University of Southern California Dissertations and Theses

Evidence for spontaneous situational inferences

(USC Thesis Other)

Evidence for spontaneous situational inferences

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter face, while others may be
from any type o f computer printer.
The quality of th is reproduction is dependent upon th e quality of the
copy subm itted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back o f the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly to
order.
UMI
A Bell & Howell Information Company
300 North Zed) Road, Ann Arbor MI 48106-1346 USA
313/761-4700 800/321-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
EVIDENCE FOR SPONTANEOUS SITUATIONAL INFERENCES
by
Yijing Yang
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment o f the
Requirements for the Degree
Master of Arts
(Psychology)
August 1997
© 1997 Yijing Yang
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 1387849
UMI Microform 1387849
Copyright 1998, by UMI Company. All rights reserved.
This microform edition is protected against unauthorized
copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, MI 48103
I
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UNIVERSITY O F S O U T H E R N C A L IFO R N IA
T H E G R A D U A TE S C H O O L
U N IV E R S IT Y P A R K
L O S A N G E L E S . C A L IF O R N IA 8 0 0 0 7
This thesis, w ritten by
under the direction o f e.r—Thesis Comm ittee,
and approved by all its mem bers, has been pre
sented to and accepted by the D ean o f The
Graduate School, in partial fu lfillm en t o f the
requirements fo r the degree o f
y > J
THESIS COMMITTEE
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
p. iii
p. iv
p .l - p.8
p.8 - p.25
p.8- p. 10
p.10 - p.15
p. 15 - p.20
p.20 - p.25
p.25 - p.28
p.28
p.29 - p.31
»
J L
Table of Contents
I. List of Figures & Tables
II. Abstract
II. Introduction
m . Method
A. Paradigm & Stimuli
B. Study 1.
C. Study 2.
D. Study 3.
IV. General Discussion
V. Conclusion
VI. References
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List o f Figures & Tables
Figure 1 — Read & Miller’s (1993) Parallel Processing Model of Social Inference, p.6
Table 1 — Sample Vignettes. p.9
Table 2 — Mean Reaction Times in Study 1. p. 13
Table 3 — Median Reaction Times in Study 1. p. 13
Table 4 — Mean Reaction Times Adjusted for Reading Times in Study 1. p . 14
Table 5 — Mean Reaction Times in Study 2. p. 17
Table 6 — Median Reaction Times in Study 2. p. 17
Table 7 — Mean Reaction Times Adjusted for Reading Times in Study 2. p. 18
Table 8 — Mean Reaction Times by Cultural Beliefs in Study 2. p. 19
Table 9 — Mean Reaction Times in Study 3. p.23
Table 10 — Median Reaction Times in Study 3. p.23
Table 11— Mean Reaction Times Adjusted for Reading Times in Study 3. p.23
£
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
iv
Abstract
The current paper seeks to offer initial evidence that situational inferences can occur
spontaneously. Based on a reaction time paradigm used by Singer, Revlin and Halldorson
(1990), subjects in three studies read target sentences paired with either a dispositional
context that elicits dispositional explanations or a situational context that elicits situational
explanations and then answered dispositional or situational inference questions. A no
context condition was included in the third study. The consistent pattern o f results across
the three studies indicated that subjects were much faster in responding to the questions
when they matched the contexts as compared to the no context or mismatching context
condition. Moreover, cognitive load did not have a differential effect on the extent to which
spontaneous dispositional or situational inferences were made. Taken together, these
results suggest that under the right information configuration, situational inferences can be
as spontaneous as dispositional inferences.
i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Introduction
Suppose you see someone leaving a restaurant without giving a tip, do you think
that person is mean and stingy? Suppose you also learn that the waiter in the restaurant
gave very bad service to that person, would you still think the reason why that person did
not leave a tip is due to his mean personality or would you instead think he did it in
response to the bad service? Although both dispositional inferences and situational
inferences are valid explanations for people’s behaviors in daily life, most researchers in
social psychology have tended to focus on dispositional inferences, with the most
pronounced example being Gilbert's three stage model of social inference, which proposes
that our brain is hard-wired to preferentially make dispositional inferences and not
situational inferences (Gilbert, 1989). Similarly, the research on spontaneous trait
inferences (e.g. Uleman, Newman, & Winter, 1992) has exclusively looked at trait
inferences without inspecting the spontaneity of situational inferences. Indeed, psychology
has almost exclusively looked at dispositional inferences while little has been done to
examine the role of situational inferences which are arguably just as important as
dispositional inferences.
The current paper seeks to fill this gap by providing initial evidence that situational
inferences do occur spontaneously and can be as spontaneous as dispositional inferences.
We will first briefly review Gilbert's three stage model of social inference and the literature
on spontaneous trait inferences to see why it is of paramount importance to investigate
spontaneous situational inferences. We will then spell out an alternative model of social
inference proposed by Read and Miller (1993), in which both dispositional and situational
inferences can occur spontaneously. Three studies were conducted to offer initial support
for the Read and Miller model. Finally, the results of these studies and their implications
will be discussed.
i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
!
2
Gilbert's Three-Stage Model of Social Inference
One of the most robust phenomena in social psychology is the fundamental
attribution error, the tendency to underestimate the situational pressures that might have
contributed to another's behavior and overattribute the behavior to another's personality
(e.g. Heider, 1958; Ross, 1977). Many models of social inference have been proposed to
explain this counter-intuitional phenomenon (e.g. Gilbert, 1989; Quattrone, 1982; Reeder,
1993; Trope, 1986). One influential view (e.g. Gilbert, 1989; Gilbert, Peelham, & Krull,
1988) suggests that a three-stage process occurs when people draw causal inferences.
First, people categorize the actor's behavior ( e.g. The woman is behaving in an anxious
manner.), next they characterize, or draw a dispositional inference that corresponds to the
behavior (e.g. The woman must have an anxious personality.), and finally they may correct
the dispositional inference by taking into account situational pressures ( e.g. The woman is
being asked embarrassing questions. So the situation she is in and not her personality
might have contributed to her anxiety.). According to Gilbert, because the dispositional
inference stage is an automatic process while the correction stage is more of a controlled
process, people usually underadjust for situational pressure in everyday interaction, where
people are typically cognitively busy. Thus, Gilbert argues that the fundamental attribution
error occurs, at least partially, because the necessities of social interaction reduce the
cognitive capacity available for controlling situational factors.
Gilbert’s model is based on the assumption that dispositional inferences are special,
and that somehow they are cognitively privileged over situational inferences (Gilbert,
1989). However, neither evolutionary science nor neuroscience has provided clear bases
for the claim that our brain is hard-wired to preferentially make trait inferences and not
situational inferences. Figuring out the environment is as important as figuring out other
people in helping us to survive, if not more important. Thus, Gilbert’s model is not
plausible, at least not at a biological level.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3
Secondly, Gilbert’s model fails to explain the finding in the cross cultural literature
that the preference for trait inferences exhibited in the fundamental attribution error differs
from culture to culture (e.g. Zarate & Uleman, 1994). Specifically, the cultural differences
are typically described in terms of individualism and collectivism (e.g. Triandis, 1989,
1990). In individualistic cultures, the uniqueness and independence of an individual are
emphasized while collectivist cultures pay much more attention to interdependence and
cooperation among people. American culture and most European cultures are individualistic
cultures; Asian and Latino cultures are more collectivist (Hofstede, 1980). Because o f the
differences in the belief systems of the two kinds of cultures, attributions of causality and
responsibility are made more in terms of traits in individualistic than in collectivist cultures
(e.g. Markus & Kitayama, 1991). For instance, Miller (1984) found the reverse pattern of
the fundamental attribution error among her Hindu subjects. Members of collectivist
cultures tend to make situational inferences from behaviors while putting less emphasis on
cross-situationally consistent person characteristics. Miller's research demonstrates the
fundamental attribution error is not that fundamental after all, for it may be a phenomenon
that is specific to the Western culture.
In sum, these findings suggest that situational inferences are not less basic and
hence less spontaneous than dispositional inferences by nature. The fact that spontaneous
trait inferences are more frequently found in research does not necessarily mean
spontaneous situational inferences do not exist, or to a lesser degree. Rather, it may be a
product specific to the Western culture, in which people have practiced and have been
conditioned to focus on the individual. Clearly, Gilbert's model fails to account for this
cultural difference in the spontaneity of making causal inferences. The fact that Gilbert’s
model falls short at both the biological level and the cultural level calls for a reexam ination
of the premise that dispositional inferences are special and more spontaneous than
situational inferences.
i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
4
Spontaneous Trait Inferences
Another line of research that specifically looks at dispositional inferences is the
study of spontaneous trait inferences. Looking through the literature, we found an
abundance of research that has shown people make spontaneous trait inferences (for a
review see Uleman, Newman, & Moskowitz, 1996), while little evidence exists for
spontaneous situational inferences. The fact that much of the research has found
spontaneous trait inferences, however, does not mean spontaneous situational inferences
do not exist. On the contrary, the lack o f research in spontaneous situational inferences may
be partially responsible for the fact that trait inferences are thought to be more spontaneous
than situational inferences.
The only published study that investigated both spontaneous trait inferences and
spontaneous situational inferences was done by Lupfer, Clark, and Hutcherson (1990).
Subjects in the study received either background information designed to facilitate
dispositional inferences, information to facilitate situational inferences, or no background
information. All subjects were instructed to memorize a series of digits that were either
difficult or easy. Cued recall revealed that dispositional background information facilitated
trait inferences but did not affect situational inferences. Similarly, situational information
facilitated situational inferences but had no impact on trait inferences. Moreover, subjects
showed no awareness of making either dispositional or situational inferences, suggesting
both dispositional and situational inferences can occur spontaneously outside people's
awareness. Lupfer and his colleagues (1990) concluded that the study "yielded strong
evidence for the spontaneity o f trait-relevant inferences and qualified evidence for the
spontaneity of situational inferences".
Lupfer et al. also found that situational inferences are more affected by cognitive
load than dispositional inferences, suggesting that dispositional inferences might be more
spontaneous than situational inferences. However, caution must be taken in interpreting
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that result. Lupfer et al. used stimulus materials developed by Winter, Uleman, & Cunniff
(1985) and Uleman, Newman, & Winter (1987), which were explicitly designed to
strongly elicit dispositional inferences (e.g. The mailman picks his teeth during dinner at
the fancy restaurant.). Their study, therefore, fell short in assessing whether situational
inference could be as spontaneous as dispositional inference. As Lupfer and his colleagues
pointed out in their discussion, "In order to assess whether the spontaneity of trait-relevant
and situational inferences are equipotent, one would need to include target sentences that
had been shown independently to elicit situational attributions, along with the sentences that
elicit dispositional attributions." Thus, the possibility exists that when there is a strong
situational context to elicit situational explanations, situational inferences can be as
spontaneous as situational inferences.
In the current paper, we aim to fill in the gap of our knowledge about the
spontaneity of situational inferences and present initial evidence for the existence of
spontaneous situational inferences. To provide theoretical background for this work, we
discuss a parallel processing model o f social explanation proposed by Read and Miller
(1993), in which not only can dispositional and situational inferences occur spontaneously,
but they can both occur at the same time. In this model, the extent to which we make one
type of inference versus the other depends on various factors including preexisting
knowledge structures, information configurations, and salience.
Specifically, this model proposes that constructing causal explanations involves the
repeated application of two steps. First, input activates a host of concepts that are organized
in a loose, heterogeneous network, composed of both relevant and irrelevant concepts.
Second, activated concepts are organized into a coherent representation o f the causal
relationships by a parallel constraint satisfaction process that evaluates the explanatory
coherence of the network (Thagard, 1989, 1992). As new information is received, these
steps are repeated.
i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
A simplified representation of the model is presented in Figure 1 as it applies to the
Gilbert et al. (1988) paradigm, in which the subjects were trying to determine whether an
actor's anxious behavior is due to an anxious disposition or to an anxiety provoking
situation. Each node represents what may be a set of related concepts. The nodes at the
bottom represent the overt behavior cues that help to activate the behavior identification
node that this is an anxious behavior. Positively linked to the anxious behavior node are the
anxious disposition and the anxious situation nodes, whereas the two nodes are negatively
linked themselves, for they are rival although possible explanations for the behavior. Thus,
both nodes are activated at the same time, and the strength of the explanation is evaluated
by passing activation among the nodes in parallel until the activation asymptotes. Which
explanation is best depends on the weights, and the relative amount of attention paid to the
various nodes.
Figure 1— Read & Miller’s (1993) Parallel Processing Model of Social Inference.
Fidgeting"] Sweating ~|
Anxious Behavior
Anxious Disposition Anxiety-Provoking Situation
This model enjoys several advantages over Gilbert’s model. First, this model not
only predicts that people make both spontaneous trait inferences and spontaneous
situational inferences, but also offers an explanation of why they may both occur and why
sometimes one is stronger than the other. Read and Miller’s model predicts that the
likelihood of making dispositional or situational inferences depends on the extent to which
they are supported by the stimulus array. Thus, to the extent that situational causes are
highly salient or have a stronger link to the effect, people should be more likely to make
, spontaneous situational inferences. On the other hand, when the dispositional causes are
i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7
highly salient or have a stronger link to the effect, people should be more inclined to make
spontaneous dispositional inferences. This is consistent with Krull's (1993, 1996) finding
that which kind of inference is more affected by cognitive load is based on the inferential
goal. Krull noted that in most of Gilbert's research, subjects were explicitly given the goal
of diagnosing the actor's personality, which he argued could be contributing to the fact that
dispositional inference was less affected by cognitive load than situational inference.
Consistent with this hypothesis, Krull showed that individuals with a goal to diagnose the
impact of the situation underadjust for dispositional factors under high cognitive load.
Second, both Gilbert's model and Krull’s later refinement of the model (Krull,
1993) argue that the consideration of dispositional and situational information is serial
rather than parallel. Krull (1996) specifically proposed that the order in which this
information is considered and which kinds o f inferences are relatively automatic is a
function o f explicit inferential goals to either diagnose personality or the situation.
However, there are logical problems with such serial models. First, such serial
models argue that in the characterization stage only one of the two types of inferences can
be processed and the other must be delayed until the correction stage. This is inconsistent
with Trope (1986) and others' finding that both dispositional and situational information
are integrated quite early when behavior is identified. Second, when we explain both
parties' behavior in a dyadic interaction, each person's behavior is simultaneously part of
the situation for the other. Thus, exactly the same behavior is both dispositional and
situational information. According to the serial model, the same behavior would som etim es
be in the characterization stage as dispositional information and other times in the correction
stage as situational information. The categorization of the behavior would continually
bounce around, which is hardly plausible. On the contrary, Read and Miller's model argues
that both dispositional and situational information can be integrated in parallel and quite
quickly, resulting in a spontaneous inference that integrates both types of information.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5
t
8
Thus, this model does not suffer from the logical problems that Gilbert’s and Krull's
models have.
Finally, the Read and Miller model also explains the finding that trait inferences are
more prevalent in Western cultures than in Eastern cultures, for they are more salient in
Western cultures than in Eastern cultures and they are better learned. Such evidence is not
supported by Gilbert’ s model of dispositional inference or any other model that proposes
situational inferences are a controlled process.
Three studies were done to provide initial evidence that supports Read and Miller's
model of social explanation. The first study was conducted to test the hypothesis that under
the right information configuration, people make spontaneous situational inferences as well
as spontaneous trait inferences. The second and third studies looked at factors that affect
the extent to which people make one type of inference versus the other. Specifically, the
factors investigated were cognitive load and cultural differences (individualism vs.
collectivism). We predicted that because situational inferences can be as spontaneous as
dispositional inferences, cognitive load should not have a differential effect on the extent to
which these two types of inferences are made. On the other band, cultural beliefs should
have an impact in that subjects with collectivist beliefs should be more likely to make
situational inferences whereas individualistically oriented subjects should preferentially
make dispositional inferences.
Method
Paradigm and Stimuli
The general paradigm of the three studies was adapted from a reaction time
paradigm used by Singer, Revlin and Halldorson (1990) in which subjects read short
vignettes consisting of sequences of behaviors. Specifically, each vignette was composed
of a context statement followed by a target statement describing the target individual’s
behavior. In pilot testing, we developed target statements that did not produce strong
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
9
situational or dispositional inferences in and of themselves. And for each o f these target
statements, we developed (1) a situational context that consistently elicits situational
explanations, and (2) a dispositional context that consistently elicits dispositional
explanations. Thus, when the target sentences are paired with a situational context, they are
likely to induce a situational inference. Similarly, when the target sentences are paired with
a dispositional context, they should lead to dispositional inferences.
After reading each vignette, subjects were asked to answer a question which could
always be answered by either YES or NO. The questions asked about general world
knowledge that subjects would have needed to activate to understand the vignette, such as a
causal link between two events in the vignette. Thus, if the subjects are faster in responding
to dispositional inference questions after reading vignettes that elicit dispositional
inferences, this suggests that subjects have indeed made a spontaneous dispositional
inference. Similarly, if the subjects are faster in answering situational inference questions
after reading vignettes that elicit strong situational inferences, this indicates that they have
made spontaneous situational inferences. An example of the materials we used in our
studies is shown in Tabie 1.
Table 1— Sample Vignettes.
Dispositional Context Julia's husband was ten minutes late getting
home.
Situational Context Julia's husband had an affair with another
woman.
Target Statement Julia was furious.
Dispositional Inference Question Do hot-tempered people get angry easily?
Situational Inference Question Do husbands' unfaithful behaviors make
wives angry?
As you can see from this example, the target statement "Julia was furious." does
not elicit either a strong dispositional inference or a strong situational inference. However,
given the dispositional context that "Julia's husband was ten minutes late getting home",
we are likely to infer that Julia is a hot-tempered person. In making that inference we
i
i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
10
should activate the general world knowledge that hot-tempered people are easily angered.
Given the situational context that "Julia's husband had an affair with another woman", the
same target statement is likely to lead to the situational inference that "Julia is angry because
of her husband's unfaithful behavior". To make this inference we should activate the causal
knowledge that husbands’ unfaithful behaviors make wives angry. Thus, we predicted that
subjects should say YES faster to the dispositional inference question “Do hot-tempered
people get angry easily?” after they have read the dispositional context compared to the
situational context while they should be faster to the situational inference question “Do
husbands’ unfaithful behaviors make wives angry?” given the situational context.
Study 1
Overview. The basic design was a Context (Dispositional vs. Situational) X Question
(Dispositional vs. Situational) X Version (1 of 4), mixed design with context and question
as within subjects factors.
There were 7 target statements for each of the four conditions (Dispositional
Context, Dispositional Question; Dispositional Context, Situational Question; Situational
Context, Dispositional Question; Situational Context, Situational Question), resulting in a
total of 28 items. Subjects responded to one of four versions o f the stimuli, constructed by
dividing the 28 items into 4 sets of 7 and then using a Latin square to insure that each target
item occurred in each of the four conditions across subjects. The order of the vignettes was
counterbalanced across the conditions.
Because all of the responses to the experimental questions were YES, in addition to
the 28 experimental vignettes, we included 28 filler vignettes followed by knowledge
questions for which the answers were half YES and half NO, to prevent the use of
response strategies. All the vignettes and questions were programmed on the computer
using the computer software PS YSCOPE.
i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
11
Subjects. A total of 74 students at the University of Southern California participated in
the experiment for course credit.
Procedure. Upon arrival at the laboratory, each subject was seated by an experimenter
in front o f a Macintosh computer in a small cubicle. The subjects were told that we were
interested in studying how quickly people respond to questions about social behaviors. The
experimenter read the instructions which informed the subjects that they would read a
number of short vignettes consisting of sequences o f social behaviors on the computer.
Each vignette was made up of two sentences. The vignettes would be presented a sentence
at a time and they would press the SPACE BAR when they were ready to proceed to the
next sentence. When a vignette was completed, a question was Clashed on the screen. All
the questions could be answered by either YES or NO. The "YES" or "NO" keys were
indicated on the keyboard ("Z" key for NO, and the "/" key for YES). The subjects were
told to answer the questions as quickly and as accurately as possible by pressing the "YES"
or "NO" key as marked on the keyboard. After each question was answered, there was a
500ms short break, during which the words "Please wait for the next vignette." were
flashed on the screen. The first sentence of the next vignette would appear automatically on
the screen afterwards. There was also a 2 second break when they finished half of the
vignettes and the words "You can take a break now" would appear on the screen. After 2
seconds, the words "You can press any key to continue" appeared. They could press any
key and the next vignette would appear on the screen. The computer recorded their
response times to all questions in milliseconds. Subjects were debriefed at the end of the
experiment.
Results and Discussion. Reaction time data are notorious for their skewness. In
dealing with this problem, one method that has been typically employed is to throw out
a
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
12
outliers. In the present study, three subjects were excluded from the data analysis, because
their response times were either consistently too fast (lower than 1000ms) or consistently
too slow (higher than 10000ms). Since subjects could not possibly finish reading the
questions and answer them within 1000ms, nor should they have waited so long as
10000ms to answer the questions, these scores indicate that the subjects were not paying
attention to the stimuli and thus they should not be taken as accurate representations of the
true response times.
However, because the design of the study was a repeated measure design, another
problem arises. That is, some individuals had a few scores that were either too fast or too
slow but the majority of their scores were within the normal range and these individuals
could not be discarded as outliers. In order to solve this problem, we used both medians
and means to analyze the data. The reason for using medians is two-fold. First, the median
is found to be one of the most robust measures of location (Wilcox, 1996), especially when
the distribution is highly skewed (Bush, Hess, & Wolford, 1993). Second, one of the
purposes of using the repeated measure paradigm is to find a measure that is most
indicative of the subject's typical response pattern. And in our case, the score that is most
representative of the subjects' response time might be the medians of their response times
instead of their average response times.
The means and medians are presented in Table 2 and Table 3. A repeated measure
MANOVA supported our hypothesis that in addition to the main effect of context, E( 1,67)
= 8.65, j> = .004 (Median: E(l,67) = 8.45, j> = .005), there was a highly significant
interaction between context and question, E(3, 67) = 44.74, p = .000 (Median: E(3,67) =
66.44, p = .000). Subjects were much faster in responding to dispositional inference
questions given the dispositional context than given the situational context. Similarly, a
situational context speeded up their responses to situational inference questions. In other
words, the extent to which subjects made spontaneous dispositional or situational
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
13
inferences were highly affected by the type of contexts they read. The effect of version,
version by question, version by question by context, were all non significant (F(3,67) =
.86, g = .469; F(3,67) = .59, g = .626; F(3,67) = .29, g = .834 respectively). The
version by context interaction was found to be significant, E(3,67) = 3.54, g = .02,
although the nature o f this effect remains unclear.
Table 2 — Mean Reaction Times in Study 1. (in milliseconds) N=71
context Question
Dispositional Situational
Dispositional 3267.76 4333.05
(892.97) (1365.74)
Situational 3633.84 3588.36
(1106.88) (1139.71)
* The numbers in parenthesis are standard deviations.
Table 3 — Median Reaction Times in Study 1. (in milliseconds) N=71
Context Question
Dispositional Situational
Dispositional 2976.32 4043.45
(832.59) (1171.57)
Situational 3417.62 3235.23
(1137.26) (880.55)
* The numbers in parenthesis are standard deviations.
We also found an unexpected significant effect of question, F(l,67) = 39.25, g =
.000 (Median: E(1.67) = 47.77, g = .000). We suspected that part of the reason might be
due to the fact that situational inference questions were on the average 1.5 words longer
than dispositional inference questions. The differences in the length of the questions and
hence the differences in reading times might be responsible for the significant effect of
question.
In order to test this possibility, we performed a separate study in which 11 subjects
were asked to read the questions without the contexts. Their reading times were recorded in
the computer. The mean reading time for a dispositional inference question was 2317.79ms
while the mean reading time for a situational inference question was 2674.21ms. We then
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[
i
14
subtracted the reading time of each individual item from its mean response time and came
up with an adjusted score. (Since the pattern of results were essentially the same for means
and medians, we only adjusted for the mean response times in the reading time analysis.)
The adjusted scores are shown in Table 4.
Table 4 — Mean Reaction Times Adjusted for Reading Times (in milliseconds) N=71
Context Question
Dispositional Situational
Dispositional 956.40 1671.79
(923.74) (1362.98)
Situational 1284.84 921.97
(1081.73) (1137.93)
* The numbers in parenthesis are standard deviations.
A repeated measure MANOVA yielded a close-to significant effect of question,
F(l,67) = 3.57, ^ = .063, while the effect of context ( F(l, 67) = 8.65, g = .004) ) and the
interaction between context and question ( F(3,67) = 44.74, g = .000 ) remained highly
significant. As the adjusted means in the matching contexts conditions do not differ from
each other (M = 956.40, M = 921.97 respectively), this marginally significant effect of
question is primarily caused by the difference in the mismatching contexts conditions.
Specifically, subjects in the situational context, dispositional question condition (M =
1284.84) tend to be a lot faster than subjects in the dispositional context, situational
question condition (M = 1671.79).
The interpretation of this result is not immediately obvious. There are several
possible explanations: First, because of the nature of the current experimental design, we
could not perform a covariate analysis to see how much of the significant effect of question
could be accounted for by the difference in reading times. Subtracting the reading times
from the reaction times could not give us an estimate of the extent to which differences in
reading times accounted for the variance. Second, by looking at the vignettes, we
discovered that a lot of times the trait terms used in the dispositional inference questions
i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
were synonyms of the target behaviors. Take the vignette in Figure 1 for example, the trait
term “hot-tempered” is a synonym of the target behavior “be furious”. In other words,
subjects could have been primed for the concept after reading the target sentence, thus
resulting in a faster than expected response to the dispositional inference questions. Lastly,
the possibility exists that the current result is not due to a lack of appropriate statistical tools
or to the specifics of the vignettes we used, but that there is still something special about
dispositional inferences and they are at least to some extent more spontaneous than
situational inferences. We will further examine these possibilities in our subsequent
studies.
Study 2
Overview. Study 2 aimed to investigate the effect of cognitive load, and preexisting
cultural beliefs on the extent to which people make spontaneous trait and situational
inferences. The basic paradigm of the study was the same as Study 1 except that we added
another two between subject factors, cognitive load and cultural beliefs, resulting in a
Context (Dispositional vs. Situational) X Question (Dispositional vs. Situational) X
Version (1 of 4) X Cognitive load (High load vs. Low load) X Cultural beliefs
(Individualism vs. Collectivism), mixed design with context and question as within
subjects factors.
We predicted that cognitive load should not have a differential effect on the extent to
which dispositional and situational inferences were made, while cultural beliefs should
have an effect. Specifically, subjects with more collectivist beliefs should be more likely to
make spontaneous situational inferences and subjects with individualistic beliefs should be
more likely to make spontaneous dispositional inferences.
Subjects. A total of 99 students at the University o f Southern California participated in
i the experiment in return for course credit.
i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
16
Procedure. Most of the procedure was the same as in Study 1. However, at the
beginning of the experiment, subjects were also told that while they were performing the
reaction time task, they were also required to remember a set of numbers and rehearse them
continuously. They were told that because we were interested in learning how people could
do multiple things at the same time, these two tasks were equally important. Each subject
was presented with an eight digit number which was either easy to remember (e.g.
88888888) or difficult to remember (e.g. 26513847). Instead of having one break when
subjects finished half of the vignettes, there were seven short breaks, each of which lasted
for 2 seconds before the words "You can press any key to continue." appeared on the
screen. Subjects were asked to record the number they received previously and to
remember a new number during the break and then press any key to continue. At the end of
the experiment, subjects filled out the INDCOL scale developed by Triandis (1988).
Subjects were debriefed afterwards.
M anipulation Check. The percentage of numbers that were perfectly recalled indicated that
subjects who were given the easy numbers remembered the numbers (M = .95)
significantly better than subjects who were given the difficult numbers (M = 58), t(l,95) =
11.71, p = .000, 2-tail. It is important to note, however, that even in the difficult number
condition, the mean percentage of completely correct responses fM = .58) was pretty high.
The implication of this will be discussed in the next section.
R esults and Discussion. A s in Study 1, four outliers were excluded from analysis for
having response times that were either consistently too fast (lower than 1000ms) or too
slow (higher than 10000ms).
i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
17
Table 5 — M ean R eaction Times in Study 2. (in m illiseconds) N=95
Context Question
Dispositional Situational
Dispositional 3011.07 4085.01
(933.84) (1321.17)
Situational 3308.73 3443.13
(1159.66) (1283.94)
* The numbers in parenthesis are standard deviations.
Table 6 — Median Reaction Times in Study 2. (in milliseconds) N=95
Context Question
Dispositional Situational
Dispositional 2767.66 3787.25
(926.84) (1122.01)
Situational 3008.39 3149.08
(1009.18) (1169.53)
* The numbers in parenthesis are standard deviations.
As shown in Table 5 and Table 6, the overall pattern o f results was essentially the
same as in Study 1. There were a significant effect of context, E(l,91) = 7.50, g = .007
(Median: E (l,91) = 9.66, g = .003), and a significant question by context interaction
effect, E(3,91) = 55.69, g = .000 (Median: E(3,91) = 69.31, g = .000).
The adjusted means are shown in Table 7. Like Study 1, given the matching
contexts, subjects did not differ in their response times to the dispositional and situational
questions (M = 688.08, M = 744.68) while their response times differed in the two
mismatching context conditions (M = 1427.45, M = 1014.82), except that the question
effect was more significant than in Study 1 after adjusting for differences in reading times
(F(l,91) = 13.96, g = .000). However, since we failed to include a no context condition in
both Study 1 and Study 2, we did not have an appropriate baseline to assess whether this
difference was caused by an unusually fast response in the situational context, dispositional
question condition, or an usually slow response in the dispositional context, situational
question condition. W e will deal with this problem in Study 3 by adding to the existing
paradigm a no context condition.
i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
\
18
Table 7 — M ean Reaction Times Adjusted for Reading Times (in m illiseconds) N=95
Context Question
Dispositional Situational
Dispositional 688.08 1427.45
(922.13) (1271.70)
Situational 1014.82 744.68
(1169.61) (1254.29)
* The numbers in parenthesis are standard deviations.
As predicted, the cognitive load manipulation did not have a differential effect on
the extent to which subjects made dispositional versus situational inferences. The main
effect of cognitive load was non significant F(l,91) = .78, p = .378, as were all the other
interaction effects (loadby context, F(l,91) = .73, £ = .396; load by question, F(l,91) =
.72, p = .398; load by context by question, F(l,91) = .00, p = .991). The results were
similar for the median analysis. This finding contradicts Gilbert's prediction that trait
inferences are more spontaneous than situational inferences. According to Gilbert,
dispositional inferences should be less affected by cognitive load than situational
inferences, which is apparently not the case here. However, one could argue that our
cognitive load manipulation might not be a particularly effective one and thus may not be
sensitive enough to detect the difference. People might argue that 58% recall means that
many subjects aren’t doing the task properly. We seek to rule out this possibility in our
third study by using a different cognitive load manipulation.
A median split was performed on the INDCOL scale to divide the subjects into two
groups, those who were more individualistically oriented and those who were more
collectivistically oriented. We then used the repeated measure MANOVA to test the effect of
cultural beliefs and found non significant effects of INDCOL, E(L 92) = 1.33, p = .252,
and all possible interactions (INDCOL by context, E (l,92) = .42, p = .521; INDCOL by
question, E(1.92) = 3.64, p = .060; INDCOL by context by question, E(l»92) = .64, p =
.424). The results were essentially the same for the median analysis.
i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
*
r
*
19
Examining the pattern of results in Table 8, we found that collectivists tend to be
slower than individualists in responding to the questions regardless of the question type,
which has contributed to the marginally significant INDCOL by question interaction effect.
We suspect that most of the subjects who hold collectivist beliefs are students from Asian
and Latin American countries, for whom English is not their first language. Language
difficulty may thus have caused a slower response for collectivists. Unfortunately, as we
did not collect data on the ethnicity of the subjects, we could not directly test this
hypothesis. Future studies should look into this possibility.
Table 8 — Mean Reaction Times by Cultural Beliefs in Study 2. (in milliseconds) N=95
Context Question
Dispositional Situational
Individualist Collectivist Individualist Collectivist
Dispositional 2920.86 3042.88 3887.26 4151.16
(851.64) (966.00) (1119.16) (1322.66)
Situational 3209.69 3314.35 3157.95 3616.54
(1113.87) (1093.87) (960.88) (1433.39)
* The numbers in parenthesis are standard deviations.
Although the finding that collectivists and individualists did not differ in the extent
to which dispositional and situational inferences were made runs counter to our prediction,
caution must be taken in interpreting the result. First, a non significant finding does not
necessarily indicate the null hypothesis is true unless there is adequate power (Wilcox,
1996). Second, because our subjects were American college students, our sample might
have a restricted range, which could vastly reduce statistical power. Indeed, the mean
INDCOL scores ranged from 2.70 to 4.65 on a 1-7 point scale (the lower the score, the
more individualistic), indicating our subjects were highly individualistic. Third, Triandis's
INDCOL scale has been typically found to have very low reliability (Gudykunst,
Matsumoto, Ting-Toomey, Nishida, Kim, & Heyman, 1994). The cross-cultural literature
has so far been unable to come up with a widely recognized scale to reliably measure the
differences in cultural beliefs. Thus, it is unclear whether the non significant finding is due
i
i
i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
20
to the lack of difference between the two groups or to the error in measurement. In Study
3, we attempted to deal with the measurement problem by using a different INDCOL scale
that had better reliability.
gtpdy-3
O verview. In response to the non significant findings o f cognitive load and cultural
beliefs in Study 2, we employed a different cognitive load manipulation and a different
INDCOL scale in Study 3. We also included a no context condition in which only the target
sentences were presented without being paired with situational or dispositional contexts.
This was done for the following two reasons: First, as discussed earlier, a no context
condition would provide us with a baseline to estimate whether the unexpected difference in
the mismatching contexts conditions was due to an unusually fast response in the
situational context, dispositional question condition or an unusually slow response in the
dispositional context, situational question condition. Because mismatching contexts might
hinder the information configuration process and produce an inhibition effect, the response
times in the mismatching contexts condition should be slower than those in the no context
conditions for both situational and dispositional inference questions. Thus, if we found that
for dispositional inference questions, response times did not differ in no contexts and
mismatching contexts conditions while they did differ for situational questions, this would
indicate that the response times in the situational context, dispositional inference condition
might be unexpectedly fast, which could be explained by a priming effect caused by the use
of synonyms in the dispositional questions. On the other hand, if the response times in the
mismatching contexts were significantly slower than those in the no contexts conditions for
both situational and dispositional questions, but the difference was larger for situational
questions, this would suggest the response times for the dispositional context, situational
question might be unusually slow. Accordingly, it would be harder to make a case that this
A
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
was caused by a priming effect, while the result would be more germane to the explanation
that there is indeed something special about dispositional inferences.
Second, a no context condition could help to rule out a context checking explanation
of our results. According to the context checking explanation, when answering the
questions, subjects might have developed a response strategy in which they went back to
see if the questions fitted in with the contexts they read previously. Because the
mismatching contexts did not fit with the questions while the matching contexts did,
subjects might have been slower in answering the mismatching questions than the matching
questions. In other words, the interaction effect between question and context was caused
by a slow-down of mismatching contexts (response strategy argument) rather than a speed
up of matching contexts (information configuration argument). While this alternative
explanation would predict the same pattern of results as we got in Study I and Study 2, it
would make a different prediction when we compare the matching contexts condition with
the no contexts condition. According to the response strategy argument, subjects' response
times in the no context condition should not differ from those in the matching context
condition, because subjects should not be slowed down by checking back when there was
no context. On the other hand, we argue that because matching contexts help with the
information configuration process, they should speed up the response times compared to
the no context conditions.
The basic design of the study was a Context (Dispositional vs. No vs. Situational)
X Question (Dispositional vs. Situational) X Version (1 of 4) X Cognitive load (High load
vs. No load) X Cultural beliefs (Individualism vs. Collectivism), mixed design with
question as within subjects factors.
Subjects. A total of 156 students at the University of Southern California participated
in the experiment in return for course credit.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Procedure. The procedure was the same as that o f Study 2, except for the cognitive load
manipulation. In this study, subjects in the high load condition were asked to listen to a
tape about life on Mars when they were performing the reaction time task. They were asked
to put on headsets so that they could listen to the contents carefully and were told that they
would be given a questionnaire about the contents later. The tape was about 20 minutes
long, but subjects were asked to stop the tape once they finished the reaction time task,
which typically took about IS minutes. Subjects in the no load condition simply performed
the reaction time task without listening to the tape. All subjects were given two
questionnaires at the end of the experiment. One was an INDCOL scale developed by
Gudykunst et al. (1994), which was found to have a reliability of .7 to .8. The other was a
10-item multiple choice questionnaire designed to test subjects listening comprehension of
the Mars tape. All the items asked about the first half of the tape to ensure that subjects in
the cognitive load condition had heard all the answers before they stopped the tape.
Subjects in the no load condition were told that the questionnaire was designed for another
study. Subjects were debriefed afterwards.
M anipulation Check. The manipulation check indicated that subjects who listened to the
tape indeed were better at answering the questions than subjects who did not listen to the
tape, t( 1,141) = 3.58, g = .000, 2-tail, although the absolute means of the two groups did
not differ that much (M = 2.9167, M = 2.1268 respectively).
Results and Discussion. As in previous studies, 8 subjects whose response times
were consistently too quick or too slow were discarded as outliers and were excluded from
further analysis.
i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
23
Table 9 — M ean Reaction Times in Study 3. (in m illiseconds) N=148
Context Question
Dispositional Situational
Dispositional 2942.34 4034.75
(799.08) (1012.86)
No 3201.04 3688.78
(715.40) (1090.01)
Situational 3175.72 3201.14
(884.07) (1089.17)
* The numbers in parenthesis are standard deviations.
N=148
Context Question
Dispositional
No
Situational
Dispositional Situational
2687.08 (717.43) 3649.06 (896.75)
2865.78 (972.18) 3300.81 (880.53)
2912.71 (805.63) 2830.11 (570.08)
* The numbers in parenthesis are standard deviations.
Again, Study 3 replicated the basic findings of Study 1 and Study 2 as shown in
Table 9, Table 10 and Table 11 (for adjusted means). More importantly, the context by
question interaction was still highly significant E(2,146) = 26.50, p = .000 (Median
E(2,146) = 39.54, p = .000), indicating that there was indeed a speed-up in response when
subjects were given the matching contexts (M = 626.25, M = 530.05) as compared to no
contexts (M = 882.62, M = 1017.22). As the response strategy argument predicted no
speed-up in the matching context condition, this result clearly ruled out that explanation.
Table 11 — Mean Reaction Times Adjusted for Reading Times in Study 3.
(in milliseconds) N=148
Context Question
Dispositional Situational
Dispositional 626.25 1362.01
(800.26) (1005.57)
No 882.62 1017.22
(1072.16) (1086.08)
Situational 856.19 530.05
(855.43) (714.58)
* The numbers in parenthesis are standard deviations.
i!
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
24
Moreover, comparing the means in the mismatching context conditions with those
in the no context conditions after we adjusted for the reading time, we found an expected
inhibition effect for situational questions. In other words, dispositional contexts slowed
down the response times to the situational questions compared to the no context condition
fM = 1362.01 vs. M = 1017.22). On the other hand, situational contexts did not slow
down the response times to the dispositional questions compared to the no context
condition (M = 856.19 vs. M = 882.62). We suspect that this is because the expected
inhibition effect is offset by a priming effect in the situational context, dispositional
question condition. This suggests that the significant effect of question found in the first
two studies was primarily caused by an unusually fast response in the situational context,
dispositional question condition rather than an unusually slow response in the dispositional
context, situational question condition. This renders some indirect support for the
explanation that subjects might have been primed by the synonyms in the dispositional
questions, leading to the unexpected pattern of results.
The effect of cognitive load also replicated the finding in Study 2. There was no
significant effect of cognitive load £(1,141) = 1.14, p = .287, and it did not interact with
any of the other variables (toad by context, £(2,137) = .36, p = .701; toad by question,
E(l,141) = .13, p = .716; toad by context by question, E(2, 137) = 1.00, p = .371).
Again, this replication does not preclude the possibility that a more powerful toad
manipulation might show a different result. The fact that the control group and the
experimental group did not differ that much in the number of correct answers in the
manipulation check indicates that our toad manipulation might not have been that effective.
Similarly, we failed to find any significant effect of INDCOL, E(l,146) = 2.16, p =
. 144, or any of the possible interaction effects — INDCOL by Question, E(l, 146) = 1.34,
p = .248; INDCOL by Context, E(2,142) = .33, p = .721; INDCOL by Context by
Question, E(2,142) = .28, p = .760. As discussed earlier, the nature of this finding is
jli
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
«
25
ambiguous. Even though we employed a relatively reliable scale, the restricted range of
subjects we had ( The range o f scores was from 3.20 to 6.07 on a 1-7 point scale, with 1
indicating extremely collectivist, 7 indicating extremely individualistic. Median was 4.40. )
could still contribute to the non significant result
General Discussion
In Study 1, based on a reaction time paradigm used by Singer, Revlin and
Halldorson (1990), subjects read target sentences paired with either a dispositional context
that elicits dispositional explanations or a situational context that elicits situational
explanations and then answered dispositional or situational inference questions. Subjects
responded to the questions significantly faster when the contexts matched the questions
than when the contexts did not match the questions. This pattern of results was replicated in
both Study 2 and Study 3. Taken together, the consistent pattern across the three studies
that matching contexts significantly speeded up the response times to both dispositional
inference questions and situational inference questions provides strong support for the
hypothesis that situational inferences can occur spontaneously.
Effect of Cognitive Load
Experiment 2 investigated the effect of cognitive load on the extent to which
situational and dispositional inferences were made and showed that contrary to Gilbert's
prediction, situational inferences were not more affected by cognitive load than
dispositional inferences. However, it also raised the concern that our cognitive load
manipulation may not have been particularly effective. Study 3 was done to address this
issues. We used a different cognitive load manipulation in Study 3 and again found a non
significant result. Although the possibility exists that the load manipulation used in Study 3
was still not sensitive enough, these two studies offered some converging evidence that
dispositional and situational inferences may not differ in their spontaneity, at least not as
much as some social psychologists thought (Gilbert, 1989). This is also reminiscent of
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
26
Krull's (1993) finding that given situational inferential goals, situational inferences are less
affected by cognitive load than dispositional inferences whereas given dispositional
inferential goals, dispositional inferences are less affected by cognitive load than situational
inferences. However, the current research has extended Krull's finding from intentional
social inferences to spontaneous social inferences.
Effect o f Cultural Beliefs
Both Study 2 and Study 3 revealed a non significant effect of cultural beliefs on the
extent to which dispositional and situational inferences were made, although it remained
unclear whether this finding was due to a lack of appropriate measurement tools or to the
restricted range of subjects we had. While this finding does not confirm our hypothesis, it
is not completely at odds with the existing literature which have often yielded ambiguous
results in the effect of cultural differences on spontaneous dispositional inferences. For
instance, Newman (1993) reported two studies that reported differences in spontaneous
trait inferences due to differences in preexisting belief systems at the individual level.
Newman found that the expected effect of spontaneous trait inferences appeared only
among individualistically oriented men, and not among individualistically oriented women
or nonindividualistically oriented men and women. Similarly, Duff, Newman and Wolsko
(1995) showed that individualism and trait-cued recall correlated positively among men but
not among women. Moreover, sex differences occurred for the delayed cued recall measure
but not for the immediate recognition probe measure. The inconsistencies and ambiguities
in this area suggest there might be other variables, such as gender, which could moderate
the effect of cultural beliefs on social inferences. Future research should aim to identify
these potential moderators.
Are Dispositional Inferences More Spontaneous Than Situational Inferences?
So far, the evidence has shown that spontaneous situational inference does occur.
But are situational inferences as spontaneous as dispositional inferences, or as Lupfer et al.
JL
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
27
(1990) suggested, are they less spontaneous than dispositional inferences? The result of
our studies seems to support the former. Across the three studies, we found a consistent
pattern that under matching contexts conditions, subjects did not differ in their response
times to the dispositional and situadonal questions, suggesting that under the right
information configuration, situational inferences can be as spontaneous as dispositional
inferences. However, complication results as we interpret the finding that under
mismatching contexts conditions, subjects were consistendy slower in responding to
situational inference questions than to dispositional inference questions. This pattern of
results is open to two alternative explanations. First, dispositional inferences are more
spontaneous than situational inferences and responses to dispositional questions were
affected by a mismatching context to a lesser degree than situational questions, thus a
mismatching context had less of a slow-down effect on the dispositional questions.
Second, the mismatching situadonal contexts did slow down the responses to the
dispositional questions, but this slow-down effect was overridden by a priming effect
caused by the use of synonyms in the dispositional questions, resulting in an unusually fast
response in the situational context, dispositional question condition.
Two findings render the former explanation unlikely: First, if dispositional
inferences were indeed more spontaneous than situational inferences, response times to
dispositional questions would have been faster than those to situational questions when
there was no context. On the contrary, we found that once we adjusted for the difference in
reading times, response times to dispositional and situational questions did not differ under
no context condition. Second, if dispositional inferences were more spontaneous than
situational inferences, cognitive load would have more of an effect on situational inferences
than on dispositional inferences. Again, this is not what we found in Study 2 and Study 3.
Taking these findings into account, it is unlikely that dispositional inferences are
more spontaneous than situational inferences. Rather, it is possible that the priming effect
i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
I .
1 )
28
has contributed to the unusually fast response to the dispositional questions. Future
research should be wary of such possibilities and try to use experimental stimuli that do not
produce priming effects.
Conclusion
Since the discovery of the fundamental attribution error, social psychology has long
been biased toward the study of trait inferences as exemplified in Gilbert's three stage
model of social inference and the research on spontaneous trait inferences. Few attempts
have been made to examine situational inferences. Now after thirty years of research on the
fundamental attribution error, psychology has found itself a victim of this human bias. As
Lewin (1931) pointed out, until psychology stopped thinking of behavior as the expression
of dispositional properties of the person and began to think of it as an interaction between
the person and the environment, it would be doomed to remain in its infancy. The current
research has made a first attempt to fill in this gap by offering converging evidence that
given the right information configuration, situational inferences can be as spontaneous as
dispositional inferences. Hopefully, we can finally grow out of our prolonged infancy and
into a prosperous adulthood.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
29
References
Bush, L.K., Hess, U., & Wolford, G. (1993). Transformations for Within-Subject
Designs: A Monte Carlo Investigation. Psychological Bulletin, 3, 566-579.
Duff, K.J., Newman, L.S., & Wolsko, C. (1995). Culture and spontaneous inferences.
Paper presented at the annual meeting of the Midwestern Psychological
Association, Chicago, IL.
Gilbert, D.T. (1989). Thinking lightly about others: Automatic components of the social
inference process. In J.S. Uleman & J.A. Bargh (Eds.), U nintended thought
(pp. 189-211). New York: Guilford.
Gilbert, D.T., Peelham, B.W., & Krull, D.S. (1988). On cognitive busyness: When
person perceivers meet persons perceived. Journal o f Personality and Social
Psychology, 54, 733-740.
Gudykunst, W.B., Matsumoto, Y., Ting-Toomey, S., Nishida, T., Kim, K., & Heyman,
S. (1994). Measuring se lf construals across cultures. Paper presented at the
International Communication Association convention, Sydney, Australia.
Heider, F. (1958). The psychology o f interpersonal relations. New York: Wiley.
Hofstede, G. (1980). Culture’ s consequences: International differences in work-related
values. Beverly Hills, CA: Sage.
Krull, D.S. (1993). Does the grist change the mill? The effect of the perceiver’s inferential
goal on the process of social inferences. Personality and Social Psychology
Bulletin, 19, 340-348.
Krull, D.S. (1996). On thinking first and responding fast: Flexibility in social inference
processes. Personality and Social Psychology Bulletin, 9, 949-959.
Lewin, K. (1931). The conflict between Aristotelian and Galileian modes o f thought in
contemporary psychology. Journal o f General Psychology, 5, 141-177.
Lupfer, M.B., Clark, L.F., & Hutcherson, H.W. (1990). Impact of context on
spontaneous trait and situational attributions. Journal o f Personality and Social
Psychology, 58, 239-249.
Markus, H.R., & Kitayama, S. (1991). Culture and the self: Implications for cognition,
emotion, and motivation. Psychological Review, 98, 224-253.
Miller, J.G. (1984). Culture and the development of everyday social explanation. Journal
o f Personality and Social Psychology, 46, 961-978.
Newman, L.S. (1993). How individuals interpret behavior: Idiocentrism and spontaneous
trait inference. Social Cognition, 11, 243-269.
Quattrone, G.A. (1982). Overattribution and unit formation: When behavior engulfs the
person. Journal o f Personality and Social Psychology, 42, 593-607.
£
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
30
Read, S.J., & Miller, L.C. (1993). Rapist or “regular guy”: Explanatory coherence in the
construction of mental models o f others. Personality and Social Psychology
B ulletin, 19, 526-541.
Reeder, G.D. (1993). Trait-behavior relations and dispositional inference. Personality and
Social Psychology Bulletin, 19, 526-541.
Ross, L.D. (1977). The intuitive psychologist and his shortcomings: Distortions in the
attribution process. In L. Berkowitz (Ed.), Advances in experim ental social
psychology (Vol. 10, pp. 174-221). New York: Academic Press.
Singer, M ., Revlin, R., & Halldorson, M. (1990). Bridging inferences and enthymemes.
In A.C. Graesser & G.H. Bower (Eds.), Inferences and text com prehension. New
York: Academic Press.
Thagard, P. (1989). Explanatory Coherence. Behavioral and Brain Sciences, 12,435-467.
Thagard, P. (1992). Conceptual Revolutions. Princeton, NJ: Princeton University Press.
Triandis, H.C. (1988). Collectivism and development. In D. Sinha & H.S. Kao (Eds.)
Social values and development: A sian perspectives (pp. 285-303). Newbury Park,
CA: Sage.
Triandis, H.C. (1989). The self and social behavior in differing cultural contexts.
Psychological Review, 96, 506-520.
Triandis, H.C. (1990). Cross-cultural studies of individualism and collectivism. In J.
Berman (Ed.), Nebraska Symposium on M otivation, 1989 (pp.41-133). Lincoln:
University of Nebraska Press.
Trope, Y. (1986). Identification and inferential processes in dispositional attribution.
Psychological Review, 93, 239-257.
Uleman, J.S., Newman, L.S., & Moskowitz, G.B. (in press). People as flexible
interpreters: Evidence and issues from spontaneous trait inference. In M.P. Zanna
(Ed.), Advances in experimental social psychology (Vol.29). San Diego: CA:
Academic Press.
Uleman, J.S., Newman, L., & Winter, L. (1987). M aking spontaneous tra it inferences
uses some cognitive capacity at encoding. Unpublished manuscript, New York
University, New York, NY.
Uleman, J.S., Newman, L., & Winter, L. (1992). Can personality traits be inferred
automatically? Spontaneous inferences require cognitive capacity at encoding.
C onsciousness and Cognition, 1, 77-90.
Wilcox, R.R. (1996). Statistics fo r the social sciences. San Diego, CA: Academic Press.
Winter, L., Uleman, J.S., & Cunniff, C. (1985). How automatic are social
judgments? Journal o f Personality and Social Psychology, 49, 904-917.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
31
Zarate, M.A., & Uleman, J.S. (1994). Lexical decision evidence that Anglos do, and
Chicanos do not make spontaneous trait inferences. Unpublished manuscript,
University of Texas at El Paso.
i
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Asymmetries in the bidirectional associative strengths between events in cue competition for causes and effects

PDF

Effect of affective reactions by an outgroup on preferences for crossed categorization discussion partners

PDF

Adaptation to sine-wave gratings selectively reduces the sensory gain of the adapted stimuli

PDF

Displaced aggression as a function of target power

PDF

Alcohol's impact on triggered displaced aggression

PDF

Insights into the nature of phonological and surface dyslexia: Evidence from a novel word learning task

PDF

Articulated thoughts regarding cognitions toward older adults

PDF

Articulated thoughts about intentions to commit anti-gay hate crimes

PDF

Applying the theory of reasoned action to condom use: The effect of immediate consequences on intention-behavior consistency

PDF

Attachment, accommodation, and the outcomes of romantic relationships

PDF

Cognitive functioning and dementia following cancer: A Swedish twin study

PDF

An examination of affective modulation, psychopathy, and negative schizotypy in college and community samples

PDF

Effects of personal resource sufficiency on perceived difficulty and desirability of earthquake preparedness

PDF

Hedonic aspects of conditioned taste aversion in rats and humans

PDF

Head injury and dementia: A co-twin control study of Swedish twins

PDF

Rationalizing risk: sexual behavior of gay male couples

PDF

Meta-analysis on the misattribution of arousal

PDF

Effects of bilateral stimulation and stimulus redundancy on performance in processing nonword letter trigrams

PDF

Cognitive Assessment Of Reactance Using The Articulated Thoughts In Simulated Situations Paradigm

PDF

Auditory word identification in dyslexic and normally achieving readers

Asset Metadata

Creator Yang, Yijing (author)

Core Title Evidence for spontaneous situational inferences

School Graduate School

Degree Master of Arts

Degree Program Psychology

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag OAI-PMH Harvest,psychology, cognitive,psychology, social

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Read, Stephen (committee chair), Hellige, Joseph (committee member), Miller, Lynn Carol (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-17991

Unique identifier UC11337003

Identifier 1387849.pdf (filename),usctheses-c16-17991 (legacy record id)

Legacy Identifier 1387849.pdf

Dmrecord 17991

Document Type Thesis

Rights Yang, Yijing

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA