U1 TFROM ACTS TO DISPOSITIONS

h e Attribution Process in

I Person Perception'

Edward E. Jones and Keith E. Davis

DEPARTMENT OF PSYCHOLOGY DEPARTMENT OF PSYCHOLOGY
DUKE UNIVERSITY UNIVERSITY OF COLORADO
DURHAM, NORTH CAROLINA BOULDER, COLORADO

I. The Naive Explanation of Human Actions: Explanations by Attributing

Intentions . . . . . . . . . . . . . . . . . . 220
11. A Theory of Correspondent Inferences . . . . . . . . . . 222
A. The Concept of Correspondence . . . . . . . . . . 223
B. Acts and Their Effects . . . . . . . . . . . . 225
C. The Assumed Desirability of Effects . . . . . . . . . 226
D. The Determinants of Correspondence . . . . . . . . 227
E. The Calculation of Commonality . . . . . . . . . . 230
F. Conditions Affecting Desirability Assumptions . . . . . . . 232
G. Information Gain and the Role of Prior Choice . . . . . . 233
H. Correspondence, Choice, and Role Assignment . . . . . . 234
111. Personal Involvement and Correspondence . . . . . . . . . 237
A. The Hedonic Relevance of the Action to the Perceiver . . . . 237
B. Personalism: The Actor's Intention to Benefit or Harm the Perceiver 246
IV. Summary and Conclusions . . . . . . . . . . . . . 263
References . . . . . . . . . . . . . . . . . . 266

Many social psychologists have expressed a central interest in the

ties between person perception and interpersonal behavior. The writings of
Fritz Heider have exerted a predominant and continuing influence on
research designed to illuminate these ties. From his 1944 paper on
phenomenal causality to his more recent (1958) book on The PsychoZogy
of Interpersonal Relations, Heider has persistently concerned himself with
the cognitive aspects of social interaction. His writings are especially im-
portant for recognizing and identifying the major problems with which
'Much of the research reported herein was supported by National Science
Foundation Grants 8857 and 21955 to the first author.
219
220 EDWARD E. JONES AND KEITH E. DAVIS
any theory of person perception must contend: causal attribution, cogni-
tion-sentiment relations, taking the other’s perspective, and so on. Heider’s
comments are comprehensive, perceptive, and provocative. His exposition
does not lend itself readily, however, to the formulation of interrelated
propositional statements. Thus the research which has been done to date
is largely demonstrational in significance and dismayingly sparse in
quantity. While the studies which we intend to review in this chapter
may be seen as islands in the same phenomenological sea, it is not very
clear how one navigates between them.
We believe that the kind of systematic, conceptual structure that
is needed must involve an analysis of phenomenal causality, or the determi-
nants and consequences of attributing causation for particular actions.
In the central portion of this chapter we shall attempt to review, and
to some extent reformulate, much of the recent research concerning phe-
nomenal causality and the attribution of intentions. Our first task, how-
ever, is to introduce the notion of explaining an action by assigning an
intention and to set the stage for the theory of inference which follows.
I. The Naive Explanation of Human Actions:
Explanation by Attributing Intentions
At the he,art of Heider’s analysis of naive or “common sense psycho-
logy” is the distinction between personal and impersonal causality. We
assume that the person-perceiver’s fundamental task is to interpret or
infer the causal antecedents of action. The perceiver seeks to find suficient
reason why the person acted and why the act took on a particular form.
Instead of the potentially iniinite regress of cause and effect which char-
acterizes an impersonal, scientific analysis, the perceiver’s explanation
comes to a stop when an intention or motive is assigned that has the
quality of being reason enough. “He eats because he is hungry” would
not ordinarily bring a request for further explanation. After all, eating
is something one would do if one were hungry.
The cognitive task of establishing sufficient reason for an action
involves processing available information about, or making assumptions
about, the links between stable individual dispositions and observed ac-
tion. Let us start with the case in which a perceiver observes an action
and at least some of its effects. His basic problem as a perceiver is
to decide which of these effects, if any, were intended by the actor.
Let us first address ourselves to the problem of “if any.’’ In order to
conclude that at least some of the effects achieved by an action were
intended, the perceiver must first believe that the actor was aware his
action would have the observed effects. Thus a first condition in the
inference process is the assumption of knowledge on the part of the
 FROM ACTS TO DISPOSITIONS 221
actor. Consequences of an action which the actor could not have forseen
do not qualify as candidates for what he was trying to achieve. The
condition of knowledge is of critical importance within our Iegal system
where it is customary to distinguish among levels of responsibility for a
crime: (1) intentional (P did X to enjoy the immediate effects of XI,
(2) incidental (P did X as a means of getting to Y), and (3) accidental
(Xwas a consequence of P’s action that he neither intended nor expected).
In addition to assumptions about knowledge of consequences, de-
cisions linking intentional attributes to the effects of action are also affected
by the perceiver’s judgments of the actor’s ability to bring about the
effects observed. Simply put, an actor cannot achieve his objectives solely
by desiring to achieve them. He must have the capacities or skill to
move from his present condition of desire to a subsequent condition
of attainment and satisfaction. When a person’s actions have certain conse-
quences, it is important for the perceiver to determine whether the person
was capable of producing these consequences in response to his intentions.
Especially in the case where an actor fairs to produce certain effects
that might have been anticipated by the perceiver, there may be ambiguity
as to whether the actor did not want to produce the effects, or wanted
but was not able to.
Even when effects are achieved, however, the perceiver may have
the problem of assessing the relative contribution of luck or chance.
When a novice archer hits the bull’s eye, we are more apt to attribute
this to luck than to skill. There are other occasions when we do not
assign intentions to correspond with effects achieved because we do
not consider the actor capable of producing those effects at will. A
jury is more likely to believe that a killing is accidental if the average
person would have lacked the skill (the marksmanship, the strength, etc.)
to bring about the crime deliberately. It was quite possible to believe
that Oswald intended to kill President Kennedy and not Mrs. Kennedy
or a secret service man, because he was known to be an expert marksman.
For a further discussion of the problems involved in judging ability relative
to difficulty and luck, the reader is referred to Heider (1958).
The perceiver may have certain information about knowledge and
ability (he may be informed that Oswald knew the gun was loaded and
that Oswald often practiced on a local rifle range), or he may merely
assume that knowledge and ability were probably present or probably
absent. Whether the perceiver’s conclusion about such matters is correct
or incorrect, the conclusion obviously will affect his decisions about the
actor’s intentions in the situation. Knowledge and ability are preconditions
for the assignment of intentions. Each plays a similar role in enabling
the perceiver to decide whether an effect or consequence of action was
222 EDWARD E. JONES AND KEITH E. DAVIS
accidental. The assignment of intention, in turn, is a precondition for
inferences concerning those underlying stable characteristics toward which
the perceiver presses in attaching significance to action. As Heider (1958)
argued, the perceiver ordinarily strives to discover the invariances which
underlie manifest actions in order to stabilize the environment and render
it more predictable.
We may attempt to summarize the foregoing remarks by the diagram
presented as Fig. 1. It is assumed that the perceiver typically starts with
the overt action of another; this is the grist for his cognitive mill. He
then makes certain decisions concerning ability and knowledge which
will let him cope with the problem of attributing particular intentions
to the actor. The attribution of intentions, in turn, is a necessary step
in the assignment of more stable characteristics to the actor.

INFERRED OBSERVED

Disposition [ -Intention ~ o wAbility

l e ' g ~ * ~ ~ i o ~ c - :
Effecc

FIG. 1. The action-attribute paradigm.

Fig. 1 attempts to clarify the circumstances under which any inten-

tions will be assigned to explain action. But we now seek to extend
the analysis in order to account for the attribution of particular intentions
and dispositions on the basis of particular actions. We shall here largely
ignore the problems involved in imputing knowledge and ability and con-
centrate on specific linkages between effects achieved and intentions re-
vealed. We assume that those consequences of action obviously neither
intended by the actor nor within the range of his capabilities will be
considered irrelevant by the perceiver.
11. A Theory of Correspondent Inferences
Our purpose is to construct a theory which systematically accounts
for a perceiver's inferences about what an actor was trying to achieve
by a particular action. In achieving this purpose we view the action
as occurring within a particular situational context which defines, in large
part, its meaning for the perceiver. In particular, as we shall attempt
to show in greater detail below, the meaning of an action-its intentional
significance-derives from some consideration of the alternative action
 FROM ACTS TO DISPOSITIONS 223
possibilities available to but foregone by the actor. As perceivers of action,
we can only begin to understand the motives prompting an act if we
view the effects of the act in the framework of effects that other actions
could have achieved.
Perhaps an example will further clarify our purpose and approach.
Let us imagine ourselves as silent observers of an interaction episode
in which A and B are working together on a task. We observe that
A gives orders to B, monitors his performance, and shows his displeasure
with the quality and quantity of B’s work. The inferences about A we
would most likely draw from this episode would depend critically on
the action alternatives seen to be available to him. If A and B had
come together in a free situation, we would be inclined to see A as
quite arrogant and domineering. If we were informed that A had been
given instructions to take a directive leadership role, we would be less
likely to regard his dominating behavior as an indication of his personal
qualities: that is, our inferences about dominance from his action would
be much less correspondent.
Such role-playing instructions presumably limit A’s freedom to be-
have in a “revealing” way, that is, in a way which is characteristic of
him rerative to others. The theory which follows attempts to imbed this
consideration of perceived freedom of choice in a systematic framework.
We will attempt to extract conceptual commonalities from empirical situa-
tions involving different varieties of environmental constraint. Our ap-
proach is to cast these “conceptual commonalities” in a form which is
amenable to cumulative experimental research.
A. THE OF CORRESPONDENCE
CONCEPT
When the perceiver infers personal characteristics as a way of ac-
counting for action, these personal characteristics may vary in the degree
to which they correspond with the behavior they are intended to explain.
Correspondence refers to the extent that the act and the underlying char-
acteristic or attribute are similarly described by the inference. In the
example provided in the preceding section, the most correspondent in-
ference is that which assumes with high confidence that domineering be-
havior is a direct reflection of the person’s intention to dominate, which
in turn reflects a disposition to be dominant. Thus, to anticipate the
broad outlines of the theory to come, correspondence of inference declines
as the action to be accounted for appears to be constrained by the setting
in which it occurs.
To say that a person is dominant is to say that he is disposed
to behave in a dominant fashion in a variety of settings, Of course,
the perceiver in the above example would not infer such a dominance
224 EDWARD E. JONES AND KEITH E. DAVIS

disposition if he had not first inferred an intention to dominate. The

actor’s intention may or may not be conscious and deliberate, but it
is marked by some aspect of desire or volition which comes from the
person and is not predetermined by environmental forces. Our theory
assumes, in using the two concepts of intention and disposition, that
correspondence declines as the perceiver moves from inferring intentions
to more elaborate inferences about dispositional structures. If the per-
ceiver, having observed a single action, infers intention X with moderate
confidence, he cannot be more confident in inferring the underlying dis-
position X’ from the intention X. This would appear to be so because
intentions are the data for inferring dispositions, and because an intention
may reflect any of several dispositions.
Hopefully, the foregoing discussion has given the reader some general
feeling for the meaning of correspondence in the present context. For
the sake of theoretical clarity, however, more precise and formal explica-
tion is in order. Such an explication may provide a clearer path toward
understanding the theory.
All actions have effects on the environment. From the perceiver’s
point of view, any effect of another person’s action is a potential reason
why this person had engaged in that action. To infer that the action
occurred for X reason is to specify the actor’s intention and, indirectly,
an underlying disposition. Both intentions and dispositions are attributes
of the person. The perception of a link between a particular intention
or disposition and a particular action may therefore be called an attribute-
effect linkage.
Let us now attempt a more formal definition of correspondence.
Given an attribute-eflect linkage which is oflered to explain why an act
occurred, correspondence increases as the judged value of the attribute
departs from the judge’s conception of the average person’s standing on
that attribute. Turning to the illustration used earlier, the inference that
domineering action reflects an underlying trait of dominance is corre-
spondent to the extent that the actor’s dominance is seen as greater
than that of the average person. This implies, incidentally, that the in-
tention to dominate is out of the ordinary-somehow more intense and
noteworthy than we would normally expect.
As a simple example of how the concept of correspondence can
be put to use in a research setting, we may provide the perceiver with
rating scales designed to measure the strength of the trait attributed to
the actor and his confidence in making his rating. The perceiver’s certainty
that the actor is extreme on a trait which provides sufficient reason for
the action’s occurrence is, then, the level of correspondence of his
inference.
 FROM ACTS TO DISPOSITIONS 225

B. ACTSAND THEIREFFECTS
An act is conceived of as a molar response which reflects some
degree of personal choice on the part of the actor (if only between
action and inaction, though more typically between alternative courses
of action) and which has one or more effects on the environment or
the actor himself. Eflecrs are distinctive (or potentially distinctive) conse-
quences of action. Stated in the broadest terms, they are discriminable
changes in the pre-existing state of affairs that are brought about by
action. Delimiting the unit with which we shall be concerned is more
a problem in theory than in practice. If we observe that a man leaves
his chair, crosses the room, closes the door, and the room becomes less
noisy, a correspondent inference would be that he intended to cut down
the noise. One might ask whether the inference that the man intended
to reach the door is not also a correspondent inference since “reaching
the door” is an effect of crossing the room. But the subordinate parts
of a meaningful action sequence do not have to be confused with the
effects of an action. In this case, the perceiver is likely to “organize”
the action in his mind as beginning with the decision to leave the chair
and ending with the closing of the door. It is the effects of the terminal
act in a meaningful sequence, then, that provide the grist for our theory.
An act may have only one effect, but usually has multiple effects.
When the man closes the door this may reduce the draft, reduce the
illumination in his office, and make two students talking in the hall feel
a little guilty for interrupting his work. Thus, we are usually dealing
with choice areas rather than single choices. Important implications for
the theory are contained in the fact that “the bitter often comes with
the sweet”-an action may be performed to achieve effect x, but effects
m, p, t, and z are inextricably produced by the act as well. A choice
between two choice areas, then, is a choice between two multiple-effect
clusters. The multiple effects in one cluster may or may not overlap
extensively with the multiple effects in the second cluster. That is, certain
effects may be common to the chosen alternative and to the nonchosen
alternative.
If the promising young psychologist Dr. Smedley accepts a position
at Harvard rather than Yale, the following effects are obviously common
to these two areas of choice: being in the Ivy League, living in New
England, joining a university with high prestige and good salaries, living
near the sea coast, etc. The theory assumes, then, that these common
effects could not have been decisive in the choice, and thus do not provide
information which could contribute to correspondent inferences. There
are also, of course, distinctive differences between the setting at Yale
226 EDWARD E. JONES AND KEITH E. DAVIS

and Harvard-especially if the perceiver were intimately knowledgeable

about the psychology departments of the two institutions-and the per-
ceiver’s cognitive accounting of these differences would be the critical
determinant of whatever inference was made.
For convenience in representing the structure of the situation in
which action occurs, we shall from time to time diagram each perceived
choice area as a circle within which the effects of the choice expressed
as alphabetical letters may be circumscribed. Common effects may then
be represented by the appearance of the same alphabetical letters in
different “choice circles.” Our hypothetical example of Dr. Smedley’s
dilemma might be diagrammed as in Fig. 2.

Dr. Smedley

Harvard
/ \Yale

a. Ivy League
b. New England
c. prestige
d. good salary
e. close to New York plays
f. emphasis on interdisciplinary research
g. emphasis on experimental approaches
to learning, etc.
Fro. 2. Smedley’s choice.

C. THEASSUMEDDESIRABILITY OF EFFECTS
As the perceiver considers the multiple effects of action, he will
usually assume that some of the effects were more desirable to the actor,
and therefore more diagnostic of his intentions than others. In fact, it
is almost always the case that some of the effects of the chosen alternative
action are assumed to be undesirable to the actor and some of the effects
of nonchosen action are assumed to be desirable. The two major effects
of a man’s buying a car, for example, are the acquisition of an automobile
and the incurring of a substantial debt. The average perceiver, given
evidence of such a purchase, will probably assume that the individual
desired the car so much that he was willing to go into debt for it,
not that he was willing to accept the burden of an automobile for the
privilege of being a debtor.
 FROM ACTS TO DISPOSITIONS 227
These assumptions by the perceiver tend to operate as hypotheses
which bias the inference process. Thus upon observing that an action
leads to a normally desirable effect, the perceiver usually will believe that
most persons, including the present actor, find that effect desirable. The
achievement of this effect will therefore be regarded as the actor’s most
likely intention. The perceiver may, of course, be wrong in his assumptions
about people in general. This particular actor may have intended to pro-
duce effects in the choice area that most people would be indifferent about
or even feel negatively toward. Thus, cultural assumptions or social
stereotypes may obscure the true significance of an action.
Let us take a closer look at the consequences for the inference
process of assumptions about the desirability of effects. The first step
is to distinguish clearly between effects which are assumed to be desirable
and those assumed to be undesirable consequences for the average actor.
Unless he has evidence to the contrary, the perceiver will assume that
the actor has acted in spite of, rather than because of, any negative
effects in the choice area. We may go beyond this to assert that any
effects in the choice area which are not assumed to be negative will
take on greater importance the more negative the remaining effects. Infer-
ences concerning the intention to achieve desirable effects will increase
in correspondence to the extent that costs are incurred, pain is endured,
or in general, negative outcomes are involved.
Within the range of the supposedly desirable consequences, we still
must recognize that effects assumed to be highly desirable are more likely
to enter into attribute-effect linkages than effects assumed to be variable
or neutral in desirability. However, it is also clear that attribute-effect
linkages based on universally desired effects are not informative concerning
the unique characteristics of the actor. To learn that a man makes the
conventional choice is to learn only that he is like most other men.
By the definition of correspondence stated above, an inference must char-
acterize the actor’s standing as high or low on an attribute relative to
the average person, in order to qualify as correspondent. If a choice
is explained on the basis of effects in the choice area which anyone
would like to produce and enjoy, an attribute inferred to account for
that choice will be low in correspondence. In general, we learn more
about uniquely identifying intentions and dispositions when the effects
of a chosen action are no more universally desired than the effects of
a nonchosen action.
D. THEDETERMINANTS OF CORRESPONDENCE
It may be helpful to divide the inference process into two aspects
(which may in fact be seen as stages in the process). Given an act
228 EDWARD E. JONES AND KEITH E. DAVIS

which leads to multiple effects, the perceiver implicitly attaches a proba-

bility value to each effect as a candidate for launching the inference
process. In other words, the perceiver assumes that certain of the effects
achieved are more likely to have been the goal of action than others.
In certain cases the probability of all effects but one may be zero, indi-
cating extreme confidence in attributing causation to a particular intention.
In other cases, the probability vaIues may be distributed among a range
of effects, and since it is possible that the target person acted for more
than one reason the probabilities may add up to more than 1.00. The
probability value for any given effect should vary directly as a function
of the assumed desirability of the effect and inversely as a function of
the number of other effects competing for the perceiver’s attention. If the
perceiver is asked, then, “what was A trying to accomplish?”---his response
should reflect this combination of assumed desirability and the number
of noncommon effects.
The second aspect or phase is to attach personal significance to
the effect or effects singled out as most probable courses of action. In
short, what does the action reveal about this particular actor that sets
him apart from other actors? By our definition of correspondence, relative
extremity of perceiver rating is the crucial measure. Here it seems quite
clear that assumed desirability and the number of noncommon effects
have conflicting implications. The greater the assumed desirability of the
effect in question, the less warrant there is for ratings of relative extremity.
The smaller the number of effects in contention, on the other hand,
the greater the warrant for extreme ratings. Assumed desirability, then,
positively affects the probability of an attribute-effect linkage being chosen
to begin the inference process and negatively affects the tendency to
assign extreme values to personal characteristics.
The correspondence of an inference, which should directly reflect
the amount of information revealed by an action, is thus a function
of the two conditions covered by the following explicit formulation:
An inference from action to attribute is correspondent as an inverse
function of ( a ) the number of noncommolt effects following the action,
and ( b ) the assumed sociat desirability of these eflects. This relationship
may be stated in simpler terms as a near tautology: the more distinctive
reasons a person has for an action, and the more these reasons are
widely shared in the culture, the less informative that action is concerning
the identifying attributes of the person.
It should be reiterated that correspondence has nothing to do, neces-
sarily, with the accuracy of the inference. The actor may have had no
intention of producing an effect which is seen by the perceiver as a
prominent consequence of his action. Being able to predict the effects
 FROM ACTS TO DISPOSITIONS 229
of one’s own actions is an important precondition for being accurately
perceived by others. The theory does not assume, then, that the perceiver
and the actor agree on the effects of the latter’s action, but focuses
on those effects of which the perceiver assumes the actor was aware.
The knowledge portion of the attribute-action paradigm (see Fig. 1)
is relevant only in determining which effects the perceiver will include
in the choice circle (see Fig. 2).
There is an interesting relationship between imputations of knowl-
edge and actions leading to socially undesirable effects. We have aIready
noted that when there are both desirable and undesirable effects of an
action, more significance is attached to the desirable effects the more
numerous and distasteful the undesirable effects “incurred.” In a case
where the actor produces solely negative effects, however, the situation

ASSUMED DESIRABILITY

High how
Trivial Intriguing
NUMBER OF
NONCOMMON
EFFECTS

FIG.3. Effect desirability and effect commonality as determinants of correspondence.

is quite different. Here the perceiver has two obvious options: He

may decide that the actor is truly a deviant type, that he desires those
goals which are shunned by others; or he may decide that the actor
was unaware of the effects’.ofhis action. It seems reasonable to pro-
pose, therefore, that in cases where the action-choice circle contains
only effects judged to be socially undesirable, the more undesirable
these effects the greater the perceiver’s tendency to impute ignorance
or lack of awareness to the actor. The possibility of imputing ignorance
thus sets a limit on the degree of correspondence predicted when actions
lead only to socially undersirable effects. (Note that this is quite dif-
ferent from what one would predict in the case where an action leads to
effects which are undesirable for the perceiver but recognized by him to
serve some purpose of the actor. We shall discuss this latter case below
under the heading of Hedonic Relevance.)
It may be helpful in summarizing the theory to consider the joint
operation of effect desirability and effect commonality as determinants
of correspondence, and to do so in the framework of a fourfold table.
Such a table .is presented in Fig. 3. As the figure shows, actions which
230 EDWARD E. JONES AND KEITH B. DAVIS

lead to effects deemed highly desirable to most persons cannot help but
be trivial from an informational point of view. Also, when the number
of noncommon effects is high, the perceiver cannot escape from the am-
biguity of his data in making inferences either to common or idiosyncratic
personal characteristics. In line with the stated theoretical relationship,
the high correspondence cell is that in which assumed desirability and
the number of noncommon effects are both low.
E. THE CALCULATION OF COMMONALITY
Since the theoretical statement refers to the number of noncommon
effects in the chosen alternative, it is important to clarify how effects
are identified and their commonality assessed. There are two rather differ-
ent clusters of problems involved here. One cluster concerns identifying
effects and determining whether or not two effects so identified are to
be viewed as common. This will simply have to be a matter of cumulative
research experience in fractionating and labeling consequences.
The other cluster of problems assumes that we can somehow achieve
the necessary identifications and concerns the method of determining how
the noncommon effects in the chosen alternative will be sorted out from
all others in order to generate reasonably precise predictions. There are
essentialIy three steps involved in sorting out the number of noncommon
effects. The first is for the experimenter to lay out (for example, in
the graphic manner of Fig. 2) the different action alternatives that the
perceiver is likely to envision in the actor’s situation. For each of these
alternatives there will be a choice area, a circle containing the distinctive
effects of that action. Having identified these circles and the effects they
circumscribe, the second step is to pool all the effects associated with
the nonchosen alternatives and to compare them with the effects in the
choice circle of the chosen alternative. Having done this, one follows
through on the assumption that effects which appear in both chosen and
nonchosen alternatives cannot serve as a basis for inferences about per-
sonal dispositions. One is then left with effects which appear only in
the chosen alternative and only in the nonchosen alternative(s). The
third and final step is to view the noncommon effects in the nonchosen
alternatives as effects which the actor may be trying to avoid. They
may thus be transposed, with their sign reversed, and regrouped among
the noncommon effects of the chosen alternative. The new total of non-
common effects now serves as the basis for making predictions about
the level of correspondence.
In order to illustrate these three stages in the calculation of non-
common effects, let us invent a case involving the choice of a marital
partner. We are informed that a Miss Adams, who is ripe for matrimony,
 FROM ACTS TO DISPOSITIONS 23 1
has received proposals from three suitors, Bagby, Caldwell, and Dexter.
All we know about Bagby is that he is wealthy, he has high status
in the community, and he is physically attractive. Caldwell is also quite
wealthy, he is physically attractive, and he has often expressed his longing
for a houseful of his own children. Dexter is physically attractive like

Effects of marriage to be considered:

a. wealth
b. social position
c . sexual enjoyment
d . children
e. intellectual stimulation

A. The Choice
M i s s Adams
Bagby Caldwell Dexter

@ @ @
B. Elimination of Common Effects (c)

@ @ @
C. Regrouped Noncommon Effects
If choice is:
Bagby b wanted, d and e not important
Caldwell d wanted, b and e not important
Dexter e wanted, a not wanted, b and d not important

D. Inferences:
If Bagby, M i s s Adams i s a snob; if Caldwell, Miss Adams i s the
maternal type; if Dexter, M i s s Adams i s an intellectual.
FIG.4. Miss Adams chooses a husband.

the others, and he is very much the intellectual-widely read and conversa-
tionally scintillating. These are the only characteristics of these men that
we know anything about, and on the basis of this knowledge we may
diagram Miss Adam’s choice as in Fig. 4A.
First we lay out those effects presumed to follow from the choice
of each man to the exclusion of the other. Then we notice that the
three men are physically attractive so we rule out the common effect
of sexual enjoyment-we have no information about the importance of
232 EDWARD E. JONES AND KEITH E. DAVIS

this particular characteristic as a determinant of Mss Adams’ choice.

The remaining effects are regrouped after we receive information about
which man has been chosen, The only further distinction is that between
effects which are judged to have been wanted by Miss Adams, effects
judged to have been unimportant to her, and effects judged to have been not
wanted or actively avoided. The distinction between lack of importance
and undesirability is new; it is based on the commonality of particular
effects among the nonchosen alternatives, Thus, if Dexter is chosen, we
have no more reason to assume that Miss Adams wanted intellectual
stimulation than that she didn’t want the responsibilities of wealth. After
all, she has avoided two wealthy suitors in favor of one who is not
wealthy. Our inference about not wanting wealth, then, would be more
correspondent than an inference about not wanting children or not caring
about social position. If it were just a matter of not having children,
Miss Adams could just as easily have married Bagby. If she was anxious
to avoid social position she could have accomplished this by marrying
Caldwell. We say these these two effects were not important to her to
suggest that she is willing to forego them in favor of intellectual stimula-
tion, but our evidence that she wanted to avoid these effects is weaker
than in the case of wealth.
A further complexity which must be acknowledged is the fact that
some or all of the noncommon effects may be correlated or seen by
the perceiver to express the same general purpose. At the present stage
of developing the theory of correspondence, no formal provision is made
for this possibility, To some extent, our flexibility in deciding what we
shall call an effect reduces the magnitude of the problem. Thus, certain
combinations of discriminable effects may be treated as a more general,
unitary effect if each member of the combination has a common sig-
nificance for the perceiver. The result would be to increase the corre-
spondence and the generality or importance of the ensuing inference.
We shall examine this possibility in greater detail when we consider,
below, the contribution of hedonic or affective relevance of effects in
the inference process and again when we consider sequences of action
choices over time.
F. CONDITIONS
AFFECTINGDESIRABILITY
ASSUMPTIONS
We have remarked that the perceiver’s assumptions or hypotheses
about which effects of action were most likely desired by the actor play
an important role in the inference process. Those effects perceived to
have been high in desirability (i.e., commonly desired by all persons
or by all members of a particular cultural group) play a smaller role
in the determination of correspondence than those effects which are less
 FROM ACTS TO DISPOSITIONS 233
universally sought, Without going too deeply into the problem, there are
a number of variables which might condition the perceiver’s assumptions
that the actor desires the same effects as most persons. Even if we restrict
ourselves to the case in which the perceiver confronts the actor for the
first time, there may be cues in the circumstances of their encounter
and in the appearance of the actor which affect the likelihood of his
being seen as desiring the same things most persons would in a given
situation. If the situation is so structured that the actor and the perceiver
are working for the same objectives, the perceiver may reflect on his
own intentions to draw inferences about the most likely aspirations of
the actor. Cues about shared perspectives should thus facilitate the forma-
tion of definite hypotheses about the actor’s motives and desires. Other
cues, perhaps reflected in the features of the actor’s appearance, might
lead the perceiver to assume similarity of intention and disposition before
any action has occurred. Or the perceiver’s stereotypes about the members
of identifiable classes or cultural groups may be triggered by such appear-
ance cues.
G. INFORMATION GAINAND THE ROLEOF PRIORCHOICE
We now come to one of the most frustrating sources of complexity
in calculating noncommon effects in the chosen alternative, both for the
individual perceiver whose actions the theory concerns and for the theory
itself. A person who confronts certain behavior choices has often made
previous choices which have brought him to his present decision. There
is often a great deal of information contained, then, in knowledge about
what alternatives are being considered, above and beyond the information
revealed by the actual decision which is made.
Once again, this point can be clarified by an example. Let us consider
Miss Adams again, this time caught between the options of going to
medical school or to law school. She has been accepted for admission
in the two professional schools of comparable universities and the choice
is, in that sense, entirely up to her. In comparison, we come upon Mr.
Bagby, poised before the choice of going to Duke or to Colorado for
graduate study in psychology. Since the Duke and Colorado psychology
departments would seem to have much more in common than medical
school versus law school, our theory would seem to suggest that any
inference we might make about Mr. Bagby after he has chosen would
be more correspondent than any comparable inference about Miss Adams.
There are many noncommon effects in Adams’ chosen alternative, whereas
the number of noncommon effects in Bagby’s chosen alternative is un-
questionably smaller. The hidden factor in this comparison is the fact
that considerably more information is contained in the datum that Bagby
234 EDWARD E. JONES AND KEITH E. DAVIS

had already ruled out everything besides psychology, than in the datum
that Adams is still struggling with the choice between two basic pro-
fessions, professions which differ from each other on many different di-
mensions. The example is thus misleading because Bagby is at a later
stage of the choice process than Adams. In order to render them com-
parable we would have to have more information concerning Bagby’s pre-
ceding choices, choices wlich have narrowed the field to psychology.
The example points up the importance of defining correspondence
in terms of the information gained through the observation of behavior,
not in terms of the confidence one has in drawing an inference which
may be based on prior information or on knowledge of the culture.
If all we know is that Bagby has chosen the Colorado Department of
Psychology over the Duke Department of Psychology, we have learned
nothing from the choice itself concerning the strength of his motivation
to become a psychologist. On the other hand, we might be willing to
make some rather confident statements about his love of mountains or
the degree of his dislike for hot and humid summers. The facts that
there are high mountains and low humidity in Colorado are certainly
two of the noncommon effects in the Colorado versus Duke choice.
In dealing with cases of complex behavioral choice in the natural
environment, the matter of prior choice is destined to create enormous
dBculties in the application of the theory to individual cases. However,
in the realm of experimental planning, prior choice presents opportunities
for empirical exploration rather than disruptive trouble. The probability
of imputing prior choice can either be held constant or systematically
varied by suitable experimental arrangements. In the former case for
example, subject-perceivers, starting from the same baseline of relative
ignorance about a stimulus person, may be exposed to a choice made
by that stimulus person under conditions which emphasize the stimulus
persons’s lack of control over the choice presented him.
H. CORRESPONDENCE, CHOICE, AND ROLEASSIGNMENT
A recent experiment by Jones et al. (1961 ) exemplifies the reasoning
which underlies the foregoing theory. It is worth reviewing this study
before considering some of the more subtle and tentative extensions of
the theory into interaction settings marked by personal involvement. The
investigation’s central purpose was to demonstrate that behavior which
conforms to clearly defined role requirements is seen as uninformative
about the individual’s personal characteristics, whereas a considerable
amount of information may be extracted from out-of-role behavior. In
other words, inferences based on out-of-role behavior were predicted to
be higher in correspondence than inferences from in-role behavior. The
 FROM ACTS TO DISPOSITIONS 235
reasons for this will become more apparent after the procedures of the
experiment are described.
Male undergraduate subjects were exposed to one of four tape re-
corded “job interviews” in which the interviewee was instructed (on the
tape) to appear very interested in qualifying either as a prospective sub-
mariner or as an astronaut. The subjects were aware that the interviewee
was being invited to play a role in a fictitious interview situation, but
they were not told that the entire interview was carefully written as
a prearranged script and was recorded by an experimental accomplice
serving as the interviewee.
Those subjects who listened to the recording involving the sub-
mariner role, heard the interviewer describe the ideal submariner as obedi-
ent, cooperative, friendly, gregarious-in short, as “other directed.” The
remaining subjects listened to a description of the ideal astronaut as
one who does not need other people, who has inner resources-in short,
a rather “inner-directed” person. These two interview beginnings were
spliced into two different endings, thus creating the four experimental
groups. The interviewee either responded with a series of statements indi-
cating extreme other-directedness or he responded with a series of inner-
directed statements. On half of the recordings, then, the interviewee-
accomplice behaved very much in line with the requirements of the occupa-
tional role (astronaut-inner condition, submariner-other condition). On
the other half the behavior was distinctly out of line with these require-
ments (astronaut-other condition, submariner-inner condition).
After listening to these tape recordings, the subjects were asked
to rate the interviewee (“what do you think he is really like as a person?”)
and indicate their confidence in the traits they evaluated on their rating
scale. The results were striking and unequivocal (see Table I ) . After
the two in-role recordings, the stimulus person was rated as moderately
affiliative and moderately independent. In each case the confidence ratings
were extremely low. On the other hand, the astronaut-other was seen
as very conforming and affiliative, and confidently rated as such. The
submariner-inner was seen as very independent and non-affiliative, again
with high confidence. Thus the actual responses of the interviewee were
clearly evaluated in the context of the structured setting from which they
emerged. If other-directedness is called for, an inner-directed response
is highly informative. Inner-directedness in the face of a situation which
seems to require it, on the other hand, is difficult to interpret. The same
kind of contrast applies to other-directedness in the two settings described,
providing a replication of the basic hypothesis within the single experiment.
Now we may ask how the results of the Jones et al. (1961) experi-
ment are to be explained in terms of the foregoing theoretical statement
236 EDWARD E. JONES AND KEITH E. DAVIS

relating cultural desirability and the number of noncommon effects to

the degree of correspondence of an inference about personal dispositions.
In-role behavior does not lead to confident, correspondent inferences be-
cause such behavior has multiple consequences and many of these are
high in cultural desirability. Most people want to avoid embarrassing
others by not meeting their expectations, most people want to gain the
rewards impkit in approval from authority figures, most people wish
to manifest their intelligence by showing that they understand what is
required of them, and so on. Each of these effects is a “plausible reason”
TABLE I
PERCEPTIONS OF AFFILIATION
AND CONFORMITY”’

Comparisons
Astro- Astro- Sub- Sub-
other (AO) inner (AI) other (SO) inner (SI) Direction t

N: 33 33 51 37

Affiliation
R 15.27 11. la 19.00 8.64 A0 >SO 4.0ad
SD 2.99 3.81 3.53 4.73 A1 > SI a . la‘
Conformity
B 15.91 13.09 1%.58 9.41 A0 > SO 4 . Oad
SD 3.92 3.4% 3.39 4.95 A1 > SI 3.

Data from Jones d al. (1961).

The higher the mean value, the greater the perceived affiliation or conformity.
Comparisons between A 0 and SI are not tabled, but the differences between these condi-
tions would of course be highly significant.
p < .05.
d p < .001.

for in-role behavior in the experiment just described. On the other hand,
plausible reasons for out-of-role behavior (Le., those with a reasonabIe
degree of assumed social desirability) are comparatively scarce. One of
the few noncommon effects of behavior at variance with role demands
is the satisfaction of expressing one’s true nature. This effect is also
a possible accompaniment of in-role behavior, but in that case it exists
in the choice circle along with many other effects. Since there are fewer
noncommon effects in the astronautsther and submariner-inner choices,
the effect of “being oneself” forms the basis of a more correspondent
inference in these conditions and the interviewee’s behavior tends to be
taken at face value.
The implications of this study can probably be extended quite gen-
erally to cover behavior which is or is not constrained by a well defined
 FROM ACTS TO DISPOSITIONS 237
social situation. When certain role requirements are salient, conformity
is more rewarding to the actor, more likely to avoid embarrassment and
social disapproval, than is nonconformity, The actor may conform for
many other reasons as well. Thus, in the case of conformity to role
requirements, we do not know the exact reason why the individual be-
haves the way he does, but there is really no particular mystery in his
behavior. This is an example of “trivial ambiguity.” On the other hand,
behavior which departs from clearly defined role requirements cries for
explanation. The fact that the effects of such behavior are presumably
low in cultural desirability makes the behavior intriguing to the perceiver.
The fact that there are few reasons why a person would behave that
way (the action leads to a limited number of noncommon effects) provides
the basis for a correspondent inference concerning the intentions and
dispositions of the actor.

III. Personal Involvement and Correspondence

In the remaining sections of the chapter, we shall turn to those
factors of personal involvement which affect the inference process in
person perception. The theory of action implied in the discussion thus
far obviously assumes that the actor is concerned with the consequences
of his action. It is the very fact that his action choices have motivational
significance for him that makes these choices informative for the perceiver.
But a special and enormously important feature of many person perception
settings is that the choice of an actor has significant rewarding or punishing
implications for the perceiver. We turn to examine this feature and
to consider its implications for our theory of correspondence.

A. THE HEDONIC
RELEVANCE
OF THE ACTIONTO THE PERCEIVER

The actor’s behavioral choices may or may not contain effects which
have hedonic relevance for the perceiver. The hedonic relevance of an
effect is a function of its motivational significance for the perceiver: does
the particular action consequence promote or undermine the perceiver’s
values; does it fulfill or obstruct his purposes? Effects which fulfill a
purpose have positive relevance; those which obstruct a purpose have
negative relevance. For a choice to have relevance means that the algebraic
balance of positive and negative effects in the chosen alternative is not
equal to the algebraic balance of positive and negative effects in the
nonchosen alternative(s). Simply put, the choice proved gratifying or
disappointing to the perceiver.
An experiment by Steiner and Field (1960) is conceptually quite
similar to the astronaut-submariner study but contains a strong dash of
238 EDWARD E. JONES AND KEITH E. DAVIS

hedonic relevance as an added ingredient. In this study, University of

Illinois students met in groups of three to “discuss the desirability of
desegregation of public schools and . .. attempt to reach agreement among
themselves.” The major manipulation varied the extent to which the re-
sponsibiIity for presenting certain points of view was assigned by the
experimenter. For half of the groups, a confederate of the experimenter
was always assigned the role of “a typical Southern segregationist.” In
the other groups, subjects were encouraged to take into consideration the
viewpoint of an N.A.A.C.P. member, a Northern clergyman, and a
Southern segregationist, but no role assignments were made. In both cases,
however, the confederate gave an identical, prosegregation performance.
From the perceiver’s point of view, he apparently chose to express pro-
segregation beliefs where no role assignment was made, whereas he had
little choice but to express the same beliefs in the role assignment condi-
tion. Since the subjects themselves were all in favor of integration, the
expression of prosegregationist beliefs would, we assume, be relevant in
the negative direction.
The following results would be expected given the theoretical state-
ment that we have developed thus far: (1) perceivers should attribute
more intense prosegregation beliefs to the actor in the choice condition
than to the same actor in the assignment condition; (2) perceivers should
be more confident of their inferences in the choice condition; and, ( 3 )
they should evaluate the chooser less favorably than the actor who had
the role assigned to him. The investigators do not report the data bearing
on the first hypothesis, but the other data make sense only if it were
supported. Hypothesis 2 was strongly supported, and both indices bearing
on hypothesis 3 were in the predicted direction, though only one treatment
difference was significant. In addition, while the fact is not particularly
relevant in the present theoretical context, the subjects were apparently
more influenced by the remarks of the actor when he chose the role
than when he was assigned to it, even though he was better Iiked in
the latter case.
The results confirm very well the expectation derived from the theory
of correspondent inferences, and the subjects show the same uncertainty
in the role assignment condition as was observed in the in-role treatments
of the astronaut-submariner study. Since there are so many objectives
served when the actor in the role assignment condition follows his assign-
ment, and since most of these objectives are quite culturally desirable,
the perceiver learns very little from the actor’s compliance.
Hedonic relevance is involved because the position taken by the
actor is contrary to the view held by all perceivers in the experiment.
While the experiment does not manipulate the relevance of the action
 FROM ACTS TO DISPOSITIONS 239
directly, it does alter the subject’s evaluative response to the action by
altering his interpretation of the act.
It is not as yet clear, however, precisely how relevance enters into
the inference process. At the outset, it may be useful to distinguish between
the effects of relevance on correspondence, and the joint effects of rele-
vance and correspondence on evaluation by the perceiver, Let us consider
each of these in turn. We propose that as relevance increases there is
also an increase in the likelihood that inferences will be correspondent.
This is because effects which might appear to have little in common
in the eyes of most observers might be functionally equivalent to a par-
ticular perceiver. Thus, relevance may provide a potent criterion for
grouping and packaging the effects of action, thereby reducing the number
of unrelated or noncommon effects in the choice circle. The result is
an increase in the correspondence of any inference based on that particular
choice. This reasoning does not apply in the event that a nonrelevant
effect is seen as the probable goal of the action. However, we may assume
that the probability of launching the inference from a relevant effect
increases directly as a function of the degree of relevance involved.
In addition to the packaging of effects in terms of their positive
or negative significance for the perceiver, the number of noncommon
effects may be further reduced by assimilation to the predominant hedonic
value. When the actor makes a choice which is relevant to the perceiver,
there will be a tendency for the remaining more or less neutral effects
to take on the sign of other effects in the choice circle, This assimilation
should operate in such a way as to increase the differentiation between
chosen and nonchosen courses of action. The process may be illustrated
by changes in the connotative meaning of attributed dispositions. Let
us assume that we have identical information concerning the moderately
high risk-taking tendencies of Adams and Bagby. If Adams does something
which, on balance, goes against our interests, the assimilation hypothesis
proposes that risk-taking proclivity might be construed as recklessness
and irresponsibility. If Bagby does something that supports our interests
and benefits us, riskiness might take on connotations of creativeness and
inventive autonomy. This would seem to be an expression of Heider’s
(1958) general balance principle: Bad actions come from bad people
and good is achieved by the good.
Turning now to the joint effects of relevance and correspondence
on evaluation by the perceiver, the following proposition suggests itself:
If the consequences of an act are predominantly positive, the perceiver
will be more favorably disposed toward the actor, the greater the corre-
spondence value of the action. The converse will be true of actions whose
effects are negative. In general, ignoring direction for the moment, the
 240 EDWARD E. JONES AND KEITH E. DAVIS

evaluation of an individual will be more extreme as a joint function

of increases in relevance and correspondence.
Since relevance increases correspondence, and since relevance and
correspondence affect evaluation, it might seem reasonable to link relevance
directly to evaluation. However, relevance may well affect only one condi-
tion of correspondence-the commonality of effects-and not the other,
the cultural desirability of effects. For this reason it is possible to have
high relevance and only moderate correspondence. When, for example,
the Russian ambassador to the United Nations makes a speech accusing
America of imperialistic ambitions, dollar diplomacy, exploitation of the
worker, and so on, it is easy for us to put these remarks into a single
package under the label of negative hedonic relevance. And yet, we are
sufficiently aware of the norms of cultural desirability among Russian
public spokesmen to recognize that none of the ambassador’s statements
departs very far from these norms.
In terms of the fourfold table presented as Fig. 3, we are dealing
with a case of trivial clarity. Note that relevance is high-the statements
chosen by the ambassador have effects almost all of which axe an affront
to our values as American perceivers. The number of noncommon effects
is low-the disparate remarks may be readily packaged as anti-American;
and assumed cultural desirability (for a Russian) is high. But, we would
not predict a particularly intense negative evaluation in this case. Since
this particular Russian is just saying what any other Russian would say
under the same circumstances, it is rather hard to take special umbrage
at his “negatively relevant” remarks, The example helps us to see, then,
that a combination of relevance and high correspondence is prerequisite
for extreme evaluations to occur. Relevance controls the direction of
evaluation, but is only one of two contributing determinants of its
extremity.
I . Relevance Increases Correspondence: Empirical Support
In order to test the hypothesis that relevance increases correspond-
ence in the inference process, it is necessary to present the same action
or series of acts in contexts of differing personal relevance for the per-
ceiver. This was done in an experiment by Jones and deCharms (1957)
and in another by Kleiner (1960), the results of which we shall briefly
summarize.
Two separate experiments, sharing certain basic procedural features
were conducted by Jones and deCharms (1957). In the first experiment
a trained accomplice was the only member of a group, including four
or five naive subjects, who failed the assigned experimental task. In
one condition, individual fate, the relevance of this failure was minimized.
 FROM ACTS TO DISPOSITIONS 24 1
The subjects all received the rewards promised them for succeeding and
this was in no way contingent on the accomplice’s performance. In another
condition, common fate, the accomplice’s failure prevented anyone from
reaping the rewards available. This was, then, a condition of negative
hedonic relevance for the naive subjects. The subjects rated the accom-
plice twice; first, prior to the main experimental inductions and again
after his failure was established. We would expect to find, in an index
of change in ratings of the accomplice, indications of greater correspond-
ence in the negative relevance than in the minimal relevance condition.
In line with this prediction, the accomplice was regarded as being less
competent, less dependable, and generally judged in less favorable terms
in the common fate (negative relevance) than in the individual fate (mini-
mal relevance) condition, Contrary to expectation, no differences in likea-
bility or friendliness occurred as a function of relevance. Perhaps we
may cite this pattern of findings to illustrate that relevance may affect
certain attributions without necessarily affecting personal evaluation.
A study by Kleiner (1960) varied the positive relevance of con-
structive member actions by varying the probability of group failure. A
previously instructed accomplice then facilitated group goal achievement by
solving problems too difficult for the others. We assume that the degree
of positive relevance varies directly with the degree of initial threat to
the group. Unfortunately for our purposes, Kleiner did not get extensive
impression ratings over a variety of traits, but he did get evidence con-
cerning changes in perceived importance of group members. Consistent
with the relevance-correspondence hypothesis, the greater the group’s need
for help, the greater importance attributed to the helpful confederate.
Consistent with the second evaluation, the rated likeability of the con-
federate as both a teammate and as someone to socialize with was posi-
tively related to the degree of initial need for help.
While there are no other investigations (to our knowledge) that
concern themselves directly with the relevance-correspondence hypothesis,
there are several closely related studies which increase our confidence
in its validity. These are studies in which conditions of potential relevance
are created by the anticipation of further interaction, but in which im-
pression ratings are taken before the direction of relevance has been
established by final action. The general pattern of findings from these
studies has been called “facilitative distortion” of perceived attributes-the
stimulus person is assigned attributes that are consistent with the positive
outcome hoped for in the interaction.
The classic study was done by Pepitone (1950). Variations in moti-
vation (relevance) were established by having high school students think
that their ideas about athletics would be instrumental in obtaining either
242 EDWARD E. JONES AND KEITH E. DAVIS

very desirable championship basketball tickets (high relevance) or much

less desirable tickets (low relevance). The three judges who evaluated
the students’ ideas were again accomplices of the experimenter. They
varied their apparent approval of the subject in some conditions, and
their apparent power to grant him a ticket in others. On the whole,
there was a strong tendency for subjects to view the more favorable
judges as more powerful than the less favorable judges, though power
was ostensibly equated by instructions and the accomplices’ careful atten-
tion to their prescribed roles. Similarly, when there was a deliberate at-
tempt to vary the judges’ power, the more powerful judges were regarded
as more approving, We may only assume that such “facilitative distortion”
would not have occurred in a no-relevance control group. It was true
that the high relevance subjects saw the approving judge as more powerful
than the low relevance subjects did, which provides direct support for
the relevance-correspondence hypothesis, but the remaining differences
which might test the hypothesis were not significant.
A similar pattern of facilitative distortions was found in one phase
of Davis’ (1962) study. Subjects were given preinformation about an
individual with whom they were to engage in a series of either cooperative
or competitive interactions. For half of the subjects, this information
portrayed an essentially submissive person; for the other half, an essentially
ascendant person. When the submissive person was to be a partner, she
was seen as more active, outgoing, forceful, tough, and as less passive, shy,
and uncertain of herself than when she was to be an opponent. These dif-
ferential effects did not approach statistical significance in the condition of
ascendant information. Perhaps the constraints of clear information were
strong enough to inhibit distortion in the latter case.
Other studies also show both facilitative distortion effects and the
absence of such effects, but in no study do “pessimistic” distortions occur.
There seems little doubt that relevance may increase distortion by causing
increases in correspondence which are not based on added information.
What we now need is to determine other parameters which influence
the relevance-correspondence relationship and thus affect the perceiver’s
reliance on the available data he obtains from observed action.
2 . Relevance and Correspondence Determine Evaluation: Empirical
Support
The second hypothesis concerning relevance was that evaluation
is a joint function of the degree of relevance and the level of corre-
spondence. Under a variety of different guises, this hypothesis has received
greater empirical attention than the prior hypothesis linking relevance
to correspondence. It is not difficult to find ample support for the proposi-
 FROM ACTS TO DISPOSITIONS 243
tion that people like others who benefit them in some way and dislike
others who are harmful. But since we have already argued that relevance
increases correspondence, this proposition is not a very precise rendition
of the second hypothesis. The second hypothesis requires the demonstra-
tion that both relevance and correspondence are necessary conditions
for evaluation, or at least that evaluation will be more extreme when
both are present at a high level. Our reasoning implies that evaluation
will become more extreme as a function of increases in either relevance
or correspondence, as long as the other variable is held constant at some
value greater than zero. If an action is expected to be positively or
negatively relevant for a perceiver, for example, the perceiver’s evaluations
should become more extreme when the conditions of judgment give rise
to high correspondence.
As one test of this hypothesis, we may return to the second experi-
ment reported in Jones and deCharms ( 1957). Cross-cutting the common
fate-individual fate variation which characterized both experiments, an
additional instructional variation was introduced. Half of the subjects
were led to believe that the task was such that failure should be primarily
attributed to lack of ability. The remaining subjects were told that failure
on the particular problems to be solved could only reflect a lack of
motivation, a lack of willingness to try hard. In retrospect, we might
now see the ability condition as involving less choice for the actor than
the motivation condition. After all, if ability and not motivation is in-
volved, then the subject may try heroically, knowing that others are de-
pendent on him and that doing well is important-but still fail. In the
ability condition, therefore, his failure would not provide a basis for
correspondent inferences about his attitudes toward the group. The indi-
vidual must have some degree of choice among action alternatives before
one may begin to speak of noncommon effects in the chosen alternative.
The results bear out the prediction quite well. An evaluation change
index was composed from the combined ratings of the accomplice made
by each naive subject before and after the experimental variables were
introduced. The traits involved were deliberately chosen to reflect an
evaluative “halo effect”: competent, intelligent, conscientious, likeable, de-
pendable, and so on. When the accomplice supposedly had no choice
(in the ability conditions), variations in personal relevance for the per-
ceiver did not lead to differential changes in evaluation. Thus the evalua-
tion change scores in the common fate-ability condition were almost identi-
cal to the evaluation change scores in the individual fate-ability condition
(see Table 11). When the accomplice was presumed to have a choice,
on the other hand, relevance was a crucial determinant of evaluation.
Subjects in the common fate-motivation condition were significantly more
244 EDWARD E. JONES AND KEITH E. DAVIS

negative in their evaluation change scores than subjects in the individual

fate-motivation condition.
It should be emphasized, of course, that the actual behavior of
the accomplice was as nearly the same in all conditions as careful pre-
training and periodic monitoring could make it. In conclusion, then, the
accomplice was negatively evaluated if his failing performance prevented
the others from obtaining rewards and he could have avoided failure
by trying harder. Not trying hard in this case may have been equivalent
to an attitude of indifference to the group, an attitude which (once in-
ferred) would be resented by the group’s members.

TABLE 11”
CHANGEIN “HALO
EFFECT,”EXPERIMENT
I1

Groups

Common fate Individual fate

-
Motivation Ability Motivation Ability
-
’p 23.0 17.3 14.4 18.9
SD 10.68 6.27 2.12 9.63

Data from Jones and decharms (1957).

b The greater the mean change, the more negative the “after” evaluation,

Perhaps the most celebrated study linking causal attribution and

evaluation is that of Thibaut and Riecken (1955). They conducted two
separate experiments to explore the proposition that an act of benevolence
which is “internally caused” is more appreciated than one which is the
inevitable result of environmental circumstances. We would now view in-
ternal causation as another way of talking about the perception of choice
alternatives available to the actor. A person “internally causes” certain
effects in the environment only when he had the option of causing other
effects and did not do so.
In each of the Thibaut-Riecken experiments, an undergraduate subject
was introduced to two experimental accomplices or confederates, one
of whom was apparently much higher in academic or social status than
himself and one of whom was lower in status. The subject soon found
himself in the position where he needed the help of at least one of
the confederates. The experimenter encouraged him to ask for help and
required only that he make an identical request of both the high status
and the low status confederate. When both of the confederates eventually
complied with his request, the subject was asked to explain the compliance
 FROM ACTS TO DISPOSITIONS 245
and to evaluate each confederate. Since the differences in experimentally
manipulated status were perceived as differences in the ability to resist
persuasion, the high status confederate was regarded as having more choice
in his decision about compliance. The low status confederate, on the
other hand, was more likely to be viewed as complying because he felt
“coerced” by the more powerful subject. The norms governing a low
status position are such that compliance to those higher in status is often
expected. In our terms, then, the behavior of the high status person
should lead to more correspondent inferences concerning the intention
to help the subject out of a disposition of spontaneous affection or good
will.
As our hypothesis would predict, holding relevance constant (the
subject is benefitted equally by the two confederates), as correspondence
of inference (about spontaneous good will) increases, positive evaluation
also increases. The benevolence of the high status confederate earns him
a greater increase in attractiveness than does the benevolence of the
low status confederate. Relevance in the positive direction, coupled with
high correspondence in the form of perceived internal causation, results
in more positive evaluation.
Incidental findings from two other studies may be mentioned as
well. These findings also bring out the relationship between relevance,
correspondence, and evaluation. In the study by Davis (1962) briefly
referred to above, control groups were run in which subjects anticipated
either cooperative or competitive discussions with each other, but no pre-
information about the partners was provided beforehand. Each subject
rated the other person prior to the interaction on traits which could be
combined into an ascendance-submission index, and on likeability. In
the competitive condition, the more ascendant one’s opponent, the greater
the probability of one’s own failure; ascendance has negative hedonic
relevance. In the cooperative condition, on the other hand, ascendance
has positive relevance since it implies a greater probability of team success.
Comparing individual differences in the tendency to assign high first im-
pression ratings on ascendance, we should expect a positive correlation
between perceived ascendance and likeability in the cooperative condition
and a negative correlation between these two sets of ratings in the competi-
tive condition. The correlational values were actually f.60 and -.18,
respectively, reflecting a difference between conditions which is significant.
Finally, in an experiment by Jones and Daugherty (1959), some
subjects were led to anticipate interacting with one of two persons about
whom a fair amount of information was provided via a tape-recorded
interview. Others received the same information about the two persons,
but it was clear that no subsequent interaction would take place. One
246 EDWARD E. JONES AND KEITH E. DAVIS

of the two interviewees was presented as a rather intellectual, somewhat

dif€ident person, with moderately strong aesthetic interests. The other was
presented as a rather opportunistic and conforming, but obviously sociable
person. In the no-anticipation condition, in which we may assume that
the characteristics of both persons were of minimal relevance, the diffident
esthete was more highly evaluated on a variety of dimensions than was
the sociable politician. In contrast, when the subjects were led to anticipate
interacting with one of the two, making the relevance of sociability more
salient, the subjects’ evaluation of the politician markedly and signifi-
cantly increased, These results, then, suggest that a particular personal
attribute (sociability) was assigned approximately the same ratings in
the two conditions (varying in the anticipation of interaction) but vari-
ations in the relevance of that attribute were associated with shifts in
evaluation. If there had been no evidence that the “politician” was soci-
able, correspondent inferences about him would not have been drawn
regarding that disposition and evaluation would not have varied with rele-
vance, It should be emphasized that the obtalned differences were not
anticipated. However, we view the interpretation as the most plausible
one available and are encouraged to think that a replication specifically
addressed to the present hypothesis would show the same pattern.
In summary, there seems little question that variations in the rele-
vance of an action to the perceiver have an effect on the process of
inferring dispositions which explain the action. Our first hypothesis was
that relevance tends to increase correspondence by reducing the number
of noncommon effects in the action alternative chosen. We have presented
some evidence in favor of this hypothesis, although it is clear that the
strength of confirmation depends on other conditions, such as the am-
biguity of available information about the actor and the consequent leeway
for facilitative distortion. The second hypothesis, which states that personal
evaluation varies as a joint function of relevance and correspondence,
has received stronger support than the first. Here again, however, much
of the evidence is indirect and circumstantial. Hopefully, the present theo-
retical analysis will point the way toward more precise tests of both
hypotheses.
THE ACTOR’SINTENTION
€3. PERSONALISM: TO BENEFITOR HARM
THE PERCEIVER
An act or a choice may be hedonically relevant to the perceiver
even though it is quite clear to the latter that the choice was not conditioned
by his unique presence. An actor might express opinions which differ
radically from the perceiver’s without having any knowledge of the latter’s
views. Such a choice of opinions may have hedonic relevance for the
 FROM ACTS TO DISPOSITIONS 247
perceiver, but may not have been offered with any intention to gratify
or to spite [Link] variable of personalism is introduced to distinguish
between choices which are conceivably affected by the presence of the
perceiver and choices which are not conceivably so affected.
t a perceiver to judge whether a choice was
It is usually d ~ c u l for
affected by personalistic considerations. He may, in effect, experimentally
arrange conditions of his own presence and absence in an attempt to
detect differences in the choice made by the stimulus person. This is often
done indirectly, as when the perceiver compares reports of choices made
in his absence with his own observations of choices made in his presence.
We may try, for example, to find out what others say about us and
our beliefs behind our backs, When the actions of another person obstruct
our interests, it becomes important for us to determine whether the other
sets out specifically to make life unpleasant for us or whether we have
been disadvantaged as a by-product of actions primarily directed toward
other objectives. Similarly, when others go out of their way to help us,
we have an interest in establishing whether they did this because of
our uniquely attractive personality or because they would have helped
almost anyone under the circumstances.
The distinction between relevance and personalism hinges on the
perceiver’s imputation of a certain kind of knowledge to the actor: the
actor’s awareness that the interests of the perceiver are positively or
negatively af€ected by his actions. If such knowledge is not imputed by
the perceiver to the actor. then we are dealing with a case of ‘‘impersonal
hedonic relevance.”
When a hedonically relevant action is produced in the presence
of the perceiver, the latter’s problem is to decide whether the act was
uniquely conditioned by the fact that he was its target. When there is
such evidence of a “unique conditioning,” the perceiver is likely to draw
strong inferences of malevolence or benevolence, stronger than he would
as a bystander. He and only he is the target of the other’s highly relevant
action; therefore, it is assumed that the other has a special interest in
making life easy or difficult for him as a person.
Since the perceiver is going to be so vitally concerned with relevant
effects that were deliberately produced for his consumption, such effects
should clearly play a special role in shaping his inferences about the
actor. We propose that action which is both relevant and personal has
a direct and dramatic effect on evaluative conclusions about the actor.
One reason for this is that personalism clearly implies choice. If an actor
benefits a perceiver, this is a personalistic episode only if it reflects
the selection of that particular perceiver as a worthy beneficiary in the
face of opportunities to select other targets or other actions. The combina-
248 EDWARD E. JONES AND KEITH E. DAVIS

tion of personalism and positive relevance, then, insures a positive evalua-

tion simply by insuring a correspondent inference of focused benevolence.
The special significance of such focused benevolence may lie in the fact
that it satisfies the perceiver’s needs for information about his worthiness,
as well as other needs for security, power over others, and so on. In
any event, the receipt of focused benefit or focused harm should generate
“halo” effects in the inference process which go beyond the assimilation
to hedonic value predicted in the case of impersonal hedonic relevance.
Personalism may, of course, be incorrectly assumed by the perceiver.
The most extreme form of distortion along these lines may be seen in
paranoia, where innocent actions and actions not conditioned by the per-
ceiver’s presence, become the data for inferences concerning ulterior
malevolent motivation.
Surprisingly, there are few experiments which precisely assess the
role of personalism in the inference process. The above proposition implies
that hedonically relevant actions which the perceiver judges to be uniquely
affected by his presence will give rise to correspondent inferences to
all those attributes captured by a positive or negative “halo” effect. In
an experiment specifically concerned with variations in personalism, Gergen
(1962) arranged to have coeds receive uniformly positive, reinforcing
remarks from another coed under personal versus impersonal conditions.
Such remarks probably are hedonically relevant and positive as far as
the first coed is concerned.
In the personal treatment, the girls had been previously introduced
to each other, had engaged in a pleasant and informative interaction,
and the reinforcing person (actually an experimentaI accomplice) had
quite a bit to go on in expressing her positive feelings about the subject.
In the impersonal treatment, on the other hand, the subjects were informed
that the accomplice had been through some intensive training designed
to help her establish rapport in a social interaction. In addition, and
in clear contrast with the personal treatment, the accomplice never saw
the subject, but interacted with her through a microphone-speaker system
while the subject observed her through a one-way mirror. Each subject
was ultimately asked to record her impression of the accomplice on a
series of evaluative scales. From our proposition concerning the role of
personalism in producing high correspondence for evaluative character-
istics, we would expect a more positive halo effect in the personal than
in the impersonal experimental treatment. There were, however, no sig-
nificant differences between the subjects’ evaluative ratings of the accom-
plice in the two conditions.
There was some evidence that the subjects felt sorry for the accom-
price in the impersonal condition, since she had to operate under the
 FROM ACTS TO DISPOSITIONS 249
rather embarrassing handicap of being seen by the subject without being
able to see her. There was also some confusion about whether the subject
was to rate the accomplice as she appeared to be or as she “really
was.” We do not feel, therefore, that the Gergen experiment is a crucial
test of the personalism proposition, though some variation of Gergen’s
procedure would seem to have promise as a fairly direct approach to
the problem. It is at least conceivable that the proposition only holds
for harmful actions, and that persons are much less sensitive to variations
in personalism when positive actions are involved. This may be especially
true when these positive actions involve verbal compliments. The reluc-
tance of subjects to assume that a compliment was not intended for
them personally is discussed in detail by Jones ( 1964). In this same source
the reader will find a fuller exposition of the Gergen experiment along
with results of other dependent variable measures which were more central
to his concerns.
1 . Factors Mitigating One’s Evaluation of an Aggressor
More indirect and yet more promising evidence on the role of person-
alism comes from experiments concerned with the factors which mitigate
one person’s reactions to being verbally attacked or insulted by another.
The basic paradigm involves comparing perceivers’ reactions to the same
attacking action when it occurs in different settings. Typically, one setting
is designed to bring out reciprocal hostility in the subjects (in the form
of highly negative impression ratings) while other settings are arranged
to check whether factors which theoretically should mitigate a hostile
reaction in fact do. We are especially interested in those studies within
this paradigm which exemplify variations in the perceived personalism
of the attack.
a. Provocation by the perceiver. An obvious mitigating variable which
comes to mind is the extent to which the attack is seen as justified
by the target person. If the perceiver believes he has done something
to earn attack, insult, or rejection, he will presumably be less inclined
to appraise his attacker negatively than if the attack was unreasonable
or arbitrary. For exampIe, Deutsch and Solomon (1959) found that sub-
jects who were led to believe they had performed poorly on a task were
less negative in appraising a stimulus person who rejected them as future
work partners than subjects who were led to believe they had performed
well.
A similar point is brought out by the results of an experiment
by Strickland et al. (1960) on the effects of group support in evaluating
an antagonist. Each subject met first with two other subjects who shared
his opinions (pro or con) about the role of big-time athletics in university
250 EDWARD E. JONES AND KEITH E. DAVIS

life. He then privately chose a series of five arguments to support his

position. These were to be transmitted to a person in the next room
who was presumably neither for nor against big-time athletics, After this
person had a chance to study the arguments, he was interviewed by
the experimenter who probed his feelings about the person who sent
the arguments. This interview was broadcast into the subject’s room and
it contained a strong attack on his intelligence and integrity.
However, prior to his exposure to the broadcast interview (which
was actually a standardized tape-recording), the subject learned that his
fellow group members would either have chosen the same arguments he
did (group support) or would have chosen a very different set (no sup-
port). The subject’s final ratings of the person in the next room-which
tapped into such dispositional characteristics as intelligence, warmth, ad-
justment, conceit, and likeability-were affected by this variation in group
support. These rating differences were corroborated by free response
sketches in which each subject expressed his private feelings about the
person. Those whose arguments were supported were more negative in
their evaluations of the person in the next room than those whose argu-
ments were not supported.
In neither the Deutsch and Solomon (1959) nor the Strickland
et al. (1960) study was the potential for perceiver personalism particularly
high. In each, regardless of the experimental condition, a very limited
sample of the subjects’ behavior was the stimulus occasion for attack
or rejection. The subject did not, in other words, expose the full range
of his personal characteristics as a preface to the attack received. Neverthe-
less, the attack was directed toward him ostensibly because of behavior
for which he must bear at least some of the responsibility.
We now suggest that an attack in the face of good performance
(or group support) is more apt to be viewed as an attempt to harm
or to disadvantage the subject than an attack in response to poor per-
formance, because after a poor performance (including the sending of argu-
ments defined as inferior by the group), the attacker will be seen as
more constrained to respond negatively. The correspondence value of his
hostile action, in other words, will be lowered by the presence in the
choice area of effects having more to do with fulfillment of task require-
ments, candor, and realism than with hostility. Since there are fewer
reasons for the antagonist’s attack in the good performance setting, and
one of these is presumably t%e antagonist’s desire to hurt the subject as
a person, correspondence and therefore unfavorability of general impres-
sion are high in this latter case.
b. Evidence of chronicity. Another factor which mitigates a per-
ceiver’s evaluation of an antagonist is any evidence conceming the latter’s
 FROM ACTS TO DISPOSITIONS 25 1
general tendency to be indiscriminately aggressive. If the antagonist is
known to be or gives fairly good evidence of being a chronically dyspeptic
or uncontrollably negative person, his derogation will have less sting for
the target person who bears its brunt.
Two recent experiments by Berkowitz (1960), conducted in quite
a different framework from the one we are here proposing, shed some
light on the effects of a perceiver’s prior knowledge of a particular
attack. Since only the first of the Berkowitz studies is particularly relevant
to our present concerns, we shall confine ourselves to that.
Pairs of subjects were brought together for a study of first im-
pressions. Through a bogus note exchange, the subject received informa-
tion first, indicating that the partner was either generally hostile or gen-
erally friendly and second, indicating that the partner either liked or
disliked the subject personally, The subject recorded his impression of
the partner once at the outset of the experiment, once after the general
information, and once after the personal evaluation from the partner.
The results showed that if the partner was perceived to be hostile
initially, the partner’s favorable evaluation of the subject had a decidedly
ameliorating effect on the subject’s impression, while the unfavorable eval-
uation changed this impression very little. Similarly, if the subject initially
perceived the partner to be friendly, the unfavorable evaluation received
from the partner created a striking change of impression in the direction
of perceived unfriendliness, while the favorable evaluation resulted in
minimal change.
It would appear, then, that the fact of prior knowledge concerning
the hostility of the attacker reduces the personal significance of the attack.
If we look more closely at the Berkowitz results, however, the point
they make is actually rather different from the one we are presently
pursuing. If a person who is already seen as generally hostile attacks
the perceiver, there will be less of a decline (from the second to the
third rating) than if a friendly person attacks the perceiver. However,
the subject actually ends up liking the hostile attacker less than the friendly
attacker, presumably because the evidence concerning the undesirable char-
acteristic, hostility, summates: two hostility indicators are worse than one.
The Berkowitz results really do not suggest that the perceiver is less
bothered or upset by the attack if he has already decided that the attacker
is generally hostile. They merely tell us that the attack is not as unexpected
from a hostile person and therefore it contributes less to a change in
impression from a point that is more negative to start with. We are
dealing here, then, with the attempt on the part of the perceiver-subject
to appraise the significance of a particular action choice against background
information about different prior choices.
252 EDWARD E. JONES AND KEITH E. DAVIS
In order to confirm the significance of perceived general hostility
as a prior choice factor mitigating the significance of the attack, the
results would have to show that the generally hostile attacker is better
liked by the recipient of the attack than the friendly attacker. However,
such a finding might be d ~ c u l to t obtain experimentally. After all, the
fact that he is hostile does not make the attacker likeable to anybody.
It merely means that the attack itself will cause less of a stir.
Evidence from a recent experiment does indicate that someone who
starts out being derogatory and continues to act that way is better liked
than someone who starts out being favorably disposed to the subject
and becomes increasingly derogatory (Aronson and Linder, 1965). In
this experiment the subject believed he was over-hearing a series of ap-
praisals referring to him with short episodes of social interaction inter-
vening; he was not the target of openly expressed hostility. This may
be a critical difference between the Berkowitz design and the Aronson
and Linder design. Another difference that may have been ciucial is that
the former study asked for an intervening rating (which might have “com-
mitted” the subject to a particular rating of the attacker) while the latter
study did not.
If we return to the conditions of the Berkowitz experiment, it may
be too much to expect the hostile attacker to be better liked than the
friendly attacker. What is needed is a comparison between the target
of the attack and an “innocent” bystander as regards their impressions
of the attacker. Because of the general negative significance attached
to being hostile, it does not make sense to predict that either the involved
subject or the bystander would like the hostile attacker better than the
friendly one. However, a more refined and promising hypothesis, still
in the spirit of our earlier remarks on the role of personalism, is that
the involved subject will dislike the hostile attacker less than the bystander,
relative to the discrepancy between their impressions of the friendly attacker.
c. Emotional adjustment of the attacker. Such an experimental com-
parison has yet to be made, unfortunately, but the results of an earlier
experiment by Jones et al. (1959) can be interpreted quite nicely in
these terms. The procedures of the study by Jones et al., were roughly
as follows. At a given experimental session, a pair of female subjects listened
to two female stimulus persons allegedly conversing about one of the
subjects in an adjacent room. The conversation was actually a carefully
written and skillfully acted tape-recording. The stimulus persons were al-
legedly enrolled in a “senior course in personality assessment” and it
was their duty to observe a designated subject through the one-way mirror
for a period of time and then to discuss their impressions of that subject.
It was clear to them that their remarks would be overheard by the
 FROM ACTS TO DISPOSITIONS 253
subjects in the adjacent room. One of the stimulus persons was generally
neutral or mildly favorable in her comments, but the other stimulus person
(“the derogator”) was decidedly hostile and clearly had a low opinion
of the subject. The subject whose characteristics were not being discussed
was instructed to sit aside as a bystander and to pay close attention
to the proceedings.
Prior to the attack, both the involved and bystander subject were
given some information about the two students who would be observing
them from the next room. It was clear upon reading this information
that one of these students was quite maladjusted: She had an unstable
home life, inadequate emotional resources, and underlying anxiety. The
other student was presented as an effective, well-rounded, insightful under-
graduate who had reached her present station from a home life that had
been happy and rich with support and affection, For one group of subjects
(the derogator-ma1 group), the data sheets presented the stimulus person
who did not derogate as well adjusted. For a second group of subjects
(the derogator-well group), the background information sheets were simply
reversed.
After the involved and the bystander subjects listened to the tape-
recorded discussion, including the derogatory remarks, they were each in-
structed to rate the two stimulus persons on a number of items. The
items with which we are particularly concerned at the present are two
reflecting the perceived likeability of the stimulus person. The subjects
were asked to indicate the extent of their agreement with the two state-
ments: “As a person, she is extremely likeable,” and ‘? find it hard
to like this person to any extent.”
Thus the experimental manipulations created a standard situation
in which a subject was derogated by a well-adjusted or a maladjusted
person while another subject looked on, The variable of perceived malad-
justment was included, in effect, to see whether it would serve as a factor
mitigating the subject’s response to the derogator. Jones et d.,reasoned
that the involved subject would be less upset by an attack from a malad-
justed person than by an attack from a well-adjusted one. The bystander,
on the other hand, was expected to be less concerned in general with
the derogation and its implications for inferences about the derogator’s
personality, and more inclined to prefer the well-adjusted person because
she was probably more appealing and talented. The prediction, then,
was that there would be a statistical interaction between role (involved
versus bystander) and condition of the derogator (maladjusted versus
well adjusted).
The data were analyzed in terms of each subject’s relative preference
for the nonderogator over the derogator, a procedure adopted to reduce
254 EDWARD E. JONES AND KEITH E. DAVIS

that portion of rating variability due to individual differences in scde

interpretation or style of responding to the scale items. (Some such device
is usually essential in an “after-only” design.) The crucial results are
summarized in Table 111. It is evident that there is a general dislike
for people who are derogatory and a general preference for people who
are well adjusted. When these two factors work in the same direction (as
in the first column of the table), the discrepancy scores are understandably
large. Of greater theoretical interest, when the derogator is well adjusted,
the involved subject obviously likes her much less than the bystander sub-
ject ( p < .025). To a slight extent, the average bystander even prefers

TABLE I11
JOINT EFFECTSOF DEROOATOR
ADJUSTMENT AND SUBJECT INVOLVEMENT
ON ’
“LIKEABILITY”~~

Condition

Derogator-ma1 Derogator-well

Involved subjects
x 3.9% 2.17
SD (1.78) (2. as)
Bystander subjects
s 4.33 - .67
SD (3.17) (3.96)
-~ ~~ ~~~~ ~ ~~ ~ ~~

0Baaed on discrepancy between ratings of derogator and nonderogator. The larger

the mean value, the greater the tendency to dislike the derogator relative to the nonderogator.
Data from Jones et &. (1969).

the well-adjusted derogator to the maladjusted nonderogator (as indicated

by the minus sign in that cell). When the derogator is maladjusted, how-
ever, the involved subject actudy likes her better than the bystander
does (though this difference does not approach statistical significance).
The predicted interaction effect is minimally significant (c = 1.877; p < .05,
one-tailed test).
In the context of the present discussion, we would argue that the
personalistic significance of the derogation is obviously greater for the
involved subject than for the bystander, and that it is greater for the
involved subject when the derogator is well adjusted than maladjusted.
When the derogator is maladjusted, the involved subject can take comfort
in the hypothesis that the attacker’s hostility is a symptom of her own
problems and she would express similar insults to anyone who came within
range. Perceived personalism should be fairly low. When the derogator is
well adjusted, the involved subject will be more likely than the bystander
 FROM ACTS TO DISPOSITIONS 255
to package the insulting remarks into one cluster of highly related hedonic
effects, and therefore to assign more correspondent, personalistic meaning
to the attack. There is no easy way to escape the inference that the
derogator finds the subject personally offensive and is “against her.”
The maladjustment treatment in the preceding experiment may be
construed in terms of the reduction of freedom to choose which ac-
companies poor adjustment, and the perception of these restraints on choice
by the perceiver. Perhaps the prevailing stereotypes of mental health and
mental illness contribute to the tendency to perceive the maladjusted person
as not responsible for the trouble he may cause others. Under the proper
circumstances, however, he may be seen as more responsible for causing
trouble than the normal, well-adjusted actor. At least such is the implica-
tion of some results from the experiment by Gergen and Jones (1963).
d . Amplification b y ambivalence. Gergen and Jones set out to test
a set hypotheses deriving from the assumption that people are ambivalent
of
toward the mentally ill. Many persons expect the mentally ill to have
annoying characteristics but inhibit their annoyance because they acknowl-
edge the fact that they are not responsible for their condition and its
consequences. Gergen and Jones reasoned that the ambivalence toward
a particular mentally ill person would be “split” if a situation were arranged
in which his behavior had clear positive or negative consequences for
the perceiver. Thus a perceiver should like a benevolent mentally ill person
better than a benevolent normal person, and dislike a malevolent mentally
ill person more than a malevolent normal person.
In order to test this hypothesis (which was loosely derived from
psychoanalytic writings on ambivalence), 64 ambulatory V.A. hospital
patients (nonpsychiatric) were given the task of predicting a series of
hypothetical consumer choices being made by a patient in the adjoining
room. The patient in the next room was alleged to be in the hospital
either with a psychiatric illness or with a minor organic ailment. Actually
there was no person in the next room, and all the information about
him was conveyed by a combination of tape-recorded interviews and
feedback through equipment controlled by the experimenter.
The choices of the patient in the adjoining room (hereafter called
the stimulus person) were either very hard or very easy to predict. In
the low consequence (i.e., low relevance) condition, the stimulus person
(actually, by a ruse, the experimenter) provided corrective feedback by
an informative signal light whenever a prediction error was made. In
the high consequence condition, prediction errors called forth a raucous
buzzer of unpredictable duration. The experimenter also made it clear
that he found the buzzer very annoying, implying that it was up to the
subject to keep him happy by making the correct predictions. Both before
25 6 EDWARD E. JONES AND KEITH E. DAVIS

the prediction task and after it was completed, the subject filled out
an impression rating scale indicating his current feelings about the stimulus
person.
The experimental hypothesis was stated as follows: “Evaluative
judgments of a mentally ill stimulus person vary little as a function of
predictability unless affective consequences are attached to success and
failure of prediction. The role of affective consequences is less important

0 NORMAL

-
--
MENTAllY Ilk
PREDICTABLE
UNWEDICTABLE

I
I I I
HI LO
CONSEQUENCE
Fro. 5. Changes in evaluation as a function of mental status, predictability.
and consequence. (Data from Gergen and Jones, 1963.)

in evaluative judgments of a nornial stimulus person. Judgments of the

normal should directly reflect variations in predictability, regardless of
the consequences of judgment” (Gergen and Jones, 1963, p. 70).
The results presented in Fig. 5 quite strikingly coniirm this compli-
cated hypothesis. There is, as implied by the hypothesis, a significant statisti-
cal interaction between the three factors of normality, predictability, and
consequence. The most striking thing to note is the extent to which
consequence determines perceptions of the mentaIly ilI person. When he is
in a position to hurt the subject or to spare him pain-in short, when he is
either benevolent or malevolent toward the perceiver-the mentally ill
 FROM ACTS TO DISPOSITIONS 257
stimulus person is judged very favorably or unfavorably. When there are no
such personal consequences for the perceiver, the stimulus person’s
predictability is not a relevant factor in judging him. (There is also an
overall effect of predictability such that, across conditions, the predictable
person tends to be evaluated more positively than the unpredictable
person.)
Furthermore, it may be shown that the perception of benevolent
versus malevolent intentions is involved in the subject’s judgments of the
high consequence-mentally ill person. A variation of the experiment was
run in which the stimulus person had supposedly made his consumer choice
days before and was not, as alleged in the main experiment, actually in
the next room at the time and responding through his own actions to
the subject’s predictions. In other words, all feedback to the subjects
(including the unpleasant buzzer) was openly controlled by the experi-
menter. The experiment was in all other respects a precise replication of
the first version. In this variation, predictability again had a strong effect
on average evaluations, but there were no main or interaction differences
as a function of consequence or normality. There must, then, be some
possibility that the stimulus person is deliberately hurting or sparing the
subject for the complex effects noted in the first experiment to occur.
There are many questions raised by the Gergen and Jones experi-
ment. There are also special problems involved in relating these results
to the Jones et al. (1959) findings. In interpreting those findings we
argued, in effect, that an attack by a maladjusted stimulus person was
less devastating than an attack by a well-adjusted stimulus person, because
the normal person is perceived to have greater freedom of choice. It
is as if evidence concerning maladjustment acts as a damper on the intensity
of personal feeling toward the maladjusted person. The Gergen and Jones
results, however, seem to show that under certain conditions evidence
concerning maladjustment (i.e., mental illness ) amplifies rather than con-
stricts the intensity of the perceiver’s personal feelings. Does this mean
that the mentally ill person is assumed to have greater freedom of choice
than the normal person in the Gergen and Jones study?
In spite of certain superficial resemblances, the experiments are really
quite different in several, crucial respects. In the earlier experiment the
very meaning and intensity of the attack is presumably a function of the
attacker’s adjustment status. To be insulted by a pathetic, perhaps mildly
paranoid person is hardly to be insulted at all. In the Gergen and Jones
experiment, however, the consequences of the attack are embarrassing,
painful, and irritating, regardless of their source. In this case, furthermore,
the question of freedom to choose may actually exacerbate rather than
mitigate the response to the mentally ill stimulus person.
25 8 EDWARD E. JONES AND KEITH E. DAVIS

While the data do not force this interpretation on us, there is at

least nothing inconsistent in the rating or the post-experimental question-
naire results with the following speculations: When unpredictability hurts,
the “why” of the unpredictability becomes a more important issue to the
one who suffers. Two possible effects of “buzzing” the subject in the
high consequence condition are especially salient, hurting the subject and
being honest to one’s true preferences for certain consumer objects. In
the replication of the experiment, the first of these effects is ruled out
by the change in information about the role of the stimulus person.
Perhaps it is the case that when the stimulus person is normal, the most
likely hypothesis is that he is making “normal choices” and, therefore,
that the subject must take at least some of the blame for not being
able to figure these out. When the stimulus person is mentally ill, on
the other hand, the abnormal choice becomes an instrument of malevolence.
Since the choice is a function of abnormality, it is difficult for the subject
to maintain the feelings of sympathy which, in the low consequence condi-
tion, are sufficient to keep his impression a fairly neutral one. The im-
pression of malevolence may be heightened by the feeling that since the
mentally ill person is confused about his choices anyway, the least he can
do is to go along with the subject’s predictions and not lean on the error
buzzer.
Such speculations are obviously no substitute for clear and compelling
data on the roIe of perceived choice in the assignment of benevolent or
malevolent intentions. The preceding review of studies involving the effects
of personalism, studies mainly focusing on factors which mitigate a target
person’s response to being attacked, points up the need for additional
research into the cognitive consequences of being singled out for benefit
or harm. Those of us who have done research on this problem have
lacked the kind of integrating framework which is needed to carry out
a series of related studies. Perhaps the theory of correspondent inferences
will help to provide a focus for the parametric experiments which are
needed in cIarifying the basic facts about personalism and its implications.
2 . Personalism and Ingratiation
We have defined as personalistic those actions which are relevant
to the perceiver’s interests and, as far as the perceiver can tell, are de-
liberately carried out by the actor because of this relevance. The concept
of personalism inevitably implies a certain degree of choice on the part
of the actor which is not inherent in the concept of relevance. Since the
condition of choice increases the likelihood of correspondent inferences,
the coexistence of relevance and personalism should produce rather extreme
evaluative judgments. It seems intuitively plausible that someone who
helps us will be seen as more generous, helpful, friendly, etc., than someone
 FROM ACTS TO DISPOSITIONS 259
who helps another in our presence. A comparable line of thought applies
to the case of malevolent actions when we have been singled out as their
target.
In an attempt to provide a more rational basis for this intuition, we
suggest the following distinctions and their consequences. There are three
basic decisions which lurk in the wings during an interaction episode:
whether to approach the person further and open oneself to him, whether
to avoid or ignore him, or whether he must be “coped with,” i.e., attacked
or fended off. When the effects of an action are relevant in the positive
direction, the decision is typically made to approach further. Such a decision
is likely to be made whether or not the action is personalistic in addition
to being relevant. When the effects of an action are negative, however,
the imputation of personalism means that coping may be necessary. In
other words, if a person’s actions happen to offend you, he can merely
be avoided-unless he is intent on offending you and will go out of
his way to accomplish this objective. To the extent that coping is required,
we would expect greater hostility to be aroused toward the threatening
person, and those characteristics associated with malevolence should be
inferred with higher correspondence. This might be partly a matter of
justifying the hostility and partly a matter of keeping it at a high enough
level (through self-reminding instigations) to support coping behavior.
On the basis of this reasoning, it would appear that personalism
plays a larger role when an action is harmful than when it is beneficial.
The examples we have cited, along with most of the relevant experiments
in the literature, describe the response of a perceiver to some form of
attack or rejection. However, a number of experiments have recently been
completed which deal with first impressions in response to beneficial ges-
tures, compliments, and agreements. These studies raise a new set of con-
siderations which we shall discuss in concluding our treatment of the role
of personalism.
We have argued that negative actions lead the target person either
to avoid the actor or to mobilize cognitive support for the actions involved
in coping with him. Positive actions, on the other hand, lead to approach
behaviors, personal openness, and reciprocation in kind. But a new and
complicated problem arises with respect to positive actions. The perceiver
must determine their credibility; to what extent does the beneficial act
correspond with the intention really to improve the situation of the perceiver
as an end in itself? As Jones (1964) has argued in his extended discussion
of ingratiation, beneficial actions tend to be much more ambiguous than
harmful actions when it comes to deciding on the actor’s true intention
or his ultimate objectives in the situation. The ambiguity of beneficial
actions centers around the extent to which ulterior, manipulative purposes
may be served by them.
260 EDWARD E. JONES AND KEITH E. DAVIS

We may now ask, what implications are contained in the theory

of correspondent inferences for predicting the cognitive impressions of
someone faced with beneficial action? First of all, it is clear that the
ambiguity arises because there are at least two classes of effects following
from those actions, Actions such as compliments, agreements, and favors
may validate the perceiver’s self-concept, reduce his uncertainties, offer
support against antagonists. Alternatively, or in addition, such actions may
have the effect of obligating the perceiver to benefit the actor in return.
If the first class of effects is the most salient, the perceiver will attribute
to the actor the intention to express his true feelings. From this starting
point, correspondent inferences will be drawn to such dispositions as
candor, friendliness, likeability, and generosity. In short, the perceiver’s
evaluation of the actor who has complimented him or agreed with him
will be positive. If, on the other hand, the second effect, creating obligations
to benefit, is salient and noncommon, the actor may be seen as manipula-
tive, self-seeking, conforming, lacking in candor, etc.
Whether the inference process is tipped in the first, positive direction
or the second, negative direction depends on the perceiver’s reconstruction
of the action alternatives available to the actor. This cognitive reconstruc-
tion will depend, in turn, on the perceiver’s own role as one of the
components in the actor’s situation. Specifically, if the perceiver does
not control any resources which are important to the actor, then the
circle containing the effects of the chosen alternative will not contain the
effect of “obligating the perceiver to benefit the actor in return.” Pre-
sumably, then, some such effect as “validating the self-concept” will be
salient and the perceiver will be seen as intending an honest compliment
or expressing his genuine agreement with the perceiver’s opinions.
If the perceiver does control resources important to the actor (i,e.,
if the actor is dependent on him), it will be hard for him to decide
whether he is merely the target of an ingratiation attempt or the target
of honest compliments. At the very least, the correspondence of inference
to favorable dispositions will be reduced as a function of his own position
as a dispenser of valuable resources. Depending on the circumstances,
the perceiver may infer flattering or manipulative intentions and assign
unfavorable dispositions, or he may infer benevolent intentions reflecting
favorable dispositions.
Let us now consider three recent experiments which support the
conclusions of the above line of reasoning. Jones et a!. (1963) conducted
an experiment in which upper classmen in a campus R.O.T.C. unit ex-
changed written messages with freshmen in the same unit. This exchange
occurred in response to two different sets of instructions, constituting
the major treatments of the experiments. In the ingratiation condition,
both the high and low status subjects were given instructions concerning
 FROM ACTS TO DISPOSITIONS 261
the importance of compatibility. The experimenter said he was trying to
find a number of highly compatible leader-follower pairs to participate
in some crucial studies on leadership later in the year. In the control
condition, the message exchange was presented as part of a first impression
study and the importance of “not misleading your partner” was stressed.
The messages that were sent concerned opinions on a variety of
issues and eventually contained ratings by each subject of himself and
his partner. These messages were actually intercepted and standard infor-
mation about each subject was conveyed to his partner. Each found the
other agreed with him on a variety of opinion issues, presented a rather
modest view of himself, and expressed a complimentary view of his part-
ner, the message recipient. On a post-experimental questionnaire (not to
be seen by the partner) he was asked to rate the partner with respect
TABLE IV
PERCEPTION
OF FLATTERYO.&

Perceiver group
~~

HS Ls

M SD N M SD N
- Pdiff

Ingratiation 13.05 4.14 19 9.62 3.16 21 .01

Control 11.68 4.46 19 11.85 3.41 90 11s

a Mean post-experimental ratings in each condition and differences between them.

The higher the mean score, the greater the pcrreivrd flnttcry.
b Data from Jones et al. (1963).

to the following trait dimensions: completely sincere-on the phony side,

trustworthy-unreliable, and brutally frank-flatterer. Each pair of
antonyms was separated by a twelve-point scale. By adding a subject’s
rating on each of the three traits, he could be given a score ranging from
3 to 36 with a “perceived average” value at 19.5.
The results are presented in Table IV. They show that high and
low status subjects perceived each other to be equally sincere in the control
condition, but that the low status subjects attributed significantly greater
sincerity to the highs than the highs did to the lows in the ingratiation
condition. Restricting our concern to the ingratiation condition, we would
say that the inference concerning sincerity is more correspondent for
the low status perceivers than for the highs. Relative to the hypothetical
average value of 19.5, the empirical mean of 9.62 is more extreme than
the mean of 13.05.
This could have been predicted from the theory of correspondent
inferences on the grounds that fewer noncommon effects were involved
in the high status person’s decisions to compliment the low status person
262 EDWARD E. JONES AND KEITH E. DAVIS

than vice versa. Since the low status person was, presumably, in greater
need of approval than the high status person, the latter may have been
more apt to include “reciprocation of approval” among the effects serving
as grist for the inference process, This could, then, have led to reduced
correspondence of inference, i.e., ratings of sincerity which were closer
to the mean or, in effect, greater perceived flattery.
Such an interpretation is quite post hoc and we offer it to illustrate
how the theory might account for such findings rather than as confirmation
of the theory. The problem is that other assumptions (which we believe
are plausible) must be introduced to account for the fact that low status
subjects perceive the highs to be more sincere in the ingratiation than
in the control condition.
A study by Jones et a2. (1963) was more explicitly designed to
test the hypothesis that positive, supportive behavior will be taken more
at face value as a genuine indication of sincerity and good intentions
when the actor is not dependent on the target person. The supportive
behavior, in this case, was consistently high agreement with the latter’s
opinions. Dependence was manipulated in a manner simular to the preced-
ing experiment. Unlike the conditions of that experiment, the subjects were
not themselves the targets of agreement, but served in the role of by-
standers. Their task was to evaluate a stimulus person who agrees very
closely with another person on whom he is obviously very dependent or
not dependent at all.
In general, when the agreeing person was presented as dependent
for approval on the other, he was better liked and was assigned more
positive characteristics when he did not agree too closely. When dependence
was low, on the other hand, the degree of agreement did not af€ect the
ratings to any significant extent. Once again, the actor’s condition of de-
pendence affected the significance attached to highly “ingratiating” be-
havior. The subjects felt neutral about the high dependent conformist,
because they did not know whether he was conforming for strategic ad-
vantage or whether his opinion agreement was coincidental. The fact
that he was dependent, thus, increased the ambiguity of his behavior
by adding the granting of approval to those possible effects of action
with which the perceiver had to come to terms in his evaluation.
A study by Hilda Dickoff (discussed in Jones, 1964) also showed
quite clearly that an actor who consistently compliments the perceiver
is better liked when he is not dependent on the latter. Dependence has
no effect when the evaluation received is still positive but contains a few
plausible reservations.
The obvious feature of all of these studies is the fact that the
same behavior (actions which can be seen as ingratiating in intent) results
in quite different inferences depending on the context in which it occurs.
 FROM ACTS TO DISPOSITIONS 263
More specifically, the studies on ingratiation which we have cited share a
concern with the variable, dependence, as the contextual conditions whose
presence or absence affects causal attribution. Our inference that a compli-
mentary or agreeable person really likes us is apt to be stronger if we
are unable to think of anything we have that he might covet. In other
words, the compliments or expression of opinions will be taken at face
value and correspondence will be high when the actor has no apparent
reason to choose the compliment other than his belief that it applies to us.
Now let us return to the notion of personalism and note an apparent
qualification of our proposition that personalism increases correspondence.
It may appear that when we are dealing with actions that are potentially
ingratiating, correspondence declines as a function of personalism. Com-
pliments to one’s face are harder to evaluate than the same positive
statements said behind our back. Opinions which agree with our own
are more apt to be taken at face value when expressed prior to our
opinion avowals than after such avowals.
We would not argue with the above interpretations in these hypo-
thetical cases. We would claim, however, that personalism is involved in
quite different ways in the kind of face-to-face confrontation where ingrati-
ation is an issue and in the case of negative or neutral information.
In fact, ingratiation only becomes an issue in the absence of indications
that personalism is involved. The person who receives a face-to-face
compliment must decide whether that compliment was meant for him
because of his unique personal qualities, or was meant for anyone who
happened to occupy a position as a potential dispenser of resources.
The high status person may have a difficult problem arranging conditions
to test the reactions of his subordinates to him as a person; it may
be hard for him to get certain kinds of self-validating information. The
important point is that actions that may be seen as directed toward him
as an occupant of a social position may therefore not be personalistic
and the correspondence value of inferences derived from such ambiguous
actions is apt to be low.
IV. Summary and Conclusions
In the present essay we have attempted to develop a systematic
conceptual framework for research on person perception. We have been
especially interested in specifying the antecedent conditions for attributing
intentions or dispositions, having observed an action. Dispositional attri-
butes are in a general way inferred from the effects of action, but not
every effect is equally salient in the inference process. Even if we assume
as perceivers that the actor knew what the effects of his action would
be, we must still engage in the complex analytic process of selectively
linking certain effects achieved to certain effects intended. This assignment
264 EDWARD E. JONES AND KEITH E. DAVIS

of intentions can provide sufficient reason for (or explanation of) the
action, so that the perceiver may go about his interpersonal business un-
fettered by a concern with ultimate or infinitely regressive causes.
Our most central assumption in considering the attribution of in-
tentions is that actions are informative to the extent that they have emerged
out of a context of choice and reflect a selection of one among plural
alternatives. When we pursue the implications of this assumption in some
detail, it is apparent that the distinctiveness of the effects achieved and
the extent to which they do not represent stereotypic cultural values de-
termine the likelihood that information about the actor will be extracted
from an action. We have used the term “correspondence of inference”
to refer to variations in this kind of informativeness. To say that an
inference is correspondent, then, is to say that a disposition is being rather
directly reflected in behavior, and that this disposition is unusual in its
strength or intensity. Operationally, correspondence means ratings toward
the extremes of trait dimensions which are given with confidence.
Having formulated the inference problem in these terms, an obvious
research question arises. What are the factors which control the perceiver’s
judgment that the actor had a choice? Or, more precisely, what conditions
influence his judgment concerning the number and distinctiveness (non-
commonness) of effects? It is our hope that cumulative, perhaps even para-
metric, research will be stimulated by posing the inference problem in
these terms. A study in which the stimulus person either went along
with or resisted clearly stated role-demands, was presented to exemplify
some of the more obvious implications of the theory. The results of the
study may be interpreted as showing that a low degree of “psychological”
choice is functionally the same thing as having many reasons for making
a choice. In-role behavior is supported by too many reasons to be informa-
tive about the actor; out-of-role behavior is more informative because the
effects of such actions are distinctive (few in number) and not to be
dismissed as culturally desirable.
In the latter portions of the present essay we have considered the
further complexities associated with perceiver involvement which affect
theoretical predictions concerning inferred attributes. Our analysis distin-
guished between two levels of involvement: hedonic relevance and per-
sonalism. An actor’s choice is hedonically relevant for the perceiver if,
on balance, it promotes or thwarts his purposes. An action is personalistic,
in the perceiver’s view, if it was uniquely conditioned by the latter’s
presence: if conditions are such that the perceiver believes he is the in-
tended consumer of the effects produced by the actor.
In discussing the various effects of relevance, we argued that corre-
spondence generally increases with increasing relevance. Evaluation, in
turn, is a joint function of both relevance and correspondence. A small
 FROM ACTS TO DISPOSITIONS 265
number of studies were discussed which seem to shed some light on the
impact of relevance. In particular, it was noted that if one holds relevance
constant (at some value other than zero) and manipulates the variables
alleged to increase correspondence, evaluation becomes more extreme.
Similarly, by pegging correspondence at a particular level and increasing
relevance, the same increase in evaluation extremity may be observed. It
should be emphasized, however, that much of the research cited was only
indirectly concerned with variations in relevance. More systematic re-
search is needed to establish the conditions under which relevance calls
forth positive or negative evaluations. In addition, we need to know much
more about the relations between affective and cognitive processes implied
by the linkage between relevance and correspondence.
In the final section on personalism, we discussed a study in which
this variable was directly and dramatically manipulated, only to acknowl-
edge that the effects were negligible. It will be important to establish
the reasons for this curious result by designing other experiments which
directly approach different facets of the complex personalism variable.
There is, however, indirect or circumstantial evidence which encourages
the conviction that personalism and hedonic relevance are not identical
in their effects. Specifically, we discussed several experiments which were
concerned with the mitigation versus amplification of hostility toward an
attacker. Here it was seen that the intensity of hostile reciprocation was
affected by factors in the situation which made it more or less likely
that personalism was involved. Such factors as sufficiency of provocation
and indiscriminateness of the attack, were shown to affect the recipient’s
evaluation of the attacker. These conditions could be, and were, discussed
in terms of the correspondence-noncommon effect theory. Several experi-
ments on ingratiation and the perception of flattery were also discussed
in these terms. The main dilemma of the perceiver, when he becomes the
target of actions which may be ingratiating in intent, is to determine
whether he is being benefited because of his unique personal qualities
or because of the resources which he may control.
This essay, long .as it is, could have been much longer if we
had hedged our statements with proper qualifications and dealt fully with
the problems which remain in our formulation. We have no illusions that
we have finally opened the main door on the mysteries of causal attribu-
tion. Our formulation has changed considerably since our work on this
essay began, and it will undoubtedly change much more with further
thought. We trust it is also obvious that the ability to accommodate old
findings from complex experiments is an easy hurdle for any theory to
jump. We remain optimistic, however, that the present framework en-
courages systematic thinking about inferring dispositions from actions and
suggests some of the major variables that merit initial consideration.
266 EDWARD E. JONES AND KEITH E. DAVIS

REFERENCES
Aronson, E. and Linder, D. (1965). Gain and loss of esteem as determinants
of interpersonal attractiveness. J . exp. soc. Psychol. 1, 156-172.
Berkowitz, L. (1960). Repeated frustrations and expectations in hostility arousal.
J . abnorm. soc. Psychol. 60, 422-429.
Davis, K. E. (1962). Impressions of others and interaction context as determinants
of social interaction 5nd perception in two-person discussion groups. Unpub-
lished doctoral dissertation, Duke Univer.
Deutsch, M., and Solomon, L. (1959). Reactions to evaluations by others as
influenced by [Link] 22, 93-1 12.
Gergen, K. I. (1962). Interaction goals and personalistic feedback as factors
affecting the presentation of the self. Unpublished doctoral dissertation, Duke
Univer.
Gergen, K. J., and Jones, E. E. (1963). Mental illness, predictability, and affective
consequences as stimulus factors in person perception. J. abnorm. SOC. Psychol.
67, 95-104.
Heider. F. (1944). Social perception and phenomenal causality. Psychol. Rev.
51, 358-374.
Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley.
Jones, E. E. (1964). Ingratiation. Appleton, New York.
Jones, E. E., and Daugherty, B. (1959). Political orientation and the. perceptual
effects of an anticipated interaction. J . abnorm. SOC. Psychol. 59, 340-349.
Jones, E. E., and decharms, R. (1957). Changes in social perception as a function
of the personal relevance of behavior. Sociometry 20, 75-85.
Jones, E. E., Hester, S. L. Farina, A., and Davis, K. E. (1959). Reactions to un-
favorable personal evaluations as a function of the evaluator’s perceived adjust-
ment. J . abnorm. SOC. Psychol. 59, 363-370.
Jones, E. E., Davis, K. E., and Gergen, K. J. (1961). Role playing variations and
their informational value for person perception. J. abnorm. SOC. Psychol. 63,
302-310.
Jones, E. E., Gergen, K. J., and Jones, R. G. (1963). Tactics of ingratiation
among leaders and subordinates in a status hierarchy. Psychol. Monogr. 77,
Jones, E. E., Jones, R. G., and Gergen, K. J. (1964). Some conditions affecting
No. 3 (Whole No. 566).
the evaluation of a conformist. J . Pers. 31, 270-288.
Jones, E. E., Gergen, K. J., Gumpert, P., and Thibaut, J. W. (1965). Some
conditions affecting the use of ingratiation to influence performance evaluation.
1. pers. soc. Psychol. 1, 613-626.
Kleiner, R. J. (1960). The effects of threat reduction upon interpersonal attractive-
ness. J . Pcrs. 28, 145-156.
Pepitone, A. (1950). Motivational effects in social perception. Hum. Relat. 1,
57-76.
Steiner, I. D., and Field, W. L. (1960). Role. assignment and interpersonal influence.
J . abnorm. SOC. Psychol. 61,239-246.
Strickland, L. H., Jones, E. E., and Smith, W.P. (1960). Effects of group support
on the evaluation of an antagonist. J . abnorm. SOC. Psychol. 61, 73-81.
Thibaut, J. W.,and Riecken, H. W. (1955). Some determinants and consequences
of the perception of social causality. 1. Pers. 24, 113-133.