispositional inference.  

McArthur and Post. 

 

McArthur and Post (1977) 

reported the results of five studies in which the salience of an 

actor was manipulated in a variety of innovative ways (e.g., shirt color, motion, and 
brightness). Observers rated the actors on the person—situation scale, and McArthur and 
Post presented a mixed bag of results: Some of their experiments showed an increase in 
the relative contribution of personal (as opposed to situational) causes, but most showed 
unexpected reversals of this effect. Separate analyses of the dispositional and situational 
measures revealed that all of the "action" involved the latter measure. The investigators 
appropriately concluded that "being physically conspicuous ... does not seem sufficient to 

have a significant influence on attributions of behavior to dispositional causes" (p. 534). 
In short, McArthur and Post found no reliable evidence for the contention that salient 
behavior facilitates dispositional inference.  

Arkin and Duval. 

 

Arkin and Duval (1975) 

manipulated the self-focus of an actor (either by videotaping or 

not videotaping the actor as he or she chose the most appealing of several pieces of 
artwork) and also the salience of the actor's environment (by allowing the actor to peruse 
stationary photographs of the artwork or by showing the actor a dynamic video 
presentation of the artwork). Observers made attributions about actors on the person—
situation scale. Of course, neither of the independent variables was a manipulation of the 
actor's physical salience for the observer, and thus this well-cited experiment is not 
directly relevant to the salience explanation. Nonetheless, the latter manipulation (it could 
be argued) manipulated the relative salience of the actor by manipulating the salience of 
his or her environment. When 

Arkin and Duval (1975) 

analyzed observers' dispositional 

attributions for the actor's behavior, they found "no significant main effects or 
interactions" (p. 434). Most important, when the actor's environment was made salient, 
observers showed no attenuation of their tendency to draw dispositional inferences about 
the actor.  

Storms. 

 

Storms (1973) 

asked subjects to engage in a conversation with another subject, and he 

then asked both of these actors and a matched set of observers to make attributions for the 
actors' behavior on the person—situation scale. Storms explained to subjects that 
personal causes included "personality, traits, character, personal style, attitudes, mood" 
(p. 168). Subjects displayed the well-known actor—observer effect ( 

Jones & Nisbett, 

1972 

); that is, actors were less likely than were their matched observers to attribute their 

own behavior to dispositional causes. Some actors and observers were then shown a 
videotape of the interaction that was shot from the visual perspective of the other actor 
and were asked to make attributions anew. Unlike most other investigators, Storms did 
indeed find that this change of visual perspective reversed the actor—observer effect; that 
is, the reoriented actors were, in fact, more likely than the reoriented observers to 
attribute their behaviors to dispositional causes.  

But 

Storms (1973) 

was prescient in his concern that the person—situation scale might not 

adequately measure dispositional inference, and he therefore included a much better 
measure. Storms asked subjects to rate how friendly, talkative, nervous, and dominant the 
actors had been during the conversation and then to rate how the actors generally behaved 
on each of these dimensions. Storms correctly reasoned that "if a subject had perceived 
that the actor's behavior in the conversation was due to a stable personal disposition, then 
the observer would likely have predicted that the actor behaved the same way in general" 
(p. 168). This measure is superior to the person—situation scale inasmuch as it requires 
subjects to predict future behavior, a task that 

Ross (1977) 

later suggested is the sine qua 

non of dispositional inference. What did Storms find when he analyzed this superior 

measure? "Although the direction of differences ... was as expected, none of the 
individual comparisons between cells reached significance" (p. 171). In fact, ratings on 
the problematic person—situation scale explained only 13% of the variance in 
predictions of behavior.  

Unengulfing the field. 

 

The findings of 

Arkin and Duval (1975) 

, 

McArthur and Post (1977) 

, and 

Taylor and 

Fiske (1975) 

are particularly troubling for the hypothesis that the salience of behavior 

facilitates dispositional attribution. All three studies used the person—situation scale (or 
two separate scales) and found that salience did not increase the likelihood that observers 
would implicate "something about the person" as the cause of an actor's behavior. 

Ellis 

and Holmes (1982) 

found the same thing for participants in an interaction. 

Storms (1973) 

was the only one of these investigators to find differences on the difficult-to-interpret 
person—situation scale, and he found only marginal (and weakly correlated) differences 
on a clearly superior measure. Taken together, what does all of this mean? It does not 
mean that salience has no effects on human judgment (see 

Taylor & Fiske, 1978 

). 

Indeed, every one of the aforementioned investigators found that salience did something 
interesting to subjects' responses (e.g., to their perceptions of the interactions or to their 
ratings of situational causality). But they did not find that salience increased their 
subjects' willingness to infer dispositions from behavior. Our reading of the literature 
leads to a simple conclusion: The relative salience of behaviors and situations facilitates 
dispositional inference only to the extent that it prevents subjects from possessing or 
using information about the actor's situation. Above and beyond the effects that salience 
exerts through the mechanisms of lack of awareness and incomplete corrections, we have 
found little evidence to suggest that it exerts an independent effect. From our perspective, 
the salience of behavior does not qualify as a fifth, independent cause of correspondence 
bias.  

Consequences of Correspondence Bias  

We have described four mechanisms that can produce the correspondence bias. Such 
mechanisms are proximal causes; that is, they explain how the bias is produced, but they 
do not explain why. What are the ultimate causes of correspondence bias? Most modern 
psychologists are functionalists in that they define the ultimate cause of a behavior in 
terms of its beneficial consequences for the organism. It is said that a particular 
phenomenon occurs because it (or the more basic process in which it is grounded) fills a 
need of the individual and is thus selected (or, at least, not selected against) at the 
ontogenetic or phylogenetic level. What do mechanisms that produce correspondence 
bias do that some other mechanisms might not? Are there benefits to having–or being–the 
sort of machinery that makes logically unwarranted dispositional inferences? We believe 
that there are fewer negative and more positive consequences than an unreflecting 
analysis might at first suggest and that this relative immunity to the consequences of 
correspondence bias may partially explain its persistence.  

Negative Consequences 

 

No one doubts that inferential errors can have suboptimal, maladaptive, and even tragic 
consequences. An unwarranted dispositional inference may constitute a "dangerous 
epistemological stance" that places one in "dire peril" ( 

Nisbett, 1987, pp. 103—104 

). 

Nonetheless, it is worth reminding oneself that, like the incorrect solution to a 
mathematical puzzle, the correspondence bias is a logical error–an inference about the 
existence of an attribute whose existence logic places in doubt–and that errors on the 
plane of pure reason do not always count as disasters on earth. As 

H. G. Wells (1932, p. 

76) 

wryly noted, "No appreciable effect has been produced upon the teaching of machine 

drawing by the possibility that space is curved and expanding." Indeed, it would not be 
terribly cost-effective for an architect to worry about the curvature of space and use 
Riemannian geometry when designing a split-level ranch house because the strictly 
incorrect (but elegantly simple) assumptions of Euclidian geometry serve quite nicely 
those who measure spaces shy of parsecs. Similarly, if an inferential process produces an 
occasional logical error but also a significant savings of time and energy, it may provide a 
net benefit to the mental system that uses it ( 

Hogarth, 1981 

; 

Nisbett & Ross, 1980 

). It is 

strictly incorrect, for example, to assume that all people with low voices and beards are 
male and that all people with high voices and enlarged breasts are female. Nonetheless, 
this assumption is so close to being perfectly true that it would hardly behoove the 
bachelor to abandon it and insist that his dates undergo genetic testing. The time and 
energy that one saves by using such heuristics is probably worth the cost of their rare 
failures.  

Is there any reason to believe that the inference of corresponding dispositions is a 
similarly useful heuristic? No one can say how often people are as they act; once again, 
social psychology experiments are particularly incapable of determining whether 
dispositional inferences are warranted on most occasions, some occasions, or any 
occasions. But it is worth noting that there are three easily imaginable circumstances 
under which an observer's tendency to ignore the situational constraints on another's 
behavior will–like the architect's tendency to ignore the curvature of space–cause no 
structural damage. These are the cases of self-induced constraint, omnipresent constraint, 
and superfluous constraint.  

Self-induced constraint. 

 

Subjects in social psychology experiments are usually assigned randomly to the situations 
in which they behave; subjects in real life may or may not be. At birth, one inherits a 
national identity, a cultural and racial heritage, and a socioeconomic circumstance. Surely 
these assignments are "random" inasmuch as one does not choose them, but many of the 
important situations that shape one's life are situations that one does, in fact, enter by 
choice or is drawn into by proclivity. In their comprehensive discussion of situational 
choice, 

Snyder and Ickes (1985, p. 918) 

concluded that "individuals appear to gravitate 

actively toward social situations that will foster and encourage the behavioral expression 
of their own characteristic dispositions and interpersonal orientations." In other words, 
people seek situations that will "push" them in the same direction as do their own 
dispositions.  

To the extent that the constraints on a person's behavior are freely chosen or otherwise 
self-induced, it may do the observer little harm–and even much good–to ignore the 
effects of these constraints when making attributions. For example, the role of banker 
demands conservative dress, a preoccupation with finances, and a somewhat formal 
demeanor. If a person were randomly assigned to that occupation, then the 
correspondence-biased observer would attribute dispositional conservatism to the banker 
at his or her own inferential peril. Despite what the savings and loan crisis might suggest, 
bankers are not randomly assigned to their professions. In fact, it is probably the 
dispositionally conservative, formal, and economically minded person who is most likely 
to be drawn to a career in banking. In this case, the situational forces do not elicit the 
actor's behavior so much as the actor's dispositions elicit the situational forces.  

An observer's failure to discount behaviors performed under self-induced constraints will 
not necessarily lead to correspondence bias. In fact, when situational forces are entirely 
self-induced, the use (and not the ignorance) of the discounting principle may actually 
lead to serious inferential error. For example, if one assumes tha t the role of professor 
demands intellectual curiosity and that professors are therefore not more dispositionally 
curious than grocery clerks and undertakers, one will be wrong on several days of the 
week. One will have ignored not a situational cause of the actor's behavior (i.e., role 
demands) but a behavioral effect of the actor's dispositions (i.e., occupational choice). As 

Wachtel (1973) 

suggested:  

The understanding of any one person's behavior in an interpersonal 
situation solely in terms of the stimuli 

presented to 

him gives only a 

partial and misleading picture. For to a very large extent, these stimuli are 

created by 

him. They are responses to his own behavior, events he has 

played a role in bringing about, rather than occurrences independent of 
who he is and over which he has no control. ... [Situations are] largely of 
one's own making and [are] themselves describable as a characteristic of 
one's own personality. (p. 330)  

Attributionists prize the discounting principle. And it is indeed a handsome logical tool. 
But one must not forget that the discounting principle is valid only when situations and 
dispositions are independent causes of behavior that do not affect each other. This is 
usually the case in psychology experiments, in which subjects are randomly assigned to 
experience short-lived situational constraints. But to assume that the effects of situational 
forces must always be subtracted out of the behavior when one diagnoses an actor's 
dispositions is to overlook the fact that, outside the psychology experiment, such forces 
may be telltale effects of the very dispositions one hopes to diagnose. They may, in fact, 
constitute useful information. Of course, no one knows if the situational forces of 
everyday life are "largely of one's own making." But certainly they are sometimes of 
one's own making, and, when they are, the observer who ignores the discounting 
principle will save time, save energy, and make the right call to boot.  

Omnipresent constraints. 

 

Just as one may ignore self-induced situational constraints and end up with an accurate 
inference nonetheless, so may one ignore omnipresent constraints and end up with an 
adequate inference. As 

Swann (1984) 

noted, many of one's interactions with others take 

place in a restricted set of situations: One sees one's students in the classroom but not in 
the bathtub, one's loan officer at the bank but not at the ballpark, and so on. As such, the 
situational forces that shape an actor's behavior in one instance may continue to shape 
that behavior in every instance in which one observes it; thus, one may neither wish nor 
need to subtract out the effects of these forces on behavior (this is especially true when 
one actually constitutes the situational force that constrains another; see 

Gilbert & Jones, 

1986 

). Perhaps working as a night manager at an inner-city 7-Eleven is enough to make 

even the most trusting soul behave like the warden of a maximum security prison. If one 
ignores the role that such a distrust-inducing situation plays and concludes that the night 
manager is, in fact, dispositionally distrustful, then that correspondence-biased inference 
(which is technically incorrect) will still allow one to predict the night manager's 
behavior with enviable accuracy. Indeed, if one were to spend the extra time and energy 
necessary to conduct a full-scale attributional analysis, it is not clear that one would reap 
any additional inferential reward. After all, when will one ever need to predict the night 
manager's behavior except in the fluorescent landscape of that particular convenience 
store?  

When the person and situation are perfectly confounded and the observer is willing to 
settle for "circumscribed accuracy" rather than "global accuracy" ( 

Swann, 1984 

), it may 

not matter whether the situation or the actor's disposition is the true cause of the actor's 
behavior. In such cases, a dispositional inference delivers a lot of bang for the buck. It is 
interesting to note that during most of human history, situations and roles have been 
rather well confounded with the individuals who have occupied them ( 

Stavrianos, 1989 

). Members of hunter—gatherer and early agrarian societies probably had little need to 
predict the behavior of individuals outside of their well-known social roles because 
individuals rarely existed outside of these roles. Only in a modern, mobile, multicul tural 
society can people move easily from one role or situation to another; therefore, only in 
such a society will correspondence-biased inferences have potentially troublesome 
consequences. Until recently, the omnipresence of situational constraints may have 
rendered the correspondence bias little more than a logical faux pas.  

There is a second reason why observers may ignore omnipresent constraints with relative 
impunity. As 

Higgins and Winter (1993) 

argued, it makes good attributional sense for an 

observer to subtract out a situation's effects on an actor's behavior when that situation is a 
fleeting force that may be working in opposition to the actor's enduring personal 
characteristics. But some situations do not fleet as quickly as others. In fact, when 
situations are enduring, they may shape behavior not by facilitating or opposing the 
actor's dispositions but by creating them. Drinking beer with a group of longshoremen 
may induce a timid young man to offer a few uncharacteristically bawdy stories, but 
being raised by the same group of longshoremen may cause the young man to relish such 
stories. In other words, when situations are temporary, they encourage temporary 
fluctuations in overt behavior, and one says that the behavior has been changed by the 
situation. But when situations are enduring, they may foster enduring behavioral 

tendencies, and one says that the actor has been changed by the situation. At what point 
does acting end and being begin? The answer to this philosophical riddle is one that 
attributionists have yet to find (probably because it is one they have yet to seek). But 
clearly, to the extent that omnipresent constraints can create dispositions, observers who 
ignore those constraints will not suffer. In fact, observers who attempt to use the 
discounting principle to subtract out the effects of disposition-generating situations (e.g., 
"The battered child isn't dispositionally fearful, she's just been in a scary situation for 10 
years") will end up with an erroneous inference. Once again, when situations and 
dispositions are causally related, the discounting principle is not a valid logical tool, and 
those who ignore it will reap inferential rewards for doing so.  

Superfluous constraints. 

 

When observers take into account the influence of a situational constraint on an actor's 
behavior, they can, of course, take into account only the influence that they believe the 
constraint has exerted. Sometimes these beliefs are wrong because sometimes situational 
constraints are superfluous; that is, they coerce an actor to do what she or he would have 
done anyway. For example, children in the 

Lepper et al. (1973) 

study were given an 

award for playing with toys that they would have played with even if the award had not 
been offered. Likewise, subjects in a study conducted by 

Gilbert and Silvera (1993) 

were 

given help with an anagram test even though they would have attained perfect scores 
without that help. In both of these cases, actors were laboring under constraints that were, 
for them, mere window dressing, and when observers behaved like logical attributers 
(i.e., when they used the discounting principle), they mistakenly concluded that the actors 
lacked a predisposition to engage in their respective toy-playing and test-passing 
behaviors. Because the constraints in these instances were superfluous, observers would 
have done well to ignore them. Indeed, when Gilbert and Silvera put some observers 
under cognitive load and thereby impaired their ability to correct their dispositional 
inferences, the loaded observers made judgments that were logically superior to those of 
their unimpaired counterparts. In short, observers who ignore constraints that are not 
actually controlling an actor's behavior will not suffer for that ignorance; in fact, that 
ignorance will increase the accuracy of their attributions. 

4 

 

Positive Consequences 

 

People make attributions because doing so enables them to achieve certain ends, for 
instance, to predict others and thereby control the extent to which others' behavior can 
affect them. 

Heider (1958) 

was adamant in his contention that dispositional inferences are 

a way of gaining power over one's world: "Man grasps reality, and can predict and 
control it, by referring transient and variable behavior and events to relatively unchanging 
underlying conditions, the so-called dispositional properties of his world" (p. 79). Also, 
"a personality characteristic enables one to grasp an unlimited variety of behavioral 
manifestations by a single concept ... [and] insofar as personal dispositions are connected 
in lawful ways with other features, predictions about behavior of the other person become 
possible" (p. 30).  

According to 

Heider (1958) 

, dispositional inference is like a naive factor analysis, a data-

reduction technique that enables a large array of behaviors to be understood in terms of a 
few underlying commonalities that he called dispositions. Because dispositional 
inferences are so economical, observers want to make them, and, as every thinker from 
Plato to Freud has acknowledged, when people want to believe that something is the case 
they often find ways to do so. In this sense, correspondence bias is a sort of "wishful 
thinking" that gratifies the individual who wishes to predict the behavior of others. In 
support of this contention, a number of studies do suggest that, when the individual's 
need for control is piqued (e.g., by making the observer's outcomes dependent on the 
actor), the tendency toward dispositional inference may be exacerbated (e.g., 

Berscheid, 

Graziano, Monson, & Dermer, 1979 

; 

D. T. Miller, Norman, & Wright, 1978 

; but see 

Pittman & D'Agostino, 1985 

, for the opposite effect).  

As interesting as this account may be, it is not quite complete. Specifically, it is not clear 
why the "unchanging underlying conditions, the so-called dispositional properties of his 
world" must refer to characteristics of persons rather than to the characteristics of 
situations (see 

Nisbett, 1987, p. 109 

). One can surely imagine an extraterrestrial who 

tends to attribute the behavior of human beings to the enduring and unique characteristics 
of their situations and thereby gains the ability to predict and control people through an 
expert understanding of situational influence. (Note that the extraterrestrial would be in 
very much the same business as the social psychologist.) In other words, factor analysis is 
a powerful technique because it extracts a few factors from large amounts of data, not 
because it extracts factors of a particular kind. Why, then, do observers satisfy their need 
for control by reducing behavioral observations to personal dispositions rather than to 
situational characteristics?  

As with so many things, Western culture may be the culprit inasmuch as it encourages 
people to use one control-enhancing strategy (attribution of behavior to dispositions) 
rather than the other (attribution of behavior to situational characteristics). As 

Nisbett 

(1987, p. 110) 

has argued, "Much of Western culture, from the Judeo-Christian insistence 

on individual moral responsibility to the intellectual underpinnings of capitalism and 
democracy, emphasize the causal role of the actor." Some writers have argued that 
capitalist societies maintain an illusion of fairness by teaching their members that they 
are both the proximal and ultimate causes of their own behavior; as such, both the 
"haves" and the "have nots" are socialized to believe that they are responsible for their 
respective successes or failures (e.g., 

Lukes, 1973 

; 

Weber, 1930 

). Presumably, if 

capitalist societies embraced the Marxist view that behaviors (and their consequences) 
are essentially the products of the sociopolitical contexts in which they occur, then 
dissatisfaction among the lower classes would invite revolutionary upheaval. If one finds 
this particular claim a shade too sinister, there is surely a very long list of other, less 
cynical reasons why Western cultures promote a dispositionist view of human behavior 
(e.g., 

Markus & Kitayama, 1991 

; 

Ross & Nisbett, 1991, pp. 169—203 

; 

Spence, 1985 

). 

We need not review them here. The point is simply this: Drawing dispositional inferences 
may be only one way of satisfying the need for control, but it seems to be the one way 
prescribed by Western culture (see 

J. G. Miller, 1984 

; 

Newman, 1993 

; 

Schweder & 

Bourne, 1982 

). When dispositional inferences are unwarranted, this sense of control may 

be illusory, but even illusory control can have sanguine effects ( 

Alloy & Abramson, 

1979 

; 

Langer, 1975 

; 

Taylor & Brown, 1988 

), and thus the mechanisms that produce it 

may have an advantage.  

Let us return to the question that opened this section: Why the correspondence bias? A 
few answers have emerged. First, dispositional inferences are easy to make and are 
undoubtedly correct on some occasions. Second, even when they are incorrect in the 
strictly logical sense, they may have few unfavorable and many favorable consequences 
for the observer as long as the situation that she or he has ignored is an effect of the 
actor's dispositions, a cause of the actor's dispositions, or simply the same situation within 
which he or she wishes to predict the actor's behavior. Finally, dispositional inferences 
afford the observer a culturally acceptable way of gaining a sense of control over her or 
his environment, and feelings of control, however illusory, may ultimately yield greater 
psychological benefits than would logically impeccable inferences. The goodness of the 
answer to any question about ultimate causes depends, of course, on what satisfies the 
person who posed it. To the extent that a surfeit of positive and a lack of negative 
consequences can be said to explain why a psychological phenomenon exists, the 
ultimate causes of correspondence bias seem tractable.  

Coda  

We may strive to see others as they really are, but all too often the charlatan wins our 
praise and the altruist our scorn. Juries misjudge defendants, voters misjudge candidates, 
lovers misjudge each other, and, as a consequence, the innocent are executed, the 
incompetent are elected, and the ignoble are embraced. In this article, we have examined 
one of the errors to which human beings are prone: the correspondence bias. We have 
argued that this tendency to draw logically unwarranted inferences about the dispositions 
of others can be caused by four distinct mechanisms, all of which fall out of a basic 
model of attributional process. We have tried to say what the correspondence bias is and 
how it comes to be. But the question we cannot answer is a pressing one: How prevalent 
is this bias in everyday life? Unfortunately, social psychology experiments are especially 
poor tools for answering questions about prevalence. This is unfortunate because it tends 
to make such matters the constant target of intuitive appeals. These appeals often take the 
form of suggesting that inferential errors cannot be pervasive or problematic because one 
need only "look around" to see that people navigate their social worlds with ease and 
aplomb. Every day, people meet people, make judgments, make friends, and conduct the 
other dull business of ordinary life, all without any obvious impairment. Can the 
correspondence bias be more than a hothouse phenomenon if people in the real world do 
just fine?  

As with most intuitive appeals, this one rests on a tenuous assumption, namely, that 
people do just fine. In the past year, 1,000 people who thought they knew their 
acquaintances have been raped by them, 10,000 people who thought they knew their 
mates have divorced them, and 100,000 people who thought they knew their sovereigns 
have died as pawns in their wars. Just how capably do we navigate our social worlds? 
Just how accurate are our understandings of those around us? We do not know. Nobody 

does. But before we accept the stale contention that people do just fine when 
psychologists are not manipulating and measuring them, we should probably look 
around.  

References 

 
Ajzen, I. & Fishbein, M. (1975). A Bayesian analysis of attribution processes.( 

Psychological Bulletin, 82, 

261—277.)  

Alloy, L. B. & Abramson, L. Y. (1979). Judgments of contingency in depressed and 
nondepressed students: Sadder but wiser?( 

Journal of Experimental Psychology: General, 

108, 

441—485.)  

Arkin, R. M. & Duval, S. (1975). Focus of attention and causal attributions of actors and 
observers.( 

Journal of Experimental Social Psychology, 11, 

427—438.)  

Bargh, J. A. (1989). Conditional automaticity: Varieties of automatic influence in social 
perception and cognition.(In J. S. Uleman & J. A. Bargh (Eds.), 

Unintendedthought 

(pp. 

3—51). New York: Guilford Press.)  
Baron, R. M. (1988). An ecological framework for establishing a dual-mode theory of 
social knowing.(In D. Bar-Tal & A. W. Kruglanski (Eds.), 

The social psychology of 

knowledge 

(pp. 48—82). Cambridge, England: Cambridge University Press.)  

Bassili, J. N. & Smith, M. C. (1986). On the spontaneity of trait attribution: Converging 
evidence for the role of cognitive strategy.( 

Journal of Personality and Social 

Psychology, 50, 

239—245.)  

Bem, D. J. (1972). Self-perception theory.(In L. Berkowitz (Ed.), 

Advances in 

experimental social psychology 

(Vol. 6, pp. 1—61). San Diego, CA: Academic Press.)  

Berscheid, E., Graziano, W., Monson, T. & Dermer, M. (1979). Outcome dependency: 
Attention, attribution, and attraction.( 

Journal of Personality and Social Psychology, 34, 

978—989.)  
Bierbrauer, G. (1979). Why did he do it? Attribution of obedience and the phenomenon 
of dispositional bias.( 

European Journal of Social Psychology, 9, 

67—84.)  

Buss, A. R. (1978). Causes and reasons in attribution theory: A conceptual critique.( 

Journal of Personality and Social Psychology, 11, 

1311—1321.)  

D'Agostino, P. R. & Fincher-Kiefer, R. (1992). Need for cognition and correspondence 
bias.( 

Social Cognition, 10, 

151—163.)  

Dunning, D., Griffin, D. W., Milojkovic, J. & Ross, L. (1990). The overconfidence effect 
in social prediction.( 

Journal of Personality and Social Psychology, 58, 

568—581.)  

Ellis, R. J. & Holmes, J. G. (1982). Focus of attention and self-evaluation in social 
interaction.( 

Journal of Personality and Social Psychology, 43, 

67—77.)  

Fein, S., Hilton, J. L. & Miller, D. T. (1990). Suspicion of ulterior motivation and the 
correspondence bias.( 

Journal of Personality and Social Psychology, 58, 

753—764.)  

Festinger, L. & Carlsmith, J. (1959). Cognitive consequences of forced compliance.( 

Journal of Abnormal and Social Psychology, 58, 

203—210.)  

Fischoff, B. (1975). Hindsight ?  foresight: The effects of outcome knowledge on 
judgments under uncertainty.( 

Journal of Experimental Psychology: Human Perception 

and Performance, 1, 

288—299.)  

Fiske, S. T. & Taylor, S. E. (1991). 

Social cognition 

(2nd ed.).(Reading, MA: Addison-

Wesley)  
Fodor, J. A. (1983). 

The modularity of mind. 

(Cambridge, MA: MIT Press)  

Funder, D. C. (1987). Errors and mistakes: Evaluating the accuracy of social judgment.( 

Psychological Bulletin, 101, 

75—90.)  

Gilbert, D. T. (1991). How mental systems believe.( 

American Psychologist, 46, 

107—

119.)  
Gilbert, D. T. & Jones, E. E. (1986). Perceiver-induced constraint: Interpretations of self-
generated reality.( 

Journal of Personality and Social Psychology, 50, 

269—280.)  

Gilbert, D. T., Krull, D. S. & Pelham, B. W. (1988). Of thoughts unspoken: Social 
inference and the self-regulation of behavior.( 

Journal of Personality and Social 

Psychology, 55, 

685—694.)  

Gilbert, D. T., McNulty, S. E., Giuliano, T. A. & Benson, J. E. (1992). Blurry words and 
fuzzy deeds: The attribution of obscure behavior.( 

Journal of Personality and Social 

Psychology, 62, 

18—25.)  

Gilbert, D. T. & Osborne, R. E. (1989). Thinking backward: Some curable and incurable 
consequences of cognitive busyness.( 

Journal of Personality and Social Psychology, 57, 

940—949.)  
Gilbert, D. T., Pelham, B. W. & Krull, D. S. (1988). On cognitive busyness: When person 
perceivers meet persons perceived.( 

Journal of Personality and Social Psychology, 54, 

733—740.)  
Gilbert, D. T. & Silvera, D. H. (1993). 

How to help people ruin their reputations. 

(Unpublished manuscript, University of Texas at Austin)  
Griffin, D. W., Dunning, D. & Ross, L. (1990). The role of construal processes in 
overconfident predictions about the self and others.( 

Journal of Personality and Social 

Psychology, 59, 

1128—1139.)  

Griffin, D. W. & Ross, L. (1991). Subjective construal, social inference, and human 
misunderstanding.(In M. Zanna (Ed.), 

Advances in experimental social psychology 

(Vol. 

24, pp. 319—356). San Diego, CA: Academic Press.)  
Heider, F. (1958). 

The psychology of interpersonal relations. 

(New York: Wiley)  

Higgins, E. T. & Winter, L. (1993). The "acquisition principle": How beliefs about a 
behavior's prolonged circumstances influence correspondent inference.( 

Personality and 

Social Psychology Bulletin, 19, 

605—619.)  

Hilton, D. J. & Slugoski, B. R. (1986). Knowledge-based causal attribution: The 
abnormal conditions focus model.( 

Psychological Review, 93, 

75—88.)  

Hinsz, V. B. & Tomhave, J. A. (1991). Smile and (half) the world smiles with yo u, frown 
and you frown alone.( 

Personality and Social Psychology Bulletin, 17, 

586—592.)  

Hogarth, R. M. (1981). Beyond discrete biases: Functional and dysfunctional aspects of 
judgmental heuristics.( 

Psychological Bulletin, 90, 

197—217.)  

Ichheiser, G. (1949). Misunderstandings in human relations: A study in false social 
perception.( 

American Journal of Sociology, 55. 

)  

Jacoby, L. L., Kelley, C. M. & Dywan, J. (1989). Memory attributions.(In H. L. Roediger 
& F. I. M. Craik (Eds.), 

Varieties of memory and consciousness: Essays in honor of 

Endel Tulving 

(pp. 391—422). Hillsdale, NJ: Erlbaum.)  

Jacoby, L. L., Toth, J. P., Lindsay, S. D. & Debner, J. A. (1992). Lectures for a 
layperson: Methods for revealing unconscious processes.(In R. Bornestein & T. Pittman 
(Eds.), 

Perception without awareness 

(pp. 81—120). New York: Guilford Press.)  

Jones, E. E. (1979). The rocky road from acts to dispositions.( 

American Psychologist, 

34, 

107—117.)  

Jones, E. E. (1990). 

Interpersonal perception. 

(New York: Macmillan)  

Jones, E. E. & Davis, K. E. (1965). From acts to dispositions: The attribution process in 
person perception.(In L. Berkowitz (Ed.), 

Advances in experimental social psychology 

(Vol. 2, pp. 219—266). San Diego, CA: Academic Press.)  
Jones, E. E. & Harris, V. A. (1967). The attribution of attitudes.( 

Journal of Experimental 

Social Psychology, 3, 

1—24.)  

Jones, E. E. & Nisbett, R. E. (1972). The actor and the observer: Divergent perceptions of 
the causes of behavior.(In E. E. Jones, D. E. Kanouse, H. H. Kelley, R. E. Nisbett, S. 
Valins, & B. Weiner (Eds.), 

Attribution: Perceiving the causes of behavior 

(pp. 79—94). 

Morristown, NJ: General Learning Press.)  
Kahneman, D. & Miller, D. T. (1986). Norm theory: Comparing reality to its 
alternatives.( 

Psychological Review, 93, 

136—153.)  

Keller, E. (1992). Nature, nurture, and the human genome project.(In L. Hood & D. J. 
Kevles (Eds.), 

The code of codes 

(pp. 253—288). Cambridge, MA: Harvard University 

Press.)  
Kelley, H. H. (1967). Attribution theory in social psychology.(In D. Levine (Ed.), 

Nebraska Symposium on Motivation 

(Vol. 15, pp. 192—238). Lincoln: University of 

Nebraska Press.)  
Keysar, B. (1993). 

A process model of linguistic perspective-taking: Toward a theory of 

language use. 

(Unpublished manuscript, University of Chicago)  

Keysar, B. (1994). The illusory transparency of intention: Perspective taking in text.( 

Cognitive Psychology, 26, 

165—208.)  

Kruglanski, A. W. (1975). The endogenous-exogenous partition in attribution theory.( 

Psychological Review, 82, 

387—406.)  

Kruglanski, A. W. (1989). The psychology of being "right:" The problem of accuracy in 
social perception and cognition.( 

Psychological Bulletin, 106, 

395—409.)  

Krull, D. S. (1993). Does the grist change the mill? The effect of the perceiver's 
inferential goal on the process of social inference.( 

Personality and Social Psychology 

Bulletin, 19, 

340—348.)  

Langer, E. J. (1975). The illusion of control.( 

Journal of Personality and Social 

Psychology, 32, 

311—328.)  

Lepper, M. R., Greene, D. & Nisbett, R. E. (1973). Undermining children's intrinsic 
interest with extrinsic rewards: A test of the "overjustification" hypothesis.( 

Journal of 

Personality and Social Psychology, 28, 

129—137.)  

Lewin, K. (1931). The conflict between Artistotelian and Galileian modes of thought in 
contemporary psychology.( 

Journal of General Psychology, 5, 

141—177.)  

Lombardi, W. J., Higgins, E. T. & Bargh, J. A. (1987). The role of consciousness in 
priming effects on categorization: Assimilation versus contrast as a function of awareness 
of the priming task.( 

Personality and Social Psychology Bulletin, 13, 

411—429.)  

Lorge, I. (1936). Prestige, suggestion, and attitudes.( 

Journal of Social Psychology, 7, 

386—402.)  
Lukes, S. (1973). 

Individualism. 

(Oxford, England: Basil Blackwell)  

Lupfer, M. B., Clark, L. F. & Hutcherson, H. W. (1988). Impact of context on 
spontaneous trait and situational attributions.( 

Journal of Personality and Social 

Psychology, 58, 

239—249.)  

Markus, H. R. & Kitayama, S. (1991). Culture and the self: Implications for cognition, 
emotion, and motivation.( 

Psychological Review, 98, 

224—253.)  

Martin, L. L., Seta, J. J. & Crelia, R. A. (1990). Assimilation and contrast as a function of 
people's willingness and ability to expend effort in forming an impression.( 

Journal of 

Personality and Social Psychology, 59, 

27—37.)  

McArthur, L. Z. & Baron, R. M. (1983). Toward an ecological theory of social 
perception.( 

Psychological Review, 90, 

215—238.)  

McArthur, L. Z. & Post, D. L. (1977). Figural emphasis and person perception.( 

Journal 

of Experimental Social Psychology, 13, 

520—535.)  

McGill, A. L. (1989). Context effects in judgments of causation.( 

Journal of Personality 

and Social Psychology, 57, 

189—200.)  

Medcof, J. W. (1990). PEAT: An integrative model of attribution processes.(In M. P. 
Zanna (Ed.), 

Advances in experimental social psychology 

(Vol. 23, pp. 111—209). New 

York: Academic Press/Harcourt Brace Jovanovich Publishers.)  
Milgram, S., Bickman, L. & Berkowitz, O. (1969). Note on the drawing power of crowds 
of different size.( 

Journal of Personality and Social Psychology, 13, 

79—82.)  

Miller, A. G., Ashton, W. & Mishal, M. (1990). Beliefs concerning the features of 
constrained behavior: A basis for the fundamental attribution error.( 

Journal of 

Personality and Social Psychology, 59, 

635—650.)  

Miller, D. T., Norman, S. A. & Wright, E. (1978). Distortion in person perception as a 
consequence of the need for effective control.( 

Journal of Personality and Social 

Psychology, 36, 

598—607.)  

Miller, F. D., Smith, E. R. & Uleman, J. (1981). Measurement and interpretation of 
situational and dispositional attributions.( 

Journal of Experimental Social Psychology, 17, 

80—95.)  
Miller, J. G. (1984). Culture and the development of everyday social explanation.( 

Journal of Personality and Social Psychology, 46, 

961—978.)  

Mook, D. G. (1983). In defense of external invalidity.( 

American Psychologist, 38, 

379—

387.)  
Newman, L. S. (1991). Why are traits inferred spontaneously? A developmental 
approach.( 

Social Cognition, 9, 

221—253.)  

Newman, L. S. (1993). How individualists interpret behavior: Idiocentrism and 
spontaneous trait inference.( 

Social Cognition, 11, 

243—269.)  

Nietzche, F. (1984). 

Human all too human 

(M. Faber & S. Lehmann, Trans.).(Lincoln: 

University of Nebraska Press. (Original work published 1886)  
Nisbett, R. E. (1987). Lay personality theory: Its nature, origin, and utility.(In N. E. 
Grunberg, R. E. Nisbett, J. Rodin, & J. E. Singer (Eds.), 

Adistinctive approach to 

psychological research: The influence of Stanley Schachter 

(pp. 87—117). Hillsdale, NJ: 

Erlbaum.)  
Nisbett, R. E. & Ross, L. (1980). 

Human inference: Strategies and shortcomings of social 

judgment. 

(Englewood Cliffs, NJ: Prentice Hall)  

Osborne, R. E. & Gilbert, D. T. (1992). The preoccupational hazards of social life.( 

Journal of Personality and Social Psychology, 62, 

219—228.)  

Pittman, T. S. & D'Agostino, P. R. (1985). Motivation and attribution: The effects of 
control deprivation on subsequent information-processing.(In J. Harvey & G. Weary 

(Eds.), 

Attribution: Basic issues and applications 

(pp. 117—141). San Diego, CA: 

Academic Press.)  
Quattrone, G. A. (1982a). Behavioral consequences of attributional bias.( 

Social 

Cognition, 1, 

358—378.)  

Quattrone, G. A. (1982b). Overattribution and unit formation: When behavior engulfs the 
person.( 

Journal of Personality and Social Psychology, 42, 

593—607.)  

Reeder, G. D. & Brewer, M. B. (1979). A schematic model of dispositional attribution in 
interpersonal perception.( 

Psychological Review, 86, 

61—79.)  

Reeder, G. D., Fletcher, G. J. O. & Furman, K. (1989). The role of observers' 
expectations in attitude attribution.( 

Journal of Experimental Social Psychology, 25, 

168—188.)  
Ross, L. (1977). The intuitive psychologist and his shortcomings.(In L. Berkowitz (Ed.), 

Advances in experimental social psychology 

(Vol. 10, pp. 173—220). San Diego, CA: 

Academic Press.)  
Ross, L., Amabile, T. M. & Steinmetz, J. L. (1977). Social roles, social control, and 
biases in social-perception processes.( 

Journal of Personality and Social Psychology, 35, 

485—494.)  
Ross, L., Greene, D. & House, P. (1977). The "false consensus effect": An egocentric 
bias in social perception and attribution processes.( 

Journal of Experimental Social 

Psychology, 13, 

279—301.)  

Ross, L. & Nisbett, R. (1991). 

The person and the situation: Perspectives of social 

psychology. 

(New York: McGraw-Hill)  

Schul, Y. & Burnstein, E. (1985). When discounting fails: Conditions under which 
individuals use discredited information in making a judgment.( 

Journal of Personality 

and Social Psychology, 49, 

894—903.)  

Schweder, R. A. & Bourne, E. (1982). Does the concept of the person vary cross-
culturally?(In A. J. Marsella & G. White (Eds.), 

Cultural conceptions of mental health 

and therapy 

(pp. 97—137). Boston: Reidel.)  

Sherman, S. J. (1980). On the self-erasing nature of errors of prediction.( 

Journal of 

Personality and Social Psychology, 39, 

211—221.)  

Skinner, B. F. (1971). 

Beyond freedom and dignity. 

(New York: Knopf)  

Snyder, M. L. & Frankel, A. (1976). Observer bias: A stringent test of behavior engulfing 
the field.( 

Journal of Personality and Social Psychology, 34, 

857—864.)  

Snyder, M. & Ickes, W. (1985). Personality and social behavior.(In G. Lindzey & E. 
Aronson (Eds.), 

The handbook of social psychology 

(3rd ed., Vol. 2, pp. 883—947). 

Hillsdale, NJ: Erlbaum.)  
Solomon, S. (1978). Measuring dispositional and situational attributions.( 

Personality 

and Social Psychology Bulletin, 4, 

589—594.)  

Spence, J. T. (1985). Achievement American style: The rewards and costs of 
individualism.( 

American Psychologist, 40, 

1285—1295.)  

Stavrianos, L. S. (1989). 

Lifelines from our past: A new world history. 

(New York: 

Pantheon Books)  
Storms, M. D. (1973). Videotape and the attribution process: Reversing actors' and 
observers' points of view.( 

Journal of Personality and Social Psychology, 27, 

165—175.)  

Strickland, L. H. (1958). Surveillance and trust.( 

Journal of Personality, 26, 

200—215.)  

Swann, W. B. (1984). Quest for accuracy in person perception: A matter of pragmatics.( 

Psychological Review, 91, 

457—477.)  

Taylor, S. E. & Brown, J. D. (1988). Illusion and well-being: A social-psychological 
perspective on mental health.( 

Psychological Bulletin, 103, 

193—210.)  

Taylor, S. E. & Fiske, S. T. (1975). Point-of-view and perceptions of causality.( 

Journal 

of Personality and Social Psychology, 32, 

439—445.)  

Taylor, S. E. & Fiske, S. T. (1978). Salience, attention, and attribution: Top of the head 
phenomena.(In L. Berkowitz (Ed.), 

Advances in experimental social psychology 

(Vol. 11, 

pp. 249—288). San Diego, CA: Academic Press.)  
Trope, Y. (1986). Identification and inferential processes in dispositional attribution.( 

Psychological Review, 93, 

239—257.)  

Trope, Y. & Cohen, O. (1989). Perceptual and inferential determinants of behavior-
correspondent attributions.( 

Journal of Experimental Social Psychology, 25, 

142—158.)  

Trope, Y., Cohen, O. & Maoz, Y. (1988). The perceptual and inferential effects of 
situational inducements on dispositional attributions.( 

Journal of Personality and Social 

Psychology, 55, 

165—177.)  

Tversky, A. & Kahneman, D. (1973). Availability: A heuristic for judging frequency and 
probability.( 

Cognitive Psychology, 5, 

207—232.)  

Uleman, J. S. (1987). Consciousness and control: The case of spontaneous trait 
inferences.( 

Personality and Social Psychology Bulletin, 13, 

337—354.)  

Wachtel, P. (1973). Psychodynamics, behavior therapy, and the implacable experimenter: 
An inquiry into the consistency of personality.( 

Journal of Abnormal Psychology, 82, 

324—334.)  
Walden, T. A. & Ogan, T. A. (1988). The development of social referencing.( 

Child 

Development, 59, 

1230—1240.)  

Weber, M. (1930). 

The Protestant ethic and the spirit of capitalism. 

(Winchester, MA: 

Allen & Unwin)  
Weiner, B., Frieze, I., Kukla, A., Reed, L., Rest, S. & Rosenbaum, R. M. (1972). 
Perceiving the causes of success and failure.(In E. E. Jones, D. E. Kanouse, H. H. Kelley, 
R. E. Nisbett, S. Valins, & B. Weiner (Eds.), 

Attribution: Perceiving the causes of 

behavior 

(pp. 95—120) Morristown, NJ: General Learning Press.)  

Wells, H. G. (1932). 

The work, wealth, and happiness of mankind. 

(London: William 

Heinemann)  
White, P. A. (1991). Ambiguity in the internal/external distinction in causal attribution.( 

Journal of Experimental Social Psychology, 27, 

259—270.)  

Whitney, P., Waring, D. A. & Zingmark, B. (1992). Task effects on the spontaneous 
activation of trait concepts.( 

Social Cognition, 10, 

377—396.)  

Wilson, T. D. & Brekke, N. (1994). Mental contamination and mental correction: 
Unwanted influences on judgments and evaluations.( 

Psychological Bulletin, 116, 

117—

142.)  
Winter, L. & Uleman, J. S. (1984). When are social judgments made? Evidence for the 
spontaneousness of trait inferences.( 

Journal of Personality and Social Psychology, 47, 

237—252.)  
Winter, L., Uleman, J. S. & Cunniff, C. (1985). How automatic are social judgments?( 

Journal of Personality and Social Psychology, 49, 

904—917.)  

Wyer, R. S. & Unverzagt, W. H. (1985). The effects of instructions to disregard 

information on its subsequent recall and use in making judgments.( 

Journal of 

Personality and Social Psychology, 48, 

533—549.)  

 

1 

 

Some psychologists (e.g., 

Funder, 1987 

; 

Kruglanski, 1989 

; 

Swann, 1984 

) have 

questioned the status of logical errors, and they have argued that inferential departures 
from rational baseline models are not necessarily damaging to the organism and, 
furthermore, that experimental procedures that claim to document these so-called "errors" 
do not always give subjects an adequate opportunity to avoid them. We do not assume 
that errors of judgment necessarily lead to negative consequences (indeed, we argue later 
that quite the opposite is sometimes the case), nor do we assume that such errors are 
necessarily the "fault" of the experimental subject. However, we do continue the tradition 
of referring to departures from rational baselines as errors, regardless of their causes or 
consequeces.  

 

2 

 

Just as the results of this study were not due to lack of awareness, subsequent studies 
have shown that they were also not due to inflated categorizations (see 

Gilbert, McNulty, 

Giuliano, & Benson, 1992 

; 

Gilbert & Osborne, 1989 

).  

 

3 

 

The person—situation measure was state of the art when these investigators used it, and 
each of these studies makes an extremely important contribution to the attribution 
literature. We have used direct quotes from these studies to emphasize the fact that none 
of the investigators misstated or misinterpreted their own results. All were extremely 
clear about what they did and did not find.  

 

4 

 

It is interesting to note that when omnipresent constraints create dispositions, they may 
eventually become superfluous constraints. If a small child is coerced into brushing her 
teeth before bed, the habit may become so well ingrained by middle childhood that it 
does not require the threat of punishment that still happens to persist. The 
correspondence-biased observer who ignores that superfluous threat and concludes that 
the older child is dispositionally compelled to brush her teeth will, in fact, be right. 
Ironically, a situational constraint will have endured long enough  to warrant being 
ignored.  

Figure 1. The sequence of events that occurs when an attribution is made.  

 

 

 
 
Figure 2. An expanded version of the sequence of events that occurs when an attribution 
is made.