Inhibiting and facilitating conditions of the human smile: a nonobtrusive test of the facial feedback hypothesis.

Abstract

We investigated the hypothesis that people's facial activity influences their affective responses. Two studies were designed to both eliminate methodological problems of earlier experiments and clarify theoretical ambiguities. This was achieved by having subjects hold a pen in their mouth in ways that either inhibited or facilitated the muscles typically associated with smiling without requiring subjects to pose in a smiling face. Study 1's results demonstrated the effectiveness of the procedure. Subjects reported more intense humor responses when cartoons were presented under facilitating conditions than under inhibiting conditions that precluded labeling of the facial expression in emotion categories. Study 2 served to further validate the methodology and to answer additional theoretical questions. The results replicated Study 1's findings and also showed that facial feedback operates on the affective but not on the cognitive component of the humor response. Finally, the results suggested that both inhibitory and facilitatory mechanisms may have contributed to the observed affective responses.

Full text

Page 1

Journal of Personality and Social Psychology
1988, Vol. 54, No. 5, 768-777
Copyright 1988 by the American Psychological Association, Inc.
0022-35l4/88/$00.15
Inhibiting and Facilitating Conditions of the Human Smile:
A Nonobtrusive Test of the Facial Feedback Hypothesis
Fritz Strack
Universitat Mannheim
Mannheim, Federal Republic of Germany
Leonard L. Martin
University of Illinois
at Urbana-Champaign
Sabine Stepper
Universitat Mannheim
Mannheim, Federal Republic of Germany
We investigated the hypothesis that people's facial activity influences their affective responses. Two
studies were designed to both eliminate methodological problems of earlier experiments and clarify
theoretical ambiguities, This was achieved by having subjects hold a pen in their mouth in ways that
either inhibited or facilitated the muscles typically associated with smiling without requiring subjects
to pose in a smiling face. Study 1 's results demonstrated the effectiveness of the procedure. Subjects
reported more intense humor responses when cartoons were presented under facilitating conditions
than under inhibiting conditions that precluded labeling of the facial expression in emotion catego-
ries. Study 2 served to further validate the methodology and to answer additional theoretical ques-
tions. The results replicated Study 1 's findings and also showed that facial feedback operates on the
affective but not on the cognitive component of the humor response. Finally, the results suggested
that both inhibitory and facilitatory mechanisms may have contributed to the observed affective
responses.
Research on the role of peripheral physiological reactions in
the experience of emotion has placed its main emphasis on the
influence of facial muscular activity. A great number of studies
have dealt with whether and how people's facial expressions in-
fluence their affective experience. The basic hypothesis of these
studies is derived from Darwin's (1872) early contention that
an emotion that is freely expressed by outward signs will be
intensified, whereas an emotion whose expression is repressed
will be softened (p. 22). In other words, Darwin suggested that
in the presence of an eliciting emotional stimulus a person's
emotional experience can be either strengthened or attenuated
depending on whether it is or is not accompanied by the appro-
priate muscular activity.
Darwin's statement is the predecessor of the current facial
This research was supported by grants from the Deutsche For-
schungsgemeinschaft to Fritz Strack and Norbert Schwarz and by
NIMH postdoctoral research fellowship F32 MH-09194 to Leonard
Martin. Study 2 is based on Sabine Stepper's Diploma thesis at the Uni-
versity of Mannheim. Part of this research was presented at the 28th
Tagung experimentell arbeitender Psychologen in Saarbriicken (March
1986) and at the 94th convention of the American Psychological Associ-
ation in Washington, DC (August 1986).
Appreciation is extended to Paul Kodros and Gerlinde Willy for their
assistance in data collection; to Vera Strack and Gerald Clore for sugges-
tions concerning the study's design; to Robert S. Wyer, Thomas Ostrom,
Eva Me, Norbert Schwarz, Michaela Wanke, Charles Judd, several
anonymous reviewers, and to the University of Illinois Social Cognition
Group for their helpful comments.
Correspondence concerning this article should be addressed to Fritz
Strack, Fakultat fur Sozialwissenschaften, Universitat Mannheim, D-
6800 Mannheim, Federal Republic of Germany.
feedback hypothesis. Although distinctions were made among
several variants of this hypothesis (e.g., Buck, 1980; Winton,
1986), its core is the "causal assertion that feedback from facial
expressions affects emotional experience and behavior" (Buck,
1980, p. 813).
The effects of facial activity have been investigated by using
three classes of dependent variables: self-reports, recall mea-
sures, and autonomic indexes. Typically, it has been found that
(a) facial expressions influence affective self-reports and ratings
of affective stimuli in the direction of the hedonic value of the
expressed emotion (e.g., Laird, 1974); (b) emotional facial ex-
pressions improve the recall of hedonically consistent material
(Laird, Wagener, Halal, & Szegda, 1982); and (c) facial expres-
sions have autonomic consequences, such as changes in heart
rate, skin temperature, skin conductance, and blood volume
(e.g., Ekman, Levenson, & Friesen, 1983; McCaul, Holmes, &
Solomon, 1982; Zuckerman, Klorman, Larrance, & Spiegel,
1981).
Although reviews of this research differ in their assessment
of the presently available results (cf. Buck, 1980; Laird, 1984;
Winton, 1986), there seems to be agreement with the conclu-
sion that "facial feedback has a small but reliable moderating
effect on the emotional experience and on the evaluation of
emotional stimuli" (Kraut, 1982, pp. 861f). No agreement ex-
ists, however, on how people's facial expressions influence their
emotional reactions.
Two classes of mechanisms have been proposed as possible
mediators. Some theorists hold that cognitive processes are re-
sponsible for the effect. Laird (1974), for instance, assumed that
a self-perception mechanism underlies the facial feedback phe-
768

Page 2

INHIBITING AND FACILITATING CONDITIONS
769
nomenon. According to this view, people who perceive them-
selves to be smiling infer that they are probably happy, whereas
people who perceive themselves to be frowning infer that they
are probably sad.
A second group of theorists (e.g., Ekman et al., 1983; Izard,
1977; Tomkins, 1962, 1979) hold that cognitive mediation is
not necessary for the facial feedback effect to occur. According
to these theorists, physiological mechanisms may be sufficient
to generate the affective reactions. From this perspective, a fa-
cial expression may affect people's emotional experience with-
out them being aware of their expression.
Facial Simulation Procedure
The facial feedback hypothesis has typically been tested by
inducing subjects to simulate facial expressions that represent
particular emotions and then measuring their emotional states.
This technique was introduced by Laird (1974), who used a
cover story about measuring facial muscular activity and at-
tached surface electrodes to subjects between their eyebrows, at
the corners of their mouths, and on their jaws. Then a set of the
electrodes was touched and subjects were asked to contract their
muscles at these points. By using this procedure, Laird was able,
without ever mentioning an emotion or an emotional expres-
sion, to induce subjects to either smile or frown. He found
differences in the subjects' reported moods and in their funni-
ness ratings of cartoons. Specifically, subjects in the smile con-
dition reported feeling happier and rated cartoons as funnier
than did subjects in the frown condition.
This facial simulation procedure, or variants of it, has fre-
quently been used to investigate the effects of facial feedback.
Unfortunately, however, ambiguities associated with that meth-
odology have clouded the theoretical implications of the find-
ings that have been obtained. More specifically, the facial simu-
lation procedure may not effectively prohibit subjects from rec-
ognizing the emotional meaning of the manipulated facial
activity. On the basis of this possibility, the issue has been raised
whether subjective responses produced by the facial simulation
procedure may have been the result of situational demands
(e.g., Buck, 1980; Ekman & Oster, 1982; Zuckerman et al.,
1981). In his original studies, Laird (1974) assessed subjects'
awareness of the experiments' purpose. As a result, 16% of the
subjects in the first experiment and 19% of the subjects in the
second experiment indicated awareness of the relation between
the manipulated expressions and their feelings and had to be
excluded from the analysis.
Even if motivational influences to act in a manner consistent
with the facial feedback hypothesis can be avoided, there is a
second way in which the recognition of the emotional meaning
of one's facial expression may affect subsequent ratings (cf.
Winton, 1986). Research on category accessibility (e.g., Hig-
gins, Rholes, & Jones, 1977; Srull & Wyer, 1979) as well as on
priming of emotion categories (cf. Bower, 1981; Clark & Isen,
1982) suggests that the recent activation of such a category (e.g.,
to interpret one's facial activity) may render this and semanti-
cally related categories, as well as related episodic information,
more accessible for later use.
In sum, the facial-posing procedure leaves open the possibil-
ity that the experimentally induced facial activity can be inter-
preted in emotion categories and that this interpretation, by
itself and in combination with motivational influences, may
confound the operation of direct physiological feedback.
It is important to note that autonomic reactions as a conse-
quence of facial expressions do not rule out the possibility of a
cognitive mediation. One reason is that physiological changes
may follow effortful but nonemotional facial activities such as
puffing out the cheeks or closing one eye (Tourangeau & Ells-
worth, 1979). So if emotional expressions differ in this respect,
it is possible that autonomic reactions are caused by the physi-
cal effort they require rather than by the emotional quality of
the expression (cf. McCaul et al., 1982).
A second reason why autonomic responses do not rule out
cognitive mechanisms is that thinking has been found to be an
effective means of modifying physiological responses. Imagin-
ing emotionally relevant events influences not only affective
self-reports (cf. Strack, Schwarz, & Gschneidinger, 1985) but
also autonomic reactions (e.g., Lang, Kozak, Miller, Levin, &
McLean, 1980; Miller et al., 1981; Roberts & Weerts, 1982).
Findings of physiological changes in response to facial posing
do not, therefore, necessarily imply a primacy of physiological
processes, nor do they convincingly demonstrate that physio-
logical mechanisms are sufficient to produce the facial feedback
effect.
Dissimulation/Exaggeration Paradigm
A line of research conducted by Lanzetta, Kleck, Zucker-
man, and their colleagues (e.g., Kleck et al., 1976; Lanzetta,
Cartwright-Smith, & Kleck, 1976; Zuckerman etal., 1981)also
addresses the facial feedback hypothesis. These researchers in-
duced subjects to suppress or to exaggerate facial expressions
in the presence of emotional stimuli such as electric shocks or
pleasant versus unpleasant videotapes.
In these studies, subjects' facial muscles were not manipu-
lated into emotional expressions. Rather, subjects were directly
asked to modify the expressions they would normally have in
response to a stimulus situation. Zuckerman et al. (1981), for
instance, instructed subjects in the suppression condition not
to reveal by their facial expression which of two videotapes
(pleasant vs. unpleasant) they were watching. In the exaggera-
tion condition, subjects were told to "pose appropriate facial
expressions" (p. 933) so that observers would be able to identify
the presented tape from their faces.
Typically, the instructions to dissimulate or exaggerate facial
expressions resulted in corresponding subjective responses and
in autonomic reactions. These findings support the facial feed-
back hypothesis. However, they do not allow a discrimination
between physiological and cognitive feedback mechanisms.
Zuckerman et al. (1981) admit that "the present procedures do
not permit a clear identification of the processes mediating the
observed relationships" (p. 942). That is, the facial expressions
may have given rise to cognitions ("attributions about affective
states," p. 942) which in turn may have affected physiological
activity.
Moreover, it is possible that subjects may have used cognitive
strategies to support the required facial expressions. For exam-

Page 3

770
E STRACK, L. MARTIN, AND S. STEPPER
pie, they may have directed their attention away from the emo-
tional stimulus in an effort to suppress their expression of the
emotion elicited by it. Correspondingly, they may have inten-
tionally increased the emotional intensity of their thoughts
about the stimulus in order to generate the appropriate emo-
tional expression. Such cognitive mechanisms may have con-
tributed to observed emotional responses.
Methodological Alternative
The main obstacle to precluding possible situational influ-
ences and investigating the underlying mechanisms of facial
feedback seems to be that the emotional meaning of the manip-
ulated facial expressions is either explicit or can be recognized
by the subjects. A procedure that induces subjects to contract
muscles associated with certain expressions in a way that does
not simulate the expression itself could therefore be an impor-
tant improvement to the experimental methodology that would
help us clarify the processes that mediate the facial feedback
effect.
This study is an attempt to provide a clearer test with a new
procedure. If the facial feedback hypothesis is assumed to be
correct, and if no interpretational mediation is required for the
effect to occur, then the inhibition or facilitation of the muscle
contractions associated with a particular emotional expression
should be sufficient to modify a person's affective experience,
even though the muscles are not contracted in the simulation of
an emotion-relevant expression. This strategy deviates in im-
portant ways from that typically used to test the facial feedback
hypothesis. It differs from the facial simulation studies in that
no correct expression has to be generated. It requires only that
the muscles typically involved in an expression be activated.
This strategy also differs from the dissimulation/exaggeration
studies in that subjects' facial actions are induced without re-
quiring them to modify an emotional expression. These meth-
odological differences minimize the likelihood that the subjects
interpret their facial activity as representative of a particular
emotion and therefore use cognitive strategies to modify their
emotional experiences.
To create the appropriate facial responses, subjects were in-
duced to hold a pen with their lips only, with their teeth only, or
with their nondominam hand. We assumed that holding the pen
with the lips only would contract the orbicularis oris muscle.
This would be incompatible with contracting the zygomaticus
major or the risorius muscles that are used in smiling. Thus,
muscle activity associated with smiling would be inhibited.
Holding the pen with the teeth only would mainly contract the
zygomaticus major or the risorius muscle that is part of the
smiling response (cf. Hager, 1982). This would facilitate smil-
ing. Holding the pen in the nondominant hand, of course, would
not affect a particular set of facial muscles.
Because a request to hold a pen in one's mouth would cer-
tainly stimulate a variety of speculations about the purpose of
the experiment, great care was taken to create a situation of
high experimental realism (Carlsmith, Ellsworth, & Aronson,
1976). This situation should allow the manipulation of subjects'
facial expressions without directing their attention toward their
facial activity and without inducing them to associate their fa-
cial response with a particular emotion. It appeared possible to
accomplish this in an experiment whose ostensible goal was to
investigate psychomotoric coordination. Specifically, subjects
were told that they were to be in an experiment investigating
people's ability to perform different tasks with parts of their
body not normally used for those tasks, as injured or handi-
capped persons often have to do. Subjects were then asked to
perform a variety of tasks by holding a pen with their lips only,
with their teeth only, or with their nondominant hand. The task
of interest was subjects' ratings of the funniness of cartoons.
Pilot testing had indicated that this cover story was plausible
and that it succeeded in getting subjects seriously involved in
the tasks they were to perform. Thus, we assumed that suspi-
ciousness and demand characteristics were minimized. More-
over, the subjects' attention was directed toward the tasks they
were to perform and not toward their own expressions, as in
previous studies.
It was hypothesized that more or less positive affective states,
as reflected in the experience of humor, would be influenced by
muscle contractions associated with smiling. Specifically, po-
tentially humorous stimuli should be rated least funny when
the muscles associated with smiling are inhibited (lips condi-
tion), but should be rated most funny when this muscular activ-
ity is facilitated (teeth condition). With no manipulation of rele-
vant facial muscles (nondominant hand condition), humor rat-
ings should not be affected.
Two studies were conducted to test this hypothesis and to an-
swer related questions. In the first experiment, cartoons were
rated under different pen-holding conditions. The second exper-
iment was conducted to replicate the previous findings and to
obtain information about the relevant dimensions of the elic-
ited affect and about the relative contribution of inhibiting and
facilitating influences.
Study 1
Method
Subjects. Ninety-two male and female undergraduates of the Univer-
sity of Illinois participated in the experiment in partial fulfillment of a
course requirement. Subjects were run in groups of 4. After their arrival,
the subjects were each assigned a cubicle that prevented communication
between the subjects but allowed them to communicate with the experi-
menter through an open space. On their desks, subjects found a felt-
tipped marker 12 mm in diameter, an alcohol swab, and a paper tissue.
Procedure. After the subjects arrived, the experimenter introduced
the study with the following words:
The study you are participating in has to do with psychomotoric
coordination. More specifically, we are interested in people's ability
to perform various tasks with parts of their body that they would
normally not use for such tasks. As an example, you may have
seen pictures of physically impaired people who use their mouth to
write or use the telephone. Obviously, the ability to do the same
task with different parts of their body has important implications
for these people. For them, the quality of their future life is greatly
dependent on whether they can continue to exercise control over
their environment by being able to perform basic tasks by them-
selves. The tasks we would like you to perform are actually part of
a pilot study for a more complicated experiment we are planning
to do next semester to better understand this substitution process.
The tasks are a sample of a much larger number of tasks, and they

Page 4

INHIBITING AND FACILITATING CONDITIONS
771
involve very different aspects of psychological functioning. Some
of the tasks are related to physical skills, like drawing lines, and
others are related to more "normal" mental activities that people
with a bodily impairment might do during a typical day, like read-
ing a magazine.
The experimenter went on to explain that several methods of holding
a pen other than with one's dominant hand would be tried out. Depend-
ing on the experimental condition, subjects were told to hold the pen
with the nondominant hand, with the teeth, or with the lips. For reasons
of procedural logistics, subjects in the same session were in the same
experimental condition.
Before the first task, subjects were told to disinfect the felt pen with
the alcohol swab provided. Then, under the lips condition, they were
instructed to hold the pen tightly with their lips. It was emphasized that
they should not touch the pen with their teeth. In addition, the experi-
menter demonstrated the correct way to hold the pen, by protruding
the lips as shown in the left half of Figure 1. He also demonstrated the
incorrect way to hold the pen, by compacting the lips tightly against the
teeth. Under the teeth condition, subjects were told to hold the pen with
their front teeth. It was emphasized that they should hold the pen gently,
without touching it with their lips. Again, the experimenter demon-
strated the correct technique (see the right half of Figure 1). Under the
nondominant hand condition, the experimenter asked subjects to hold
the pen with the hand they would not normally use for writing.
The experimental task consisted of four parts that were printed on
separate sheets of paper and presented to subjects in a single booklet.
The first task was depicted as a practice task and involved drawing a
straight line between two points. The second task, which ostensibly be-
gan the experiment proper, involved drawing a line between 10 ordered
digits printed randomly about the page. The digits were printed on
graph paper to increase the task's face validity. In addition, a 10-point
scale was printed on the bottom of this page, and subjects were in-
structed to indicate on this scale how difficult it was for them to perform
the digit-connection task. They were asked to mark the scale holding
the pen as they had while performing the task. This difficulty rating
was included for two reasons. First, it familiarized subjects with making
scale ratings while holding the pen in these positions. Second, we ex-
pected that knowledge of the perceived difficulty of the various ways of
holding a pen would be potentially useful in discriminating between
alternative explanations of any observed effects.
The third task consisted of eight consonants and nine vowels ran-
domly printed on the page. The subjects' task was to underline only the
vowels. After doing so, subjects were again asked to rate the difficulty
of the task on a 10-point scale printed on the bottom of the page.
The fourth task involved the ratings of main interest. In this task,
subjects were told that they would see several cartoons of the type typi-
cally found in magazines and that, as usual, some would seem funnier
than others. Subjects were asked to rate each on a 10-point scale ranging
from not at all funny (0) to very funny (9). Subjects read and rated each
cartoon with the pen held in the original position (i,e., lips, teeth, or
nondominant hand). The four cartoons were taken from Gary Larson's
series The Far Side and had been prerated as being moderately funny
(Jl/=6.61).
Pretesting. To ensure that the experimental task and its cover story
did, in fact, prevent subjects from interpreting their facial activities in
terms of emotion-relevant categories, extensive pretesting was con-
ducted. Twelve subjects similar to those used in the actual experiment
were paced through the experimental procedure and intensively inter-
viewed about their perceptions of the experimental situation. The pre-
testing indicated that the cover story was extremely effective. None of
these subjects suspected the actual purpose of the experiment. In fact,
even after these subjects were told that the study's goal was not the one
that had originally been described to them, their forced guesses con-
tained no reference to either facial expressions or moods. Moreover,
after they were informed of the study's actual purpose and explicitly
asked whether they had entertained the correct hypothesis at any time
during the experiment, all subjects indicated that the only hypothesis
they had considered was the one initially provided by the experimenter.
This unanimous reaction assured us that the cover story was effective,
that the experimental setting did not elicit alternative hypotheses on the
part of the subjects, and that the facial activity was not spontaneously
associated with a particular emotion.
Results
Funniness ratings. The dependent variables of main concern
were the funniness ratings of the four cartoons. Surprisingly, the
average rated funniness from the experimental subjects (M =
4.75) was considerably lower than the pretest ratings (M -
Figure 1. Illustration of the technique used to contract the different facial muscles: left, lips condition;
right, teeth condition.

Page 5

772
E STRACK, L. MARTIN, AND S. STEPPER
Table 1
Ratings ofFunniness and Difficulty: Study 1
Position of pen
Cartoon
Lip
Hand
Teeth
First
Second
Third
Fourth
Mean funniness
Mean difficulty
3.90
4.00
4.47
4.90
4.32
4.47
5.13
4.10
4.67
5.17
4.77
2.72
5.09
4.19
5.78
5.50
5.14
4.91
Note. All ratings were made on a scale from 0 to 9, where a lower value
stands for lower funniness and difficulty, a higher value for higher funni-
ness and difficulty.
6.61). One reason for this unexpected difference may be found
in the divergent ranges of the funniness ratings and the resulting
use of the response scale. The least funny cartoons in the pretest
were far less funny than those in the experiment proper. Conse-
quently, subjects may have adjusted the response scale to ac-
commodate these cartoons and thus assigned higher ratings to
the funnier ones than they did in the main experiment (cf. Os-
trom & Upshaw, 1968).
On the basis of the facial feedback hypothesis, we predicted
that the cartoons would be rated least funny when the activity
of the muscles associated with smiling was inhibited (lips condi-
tion), but would be rated funniest when this activity was facili-
tated (teeth condition). Table 1 gives an index of the funniness
ratings for the four cartoons and for each cartoon separately. As
can be seen, the results clearly support the predictions. Subjects
who held the pen with their lips gave the lowest overall ratings
(M = 4.32), whereas subjects who held it with their teeth gave
the highest ratings (M = 5.14). The ratings for subjects who
held the pen in their nondominant hand fell between these two
extremes (M = 4.77). The predicted differences were confirmed
by a significant linear contrast, f(89) = 1.85, p = .03.'
The predicted pattern proved to be largely consistent for all
individual cartoons. As Table 1 shows, all four cartoons were
rated least funny in the lips condition, and three of the four
cartoons (except Cartoon 1) were rated funniest under the teeth
condition. No significant interaction between the experimental
conditions and the four cartoons resulted from a mixed-model
analysis of variance (ANOVA) treating the cartoons as a within-
subjects factor, F < 1. Taken together, these findings suggest that
inhibiting the muscular activity associated with smiling damp-
ened subjects' experience of humor, whereas facilitating this ac-
tivity intensified their experience. Although the size of the effect
was small, it proved to be quite consistent over the series of
stimuli.
Difficulty ratings. One alternative explanation for these find-
ings may be found in the different degrees of difficulty for the
three experimental conditions. It could be argued thst the more
difficult it was for subjects to hold the pen, the more they were
distracted from the cartoon's humorous content and the less
funny the cartoons were rated. This alternative hypothesis can
be tested by looking at the difficulty ratings from the two tasks
that immediately preceded the funniness ratings.
The bottom row of Table 1 shows the combined difficulty
ratings for the two tasks. As can be seen, there is no correspon-
dence between the pattern of the mean difficulty ratings and the
pattern of the rated funniness. An analysis of covariance yielded
no significant effect for difficulty as a covariate, F < 1. The
effects of the independent variable on funniness ratings in this
analysis were about the same as when task difficulty was not
controlled for. The results clearly indicate that the difference in
funniness ratings was not produced by differences in the diffi-
culty of the three experimental conditions.
Study 2
To strengthen the empirical basis of the results and to sub-
stantiate the validity of the methodology, a second study was
conducted in a different cultural setting. The same pen-holding
procedure was used and subjects had the same task of assessing
a series of cartoons. However, the second study differed from
the first in two important respects. These modifications were
introduced to achieve a better understanding of the processes
that underlie the observed relationship between holding a pen
in a particular way and the differences in the ratings of the car-
toons.
First, the question arises as to whether the pen-holding proce-
dure also affects subjects' emotional feelings in the absence of
an external eliciting stimulus. Specifically, holding the pen un-
der these different conditions may not only modify an existing
emotional experience by inhibiting or facilitating the appropri-
ate facial reaction as proposed by Darwin (1872). It may actu-
ally induce emotional feelings that are not elicited by the hu-
morous stimuli, as proposed by James (1890). This is particu-
larly true for the teeth condition, where holding the pen in the
experimental position not only permits a smile to occur but at
the same time requires the muscles necessary to produce a
smile to be contracted. To evaluate this possibility, we varied
the point at which the subjects were told to hold the pen in the
appropriate position. Half of the subjects held the pen with their
lips (or teeth) both when they were presented with the humor-
ous stimuli and when they rated them. The remaining subjects
were instructed to hold the pen in the appropriate position only
when they gave their ratings.
Three possibilities are theoretically conceivable. First, sub-
jects may use the affect they experience at the time of judgment
as information about their feelings toward the cartoons (cf.
Schwarz & Clore, 1983). If this is the case and if the pen-holding
procedure itself induces different affective reactions, the proce-
dure should influence the ratings in the same direction, regard-
less of whether it is used at the time of the stimulus presentation
1 Although subjects were visually isolated and could not communi-
cate with each other, one might argue that the experimental groups con-
stitute the appropriate unit of analysis. Unfortunately, group codings of
Study 1 were lost through a clerical oversight. However, an analysis of
variance that used sessions as unit of analysis in Study 2 yielded basi-
cally the same effects as using the subjects as unit of analysis (see Foot-
note 2).