-
The Moral Emotions: A SocialFunctionalist Account of
Anger,Disgust, and Contempt
Cendri A. Hutcherson and James J. Gross
Stanford University
Recent research has highlighted the important role of emotion in
moral judgment and decision making(Greene, Sommerville, Nystrom,
Darley, & Cohen, 2001; Haidt, 2001). What is less clear is
whetherdistinctions should be drawn among specific moral emotions.
Although some have argued for differencesamong anger, disgust, and
contempt (Rozin, Lowery, Imada, & Haidt, 1999), others have
suggested thatthese terms may describe a single undifferentiated
emotional response to morally offensive behavior(Nabi, 2002). In
this article, we take a socialfunctionalist perspective, which
makes the prediction thatthese emotions should be differentiable
both in antecedent appraisals and in consequent actions
andjudgments. Studies 13 tested and found support for our
predictions concerning distinctions amongantecedent appraisals,
including (a) a more general role for disgust than has been
previously beendescribed, (b) an effect of self-relevance on anger
but not other emotions, and (c) a role for contempt injudging
incompetent actions. Studies 4 and 5 tested and found support for
our specific predictionsconcerning functional outcomes, providing
evidence that these emotions are associated with
differentconsequences. Taken together, these studies support a
socialfunctionalist account of anger, disgust, andcontempt and lay
the foundation for future research on the negative interpersonal
emotions.
Keywords: moral emotions, anger, disgust, contempt,
sociofunctional
How distinct are discrete emotions? This question has beenone of
the most hotly debated within emotion research (L. F.Barrett, 2006;
L. F. Barrett et al., 2007; Izard, 2007). Over thedecades,
researchers have attempted to create a theoretical andempirical
framework describing when and why the different emo-tion words we
use should be considered to reflect true differencesin underlying
psychological states. Yet despite an enormous cor-pus of research,
disagreements over the structure of emotionalspace (e.g.,
dimensional vs. discrete models; Russell, 2003), aswell as over
individual emotions, still abound. For the triad ofnegative social
emotions concerned with judging the actions anddispositions of
others (anger, disgust, and contempt; Haidt, 2003),the proper
division of space has seemed particularly elusive.1
The link between these negative emotions and our responses
tosocially relevant behavior is easy to detect. We might say that
weare angry at injustice, disgusted by a heinous murder, or
contemp-tuous of corrupt politicians. On a more mundane level, too,
thedriver who fails to signal before swerving across lanes, the
man-ager who picks his nose, the inept store clerk who drops
ourgroceriesall can be targets of our anger, disgust, and
contempt.
Yet even though we use different words to describe these
reac-tions, it is unclear whether they correspond to different
underlyingpsychological states with different consequences. More
often thannot, these terms seem to bleed into one another (e.g.,
Marzillier &Davey, 2004; Nabi, 2002; Simpson, Carter, Anthony,
& Overton,2006), raising this question: Are there meaningful
differencesamong the emotions described by these three terms, and
if so, whatare they?
Defining the Space of Moral Emotions: An Outline ofPossible
Models
Figure 1 presents a range of possible models depicting
therelationship among these three moral emotions, from least to
mostdifferentiated. In the simplest model, anger, disgust, and
contempt
1 In addition to the other-condemning emotions, Haidt (2003)
alsodescribed three additional families of sociomoral emotions,
including theother-praising emotions (gratitude, awe), which are in
many ways thefunctional opposites of the moral-condemning emotions;
the other-suffering emotions (such as sympathy or compassion),
which are negativefeelings associated with another persons pain or
misfortune; and theself-conscious emotions (such as shame, guilt,
and pride), which fall withinthe domain of the sociomoral emotions
because they involve appraisals ofthe self as a social or moral
object. The extent to which each of thesefamilies includes
separate, distinct emotions also has yet to be
completelydetermined; this article focuses on the other-condemning
emotions, al-though the results presented here may have
implications for understandingemotions within the other families.
For a fascinating discussion of theseparability of the negative
self-conscious emotions, see Tangney, Wagner,Hill-Barlow,
Marschall, and Gramzow, 1996.
This article was published Online First January 31, 2011.Cendri
A. Hutcherson and James J. Gross, Department of Psychology,
Stanford University.We thank Anne Cherniss, Ken Ferry, Rebecca
Roediger, and Hants
Williams for assistance with data collection.Correspondence
concerning this article should be addressed to Cendri A.
Hutcherson, who is now at the Division of Humanities and Social
Sciences,California Institute of Technology, MC 228-77, Pasadena,
CA 91125.E-mail: [email protected]
Journal of Personality and Social Psychology 2011 American
Psychological Association2011, Vol. 100, No. 4, 719737
0022-3514/11/$12.00 DOI: 10.1037/a0022408
719
-
are semantic equivalents, synonyms for the same underlying
neg-ative emotional state evoked by interpersonal concerns (e.g.,
Nabi,2002). In a partially differentiated model, two emotion
families aredistinguishable, typically on the basis of the
dimension of attack/approach and avoid/withdraw: anger on the one
hand and disgust/contempt on the other. Models of this type have
begun to appearin the stereotyping and prejudice literature (e.g.,
Fiske, Cuddy,Glick, & Xu, 2002; Mackie, Devos, & Smith,
2000). In a fullydifferentiated model, the three emotions are
distinct. One instan-tiation of this modeldescribed belowis the CAD
triad model(Rozin, Lowery, Imada, & Haidt, 1999), in which
these threedifferent emotions result from a single appraisal (moral
domain)with three categories (autonomy, divinity, community). Each
ofthese models has received some empirical support, but it is not
yetclear how to reconcile conflicting findings.
Distinguishing the Triad: A SocialFunctionalistAccount
Our perspective on this issue, favoring the fully
differentiatedmodel, derives from a functionalist account of
emotion, whichargues that emotions are adaptive solutions
comprising a coordi-nated set of appraisals, communicative
gestures, physiologicalresponses, and action tendencies tailored to
respond to crucialproblems faced by our species over the millennia
(Cosmides &Tooby, 2000; Tooby & Cosmides, 1990). One may
thus differen-tiate emotion families by considering the extent to
which theyserve different purposes (K. Barrett & Campos, 1987).
Socialfunctionalism examines the adaptive role that emotions play
inrelations between individuals, groups, and cultures (Ekman,
1992;Keltner & Haidt, 1999; Keltner, Haidt, & Shiota,
2006). Thenumber of different social/moral emotions can be inferred
byidentifying distinct dimensions of socially relevant actions
forwhich advantageous response requires different sets of changes
inbehavior, cognition, and/or motivation. The question is, within
theinterpersonal domain, how many unique situations,
requiringunique sets of responses, are there?
Anger, Appraisals of Self-Relevance, and
DefensiveAttack/Approach Behavior
One of the most basic social challenges may involve the
abilityto discern the ways in which anothers actions immediately
impact
the self. When those actions lead to direct threat, vigorous
defenseor attack can be the quickest way to resolve the danger,
despite thepotential risks and energy expenditure. When the self is
not di-rectly involved, passive avoidance may be a less costly
means toreduce threat. To successfully navigate the social world,
an indi-vidual must be able to distinguish and respond
appropriately tothose circumstances in which the threat posed by a
social actormerits more proactive, but potentially costly,
strategies or morepassive, conservative tactics. In our view, anger
serves this func-tion.
Unlike other negative emotions, which typically prompt
avoid-ance behavior, anger tends to promote approach tendencies, in
theform of attack (Harmon-Jones & Allen, 1998; Roseman, Wiest,
&Swartz, 1994), although others have also noted an increase
inavoidance as well (Averill, 1982). Consequently, anger results
inhigher energy expenditure, evidenced by greater autonomicarousal
and behavioral activation (Levenson, Ekman, & Friesen,1990) and
a greater willingness to take risks (Lerner & Keltner,2001;
Lerner & Tiedens, 2006). These higher cost behaviors maybe
designed to prevent or terminate specific behaviors that
areperceived as immediate threats to the self, and anger may
alsoincrease the willingness to incur costs in order to punish
betrayal(de Quervain et al., 2004; OGorman, Wilson, & Miller,
2005).One implication of this view is that the morality per se of a
socialbehavior may be less important in determining an angry
responsethan the extent to which it impacts the self. Moreover,
despite itsmoral overtones, anger may be concerned less with how to
respondto peoples moral intentions and dispositions and more with
howbest to actively respond to their behavior.
Moral Disgust, Contempt, and Appraisals ofCharacter: Morality
Versus Competence
Although anger may fill the need to respond directly to
imme-diate social threats, there is a complementary need to
respondappropriately to individuals who are not currently engaged
inthreatening behavior but whose past behavior suggests they
shouldbe avoided, a less costly but more widely applicable
response.This, we argue, is the primary function of both moral
disgust andcontempt: to mark individuals whose behavior suggests
that theyrepresent a threat and avoid them, thereby reducing the
risk ofexposure to harm.
Figure 1. Possible relations among anger, disgust, and contempt
in the social domain. A: Equivalent model:Moral judgment of others
is represented by a single underlying emotional state. B: Partially
distinct model:Anger is distinct, but moral disgust and contempt
are overlapping terms for a single underlying state. C:
Fullydifferentiated model: Social judgment is captured by three
different, nonoverlapping emotions.
720 HUTCHERSON AND GROSS
-
However, a simple mapping between approachanger and
with-drawalcontempt/moral disgust may not be sufficient to
encom-pass the full range of socially relevant situations and
requisiteadaptive responses. A rich body of research suggests that
socialjudgments are composed of two orthogonal dimensions. The
firstdimension, referred to in different literatures as morality,
warmth,communion, or affiliation, describes the benevolence or
malevo-lence of an actors intentions. The second dimension,
referred to ascompetence, dominance, or agency, describes the
abilities, skills,and resources an actor has at his or her disposal
to carry out thoseintentions (Horowitz et al., 2006; Leary, 1957;
Skowronski &Carlston, 1987; Wojciszke, 2005). Approaching those
with exper-tise or knowledge and avoiding those who may be a
burdenbecause they lack the ability to contribute to the social
group alsohas inherent adaptive value. In our view, the need to
distinguishthreats posed by immorality or incompetence,
respectively, maycorrespond to the functions of moral disgust and
contempt.
Haidt, Rozin, McCauley, and Imada (1997) treated moral dis-gust
as a socially elaborated version of nonsocial disgust (elicitedby
death, decay, and disease). Nonsocial (core) disgust evokes
adeceleration in heart rate (Levenson et al., 1990), avoidance
(Ola-tunji & Sawchuk, 2005; Rozin & Fallon, 1987), and a
facialexpression designed to limit contact with the environment
(Rozin& Fallon, 1987; Susskind et al., 2008). Disgust also
results in alasting reluctance to come into contact with objects
themselvescontaminated by contact with something disgusting.
Althoughsimilarities between nonsocial and social disgust have
receivedonly indirect empirical support, such similarities could
have func-tional significance (Tybur, Lieberman, &
Griskevicius, 2009).Once marked as capable of malicious behavior,
an individualshould be consistently avoided, regardless of whether
he or shesubsequently performs a few benevolent actions (Martijn,
Spears,Vanderpligt, & Jakobs, 1992; Reeder & Coovert,
1986). Moraldisgust may serve this function.
Contempt, on the other hand, may be more strongly concernedwith
appraisals of incompetence. Reliance on the help of anincompetent
individual risks wasting time or resources. Moreover,seeking
outside expertise and accepting the obligations entailed byaid is
useful only in situations where one does not oneself have
theresources to accomplish a task. Contempt may function to
diminishinteraction with individuals who cannot contribute in a
meaningfulway to the group, especially those individuals judged to
be lower
or less capable than the self (Izard, 1977), yet who do not
behavein intentionally malicious ways.
As with anger, several predictions concerning the effects
ofthese emotions on judgment and decision making stem from theview
presented above. In particular, both moral disgust and con-tempt
may encourage a focus on a persons intentions and dispo-sitions,
leading to more stable, longer lasting attributions concern-ing a
persons character. In contrast, anger may be more likely
todissipate once the immediate threat or injustice has been
properlypunished or the appropriate restitution has been made
(Ohbuchi,Kameda, & Agarie, 1989).
Empirical Evidence: Differentiation and Overlap
The account we have outlined above suggests that anger,
moraldisgust, and contempt should be differentiable on the basis
ofappraisals that elicit one emotion more strongly than the
others,with corresponding differences in the behaviors consequent
toeach emotion (see Figure 2A). Unfortunately, little empirical
re-search has directly tested distinctions between these
emotions.
Evidence for Differentiation
The most comprehensive evidence for differentiation among
thethree other-condemning emotions comes from work on the CADtriad
hypothesis (Haidt, 2003; Rozin et al., 1999). Informed bycultural
and anthropological discourse, and motivated as we are bya
socialfunctionalist perspective, Rozin et al. (1999) argued
thatthese emotions function specifically within the moral domain
todistinguish behaviors that violate three moral ethics (see
Figure2B). Humans feel contempt for violations of community
(respectfor hierarchical and communal obligations of an individual
tosociety). They feel anger for violations of autonomy (disregard
forthe personal rights or freedoms of the individual). Finally,
they feeldisgust for violations of divinity (disrespecting the
sacredness ofGod or causing degradation or pollution to oneself or
another). Ina series of studies, these researchers presented
participants in bothJapan and the United States with actions
representing one of thethree moral codes (examples of which can be
seen in Table 1) andasked them to select the most applicable facial
expression andverbal label from examples of disgust, contempt, and
anger. Par-ticipants linked the expression and label of contempt
with com-
Figure 2. Two different fully differentiated models. A: The
socialfunctionalist model. B: The CAD triadmodel.
721SOCIALFUNCTIONALIST ACCOUNT OF MORAL EMOTIONS
-
munity violations, anger with autonomy violations, and
disgustwith divinity violations.
Evidence for Overlap
Although the data for distinctions among anger, moral
disgust,and contempt are appealing, a growing body of empirical
evidence
is at odds with this view. Whereas participants reported
feelingonly disgust toward nonsocial disgust elicitors, they
endorsed arange of emotion terms, including disgust and anger,
toward so-ciomoral violations (Marzillier & Davey, 2004). Anger
and moraldisgust are usually highly correlated (Simpson et al.,
2006), per-haps because people employ the terms anger and
disgust
Table 1Emotion Intensity Ratings in the Context of Community,
Autonomy, and Divinity Violations (Study 1)
Item Code
Anger Moral disgust Contempt Sadness Fear Grossed out
M SD M SD M SD M SD M SD M SD
A child hits another child A 2.46a,b 1.79 2.29b 1.90 1.79c 1.79
2.78a 1.94 1.65c 1.80 0.73d 1.33Someone edges ahead in a very
long
line A 3.38a 1.58 2.79b 1.68 2.96b 1.69 0.86c 1.29 0.71c 1.26
0.62c 1.31A person is smoking in the
nonsmoking section of a smallwaiting area A 3.65a 1.79 3.18b
1.79 3.28b 1.83 1.21d 1.60 0.99d 1.44 2.65c 2.20
Someone embezzles from a bank A 3.35b 1.54 4.27a 1.25 3.52b 1.73
1.85c 1.85 1.25d 1.73 0.83e 1.50A person fakes an injury after
an
automobile accident in order tocollect on insurance A 3.67b 1.52
4.63a 1.30 3.80b 1.66 1.86c 1.85 0.84d 1.37 1.08d 1.75
Someone is a member of the Ku KluxKlan A 4.74b 1.51 5.35a 1.12
4.78b 1.59 3.73c 2.20 3.24d 2.30 2.25e 2.53
A man comes home drunk and beatshis wife A 5.22b 1.23 5.41a 0.95
4.94c 1.61 4.88c 1.51 3.29d 2.34 2.43e 2.46
A person puts cyanide in a yogurtcontainer at the supermarket A
5.04b 1.46 5.50a 1.00 4.71c 1.84 3.34d 2.33 4.46c 1.96 3.17d
2.53
A person steals a purse from a blindperson A 5.08b 1.24 5.52a
0.82 4.82c 1.57 4.12d 1.94 2.02e 2.23 1.88e 2.35
A cleaning person, who thinks no oneis watching, sits in the
chair of thecompany president C 0.47b 1.01 0.48b 0.97 0.57b 1.14
1.05a 1.57 0.54b 1.21 0.23c 0.69
A salesman addresses someone byhis/her first name after just
meetinghim/her C 0.96a 1.42 1.05a 1.35 1.11a 1.33 0.39b 0.81 0.39b
0.88 0.48b 1.10
An oversensitive employee directlycriticizes his boss C 1.54b
1.51 1.54b 1.48 1.93a 1.56 1.09c 1.39 1.10c 1.65 0.47d 1.04
A teenager begins to eat dinnerbefore everyone else at the table
isserved C 1.25c 1.34
1.55a,
b 1.471.37b,
c 1.50 0.88d 1.30 0.44e 0.92 0.40e 0.87An 8-year-old student
speaks to his
teacher in the same way as he talksto his friends C 1.46b 1.54
1.69a 1.69 1.32b 1.47 1.43b 1.57 0.73c 1.23 0.49d 0.97
An employee unjustifiably complainsto his/her boss C 2.24a 1.63
1.92b 1.69 2.31a 1.69 0.98c 1.36 0.76d 1.31 0.60d 1.22
A person burns the American flag C 2.81a 2.02 2.75a 2.05 2.47b
2.06 2.64a,b 2.04 1.62c 1.72 0.66d 1.36Someone who regularly leaves
work
an hour early when no one isaround C 2.81b 1.70 3.07a 1.66
3.01a,
b 1.77 0.95c 1.31 0.46d 0.95 0.62d 1.18A 10-year-old says dirty
words to his
parents C2.32b,
c 1.86 3.16a 1.87 2.11c 1.82 2.54b 1.93 1.01d 1.50 0.75e
1.26Someone doesnt go to his own
mothers funeral C 2.83c 2.11 3.71b 2.03 2.88c 2.11 4.66a 1.64
0.97d 1.59 0.94d 1.68A company executive refuses to sit
next to a laborer on a train C 4.27b 1.51 4.58a 1.47 4.16b 1.66
3.48c 1.83 0.91e 1.40 1.53d 2.07A 16-year-old refuses to give up
his/
her seat on the bus to a crippledold lady C 4.25b 1.50 4.90a
1.18 4.02b 1.74 3.41c 1.88 0.90e 1.37 1.24d 2.02
A person has an incestuousrelationship D 2.79c 2.06 4.77a 1.54
3.07b 2.21 2.71c 2.26 1.48d 1.86 4.82a 1.71
A 70-year-old male has sex with a17-year-old girl D 3.17d 2.13
4.85b 1.60 3.45c 2.16 2.96d 2.26 1.56e 1.94 5.23a 1.41
Note. Emotions within a row that share a superscript do not
differ significantly at p .05 for that item. Significant
differences were determined on thebasis of the results of paired t
tests (N 151). The highest emotion intensity rating for each item
is shown in boldface. A autonomy violation; C community violation;
D divinity violation.
722 HUTCHERSON AND GROSS
-
interchangeably. Although participants almost exclusively
de-scribe moral offenses when prompted with the word angry
andobjects related to disease or decay for grossed out, the
worddisgusted without further specification elicits descriptions
ofboth nonsocial and moral events (Nabi, 2002). Furthermore,
whenthese disgusting events are moral in nature, they evoke
anger-like action tendencies, such as the desire to lash out, to
get back atsomeone, and to overcome some obstacle. One implication
of thisresearch is the importance of precise wording. Failure to
specify toa lay audience whether the term disgust refers to social
ornonsocial feelings may result in spurious conclusions.
Similarly,failure to measure anger casts some doubt on research
that claimsto find specific forms of disgust related to social and
nonsocialviolations (Tybur et al., 2009).
Contempt and disgust are even more often conflated with
eachother in the literature (e.g., Harris & Fiske, 2006; Mackie
et al.,2000). In some cases, they seem to be used simply to
describediffering intensities of the same underlying negative
social emo-tional state, much as happy and ecstatic are used.
Evidencelinking contempt to a unique facial expression, separate
fromdisgust and anger (e.g., Ekman & Friesen, 1986; Matsumoto
&Ekman, 2004), has been questioned on a number of counts.
Thecontempt expression, a unilateral lip curl, is frequently
confusedwith the expressions for disgust and anger (e.g., Shioiri,
Someya,Helmeste, & Tang, 1999), or, more disturbingly, given a
label ofneutral by a ratio of nearly 2:1 when the option is
available(Wagner, 2000). It is worth noting that these same studies
alsoobserved a fair degree of confusion between anger and
disgustexpressions. For example, approximately 30% of participants
la-beled the prototypical anger expression as disgust.
Similarly,nearly 40% of participants labeled the disgust facial
expressionwith either anger or contempt (Shioiri et al., 1999).
Research on action tendencies also suggests less
differentiationbetween contempt and disgust. Expressions of either
contempt ordisgust during episodes of marital conflict predict
later maritaldissolution (Gottman, 1994). Experiences of contempt
and disgustalso both predict tendency to withdraw from rather than
confrontan antagonistic social group (Mackie et al., 2000), and
both may beassociated with prejudice toward the most stigmatized,
dehuman-ized minorities, such as the homeless or drug addicts
(Fiske et al.,2002; Harris & Fiske, 2006; Hodson &
Costello, 2007). The failureto demonstrate either a unique facial
expression or set of actiontendencies for contempt separate from
disgust or anger (Rosemanet al., 1994) suggests that these terms
may define a single under-lying state.
The Present Research
It is not clear how to reconcile these conflicting findings.
Whydo some studies find evidence for distinctions among the
emotionsbut others find little, if any? How do these observations
fit with thefunctionalist account outlined above? One way of posing
thesequestions is depicted in Figure 1, which suggests a
continuumanchored by two extreme viewpoints, the first that views
anger,moral disgust, and contempt as semantically equivalent terms
fora single underlying emotional state (e.g., moral indignation),
andthe second that views these emotions as three completely
separate,independent states. Our goal was to determine where on
thiscontinuum it makes sense to draw this line.
A comparison of the predictions of our
socialfunctionalistaccount, which argues that the functions of
these emotions extendbeyond a strict concern with morality, to the
more strictly moralfunctionalist CAD triad hypothesis is
instructive (see Figure 2).Both views are concerned with the
adaptive functions served bythese three emotions but differ
regarding each emotions function.Both accounts link contempt to
status and hierarchy, although weargue that contempt may be
associated with status through itsfunction in judging nonmoral,
incompetent behavior. Similarly,both theories implicate anger in
response to threats against rightsand property, although our
perspective makes the further specifi-cation that the threat be
directly and immediately relevant to theindividual. The two differ
most markedly with respect to theirtreatment of disgust. Whereas
the CAD triad hypothesis linksdisgust specifically to offenses
against the sacred and the pure, weattribute a more general role to
moral disgust. Immoral behavior,no matter whether it involves
offenses against autonomy, commu-nal obligations, bodily purity, or
even other ethics (Graham, Haidt,& Nosek, 2009), warrants
avoidance. Two actions that indicate asimilar level of immoral or
socially exploitative behavior shouldelicit the same amount of
moral disgust, regardless of whether theyrepresent different kinds
of ethical violations, and two actions thatdiffer in their
implications for moral character should elicit differ-ent levels of
moral disgust, even if they represent the same moralethic. This
distinction between our account and the CAD triadaccount in part
parallels the distinction between the functions ofnonsocial and
moral disgust. Many exemplars of divinity viola-tions used by Rozin
et al. (1999; such as biting into an apple witha worm in it or
eating rotten meat) appear to more strongly tapnonsocial
disgust.
Study 1: The Role of Moral Disgust in MoralViolations
Because a lumpers perspective (Panel A in Figure 1), ourextended
socialfunctionalist account, and a different,
primarilymoralfunctionalist account (the CAD triad) all make
competingpredictions regarding disgust, we begin with this emotion.
Usingquestionnaire-based methods, our goal in this study was to
addresswhether there is a specifically moral version of disgust,
responsiveto purely social offenses and differentiable from feeling
grossedout (Nabi, 2002). On the basis of our socialfunctionalist
per-spective, we hypothesized that when the distinction between
socialand nonsocial disgust was made clearer to participants,
moraldisgust would be associated with all forms of moral
offenses.
Method
Participants. One hundred and fifty-one students (meanage 18.8
years, range 1725; 62% female and 38% male; 58%Caucasian, 22% Asian
or Asian American, 6% Hispanic or Latino,3% Black, 11% other or
declined to answer) enrolled in an intro-ductory psychology class
participated in exchange for coursecredit.
Materials and procedure. In large groups, participants
com-pleted a questionnaire consisting of a subset of the moral
violationsused in Rozin et al. (1999). These included all 12
actions catego-rized as community violations (Items 4, 8, 10, 11,
13, 14, 19, 27,28, 33, 34, and 38 from Rozin et al., 1999), all
nine actions
723SOCIALFUNCTIONALIST ACCOUNT OF MORAL EMOTIONS
-
categorized as autonomy violations (Items 12, 16, 25, 35, 36,
40,41, 43, and 44), and two actions categorized as divinity
violations(Items 3 and 22). Because we were concerned primarily
with moralbehavior, we selected the only two divinity violations
that repre-sented obviously intentional sociomoral violations. The
three itemsnot selected included (a) a person biting into an apple
with a wormin it, (b) a person eating rotten meat, and (c) a person
touching acorpse. These were deemed to be less clearly moral in
nature andambiguous with respect to moral intention.
All items were worded to describe a person committing
aparticular offense (e.g., Someone embezzles from a bank).
In-structions for the questionnaire were as follows:
Often, when we learn about an immoral event (something bad,
wrong,or evil) we experience strong feelings. We might feel a range
ofemotions, including anger/rage, moral disgust/revulsion,
contempt, orother emotions. We are interested in the reactions you
might have tothe following situations. There are no right or wrong
answers. Pleasejust indicate how you would feel in each
situation.
For each situation described below, indicate how much of each of
theemotions listed you feel at the actions/events described, on a
scalefrom 0 (not at all) to 6 (extremely). Please also circle the
space of theemotion that best describes your overall reaction to
the behavior.
We assessed responses using six different emotion terms
(anger,contempt, moral disgust, sadness, fear/anxiety, and grossed
out).Of these, anger and contempt are identical to the terms used
inRozin et al. (1999). Disgust, for reasons mentioned above,
wasbroken into two terms, moral disgust and grossed out, to
bettercapture the distinction between social disgust and nonsocial
dis-gust, respectively. Finally, we included the terms sadness
andfear/anxiety to assess the specificity of the emotional response
tomoral violations.
Before the target questions, participants were provided with
anexample, illustrating both the emotional intensity ratings and
thebest descriptive rating. Text of the example read
****Exam-ple**** A man is caught selling drugs to minors.
****Exam-ple**** and was followed by numbers under the column for
eachemotion, to indicate possible responses. The numbers for
angerand moral disgust were the highest of the six emotions and
equalto each other. A circle around the number in the moral
disgustcolumn indicated to participants both how to indicate the
bestdescriptive emotion and that they should circle a single
emotion asbest descriptive even in cases of a tie in intensity.
Results
For each item, we conducted paired t tests on emotional
inten-sity ratings for all 15 combinations of emotion pairs,
reportingsignificant differences below a threshold of p .003,
theBonferroni-corrected alpha level for 15 comparisons (.05/15). As
asecondary analysis, we analyzed the frequency with which
eachemotion was selected as the best descriptive term, using an
ap-proach similar to that employed by Rozin et al. (1999).
Althoughthere was a main effect of gender on emotion intensity,
indicatingthat women (M 2.92, SD 1.03) responded more strongly
thanmen (M 2.58, SD 1.23), F(1, 127) 5.53, p .02, p
2 .04,there were no interactions between gender and the main
effects ofinterest. We therefore report the results below
collapsing across
gender. Results are presented by moral code. Each action and
thecorresponding emotion intensities appear in Table 1.
Across the community violations (n 12), moral disgust re-ceived
higher intensity ratings than any other emotion (M 2.52,SD 1.02),
all ts(150) 3.96, all ps .001, all ds 0.26.Participants reported
greater levels of moral disgust than anger(M 2.26, SD 1.02),
contempt (M 2.24, SD 1.07), sadness(M 1.96, SD 0.96), fear (M 0.82,
SD 0.89), or grossedout (M 0.70, SD 0.90). Frequency analysis of
the bestdescriptive term painted a similar picture. Averaging over
all of thecommunity violations, moral disgust was chosen as the
best de-scriptive term (M 31%), followed by contempt (M
23%),sadness (M 22%), anger (M 17%), fear (M 6%), andgrossed out (M
2%).
Across the autonomy violations (n 9), moral disgust (M 4.33, SD
0.82) was rated significantly higher than any otheremotion, all
ts(148) 4.98, all ps .001, all ds 0.29, includinganger (M 4.07, SD
0.97), contempt (M 3.74, SD 1.24),sadness (M 2.74, SD 1.29), fear
(M 2.05, SD 1.29), andgrossed out (M 1.74, SD 1.52). Averaging over
all of theautonomy violations, moral disgust was chosen as the best
descrip-tive term (M 42%), followed by anger (M 22%), contempt(M
17%), sadness (M 9%), fear (M 7%), and grossed out(M 3%).
Across the divinity violations, grossed out was rated highest(M
5.03, SD 1.39) but was not significantly different frommoral
disgust (M 4.81, SD 4.8), t(146) 2.04, p .60,corrected for multiple
comparisons. Both emotions were ratedsignificantly higher than any
other emotion, all ts(146) 10.15, allps .001, all ds 0.9, including
contempt (M 3.26, SD 1.98), anger (M 2.96, SD 1.89), sadness (M
2.84, SD 1.98), and fear (M 1.52, SD 1.67). Averaging over both of
thedivinity violations, grossed out was chosen 58% of the
time,followed by moral disgust (M 31%), contempt (M 5%),sadness (M
4%), fear (M 3%), and anger (M 2%).
Discussion
Contrary to the CAD triad model (in which moral disgustresponds
specifically to violations involving sanctity and purity),our
account suggests that moral disgust should be an adaptiveresponse
to moral violations more generally. The results of Study1 supported
this hypothesis. When moral disgust was linguisti-cally
differentiated from grossed out, it emerged strongly in
bothautonomy and community violations and rivaled grossed out in
thedivinity violations. This relationship held whether we
measuredemotion using a single rating of the best descriptive
emotion (asRozin et al., 1999, did) or using a more detailed rating
of theintensities of a range of emotions. Both methods also
distinguishedmoral disgust from nonsocial disgust. Although moral
disgust wasuniformly high across all three codes, the term grossed
out wasendorsed strongly only in divinity violations.
The finding that disgust was best descriptive of all three types
ofmoral violations is consistent with our view. However, the
patternacross the second-most intense emotions followed the
predictionsmade by the CAD triad hypothesis (Rozin et al., 1999),
identifyingcontempt, anger, and moral disgust in community,
autonomy, anddivinity violations, respectively. Moreover, although
the data dosuggest that moral disgust and grossed out are different
emotional
724 HUTCHERSON AND GROSS
-
entities, they do not yet rule out an account in which moral
disgustis simply used to express the sum of other negative
emotions,including anger and contempt. Studies 2 and 3 identified
specificappraisals that differentiate anger, disgust, and
contempt.
Study 2: Differentiating Anger and Moral Disgust
As outlined above, we hypothesized that in cases in which oneis
directly endangered by an immoral act, a more proactive re-sponse
than disgust, such as anger, may be warranted. In caseswhere one is
not directly harmed by immoral behavior, the low-cost avoidance
associated with disgust may be more adaptive.Alternatively, if
anger responds merely to whether an action is aviolation of the
ethic of autonomy, it may depend less on who thevictim is and only
on the nature of the crime. Recent evidenceshowing that
participants report greater anger when a moral of-fense is
self-relevant than when it is not (Batson, Chao, & Givens,2009;
Batson et al., 2007) is consistent with this hypothesis.However,
other moral emotions, such as disgust and contempt,were not
assessed in these studies. It therefore remains possiblethat,
contrary to our predictions, self-relevance is not an
appraisalunique to anger but is rather a general intensifier of
morallyrelevant emotions. In order to test these possibilities, we
presentedvignettes of moral violations, systematically varying the
degree ofself-relevance of the immoral act, and assessed response
acrossseveral different emotions.
Method
Participants. One hundred and thirty-one students (age range1731
years; 63% female and 37% male; 51% Caucasian, 25%Asian or Asian
American, 9% Black, 6% Hispanic or Latino, 9%other) enrolled in an
introductory psychology class at StanfordUniversity completed one
of three alternate versions of the ques-tionnaire. Forty-five
students completed Version 1, 45 completedVersion 2, and 41
completed Version 3.
Materials and procedure. Nine new moral violations werecreated
for the questionnaire used in Study 2 (see Appendix A fora full
list of items). Of these, three control violations appeared
inconsistent position and wording (always presenting the victim as
ageneric other) in each questionnaire, to ensure that when
violationswere equivalently presented, the three groups did not
differ. Theother six violations varied in the degree of
self-relevance acrossthe three versions of the questionnaire. In
Version 1, violationswere worded such that the participant was
presented as the victim;in Version 2, the victim was a friend of
the participant; and inVersion 3, the victim was a generic other.
For example, in the itemA student steals your exam and copies it
(emphasis added), theitalicized word would be replaced either with
your friend or astudent in the two alternate versions. As in Study
1, participantsindicated the intensity of six different emotions
(anger, moraldisgust, contempt, sadness, fear/anxiety, and grossed
out). Theyalso circled the emotion term best descriptive of their
overallexperience.
Results
In order to determine whether there was a difference betweenthe
control (identical perspective) and test (varied perspective)
items, an average was created across the three control
questionsand across the six test questions for each emotion for
each partic-ipant. We hypothesized that this average would reveal
higherexperiences of anger when the self was implicated than
whenunrelated individuals were harmed by an immoral offense. On
theother hand, moral disgust, responding to the perpetrators
moralcharacter, should remain relatively constant across the three
per-spectives.
As in Study 1, we observed a significantly stronger
emotionalresponse in women (M 2.73, SD 0.72) than in men (M 2.36,SD
0.67), F(1, 112) 7.66, p .007, p
2 .06; however,gender did not interact with any of the effects
reported below andwas thus dropped as a factor. A 3 2
repeated-measures analysisof variance (ANOVA) was conducted, with
version (self, friend,other) as a between-subjects factor and
question type (control vs.test) as a repeated, within-subject
measure. For all emotions exceptfear, there were significant
effects of question type: The controlviolations were consistently
rated as slightly more intense, allFs(1, 128) 9.61, all ps .002,
all p
2s .07. These effects werequalified by a significant interaction
between version and questiontype for anger, contempt, sadness, and
fear, all Fs(2, 128) 4.87,all ps .01, all p
2s .07, as well as a marginally significantinteraction for moral
disgust, F(2, 128) 2.5, p .09, p
2 .04.Follow-up between-groups t tests indicated that, for
control
questions, there was no significant effect of version for any
emo-tion. Even though these items were regarded as more
negative,they were responded to in an equivalent manner across all
groups.However, as predicted, items in which the self-relevance of
thevictim was manipulated showed strikingly different
emotionalprofiles across versions. There was a significant linear
effect ofversion for anger, t(128) 4.12, p .001, d 0.86, and for
fear,t(128) 3.98, p .001, d 0.85. For moral disgust, this effectwas
not significant, t(128) 1.45, p .15. Although fear wasconsiderably
lower overall than anger, both were highest in the selfcondition
(MAnger 4.24, SD 0.93; MFear 1.43, SD 0.98),lower in the friend
condition (MAnger 3.62, SD 0.88; MFear 0.71, SD 0.72), and lowest
in the other condition (MAnger 3.41, SD 1.01; MFear 0.62, SD 0.76).
Moral disgust showedthe opposite pattern, being higher in the other
condition (M 3.71,SD 1.05) than in the self condition (M 3.38, SD
0.96).Contempt and sadness did not show a clear pattern, both
beinglower in the friend condition than in either of the other
conditions.None of the emotions showed a main effect for version,
suggestingthat the groups did not differ in a general way in their
emotionalresponding.
Analysis of the frequency with which each emotion label
wasselected as best descriptive revealed a similar pattern of
findings.In control (identical perspective) items, moral disgust
was selectedas the best descriptive term in a similar manner across
all threegroups. For the test (perspective varied) items, the
percentage ofpeople selecting anger as the best descriptive term
was highest inthe self condition (42%), intermediate in the friend
condition(40%), and lowest in the other condition (27%). Moral
disgustshowed the opposite pattern of results. The percentage of
partici-pants selecting this emotion as best descriptive was lowest
in theself condition (20%), intermediate (but lower than anger) in
thefriend condition (27%), and highest in the other condition
(34%;see Figure 3). All other emotions were selected by an
equivalentpercentage of participants across the three groups.
725SOCIALFUNCTIONALIST ACCOUNT OF MORAL EMOTIONS
-
Discussion
The results of Study 2 strongly confirmed the predicted
patternof results. Even though the types of moral violations were
heldconstant across all three versions of the questionnaire,
varying therecipient of harm had a substantial effect on the
emotions felt.Consistent with the prediction that anger and its
associated actiontendencies might be most adaptive when the harm
directly impli-cates the self, anger was highest for the self
condition, lower in thefriend condition, and lowest in the other
condition. Additionally, asendorsement of anger decreased across
the three conditions, en-dorsement frequency of moral disgust
increased. In the intensityresults, moral disgust remained roughly
constant across the threeconditions, trending up slightly. This
pattern is consistent withmoral disgust being concerned with the
intentions or maliciousnessof the perpetrator, whereas anger is
concerned with attack againstdirect threat. It clearly
differentiates anger from moral disgust anddemonstrates that people
use these terms in different ways, con-trary to the conclusions
drawn by some researchers (Nabi, 2002;Simpson et al., 2006).
However, this study does not clearly allow us to
differentiatemoral disgust and contempt. Endorsement of each
emotion as bestdescriptive revealed a different trend in moral
disgust and con-tempt across the three versions of the
questionnaire. Intensity datashow that moral disgust and contempt
were both relatively flatacross conditions, although moral disgust
was consistently higherthan contempt. These data are thus
consistent with a model inwhich contempt is simply a weaker version
of disgust, consistentwith some models presented in the prejudice
literature, which oftenlump together contempt and disgust (e.g.,
Fiske et al., 2002;Mackie et al., 2000).
Study 3: Differentiating Contempt and Moral Disgust
Although contempt and moral disgust have often proven diffi-cult
to separate, our extended socialfunctionalist perspective sug-gests
a possible distinction between them. Whereas disgust mayconcern the
likelihood that an individual harbors benevolent ormalicious
intentions, contempt may concern an individuals com-petence to
achieve their aims. We thus predicted that actions thatsignal
immorality would be associated with a greater balance ofmoral
disgust relative to contempt, whereas actions that
signalincompetence would be associated with a greater balance of
con-tempt relative to moral disgust. Such a pattern would clearly
showthat moral disgust and contempt are not simply terms
associatedwith different levels of the same emotion but respond to
differentsocially relevant signals. If, on the other hand, moral
disgust andcontempt are only synonyms for a general motivation to
withdrawfrom morally offensive individuals, there should be no
situations inwhich the emotions differ.
Method
Participants. One hundred and six undergraduate students(age
range 1729 years; 71% female and 29% male; 43% Cauca-sian, 24%
Asian or Asian American, 9% Black, 9% Hispanic orLatino, 15% other
or declined to answer) enrolled in an introduc-tory psychology
class at Stanford University completed one of twoquestionnaires
(Version 1, n 56; Version 2, n 50).
Materials and procedure. The questionnaires used wereidentical
in format to those used in Studies 1 and 2. To assessresponses to
moral violations, we included several items used byRozin et al.
(1999) that were representative of the ethic of com-munity (e.g., A
teenager begins to eat dinner at the table beforeeveryone else is
served). In order to assess whether contempt and
Figure 3. Selection of the best descriptive emotion in items
with varied and common perspective (Study 2).Changes in endorsement
of anger, contempt, and moral disgust are shown across three groups
of participantsgiven moral violations that were similar except for
the nature of the victim (left panel) or that were identical(right
panel).
726 HUTCHERSON AND GROSS
-
moral disgust could be differentiated by focusing on
incompe-tence, we created several examples of incompetent behaviors
(e.g.,A person who tries to use large words to sound smart,
butmispronounces them). Finally, in order to assess replicability
offindings, we created two versions of the questionnaire, each
con-sisting of several community violations and competence
failures.There was no overlap of items between the two
questionnaires (seeAppendix B for a full list of items).
Participants were givenidentical instructions to those used in
Studies 1 and 2 and wereprovided with an identical example.
Responses were assessedusing seven different emotion terms (anger,
contempt, moral dis-gust, sadness, pity, grossed out, and
amusement).
Results
Separate averages across the morality and competence itemswere
created for each questionnaire version, for each emotion, andfor
each participant. In initial analyses, questionnaire version
andgender were included as factors. There were no effects
involvingversion, so results were collapsed across the two
versions. Al-though there was only a marginally significant main
effect ofgender on overall intensity of response, F(1, 90) 2.9, p
.09,p
2 .03, there was a significant interaction between sex
ofrespondent and action type in the overall intensity of
emotionalresponse, F(1, 90) 4.48, p .04, p
2 .05. Post hoc analysesindicated that although moral violations
elicited greater emotionthan competence violations for both men
(MMoral 1.58, SD 0.82; MCompetence 1.37, SD 0.66), t(27) 2.18, p
.04, d 0.28, and women (MMoral 1.97, SD 0.64; MCompetence 1.52,SD
0.75), t(64) 6.98, p .001, d 0.65, this difference wasstronger in
women, t(91) 2.16, p .03, d 0.49. However,type of emotion did not
interact with this effect, and gender wasthus dropped as a factor
in analyses reported here. Using theemotion intensity averages, we
conducted a 2 (action type) 7(emotion type) repeated-measures
ANOVA. The main effects ofaction and emotion type were significant,
as was their interaction(all Fs 53.47, all ps .001, all p
2s .353). Follow-uppaired-sample t tests, Bonferroni corrected
for multiple compari-sons, indicated that, in both studies, with
the exception of pity,grossed out, and amusement, moral violations
elicited more in-tense emotion than did competence violations, all
ts(99) 7.33, allps .001, all ds 0.78. Pity and amusement were
consistentlyrated higher in failures of competence than in
violations of mo-rality, both ts(99) 6.9, ps .001, ds 0.74, and
there was nosignificant difference for the term grossed out (see
Table 2 formeans and standard deviations).
Analysis of the three emotions concerned with negative
socialjudgment indicated that moral disgust was significantly
higherthan any other emotion in the community violations, all
ts(100) 5.9, all ps .001, all ds 0.35, followed by anger and
thencontempt. In contrast, participants reported a greater amount
ofcontempt than moral disgust toward incompetent acts, t(102) 6.47,
p .001, d 0.62 (see Figure 4). Moreover, communityviolations
elicited marginally significantly higher amounts of an-ger than
contempt, t(105) 2.4, p .02 uncorrected, d 0.19,whereas incompetent
behaviors elicited marginally higher amountsof contempt than anger,
t(105) 2.5, p .01 uncorrected, d 0.24.
Analysis of percentages of participants selecting particular
emo-tions as best descriptive of their overall experience confirmed
theresults of the emotion intensity ratings (see Table 3). For
violationsof community, the term moral disgust was selected most
fre-quently (39%). In comparison, contempt was selected
relativelyless frequently (9%) and, in both cases, was selected
less oftenthan anger. In contrast, incompetent actions were
associated pri-marily with amusement and pity, whereas a direct
comparison ofanger, moral disgust, and contempt revealed that, of
these three,contempt was selected more often than either of the
other twoemotions (contempt 12%, anger 8%, moral disgust 1%).
Discussion
The primary goal of Study 3 was to test whether
specificappraisals distinguished moral disgust from contempt. We
hypoth-esized that two dimensions relevant to the other-judgmental
emo-tions of moral disgust and contempt might be morality and
com-petence, respectively. Results were consistent with this
hypothesis.Restricting the focus to the three moral judgmental
emotions alone,moral disgust was clearly the emotion most strongly
felt in re-sponse to violations of the ethic of community, assessed
eitherwith intensity reports or by selection of the best
descriptive term,as in Study 1. By contrast, for incompetent
actions, contempt wasrated most highly, measured either using
intensity or the bestdescriptive term.
At the same time, however, several other patterns in the data
areworth noting. First, other emotions, such as pity and
amusement,emerged as more highly activated in incompetent acts than
any ofthe three sociomoral emotions, including contempt. Second,
al-though comparing the balance of emotions within either
commu-nity violations or incompetent acts leads to a conclusion
thatdisgust is associated with immorality, and contempt with
incom-petence, comparing between these two classes of actions
paints a
Table 2Emotion Intensity Ratings of Community Violations and
Incompetent Actions (Study 3)
Action type
Anger Moral disgust Contempt Sadness Pity Grossed out Amused
M SD M SD M SD M SD M SD M SD M SD
Community violation 2.70b 1.23 3.11a 1.11 2.46b 1.24 1.74c 0.94
1.36d 0.96 0.58e 0.83 0.86e 0.76Incompetent action 1.18c,d 0.98
0.76d 0.90 1.45c 1.31 1.02d 0.98 2.18b 1.24 0.41e 0.76 2.98a
1.09
Note. Emotions within a row that share a superscript do not
differ significantly at p .05 for that type. Significant
differences were determined on thebasis of the results of paired t
tests (N 106).
727SOCIALFUNCTIONALIST ACCOUNT OF MORAL EMOTIONS
-
slightly different picture. Both emotions are higher in
immoralthan incompetent actions.
We believe that these findings are also consistent with
oursocialfunctionalist approach. First, our account does not
prohibitthe coactivation of emotions. To the extent that pity or
amusementshares an appraisal with contempt, one might indeed expect
themto co-occur. Second, it is also important to note that
competencemay refer not just to cognitive domains but also to
social ones.Knowing how to respond to another person at the
appropriate time,or in the appropriate mannerhow to entertain,
appease, flatter, orcritique for maximum social benefitmay be just
as important askill as knowing how to solve difficult mathematical
problems.Indeed, given evidence that people weight social
information morehighly than competence information in evaluating
people (Woj-ciszke, 2005), it may be less surprising to find that
contempt ishigher in community violations, which may demonstrate as
mucha lack of knowledge or facility with social conventions as
amalevolent intention.2
Study 4: Perceived Consequences Associated With theMoral
Emotions
In Studies 13, we focused on antecedent appraisals
distinguish-ing anger, disgust, and contempt; however, our extended
socialfunctionalist account also makes specific predictions about
thelikely consequences of these emotions. If people have
observedeach of these emotions to have different social
consequences, theyshould distinguish among them and have clear
preferences regard-ing being the target of one versus another of
these emotions.
In Study 4, we tested this prediction by examining beliefs
aboutconsequences for someone evoking these three emotions.
Moraldisgust, to the extent that its properties are similar in
consequenceto nonsocial disgust, may be associated with a
long-lasting judg-ment of the inherent moral worth of a target that
is difficult toundo, and avoidance or ostracism and a loss of
regard may resultfrom a person being labeled morally disgusting.
Given evidencethat social rejection can be extremely painful
(Eisenberger, Lieber-man, & Williams, 2003), disgust may in
some sense be the worstemotion one could inspire in others.
Contempt, as another emotionconcerned with judgments about a
persons character, may also beundesirable, but perhaps less severe
because it is associated with
incompetence, a category weighted less heavily in overall
impres-sions (Wojciszke, 2005). Finally, although its aggressive
actiontendencies would seem to make it the most dangerous
emotion,anger might actually be the least negative, because it may
befocused on temporary behavior rather than lasting judgments.
Note that these predictions differ from the predictions of
eitherthe CAD triad view or the undifferentiated view. The CAD
triadview argues that different individuals and cultures may
emphasizedifferent ethics over others. In modern Western cultures,
divinityand community ethics tend to be emphasized less in
morality, andautonomy tends to be emphasized more (Haidt, Koller,
& Dias,1993; Rozin et al., 1999). Thus, a prediction of the CAD
triadhypothesis might be that, at least in Western cultures where
au-tonomy is prioritized, anger should be the worst emotion,
whereascontempt and disgust might be comparatively less bad.
Alterna-tively, the CAD triad might predict that, given that all
threeemotions are about morality (albeit different kinds), there
shouldbe no differences between them. The semantic equivalents
viewlikewise predicts no difference in the desirability of these
emo-tions. To determine which account best fits the pattern of
perceivedconsequences, we asked participants to make judgments
aboutwhich emotion they would least like to inspire and to justify
theirresponse.
Method
Participants. Forty-six participants (age range 1823 years;70%
female and 30% male; 41% Caucasian, 31% Asian or AsianAmerican, 4%
Black, 15% Hispanic or Latino, 9% other) were
2 This distinction may also help to place the association
between com-munity violations and contempt in both Rozin et al.s
(1999) work and inour Study 1 (secondary to moral disgust) in the
proper light. Communityviolations may be seen less as signs of
immorality and more as signs ofsocial incompetence. Indeed, we have
shown that the community violationsthat were more strongly
associated with moral disgust in Study 1 are alsojudged as being
more strongly indicative of immorality than incompetence,whereas
the community violations associated with contempt in Study 1
areperceived as more indicative of stupidity or a lack of
competence than ofimmorality (Hutcherson, 2007).
Figure 4. Intensity of emotions in items representing community
violations and incompetent actions (Study 3).Self-reported
intensities of anger, moral disgust, and contempt are shown in the
context of violations of the ethicof community (left panel) or of
failures of competence (right panel). Error bars represent the
standard error ofthe mean.
728 HUTCHERSON AND GROSS
-
recruited from the Stanford University campus and
surroundingcommunity.
Materials and procedure. Participants were given
informedconsent, filled out a short demographic questionnaire, and
thencompleted a series of questionnaires presented on a computer,
inwhich participants preferences about being the target of
severaldifferent negative emotions, including moral disgust, anger,
con-tempt, grossed out, fear, and sadness, were assessed.
Participantsmade a series of choices comparing these negative
emotions pair-wise and indicated for each pair which emotion they
would rathersomeone to feel toward them (e.g., Would you rather
someone bemorally disgusted with you or angry with you?).
Participantsmade choices for six pairs: moral disgust versus anger,
moraldisgust versus sadness, moral disgust versus contempt,
moraldisgust versus grossed out, contempt versus anger, and
sadnessversus anger. After making their choice, they were asked to
ex-plain why they chose their answer. These answers, with all
refer-ences to emotion words blanked out, were coded by two
indepen-dent raters, blind to participants choices, for their
references tocertain categories of response. The following nine
themes (andsample illustrative responses) were used to code
participantsjustifications: whether an emotion (a) was temporary or
could befixed (e.g., People get over anger easier. Time usually
takes careof it.); (b) was indicative of ones moral sense or
character (e.g.,Anger is less deep than disgustdisgust would mean
they wereupset with my ideals and my personality); (c) was
responding toa single, relatively localized action (e.g., Anger
usually involvesa single action. For example, I am angry at you for
breaking myiPod.); (d) could be a result of a misunderstanding or
idiosyn-cratic response (e.g., Many times anger is unjustified and
easilysolved); (e) would make the participant feel bad about him-
orherself (e.g., Having someone grossed out would be
embarrassingtoward myself); (f) would make the participant feel
sympathy forthe pain of another person (e.g., The other person
would feel hurt,and I never try to hurt anyone); (g) indicated
physical rather thanmoral elicitors (I could gross out someone
simply by picking upa worm; for someone to be morally disgusted
with me, Id have todo something shameful); (h) was less intense or
less meaningful(e.g., Lesser of the two); or (i) was indicative of
a superiority orstatus judgment (e.g., Because contemptuous
indicates air ofsuperiority). Justifications often invoked multiple
themes andwere given a score of 1 or 0 for each theme.
Justifications noteasily classified into one of these categories
were given a score forother. Reliability between the two coders
overall was substantial(Cohens .66), and reliability for individual
categories rangedfrom .47 (feeling bad about the self) to .93
(physical elicitors).Disagreements between the two coders were
resolved by a thirdcoder.
Results
Emotion preferences. Participants preferences strongly
sup-ported our socialfunctionalist account. Compared with
moraldisgust, participants by a wide margin preferred to be the
recipientof someones anger (43 vs. 3), 2(1, N 46) 34.78, p
.001;contempt (38 vs. 8), 2(1, N 46) 19.56, p .001; sadness (31vs.
15), 2(1, N 46) 5.6, p .02; and basic disgust (40 vs. 6),2(1, N 46)
25.81, p .001. Participants also stronglypreferred to be the target
of someones anger than contempt (40 vs.6), 2(1, N 46) 25.13, p
.001, and to anger rather thansadden someone (38 vs. 8), 2(1, N 46)
19.56, p .001.
Justifications for emotion preferences.Anger versus moral
disgust. Among the majority of partici-
pants preferring to be the target of anger versus moral disgust,
58%argued that anger was of shorter duration or more easily
reversed.Sixty percent further specified that anger was less
indicative that apersons inherent moral character was lacking or
that anger couldbe responsive to a single action (21%). A sense of
self-shame orsadness was also cited as justification by 12% of
participants. Thethree participants who indicated they would rather
induce moraldisgust generally gave idiosyncratic responses not
easily catego-rized (e.g., [I prefer moral disgust] because I
always have thisfeeling that I am not like other people and am
morally different inmy actions).
Contempt versus moral disgust. A similar set of
justificationswas provided by participants who preferred to be the
target ofcontempt to moral disgust: 30% argued that moral disgust
wouldlast longer or be more difficult to remedy; 35% thought that
moraldisgust was more telling of a persons character; and 14% said
thatcontempt could simply be the result of idiosyncratic values on
thepart of someone experiencing it. Sixteen percent of
participantsargued that contempt was simply less intense. Of the
eight partic-ipants preferring to elicit moral disgust rather than
contempt, 42%gave idiosyncratic responses. Twenty-two percent
argued that con-tempt rather than moral disgust was more difficult
to undo. Noneargued that contempt would be more indicative of ones
character.
Sadness versus moral disgust. Of the participants indicatingthat
moral disgust was a worse emotion, 29% justified their
choicebecause disgust was a stronger judgment of their moral
character,was less reversible (29%), was more intense or meaningful
(20%),or would make them more ashamed of themselves (13%).
Amongparticipants selecting sadness as the worse emotion to
inspire, mostindicated that they did not want to be the source of
anotherpersons pain (40%), whereas smaller numbers indicated that
sad-ness would last longer (20%).
Grossed out versus moral disgust. Of the majority of
partic-ipants preferring someone to feel grossed out, most
thought
Table 3Percentage of Subjects Selecting an Emotion as Best
Descriptive for Community Violations and Incompetent Actions (Study
3)
Action type Anger Moral disgust Contempt Sadness Pity Grossed
out Amused
Community violation 16.25 39.08 13.57 9.76 7.68 0.21
13.46Incompetent action 7.98 1.38 11.79 3.30 28.86 1.33 47.32
Note. N 106.
729SOCIALFUNCTIONALIST ACCOUNT OF MORAL EMOTIONS
-
grossed out was less indicative of ones moral character
(18%),would be a less severe or meaningful emotion (30%), would be
ofshorter duration (25%), or could be felt to purely physical
breachesof etiquette (23%). Of the minority who preferred someone
feelmorally disgusted, two cited idiosyncratic reasons, and one
personeach thought that being grossed out was more indicative of
onesmoral character, less subject to idiosyncratic perspectives,
wouldhurt the other person more, or would inspire more
self-shame.
Anger versus contempt. Of the participants preferring angerto
contempt, 60% argued the contempt would last longer or wouldbe
harder to remedy than anger. However, compared with moraldisgust,
fewer participants argued that contempt was a strongerstatement
about a persons inherent moral character (8%). Inaddition, whereas
no participants mentioned issues of status orsuperiority when
comparing anger to moral disgust, 8% of partic-ipants mentioned
this issue in distinguishing anger from contempt.Of the six
participants preferring to be the target of contempt, twoargued
that contempt would be easier to undo.
Anger versus sadness. Participants who preferred makingsomeone
angry justified their choice in several ways: 50% arguedthat
sadness was longer lasting or harder to undo, whereas 21%argued
that they would feel worse for having hurt someone. Ofparticipants
who preferred making someone sad, 25% said thatthey wished to avoid
the violence or aggression of anger, 25%thought anger would be more
difficult to undo, 13% cited greatersympathy for the persons anger,
and 13% cited idiosyncraticreasons.
Discussion
The results of Study 4 strongly supported the predictions of
oursocialfunctionalist perspective. Participants beliefs
concerningthe consequences of each emotion painted a clear picture
of moraldisgust as the most damaging emotion, both because it
seemshardest to undo and is most indicative of a persons inherent
moralcharacter. Anger, according to participants, is the most
easilyremedied, is more concerned with specific actions, and is
mostlikely to arise from a misunderstanding or individually held
atti-tude. Contempt lies somewhere in the middle, being
considerablyworse than anger, but considerably better than moral
disgust.Moreover, though relatively few participants mentioned
issues ofstatus or superiority for any emotion, contempt was the
onlyemotion to receive justifications consistent with such an
appraisal.
Study 5: Action Tendencies and Judgments inRecalled Events
Study 4 supported the predictions of our account indirectly,
bydemonstrating that people distinguish these emotions when
imag-ining being on the receiving end of these emotions. Whether
theseemotions are associated with different action tendencies and
judg-ments where actual events are concerned remains unclear. In
Study5, we asked participants to recall events from their lives and
ratethe responses, judgments, and emotions associated with
each.
We made several predictions. First, we expected anger to
beuniquely associated with active attempts to change or prevent
harmto the self and to be more sensitive than either contempt or
moraldisgust to attempts by the offending party to make
reparations. Itshould not be as strongly correlated with character
judgments of
either incompetence or immorality. Second, in contrast to
anger,moral disgust should be most strongly associated with the
judg-ment that a person is immoral. To the extent that it is a less
activeemotion than anger, it should be associated with avoidance
re-sponses but not with attempts to stop an offensive act. Finally,
tothe extent that it is harder to undo and evokes more
lastingcharacter judgments, it should not be sensitive to an
apology andshould render attempts at reparations less acceptable.
Third, incontrast to both anger and moral disgust, contempt should
beassociated with the judgment that a person is incompetent, as
wellas with feelings of superiority, but should not correlate
withjudgments that the person is immoral.
Method
Participants. Thirty participants, recruited online through
theStanford University Psychology Department participant
database,completed a set of web-based questionnaires in exchange
for $5.Of these, three were excluded for response times that
suggestedthey were not completing the questionnaires in good faith
(i.e.,taking less than 10 s to recall an event and answer 19
questions).The final sample consisted of 27 participants (mean age
34.6 years,range 1960; 70% female and 30% male; 67% Caucasian,
19%Asian or Asian American, 11% Black, 3% Hispanic or Latino).
Materials and procedure. Participants gave informed con-sent,
completed three questionnaires, and then filled out a
shortdemographic form. For each questionnaire, participants were
givenone of three instructions, with order counterbalanced across
sub-jects:
1. Think of the most recent time you can remember whensomeone
did something that had a negative impact onyou. This does not have
to be anything necessarilyimmoral (although it can be), but just
something thataffected you in a bad way. (Self-relevant event)
2. Think of the most recent time you can remember whenyou saw or
heard about someone else (not you) doingsomething immoral, wrong,
or offensive, but that did notdirectly affect you. (Immoral
event)
3. Think of the most recent time you can remember whenyou saw or
heard about someone else (not you) doingsomething stupid, idiotic,
or incompetent. (Incompe-tent event)
After describing the particular event they had recalled,
participantsanswered a set of 19 questions. The first four
questions askedabout specific features of the event: how long ago
the event hadoccurred, whether the participants had seen the event
occur them-selves (or simply heard about it through someone else),
whetherthey had been directly affected by the event, and whether
someonethey knew personally had been directly affected by the
event.
To examine responses hypothesized to relate to anger, we
askedparticipants to indicate on a Likert scale from 0 (not at all)
to 4(extremely) how much they felt like personally putting in
effort toget the person(s) responsible to stop or change what they
weredoing and how much they felt that the person(s) responsible
shouldbe punished for what they did. We also asked them to
indicatewhether the person(s) responsible had apologized or tried
to make
730 HUTCHERSON AND GROSS
-
amends for what they did, with yes, no, and dont know asresponse
options.
To examine responses hypothesized to relate to moral disgust,we
asked participants to indicate on a 4-point Likert scale howmuch
they currently felt like they would go to some effort to
avoidcontact with the person(s) responsible and whether an
apology/amends would (or did) suffice to make up for what the
person did.We also asked them to indicate how much the action had
nega-tively affected their impression of the persons moral
character andhow generally immoral, unethical, or offensive they
thought theactor was.
To examine responses hypothesized to relate to contempt, weasked
participants to indicate on a 4-point Likert scale how muchthey
felt superior to the person(s) described in the event, how muchthe
action had negatively affected their impression of the
personsintelligence or competence, and how generally stupid,
idiotic, orincompetent they thought the person was.
Participants then reported how much they had felt angry,
mor-ally disgusted, contemptuous, and grossed out at the time of
theevent, on a 4-point Likert scale. Because retrospective
accountsmay be biased, and because some of the responses,
particularlycharacter judgments, might be more strongly related to
currentemotions, we also asked participants to report how much of
eachof the four emotions they currently felt about the event.
Results
Manipulation check. Examination of the event
descriptionsprovided by participants confirmed that they followed
instructions.Self-relevant events typically evoked descriptions of
minor of-fenses affecting the participant, such as insensitivity or
irrespon-sibility, but also more serious crimes, such as theft and
cheating.Immoral events evoked descriptions of lying, cheating,
stealing,and cruelty. Most immoral events directly affected someone
otherthan the participant, although a minority (28%) were reported
asaffecting the self as well. Incompetent events evoked
descriptionsof carelessness, illiteracy, and irresponsibility, and
a substantialminority cited dangerous behavior in traffic. Just
under half (46%)reported being affected by the incompetent event
directly. Themajority of events (71%) had occurred within the last
month.
Average emotion intensity across event types. We pre-dicted
that, as in the hypothetical scenarios used in Studies 13,
anger should be highest in self-relevant offenses, moral
disgustshould be highest in moral offenses, and contempt should
behighest in incompetent offenses. Table 4 reports means and
stan-dard deviations for each emotion for each event type. As
predicted,anger was significantly more intense in self-relevant
events thanany other emotion, for both past and current time
periods, allts(26) 4.18, all ps .001, all ds 1.07. Also as
predicted,moral disgust was the most intense emotion reported in
immoralevents, for both past and current emotion, although this was
notsignificantly greater than anger, the second highest emotion,
ateither time point, both paired ts(26) 1.1, ns.
However, contrary to predictions, contempt was only the
secondhighest emotion in incompetent events, being significantly
lowerthan anger at both past, paired t(26) 1.95, p .06, d 0.33,
andcurrent time points, paired t(26) 2.12, p .04, d 0.28, buthigher
than moral disgust at both past, paired t(26) 2.93, p .005, d 0.43,
and current time points, paired t(26) 1.76, p .05 one-tailed, d
0.29.
Association between action tendencies and emotions. Ourprimary
interest in Study 5 concerned not differences in averageemotion
across event types but the associations between specificemotions
and specific action tendencies and judgments. To exam-ine this
question, we first conducted tests to verify that the rela-tionship
between emotions and action tendencies/judgments wasnot
significantly different across event type. Using hierarchicallinear
modeling, with subjects specified as a random factor, wetested for
the presence of a significant interaction between eventtype and the
influence of each emotion (anger, moral disgust,contempt, grossed
out) on each action tendency/judgment (tried tostop, wanted to
punish, would avoid, apology would help, feltsuperior, judged
immoral, changed view of immorality, judgedincompetent, changed
view of incompetence) in either time period.These analyses
indicated that only for moral disgust, and then onlyfor avoidance
behavior, was there a significant interaction betweenevent type and
response: for time of the event, F(2, 75) 3.71,p .03; currently,
F(2, 75) 2.96, p .06. For all other emotionsand action
tendencies/judgments, there were no differences inrelation across
event type, all Fs(2, 75) 2.5, all ps .1. In allfurther analyses,
we thus examined the main effect of each emo-tion on actions and
judgments.
Table 4Average Emotional Intensity by Event Type (Study 5)
Event type
Anger Moral disgust Contempt Grossed out
M SD M SD M SD M SD
Emotion at the time of the event
Self-relevant 2.14a 0.85 0.82b 1.06 1.11b 1.07 0.29c 0.71Immoral
1.61a,b 1.10 1.79a 1.07 1.42b 1.24 0.75c 1.00Incompetent 1.39a 1.23
0.54b 0.79 1.00a 1.12 0.57b 0.92
Emotion currently felt about the event
Self-relevant 1.32a 0.98 0.53b 0.92 0.64b 0.95 0.32b 0.86Immoral
1.00a 1.02 1.25a 1.00 1.14a 1.15 0.54b 0.92Incompetent 0.93a 1.09
0.39b 0.69 0.64a 0.99 0.43c 0.69
Note. N 27. Within a row, emotions that share a superscript do
not differ significantly at the p .05 level.
731SOCIALFUNCTIONALIST ACCOUNT OF MORAL EMOTIONS
-
To determine which emotions were uniquely and
independentlyassociated with a particular action tendency or
judgment, weconstructed a hierarchical linear model with the target
actiontendency/judgment as the dependent variable and both past
andcurrent intensity of anger, moral disgust, contempt, and
grossedout as predictors. The model also included subjects
specified as arandom effect and event type as a fixed effect.
Starting from thisfull model, we then used backward elimination to
remove predic-tors that were not significantly correlated with the
target variableafter controlling for the effect of all other
included variables,removing the least significant predictor at each
step. We continuedwith backward selection until all variables
continued to signifi-cantly predict that measure. For example, in
the full model pre-dicting the action tendency related to attempts
to stop or intervenein the event (1*Past Anger 2*Current Anger
3*PastMoral Disgust 4*Current Moral Disgust 5*Past Con-tempt
6*Current Contempt 7*Past Grossed Out 8*Current Grossed Out 9*Event
Type), we found that onlypast anger was significantly related to
attempts to stop (b 0.41,SE 0.17, p .02), and no other variable
predicted this measureat the p .05 level. We then proceeded by
running the samemodel, dropping the least significant variable from
the analysis (inthis case, current anger, p .84). We proceeded in
this manneruntil all included variables were significant at p .05,
leaving inthis case only anger ( p .001). Below, we summarize the
resultsof this procedure for each action tendency/judgment. As
supple-ments to the full model, we also analyzed the effects of
theseemotions separately for intensity at each time point (e.g.,
1*PastAnger 2*Past Moral Disgust 3*Past Contempt 4*PastGrossed Out
5*Event Type). We discuss the results of thesemodels only if they
differed meaningfully from the full model.
Anger-related responses. The hypothesis that anger shouldbe
uniquely related to attempts to stop or intervene against
theactions of the perpetrator when an offense took place was
stronglyconfirmed. After elimination, only anger experienced at the
time ofthe event predicted actions taken to stop the perpetrator (b
0.54,SE 0.10, p .001).
We also predicted that if anger motivates active response until
awrong has been alleviated, it should be deactivated if the
perpe-trator has apologized or tried to make amends. To test this,
weexamined the relation between each emotion and the presence ofan
apology (coded as 1 if present, 0 if absent, and .5 if
theparticipant was unsure). As expected, only currently felt anger
wascorrelated, negatively, with the presence of an apology (b
0.12,SE 0.05, p .01).
Finally, we predicted that anger might be implicated in
wantingto punish a person in order to prevent future threat. This
predictionwas only partially confirmed. Anger at the time of the
event waspositively correlated with the desire to punish (b 0.29,
SE 0.10, p .004), but moral disgust at that time predicted
thisimpulse even more strongly (b 0.49, SE 0.10, p .001).
Moral disgust-related responses. In contrast to the approachand
attack behavior associated with anger, we expected avoidanceto be
more strongly associated with moral disgust. However, weinstead
found that only current anger correlated with being willingto
expend effort to avoid coming into contact with a person (b 0.51,
SE 0.12, p .001).
We further hypothesized that moral disgust should nullify
theeffectiveness of an apology or reparations, because the
judgment
that someone is inherently evil might make such attempts
suspect.Although no emotion correlated with the effectiveness of an
apol-ogy in the full model, when analyzing current emotion only,
moraldisgust alone marginally predicted an apologys effectiveness
(b 0.22, SE 0.13, p .09).
Finally, we predicted and found moral disgust to be the
emotionmost strongly tied to immoral character judgments. Current
moraldisgust was the emotion most strongly associated with
overallimmoral character judgments (b 0.49, SE 0.12, p
.001),whereas past moral disgust was the emotion most strongly
asso-ciated with a changed view of a persons moral character (b
0.77, SE 0.09, p .001). Current anger also predicted moralcharacter
judgments, albeit more weakly (b 0.35, SE 0.10,p .001). No other
emotions significantly predicted changes inmoral character
judgments after backward selection in the fullmodel.
Contempt-related responses. We predicted and found thatcontempt
would be most strongly related to issues relevant tocompetence and
status judgments. Only current contempt wasrelated to feeling
superior (b 0.61, SE 0.08, p .001). Inaddition, current contempt
was the emotion most strongly relatedboth to overall assessments of
incompetence (b 0.48, SE 0.11,p .001) and to changes in competence
judgments (b 0.51,SE 0.09, p .001). Past anger was also more weakly
related toboth judgments (incompetent character: b 0.20, SE 0.11, p
.07; changes in incompetency judgment: b 0.17, SE 0.09, p .05).
Discussion
Substantial support was observed for our predictions. In
recall-ing episodes from their recent past, participants reported
the great-est anger in self-relevant events and the greatest moral
disgust inimmoral events. Although contempt was only the second
highestemotion in incompetence events, behind anger, this may have
beendue in part to the nature of the events recalled, which were
oftenissues affecting the self, or related to dangerous traffic
maneuversrather than incompetence or stupidity per se.
In the more direct test linking the intensity of specific
emotionsto the intensity of specific action tendencies and
judgments, weobserved quite a bit of support for our hypotheses.
First, only angerwas associated with active attempts to stop an
offensive behavior.Anger was also the only emotion to be defused by
the verifiable actof an apology. Second, although moral disgust was
not stronglyassociated with any overt behaviors, it was the emotion
mostpredictive of attributing an immoral character to another
person.Consistent with this, it was also the emotion most likely to
renderan apology or attempts at amends insufficient to make up for
theoriginal act, corroborating the assertions of participants in
Study 4that it is the most difficult emotion to undo. Finally,
contrary to thepredictions of an account in which contempt is a
specifically moralemotion, we found that it was specifically and
uniquely associatedwith judgments related to incompetence,
stupidity, and status, butnot to immorality.
However, we did observe some noteworthy exceptions to thegeneral
confirmation of our predictions. First, we had predictedthat moral
disgust, whose nonsocial counterpart is typically asso-ciated with
withdrawal, would be most strongly associated withavoidance, but
instead we found that only anger predicted this
732 HUTCHERSON AND GROSS
-
behavior. This is consistent with work suggesting that anger
mo-tivates both approach and avoidance behavior (Averill, 1982)
andwith the idea that anger enables higher energy expenditure
insupport of defense against an offensive act. The wording of
ourquestion, which emphasized going to some effort to avoid
aperson, may have been too suggestive of this higher energy
re-sponse, whereas moral disgust may be associated with a
morepassive avoidance than the kind we tested here. Second,
contraryto our predictions, we observed a weaker but consistent
involve-ment of anger in both immorality and competency
judgments,suggesting that anger does play a role in judgments of a
personscharacter. One possibility for this weaker relationship is
that angerdoes indeed provoke negative character judgments, but, as
anemotion more sensitive to attempts at reparation, those effects
areof shorter duration and easier mitigation than the effects of
eitherdisgust or contempt.
General Discussion
The primary question addressed by the studies presented
hereconcerned the separability of the three other-condemning
emo-tions: anger, moral disgust, and contempt. Although some
theorieshave made distinctions between all three emotions (i.e.,
the CADtriad hypothesis; Rozin et al., 1999), more typical models
havemade fewer distinctions (Fiske et al., 2002; Mackie et al.,
2000),and solid evidence concerning either their distinction or
separabil-ity has remained scant. As a whole, the research
presented hereprovides evidence for the value of considering these
emotions asthree separate entities. Varying antecedent appraisals
clearly in-fluenced the emotions people considered relevant to a
particularsituation (Studies 13). People also clearly distinguished
theseemotions in their beliefs about the social consequences,
stronglypreferring anger over contempt and contempt over disgust
(Study4). Finally, these emotions were associated with unique
profiles ofresponses and judgments in real-life events (Study 5).
Althoughother distinctions, also motivated by social functionalism,
havebeen proposed (Rozin et al., 1999), we believe that they may
beimportantly incomplete and that a social functionalism that
incor-porates issues of self-relevance and broadens the domain of
theseemotions to include nonmoral issues such as competence is
re-quired to fully explain the differences between them.
The Social Antecedents and Consequences of Anger,Disgust, and
Contempt
The strongest support for the viability of the
socialfunctionalistaccount presented here was found in antecedent
appraisals thatdifferentiated the three emotions. The results of
Studies 13 sug-gest that, on a semantic level at least, people do
distinguishbetween anger, moral disgust, and contempt. Anger
appears to beevoked by appraisals of self-relevance, disgust seems
to be relatedmost strongly to appraisals that a person is morally
untrustworthy,and contempt seems uniquely related to the judgment
that someoneis incompetent or unintelligent. Given the pattern of
results inStudies 13, it will prove interesting in future work to
furtherelaborate the necessary and sufficient appraisals that
elicit theseemotions. Self-relevance clearly impacts the degree to
which angeris experienced in response to a negative event, but
other consid-erations may be necessary in order to predict whether
anger or
some other emotion will be evoked in response to
someonesbehavior. For example, though it is unclear from the
researchpresented here and elsewhere, introspection and casual
observationsuggest that any significant and consequential act,
regardless of itsmoral overtones, might elicit anger as long as it
is self-relevant.Moreover, it is clear that people can often be
angry at events thaton their surface do not directly impact the
self. Whether such angeris related to an expanded definition of
self that incorporates otherindividuals, groups, or nations, or
whether it is sensitive to otherfactors, remains to be seen.
A similar consideration of the necessary and sufficient
condi-tions for moral disgust and contempt is also instructive. Of
the two,disgust seems to most clearly emerge as specifically and
uniquelytied to intentional, immoral behavior. In Studies 13, the
moralityof the behavior seemed to be directly tied to the degree to
whichparticipants experienced disgust. One possibility is that this
is anartifact of the label moral, which in all studies was applied
to theword disgust and not to anger or contempt. Although this term
wasused deliberately to help separate moral and nonmoral forms
ofdisgust, it will be useful in future studies to linguistically
distin-guish moral and nonmoral forms of anger and contempt for
par-ticipants or to use other ways of distinguishing such
emotions,such as facial expression, to observe the effects that
this additionalspecification has on the extent to which
participants treat theseterms as similar or differentiable.
Despite many of the results supporting the
socialfunctionalistmodels envisioned role for contempt, this
emotion still remainsthe most nebulous of the three. Although
contempt was clearlylinked to incompetence in Study 3, it may be
that this is only oneof a number of necessary eliciting appraisals
for it. Simply beingincompetent may be enough to elicit sadness,
pity, or amusement(as seen in Study 3), but to elicit contempt may
require somethingmore, including but not limited to a judgment of
moral laxness, anunsympathetic nature, or a competitive
relationship to the per-ceiver (cf. Fiske et al., 2002).
Studies 4 and 5 provided complementary evidence to Studies13,
suggesting that these emotions may be linked to distinct setsof
social responses. People clearly think of these emotions ashaving
different consequences and may use them in free speech
tocommunicate to others their likely actions and responses.
Theseemotions also appear to be differentially tied to real-world
behav-iors and judgments. Although anger was most strongly
associatedwith overt responses and was the only emotion sensitive
to at-tempts by another person to repair the damage done by an
action,moral disgust was most strongly associated with moral
characterjudgments, and contempt was most strongly associated with
in-competence character judgments.
Implications for Affective Science
We have argued that it is useful to distinguish anger,
contempt,and moral disgust from one another and from other
affective statessuch as sadness, fear, or basic disgust. Yet it is
also useful to notehow much overlap was observed between these
emotions. Com-paratively large differences in the relative levels
of anger, disgust,or contempt often occurred in the context of
comparatively smalldifferences in the overall or absolute levels.
Even when peopleclaimed that disgust most strongly captured a
particular experi-
733SOCIALFUNCTIONALIST ACCOUNT OF MORAL EMOTIONS
-
ence, they also reported feeling intense contempt, anger, or
evensadness.
What does it mean for affective science that emotional
statesseem to be so frequently coactivated and, to some extent,
inter-changeable? Does it suggest that distinctions among discrete
emo-tions are the result of acquired experience of a stochastic and
oftennoisy nature, rather than the action of cleanly separated,
hard-wired biological mechanisms? Or that our current methodsinboth
lay and scientific settingsfail to capture the truly
relevantdimensions of emotional experience (L. F. Barrett et al.,
2007)?
How does the evidence presented here bear upon the
currentunderstanding of these emotions as they are typically viewed
in theliterature? The considerable overlap and co-occurrence
observedhere, combined with the observation of clear distinctions
betweenthem in antecedent appraisals and functional consequences,
lendssupport to models in which anger, disgust, and contempt share
acommon socioemotional core. These emotions are clearly moresimilar
to each other than they are to other emotions. Any differ-ences
among them may be the result of linguistic and semanticelaborations
overlaid on a single underlying emotion. However,this work also
points to the importance of understanding when, andwhy, these
emotions differ, and several avenues for future explo-ration arise
out of a consideration of the limitations of the researchand the
subtleties of the results.
Limitations and Future Directions
One important limitation of the studies presented here is
theirreliance on self-reported measures of affect and behavior,
whichrestricts any conclusions drawn from the data concerning
thepredictions of the socialfunctional account. Although
generallysupportive evid