Content uploaded by Ira Roseman
Author content
All content in this area was uploaded by Ira Roseman on Mar 31, 2015
Content may be subject to copyright.
Appraisal Determinants of Emotions: Constructing
a More Accurate and Comprehensive Theory
Ira J. Roseman
Rutgers University, Camden, USA
Ann Aliki Antoniou
New School for Social Research, New York,USA
Paul E. Jose
Loyola University Chicago, USA
In order to move toward a more accurate, complete, and integrative theory of
the causes of emotions, empirical evidence relevant to a recently proposed
appraisal theory was examined, and hypothes es from several alternative
appraisal theories were compared and tested. Given questions that focused
on the cognitive causes of emotions rather than their phenomenolo gic al
contents, 182 subje cts rated the appraisal dete rminants of emotion experi-
ences that they recalled. Results suggest that appraisals of unexpectedness
(not unexpected/unexpected), situational state (motive-inconsistent/motive-
consiste nt), motivational state (aversive/appetitive), probability (uncertain/
certain), control potential (low/high), problem source (non-characterologi-
cal/characterological factors), and agency (circumstances/other person/self ),
differentiate a large number of widely-discussed emotions. These results are
used to formulate a revised, empirically grounded, and more comprehensive
model that speci® es which appraisals cause 17 different emotions.
IN T R O D U C T IO N
In recent ye ars, a number of theories have been offered which attempt to
specify the particular appraisals of events that elicit different emotions (e.g.
Arnold, 1960; Frijda, 1986 ; Lazarus, 1991a; Oatley & Johnson-Laird,
COGNITION AND EMOTION, 1996,
10
(3), 241± 277
Requests for reprints should be sent to Dr Ira Roseman, Department of Psychology,
Rutge rs Universi ty, Camden, NJ 08102, USA. Email: roseman@crab.rutgers.edu.
We are grateful to Martha Gordon for assistance wi th data collection, and to D’ Arcy
Myjer, Roger Tourangeau, Jerry Parrott, and two anonymous reviewers for helpful comments
on previous drafts of this paper.
q
1996 Psychology Press, an imprint of Erlbaum (UK) Taylor & Francis Ltd
2 4 2
R O S E M A N , A N T O N I O U , J O S E
1987; Ortony, Clore, & Collins, 1988; Roseman, 1979, 19 84; Scherer,
1984, 1988; Smith & Lazarus, 1990; Weiner, 1985).
1
Given the proliferation of appraisal theories, our understanding of the
causes of emotions may now be advanced by careful assessments of
empirical e vidence. Proponents of any particular appraisal theory may be
asked to retain only those portions of that theory which are supported by
existing or newly gathered data; discard appraisal dimensions or hypothe-
sised appraisal-emotion relationships that have failed to ® nd support in
soundly conducted research; and incorporate appraisals from other theories
that can explain the elicitation of unaccounted for emotions whose exis-
tence and distinctiveness have been documented, with con¯ icts among
ri val apprai sal hypotheses resolved empiric ally. Indeed, results from prior
comparative e mpirical tests (see Roseman, Spindel, & Jose, 1990) suggest
that it might be pro® table to combine elements from different theories in
order to construct the best overall model of the appraisal determinants of
emotions.
Reisenz ein and Hofmann (199 3) note that it has been dif® cult to
evaluate the relative merits of different appraisal theories bec ause the
studies establishing their empirical base differ with regard to such factors
as the type of data gathered (e.g. remembere d vs. hypothetical events), the
particular emotions studied, the way in which apprai sals and emotions are
measured, and the statistical methods used to assess support for hypotheses .
As a solution to this proble m, Reisenzein and Hofmann (1993, p. 272)
point to studies in which competing appraisal dimensions are compara-
tively evaluated within the same methodological paradigm so as to control
potential confounds.
This article reports results from such a study, and shows how these
® ndings can be used to develop a more accurate, comprehensive, and
potentially integrative appraisal model.
A T h e o r y o f A p p r a is a l-E m o tio n R e la tio n s h ip s to b e
E v a lu a te d a n d R e v i s e d
Our starting point is the appraisal theory proposed in Roseman (1984),
which is diagrammed in Fig. 1. In thi s ® gure, the appraisal elements that
combine to elicit particular emotions are shown around the borders of the
1
Other theorie s of appraisal determinants of various emotions have also bee n proposed,
and empiri cal inve stigations of cognition and emotion have made signi® cant contributions to
the development of appraisal theory (see, e.g. Kemper, 1978; Manstead & Tetlock, 1989;
Mauro, Sato, & Tuc ker, 1992; Reisenzein & Hofmann, 1990; Smith & Ellsworth, 1985;
Solomon, 1976; Stein & Levine, 1987; and othe rs cited in Clore, S chwarz, & Conway, 1994;
Reisenzein & Hofmann, 1993; and Scherer, 1988).
chart, and the emotions elicited by the different combinations of appraisals
are given in the boxes. As wil l be discussed later, many, but not all, of the
appraisal-emotion interrelationships shown in Fig. 1 are found in several
current appraisal theories (for detailed comparisons of this with other
theories, see Clore et al., 1994; Roseman et al., 1990; Scherer, 1988).
Which of these Appraisal-Emotion Relationships have been Sup-
ported?
Research to date has provided evidence that: (a) positive emo-
tions are elicited by e vents appraised as motive-consistent, and negative
emotions are elicited by events appraised as motive-inconsistent (Roseman,
Dhawan, Rettek, Naidu, & Thapa, 1995; Roseman et al., 1990; Scherer,
1993; Wallbott & Scherer, 1988); (b) joy and sadness are el icited by events
appraised as relevant to appetitive motives, and relief, distre ss and disgust
are elicited by eve nts appraised as relevant to aversive motives (Roseman,
1991; Roseman et al., 1990); (c) hope and fear are elicited by events
appraised as uncertain, and joy is elicited by (motive-consistent) events
appraised as certain (Frijda, 1987; Roseman, 1991; Roseman, et al., 1990;
Smith & Ellsworth, 1985); and (d) pride, shame, guilt, and regret are
elic ited by events appraised as caused by the self; and love, dislike,
and anger are elicited by events appraised as caused by other persons
C O N S T R U C T IN G A N I M P R O V E D A P P R A IS A L T H E O R Y
2 4 3
F I G .1 .
Hypothesised structure of the emotion system (from Roseman, 19 84).
2 4 4
R O S E M A N , A N T O N I O U , J O S E
(e.g. Reisenzein & Hofmann, 1990; Rose man, 1991; Roseman et al., 1990;
Weine r, Graham, & Chandler, 1982).
Unsupported Relationships, Gaps in Theory, and Alternative Hypo-
theses.
Other appraisal-emotionrelationshipsdepicted in Fig. 1 have not
been supported.Forexample, it appears that,contraryto predictions,(a) an
appraisalof extremeuncertainty(``unknown’’ probabilityof occurrence)does
not elicit surprise (see Fridja,Kuipers,& ter Schure, 1989; Rosemanet al.,
1990); (b) anappraisalof high power(perceiving theself as strongratherthan
weak in a situation)does not elicit frustration,anger, and regret, ratherthan
other negative emotions (Rosemanet al., 1990; Wallbott& Scherer, 1988);
and (c) the appraisal that an event was caused by circumstances beyond
anyone’ s control, rather than by other persons or the self, is not necessary
for the experience of joy, relief, surprise, disgust, or frustration(Roseman,
1991; Rosemanet al., 1990; for similar ® ndings on ``situationalcontrol’’ see
Manstead& Tetlock, 1989; Smith & Ellsworth,1985).
The theory shown in Fig. 1 is also incomplete in several respects. For
example, it does not specify appraisals that distinguish distress from
disgust, or shame from guilt (see Fig. 1; for evidence on the distinctive-
ness of these emotion states, see, e.g. Fridja et al., 1989; Izard & Malatesta,
1987; Roseman, Wiest, & Swartz, 1994; Tangney, 1993). In addition, the
emotion of contempt, which has been shown to have a distinctive cross-
culturally recognisable facial expression (Ekman & Friesen, 1986; Ekman
& Heider, 1988; Izard & Haynes, 1988), is missing from the theory.
To develop an improved appraisal model, we attempted to incorpora te
appraisal components proposed in other theories to account for those
emotions seen to be inadequately explained by the theory show n in Fig. 1.
Because a variety of appraisals have been proposed to explain elicitation of
the same emotions, we tested alternative hypothes es from several promi-
nent theories.
2
Im p ro v in g M e a s u re m e n t o f th e C a u s e s o f P a r t ic u la r
E m o ti o n s
We set out to test hypotheses about appraisal determinants using proce-
dures designed to address a possibl e methodological problem with prior
re search. As a number of authors have pointed out (see Fridja, 1993;
2
Of course, no one study can examine al l the appraisals that have been or might be
proposed. From current theories, we selected appraisal compone nts that have attracted
particular attention in the literature, and that speci® cally addre ss the demonstrated inade-
quacies of the model presented in Fig. 1. In future research, it may be worth examining other
appraisals and appraisal-emotion relationships.
Parkinson & Manstead, 1992; Roseman et al., 1990), many w idely cited
appraisal studies may have failed to identify correctly the causes of
emotions because their experimental instructions ``asked for ratings
characterising the content rather than the cause of the experienced emo-
tion’’ (Parkinson & Manstead, 1992, p. 129). In the light of this objec-
tion, our approach was to ask subjects to rate the cause of an emotion
rather than the thoughts that they had once the emotion had begun.
Similar procedures have been successfully employed to elic it appraisal-
re levant emotion antecedent information in interviews conducte d by
Reisenzein and Hofmann (1993).
The procedures w e followed we re also designed to help avoid another
pitfall. As Scherer (1993) has observed, emotion episodes described by
subjec ts in appraisa l studies of ten encompass several emotions (see, e.g .
Smith & Ellsworth, 1987 ), each with its own appraisal determinants.
Unless the subject is instructed to specify the appraisals that are relevant
to the primary emotion under investigation, appraisals relevant to other
emotions may be reported, obscuring true appraisal-emotion relationships.
For example, Roseman et al. (1990) discuss a subject recalling an experi-
ence of relief whose rating scale responses re¯ ected an initial uncertainty
about whether her mother might have cancer (which led to an initial fear
re sponse) rather than the subsequent certainty that she did not have the
disease (w hich was what elicited relief). Roseman et al. recommended
correcting this problem by as king subjects about the
appraisa ls
that led
to their emotion, rather than by asking them about the
even t
that led to their
emotion, and we pursued this strategy within the present study.
To see whether appraisal-emotion relationships identi® ed in prior
re search would still be supported given more precise measurement of
appraisal causes of particular emotions, we attempted to replicate them
in addition to testing new hypotheses. Thus the present study tested all
appraisal-emotion relationships proposed as potential components of a
re vised appraisal theory.
S p e c i® c O b j e c tiv e s o f t h is R e s e a r c h
1.
What best explains the elicitation of surprise?
Instead of extreme
uncertainty
(Roseman, 1984), it has be en proposed that
unexpectedness
(Izard, 1977; Sche rer, 1984; cf. Charl esworth, 1969),
nove lty
(Sc herer,
1984), or
unfamiliarity
(Scherer, 1988) may be the crucial elicitor(s) of
surprise. Which, if any of them, most accurately distinguishes surprise
from other emotions?
2.
What differentiates positive from negative emotions?
As mentioned
earlie r, prior investigations have found that appraisals of situational state
C O N S T R U C T IN G A N I M P R O V E D A P P R A IS A L T H E O R Y
2 4 5
2 4 6
R O S E M A N , A N T O N I O U , J O S E
(whether events are consistent vs. inconsistent wi th a person’ s motives)
appear to determine whether positive vs. negative emotions are experi-
enced. But would this relationship be replicated with techniques that
focus subjects on the cause s rather than the contents of their emotion
experiences?
3.
What differentiates contact-regulating from distance-regul ating emo-
tions?
Roseman (1994b) has proposed that the emotions of joy and sadness
re spectively initiate and terminate movement toward stimuli, whereas the
emotions of distress and relief initiate and terminate movement away from
stimuli. Here we sought to test whether new appraisal measurement
procedures would support prior indications that an appraisal of motiva-
tional state distinguishes these two emotion groups, that is, appraising an
event as relevant to appetitive (reward-maxi mising) motive s elicits joy or
sadness, and appraising an event as relevant to aversive (punishment-
minimi sing) motives elicits distress or relief.
4.
What differentiates reactive from preparatory emotions?
As distinct
from the emotions of joy, sadness, distress, and re li ef, which are claimed to
initiate or terminate actions, hope and fear may involve preparatory
re sponses such as anticipation and vigilance. As noted earlie r, certainty
appraisals have been associated with joy, and uncertainty appraisals with
fear and hope. However, it is not clear whether certainty is also an elicitor
of sadness, distress, or relief (see Roseman et al., 1990, 1995), and some
studies have failed to replicate the relationship between uncertainty and
hope (El lsworth & Smith, 1988; Frijda et al., 1989; Mauro et al., 1 992;
Smith & Ellsworth, 1987). Would focusing subjects on the appraisa ls
leading to each speci® c emotion help sort out these appraisa l-e motion
linkages, as has been suggested?
5.
What best diffe rentiates emotions that contend with stimuli from
emotions that accommodate to them?
Instead of an appraisal of one’ s
own power
in the face of a negative event (e.g. Kemper, 1978; Roseman,
1984), different theories have proposed that emotions such as anger result
in part from an appraisal of either (a)
stimulus powerlessness
, the (low)
potency of the object to which one is reacting (see Osgood, May, & Miron,
1975) or noninsurm ountable degree of dif® culty involved in attaining or
re jecting the object ( Arnold, 1960); or (b)
stimulus controllability
, the
controllability (Schere r, 1988; Sel igman, 1975) or modi® ability (Frijda,
1986) of an event; or (c)
stimulus controllability by the self
, the extent to
which one can control or modify an event by one’ s own actions (Frijda,
1986; cf. Bandura, 1977); or (d)
coping potential
, the degree to which one
can cope with (Lazarus & Smith, 1988; Scherer, 1988), adjust to (Scherer,
1988), or do something about an event. Moderate to high power, or control,
or coping potential, has also been proposed as an antecedent of contempt
(Scherer, 1988), disgust (se e Rozin & Fallon, 1987), guilt (Schere r, 1988),
shame (Scherer, 1988), and frustration (Roseman, 1984), as opposed to
emotions such as sadness, fear, dislike, and re gret, which may result i n part
from an appraisal that little can be done about a negative event.
6.
What differentiates
``
attack emotions’’ from
``
exc lusi on emotions? ’’
Roseman (1994b) has argued that frustration, anger, and guilt involve
re sponses that attack a stimulus (move against it), whereas disgust, con-
tempt, and shame involve response s that exclude a sti mulus (move it
away). According to some theorists, an appraisal of problem source
(whether a negative event is attributed to be havioural vs. characterologic al
factors) may help to differentiate these emotions (which were not well
distinguished in Fig.1).
For example, Lewi s (1971) proposed that the stimulus to guilt is a
negative event, act, or thing for which one holds oneself responsible,
whereas the stimulus to shame is a perceived de® ciency of self (see
Janoff-Bulman, 1979, for discussion of a more general behavioural vs.
characterological self-blame dimension). Applied to negative events
caused by other persons, the behavioural vs. characterological distinction
might also differentiate anger from contempt: If another person’ s action
causes a negative event, anger is experienced; but if another person’ s
character causes a neg ative event, contempt is f elt. Applied to i mpersonal
objects and events, if the world ``acts’’ badly perhaps frustration will resul t;
but if an object or event is bad ``in character’’ then disgust ensues. Thus, by
utilis ing the behavioural vs. characterological distinction, we may be able
to incorporate contempt into a more comprehensive appraisal model, and
parsimoniously differentiate the distinctive causes of guilt vs. shame, anger
vs. contempt, and frustration vs. disgust. Alternatively, it may be proposed
that appraisals of illegitimacy (moral transgression) characterise emotions
such as guilt to a greater extent than emotions such as shame (see , e.g .
Ausubel, 1955; Izard, 1991) .
7.
Which appraisal best d ifferentiates self-directed and other-person-
directed emotions?
Lazarus and Smith (1988), citing studies of guilt by
Shaver (1985) and McGraw (1987), have claimed that apprai sals of who is
re sponsible for an eve nt, rather than appraisals of who caused an event
(Kemper, 1978; Roseman, 1984; Weiner, 1985), determine whether emo-
tions are felt toward the self or other persons. But other studies (e.g.
Reisenzein & Hofmann, 1990; Roseman, 1991; Roseman et al., 1990;
Weine r et al., 1982) support the agency appraisal; and Frijda (1993) has
re cently argued that responsibility is not a cause but rather part of the
phenomenology of emotions suc h as guilt, whe reas simpler perceptions of
C O N S T R U C T IN G A N I M P R O V E D A P P R A IS A L T H E O R Y
2 4 7
2 4 8
R O S E M A N , A N T O N I O U , J O S E
agency are true antec edents. Can new methods for measuring appraisals
help resolve this question?
M E T H O D
O v e r v ie w
Apart from the procedures introduced to focus subjects on the appraisal
antecedents, rather than the cognitive contents, of their emotions (see the
later discussion of appraisalmeasures),w e followed a standardapproachfor
studying appraisals and emotions that has been used successfully by other
researchers(e.g. Frijdaet al., 1989; Manstead& Tetlock,1989; Mauroet al.,
1992; Scherer, 1993; Smith& Ellsworth,1985). Subjects were askedto recall
emotionalexperiences, describe themin theirown words,and answer a series
of questionsdesigned to measure their appraisals.
S u b j e c ts
Subjec ts were 49 male and 133 female students enrolled in classes at
Loyola University Chicago (177 undergraduates) and the New School for
Social Research (5 psychology graduate students).
3
They ranged in age
from 17 to 55, with a median age of 19. Loyola students chose to
participate as one of sev eral alternatives for ful® lling a course require-
ment. The New S chool students volunteered in response to a course
instructor’ s in-class solicitation.
S tim u lu s E m o t io n s
In this study, we investigated appraisals hypothe sised to elicit 17 emotions.
As in Roseman et al. (1990), to study differences among particular discrete
emotions, rather than gros s differences between negative and positive
emotion classes, each subject was asked to recall e ither two negative
emotions (from among fear, sadness, distress, frustration, disgust, dislike
{but not anger} toward someone, anger, contempt, regret, guilt, and shame)
or two non-negative emotions (from among hope, joy, relief, affection
toward someone, pride, and surprise). We use the term non-negative here
to encompass surprise, which might be either positive or negative (see Fig.
1). For presentation to subjects, surprise was grouped with positive emo-
tions because prior research has shown that subjects tend to recall pleasant
rather than unpleasant experiences of surprise (Roseman et al., 1990; Smith
& Ellsworth, 1985).
3
The results of key hypothesis-testing contras t analyses, shown in Table 1, are unaffected
by inclusion or exclusion of the ® ve graduate students.
So as not to overload our subjects, each one was asked for experiences
of only two emotions. Within the negative and non-negative emotion
groups, all possibl e pairs of particular emotions were represented with
approximately equal frequency, and subjects were assigned at random to
an emotion pair. Within a pair, the order of emotions was counterbalanced
across subjects. Thus, each emotion was recalled approximately hal f the
time as the ® rst experience on a questionnaire, and half the time as the
second experience .
P r o c e d u re
Written questionnaire s were administered to groups of subjects by an
experimenter. For each emotion in a questionnaire, subjects were asked
to take a f ew minutes to recall a time when the emotion was felt, describe
what happened on that occasion, tell what it was in the situation they
described that caused them to feel that particular emotion, and rate the
extent to which each of the hypothesised appraisals caused the emotion.
For a more detailed account of these procedures see Roseman et al. (1990).
The emotion experiences desc ribed by our subje cts were similar to those
obtained in other research (e.g. Roseman et al., 1990; Shaver, Schwartz,
Kirson, & O’ Connor, 1987). For example, here is a recalled experience of
anger:
In doing laundry at my sister’ s I had left my clothes there and asked her to
® nish them. I’ d told her I would come and get them later that night. When I
called to tell her I was coming, I got her answering machine telling me she’ d
gone out, but I could get my clothes at 8 the
next morning
! I was angry
because I had virtually no clothes this week and was depending on my clean
clothes for the next day. I was also not about to get up early to do something
that she should have been around for me to do that night. I was angry with
her for not staying in an extra half hour to wait.
In contrast, here is a recall ed experience of contempt:
I felt contempt for a classmate when she offered to pay me to write a term
paper for her for a high school history class. She had about a month to work
on the paper but did nothing about it. She wanted me to jeopardise my
reputation for money and to help her out.
A p p ra i s a l M e a s u re s
As discussed in the Introduction, the present paper examines appraisals of
different types of what might be termed ``unusualness’ ’ (unexpectedness,
novelty, and unfamiliarity), situational state, motivational state, probability,
C O N S T R U C T IN G A N I M P R O V E D A P P R A IS A L T H E O R Y
2 4 9
2 5 0
R O S E M A N , A N T O N I O U , J O S E
own power, stimulus powerlessness, stimulus controllability, stimulus
controllability by self, coping potential, legitim acy, problem source,
agency, and responsibility.
Three ite ms were constructed to measure each hypothesised dimension
of appraisal. Each item asked whether a particular appraisal had caused the
subjec t to feel the focal emotion in the experience that he or she had
re called. To keep subjects focused on the relevant appraisals, the stem
``My {emotion term} was caused by’’ was g iven as the ® rst part of each
appraisal item. For example, an item asses sing the extent to which an
appraisal of probability (certainty vs. uncertainty) had caused a subject
to feel relief was: ``My relief was caused by:
Being certain about the
conseque nces of SPECIFIC EVENT A
(1) . . .
Being uncertain about the
conseque nces of SPECIFIC EVENT A
(9) .’’
4
An item assessing the rela-
tionship between appraisals of stimulus contollability by the self and the
emotion of guilt was: ``My guilt was caused by:
Thinking that SPECIFIC
EVENT A was controllable by me
(1) . . .
Thinking that SPECIFIC EVENT
A was uncontrollable by me
(9). Nine-point rating scales, anchored appro-
priately at eithe r end, were used for all appraisal items. Wording for the
items and response scales was taken from or based on previously used
instruments (e.g. Roseman et al., 1990; Smith & Ellsworth, 1985) wherever
possible. Items and response scales measuring each appraisal are give n in
the Appe ndix.
For each of the two emotion experiences, subjects answered a total of 54
appraisal questions. The items were arranged in a random order, c onstant
across emotions recalled. The number and wording of appraisal items was
similar to questionnaires used in previous research (e.g. Frij da et al., 1989;
Roseman et al., 1990; Smith & Lazarus, 1993). Judging from subjects’
comments, from the paucity of questions asked of the experimenter who
was present to provide assistance, and from e xperience with other, similar
questionnaires, subjects had li ttle dif® culty answ ering the appraisa l items.
R E S U L T S
O v e r v ie w
The same basic analytic strategy was followed as in our previous research
(Roseman et al., 1990). We began by calcul ating apprai sal scores for each
emotion experience. We then conducte d three sets of analyses to deter-
mine: (a) whether the emotions we were studying differed in appraisal; (b)
4
``SPECIFIC EVENT A’’ was what the subject had identi® ed as the cause of the emotion
recalled in the ® rst part of the questionnaire; ``SPECIFIC EVENT B’’ was the identi® ed
cause of the second emotion recalled.
if so, which of the appraisals differed among emotions; and (c) how much
each of these appraisals showed the theoretically predic ted pattern of
differences among emotions.
C a l c u la tio n o f A p p r a is a l S c o re s
Included in Table 1 are mean appraisal scores for the groups of subjects
re calling each emotion. Each of the 19 key appraisal items shown in the
table is represented by two rows of means.
5
The ® rst row gives appraisal
item scores for the various emotions recalled by subjects in the ® rst part of
the que stionnaire (``Experience 1’’ ). The second row shows appraisal item
scores for the various e motions recalle d by subjects in the second part of
the questionnaire (``Experience 2’’ ). As in Roseman et al. (1990), separate
analyses of data from the ® rst and sec ond experiences allowe d two te sts of
each theoretical prediction for all appraisal items, providing important
information about the replicability of observed relationships.
6
W e r e th e re D iffe r e n c e s in A p p r a is a l a m o n g
E m o ti o n s S tu d ie d ?
To determine whether these emotions differed in appraisal, we ran multi-
variate analyses of variance on the data for the ® rst and second experiences
re called by subjects. The predictor variable in these analyses was the
C O N S T R U C T IN G A N I M P R O V E D A P P R A IS A L T H E O R Y
2 5 1
5
Results for all 39 theory-relevant appraisal items are discussed and evaluated in this
paper, but due to limitations of space only the most important of these could be shown in
Table 1. Selection of means for incl usion in Table 1 was made empirically. For each
hypothesised appraisal dimension discussed in the Introduction, Table 1 shows the appraisal
item that maximised pre diction accuracy as judged by summed effect sizes (contrast eta
2
taken as a positive or negative value de pendi ng on ® t or lack of ® t with predictions) across
Experience 1 and Experie nce 2. That is, Table 1 shows the item that gave each appraisal
hypothesis its ``best shot’’ at accounting for the data on predicted differences between
emotions. Means and contrast test results for all appraisal items are available upon request
from the authors.
6
We also evaluated the internal consistency of the three items devised to measure each
appraisal dimension (see Appendix). Standardised reliability coef® cients (Cronbach’ s alpha)
across experiences of all emoti ons were 0.18 (for the three ite ms intended to measure
different types of ``unusualness’’), 0.86 (situational state), 0.57 (motivational state), 0.48
(probability), 0.74 (own power), 0.38 (stimulus powerlessnes s), 0.31 (stimulus controllabil-
ity), 0.69 (stimulus controllability by self), 0.64 (coping potential), 0.70 (legitimacy), and
0.04 (problem source). According to Nunnally (1967), groups of items with reliabilities
greater than 0.5 can be regarded as measuring the same underlying dimension of appraisal.
Alphas were not calculated for agency or responsi bility items because the self, other persons,
and circumstances beyond anyone’ s control may independently cause or be responsible for
an event.
T
A
B
L
E
1
M
e
a
n
A
p
p
r
a
i
s
a
l
R
a
t
i
n
g
s
f
o
r
R
e
c
a
l
l
e
d
E
x
p
e
r
i
e
n
c
e
s
o
f
E
a
c
h
E
m
o
t
i
o
n
,
w
i
t
h
C
o
n
t
r
a
s
t
T
e
s
t
s
o
f
C
o
m
p
e
t
i
n
g
A
p
p
r
a
i
s
a
l
P
r
e
d
i
c
t
o
r
s
Appraisal
Emotion Recalled
a
Predictors
Contras t eta
2
Resid ual
Su Ho Jy Rl Af
b
Pr Fe Sd Ds Fr Dg Dl An Ct Rg Gu Sh t-test F
c
Dimension 1: Sur pr is e vs. Othe r Emotions
Contrast
+16
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
2
1
Unexpected
Ex 1 9.00 4.30 6. 27 5.10 6.22 5.70 7. 08 7.50 6.18 2.92 7.36 6.55 7.55 6.00 7.73 7.91 6. 20 3.24*** 0.06 2.85² ² ²
d
Ex 2 8.30 6.55 6. 10 7.10 5.60 4.56 7. 09 7.64 6.73 6.92 6.50 7.00 6.64 6.67 6.82 6.80 5. 92 2.03* 0.02 0.71
e
Unfamiliar
Ex 1 7.09 4.80 3. 36 3.00 6.25 5.10 6. 83 5.90 5.20 4.69 6.73 4.82 5.60 5.30 6.91 5.00 5. 80 2.15* 0.03 1.92²
f
Ex 2 5.30 4.36 5. 40 6.00 4.10 5.22 7. 45 5.82 6.40 5.75 6.20 4.75 4.64 3.90 4.90 6.90 5. 92
2
0.20 0.00 1.36
f
Nov el
Ex 1 5.20 4.78 6. 00 4.67 5.00 5.90 5. 00 5.90 4.64 4.18 3.91 3.60 4.70 4.70 3.45 4.91 3. 40 0.66 0.00 1.16
g
Ex 2 4.89 5.73 6. 30 5.20 4.90 6.89 5. 82 4.00 5.10 3.92 3.75 4.42 2.64 5.20 4.10 5.67 3. 92 0.05 0.00 2.28³
g
Dimension 2: Positive vs. Negative Emotions
Contrast
0 +11 +11 +11 +11 +11
2
5
2
5
2
5
2
5
2
5
2
5
2
5
2
5
2
5
2
5
2
5
Improved ( Situational State)
Ex 1 4.27 6.00 8. 64 7.50 6.67 7.70 1. 67 2.30 3.10 2.54 2.36 2.82 2.18 3.20 3.36 3.91 1. 20 11.51*** 0.50 1.76²
Ex 2 5.00 7.45 8. 20 6.50 8.10 8.67 2. 27 2.45 3.27 2.33 2.70 3.25 1.73 2.10 2.45 2.80 2. 08 15.28*** 0.56 0.82
Dimension 3: Contact-Regulating v s. Distance-Regulating Emotions
Contrast
0 0 +1
2
1 0 0 0 +1
2
1 0 0 0 0 0 0 0 0
Pleasurable (Motiv ational State)
Ex 1 5.18 5.40 8. 63 3.30 6.22 6.90 3. 45 6.10 5.45 6.08 4.36 5.00 5.18 3.30 5.64 5.00 5. 70 3.32*** 0.06 1.63
Ex 2 7.20 7.55 8. 10 5.10 7.20 6.67 2. 00 6.45 5.45 4.92 3.70 4.67 5.36 5.50 6.36 4.60 6. 25 2.29* 0.03 2.92² ² ²
2 5 2
2 5 3
Dimension 4: Reactive vs. Preparatory Emotions
Contrast 0
2
2 +1 +1 0 0
2
2 +1 +1 0 0 0 0 0 0 0 0
Able to Predict (Probability)
Ex 1 3.45 3.20 4.45 5.20 2.22 3.70 1.67 2.90 4.00 6.38 3.36 5.18 3.27 3.60 3.36 4.09 5.30 2.44** 0.03 2.01²
d
Ex 2 2.20 2.45 4.70 4.90 4.10 5.89 2.00 4.64 3.36 4.08 3.40 5.67 4.82 4.40 3.27 2.80 3.17 3.09** 0.06 1.49
Dimension 5: Contending vs. Accomodating Emotions
Contrast
0 0 0 0 0 0
2
6
2
6
2
6 +5 +5
2
6 +5 +5
2
6 +5 +5
Powerful (Own Power)
Ex 1 3.00 4.40 6.73 5.60 3.44 6.80 1.67 2.00 2.09 2.58 1.91 4.91 2.09 3.60 4.45 5.73 2.50 0.11 0.00 7.15² ² ²
Ex 2 4.00 4.36 6.89 4.30 5.20 8.44 1.27 1.55 2.18 2.83 3.30 3.75 2.18 4.30 3.73 3.40 3.67 1.96* 0.02 6.58² ² ²
e
Easy (Stimulus Powerlessness)
Ex 1 4.18 4.90 7.00 5.50 5.67 5.70 2.33 2.50 1.83 4.15 3.73 3.09 3.00 6.90 4.73 5.55 3.50 3.31*** 0.06 3.30² ² ²
h
Ex 2 5.30 5.64 5.80 5.00 6.50 6.56 2.00 2.82 4.27 4.08 3.20 3.33 3.64 5.50 4.55 3.10 3.92 1.06 0.01 2.69² ² ²
Someone Could Change (Stimulus Controllability)
Ex 1 4.27 5.90 4.18 4.50 4.67 5.10 4.50 3.20 4.82 6.08 5.36 6.27 6.82 5.50 6.00 5.55 4.90 1.38 0.01 0.98
Ex 2 5.30 6.09 4.50 4.20 5.80 5.67 3.64 5.00 4.82 5.25 5.80 6.67 6.36 7.90 5.73 5.40 4.33 1.35 0.01 1.47
Controllable by Me (Stimulus Controllability by Self)
Ex 1 3.91 4.90 4.55 6.60 3.89 6.90 2.92 2.00 3.82 5.00 3.64 2.82 3.27 3.30 4.36 7.00 7.00 3.20** 0.06 2.64² ² ²
d
Ex 2 4.10 5.00 5.20 2.90 3.80 8.89 3.91 3.73 4.09 5.92 3.80 4.67 3.55 6.80 6.82 7.40 7.50 2.03* 0.03 3.38² ² ²
Something I Could Do (Coping Pot ential)
Ex 1 3.82 6.60 6.64 6.40 5.00 6.10 3.25 2.50 3.00 4.77 5.36 4.09 4.64 6.90 4.73 6.09 6.40 4.10*** 0.10 1.52
i
Ex 2 5.40 6.82 6.60 5.20 5.20 7.78 2.91 3.36 4.09 4.67 4.00 4.50 4.36 6.20 6.00 4.30 6.92 1.74* 0.02 2.39³
e
Dimension 6: Attack vs . Exclusion Emotions
Contrast
j
0 0 0 0 0 0 0 0 0 +1
2
1 0 +1
2
1 0
2
1 +1
Deserve Good (Legitimacy)
Ex 1 5.55 8.10 8.09 6.40 5.67 7.30 4.92 5.50 6.09 4.62 5.45 5.18 6.64 6.60 5.00 3.82 3.80
2
0.48 0.00 4.12² ² ²
d
Ex 2 3.80 7.36 7.90 5.80 6.60 8.75 4.82 5.09 6.64 6.00 6.40 5.55 6.09 5.30 5.45 3.60 3.25 0.03 0.00 4.78 ² ² ²
f
T
A
B
L
E
1
(
c
o
n
t
i
n
u
e
d
)
Appraisal
Emotion Recal led
a
Predictors
Contrast eta
2
Resid ual
Su Ho Jy Rl Af
b
Pr Fe Sd Ds Fr Dg Dl An Ct Rg Gu Sh t-test F
c
Dimension 6: Attack vs. Exclusion Emotions (continu ed)
Contrast
0 0 0 0 0 0 0 0 0 +1
2
1 0 +1
2
1 0
2
1 +1
Basic Nature (Problem Source)
Ex 1 5.45 6.89 7.09 7.80 5.78 6.80 6.00 4.90 6.64 6.85 6.27 4 .5 5 6.27 6.67 5.45 5.00 6.10 0.50 0.00 1.39
e
Ex 2 6.50 6.55 6.90 5.80 7.80 6.11 6.36 6.45 6.18 3.50 6.10 7 .3 3 6.55 6.90 5.82 3.60 6.00 2.94 ** 0.05 1.85²
e
Dimension 7a: Self-Directed vs. Other Emotions
Contrast
2
4
2
4
2
4
2
4
2
4 +13
2
4
2
4
2
4
2
4
2
4
2
4
2
4
2
4 +13 +13 +13
Self-Agency
Ex 1 3.45 5.50 5.18 7.70 4.67 7.30 4.75 4.70 4.45 4.77 3.82 3 .8 2 2.00 4.10 7.00 7.82 8.50 8.89 *** 0.25 2.64² ² ²
d
Ex 2 5.70 5.64 6.00 4.90 4.80 8.11 5.55 4.91 4.45 7.08 4.00 4 .5 8 6.09 4.00 7.82 7.20 6.58 4.85 *** 0.13 1.13
Self-Responsibility
Ex 1 3.45 7.00 5.18 7.80 4.44 6.80 5.00 4.80 5.82 6.15 4.91 3 .2 7 3.00 3.00 7.09 8.09 8.60 7.62 *** 0.20 3.49² ² ²
d
Ex 2 5.10 5.55 6.50 4.90 4.30 8.56 6.09 5.36 4.45 7.33 4.30 4 .7 5 5.00 3.60 7.27 7.70 6.00 4.82 *** 0.12 1.52
Dimension 7b: O ther-Person-Directed vs. Other Emotion s
Contrast
2
4
2
4
2
4
2
4 +13
2
4
2
4
2
4
2
4
2
4
2
4 +13 +13 +13
2
4
2
4
2
4
Other-Person-Agency
Ex 1 7.73 3.60 7.09 3.40 7.00 4.60 6.50 4.10 5.91 6.15 6.55 6 .7 3 8.09 7.30 5.27 3.45 2.40 4.66 *** 0.12 3.55² ² ²
d
Ex 2 6.40 5.73 4.70 5.10 6.20 3.78 4.45 6.55 6.64 4.25 6.30 7 .1 7 7.82 7.60 3.10 4.20 3.33 4.68 *** 0.12 1.95²
e
Other-Person-Responsibility
Ex 1 7.73 3.40 7.09 3.70 4.89 4.50 5.17 4.20 6.27 6.38 7.45 7 .4 5 8.00 7.60 3.73 4.55 2.40 4.08 *** 0.09 4.36² ² ²
d
Ex 2 6.20 5.73 5.80 5.20 5.80 4.44 4.09 6.45 6.09 4.67 6.60 7 .6 7 7.64 7.50 3.09 4.40 3.42 4.30 *** 0.10 1.87²
2 5 4
2 5 5
Dimension 7c: Event-Directed vs. Othe r Emot ions
Contrast
+8 +8 +8 +8
2
9
2
9 +8 +8 +8 +8 +8
2
9
2
9
2
9
2
9
2
9
2
9
Circumstance-Agency
Ex 1 6.18 4.50 5.27 5.10 4.89 4.90 4.50 6.40 4.64 3.54 4.55 3.45 3.09 4.10 2.82 3.73 2.50 3.13** 0.06 1.14
d
Ex 2 7.40 4.09 5.20 7.00 5.30 4.00 5.18 5.36 5.45 3.33 5.10 5.08 3.82 3.80 3.36 3.60 3.17 3.27*** 0.06 1.59
Circumstance-Responsibility
Ex 1 5.64 5.60 4.36 4.60 4.33 5.20 6.00 6.40 5.27 5.54 4.36 5.27 3.60 5.40 6.00 3.36 2.50 2.17* 0.03 1.47
Ex 2 7.20 5.45 6.33 5.60 4.60 3.11 5.55 5.27 5.20 3.08 4.90 5.00 5.09 3.10 3.91 4.40 4.25 3.27*** 0.06 1.52
f
Notes
. Su, surprise; Ho, hope; Jy, joy; Rl, relief; Af, affection towards someone; Pr, pride; Fe, fear; Sd, sadness; Ds, distress; Fr, frustration; Dg,
disgust; Dl, dislike toward someone; An, anger; Ct, contem pt; Rg, regret; Gu, guilt; Sh, shame.
Ex 1, Experience 1; Ex 2, Experience 2.
a
N
per cell (Ex perience 1, Experience 2): Su, Jy, Dg, Gu (11,10); Ho, Sd (10,11); Rl, Pr, Ct (10,10); Af (9,10); Fe (12,11); Ds, An, Rg (11,11); Fr
(13,12); Dl (11,12); Sh (10,12).
b
Used to measure li king (Roseman et al., 1990).
c
df
s = 15,164 unless otherwise noted.
d
df
s = 15,165.
e
df
s = 15,163.
f
df
s = 15,162.
g
df
s = 15,158.
h
df
s = 15,159.
i
df
s = 15,160.
j
As shown by contrast weights, prediction differs from Basic Nature (Problem Source), but
contrast is not signi® cant even if Problem Source weights are used.
*
P
< 0.05; **
P
< 0.01; ***
P
< 0.001. ²
P
< 0.05, two-tail ed; ³
P
< 0.01, two-tail ed; ² ² ²
P
> 0.001, two-tailed.
2 5 6
R O S E M A N , A N T O N I O U , J O S E
emotion recalled (surprise/joy/reli ef /hope/affection/pride/fear/sadness/dis-
tress/frustration/disgust/dislike/anger/contempt/regret/guilt/sham e); depen-
dent variables were all of the appraisal items listed in the Appendix.
Results of the MANOVAs showed highly signi ® cant differences among
emotions, both in the ® rst experience {
F
(624, 1631) = 1.75,
P
< 0.001}, and
in the second e xperience {
F
(624, 1720) = 1.62,
P
< 0.001}. Thus, appraisals
clearl y differed from emotion to emotion.
W h ic h o f th e H y p o t h e s is e d A p p r a is a ls D i ff e r e d
a m o n g E m o t io n s ?
To determine which of the se appraisals differentiated these emotions, we
then ran univariate anal yses. Again, the predictor variable in all of these
analyses was the emotion recalled. Each appraisal item served as the sole
dependent variable in one of these ANOVAs. Results indic ated that most,
although not all, of the appraisals we were studying did differentiate
these emotions. That is, for most of the items measuring appraisals of
situational state, motivational state, probability, own power, stimulus
powerlessness, stimulus controllability, stim ulus controllability by self,
coping potential, legitimacy, agency, and responsibility, the di fferences
among emotions were signi® cant (
P
< 0.05) in both re called experiences.
For certainty of consequences, changeability by self, circumstance-
agency, and circumstance-responsibility, signi® cance (
P
< 0.05) was
obtained in one recalled experience and marginal signi® cance (
P
5
0.06) in the other. Appraisals of unusualness and problem source dif-
fered signi® cantly among emotions in one of the two experiences. Only
appraisals of modi® ability and changeability showed no signi® cant dif-
ferences across emotions.
W h ic h A p p ra i s a ls D iff e re d a m o n g E m o t io n s in
T h e o re tic a lly P re d ic t e d P a tt e r n s ?
To test whether the various appraisals differentiated these emotions as
predicted, we ran single degree of freedom contrasts (see Rosenthal &
Rosnow, 1985, pp. 1± 18) on the main effects of emotion recalled on each
appraisal. Weights for a give n contrast were determined by the way in
which a particular appraisal was predicted to differ among emotions (see
Fig. 1 and alternative hypotheses discussed earlier), and a directional
t
-test
indicated whether the observed means conformed signi® cantly to predic-
tions. Contrast wei ghts s pecifying seven hypothesised distinctions among
emotions are shown in Table 1 above the various appraisals that might
account for each distinction.
7
Results of contrast
t
-tests are shown for each
row of appraisal means, and the relative success of each appraisal predic-
tion is indicated by the value of eta
2
.
Surprise vs. other emotions
. Of the three appraisals (unexpectedness,
unfamiliarity, and novelty) proposed to differentiate surprise from other
emotions, the
Contrast t
-
test
column of Tabl e 1 reveals that only unexpect-
edness conformed signi® cantly (
P
< 0.05) to this predicted pattern in both
experiences recalled by subjects.
8
As shown by the appraisal means, events
perceived as unexpected elic ited higher levels of surprise than of other
emotions.
Positi ve vs. negative emotions
. The appraisa l of situational state was
hypothesised to differentiate positive from negative emotions. Contrast
t
-
tests found that all three i tems designed to measure situational state showed
signi® cant corre spondence to predictions (
P
< 0.001). Events appraised as
re latively consistent with what one wanted (or as wanted, or as improving
things) el ici ted the positive emotions hope, joy, relief, affection, or pride.
Events appraised as relative ly inconsistent with what one wanted (or as
unwanted, or as making things worse) elicited the negative emotions fe ar,
sadness, distress, frustration, dislike, anger, contempt, regret, shame, or
guilt. Events eliciting surprise were appraised as intermediate in motive-
inconsistency . . . motive-consistency. The ite m that best accounted for the
distinction between positive and negative emotions, shown in Table 1, was
an appraisal of whether the emotion-eliciting event improved things or
made them worse.
Contact-r egul ating vs. distance-regulating emotions
. An appraisal of
motivational state was hypothesised to differentiate joy and sadness from
distress and relief. There was signi® cant support for this prediction (
P
<
0.05) in the sec ond recalled experience for all three items designed to
measure motivati onal state. However, only one of the motivational state
items received signi® cant support in both experiences. As shown in Table
1, motivation to get or keep something pleasurable was more characteristic
of events eliciting j oy or sadness; and motivation to get rid of or avoid
something painful was more characteristic of events eliciting distress or
re lief.
C O N S T R U C T IN G A N I M P R O V E D A P P R A IS A L T H E O R Y
2 5 7
7
For discussion relating the predicted emotion groupings in our analyses to a uni® ed
theory linking patterns of appraisal to patterns of emotional response (including emotion-
speci® c action tendencies and emotion-speci® c goals), see Roseman (1994a,b).
8
As in Roseman et al. (1990), uncertainty also failed to rel iably distinguish surpris e from
other emotions.
2 5 8
R O S E M A N , A N T O N I O U , J O S E
Reactive vs. pre paratory emotions
. An appraisal of probability was
proposed to differentiate both (a) joy and relief vs. hope, and (b) sadness
and distress vs. fear. Contrast tests indicated that each of the items designed
to measure probability corresponded signi® cantly to this pattern in both
re called experiences (
P
< 0.05). Strongest support was obtained for an
appraisal of the predictability of events (
P
< 0.01). As shown by the me ans
in Table 1, being able to predict what was going to happen in a situation
was more characteristic of events eliciting joy or rel ief than of events
elic iting hope; and more characteristic of events eliciting sadness or
distress than of events eliciting fear.
Contending vs. accommodating emotions
. Five classes of appraisals from
different theories (own power, stimuls powerlessness, stimulus controll-
ability, stimulus control lability by self, and coping potential) were pro-
posed as alternatives for diff erentiating emotions that contend with a
motive- inconsistent event (e.g. frustration, anger, guilt) from emotions
that accommodate to it (e.g. sadness, dislike, regret).
Contrast tests revealed that none of the items constructed to measure
appraisals of own power, stimulus powerlessness, or stimulus controllabil-
ity differentiated these emotions in the way that was predicted in both
re called emotion experiences. There was, however, consistent signi® cant
support (
P
< 0.05) for predictions made with one appraisal item designed to
measure stimulus controllability by the se lf and one item designed to
measure coping potential. That is, perceiving that an emotion-eliciting
event was c ontrol lable by the self, and perceiving there w as something
one could do about it, were generally characterstic of frustration and
disgust, rather than fear, sadness, and distress; anger and contempt, rather
than mere dislike toward someone; and guilt and shame, rather than regret.
Thirty-three of the 40 relevant pairwise comparisons of means for these
emotions ® t this predicted pattern (as shown in Table 1). Predictions made
with other items designed to measure stimul us controllability by the self or
coping pote ntial did not receive consis tent signi® cant support acros s both
re called e motion experiences.
Attack vs. exclusion emotions
. Two appraisals, legitimacy and problem
source, were proposed to differentiate guilt from shame, anger from con-
tempt, and frustration from disgus t. Contrast tests revealed that none of the
legitimacy items ® t these predictions in either of the two recalled emotion
experiences. Instead, as i ndicated by the legitimacy means in Table 1, the
belief that one deserved a good outcome was rated as a cause of positive
emotions more than negative e motions. In addition to this pattern, among
the negative e motions legitimacy tended to be low er in events eliciting
re gret, guilt, and shame, and higher in events eliciting dislike, contempt,
and anger. Among the positive emotions, legitimacy was particularly high
in events eliciting pride.
Appraisals of problem source had somewhat greate r success in making
the hypothesised emotion differentiations. Contrast tests found that predic-
tions for each of the appraisals intended to measure problem source
re ceived signi® cant support in one recalled experi ence. Support was stron-
gest for the item asking whether the emotion-eliciting event had revealed
the basic nature of someone or something (
P
< 0.01 in Experience 2). The
means in Table 1 indicate that, consistent with predictions, subjects were
more incline d to perceive that events revealed the basic nature of someone
or something in instances of disgust rather than frustration (in Experience
2), contempt rather than ange r (in both experiences) , and shame rather than
guilt (i n both experiences). None of the other appraisals measured in this
study reliably differentiated members of these emotion pairs from each
other.
Self-directed, other-person-directed, and event- dire cted emotions
.
Finally, two sets of appraisals were proposed to differentiate emotions
felt toward the self, emotions felt toward other persons, and emotions
that can be felt toward impersonal events or objects. The contrast tests in
Table 1 show signi® cant s upport for both agency and responsibility apprai-
sal predictions in e ach of the experiences recalled by subjects (
P
< 0.001,
except for circumstance-agency and circumstance-responsibility, where
P
< 0.01). Combining across the two recalled emotion experiences, effect
sizes were slightly higher for self -agency than for self-responsibility, for
other-person-agency than for other-person-responsibility, and for circum-
stance-agency than for circumstance-responsibility. Thus overall, predic-
tions were mostly strongly supported for agency appraisals.
The sources of support for agency predictions are shown in Table 1.
With few exceptions, in both experiences recalled by subjects causation by
the self was rated highest i n events eliciting pride, regret, guilt, and shame;
and causation by other persons was rated highest in eve nts eliciting
affection, dislike, anger, and contempt. Causation by circumstances
beyond anyone’ s control was generally higher in events e li citing s ur-
prise, joy, and relief, than in events eliciting affe ction and pride; and
higher in events e lic iting fear, sadness, distress, and disgust than in events
elic iting dislike, anger, c onte mpt, regret, guilt, and shame. Contrary to
predictions, causation by other persons was consistently high in events
elic iting surprise, and causation by circumstances beyond anyone’ s control
was low in events eliciting hope and frustration.
C O N S T R U C T IN G A N I M P R O V E D A P P R A IS A L T H E O R Y
2 5 9
2 6 0
R O S E M A N , A N T O N I O U , J O S E
A r e t h e r e A d d it io n a l R e l a t io n s h ip s b e tw e e n th e s e
A p p ra i s a ls a n d E m o ti o n s ?
Signi® cant residuals in Table 1 point to other relationships between the
appraisals and emotions beyond those that were identi® ed in our hypoth-
esis- testing contrasts. Means for motivational state, own power, and legi ti-
macy appraisals in Table 1 suggest that the extra appraisal variance often
corresponds to diffe rences between positi ve and negative emotions. This
® ts a pattern, noted in previous research (Roseman et al., 1990), that may
re ¯ ect appraisal sequences or interactions. That is, being motivated to
attain a pleasurable state, or appraising oneself as powerful or deserving
of a good outcome, may lead one to appraise a situation as more motive-
consistent, and thus generate more positive emotion.
D IS C U S S IO N
T e s tin g a n d R e v is in g A p p r a i s a l T h e o r ie s
This study te sted many speci® c hypotheses, proposed by different theorists.
Results showed that related yet distinct appraisals had differing relation-
ships to particular emotions, and that some appraisals ® t predictions more
closely than did others. What exactly have we le arned from these analyses?
How can our ® ndings be used to develop a more accurate and comprehe n-
sive theory of the causes of discrete emotions? What questions now need to
be answered, and what type of research is needed to address them?
R e l a tio n s h ip s b e tw e e n P a rt ic u la r A p p r a is a ls a n d
P a r tic u la r E m o ti o n s
Unexpectedness, rather than novelty, unfamiliarity, or uncertainty,
elic its surprise
. Although they may be plausible hypotheses, w e found
that neither novelty (Scherer, 1984), nor unfamiliarity (Scherer, 1988),
nor extreme uncertainty (Roseman, 1984) elicited surprise, as opposed to
other emotions, in both recalled experiences. Rather, it was the unexpect-
edness of a stimulus (Izard, 1977; Scherer, 1984) that reliably e voked
surprise. These results suggest that the theories proposed by Roseman
(1984) and Scherer (1984, 1988) need to be revised. Judging from our
subjec ts’ appraisal ratings, unfamiliarity or uncertainty may elicit fear, and
novelty may elicit joy or pride, rather than surprise. It would seem that only
when novel, unfamiliar, or uncertain stimuli are unexpected will they
produce surprise.
Situational state appraisals differentiate positive from negative emo-
tions
. Our data indicate that appraisals of motive-consistency elicit posi-
tive emotions, and appraisals of motive-inconsistency elicit negative
emotions, as predicted. Whereas previous research may have identi® ed
similar appraisals that occur as part of the experienti al content of emo-
tions, here motive-consistency vs. motive-inconsistency is reported to be a
cognitive cause of emoti ons.
The situational state results also shed light on the nature of surprise. In
prior studies, surprise has been associated with appraisals of pleasantness
(Smith & Ellsworth, 1985) or motive-consistency (Roseman et al., 1990).
But when we focused subjects on the appraisals that caused their surprise,
rather than the contents of the resulting emotional experience , we found
that neither moti ve- consistency nor notive-inconsistency was seen as
elic iting surprise. These data ® t with the conception of surprise as an
emotion that is not inherently positive or negative (Ortony et al., 1 988;
Roseman, 1984).
Motivational state differenti ates contact-regulating from distance -reg-
ulating emotions
. Only three of the six appraisal theories discussed i n
Scherer’ s (1988) review link different emotions to different types of
goals, and whether appraisal the ories should do so has been a matter of
some controversy (see, e.g. Smith & Lazarus, 1993, p. 263). However, data
gathered in this study provide some support for the hypothesis that differ-
ent emotions are felt when a person is appetitively as opposed to aversively
motivated (Roseman, 1979, 1984; cf. Spence, 1956). Additional research is
re quire d to determine whether distinct emotions also ¯ ow from different
speci® c appetitive and aversive concerns, such as enhanceme nt of eg o
identity (Lazarus, 1991a) or relationship goal s (Scherer, 1986).
Our results also suggest re® nements in conceptualisation of the moti va-
tional state dimension. Predictions were only consistently s upported when
the distinction was phrased in terms of wanting to get or keep something
pleasurable (eliciting joy if events were consistent with this motive, or
sadness if they were inconsistent with it) vs. wanting to get rid of or avoid
something painful (eliciting distress if events were inconsistent with this
motive, and relief if they were motive-consistent).
These ® ndings imply a less cognitively complex conce ptualisation of
motivational state. It seems that it is not so much
thinking about
a goal state
as a bene® t or a l oss, or
categoris ing
it as positive or negative, that is
crucial; but rather perceiving or simply registering it (see Frijda, 1993) as
something potentially pleasurable or painful.
A probability appraisal differentiates reactive vs. preparatory emotions
.
As noted earlier, it was not clear from prior research that uncertainty was
associated with hope as well as fear, and certainty with sadness, distress,
C O N S T R U C T IN G A N I M P R O V E D A P P R A IS A L T H E O R Y
2 6 1
2 6 2
R O S E M A N , A N T O N I O U , J O S E
and relief, as well as joy. Here, with subjects focused on the causes rather
than the contents of their emotions, each of these hypotheses received
empirical support.
A relational appraisal of control potential, rather tha n appraisals
focusing solely on the self or the stimulus event, appears to differentiate
contending vs. accommodating e motions
. As discussed earlier, many the-
ories claim that an appraisal of one’ s power, or of an event’ s controllabil-
ity, helps determine which of several negative emotions one will fe el . Our
empirical assessment of these and alternative hypotheses found instead that
only a relational appraisal (Lazarus, 1991a), comparing one’ s own power
or control to the potency or controllability of the stimulus, conformed
signi® cantly to predictions. Such relational appraisals have been proposed
by Bandura (1977), Frijda (1986), Lazarus and Smith (1988), and Scherer
(1988). This result is also in accord with ® ndings from the clinical
literature (e.g. Garber & Hollon, 1980; Weisz et al., 1989; Weisz, Weiss,
Wasserman, & Rintoul, 1987; but see also Weisz, Sweeney, Prof® tt, &
Carr, 1993) that it is not necessarily perceptions of the intrinsic uncontroll-
ability of negative events, but rather their perceived uncontrollability by
the self that is associated with depression (cf. sadness in Fig. 1).
What is the best way to conceptualise this relational judgement? Ana-
lyses of speci® c items revealed that predictions were consistently sup-
ported for: (a) an appraisal of an event’ s controllabil ity by the sel f (but
not its modi® ability or changeability); and (b) an appraisal of whether one
could ``do something about’’ an event (but not appraisals of whether one
could cope or deal with it). On the latter point, the predictive failure (in
both recalled experiences) of the item asking speci® cally about judgements
of whether one was ``able to cope with’’ the event cautions against con-
ceptualising the crucial appraisal in terms of coping potential (e.g. Lazarus,
1991a; Smith & Lazarus, 1993). Coping potential includes the capacity to
adjust to an event, for example, by such intrapsychic means as changing
one’ s goals, ® nding meaning in the event, or dealing with one’ s emotional
re sponses (Scherer, 1988). Our resul ts suggest that it is not one’ s ability to
cope with an event, but rather the perceived ability to control or do
something about its motive -inconsistent aspects that elicits an emotion
which will contend with a situation (such as frustration or anger) rather
than an emotion which will accommodate to it (such as sadness).
9
9
Having an appraisal of high vs. low control potential that determines whethe r one
experienc es a contending emotion vs. an accommodating emotion makes sense within a
functionalist perspective, in which appraisals serve to elicit the particular emotion whose
response tendencies are relatively likely to help a person adapt successfully to the type of
situation he or she is facing.
For example, one subject felt frustration because ``I knew most of the
material I did wrong’ ’ on an exam. ``If I didn’ t know the material, I would
not have be {
sic
} frustrated.’’ That is, it was precisely because his potential
control over the outcome was high that this subject f elt frustrated. In
contrast, the same subject felt sad when a friend was diagnosed as having
leukaemia. In the latter experience, the s ubject perceived there was nothing
he could do about ``the fact that a young man had to be ill . . . and that the
family of this young man had to suffer.’’
The best formulation of the crucial relational appraisal dimension might
be termed
control potential
or
in¯ uence potential
, which would appear to
combine only those subcomponents of ``stimulus controllability by self’’
and ``coping potential’’ that were supported in our data. A similar formula-
tion w as proposed by Wortman and Brehm (1975) in a review of the
literature on responses to uncontrollable events. They suggested that
uncontrollable events produce reactance (a f rustration- or anger-like
re sponse that aims to restore one’ s freedom) if a person believes that he
or she can still exert control; but learned helplessness (a sadness- or
depression-like response that fails to struggle against aversive condi-
tions) if establishing control is thought to be impossi ble.
Indeed, Roseman (1984) has argued that the reason perceived le gitimacy
is often associated with anger (see Averill, 1982) is that it confers a kind of
``moral power’’ w hich can be used to in¯ uence harmdoers to change their
behaviour, and recruit bystande rs or authoriti es to help change negative
events one could not alter by oneself. In the terms used here, having justice
on one’ s side tends to increase one’ s control potential, and can therefore be
an elicitor of anger. Frijda (1993) observes that violations of nonmoral
expectations can also give rise to the perception that something ``ought’’
not to have happe ned, and evoke an angry response. Thus, it would seem
that many types of expectations can engender a sense of control potential,
and thereby contribute to the elicitation of frustration or anger (see, e.g. the
anger story quoted earlier, in which the subject felt her sister ``should have
been around’’ so she could do her laundry).
An appraisal related to problem source, rather than legitimacy, may
differentiate attack vs. exc lusion emotions
. Of the two appraisals proposed
to distinguish guilt from shame, anger from contempt, and frustration from
disgust, we found no support for the legitimacy hypothesis. For e xample,
although several authors have contended that guilt results from wrongdoing
or the perception that one deserves a negative outcome (e.g. Izard, 1991;
Roseman, 1984), whereas shame need not involve immorality (e.g. Ausu-
bel, 1955), our data showed that events eliciting shame were generally
perceived to involve as much or more illegiti macy and blameworthiness as
C O N S T R U C T IN G A N I M P R O V E D A P P R A IS A L T H E O R Y
2 6 3
2 6 4
R O S E M A N , A N T O N I O U , J O S E
events eliciting guilt. In contrast, data for each of the three component
appraisals designed to measure behavioural vs. characterological problem
source signi® cantly ® t predictions in one recalled experience.
However, none of the probl em source appraisals showed signi® cant
correspondence to predictions in both experiences. The signi® cant but
inconsistent ® ndings for the problem s ource apprai sal items (along with
their low internal consistency) suggest that something resembling the
behavioural vs. characterological distinction might succeed in differentiat-
ing these emotions, but the conceptualisation of the appraisal dimension
must be revised somewhat.
One possible change is to reformulate the behavioural pole of the
continuum as ``noncharacterological’’ . In our data, the two problem source
items for which s upport was weakest (those with the smallest effect sizes
across emotion expe riences) were those that asked about behaviour or what
someone or something was doing. The items that ® tted predictions best
made no mention of behaviour. For example, consistent with hypotheses, in
the second recalled emotion experi ence subjects felt shame, rather than
guilt, when they belie ve d the ``basic nature’’ of someone or something had
been revealed (s ee Table 1). In contrast, they felt guilt when a negative
event was c aused, but it did not reveal the basic nature of a person or
object.
Analogously, as suggested by the emotion experience s quoted in
describing our procedures, one may feel contempt if another person is
seen as characterologically bad, whether because of actions (e.g. submit-
ting someone else’ s paper as he r own) or attributes (e.g. sel ® shness); but
feel anger if the person simply causes a motive-inconsistent outcome (e.g.
makes it impossible to get the laundry done). And people may fee l disgust
if some object or event is seen as bad or ``rotten’’ in character; but
frustration if the object or event is negative in its impact on them, and
yet not fundamental ly negative in character.
An appraisal of agency differentiates self-directed, other-person-direc-
ted, and event-directed emotions
. Results of the present study generally
supported hypotheses specifying how causal agency (Roseman, 1984;
Scherer, 1988; Weiner, 1985) and responsibility (Smith & Ellsworth,
1985) differentiate e motions. Predicti ons were most strongly supported
for the agency appraisa ls. In accord with hypotheses, the self was seen
as having caused events eliciting pride, regret, guilt, and shame; and other
persons as having caused events eliciting affection, dislike, anger, and
contempt. Circumstance causation was low when these person-directed
emotions were experienced; and relatively high in e xperiences of
surprise, joy, relief, fear, sadness, distress, and disgust.
It is worth noting that although circumstance-causation was generally
high in events eliciting emotions such as joy and relief, other- person-
agency and self-ag ency were sometimes higher (see also ratings for hope
and frustration). These data are consistent with the hypothesis of Roseman
et al. (1990) that the ``circumstance-c aused’’ emoti ons in Fig. 1 (e.g.
frustration) can be elicited: (a) when an event (e.g. getting a less than
satisfactory exam grade) is attributed to impersonal circumstances (e.g. the
dif® culty of the course material); or (b) when a causal agent is identi® ed
(e.g. the teacher who gave the exam; the stude nt who did poorly) but the
agency information is disregarded in a person’ s focus on the event itself
(cf. Ortony et al., 1988); or (c) when no causal attribution whatsoever is
made for an event (see Weiner, 1985).
M e th o d o lo g i c a l Is s u e s
As discussed earlier, some authors contend that prior studies on appraisal
may have yielded knowledge on the cognitive contents of emotional
experiences rathe r than their causes. Whe n i n the present study we there-
fore asked subjects about the appraisals that
caused
their emotional
re sponse, we found some differences from previous results (e .g. we found
that motive-consiste ncy is irrelevant to the causation of surprise) and some
re semblances (e.g. that consistency vs. inconsistency with motives does
differentiate positive vs. negative emotions). Finding similarities as well as
differences is understandable because some cognitive dimensions may be
both causes and contents of emotions. These ® ndings do sugge st, however,
that future research on the causes of emoti ons should focus subjects on the
appraisals leading to emotional responses rather than appraisals made
while feeling the emotion.
Parkinsonand Manstead(1993) have argued that studies of appraisalin
which subjects readaboutor imagine emotion-elicitingsituations(e.g. Rose-
man, 1991; Smith & Lazarus, 1993): Present appraisal informationin lan-
guage that people do not really use to process genuine emotion-eliciting
events; make subjects arti® cially detached observers rather than involved
participants;and omit importantfactors that may in¯ uence emotions, such
as the responsesof interactionpartners.Incontrasttosuch scenariostudies,in
this study of recalled emotionexperiences: Peo ple described emotion-elicit-
ing events in theirown words; were involved participantsin thoseevents; and
mentionedthe responsesof othersin describinghow the events unfoldedover
time to producetheir own emotionalreactions.These features countercriti-
cisms of arti® ciality and enhance the generalisabilityof our ® ndings to other
instances in which real emotionsare produced.
We may still have questions as to whether subjects recal ling emotion-
elic iting events have failed to re member some of the ir appraisals accurately
C O N S T R U C T IN G A N I M P R O V E D A P P R A IS A L T H E O R Y
2 6 5
2 6 6
R O S E M A N , A N T O N I O U , J O S E
or have been unaware of some factors (e.g. physiological proc esse s) that
affected their emotions (Parkinson & Manstead, 1992). But f ailure s of
re call and omitted true causes would by themselves make it more dif fic ult
to ® nd the relationships between appraisals and emotions we have seen.
Perhaps this helps explain why measured appraisals do not predict
observed emotional responses even more accurately than they do (see
Reisenzein & Hofmann, 1993, for a discussion of obstacles to prediction
accuracy).
Finall y, it is possible that s ubjects’ recollections of the appraisals that
caused them to feel an emotion were biased by or reconstructed based on
culturally sc ripted implicit theories (Parkinson & Mans tead, 1992). This
might happen if people did not remember or know why they felt a
particular emotion (cf. Nisbett & Wilson, 1977), or if they used such
theories to sort out which of many cognitions in an experience had caused
the emotion to occur. All retrospective studies are vulnerable to this
possiblity. Indeed, even resul ts from studie s asking pe ople about their
ongoing appraisals and emotions (recommended by Parkinson & Man-
stead, 19 93) may be bias ed by such cultural scripts.
However, this concern is diminished by data from cross-cultural studies,
showing similar appraisal-emotion relationships in many different coun-
tries. For example, Wallbott and Scherer (1988) report such similarities
across ``27 countries on ® ve continents’’ (see also Mauro et al., 1992;
Roseman et al., 1995). The fact that subjects report similar appraisals
leading to simi lar emotions in so many different societies, whe re different
languages are spoken and different religions, philosophies, cultural norms,
and explanations for behaviour are dominant (see e.g. Miller, 1984) , argues
that they are actual appraisa l- emotion sequences. If subjects’ responses
were reconstructed based on implicit theories of the causes of emotions,
re ported appraisal -emotion relationships should diffe r widely from culture
to culture.
Additional evidence that these appraisal s re ally do cause these emotions
comes from studies that manipulate appraisals and measure e motions. For
example, Smith and Lazarus (1993) instructed subjects to visualise scenar-
ios in which either they themselves or someone else was accountable for a
negative event. Imagined self-accountabi lity resulted in feelings of guilt,
and imagined other-accountability resulted in feelings of anger, which
parallels our ® ndings for the agency dimension. And in a recent non-
scenario experimental study, Roseman and Evdokas (1995) manipulated
subjec ts’ appraisals of motivational state and probability in a laboratory
setting and found similar effects on joy and reli ef as those reported here.
In truth, there i s no perfect methodology for studying appraisal-emotion
re lationshipsÐ or anything else. Eac h method has strengths and weak-
nesses. Scenario and laboratory studies, in which appraisal information is
manipulated and emotional responses measured, provide evidence that
appraisals can have a causal impact on emotions (in addition to the impact
of emotions on appraisals); but such studies may be weak in generalisa-
bility to real-life events. Retrospective studies allow more con® dent gen-
eralisation to actual experi ences, but are subject to biased or inaccurate
re call. Studies of emotional appraisal in ongoing social interactions are less
subjec t to memory distortion, but are liable to make subjects overly s elf-
conscious and alter normal appraisal proc esse s.
Appraisal theorists can take heart, however, in convergence of results
across scenario, retrospec tive, l aboratory, and ® eld methodologies, as well
as across cultures. In addition, when Reisenzein and Hofmann (1990) asked
people to form their own distinctions among events that had elicited
different emotions, most of the distinguishing features they generated
corresponded to appraisal dimensions found in scenario and retrospective
studies. Although methodological re® nements (such as those adopted in the
present study) modify and extend our knowledge, it does seem that some
important facts about appraisal determinants of emoti ons are being estab-
lished by researchers.
C o n s tr u c t in g a R e v is e d , E m p i r i c a ll y G r o u n d e d ,
C o m p r e h e n s iv e T h e o r y o f E m o t io n s
The most important aim of this research was to test a number of rival
hypotheses proposed by prominent theorists, in order to develop a more
accurate, complete, and integrative understanding of the causes of emo-
tions. With improved methods that focused subjects on the causes rather
than the contents of emotional experiences, we c ompare d many alternatives
and found signi® cant support for parti cular hypotheses about appraisal
dimensions proposed by Frijda (19 86), Izard (1977), Lazarus (1991a),
Lazarus and Smith (1988), Ortony et al. (1988), Roseman (1984), Scherer
(1984, 1988), Smith and Ellsworth (1985), and Weiner (1985). Hypotheses
about other often-discussed appraisal dimensions were not supported (see
Introduction and Results earlier). All appraisal theorists can be asked to
take account of these data.
Based on these ® ndings, a new, empiric ally grounded, more c ompre-
hensive model, with components derived from several theories, is proposed
in Fig. 2. The major diffe rences from the model shown in Fig. 1 are as
follows:
1. The emotion of contempt has been incorporated into the theory, and
appraisal determinants for it have been speci® ed.
2. Instead of extreme unc ertainty, it is unexpectedness that elicits
surprise.
C O N S T R U C T IN G A N I M P R O V E D A P P R A IS A L T H E O R Y
2 6 7
2 6 8
R O S E M A N , A N T O N I O U , J O S E
3. Instead of an ass essme nt of one’ s power (or the intrinsic controll-
ability of a stimulus), it is a relational appraisal of control potential
that determines whether frustration or disgust are experienced, rather
than fear, sadness, or distress; anger or c ontempt, rather than dislike
toward someone; and guilt or shame, rather than regret.
4. An appraisal of problem source has been introduced to distinguish
three emotion pairs. Appraising a problem as characterological
(intrinsic to an object, other person, or the self) elicits disgust rather
than frustration, contempt rather than anger, and shame rather than
guilt.
5. The emotions of surpris e, hope, joy, relief , fear, sadness, distress,
frustration, and disgust may result not only from events perceived as
caused by impersonal circumstances, but also f rom events which fail
to be attributed to any cause, and from events attributed to self- or
other- agency (if that agency information is disregarded in a focus on
the event itself). To convey the idea that circumstance-causati on can
be but is not necessarily a determinant of these emotions, it is placed
in parentheses in Fig. 2.
The model shown in Fig. 2 is intended to provide a
systematic
account of
appraisal-emotion relationships. It is systemati c in that: (a) it spec i® es how
a small number of appraisal dimensions combine to elicit a large number of
emotions;
10
(b) it i denti® es the particular emotions that re sult from all
possible combinations of these appraisals; and (c) it shows how these
emotions are related to e ach other within an emotion system. The system
speci® es which emotions are closely related (e.g. guilt, shame) and which
are more distantly related (e .g. sadness, shame); identi® es families of
re lated e motions whose eliciting conditions differ in a single appraisal
10
Because all emotions (exce pt surprise) result from
combinations
of appraisals, one of
which assesses motive-consistency vs. motive-incons istency, they are all caused by what
Lazarus and Smith (1988) regard as true ``appraisal’’ (assessment of the implications of
events for an individual’ s goal commitments) rather then mere ``knowledge’’. The appraisal
combinations that lead to particular emotions might also be described at a more molar level,
which Smith and Lazarus (1993) term ``core relational themes’’ and de® ne as summaries of
relational meaning derived from a con® guration of appraisals (for examples of molar level
descriptions for each emotion in Fig. 2, see the delineation of emotion-speci® c phenome n-
ology in Roseman, 1994b). Thus, important advantages of the Smith and Lazarus (1990) and
Lazarus (1991b) models are found also in the model shown in Fig. 2, which attempts to be
more systematic in its coverage of emotions and relationships between them.
With regard to surprise, based on our data, the model presented in Fig. 2 shows that
surprise is inde pendentof motive-consistency vs. motive-inc onsistency (Ortony et al., 1988)
and is thus elicited by a single dimension of stimulus evaluation, unexpectedness (Scherer,
1984).
dimension (e.g. disgust, contempt, shame); and predi cts which c hanges in
appraisal are necessary to transform one felt emotion into any other (e.g.
ceasing to belie ve that one has potential to control attainment of a re ward
changes frustration to sadness {cf. Wortman & Brehm, 1975}; coming to
believe that uncontrollable nonattai nment of a reward was caused by the
self changes sadness to regret).
F u r th e r R e s e a r c h a n d T h e o r y D e v e lo p m e n t
The results of this research, and the appraisal model that has been con-
structed from them, suggest a number of areas in which further research
and theoretical elaboration woul d be pro® table.
First, having tested hypotheses from several theories and clari® ed many
appraisal-emotion relationships, focused investigations are now needed of
those appraisal-emotion linkages that have only tentatively been estab-
lished. For example, proposed revisions of the problem source appraisal
need to be tested. Careful studies should examine whether apprai sing
motive- inconsistent aspects of events as characterological vs. noncharac-
terological problems reliably determines whether a person feels disgust vs.
frustration, contempt vs. anger, and shame vs. guilt.
C O N S T R U C T IN G A N I M P R O V E D A P P R A IS A L T H E O R Y
2 6 9
F IG . 2 .
Hypothesised structure of the emotion system, incorporating new revisions.
2 7 0
R O S E M A N , A N T O N I O U , J O S E
Sec ondly, there are additional dimensions of appraisal and additi onal
emotions that may merit f urther investigation (see, e .g. Manstead & Tet-
lock, 1989; Ortony et al., 1988; Parrott, 1992; Reisenzein & Hofmann,
1990). For example, Reisenzein and Hofmann (1990) claim that an apprai-
sal of whether another person vs. the self is primarily affected by an event
differentiates emotions such as pi ty, from emotions such as anxiety (cf.
Ortony et al., 1988).
One issue relevant to this area of research is whether particular states
should be conside red additi onal discrete emotions to be explained, or
variants of emotions shown in Fig. 2. For example, could pity be a type
of sadnessÐ the sadness that one feels at the misfortunes of a positively
valued person because one does not want suc h a person to suffer? And
Schadenfreude
(see , e.g. Ben-Ze’ ev, 1994) a variant of joy, felt in response
to the misfortunes of negatively valued others (cf. Ortony et al., 1988) ? If
pity and
Schadenfreude
are subtypes of sadness and joy, they shoul d have
the same basic re sponse pro® le s (e.g. the facial expressions of sadness and
joy identi® ed in cross-cultural research; related physiology) and s hare the
same basic appraisal structure (e.g. for pity: certain inconsistency with an
appetitive motive, or colloquially, that things are de® nitely not as good as
they could be) perhaps with some added appraisals or situation speci® ers
(e.g. that the motive-inc onsistent event affects primarily another person). If
pity and
Schadenfreude
are distinct discrete emotions rather than emotion
subtypes, they should have different response pro® les and not share
common underlying appraisals. In either case, the appraisal determinants
of these and other emotion states nee d to be delineated, and interesting
hypotheses may be offered that can add to c urrent know ledge. For example,
extrapolating from our data it may be proposed that emotions felt about the
fortunes of other people depend on whether their f ate is consistent or
inconsistent with one’ s own motives, thus providing a uni® ed explanation
of both ordinary and ``vicarious’’ emotions (see, e.g. Batson, Ful tz, &
Schoenrade, 1987) .
Finall y, to construct a truly comprehensive and i ntegrative theory of
emotion-elicitation, appraisal sequences should be studied. For example,
Frijda (1993) has argued that an appraisal of legitimacy is not a necessary
antecedent of emotions such as anger; and our legitimacy hypotheses failed
to receive support in our analyses. But as discussed earlier, there may be an
appraisal seque nce in which perceiving that one has justice on one’ s side
tends to elevate appraisals of control potential, and thereby affects emo-
tions. Indeed, we found that legitimacy was high in both anger and
contempt, the two other-person-directed negative emotions hypothesised
to result from high control potential. Thus, even if legitimacy is not a
necessary
determinant of anger (one may have high control potential for
other reasons), it may be a typical in¯ uence on the emotion.
Another example comes from juxtaposing our results with those of
Scherer (1993). Scherer (1993) found that an appraisal of the time of the
emotion-eliciting event (future vs. past or present) did not seem to ade-
quately distinguish f ear and anxiety vs. other negative emotions. He notes
(p. 349) that ``it may be necessary to go beyond the straightforw ard timing
issue and include dimensions such as certainty’’ . Consistent with this
observation and the hypothesis of Roseman (1979, 1984), we found that
the uncertain vs. certain probability apprai sal succeeded in differentiating
fear from sadness and distress. This suggests a second appraisal sequence:
Typically, perceiving a negative event as existing in the future leads to
believing that the event is uncertain, which can lead to feeling fear. But
future time-frame is not a
nece ssary
cause of fear because some future
negative events are certain. For example, one may learn that one is
certainly going to die in six months from a terminal illness and therefore
feel sadne ss.
These examples indicate that investigation of appraisal sequences may
help integrate different appraisal theories within a comprehensive model.
That is, different theories may have identi ® ed differe nt appraisals because
they focus on different points in a sequence of appraisals that elicit
emotions. We have suggested here that appraisals of legitimacy (Averill,
1982; Roseman, 1979; Scherer, 1984) tend to in¯ uence appraisals of
control potential (cf. Bandura, 1977; Frijda, 1986; Lazarus & Smith,
1988; Scherer, 1988); and that appraisals of the time of an event
(Scherer, 1984) in¯ uence appraisals of its probability. Control potential
and probability appraisals in turn affect emotions. In thi s view, the apprai-
sals in Fig. 2 are the proximal, minimally necessary antecedents of emo-
tions, and the other appraisals we have been discussing are distal, typical
determinants (cf. Lazarus & Smith, 1988).
As Frijda (1993) suggests, it is possible that each of the appraisals that
are minimally necessary to produce emotions can be processed at a
rudimentary cogni tive level; and Frijda discusse s e leme ntary ve rsions of
a number of the appraisals shown in Fig. 2. Inde ed, one reason for
abandoning legitimacy (Roseman, 1979) in favour of c ontrol potential as
a necessary antecedent of emotions such as anger (see Fig. 2) is that infants
appear to experience anger around four months of age (Stenberg &
Campos, 1990)Ð before they would seem to have the cognitive capacity
to make sophisicated judgements of legitimacy. In contrast, a ``primiti ve
appraisal’ ’ of control potential may be as s imple as recognising a situation
as one for which an action schema is available; or expecting that a given
progression of events is going to occur.
It is of considerable interest, particularly for establishing connections to
physiological mechanisms, to know the immediate, necessary, simplest
determinants of emotional responses such as sadness, fear, anger, shame,
C O N S T R U C T IN G A N I M P R O V E D A P P R A IS A L T H E O R Y
2 7 1
2 7 2
R O S E M A N , A N T O N I O U , J O S E
and guilt (e.g. primitive appraisals of control potential). But it is also
important, for the study of emotions in their natural and social context,
and for understanding sources of emotional pathology and the possibilities
for treatment and self-regulation, to know their typical distal antec edents
(e.g. appraisals of legitimacy or blameworthiness). Both distal and prox-
imal appraisals ne ed to be included in a comprehe nsive theory of the causes
of emotions.
Manuscript received 19 January 1995
Revised manuscript received 9 August 1995
R E F E R E N C E S
Arnold, M.B. (1960).
Emotion and personality
. New York: Columbia University Press.
Ausubel, D. (1955). Relationships between shame and guilt in the socializing process.
Psychological Review, 62
, 378± 390.
Averill, J.R. (1982).
Anger and aggres si on: An essay on emotion
. New York: Springer-
Verl ag.
Bandura, A. (1977). Self-ef® cacy: Toward a unifying theory of behavioral change.
Psycho-
logical Review, 84
, 191± 215.
Batson, C.D., Fultz, J., & Schoenrade, P.A. (1987). Distress and empathy: Two qualita-
tively distinct vicarious emotions with different motivational consequences.
Journal of
Personality, 55
, 19± 40.
Ben-Ze’ev, A. (1994). Typical cases of ple asure-in-others’ misfortune. In N.H. Frijda (Ed.),
Proce edin gs of the 8th Conference of the International Society for Researc h on Emotions
(pp. 190± 193). Storrs, CT: International Society for Research on Emotions.
Charlesworth, W.R. (1969). The role of surprise in cognitive development. In D. Elkind &
J. Flavell (Eds.),
Studies in cognitive development
. London: Oxford University Press.
Clore , G.L., Schwarz, N., & Conway, M. (1994). Affective causes and consequences of
social information processing. In R.S. Wyer & T.K. Srull (Eds.),
Handbook of social
cognition: Applications
. (Vol. 2, 2nd ed.). Hillsdale, NJ: Lawrence Erlbaum Associates
Inc.
Ekman, P., & Friesen, W.V. (1986). A new pan-cultural facial expression of emotion.
Motivation and Emotion, 10
, 159± 168.
Ekman, P., & Heider, K.G. (1988). The universality of a contempt expression: A replica-
tion.
Motivation and Emotion, 12
, 303± 308.
Ellsw orth, P.C., & Smith, C.A. (1988). Shades of joy: Patterns of appraisal differentiating
pleasant emotions.
Cogniti on and Emotion, 2
, 301± 331.
Frijda, N.H. (1986).
The emotions
. New York: Cambridge University Press.
Frijda, N.H. (1987). Emotion, cognitive structure, and action tendency.
Cognition and
Emotion, 1
, 115± 143.
Frijda, N.H. (1993). The place of appraisal in emotion.
Cognition and Emotion, 7
, 357± 387.
Frijda, N.H., Kuipers, P., & ter Schure, E. (1989). Relations among emotion, appraisal, and
emotional action readiness.
Journal of Personality and Soc ial Psychol ogy, 57
, 212± 228.
Garber, J., & Hol lon, S.D. (1980). Universal versus personal helplessness in depression.
Journal of Abnormal Psychology, 89
, 56± 66.
Izard, C.E. (1977).
Human emotions
. New York: Plenum.
Izard, C.E. (1991).
The psychology of emotions
. New York: Plenum.
Izard, C.E., & Haynes, O.M. (1988). On the form and universality of the contempt expres-
sion: A challenge to Ekman and Friesen’s claim of disc overy.
Motivation and Emotion,
12
, 1± 16.
Izard, C.E., & Malatesta, C.Z. (1987). Emotional deve lopment in infancy. In J. Osofsky
(Ed.),
Handbook of infant development
(Vol. 2, pp. 494± 554). New York: Wiley.
Janoff-Bulman, R. (1979). Characterological versus behavioral self-blame: Inquiries into
depression and rape.
Journal of Personal ity and Social Psychology, 37
, 1798± 1809.
Kemper, T.D. (1978).
A social interactional theory of emotions
. New York: Wiley.
Lazarus, R.S. (1991a). Progress on a cognitive -motivational-relational theory of emotion.
American Psychologist, 46
, 819± 834.
Lazarus, R.S. (1991b).
Emotion and adaptation
. New York: Oxford University Press.
Lazarus, R.S., & Smith, C.A. (1988). Knowledge and appraisal in the cognition-emotion
relati onship.
Cognition and Emotion, 2
, 281± 300.
Lewis, H.B. (1971).
Shame and guilt in neurosis
. New York: International Universitie s
Press.
Manstead, A.S.R., & Tetlock, P.E. (1989). Cognitive appraisals and emotional experience:
Further evidence.
Cognition and Emotion, 3
, 225± 240.
Mauro, R., Sato, K., & Tucker, J. (1992). The role of appraisal in human emotions: A cross-
cultural study.
Journal of Personality and Social Psychology, 62
, 301± 317.
McGraw, K.M. (1987). Guilt following transgression: An attribution of responsibility
approach.
Journal of Personal ity and Social Psychology, 53
, 247± 256.
Miller, J.G. (1984). Culture and the de vel opment of everyday social explanations.
Journal
of Personality and Social Psychology, 46
, 961± 978.
Nisbett, R.E., & Wilson, T.D. (1977). Telling more than we can know: Verbal reports on
mental proces se s.
Psychological Review, 84
, 231± 259.
Nunnally, J.C. (1967).
Psychome tric theory
. New York: McGraw-Hill.
Oatley, K., & Johns on-Laird, P.N. (1987). Towards a cognitive theory of emotions.
Cogni-
tion and Emotion, 1
, 29± 50.
Ortony, A., Clore, G.L., & Collins, A. (1988).
The cognitive structure of emotions
. New
York: Cambridge University Press.
Osgood, C.E., May, W.H., & Miron, M.S. (1975). Cross-cultural universals of affective
meaning. Urbana: University of Illinois Press .
Parkinson, B., & Manstead, A.S.R. (1992). Appraisal as a cause of emotion. In M.S. Clark
(Ed.),
Review of per sonality and social psychology
(Vol. 13, pp. 122± 149). Newbury
Park, CA: Sage.
Parkinson, B., & Manstead, A.S.R. (1993). Maki ng sense of emotion in stories and social
life.
Cognition and Emotion, 7
, 295± 323.
Parrott, W.G. (1992). The emotional experiences of envy and jealous y. In P. Salovey (Ed.),
The psychology of envy and jealousy
(pp. 3± 30). New York: Guilford.
Reisenzein, R., & Hofmann, T. (1990). An investigation of dimensions of cognitive apprai-
sal in emotion using the repertory grid technique.
Motivation and Emotion, 14
, 1± 26.
Reisenzein, R., & Hofmann, T. (1993). Discriminating emotions from appraisal-rele vant
si tuational information: Baseline data for structural models of cognitive apprais als.
Cognition and Emotion, 7
, 295± 323.
Roseman, I.J. (1979, September).
Cognitive aspects of emotion and e motional behavior
.
Paper pres ented at the 87th Annual Convention, American Psychological Association,
New York, NY.
Roseman, I.J. (1984). Cognitive determinants of emotions: A structural theory. In P. Shaver
(Ed.),
Review of personality and social psychology
(Vol. 5, pp. 11± 36). Beverly Hills,
CA: Sage.
C O N S T R U C T IN G A N I M P R O V E D A P P R A IS A L T H E O R Y
2 7 3
2 7 4
R O S E M A N , A N T O N I O U , J O S E
Roseman, I.J. (1991). Appraisal determinants of discrete emotions.
Cognition and Emotion,
5
, 161± 200.
Roseman, I.J. (1994a, Jul y) .
The discrete emotions form a coherent set: A theory of
emotional responses
. Paper presented at the 6th Annual Convention, American Psycho-
logical Society, Washington, DC.
Roseman, I.J. (1994b). Emotions and emotion families in the emotion system. In N.H.
Frijda (Ed.),
Proce edi ngs of the 8th Conference of the International Society for Research
on Emotions
(pp. 171± 175). Storrs, CT: International Society for Research on Emotions.
Roseman, I. J., Dhawan, N., Rettek, S.I., Naidu, R.K., & Thapa, K. (1995). Cultural differ-
ences and cross-cultural similarities in appraisals and emotional responses.
Journal of
Cross-Cultural Psychology, 26
, 23± 48.
Roseman, I.J., & Evdokas, A. (1995, June).
Appraisals do cause r eal emotions: Experi-
mental e vidence
. Pape r presented at the 7th Annual Convention, American Psyc hological
Society, New York.
Roseman, I.J. , Spindel, M.S., & Jose, P.E. (1990). Appraisals of emotion-eliciting events:
Testing a theory of discrete emotions.
Journal of Personality and Soc ial Psychology, 59
,
899± 915.
Roseman, I.J., Wiest, C., & Swartz, T.S. (1994). Phenomenol ogy, behaviors, and goals
diff erentiate discrete emotions.
Journal of Personality and Social Psychology, 67
,
206± 221.
Rosenthal, R., & Rosnow, R.L. (1985).
Contrast analysis: Focus ed comparisons in the
analysis of variance
. New York: Cambridge University Press.
Rozin, P., & Fallon, A. (1987). A perspective on disgust.
Psychological Review, 94
, 23± 41.
Scherer, K.R. (1984). On the nature and function of emotion: A component process
approach. In K.R. Scherer & P. Ekman (Eds.),
Approaches to emotion
(pp. 293± 317).
Hillsdale, NJ: Lawrence Erlbaum Associates Inc.
Scherer, K.R. (1986). Emotion experiences across European cultures: A summary s tate-
ment. In K.R. Scherer, H.G. Wallbott, & A.B. Summer® eld (Eds.),
Experiencing emo-
tion: A cros s-c ultural study
(pp. 172± 190). New York: Cambridge University Press.
Scherer, K.R. (1988). Criteria for emotion-antecedent appraisal: A review. In V. Hamilton,
G.H. Bower, & N.H. Frijda (Eds.),
Cognitive perspectives on emotion and motivation
(pp.
89± 126). Boston: Kluwer.
Scherer, K.R. (1993). Studying the emotion-antecedent appraisal process: An expert system
approach.
Cognition and Emotion, 7
, 325± 355.
Sel igman, M.E.P. (1975).
Helplessness: On depression, development, and death
. San Fran-
cisco: Freeman.
Shaver, K.G. (1985).
The attr ibution of blame: Causality, responsibility, and blameworthi-
ness
. New York: Springer-Verlag.
Shaver, P., Schwartz, J., Kirson, D., & O’ Connor, C. (1987). Emotion knowledge: Further
exploration of a prototype approach.
Journal of Personal ity and Social Psychology, 52
,
1061± 1086.
Smith, C.A., & Ellsworth, P.C. (1985). Patterns of cognitive appraisal in emotion.
Journal
of Personality and Social Psychology, 48
, 813± 838.
Smith, C.A., & Ellsworth, P.C. (1987). Patterns of appraisal and e motion related to taking
an exam.
Journal of Pe rs onali ty and Social Psychology, 52
, 475± 488.
Smith, C.A., & Laz arus, R.S. (1990). Emotion and adaptation. In L.A. Pervin (Ed.),
Hand-
book of personality: Theory and research
(pp. 609± 637). New York: Guilford.
Smith, C.A. , & Lazarus, R.S. (1993). Appraisal components, core relational themes, and the
emotions.
Cognition and Emotion, 7
, 233± 269.
Solomon, R.C. (1976).
The passions
. Garden City, NY: Doubleday.
Spence , K.W. (1956).
Behavior theory and conditioning
. New Haven: Yale University
Press.
Stein, N., & Levine, L. (1987). Thinking about feelings: The development and organization
of emotional knowledge. In R.E. Snow & M. Farr (Eds.),
Aptitude, learning and
instruction: Cognition, conation, and affect
(Vol. 3, pp. 165± 198). Hillsdale, NJ: Law-
rence Erlbaum Associates Inc.
Stenberg, C.R., & Campos, J.J. (1990). The development of anger express ions in infancy. In
N. Stein, B. Leventhal, & T. Trabasso (Eds.),
Psychological and biological approaches to
emotion
(pp. 247± 282). Hillsdale, NJ: Lawrence Erlbaum Associates Inc.
Tangney, J.P. (1993). Shame and guilt. In C.G. Costello (Ed.)
Symptoms of depression
(pp.
161± 180). New York: Wiley.
Wallbott, H.G., & Scherer, K.R. (1988). How universal and speci® c is emotional experi -
ence? Evidence from 27 countries on ® ve continents. In K.R. Scherer (Ed.),
Fac ets of
emotion: Recent research
(pp. 31± 56). Hillsdale, NJ: Lawrence Erlbaum Associates Inc.
Weiner, B. (1985). An attributional theory of achievement motivation and emotion.
Psy-
chological R eview, 92
, 548± 573.
Weiner, B., Graham, S., & Chandler, C. C. (1982). Pity, anger, and guilt: An attributional
analysi s.
Personality and Social Psychology Bulletin, 8
, 226± 232.
Weisz, J.R., Sweeney, L., Prof® tt, V., & Carr, T. (1993). Control-related beliefs and self -
reported depressive symptoms in late childhood.
Journal of Abnormal Psychology, 102
,
411± 418.
Weisz, J.R., Stevens, J.S., Curry, J.F., Cohen, R., Craighead, W.E., B urlingame, W.V.,
Smith, A., Weiss, B., & Parmalee, D.X. (1989). Control-related cognitions and depres-
sion among inpatient children and adolescents.
Journal of the American Academy of
Child and Adolescent Psychiatry, 28
, 358± 363.
Weisz, J.R., Weiss, B., Wasserman, A.A., & Rintoul, B. (1987). Control-re lated beliefs and
depression among clinic-referred children and adolescents.
Journal of Abnormal Psy-
chology, 96
, 58± 63.
Wortman,C.B.,& Brehm,J.W. (1975). Responsestouncontrollableoutcomes:Anintegration
of reactancetheor yandthe learnedhelplessnessmodel.In L. Berkowitz(Ed.),
Advances in
experimental social psychology
(Vol. 8, pp. 277± 336) New York:Academic Pr ess.
A P P E N D I X
S te m a n d S c a le A n c h o r s f o r It e m s M e a s u rin g A p p r a is a l s
My {emotion term}
a
was caused by:
Unusualne ss
(Unexpected)
SPECIFIC EVENT A being expected
(1) to
SPECIFIC EVENT A being
unexpected
(9).
(Unfamiliar*)
Thinking of SPECIFIC EVENT A as something unfamiliar
(1) to
Thinking of
SPECIFIC EVENT A as something familiar
(9).
(Novel)
Thinking of SPECIFIC EVENT A as s omething that was not novel
(1) to
Thinking of
SPECIFIC EVENT A as something that was novel
(9).
Situati onal State
(Motive -Consistency)
Thinking that SPECIFIC EVENT A was inconsistent with what I
wanted
(1) to
Thinking that SPECIFIC EVENT A was consistent with what I wanted
(9).
(Wanted*)
SPECIFIC EVENT A being very much wanted by me
(1) to
SPECIFIC EVENT A
being very much unwanted by me
(9).
(Improved*)
Beli eving that SPECIFIC EVENT A improved things
(1) to
Beli eving that
SPECIFIC EVENT A made things worse
(9).
C O N S T R U C T IN G A N I M P R O V E D A P P R A IS A L T H E O R Y
2 7 5
2 7 6
R O S E M A N , A N T O N I O U , J O S E
Motivational State
(Pleasurable*)
Wanting to get or keep something pleasurable
(1) to
Wanting to get rid of or
avoid something painful
(9).
(Bene® t)
Wanting to minimise some cost in SPECIFIC EVENT A
(1) to
Wanting to maximise
some bene® t in SPECIFIC EVENT A
(9).
(Positive)
My seeking less of something negative during SPECIF IC EVENT A
(1) to
My
seeking more of something positive during SPECIFIC EVENT A
(9).
Probability
(Able to Predict)
Being unable to predict what was going to happen in this situation
(1) to
Being able to predi ct what was going to happen in this situation
(9).
(Certain*)
Being certain about the consequences of SPECIFIC EVENT A
(1) to
Being
uncertain about the consequences of SPECIFIC EVENT A
(9).
(Not in Doubt*)
Being not at all in doubt about something that mattered to me
(1) to
Being
ve ry much in doubt about some thing that mattered to me
(9).
Own Power
(Powerful *)
Feeli ng that I was powerful
(1) to
Feeling that I was powerle ss
(9).
(Strong)
Believing that I was weak
(1) to
Believing that I was strong
(9).
(Respond Effectively)
Believing that I lacked the power to respond effectively to SPECIFIC
EVENT A
(1) to
Believing that I had the power to respond effective ly to SPECIFIC
EVENT A
(9).
Stimulus Powerlessne ss
(Less Powerful)
Thinking that SPECIFIC EVENT A was something more powerful than I was
(1) to
Thinking that SPECIFIC EVENT A was something less powerful than I was
(9).
(Something Weak*)
Believing that in SPECIFIC EVENT A I was dealing with something that
was weak
(1) to
Believ in g that in SPECIFIC EVENT A I was dealing with something that
was strong
(9).
(Easy*)
Believing that it would be easy to de al with SPECIFIC EVENT A
(1) to
Believing
that it would be dif® cult to deal with SPECIFIC EVENT A
(9).
Stimulus Controllability
(Control)
Believing I had no control at all over SPECIFIC EVENT A
(1) to
Believing I had
ve ry much control over SPECIF IC EVENT A
(9).
(Modi® able)
Thinking that SPECIFIC EVENT A was not modi® able
(1) to
Thinking that
SPECIFIC EVENT A was modi® able
(9).
(Someone Could Change*)
Thinking that someone could change SPECIFIC EVENT A
(1) to
Thinking that no one could change SPECIFIC EVENT A
(9).
Stimulu s Controllability by Self
(Controllable by Me*)
Thinking that SPECIFIC EVENT A was controllable by me
(1) to
Thinking that SPECIFIC EVENT A was uncontrollable by me
(9).
(Modi® able by Me)
Thinking that SPECIFIC EVENT A was not modi® able by me
(1) to
Thinking that SPECIFIC EVENT A was modi® able by me
(9).
(I Could Change*)
Thinking that I could change SPECIFIC EVENT A
(1) to
Thinking that I
could not change SPECIFIC EVENT A
(9).
Coping Potential
(Not Too Dif® cult*)
Thinking that it would not be too dif® cult to deal with SPECIFIC
EVENT A
(1) to
Thinking that it would be too dif® cult to deal with SPECIFIC EVENT A
(9).
(Able to Cope)
Believing that I was unable to cope with SPECIFIC EVENT A
(1) to
Believing that I was able to cope with SPECIFIC EVENT A
(9).
(Something I Could Do*)
Thinking that there was something I could do about SPECIFIC
EVENT A
(1) to
Thinking that there was nothing I could do about SPECIFIC EVENT A
(9).
Legitimacy
(Deserve Good)
Believing that I des er ved for something bad to happen
(1) to
Believing that I
deserved for something good to happen
(9).
(Morall y Right*)
Thinking of myself as morally right
(1) to
Thinking of myself as morally
wrong
(9).
(Praise worthy)
Thinking that I was blameworthy
(1) to
Thinking that I was praiseworthy
(9).
Problem Sour ce
(Being)
What someone or something was or was not doing
(1) to
What someone or something
was or was not being
(9).
(Character*)
Someone’ s or something’ s character as shown in SPECIFIC EVENT A
(1) to
Someone’ s or something’ s behaviour as s hown in SPECIFIC EVENT A
(9).
(Basic Nature)
Thinking that SPECIFIC EVENT A di d not reveal the basic nature of someone
or something
(1) to
Thinking that SPECIFIC EVENT A did reveal the basic nature of
someone or something
(9).
Responsibility
(Sel f-Responsibility)
Thinking that I was not at all responsible for SPECIFIC EVENT A
(1)
to
Thinking that I was very much responsible for SPECIFIC EVENT A
(9).
(Other-Person-Responsibility)
Thinking that someone else was not at all responsible for
SPECIFIC EVENT A
(1) to
Thinking that someone else was very much responsible for
SPECIFIC EVENT A
(9).
(Circumstance-Responsibility)
Thinking that circumstances beyond anyone’ s contol were not
at all responsible for SPECIFIC EVENT A
(1) to
Thinking that circumstances beyond
anyone’ s control were very much responsible for SPECIFIC EVENT A
(9).
Agency
(Self-Agency)
Thinking that SPECIFIC EVENT A was not at all caused by me
(1) to
Thinking that SPECIFIC EVENT A was very much caused by me
(9).
(Other-Person-Agency)
Thinking that SPECIFIC EVENTA was not at all caused by someone
else
(1) to
Thinking that SPECIFIC EVENTA was very much caused by someone e lse
(9).
(Circumstance-Agency)
Thinking that SPECIFIC EVENT A was not at all caused by
circumstances beyond anyone’ s control
(1) to
Thinking that SPECIFIC EVENT A was
ve ry much caused by circumstances beyond anyone’ s control
(9).
Note
. Here, hypothesised appraisal dimension names are shown in
bold
, and appraisal
item names in parentheses. An asterisk (*) i ndicates that responses were reverse -coded to
meas ure an appraisal.
a
The name of the emotion being recalled in a particular experience was inserted here.
C O N S T R U C T IN G A N I M P R O V E D A P P R A IS A L T H E O R Y
2 7 7
