Cultures of Moderation and Expression Assignment | Get Paper Help

read the following article (I have uploaded): Soto, J. A., Levenson, R. W., & Ebling, R. (2005). Cultures of moderation and expression: Emotional experience, behavior, and physiology in Chinese Americans and Mexican Americans. Emotion, 5(2), 154–165. Then, address the following: Summarize the purpose and the overall findings of the article. The summary should be in 50 words. Note: The data analysis section might be difficult to follow. Just scan this section and concentrate on the discussion and conclusion. What role do emotions play in human interactions? Briefly state what previous studies have discovered about the universality of facial expressions. Use another source to look up the purpose of the autonomic nervous system and ethnography. Write the definitions for both using 1–2 sentences for each. What do ethnographic accounts tell us about how Chinese people view emotions? What does somaticizing mean? Provide an example. Are the previous findings on the emotional expression of the Mexican (Latino) culture consistent? Summarize the findings. Look at the participants’ subsection of the Method section. Who were the participants? Do you think that the participants were appropriate for the study? Explain your answer. Summarize the procedure used in the study in 2–3 sentences. Read the discussion section. Which two hypotheses were supported by the results? Review the physiological part of the discussion section. What is the rationale given for the physiological findings? State one limitation of the study. What was the most interesting thing you learned from the article

Cultures of Moderation and Expression: Emotional Experience, Behavior,
and Physiology in Chinese Americans and Mexican Americans
Jose´ A. Soto, Robert W. Levenson, and Rachel Ebling
University of California, Berkeley
Ethnographic accounts suggest that emotions are moderated in Chinese cultures and expressed openly in
Mexican cultures. The authors tested this notion by comparing subjective, behavioral, and physiological
aspects of emotional responses to 3 (warned, unwarned, instructed to inhibit responding) aversive
acoustic startle stimuli in 95 Chinese Americans and 64 Mexican Americans. Subjective reports were
consistent with ethnographic accounts; Chinese Americans reported experiencing significantly less
emotion than Mexican Americans across all 3 startle conditions. Evidence from a nonemotional task
suggested that these differences were not artifacts of cultural differences in the use of rating scales. Few
cultural differences were found in emotional behavior or physiology, suggesting that these aspects of
emotion are less susceptible to cultural influence.
Emotions are the cornerstones of our social worlds, affecting our
interactions with others in countless ways. Emotions play a vital
role in maintaining social order, including helping regulate social
distance (drawing us toward some people and away from others),
announcing our intentions, and influencing the behavior of others.
Because emotions can be both helpful and disruptive, cultures
develop norms about what constitutes desirable and undesirable
emotional behavior. On the basis of differences in social conditions,
traditions, and ideals, these norms can show considerable
variation across cultures.
Psychologists who first studied culture and emotion were stimulated
by the question of whether the facial expressions for certain
“basic” emotions are universal. Despite prevailing socialconstructivist
views (e.g., Mead, 1975), the bulk of the empirical
evidence indicated that for emotions such as anger, disgust, fear,
happiness, sadness, and surprise, there is cross-cultural consistency
in the associated facial displays (e.g., Ekman & Friesen, 1971,
1986; Ekman, Sorenson, & Friesen, 1969; Izard, 1971). There has
been far less research on cross-cultural consistencies in other
aspects of emotion; however, our research group has reported
evidence for cross-cultural similarities in autonomic nervous system
responding (Levenson, Ekman, Heider, & Friesen, 1992).
Against the backdrop of these reports of similarities in facial and
autonomic manifestations of emotion, there are many studies that
have found cultural differences in the events that elicit emotions,
the ways emotions are labeled and understood, how individuals
cope with their emotions, and beliefs about how and when emotions
should be expressed and felt (e.g., Ekman, 1972; Hochschild,
1979; Mesquita & Frijda, 1992; Shweder, 1993).
For the present study, we have chosen two cultures that are
thought to differ quite dramatically in their views concerning
emotion. Chinese culture has been portrayed in ethnographies as
valuing emotional control and moderation (Klineberg, 1938; Potter,
1988; Wu & Tseng, 1985). Mexican culture, in contrast, has
been portrayed as valuing the free and open expression of emotion
(Garza, 1978; Guerra, 1970; Ramirez & Castan˜eda, 1974). We
believe that these kinds of ethnographic notions of emotion are
fertile sources of hypotheses that can guide laboratory research.
However, we take particular note of two aspects of these portrayals.
First, they focus primarily on cultural beliefs about emotion
(i.e., metaemotion)—it remains to be seen whether these beliefs
translate into measurable differences in emotional response under
controlled conditions. Second, they tend to deal with emotion at a
fairly general, nonspecific level—it remains to be seen whether
cultural differences are consistent across three key aspects of
emotion (i.e., facial expressions, reports of subjective experience,
and autonomic nervous system response).
Ethnographic and Empirical Studies
Chinese Culture and Emotion
According to ethnographic accounts, members of Chinese culture
view emotions as dangerous, value emotional moderation, and
emphasize social harmony over individual expression (Klineberg,
1938). Wu and Tseng (1985) noted that in traditional Chinese
medicine, extreme emotions are thought to cause illness. Given
these observations, we would expect to find evidence of moderate
emotional behavior or even emotional suppression on the part of
the Chinese. In a study of psychiatric disorders, Kleinman (1986)
concluded that Chinese culture was indeed characterized by emotional
restraint. Related to this is the notion that the Chinese
Jose´ A. Soto, Robert W. Levenson, and Rachel Ebling, Department of
Psychology, University of California, Berkeley.
This research was supported by National Institute on Aging Grant
R37-AG17766 and by National Institute on Mental Health Grant MH39895
awarded to Robert W. Levenson. We acknowledge the following researchers
for their invaluable input in helping us determine our cultural eligibility
criteria: Karl Heider, Nolan Zane, Yu-Wen Ying, Joe Martinez, and Claude
Steele.
Correspondence concerning this article should be addressed to Jose´ A.
Soto, who is now at the Department of Psychology, Pennsylvania State
University, University Park Campus, 417 Moore Building, University
Park, PA 16802. E-mail: josesoto@psu.edu
Emotion Copyright 2005 by the American Psychological Association
2005, Vol. 5, No. 2, 154–165 1528-3542/05/$12.00 DOI: 10.1037/1528-3542.5.2.154
154
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
“somaticize” their emotions. In this view, rather than expressing/
experiencing emotions verbally and behaviorally, the Chinese
manifest emotions as physical or bodily symptoms related to
illness. If true, this “somaticizing” of emotions may provide a
culturally condoned outlet for emotional expression (e.g., complaining
of a stomach ache when angry) when other forms of
expressions may be considered inappropriate (e.g., yelling at your
employer).
A somewhat different understanding of Chinese emotions, proposed
by Potter (1988), is that emotions lack social significance in
the collectivistic Chinese culture and thus are less relevant than
they are in the individualistic American culture. The underlying
assumption here is that emotions are more disruptive to the social
harmony than they are helpful. In this view, emotional expression
in Chinese culture is not so much discouraged or suppressed, but
rather it is ignored. Potter described attitudes toward emotional
displays in China as that of indifference, as evidenced by a
statement by one of her cultural informants: “How I feel does not
matter!” Although the underlying dynamic in Potter’s model is
quite different from that of Klineberg, Wu and Tseng, and others,
both models may lead to the same result: low levels of emotional
experience and sparse emotional displays.
Empirical tests of these notions based on measurement of actual
emotional responding (e.g., precisely measuring participants’ subjective,
behavioral, and physiological responses to carefully specified
emotional stimuli under controlled conditions) have been
rare. Tsai and Levenson (1997) found support for emotional moderation
in Chinese culture in a study comparing Chinese American
and European American dating couples who had unrehearsed
conversations about their relationship. Chinese American couples
reported fewer periods of positive emotion and showed less variability
in their reported emotional experience than European
American couples. Other studies of emotion in Chinese culture
have not directly measured emotional responding, but rather have
examined qualities of emotional judgments. Ekman et al. (1987)
found that college students from Asian countries (Hong Kong,
Japan, Sumatra) attributed less emotional intensity than students
from non-Asian countries to photographs of facial expressions
posed by Caucasians. Matsumoto (1993) reported a similar pattern
of lower intensity ratings for Asian Americans using photographs
of facial expressions posed by both Caucasians and Japanese.
Consistent with ethnographic views reviewed earlier, Matsumoto
also found that Asian participants rated the emotional expressions
as being less “appropriate” under various social situations than did
Caucasian participants.
Mexican Culture and Emotion
Similar to Chinese culture, Mexican culture is believed to be
collectivistic. Ethnographic accounts of Mexican culture describe
affiliation, group cohesion, and maintaining strong interpersonal
relationships as highly valued (Carrillo, 1982; see Eisenberg, 1999,
for additional references). However, beyond these similarities are
a number of important cultural differences. Carrillo (1982), for
example, noted that interpersonal relationships in Mexican culture
are characterized by high levels of affection. In Mexican culture,
affect is more openly accepted and more highly valued than in
Anglo culture (Garza, 1978; Guerra, 1970; Ramirez & Castan˜eda,
1974). The sentiment is best captured by Murillo’s (1976) statement
that in Mexican American culture, “It is through . . . an
ability to experience, in response to environment, emotional feelings
and to express these to one another and share them that one
experiences the greatest rewards and satisfactions in life” (p. 100).
Related to this emotional openness, Triandis and his colleagues
(Triandis, Marı´n, Lisansky, & Betancourt, 1984) reported a strong
normative pressure in Latin American cultures to behave in a
positive manner toward others. Triandis’s theory is consistent with
a Mexican philosophy of being not only both expressive and
emotional as a group but also collectivistic and wanting to promote
group cohesion. Inherent in this view is the belief that emotions
promote group cohesion, which contrasts rather dramatically with
the Chinese view that emotions are socially dangerous. Thus,
according to these ethnographic observations, we would expect
Mexican individuals to be more emotionally expressive than Chinese
individuals, particularly when it comes to positive emotions.
As was the case with Chinese culture, empirical studies of actual
emotional responding in Mexican culture have been rare. The few
existing studies provide mixed support for ethnographic observations
of greater emotional expression in Mexican culture. In support
of these notions, Garza (1978) found that Chicanos performed
better than Caucasians on a learning task when they learned
material they liked, concluding that this indicated greater importance
assigned to affective meaning in Chicano culture. However,
Eisenberg (1999) found no cultural differences between Mexican
American and Anglo American mother– child dyads in how much
they talked about emotions. Finally, contradicting ethnographic
accounts, Deffenbacher and Swaim (1999) found Mexican American
middle and high school students to be less likely to express
their anger via verbally or physically aggressive behavior than
Caucasian students.
Summary and Implications for the Present Study
Despite the fact that Mexican Americans and Chinese Americans
constitute the two fastest growing ethnic groups in the United
States (Buriel & De Ment, 1997), they have not been well studied
in the empirical literature on emotion. When they have been
studied, Chinese or Mexican participants have typically been compared
with European American samples (in which the ethnographic
contrasts are not as striking) rather than with each other.
Moreover, most studies have focused on beliefs or judgments
about emotions rather than on actual emotional responding and
have not distinguished between subjective, behavioral, and physiological
aspects of emotion. In the present study, we compared
emotional responding with standardized stimuli in Mexican and
Chinese participants using a multimethod approach that allows the
assessment of subjective, facial expressive and autonomic indicators
of emotion. We believe this approach provides a good test of
hypotheses derived from ethnographic accounts of emotion and
extends these notions by providing a more differentiated measurement
of emotion.
Methodological Considerations: Culture and Emotion
There are a number of critical sampling issues that need to be
addressed in empirical studies of culture. Many studies have used
CULTURES OF MODERATION AND EXPRESSION 155
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
nationality, self-identified race, or self-identified ethnicity as a
proxy for culture. Without due care, cultural variations within
groups can dwarf purported between-group cultural variations (for
a discussion of some of these issues, see Leung, 1996; Matsumoto,
1993). In laboratory studies conducted in the United States, college
students typically serve as the representatives of the cultures of
interest. These college students are often bicultural and, thus, differ
greatly in their exposure and adherence to various cultural traditions.
To us, a comparison between Chinese Americans and Mexican
Americans who have relatively little exposure and adherence
to Chinese and Mexican cultural traditions, respectively, makes
little sense in a study of cultural influences on emotional responding.
Given that the participants in the present study were college
students living in the United States, we adopted a careful selection
procedure to try to ensure that they had significant exposure to the
cultural traditions of interest. Because we can never eliminate all
within-group variation in cultural adherence, we also measured
various aspects of acculturation so that these could be used in our
analyses.
There are also a number of critical issues related to the measurement
of emotion. It is now well accepted that emotion encompasses
subjective experience, behavioral expression, and physiological
arousal (Lang, 1979). Although it could be argued that
culture influences all aspects of emotion equally, we believe that
culture influences these different aspects of emotion in different
ways. For example, cultures may be more concerned with the more
socially visible aspects of emotion (e.g., what we say we feel, what
we show on our faces) than the socially invisible aspects (e.g.,
subjective experience that is not verbalized, changes in the viscera).
Additionally, aspects of emotional responding differ in how
amenable they are to voluntary control. Thus, carrying out a
cultural mandate to regulate emotion in a particular way may be
easier for certain aspects of emotion (e.g., what we say we are
feeling) than for others (e.g., whether our blood pressure is rising).
Finally, the conditions under which emotions arise vary greatly
in terms of their context (e.g., interacting with others in public
settings vs. reading a book alone) and the complexity of the
appraisals involved (e.g., decoding the thoughts and feelings of
others vs. reacting to a sudden loud noise). In our previous work,
we have examined the impact of emotion in situations requiring
relatively complex appraisals in both interpersonal contexts (dyadic
interaction in dating couples; Tsai & Levenson, 1997) and
private contexts (watching films; Tsai, Levenson, & Carstensen,
2000; making voluntary facial expressions; Levenson et al., 1992).
In the present study, we used a situation involving relatively
simple appraisals in relatively private contexts (hearing sudden
loud noises), hoping to determine whether the impact of culture on
emotional responding extends even to these types of emotional
situations.
Hypotheses
Hypothesis 1: Self-report of emotional experience. Chinese
Americans will report experiencing less positive and less
negative emotion than Mexican Americans in response to an
acoustic startle.
Hypothesis 2: Self-report of emotional control. Chinese
Americans will report trying to control their emotions more
than Mexican Americans in response to an acoustic startle.
Hypothesis 3: Expressive behavior. Chinese Americans will
show less emotional behavior (facial expressions, gestures,
body movements) than Mexican Americans in response to an
acoustic startle.
Rationale (Hypotheses 2–3)
Ethnographic accounts suggest that Chinese culture endorses
emotional moderation and Mexican culture endorses open emotional
expression. Differences in emotional responding reflecting
cultural beliefs are most likely to be found in the most voluntarily
controllable (self-reported experience of emotion) and most socially
visible (emotional behavior) aspects of emotion.
Hypothesis 4: Autonomic nervous system. Chinese Americans
and Mexican Americans will show similar levels of physiological
reactivity to an acoustic startle.
Rationale (Hypothesis 4)
Physiological responding is under less volitional control than
either self-report or behavior. Cultural beliefs place greater emphasis
on regulating emotional experience and expressive behavior
than on physiological responding.
Hypothesis 5: Acculturation. Emotional differences between
ethnic groups will be greatest among those participants who
are most strongly identified with their culture of origin.
Rationale (Hypothesis 5)
Differences in emotional responding between ethnic groups
result from differences in cultural beliefs and attitudes toward
emotion. Those participants with the strongest identification with
their culture of origin should be most strongly influenced by these
beliefs.
Method
Participants
Participants were 159 Chinese American and Mexican American college
students. Of the participants, 95 were Chinese American (45 men and 50
women), and 64 were Mexican American (31 men and 33 women). Of the
Chinese American students, 44 were born in the United States, and 51 were
foreign-born. Fifty of the Mexican American students were born in the
United States, and 14 were foreign-born. The mean age for the sample was
20.04 (SD 1.95). Chinese American and Mexican American groups did
not differ significantly in the proportion of men and women, 2(1, N
159) .018, ns, or mean level of education, t(138)1.36, ns. Mexican
Americans were significantly older than Chinese Americans (20.7 years vs.
19.6 years), t(84.71) 3.45, p .001, and were also more likely than
Chinese Americans to come from lower income or lower-middle-income
homes (70.4% vs. 24.1%), 2(4, N 141) 32.1, p .001.
Participants were recruited via newspapers, radio station announcements,
and flyers posted around several campuses in the San Francisco Bay
Area. Interested individuals were screened over the telephone for eligibil-
156 SOTO, LEVENSON, AND EBLING
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
ity. Those who qualified were mailed a package that included questionnaires
on health, life-stress, personality, acculturation, ethnic identity, and
demographics. Upon completion of the questionnaire package, participants
were scheduled to come to the University of California, Berkeley campus
for a laboratory session. Participants were paid $15 to complete the
questionnaire booklet and $35 for the laboratory session.
Eligibility. Strict inclusion criteria were adopted, especially for ethnic
and cultural background. These criteria aimed to equalize participants
across age, level of education, socioeconomic status, and degree of exposure
to American culture and to the participant’s culture of origin. Specifically,
all participants were required to be between the ages of 18 and 30
years, to have received a high school diploma from a school in the United
States, and to be enrolled currently as full-time college students. Participants
also had to meet specific ethnicity criteria (see below).
Culture/ethnicity criteria for both groups. We developed ethnicity
criteria in consultation with researchers who had special expertise relevant
to these ethnic groups. Participants’ parents and all four of their grandparents
(two sets of grandparents) had to be of the same ethnicity (thus,
biracial or multiracial participants were excluded). At least 50% of participants’
close friends during childhood or adolescence and at least 10% of
their neighborhood during childhood or adolescence had to comprise
members of the same ethnic group. All participants were required to be
fluent in English.
Additional criteria for Chinese Americans. Chinese Americans had to
be born and raised either in the United States, China, Taiwan, or Hong
Kong. Both parents and all four of their grandparents had to have been born
and raised in China, Taiwan, or Hong Kong. In addition, all Chinese
American participants needed to have attained at least moderate proficiency
in speaking and understanding Chinese (Mandarin, Cantonese,
Taiwanese, or any other Chinese dialect).
Additional criteria for Mexican Americans. Mexican Americans had to
be born and raised either in the United States or Mexico, and both of their
parents and all four of their grandparents had to have been born and raised
in Mexico. Their current religion or religion while growing up had to be
Catholic. They also must have attained at least moderate proficiency in
speaking and understanding Spanish.
Apparatus and Materials
Self-report measures. As part of the questionnaire package, participants
completed a six-page General Ethnicity Questionnaire (Chinese or
Mexican version as appropriate) designed to measure their orientation
toward their culture of origin. Both the Chinese (Tsai, Ying, & Lee, 2000)
and Mexican versions were adapted from several existing measures of
acculturation (Cuellar, Harris, & Jasso, 1980; Mendoza, 1989; Sodowsky,
Lai, & Plake, 1991; Suinn, Rickard-Figueroa, Lew, & Vigil, 1987; Szapocznik,
Scopetta, Kurtines, & Aranalde, 1978). Each version consisted of
75 items asking participants to rate their exposure to or preference for their
culture of origins’ traditions, people, or language on a scale ranging from
1 (extremely) to 5 (not at all). A total acculturation score was derived by
summing all of the items. In each case, lower scores indicated stronger
orientation to the culture of origin.
To measure emotional experience after each trial, participants were
asked to complete a 9-point Likert-type scale ranging from 0 (not at all) to
8 (very much) to rate how strongly they felt each of the following 13
specific emotions: amusement, anxiety, contempt, contentment, disgust,
embarrassment, fear, anger, happiness, interest, relief, sadness, and surprise.
Participants used the same 9-point scale to indicate the degree to
which they tried to control their body movement, facial expression, verbal
expression, and physiological reactions during the trial.
Audiovisual. All instructions were presented visually on a 13-in. (33-
cm) TV monitor placed directly in front of the participant at a distance of
four feet. A remote-controlled, high-resolution video camera, partially
concealed behind darkened glass, was used to obtain frontal views of each
participant’s face and upper torso. Participants were informed prior to the
start of the experiment that they would be videotaped.
Startle stimuli. The acoustic startle was a 115 dB burst of white noise
(100 ms in duration) administered through two large loudspeakers located
behind the participant’s head. Participants typically liken this noise to a
gunshot. It should be noted that this kind of startle stimulus is highly salient
and noxious, much louder than the kind of background “startle probe”
stimulus typically used in startle probe studies that use mild startle stimuli
to assess stimulus valence and underlying mood (e.g., Vrana & Lang,
1990).
Physiology. Seven physiological functions were measured using a
system consisting of a Grass Model 7 12-channel polygraph and a microcomputer
(Grass Instruments, Quincy, MA). (a) Cardiac interbeat interval,
the electrocardiogram, was measured using Beckman miniature electrodes
(Beckman Instruments, Fullerton. CA), with Redux paste placed on opposite
sides of the chest. The interval between successive R-waves was
measured in milliseconds. (b) Pulse transmission time to the finger was
measured using a UFI photoplethysmograph (UFI Instruments, Morro Bay,
CA) attached to the top phalange of the second finger of the nondominant
hand. Transmission time was measured as the time interval between the
R-wave of the electrocardiogram and the upstroke of the peripheral pulse
at the finger. (c) Finger pulse amplitude, the trough-to-peak amplitude of
the finger pulse, was also measured from the UFI photoplethysmograph.
(d) Ear pulse transmission time was measured using a UFI photoplethysmograph
attached to the right ear lobe. Transmission time was measured
between the R-wave of the electrocardiogram and the upstroke of the
peripheral pulse at the ear. (e) Mean arterial blood pressure was measured
using an Ohmeda 2300 Finapres blood pressure monitor (Ohmeda Medical,
Laurel, MD). This monitor measures blood pressure on every heartbeat
using a cuff attached to the middle phalange of the fourth finger of the
nondominant hand. (f) Skin conductance level, the level of sweat gland
activity on the surface of the hand, was measured by passing a constant
voltage between Beckman regular-size electrodes attached to the palmar
surface of the lower phalanges of the first and second fingers on the
nondominant hand. The electrodes used an electrolyte paste consisting of
sodium chloride in Unibase. (g) Finger temperature was measured using a
Yellow Springs Instruments thermistor (Yellow Springs, OH) attached to
the palmar surface of the first phalange of the fourth finger of the nondominant
hand. Respiratory and somatic activity was also measured, but
those data were not examined for the present study.
Using locally developed software, the computer determined second-bysecond
averages for each physiological measure. Collectively, these seven
measures provide a broad sampling of autonomic nervous system responses
spanning cardiac, vascular, and electrodermal subsystems. Moreover,
in previous research (e.g., Gross & Levenson, 1993, 1997), we have
found these systems to be responsive to the kinds of emotional stimuli used
in the present study.
Procedure
Participants came into the laboratory for a 2-hr experimental session.
Previous studies have suggested that the ethnicity of an experimenter can
influence physiological responses to laboratory tasks (Anderson, 1989;
Murphy, Alpert, Willey, & Somes, 1988). Thus, participants only had
contact with a research assistant of their own gender and ethnicity during
the experimental session. Participants were seated in a comfortable chair
facing the video monitor described above. After signing a consent form,
they completed a pretrial inventory that gauged their current emotional
state (see below). After completing these brief measures, the research
assistant attached the physiological sensors and explained their function.
The experiment consisted of five trials. For three of the five trials,
participants were exposed to an acoustic startle under three different
instructional conditions (see below for details) in a counterbalanced order.
CULTURES OF MODERATION AND EXPRESSION 157
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This acoustic startle is of sufficient magnitude (115 dB) to produce a strong
defensive response (Landis & Hunt, 1939). Because this response is
reflexlike, and thus may be highly consistent across individuals, it was
important to include experimental conditions that would be conducive to
the emergence of cultural differences. Thus, we presented the startle under
three different instructional conditions. In the unanticipated condition,
participants were not warned of the impending startle, maximizing the
probability that the reflexive response would be observed. In the anticipated
condition, participants were told the startle would occur after a
countdown period. This anticipated condition provided participants with
time to reappraise the situation, and thus possibly modify their emotional
response in culturally appropriate ways. In the inhibited condition, participants
were also warned about the startle and were additionally asked to
suppress their observable emotional response. This inhibited condition
provided participants with time for reappraisal and provided an explicit
instruction to modify the emotional response, both amenable to influence
by cultural beliefs and practices.
For each participant, the three startle trials were separated by two
nonstartle trials: a handgrip task (always the second trial and described
below) and a mental arithmetic task (always the fourth trial). The handgrip
task (described below) was included to provide a nonemotional task that
produces a reliable physiological response. Thus, this task provides a
control for individual differences in nonemotional physiological reactivity
that can be used to sharpen analyses of physiological reactivity during
emotional tasks.
Before each trial began, detailed instructions for the trial were presented
on the video monitor. After the overview of the trial was presented,
participants were instructed via the monitor to, “Begin to relax now. Empty
your mind of all thoughts, memories and emotions.” These instructions
remained on the screen for 2 min, during which the participant rested
quietly. Each of the five trials began immediately after the 2-min relaxation
period and was followed by a 2-min rest period during which no instructions
were given. These two relaxation periods served as the pre- and
posttrial baselines, respectively. After the posttrial relaxation period, participants
provided ratings of how strongly they had felt each of 13 specific
emotions during the trial (e.g., “How sad did you feel when you heard the
loud noise”). A detailed description of the five trials follows.
Unanticipated startle. In the unanticipated startle trial, participants
were told, “In this part of the experiment I want you to relax for a while.
I’ll turn on some soft white noise now to help block out any other sounds
coming into the room.” They were then asked to confirm that they heard
the soft noise before moving on. Once they confirmed the soft noise, they
were instructed to “Please continue to look at the ‘x’ on the screen while
you are relaxing.” The startle occurred unexpectedly at the end of the 2-min
pretrial relaxation period. If the unanticipated trial was not the last startle
trial, participants were instructed at trial’s end that they would be told
exactly when any future startles would occur.
Anticipated startle. In the anticipated startle trial, participants were
instructed as follows:
In this part of the experiment, you will hear a loud noise. You will
know exactly when the loud noise will occur. You will see a countdown
from 10 to 1 on the video screen. When you see ‘1’ – the loud
noise will happen. Before beginning the countdown, I want you to
relax. I will not start the countdown until the rest period is over.
After the pretrial relaxation period, the participants were reminded of the
countdown, and then the countdown began, lasting for 20 s, with each
number presented on the screen for 2 s. After the “1” came on, it remained
on the screen for the duration of the startle and throughout the 2-min
posttrial relaxation period.
Inhibited startle. The inhibited startle trial was similar to the anticipated
startle condition in that participants were informed that the startle
would occur after a countdown. In the inhibited startle condition, however,
participants also received additional instructions prior to the countdown:
We want to see how well you can keep from showing any emotional
response when you hear the noise. Try not to feel anything, and try not
to have a physiological reaction. Also, see if you can act so that
someone seeing the video with the sound off won’t know that anything
has happened. Try not to show any visible signs or feel anything
before, during, or after the loud noise occurs. Try to look relaxed all
the way through. See if you can fool the person who will be studying
this video.
After the pretrial relaxation period, participants were reminded of the
countdown and to try and not show any reactions when the startle occurs.
After the “1” came on, it remained on the screen for the duration of the
startle and throughout the 2-min posttrial relaxation period.
Handgrip. Prior to beginning the experiment, the subject was asked to
squeeze a Jamar Hand Dynamometer (Lafayette Instruments, Lafayette,
IN) three times using maximum strength. A value computed as one half of
the average of these three attempts was used as the target grip strength for
that participant. During the actual handgrip trial, the target score was
written on a post-it note and taped to the monitor stand in front of the
participant. The participant was informed via the monitor that “in this part
of the experiment, you will be engaging in a physical task,” and they were
instructed to pick up the handgrip device and confirm the target score
aloud. The video instructions then informed them that a tone would signal
when they should start squeezing the device so that the meter pointer on the
handgrip matched the target score. They were also instructed to count aloud
by ones while they were squeezing to avoid breath holding. Once the
participants confirmed that they understood the instructions, the 2-min
pretrial relaxation period began. After the 2-min pretrial baseline, the
screen informed them that they should,
start squeezing so that the needle on the dial stays on the number
posted in front of you. Count out loud by 1’s, starting with the No. 1.
Keep squeezing on the device until you hear the second tone and see
the word “STOP” on the screen.
After one minute they were told to put the device down and relax for two
minutes.
Data Reduction
Self-report data. To reduce the number of dependent measures, selfreport
data were averaged into two composite measures: positive emotion
(amusement, content, happiness, and relief) and negative emotion (anger,
anxiety, contempt, disgust, embarrassment, fear, and sadness).
Behavioral data. A team of undergraduate research assistants used a
modified version of the Emotional Expressive Behavior coding system
(Gross & Levenson, 1993) to measure startle behavior and emotional
expressive behavior during the 1-s period containing the startle stimulus
and during the following 5 s. Judges were required to make binary
evaluations (present/not present) about the occurrence of the following
facial and upper body movements: hard eye closure, eye tightening, eye
widening, brow lowering, brow raised, lip corners down, lip corners up,
lip stretch, neck stretch, head jerk, shoulder raised, shuddering, forward
lunge, and torso raise. In addition, judges provided Likert ratings
regarding the intensity/duration of any manifestations of the following
emotions and emotionally relevant behaviors: anger, anxiety, disgust,
confusion, contempt, interest, embarrassment, fear, happiness, sadness,
surprise, crying, laughter, and overall pleasantness/unpleasantness. Intensity/
duration ratings used a 0 – 6 scale where 0 none, 1 slight
and short, 2 slight and long, 3 moderate and short, 4 moderate
and long, 5 strong and short, and 6 strong and long. Finally,
158 SOTO, LEVENSON, AND EBLING
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
judges counted the number of times that enjoyment smiles and nonenjoyment
smiles (Ekman & Friesen, 1982) were displayed by the participants.
To establish reliability, all participants were coded by two
judges. The average of the two judges was used in all analyses. For
binary codes, the percentage of agreement for pairs of judges was 86%.
For Likert ratings, reliability was computed by correlating the codes
from each pair of judges for each participant. The mean correlation
among judges across the entire sample was .83.
To reduce the number of dependent measures, we computed four composite
scores by first standardizing all behavioral codes and then aggregating
related behaviors. A positive behavior composite score was created
by combining lip corners up, pleasantness/unpleasantness, happiness,
laughter, interest, enjoyment smiles, and nonenjoyment smiles. A negative
behavior composite score consisted of eye tightening, brow lowering, lip
corners down, shuddering, pleasantness/unpleasantness, anger, anxiety,
confusion, contempt, crying, disgust, embarrassment, fear, and sadness. On
the basis of descriptions of the startle response provided by Landis and
Hunt (1939) and Ekman, Friesen, and Simons (1985), we created a composite
score for startle behavior that included hard eye closure, lip stretch,
neck stretch, head jerk, shoulder raise, forward lunge, and torso raise.
Because participants often report being surprised by the startle stimulus, we
created a composite score for surprise behavior that consisted of codes for
surprise, eye widening and brow raising.
Physiological data. For our primary physiological analyses, we used
composite scores that were calculated as the average of the standardized
scores (corrected so that positive values always indicated greater arousal)
for cardiac interbeat interval, pulse transmission time to the finger and ear,
skin temperature, pulse amplitude, skin conductance, and mean arterial
pressure. This kind of composite score has several advantages, including
(a) reducing the number of dependent measures, thus helping control for
Type I error; (b) providing a measure that is sensitive to variation across
participants in which response systems are maximally active (i.e., individual
response stereotypies); and (c) increasing reliability because of the
greater stability of aggregated scores. Moreover, in previous studies using
similar emotional stimuli, we have used these kinds of composite measures
with some success (e.g., Fredrickson & Levenson, 1998; Gross & Levenson,
1993).
For each of the three startle conditions, second-by-second physiological
averages were reduced to mean values representing five nonoverlapping
time windows: (a) 2-min pretrial baseline period, (b) 20-s prestartle countdown
period (anticipated and inhibited startles only), (c) first 5-s reactivity
period (including the 100 ms while the startle stimulus was present), (d)
next 10-s reactivity period (intended to capture responses in slow-reacting
signals such as finger temperature), (e) next 15-s reactivity period (starting
after the first two reactivity periods, when many signals begin returning to
baseline levels). These means were then combined into a composite score
for each time period, and difference scores were then computed for each
trial by subtracting the pretrial baseline composite score from composite
scores for each of the three reactivity time windows (Periods c, d, and e
above).
For the handgrip trial, the second-by-second physiological averages
were reduced to mean values representing two nonoverlapping time windows:
(a) the 2-min pretrial baseline period and (b) the 1-min period when
participants were squeezing the handgrip device. These means were then
combined into a composite score for each time window, and reactivity
scores were calculated by subtracting the pretrial baseline composite from
the handgrip period composite.
Results
Results are presented for the self-report, behavioral, and
physiological variables separately. Within each type of variable,
results from analyses of ethnic differences will be presented
first, followed by those involving sex differences. Finally,
analyses of the impact of within-group differences in
acculturation are reported.
Self-Report
Emotion. To test Hypothesis 1 that Chinese Americans would
report experiencing less emotion than Mexican Americans, participants’
positive and negative self-reported emotion composite
scores were analyzed using a 2 2 3 (Ethnicity Sex Startle
Type) multivariate analysis of variance (MANOVA), with startle
type serving as a repeated measure (see Table 1 for effects of
startle type). Evidence for ethnic differences were found in a
significant main effect for ethnicity, F(2, 150) 3.59, p .05,
and a significant Startle Type Ethnicity interaction, F(4, 602)
3.02, p .05. To explicate these findings, we conducted separate
univariate repeated measures analyses of variance (ANOVAs) for
positive and negative emotion. Across the startles, Chinese Americans
reported experiencing significantly less positive emotion and
negative emotion than Mexican Americans, thus supporting Hypothesis
1 (see also Table 1). For positive emotions, the ethnic
differences varied as a function of the type of startle, Startle
Type Ethnicity, F(2, 302) 5.34, p .01. In response to the
unanticipated and inhibited startles, Chinese Americans and Mexican
Americans did not differ in the amount of positive emotion
reported. In response to the anticipated startle, however, Chinese
Americans reported significantly less positive emotion (M 1.26)
Table 1
Mean Comparisons for Self-Reported Emotional Experience by
Startle Type and Ethnicity
Emotion
composite score
Unanticipated Anticipated Inhibited
M SD M SD M SD
Startle type
Positive emotion 1.10a 0.10 1.64b 0.13 1.50b 0.10
Negative emotion 1.99a 0.12 1.77b 0.11 1.67b 0.10
Ethnicity
Chinese American
Mexican
American
M SD M SD
Positive emotion 1.22a 0.11 1.61b 0.14
Negative emotion 1.61a 0.12 2.01b 0.15
Note. Means in the same row that do not share the same subscript differ
at p .05. The main effect for startle type was significant in the overall
multivariate analysis of variance, F(4, 610) 8.93, p .001, and in the
analyses of variance for positive emotion, F(2, 306) 13.77, p .001, and
negative emotion, F(2, 293.6) 5.86, p .05. Participants reported less
positive emotion and more negative emotion to the unanticipated startles
compared with the anticipated and inhibited startles.
CULTURES OF MODERATION AND EXPRESSION 159
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
than Mexican Americans (M 1.99), t(113.9)12.75, p .01.
For negative emotions, there was no Startle Type Ethnicity
interaction.
An analysis of sex differences revealed no significant Sex or
Sex Ethnicity effects in the amount of emotional experience
reported.
One concern in analyzing these self-report data is the possibility
that found differences between Chinese Americans and Mexican
Americans in self-reported emotion could be because of systematic
differences in how the two groups used the 9-point rating scale.
For example, Chinese Americans may have tended to use the lower
end of the rating scale, confounding actual differences in emotional
experience with a response style or bias. If such a response
bias did indeed exist, then one would expect Chinese Americans to
provide lower emotion ratings than Mexican Americans even in
nonemotional tasks. The handgrip task represented this kind of
nonemotional task. However, we found no significant differences
between the two ethnic groups on this task, supporting the conclusion
that our finding that Chinese Americans reported experiencing
less emotion than Mexican Americans in response to the
startle stimuli represents a true cultural difference in emotional
experience rather than an artifact of response style.
Emotional control. We also compared the two ethnic groups
on the degree to which they reported controlling their emotional
response, as indicated by self-reported attempts to control their
facial expressions, body movements, verbal expressions, and physiological
reactions. Participants’ emotion control ratings were analyzed
using a 2 2 3 (Ethnicity Sex Startle Type)
MANOVA, with startle type treated as a repeated measure (see
Table 2 for effects of startle type). Relevant to ethnic differences,
there was a significant main effect for ethnicity, F(4, 145) 4.40,
p .01. The Startle Type Ethnicity interaction was not significant.
To further understand these findings, we conducted separate
univariate analyses for each emotion control variable (see Table 2).
Results revealed that Chinese Americans reported trying to control
their verbal expression, F(1, 150) 4.49, p .05; body movement,
F(1, 150) 5.35, p .05; and physiological reactions, F(1,
150) 13.79, p .001; less than Mexican Americans (see
Table 2).
Following the same logic used earlier for the self-reported
emotion, we used data from the nonemotional handgrip task to
check for response style differences. During the handgrip task,
Chinese Americans also reported trying to control their body
movement, t(157) 2.59, p .05, and physiological reactions,
t(156) 2.97, p .01, less than Mexican Americans. Thus,
Chinese Americans reported exerting less control of body movements
and physiological reactions than Mexican Americans to
both the emotional (startles) and nonemotional (handgrip) tasks.
This suggests that Chinese Americans might use the scales on the
control items differently than Mexican Americans. Given this
finding, we reanalyzed the data for self-reported control of verbal
expressions, body movement, and physiological reactions during
the emotional startle tasks using an analysis of covariance
(ANCOVA) to control for the degree of emotional control reported
on the nonemotional handgrip task. Controlling for response style
in this manner eliminated almost all of the found ethnic differences
in self-reported control during the startles. The only exception was
that Chinese Americans still reported controlling their physiological
reactions to the startle significantly less than Mexican Americans,
F(1, 148) 5.57, p .05. With or without this statistical
control, these findings do not support Hypothesis 2, which had
predicted greater reported emotional control for the Chinese Americans
than Mexican Americans.
In terms of sex differences in the amount of emotional control
reported in response to the startles, there was no significant main
effect for gender, but there was a significant Sex Ethnicity
interaction in the overall MANOVA, F(4, 145) 4.40, p .05.
Follow-up univariate ANOVAs revealed that the Sex Ethnicity
interaction approached significance when comparing participants’
attempts to control their physiological reactions, F(1, 148) 3.42,
p .07. Among Chinese Americans, men and women did not
differ significantly in the extent to which they reported controlling
their physiological reactions, F(1, 89) 0.11, ns. Among Mexican
Americans, however, men reported controlling their physiological
reactions significantly more (M 4.17) than women (M 3.2),
F(1, 59) 4.75, p .05.
Behavior
To test Hypothesis 3 that Chinese Americans would display less
emotional behavior than Mexican Americans, we compared the
1 Degrees of freedom reflect adjustments for violations of underlying
assumptions for a given statistical procedure.
Table 2
Means and Standard Deviations for Self-Reported Emotional
Control by Startle Type and Ethnicity
Emotion control score
Unanticipated Anticipated Inhibited
M SD M SD M SD
Startle type
Facial expression 1.70a 0.17 3.32b 0.22 5.65c 0.17
Verbal expression 1.93a 0.21 2.76a 0.24 4.20b 0.26
Body movement 1.82a 0.19 3.56b 0.22 5.71c 0.18
Physiological reactions 1.86a 0.19 2.90b 0.23 4.62c 0.22
Ethnicity
Chinese
American
Mexican
American
M SD M SD
Facial expression 3.38a 0.17 3.74b 0.20
Verbal expression 2.56a 0.23 3.35b 0.28
Body movement 3.37a 0.18 4.02b 0.22
Physiological reactions 2.56a 0.19 3.68b 0.23
Note. Means in the same row that do not share the same subscript differ
at p .05. The main effect for startle type was significant in the overall
multivariate analysis of variance, F(8, 594) 32.73, p .001, and in the
analyses of variance for attempting to control facial expressions, F(1.98,
296.7) 139.34, p .001; verbal expression, F(2, 300) 36.60, p
.001; body movement, F(2, 300) 133.06, p .001; and physiological
reactions, F(1.99, 298.4) 61.40, p .001. Consistent with the instructions,
participants exerted the greatest effort to control their emotional
response to the inhibited startle and the least control during the unanticipated
trial.
160 SOTO, LEVENSON, AND EBLING
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
two groups on the positive behavior, negative behavior, startle
behavior, and surprise behavior composites using a 2 2 3
(Ethnicity Sex Startle Type) MANOVA, with startle type
treated as a repeated measure. Neither the ethnicity nor Startle
Type Ethnicity interaction was significant. Thus, Hypothesis 3
was not supported.
An analysis of sex differences revealed a significant main effect
of sex, F(4, 144) 5.29, p .01. Follow-up univariate ANOVAs
revealed that women displayed significantly more negative behavior,
F(1, 147) 9.27, p .01; positive behavior, F(1, 147)
10.65, p .01; and startle behavior, F(1, 147) 12.69, p .01;
than men. The mean emotional behavior scores for men and
women were 1.212 and 1.800, respectively, for negative behavior,
2.03 and 1.15 for positive behavior, and 1.023 and 0.839
for startle behavior. There were no differences in the amount of
surprise behavior displayed by men and women. The Sex
Ethnicity interaction was not significant.
Physiology
Hypothesis 4 predicted that Chinese Americans and Mexican
Americans would show similar amounts of physiological reactivity
in response to the acoustic startles. In order to test this hypothesis,
we analyzed our physiological reactivity composite scores for each
of the three time windows described earlier using a 2 2 3
3 (Ethnicity Sex Startle Type Time) ANOVA, with startle
type and time treated as repeated measures. No significant differences
were found for ethnicity or any of the interactions of ethnicity
with the other factors. Although we cannot prove a null
hypothesis, these results are such that Hypothesis 4 cannot be
rejected.
In terms of sex differences in physiological response, there were
no main effects for sex, but there was a significant Ethnicity Sex
interaction, F(1, 109) 5.91, p .05. Among Chinese Americans,
men and women did not differ significantly in their autonomic
physiology across the three startles, F(1, 69) 0.25, ns.
Among Mexican Americans, however, men showed a smaller
autonomic response (M 0.03) than women (M 0.07), F(1,
40) 7.87, p .01.
Acculturation
Hypothesis 5 predicted that ethnic differences would be strongest
among participants who were most strongly oriented toward
their culture of origin. To test this hypothesis, we used a median
split to categorize each participant as high or low on his or her
respective acculturation measure scores. We then recomputed the
MANOVAs for each of the dependent measures (as above) using
this new dichotomous acculturation variable as an additional
between-subjects factor. We were interested in examining those
instances in which significant interactions between ethnicity and
acculturation emerged, suggesting the importance of considering
within-group differences.
There were two instances of a significant Ethnicity Acculturation
interaction. In terms of how much negative emotional
behavior was displayed, Chinese Americans who were most
strongly oriented to Chinese culture displayed less negative emotional
behavior (M0.70) than those who were less oriented to
Chinese culture (M 0.99), F(1, 142) 3.80, p .05. The
reverse pattern held for Mexican Americans—those who were
most strongly oriented to Mexican culture displayed more negative
emotional behavior (M 1.86) than those who were less oriented
to Mexican culture (M 0.33). Thus, for both groups, this
finding supported Hypothesis 5 insofar as the more closely oriented
participants were to their culture of origin, the more likely
they were to show the emotional behavior consistent with ethnographic
accounts (i.e., less emotion in Chinese Americans, more
emotion in Mexican Americans).
In terms of how much positive emotional behavior was displayed,
Chinese Americans who were most strongly oriented to
Chinese culture did not differ significantly from those who were
less oriented to Chinese culture. Among Mexican Americans,
those most strongly oriented toward Mexican culture showed
significantly less positive behavior (M 1.94) than those who
were less oriented to Mexican culture (M 2.10), F(1, 142)
4.10, p .05. This finding does not support Hypothesis 5, which
predicted more emotional behavior among Mexican Americans
who were strongly oriented to Mexican culture and less emotional
behavior for Chinese Americans who were most strongly identified
with Chinese culture.
Taking all of these results together, Hypothesis 5 received only
limited support. In the domains of self-reported emotion and
physiological reactions, acculturation did not affect our findings.
In the behavioral domain, orientation to the culture of origin had an
inconsistent influence, varying as a function of whether the emotional
behavior was negative or positive.
Discussion
This study tested hypotheses derived from ethnographic accounts
of cultural beliefs about emotion by examining subjective,
behavioral, and physiological aspects of actual emotional responding
to carefully controlled stimuli administered in a laboratory
setting. Whereas many of our previous studies of cultural differences
in emotional responding have used socially complex situations
(e.g., interaction between partners in intimate relationships;
Tsai & Levenson, 1997) or stimuli that require more complex
appraisals (e.g., emotional films; Tsai, Levenson, & Carstensen,
2000), here we used a much simpler emotional stimulus (acoustic
startle presented to participants sitting alone in a room). To the
extent that cultural beliefs about emotion primarily target social/
interpersonal behavior and work in large part through influencing
appraisals, the use of startle stimuli should provide an indication of
the extent to which cultural influences can be found even with
simple stimuli that produce responses thought to occupy the middle
ground between reflex and emotion (Ekman et al., 1985).
We found support for two of our hypotheses. First, consistent
with ethnographic accounts, Chinese Americans reported experiencing
less emotion than Mexican Americans. Support for this
hypothesis was quite strong insofar as it was found for reports of
both negative and positive emotions. Second, consistent with our
rationale that cultural differences are not likely to permeate the
more involuntary and less socially visible aspects of emotion (e.g.,
physiology), we could not reject the hypothesis that Chinese
Americans and Mexican Americans would show similar levels of
physiological response. Although we found some support for our
CULTURES OF MODERATION AND EXPRESSION 161
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
fifth hypothesis that differences between Chinese Americans and
Mexican Americans would be greater among those who are most
strongly oriented toward their culture of origin, this was limited
(only found for emotional behavior) and inconsistent (different in
direction for positive and negative emotional behavior). Our other
two hypotheses (greater reported emotional control and less emotional
behavior among Chinese Americans than Mexican Americans)
were not supported.
Self-Report of Emotional Experience
Our findings that Chinese Americans reported experiencing less
emotion than Mexican Americans were quite robust, showing
consistency across the three types of startles and across both
positive and negative emotion. These findings are consistent with
ethnographic notions and prior empirical investigations that portray
Chinese culture as emotionally moderate and Mexican culture
as embracing emotion (e.g., Ramirez & Castan˜eda, 1974; Tsai &
Levenson, 1997). Compared with previous studies, the present
study is unusual in that Chinese and Mexican cultures were directly
contrasted with each other. Further confidence in this finding
is gained by considering the findings using these same selfreport
scales on the nonemotional handgrip task. The fact that no
cultural differences were found on this nonemotional task suggests
that the finding of lower reported emotion for Chinese Americans
in response to the startles could not be attributed to a response bias
but rather is a difference that only emerged in an emotional
context. The finding that cultural differences in emotional report
are found even with such simple stimuli as sudden loud noises is
an impressive endorsement of the power that culture holds over
this aspect of emotional responding. The fact that these cultural
differences in self-report emerge absent a parallel cultural difference
in behavioral or physiological response provides further
indication that there is a special relationship between culture and
the way we report our emotional experiences.
Although most of our findings in the domain of self-reported
emotional experience were consistent across the three startles,
there was one notable exception. Mexican Americans reported
experiencing more positive emotion than negative emotion in the
anticipated startle, a finding that may be culturally informative.
The anticipated startle provides a 20-s warning period in which the
participant can prepare for the upcoming noxious event. Interestingly,
it was only in response to the anticipated startle that Mexican
Americans reported more positive emotion than Chinese
Americans. Thus, it may be that normative pressures in Latin
cultures to behave positively toward others (Triandis et al., 1984)
are most likely to emerge in situations that allow ample time for
the situation to be appraised fully and culturally appropriate responses
to be prepared.
Behavior
As we have noted, ethnographies typically do not distinguish
among emotion self-report, behavior, and physiology. Perhaps
because of this and the fact that the early empirical studies of
culture and emotion focused on recognizing emotional facial expressions,
it is commonly thought that cultural influences will be
found both in what people say they feel (e.g., Hochschild, 1979)
and what emotional behaviors they show (e.g., Ekman, 1972).
Although we expect that under the proper conditions, cultural
modulation of expressive behaviors can and does occur (e.g., the
finding of modulation of facial expressions to films in the presence
of an authority figure; Friesen, 1973), it is worth remembering that
facial behaviors can be quite automatic and difficult to rein in once
they start. The startle is a stimulus that is strong, sudden, and
short-lived. In situations such as these, cultural influences may be
much more likely to be found in retrospective self-reports than in
the much more immediate and less easily controlled expressive
behaviors. Whether this is unique to the startle or whether it
extends to other similarly acute and powerful stimuli is an important
question for future studies.
Self-Report of Emotional Control
Although we had predicted that Chinese Americans would report
greater attempts to control their emotions than Mexican
Americans, we found no evidence of this. Moreover, we encountered
one opposite finding insofar as Chinese Americans reported
trying to control their physiological reactions less than Mexican
Americans. In light of the consistent ethnographic emphasis on
emotional control in Chinese culture and our own strong findings
of less reported emotional experience among Chinese American
participants, the absence of parallel findings in reported emotional
control is quite striking. We can offer two possible, albeit admittedly
post hoc, explanations for this finding. First, it is possible that
there is something about the worldview of Chinese Americans that
led them to have less subjective emotional experience in this
particular situation and thus less need to exert effort to control their
reactions. Second, it is possible that once a cultural “feeling rule”
(e.g., Do not say you feel a lot of emotion when stressed) becomes
ingrained, it becomes relatively effortless. Viewed in this light, our
findings that Chinese Americans do not report exerting greater
effort to control their emotions may reflect the automaticity that
these cultural moderators of emotional report assume in Chinese
Americans by the time they reach early adulthood.
Physiology
As predicted, physiological responses were similar in amplitude
for the two ethnic groups. All analyses between Chinese Americans
and Mexican Americans revealed no significant differences in
physiological response to the startles. We continue to believe that
autonomic physiology may be an aspect of emotional responding
that is least susceptible to cultural influence. Of course, cultural
influences on physiological responding may be found in situations
that are more amenable to protracted appraisal than was the case
with the kinds of sudden, short-duration stimuli used in the present
study. However, this pattern of finding cultural differences in
self-report and cultural similarity in physiological response are
quite consistent with our previous studies of emotion and culture
using quite different stimuli. Tsai and Levenson (1997) found few
differences in emotional physiology between European American
and Chinese American dating couples during dyadic interaction
despite finding cultural differences in the amount and variability of
self-reported affect. Levenson et al.’s (1992) study comparing the
Minangkabau of West Sumatra with European Americans found
162 SOTO, LEVENSON, AND EBLING
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
that voluntary emotional facial configurations produced similar
patterns of autonomic activation in the two groups but large
differences in self-reported emotion. Tsai, Levenson, and
Carstensen (2000) also found no differences between Chinese
Americans and European Americans in their physiological response
to amusing and sad film clips. We expect that these
findings derive from the fact that autonomic reactions in emotion
are relatively more “hardwired” and more difficult to control
voluntarily than self-report and behavioral aspects of emotion.
These differences are further reinforced by greater cultural concerns
with the more socially visible aspects of emotion such as
language and behavior and less concern with the less socially
visible physiological responses.
Acculturation
Our assessment of acculturation provided some support for the
value of accounting for within-culture variation when attempting
to understand the basis of cultural influence. In the domain of
emotional behavior, we found that acculturation predicted variations
in the magnitude of response. However, this included one
instance (negative emotional behavior) in which greater orientation
to the culture of origin was associated with behavior that was
more consistent with the cultural ideal and one instance (positive
emotional behavior) in which the behavior was inconsistent with
the cultural norm. No effects resulting from acculturation were
found in the realm of self-report. Thus, these findings do not lend
themselves to a straightforward explanation of the influences of
acculturation. However, it is worth noting that our stringent selection
criteria likely worked to limit within-group variation. Thus, in
a sample with greater variation in acculturation levels, stronger and
more consistent acculturation effects might be found.
Ethnographic Versus Empirical Data
A primary goal of this study was to use a laboratory-based
methodology that assesses self-report, behavioral, and physiological
aspects of actual emotional responding to test hypotheses
derived from ethnographic accounts of emotion. We chose Chinese
Americans and Mexican Americans because ethnographic accounts
suggest that they are quite different in their approach to and
beliefs about emotion. Our finding that Chinese Americans report
experiencing less emotion in response to these stimuli than Mexican
Americans supports ethnographic notions about the emotional
values of these two groups. However, our behavioral and physiological
findings did not parallel the self-report findings, a pattern
of disjunction we have found in our other studies of emotion and
culture.
We interpret this pattern of findings across studies as indicating
that the ethnographies of these groups are more applicable to
self-reports of emotion than to emotional behavior or physiology.
This makes good sense given that ethnographies primarily document
cultural values and attitudes (Klineberg, 1938; Murillo,
1976) and not behavior or physiology. We have noted that selfreports
of emotional experience are likely to be much more susceptible
to volitional control than behavioral and physiological
aspects of emotion. Thus, it may be considerably easier to say that
one is not sad after the death of a relative than to prevent one’s lip
corners from turning downward and one’s eyes from tearing.
Importantly, these kinds of dissociations have also appeared in
some ethnographic accounts. For example, Potter (1988) noted that
displays of emotion are abundant among some Chinese, but such
displays are trivialized. Words, actions, and internal bodily
changes can have quite different social implications, giving rise to
divergences between the attitudes and values regarding each of
these aspects of emotional responding within cultures.
Sex Differences
Sex differences were found in the amount of emotional behavior
displayed by men and women. Our finding is consistent with that
in other literature, which has documented greater emotional behavior
among women when compared with men (Hall, Carter, &
Horgan, 2000; see also LaFrance & Banaji, 1992). The present
findings extend this gender difference to two cultural groups that
have generally not been well represented in studies of sex/gender
and emotion.
We also found a striking sex difference within one of our
cultural– ethnic groups but not the other. Among Mexican Americans,
men reported exerting more effort to control their physiological
reactions and showed less physiological arousal when
compared with Mexican American women. Among Chinese
Americans, no sex differences were found. It is important to note
that, paralleling these findings, sex differences are featured more
prominently in ethnographies of Mexican American emotional and
social norms than in those of Chinese Americans. For example,
although Mexican culture as a whole may embrace emotional
experience and expression, beliefs about machismo, male honor or
male superiority (Carrillo, 1982) may consider certain aspects of
emotional expression in men to be construed as weak. Interestingly,
it was in the physiological realm (both in reported control
and actual response) that Mexican American men showed this
tendency. Perhaps the physiological aspects of emotion, which we
have consistently found to be the least influenced by culture, are
left available for influence by sex/gender.
Limitations
A full account of the influence of culture on emotion will
require a complete sampling of cultures, elicitors, situations, emotions,
and response systems. The present study represents only a
partial sampling of these important variables. Although this study
has strength in its multimethod measurement of self-report, behavioral,
and physiological aspects of emotion, the acoustic startle
stimulus selected