美国斯坦福大学心理学论文Hiding Feelings:The Acute Effects of Inhibiting Negative and Positive Emotion
James J. Gross
Stanford University
Robert W. Levenson
University of California, Berkeley
Emotion regulation plays a central role in mental health and illness, but little is known about eventhe most basic forms of emotion regulation. To examine the acute effects of inhibiting negative andpositive emotion, we asked 180 female participants to watch sad, neutral, and amusing films under1 of 2 conditions. Suppression participants (N = 90) inhibited their expressive behavior whilewatching the films; no suppression participants (N = 90) simply watched the films. Suppressiondiminished expressive behavior in all 3 films and decreased amusement self-reports in sad andamusing films. Physiologically, suppression had no effect in the neutral film, but clear effects in bothnegative and positive emotional films, including increased sympathetic activation of the cardiovascularsystem. On the basis of these findings, we suggest several ways emotional inhibition may influencepsychological functioning.
Emotion regulation and dysregulation figure prominently inmental health and illness (Gross & Mufioz, 1995). Indeed, byour count, over half of the nonsubstance related Axis I disordersand all of the Axis II personality disorders involve some formof emotion dysregulation (American Psychiatric Association,1994; see also Thoits, 1985). Thus, for example, major depressivedisorder is characterized by a deficit of positive emotionand/or a surplus of negative emotion; generalized anxiety disorderby heightened levels of anxiety; schizophrenia, disorganizedtype, by inappropriate emotional responses; and histrionic personalitydisorder by excessive emotionality.
Despite the manifest importance of emotion regulation topsychological well-being, surprisingly little has been done todocument adults' attempts to influence which emotions theyhave, when they have them, or how these emotions are experiencedor expressed. ~ This relative neglect is quite puzzling andinvites speculation. Is emotion regulation so ubiquitous that wealready know all there is to know about it? This seems unlikely,in that commonsense views of emotion regulation are remarkablyinconsistent (e.g., the injunction to count to 10 beforeacting so that your anger will disappear seems to contradict theconventional wisdom that bottling up your anger will hurt you).
A second possibility is that the diversity of emotion regulatoryJames J. Gross, Department of Psychology, Stanford University; Robert
W. Levenson, Department of Psychology, University of California,
Berkeley.
This study was submitted in partial fulfillment of the requirements ofames J. Gross's doctoral degree at the University of California, Berkeley.
This research was supported by National Institute of Mental Health(NIMH) National Research Service Award MH 1003401, National Instituteon Aging Grant AG07476, and NIMH Grant MH39895. We thankJoseph Campos, Paul Ekman, Stephen Hinshaw, and George Lakoff fortheir help in formulating this study.Correspondence concerning this article should be addressed to James
J. Gross, Department of Psychology, Jordan Hall, Building 420, StanfordUniversity, Stanford, California 94305-2130. Electronic mail may besent via Internet to simply overwhelms any attempt at scientific analysis.
If this view is correct, the reason that little has been done isthat little can be done.Without minimizing its complexity, we propose that emotionregulation can indeed be broken down into pieces that are amenableto empirical study. One way to do this is to study specificclinical populations. Examples of this approach include studiesthat examine regional lateralization of brain activity in depressedindividuals (Henriques & Davidson, 1991 ) or assessblunted affect in people with schizophrenia (Kring & Neale, inpress). A second, complementary approach examines aspectsof emotion regulation in nonclinical populations that have particularrelevance to clinical syndromes. Such studies both elucidatethe fundamental nature of these emotional responses andhelp establish norms necessary for subsequent determination ofwhether and how these responses are altered in clinical populations.Examples of this approach can be found in work studyingprocesses of emotional suppression (Gross & Levenson, 1993)or rumination (Nolen-Hoeksema, 1993) in nonclinical or subclinicalpopulations. In this article, we take the second approach,with a particular focus on the inhibition of negative and positiveemotion.
Emotional Inhibition: For Better or for Worse?Civilization seems to require that we inhibit the free play ofour emotions, and many have wondered what consequences suchemotional inhibition might have (Elias, 1978; Freud, 1961; Tomkins,1984). Some commentators have feared for the worst. Acentury ago, for example, Freud argued that emotional inhibitionwas an important cause of psychological illness, and his talktherapy was designed to release "strangulated affect" whoseexpression, for one reason or another, had been severely curtailed(Breuer & Freud, 1957/1895, p. 17). Despite the substan-95
This is true despite a recent explosion of interest in the study ofdevelopmental aspects of emotion regulation (e.g., Fox, 1994; Garber &Dodge, 1991 ).96 GROSS AND LEVENSONtial changes from Freud's original theory of pathogenesis that
occurred over the ensuing decades, the idea that emotional inhibitionmay lead to psychological distress remains a central tenetof psychodynamic psychotherapy. Accordingly, an importantgoal of expressive therapies continues to be the fuller expression
(e.g., representation in conscious awareness; experience andexpression in the context of the therapeutic relationship) ofinhibited emotional responses (Alexander & French, 1946;
Brenner, 1982; Levy, 1990).
Recently, others have cast emotional inhibition in a morepositive light, arguing that this and other forms of emotion regulation
represent an essential developmental milestone (Kopp,1989; Saarni, 1990; Thompson, 1991 ). This view is consistentwith the notion that healthy adults often must inhibit (to varyingdegrees) their ongoing emotion-expressive behavior (Tomkins,1984). It also accords well with the evidence concerning thepotent destructive effects of unregulated emotional responding,as in the precipitous expression of anger in intermittent explosivedisorder. On this view, emotional inhibition is not uniformlypathogenic. Indeed, in many circumstances, it may be the failureof emotional inhibition that is problematic.Obviously, there is a middle ground between these two perspectives,one that holds that there is an optimal level of emotionalregulation--somewhere between total strangulation andcompletely unfettered expression. The point we wish to emphasize
here, however, is the extremely important role that emotionalinhibition has played historically in clinical theory.
Empirical FindingsDespite long-standing clinical interest in the topic of emotional
inhibition, the relevant empirical literature is impoverishedand inconsistent (for a review, see Gross & Levenson,1993). Even if we limit ourselves to questions about the inhibitionof emotion-expressive behavior (as opposed to the inhibition
of subjective emotional experience), clear conclusions aredifficult to come by. What little is known, though, suggests thatinhibiting expressive behavior decreases self-reported experienceof some emotions (e.g., pain, pride, and amusement) butnot others (e.g., disgust; Bush, Barr, McHugo, & Lanzetta, 1989;Colby, Lanzetta, & Kleck, 1977; Gross & Levenson, 1993; Lanzetta,Cartwright-Smith, & Kleck, 1976; Leventhal & Mace,
1970; McCanne & Anderson, 1987; Stepper & Strack, 1993;Strack, Martin, & Stepper, 1988). Physiologically, inhibitingexpressive behavior while waiting for a painful shock leadsto decreased skin conductance reactivity (Colby et al., 1977;
Lanzetta et al., 1976), but the effects of inhibiting emotionalresponses to other negative stimuli have been mixed (Gross &
Levenson, 1993; Zuckerman, Klorman, Larrance, & Spiegel,1981).
The Present StudyInasmuch as mental health requires that we regulate powerfulemotional impulses--and deficiencies in this ability are implicatedin a wide variety of clinical conditions--it is unfortunatethat more definitive statements about emotional inhibition cannotcurrently be made. To better understand the nature and consequencesof this important emotion-regulatory process, we experimentallymanipulated the inhibition of emotional responding ina sample of healthy volunteers using a similar methodology tothe one we used previously to study disgust (Gross & Levenson,1993). This time, however, we examined both a negative emotion
(sadness) and a positive emotion (amusement), as well asa control neutral emotional state.
Our decision to study healthy participants was dictated by adesire to establish normative data relevant to this form of emotion
regulation before engaging the complexities of clinical samples.Our selection of target emotions was dictated by the desireto extend our previous study of the negative emotion of disgustto a more clinically relevant negative emotion (sadness), whoseregulation is centrally implicated in mood disorders such asmajor depressive disorder. Given the theoretical importanceattached to distinctions between negative and positive emotions,we also wished to include a /positive emotion (amusement) thatalso must at times be suppressed (e.g., to facilitate task focus).In addition, to address the possibility that the observed patternof findings might result from inhibiting any behavior (ratherthan emotional inhibition per se), we included a control conditionin which participants were instructed to inhibit their expressive
behavior while in a neutral emotional state.
Hypotheses
We view emotional inhibition as an active, effortful affair in
which inhibitory processes are recruited and then pitted against
ongoing emotional responses. This leads to the prediction thatsome signs of emotion may be visible even as higher orderinhibitory processes are activated. In addition, if emotional inhibitionrequires effort, it should have physiological consequences.
This leads to the prediction that participants who inhibitongoing emotion-expressive b~havior should show greatersigns of physiological activation that participants who do notengage in emotional inhibition. Finally, given the marked behavioraland physiological changes we anticipated would bewrought by emotional inhibition, it seemed possible that subjectiveexperience also might be affected. Drawing on previous
research, we predicted that emotional inhibition would lead todecreased amusement self-reports.
Method
ParticipantsOne hundred and eighty female undergraduates participated in individualexperimental sessions in order to fulfill a requirement of an introductory
psychology course. 2
Film Stimuli
Four films from a set of standardized emotional film stimuli (Gross &Levenson, 1995 ) were used. The first ( 1.5 min, soundless ) shows flowersin a park (Ekman, Friesen, & O'Sullivan, 1988) and elicits emotionreports that are similar to baseline. The second film (3.5 min, with
sound) shows a comedy routine (amusement film: Morra, Brezner, &Gowers, 1986). This film elicits emotion reports of amusement with
little other emotion. The third film (3.5 min, soundless) shows a geometricdisplay (neutral film: ScreenPeace screensaver). It elicits a relatively
2 Two additional participants were tested but had to be excluded dueto equipment failure (one in each experimental condition).
HIDING FEELINGS 97
neutral emotional state characterized by limited emotional responding.
The fourth film (3.5 min, with sound) shows a funeral scene with adistraught mother (sadness film: Stark & Ross, 1989). This film elicitsemotion reports of sadness with little other emotion,of the physiological systems (i.e., cardiac, vascular, somatic, respiratory,thermoregulatory) especially important to emotional responding.
Data Reduction
Procedure
On arrival, participants were informed that we were "interested in
learning more about emotion" and that their reactions would be videotaped.Participants then watched four films. Prior to each film, participants
rated their current emotional state (prefilm rating) and then satquietly for a minute (resting baseline). Participants first watched thefilm of flowers in a park (to accustom them to the laboratory). Participantsthen saw amusement, neutral, and sadness films (the order of Films2-4 was counterbalanced, and all participants saw all three films). Theinstructions participants received prior to the second, third, and fourthfilms were determined by their random assignment to one of two conditions(no suppression, suppression). In the no-suppression condition,participants received the following instructions:We will now be showing you a short film clip. It is important tous that you watch the film clip carefully, but if you find the filmtoo distressing, just say "stop".
In the suppression condition, participants received the followinginstructions:
We will now be showing you a short film clip. It is important tous that you watch the film clip carefully, but if you find the filmtoo distressing, just say "stop." This time, if you have any feelingsas you watch the film clip, please try your best not to let thosefeelings show. In other words, as you watch the film clip, try tobehave in such a way that a person watching you would not knowyou were feeling anythin~Watch the film clip carefully, but pleasetry to behave so that someone watching you would not know thatyou are feeling anything at all.
Each film was followed by a 1-min postfilm period. After this postfilmperiod, participants completed a self-report inventory to describe their
emotional responses during the film.
Apparatus
The first film (flowers in a park) was included solely to accustomparticipants to the laboratory; data from this film were not analyzed.
For the amusement, neutral, and sadness films, data reduction for behavioraldata was based on the prefilm (1 min) and film (3.5 min) periods.
Data reduction for physiological data was based on the prefilm ( 1 min),instructional (1 min), and film (3.5 min) periods.
Behavior. Participants' behavioral responses were coded by fourraters (2 men, 2 women) who were blind to participants' experimental
conditions and to the nature of the film stimuli. Raters used a behavioral
coding system that included 12 codes (Gross & Levenson, 1993): (a)
happiness; (b) sadness; (c) pleasantness; (d) intensity; (e) body movement;
(f) facial movement; (g) mouth movement; (h) yawning; (i)
smiling; (j) crying; (k) blinks; and (1) face touching. Three of these were
frequency measures (yawning, smiling, blinks; these were converted to
events per minute for analysis), and the rest were continuous measures
whose values were determined by the intensity, duration, and frequency
of response. Reliabilities (derived by considering the average correlations
among all possible comparisons among the four coders) were good
(mean Pearson r = .81; range, r = .64 for sadness to r = .96 for
smiling). Because each participant's behavioral responses were independently
coded by two coders, final values for each of the codes were
determined by averaging each of the coder's ratings for a given participant's
expressive behavior. Change scores were then created for each
variable by subtracting prefilm ratings from film ratings.
Physiology. For somatic activity, interbeat interval, and skin conductance
level, change scores were created by subtracting prefilm period
scores from film period scores for each variable. Then, to reduce the
total number of physiological variables in the analysis, we created on a
priori grounds two additional composite variables, which were designed
to assess theoretically defined patterns of activation. The first was a
measure of sympathetic activation of the cardiovascular system (excluding
heart rate, which is a joint function of both sympathetic and parasympathetic
activation). This measure was derived by combining four unitweighted
standardized change scores (pulse transit time to the finger,
finger pulse amplitude, pulse transit time to the ear, and finger temperature).
The second composite was a measure of respiratory activation,
formed using two unit-weighted standardized change scores (respiratory
period and respiratory depth).
Audiovisual. A color video camera placed behind darkened glass in
a bookshelf was used to record participants' facial behavior and upper
body movement unobtrusively.
Self-report. Participants rated how they felt before each film (prefilm
rating) and, after viewing each film, how they had felt during the
film (film rating). On each occasion, participants used a self-report
inventory consisting of 16 emotion terms (amusement, anger, arousal,
confusion, contempt, contentment, disgust, embarrassment, fear, happiness,
interest, pain, relief, sadness, surprise, tension) and 2 global terms
(pleasantness, intensity). For the 16 emotion terms, participants rated
the greatest amount of each emotion they felt using an anchored 9-point
Likert scale (0 = none, 8 = most in my life) that was adapted from
Ekman, Friesen, and Ancoli, 1980. For the global terms, participants
rated their feelings on 9-point Likert scales (-4 = extremely mild/
unpleasant, 4 = extremely intense/pleasant).
Physiological. Continuous recordings were made using a 12-channel
Grass Model 7 (Astro-Med, Inc., West Warwick, RI) polygraph. Nine
measures were obtained: (a) cardiac interbeat interval; (b) skin conductance
level; (c) finger temperature; (d) finger pulse amplitude; (e) pulse
transit time to the finger; (f) pulse transit time to the ear; (g) respiratory
period; (h) respiratory depth; and (i) general somatic activity. These
measures were selected so as to provide a broad index of the activity
Resul~
To assess the effects of emotional inhibition, we first consider
the physiological effects of simply hearing the instructions to
suppress (prior to the start of the film stimulus). Next, we
examine the impact of emotional suppression (after the start of
the film stimulus) on participants' physiological, behavioral,
and subjective emotional responses.
Instructional Period
During the instructional period, all participants knew they
would soon be seeing a film (they had no way of knowing
which film), and half of the participants knew they would be
trying to suppress their responses to the film. During this instructional
period we measured physiological responses (no
self-reports were obtained and no behavioral coding was done).
To evaluate the effects of the suppression instructions during
this instructional period, we simply conducted an overall 2 level
multivariate analysis of variance (MANOVA; Condition; sup98
GROSS AND LEVENSON
Table 1
Mean Change in Physiological Response and Standard Error of the Mean During the
Instructional Period for No-Suppression and Suppression Participants
Instructional group
No suppression Suppression
Measure M SE M SE t(179) p
Somatic activity 0.03 0.01 0.06 0.01 2.95 .004
Cardiac interbeat interval - 11.23 2.39 - 19.71 2.79 2.31 .02
Skin conductance level -0.22 0.04 0.09 0.06 4.74 <.001
Cardiovascular activation -0.17 0.02 -0.09 0.02 2.54 .01
Respiratory activation -0.11 0.04 0.05 0.04 3.10 .002
pression, no suppression) with all five physiological variables.
As we expected, there was an effect for condition, F(5, 167)
= 6.38, p < .001. To understand this multivariate effect, we
conducted two-level analyses of variance (ANOVAs; Condition;
suppression, no suppression) for each of the five physiological
variables (averaging over the three levels of emotion). As presented
in Table 1, compared with their no-suppression counterparts,
suppression participants showed greater increases in somatic
activity, greater decreases in cardiac interbeat interval
(signifying faster heart rates), greater increases in skin conductance,
greater relative sympathetic activation of the cardiovascular
system, and greater relative respiratory activation.
Film Period
During the film period, participants differed as to which instructions
they received and which film they viewed. During
this period, we coded expressive behavior and measured physiological
responses. Subjective emotional experience was assessed
following the film.
Expressive behavior. An overall 2 x 3 (Condition [ suppression,
no suppression] × Emotion [amusement, neutral, sadness])
MANOVA with all 12 behavioral variables revealed effects
for condition, F( 12, 163) = 15.49, p < .001, and Condition
× Emotion, F(24, 151 ) = 6.48, p < .001. This indicated
that the suppression instructions had an overall effect on participants'
behavioral responses and that this effect varied as a function
of film condition. To understand the effects of the suppression
instructions on participants' expressive behavior during the
films, we conducted 2 × 3 ANOVAs (Condition [ suppression,
no suppression] × Emotion [amusement, neutral, sadness]) for
each of the behavioral measures. As presented in Table 2, these
analyses revealed Condition × Emotion interactions for 11 of
the 12 variables. Follow-up univariate tests revealed that the
suppression instructions generally led participants to inhibit the
specific expressive behaviors genellted by each of the films.
Physiology. An overall 2 x 3 Condition ([ suppression, no
suppression ] x Emotion [ amusement, neutral, sadness ]) MANOVA
with all five physiological variables showed effects both
for condition, F(5, 167) = 4.43, p = .001, and Condition ×
Table 2
Mean Change in Expressive Behavior and Standard Error of the Mean During the Three Film
Periods for No-Suppression and Suppression Participants
Amusement Neutral Sadness
No No No
suppression Suppression suppression Suppression suppression Suppression
Measure M SE M SE M SE M SE M SE M SE C × E ~
Happiness 4.62 0.11 ~ 2.29 0.21 b 0.21 0.07 b 0.04 0.04 b 0.76 0.13 b 0.26 0.08 b 39.55 <.001
Sadness 0.06 0.03 b 0.27 0.09 ~ 0.50 0.09 0.65 0.08 1.96 0.19 b 1.23 0.14 b 6.92 .001
Pleasantness 1.43 0.05 b 0.67 0.08 b -0.32 0.05 -0.31 0.04 -0.57 0.08 -0.43 0.06 28.81 <.001
Intensity 4.02 0.11 h 2.14 0.17 b 0.93 0.10 0.68 0.08 2.02 0.17 b 1.14 0.13 b 25.82 < .00 l
Body movement 1.52 0.17 b 0.52 0.18 b 0.52 0.19 b 0.13 0.16 b -0.17 0.18 -0.12 0.15 5.91 =.003
Face touching 1.11 0.21 0.39 0.21 0.37 0.29 0.16 0.20 0.29 0.26 0.21 0.19 1.23 ns
Face movement 2.74 0.14 b 1.32 0.16 b 0.82 0.11 b 0.29 0.11 b 1.03 0.16 b 0.56 0.10 b 9.07 <.001
Mouth movement 3.21 0.18 b 1.59 0.21 ~ 0.91 0.17 -0.02 0.15 0.47 0.19 0.34 0.15 11.94 <.001
Yawning -0.17 0.04 -0.11 0.03 -0.12 0.04 -0.09 0.03 -0.26 0.05 b --0.07 0.03 b 3.15 .045
Smiling 4.81 0.22 b 1.39 0.19 b 0.03 0.06 -0.07 0.04 0.21 0.06 b -0.03 0.04 b 62.31 <.001
Crying 0.00 0.00 0.07 0.05 -0.02 0.06 0.02 0.01 0.92 0.19 b 0.38 0.12 b 3.42 .033
Blinks 3.49 1.25 6.79 1.34 1.17 1.15 0.34 1.24 5.62 1.44 4.66 1.30 3.47 .033
Note. C = condition; E = emotion.
a dfs = 2, 177. b Mean difference (between instructional groups within a given film) is significant at p < .05.
HIDING FEELINGS 99
Emotion, F( 10, 162) = 5.35, p < .001. To understand this
Condition x Emotion interaction, we used univariate 2 × 3
Condition ([ suppression, no suppression ] × Emotion [ amusement,
neutral, sadness ]) ANOVAs for each of the physiological
variables. These analyses revealed Condition x Emotion interactions
for somatic activity, F( 2, 174) = 18.32, p < .001, interbeat
interval F(2, 177) = 4.25, p = .016, and sympathetic activation
of the cardiovascular system, F(2, 176) = 3.79, p = .024, with
a marginal interaction for respiratory activation, F(2, 175) =
2.59, p = .078. Because this top-down approach did not allow
us to test fully whether the suppression instructions led to increased
skin conductance levels in the emotional films but not
the neutral film, we supplemented this approach with three additional
t tests that assessed whether the suppression instructions
affected skin conductance during the three films.
As presented in Figure 1, follow-up tests revealed that the
suppression instructions had physiological consequences during
the amusement and sadness films, but not during the neutral
film. During the amusement film, suppression participants
showed lesser somatic activity and slower heart rates, but greater
sympathetic activation of the cardiovascular system than nosuppression
participants; mean change in somatic activity: suppression
= 0.02, no suppression = 0.13, t(148.80) = 4.92, p
< .001; mean change in interbeat interval: suppression = 21.94,
no suppression = - 1.49, t(178) = 3.45, p = .001; mean change
in sympathetic activation of the cardiovascular system: suppression
= 0.43, no suppression = 0.11; t(177) = 3.60, p < .001.
During the sadness film, suppression participants showed lesser
somatic activity, greater skin conductance, greater sympathetic
activation of the cardiovascular system, and greater respiratory
activation than their no-suppression counterparts; mean change
in somatic activity: suppression = -0.03, no suppression =
-0.07, t(147.83) = 2.78, p = .006; mean change in skin conductance:
suppression = -.01, no suppression = -.26, t(177)
= 1.97, p = .05; mean change in sympathetic activation of the
cardiovascular system: suppression = 0.30, no suppression =
0.13, t(177) = 2.00, p = .047; mean change in respiratory
activation: suppression = 0.13, no suppression = -0.10, t(177)
= 2.27, p = .024.
Subjective experience. The overall MANOVA for the selfreport
variables revealed neither condition, F( 18, 133) = 1.19,
ns, nor Condition x Emotion, F(36, 115) = 0.16, ns, effects,
indicating that the suppression instructions had no overall effect
on participants' emotion self-reports. This was consistent with
our prediction that the effects of the suppression manipulation
in the self-report domain would be specific (as contrasted with
the more general effects expected in the domains of behavior
and physiology).
Because there was no evidence of any multivariate effects
involving condition, we did not proceed with follow-up univariate
tests for each of the 18 self-report variables. However, we
did conduct specific planned comparisons to test the a priori
hypotheses that suppression participants would report less
amusement than no suppression participants during the amusement
film (but not during the sadness and neutral films), and
that there would be no difference in self-reported sadness be-
0.3
SOMATIC ACTIVITY
g~ 0.2
0.1
0.0
-o.1
-0.2
~t
_N
~t
r&u
INTERBEAT INTERVAL
- :L°7 ,
)'20° I mlm m+ ira7
30
40
SKIN CONDUCTANCE
0.6
0.4
0.2
ON 0
,~ -0.2
-0.4
-0.6
I I I
AMU NEU SAD
SYMPATHETIC ACTIVATION
OF CARDIOVASCULAR SYSTEM
0.6
0.4
0.2
[~ 0.0
N -0.2
0.4
0.3
0.2
0.1
0.0
t.,4 -0.1
-0.2
RESPIRATORY ACTIVATION
I T F
AMU NEU SAD
1 Suppression
[--1 No Suppression
Figure 1. Mean change in physiological responding during amusement (AMU), neutral (NEU), and
sadness (SAD) films; asterisks indicate that means differ significantly (p < .05).
100 GROSS AND LEVENSON
4
1
AMUSEMENT RATINGS
I 1 Suppression [
I I No Suppression
SADNESS RATINGS
F T 1 1- i r T l
AMU NEU SAD AMU NEU SAD
aim Suppression ]
No Suppression
Figure 2. Mean change in amusement and sadness self-reports during amusement (AMU), neutral (NEU),
and sadness (SAD) films; asterisks indicate that means differ significantly (p < .05).
tween instructional groups for any of the three film conditions.
As presented in Figure 2, between-groups t tests comparing
change in amusement for the suppression and no-suppression
participants revealed that suppression participants reported less
amusement than no-suppression participants during the amusement
film and during the sadness film, but not during the neutral
film; mean change in amusement during amusement film: suppression
= 4.04, no suppression = 4.70, t(178) = -2.50, p =
.013; mean change in amusement during sadness film: suppression
= 0.07, no suppression = 0.61, t(156.37) = -2.13, p =
.034; mean change in amusement during neutral film: suppression
= 0.22, no suppression = 0.21, t(178) = 0.05, ns. Comparable
tests for sadness self-reports showed no difference between
instructional groups for any of the three film conditions;
mean change in sadness during amusement film: suppression =
-0.37, no suppression = -0.16, t(178) = 1.37, ns; mean
change in sadness during sadness film: suppression = 3.79, no
suppression = 4.09, t(177) = -1.07, n s; mean change in sadness
during neutral film: suppression = -.08, no suppression
= -.27, t(178) = 1.35, ns.
Discussion
To begin to chart the consequences of one important form of
emotion regulation, we asked participants either to suppress
their behavioral responses or to respond naturally while in negative,
neutral, or positive emotional states. In the following sections,
we review our findings and offer several suggestions as
to how emotional inhibition might promote--or impair--psychological
functioning.
Effects of Preparing to Suppress
The initial effects of emotional inhibition were evident during
the instructional period, even before participants were exposed
to the emotion-eliciting stimulus, During this period, participants
who were told that they would soon have to try to inhibit
their emotion-expressive behavior evidenced widespread physiological
activation compared with those who were not told they
would need to control their reactions. These signs of greater
physiological activation are most likely the autonomic and somatic
concomitants of a generalized preparatory mechanism for
emotion suppression. We believe this preparation can best be
described as participants' "bracing" themselves in anticipation
of what was to come, reflecting both their uncertainty about the
upcoming emotional experience and their concern about whether
they would be able to perform the assigned task of inhibiting
all visible signs of emotion.
Effects of Suppression
We conceptualize emotions as powerful biologically based
reactions that organize our responses to environmental challenges
and opportunities (Levenson, 1994). This leads to the
prediction that overriding emotion response tendencies requires
an active inhibitory process (Gross & Levenson, 1993). Given
the integration of behavior, physiology, and subjective experience
in emotion, we expected emotional inhibition to have important
consequences for behavior, physiology, and subjective
experience.
Expressive behavior. While viewing emotional stimuli that
ordinarily produce strong emotion-expressive behaviors, suppression
participants were able to decrease dramatically (but
not eliminate) their expressive behavior. This points to an impressive
ability in these college-aged participants, an ability that
spans both positive and negative emotions. The fact that this
inhibition was not complete gives credence to the view that
emotions may be controlled, but only to a point (Gross &
Mufioz, 1995).3 It is interesting to note that some participants
3 Suppression participants were rated as showing more sadness expressive
behavior than their no-suppression counterparts during the
amusement film. This may be due to suppression participants' engaging
the muscles that turn down the lip comers (the same muscles that are
naturally activated in sadness) in order to inhibit the upturning of the
lip corners associated with amusement.
HIDING FEELINGS 101
appeared to be able to inhibit their expressive behavior more
completely than others. What determines how well someone is
able to hide what he or she feels? Are people who are good at
emotional inhibition those who typically express little emotion
in their day-to-day lives (Gross & John, 1997)? Or are they
people who use their ability to inhibit emotion in particular
circumstances? Given the purported role of compromised emotion
regulatory mechanisms in psychopathology, it is important
to study emotional inhibition in clinical samples and to see if
clinical improvement, whether brought about by treatment or by
time, is accompanied by normalization of the kinds of emotional
regulatory processes documented here.
Physiology. Because we thought it was the suppression of
emotional expressive behavior that led to the physiological
changes found previously (Gross & Levenson, 1993), we expected
that inhibiting nonemotional behavior would not produce
physiological activation. Indeed, during the neutral film, which
produced very low levels of self-reported emotion (and predominantly
nonemotional expressive behavior such as lip biting and
yawning), no differences were found between no-suppression
and suppression participants for any of the physiological variables.
This suggests that the physiological impact of emotional
suppression grows out of the counterpoising of attempts to inhibit
expression against strong impulses to express. Absent a
stimulus that produces such impulses to express, inhibition of
ongoing behavior apparently has little physiological impact.
When emotional expression was inhibited, however, the impact
was clear. This included decreases in whatever facial and somatic
behavior was associated with the emotional stimulus. Thus, the
amusement film by itself produced increases in smiling and increases
in somatic activity. When participants attempted to suppress
their responses to this film, they manifested lesser increases
in smiling and lesser increases in somatic activity. Similarly, the
sadness film by itself produced increases in sad expressions and
decreases in somatic activity. Attempts to suppress responses to
this film resulted in lesser increases in sad expressions and lesser
decreases in somatic activity. As in Gross and Levenson (1993),
heart rate generally followed somatic activity, a finding in keeping
with the close relationship between cardiac and somatic activity
(Obrist, 1981 ).4 Whereas the foregoing findings showed some
specificity to the emotion elicited, the most striking feature of
emotional suppression common to all emotions we have studied
thus far (both sadness and amusement in the present study and
disgust in our previous studies) is the enhanced sympathetic activation
of the cardiovascular system. This suggests that suppression
of both positive and negative emotions exacts a palpable
physiological cost, particularly when one keeps in mind that this
sympathetic activation of the cardiovascular system occurs despite
decreased metabolic demands caused by the decrement in
manifest expressive behavior. 5
One question that we cannot answer is whether the physiological
activation associated with emotional suppression is different
in type or amount from that brought about by other forms of
emotion regulation (e.g., amplification, Zuckerman et al., 1981;
repression, Brown et al., 1996) or by other forms of inhibition
altogether (e.g., suppression of pain sensations, Cioffi & Holloway,
1993; thought suppression, Roemer & Borkovec, 1994;
Wegner, 1994). Fowles's (1980) three arousal model suggests
that diverse forms of inhibition may share certain features, but
we expect there are both similarities and dissimilarities among
the inhibition of emotional expressive behavior, other forms of
emotion regulation, and still other types of mental work. We
believe this issue clearly bears further study. 6
Subjective experience. Despite the substantial effects of emotional
suppression in the domains of expressive behavior and
physiological responding, this manipulation had only a modest
impact on participants' self-reported emotional experience. In
keeping with previous findings, suppression participants reported
lesser increases in amusement than no-suppression participants
during the amusement film. Our other finding in this domain,
that suppression participants reported lesser amusement than nosuppression
participants during the sadness film, was initially
puzzling. However, it may underscore the closeness of the relationship
between emotional expression and emotional experience
in amusement. Even though the sadness film primarily produced
sadness expressive behavior, it also produced low levels of amusement
expressive behavior (see Table 2), and the suppression
instructions led to decreased levels of amusement expressive behavior
and smiling during this film. Thus, our findings suggest
that whenever amusement expressive behavior is curtailed (regardless
of what the predominant emotion is), there is a corresponding
decrease in the subjective experience of amusement. 7
4 When somatic activity was used as a covariate, there was no longer a
reliable difference in heart rate between suppression and no-suppression
groups during the amusement film, F( 1, 174) = 1.38, p = .24.
5 Because the films elicited expressive behavior that differed both in
amount and kind, strong statements about emotion-specific effects of
suppression cannot be made. However, some evidence for qualitative
rather than quantitative effects is available if we use the difference
between mean behavioral intensity ratings of no-suppression participants
and suppression participants as an estimate of the degree of emotional
suppression in each film condition. A simple quantitative hypothesis
would predict that the effects of suppressing amusement (difference in
mean ratings of behavioral intensity for no-suppression and suppression
participants = 1.88) should be much greater than the effects of suppressing
sadness (differences in mean ratings of behavioral intensity =
0.88), but this is clearly not so.
6 One puzzle is why our physiological findings are at odds with those
of some previous workers. One explanation is methodological. For example,
Zuckerman and coworkers (1981) found that participants in the
suppression condition generally showed less physiological arousal than
participants in the spontaneous condition, but their analyses were conducted
using a composite measure of physiological arousal that commingles
sympathetic and parasympathetic activation, and they collapsed
across positive and negative film conditions. Because suppression appears
to have different effects on sympathetic and parasympathetic measures,
and in positive versus negative emotional states, it may be that
methodological factors are responsible for reported differences in findings.
Other differences may be more substantive. For example, Lanzetta
and colleagues (1976) found that inhibiting one's expressive behavior
while waiting for a painful shock decreased physiological responding.
This study differed from ours both in the induction procedure and in
the induced state, and it is not clear which difference is responsible for
the discrepant findings. One possibility, however, is that pain is best
conceptualized as a nonemotional state, in which case this study would
provide further evidence that concealing nonemotional behaviors (such
as gross motor activity associated with pain) does not produce the
physiological tug of war associated with the increased physiological
activation observed in the present study.
7 This point is reinforced by the results of secondary analyses that
indicated that the effects of the suppression instructions during the
amusement film were so specific that even happiness and contentment
self-reports were unaffected.
102 GROSS AND LEVENSON
Why might suppression decrease subjective emotional experience
for a positive emotion such as amusement but not for
negative emotions such as sadness or disgust? One explanation
stems from the observation that people seem to control negative
emotions more frequently than positive emotions (Wallbott &
Scherer, 1989). If this is so, adults would have had many more
experiences of disjunctions between emotional experience and
emotional expression for negative emotions than for positive
emotions. Conceivably, this might lead them to rely less heavily
on their expressive behavior when judging their own emotional
experience for negative emotions than for positive emotions.
Emotional Inhibition and Mental Health
Our findings suggest that for negative emotions such as sadness,
inhibiting emotional expressive behavior does not provide
relief from the subjective experience of that emotion. Thus, as
a route to the alleviation of negative feelings, hiding one' s emotions
is unlikely to help one feel better. Of course, this does not
mean that restraining emotional impulses (e.g., to yell or to hit)
is never desirable. It is in fact easy to imagine circumstances
under which it is better for one's own (and others' ) psychological
health and general welfare to curtail one's expressive behavior.
For example, decreasing displays of depressive affect may
increase the likelihood of receiving succorance (Coyne, 1976),
and interrupting strings of reciprocal negative affective displays
may improve marital interaction (Levenson & Gottman, 1983).
For a variety of reasons, there may be times when it is vital to
dissociate the emotions we feel from the behavior we express
(Gross & Mufioz, 1995).
Under other circumstances, however, emotional suppression
may interfere with successful adjustment. Particularly when
emotional inhibition is chronic, inflexible, and insensitive to the
nuances of the social environment, it may impair the efficiency
of cognitive processing, it may block adaptive action, and it may
limit the ability of our social partners to accurately track (and
thus respond appropriately to) our needs and plans. In the next
two sections, we discuss each of these potential implications of
emotional inhibition.
Emotional inhibition and cognitive performance. Emotional
inhibition leads to widespread increases in sympathetic activation.
Such heightened physiological activation has been shown
to lead to impaired sensory intake and sensorimotor integration
(Lacey & Lacey, t979). This suggests that emotional inhibition
might diminish cognitive performance. A resource allocation
perspective makes a similar prediction. If cognitive capacity is
finite, and the inhibition of ongoing emotion-expressive behavior
can be viewed as a task that requires cognitive resources,
this additional task might decrease cognitive performance. Although
we know of no study directly assessing the cognitive
consequences of emotional inhibition, indirect evidence comes
from a study of behavioral inhibition by Gilbert, Krull, and
Pelham (1988). These researchers found that participants who
were asked to regulate their gaze to avoid looking at certain
words drew more incorrect dispositional inferences about a target
than did participants not asked to regulate their gaze. The
findings were interpreted as suggesting that "individuals may
spend so much effort regulating their gaze that they are unable
to perform the resource-limited operations that accurate social
inference requires" (Gilbert et al., 1988, p. 688). Although it
is a sizable step from gaze inhibition to emotion inhibition,
this finding encourages speculation that emotion inhibition (and
perhaps other forms of emotion regulation) also might decrease
cognitive performance. If this speculation is borne out, this may
shed light on the performance decrements associated with mood
and anxiety disorders, perhaps caused in part by the cognitive
costs of ongoing attempts at emotion regulation.
Emotional inhibition, behavior, and social interaction. Because
emotions comprise integrated packages of response tendencies
designed to coordinate adaptive behavior in the face of
challenge (Levenson, 1994; Tooby & Cosmides, 1990), habitually
and inflexibly overriding these responses may compromise
an individual's ability to manage these challenges successfully.
Thus, if one is disgusted by what one is eating but inhibits the
food-expulsive behaviors associated with disgust, the resultant
continuation of eating increases the risk for illness. If the tendency
to withdraw from a dangerous situation that is part of
fear is inhibited, one may take unwelcome risks. If the anger
occasioned by mistreatment by another person is hidden, one's
treatment is unlikely to improve.
The last example highlights an essential function of emotionexpressive
behavior, namely the communication of our emotional
states to others, thereby influencing their behavior. Such
nonverbal information flow is essential for successful interpersonal
functioning. As theorists since Darwin (1872) have noted,
we rely on our social partners' emotional expressions to give
us information about their needs and preferences. For example,
if we inadvertently anger someone, their angry expression signals
the effect of our behavior and lets us know that we should
engage in corrective behaviors (e.g., apologizing). However, if
the person we have angered systematically suppresses emotionexpressive
behavior, we are less likely to be aware of the problem
and therefore less likely to change what we are doing. In
this case, the person who is doing the suppressing may continue
to have strong negative emotional responses--perhaps at even
greater intensity levels and with ever-increasing frequency because
the unsignaled emotions do not engender corrective behaviors
on the part of others. Indeed, this may be one important
element of the emotional miscommunication evident in so many
forms of psychopathology.
References
Alexander, E, & French, T. M. (1946). Psychoanalytic therapy: Principles
and applications. New York: Ronald Press.
American Psychiatric Association. (1994). Diagnostic and statistical
manual of mental disorders (4th ed.). Washington, DC: Author.
Brenner, C. (1982). The mind in conflict. Madison, CT: International
Universities Press.
Breuer, J., & Freud, S. (1957). Studies on hysteria (J. Strachey, Trans.).
New York: Basic Books. (Original work published 1895).
Brown, L. L., Tomarken, A. J., Orth, D. N., Loosen, E T., Kalin, N. H., &
Davidson, R.J. (1996). Individual differences in repressive-defensiveness
predict basal salivary cortisol levels. Journal of Personality
and Social Psychology, 70, 362-371.
Bush, L. K., Barr, C. L., McHugo, G. J., & Lanzetta, J. T. (1989). The
effects of facial control and facial mimicry on subjective reactions to
comedy routines. Motivation and Emotion, 13, 31-52.
Cioffi, D., & Holloway, J. (1993). Delayed costs of suppressed pain.
Journal of Personality and Social Psychology, 64, 274-282.
Colby, C. Z., Lanzetta, J. T., & Kleck, R. E. (1977). Effects of expresHIDING
FEELINGS 103
sion of pain on autonomic and pain tolerance responses to subjectcontrolled
pain. Psychophysiology, 14, 537-540.
Coyne, J. C. (1976). Depression and the response of others. Journal of
Abnormal Psychology, 85, 186-193.
Darwin, C. (1872). The expression of the emotions in man and animals.
London: Murray.
Ekman, P., Friesen, W. V., & Ancoli, S. (1980). Facial signs of emotional
experience. Journal of Personality and Social Psychology, 39, 1124-
1134.
Ekman, P., Friesen, W. V., & O'Sullivan, M. (1988). Smiles when lying.
Journal of Personality and Social Psychology, 54, 414-420.
Elias, N. (1978). The civilizing process (E. Jephcott, Trans.). New York:
Urizen Books.
Fowles, D. C. (1980). The three arousal model: Implications of Gray's
two-factor learning theory for heart rate, electrodermal activity, and
psychopathy. Psychophysiology, 17, 87-104.
Fox, N. A. (Ed.). (1994). The development of emotion regulation: Biological
and behavioral considerations. Monographs of the Society for
Research in Child Development, 59 (Numbers 2-3, Serial No. 240).
Freud, S. ( 1961 ). Civilization and its discontents (J. Strachey, Trans.).
New York: W. W. Norton.
Garber, J., & Dodge, K. A. (Eds.). ( 1991 ). The development of emotion
regulation and dysregulation. Cambridge, England: Cambridge University
Press.
Gilbert, D. T., Krull, D. S., & Pelham, B. W. (1988). Of thoughts unspoken:
Social inference and the self-regulation of behavior. Journal of
Personality and Social Psychology, 55, 685-694.
Gross, J. J., & John, O. P. (1997). Revealing feelings: Facets of emotional
expressivity in self-reports, peer ratings, and expressive behavior.
Journal of Personality and Social Psychology, 72, 434-447.
Gross, J~ J., & Levenson, R. W. (1993). Emotional suppression: Physiology,
self-report, and expressive behavior. Journal of Personality and
Social Psychology, 64, 970-986.
Gross, J. J., & Levenson, R. W. (1995). Emotion elicitation using films.
Cognition and Emotion, 9, 87-108.
Gross, J.J., & Mu~oz, R. E (1995). Emotion regulation and mental
health. Clinical Psychology: Science and Practice, 2, 151 - 164.
Henriques, J. B., & Davidson, R. J. (1991). Left frontal hypoactivation
in depression. Journal of Abnormal Psychology, 100, 535-545.
Kopp, C. B. (1989). Regulation of distress and negative emotions: A
developmental view. Developmental Psychology, 25, 343-354.
Kring, A.M., & Neale, J.M. (in press). Do schizophrenic patients
show a disjunctive relationship among expressive, experiential, and
psychophysiological components of emotion? Journal of Abnormal
Psychology.
Lacey, J. I., & Lacey, B. C. (1979). Somatopsychic effects of interoception.
In E. Meyer III & J. V. Brady (Eds.), Research in thepsychobiology
of human behavior (pp. 59-73). Baltimore: Johns Hopkins University
Press.
Lanzetta, J.T., Cartwright-Smith, J., & Kleck, R.E. (1976). Effects
of nonverbal dissimulation on emotional experience and autonomic
arousal. Journal of Personality and Social Psychology, 33, 354-370.
Levenson, R. W. (1994). Human emotion: A functional view. In P. Ekman
& R. Davidson (Eds.), The nature of emotion: Fundamental
questions (pp. 123-126). New York: Oxford University Press.
Levenson, R. W., & Gottman, J. M. (1983). Marital interaction: Physiological
linkage and affective exchange. Journal of Personality and
Social Psychology, 45, 587-597.
Leventhal, H., & Mace, W. (1970). The effect of laughter on evaluation
of a slapstick movie. Journal of Personality, 38, 16-30.
Levy, S. T. (1990). Principles of interpretation. Northvale, NJ: Jason
Aronson.
McCanne, T. R., & Anderson, J. A. (1987). Emotional responding following
experimental manipulation of facial electromyographic activity.
Journal of Personality and Social Psychology, 52, 759-768.
Morra, B., Brezner, L. (Producers), & Gowers, B. (Director). (1986).
Robin Williams live [Film]. (Available from Vestron Video, Stamford,
CT).
Nolen-Hoeksema, S. (1993). Sex differences in control of depression.
In D. M. Wegner & J. W. Pennebaker (Eds.), Handbook of mental
control (pp. 306-324). Englewood Cliffs, NJ: Prentice-Hall.
Obrist, P. A. ( 1981 ). Cardiovascularpsychophysiology: A perspective.
New York: Plenum Press.
Roemer, L., & Borkovec, T D. (1994). Effects of suppressing thoughts
about emotional material. Journal of Abnormal Psychology, 103,
467-474.
Saarni, C. (1990). Emotional competence: How emotions and relationships
become integrated. In R. A. Thompson (Ed.), Nebraska Symposium
on Motivation (pp. 115-182 ). Lincoln: University of Nebraska
Press.
Stark, R. (producer), & Ross, H. (Director). (1989). Steel magnolias
[Film]. (Available from Columbia/Tri-Star, Burbank, CA).
Stepper, S., & Strack, E (1993). Proprioceptive determinants of emotional
and nonemotional feelings. Journal of Personality and Social
Psychology, 64, 211-220.
Strack, E, Martin, L., & Stepper, W. (1988). Inhibiting and facilitating
conditions of the human smile: A nonobtrusive test of the facial feedback
hypothesis. Journal of Personality and Social Psychology, 54,
768-777.
Thoits, P. A. (1985). Self-labeling processes in mental illness: The role
of emotional deviance. American Journal of Sociology, 91, 221-249.
Thompson, R. A. (1991). Emotional regulation and emotional development.
Educational Psychology Review, 3, 269-307.
Tomkins, S. (1984). Affect theory. In K. R. Scherer & P. Ekman (Eds.),
Approaches to emotion (pp. 163-195). Hillsdale, NJ: Erlbaum.
Tooby, J., & Cosmides, L. (1990). The past explains the present: Emotional
adaptations and the structure of ancestral environments. Ethology
and Sociobiology, 11, 375-424.
Wallbott, H. G., & Scherer, K. R. (1989). Assessing emotion by questionnaire.
In R. Plutchik & H. Kellerman (Eds.), Emotion: Theory,
代写research, and experience (Vol. 4, pp. 55-82). New York: Academic
Press.
Wegner, D. M. (1994). Ironic processes of mental control. Psychological
Review, 101, 34-52.
Zuckerman, M., Klorman, R., Larrance, D. T, & Spiegel, N. H. (1981).
Facial, autonomic, and subjective components of emotion: The facial
feedback hypothesis versus the externalizer-internalizer distinction.
Journal of Personality and Social Psychology, 41, 929-944.
Received May 23, 1995
Revision received June 13, 1996
Accepted June 28, 1996 •