Subjective social status moderates cortisol responses to social threat

Brain, Behavior, and Immunity 20 (2006) 410–419 www.elsevier.com/locate/ybrbi

Subjective social status moderates cortisol responses to social threat 夽 Tara L. Gruenewald a,¤, Margaret E. Kemeny b, Najib Aziz c a

Department of Medicine/Geriatrics, David GeVen School of Medicine, University of California, Los Angeles b Department of Psychiatry, University of California, San Francisco c Department of Pathology and Laboratory Medicine, David GeVen School of Medicine, University of California, Los Angeles Received 6 September 2005; received in revised form 2 November 2005; accepted 22 November 2005 Available online 18 January 2006

Abstract Research has demonstrated a robust relationship between social status, physiology and health in humans and animals. However, perceptions of social status within a speciWc social group have rarely been studied in this area and may provide additional relevant information. The current investigation examines subjective perceptions of social status as a moderator of cognitive, emotional and cortisol responses to stressor tasks characterized by social-evaluative threat or its absence. As part of a larger study, 81 college students living in a residential dormitory completed a measure of their subjective perceptions of their social status within their dormitory Xoor. They were randomly assigned to undergo a standard performance stressor task either with or without social evaluation. It was hypothesized that individuals who perceived that they were of low status within their dorm group would show greater increases in negative self-evaluative emotions (i.e., shame) and cognitions (low social self-esteem) and greater cortisol responses to the stressor under conditions of social-evaluative threat. Subjective social status moderated cortisol responses to the social-evaluative stressor, but in a direction opposite that hypothesized. Individuals who perceived themselves to be of high status showed sizable and signiWcant cortisol increases (both peak and recovery), while those who perceived themselves to be of low status did not mount a signiWcant cortisol response to the stressor. Both groups showed increased negative self-evaluative responses to the tasks. A discussion of the possible health implications of the robust cortisol responses of high status individuals and the hyporesponsive cortisol reactions of low status individuals is provided. © 2005 Elsevier Inc. All rights reserved. Keywords: Cortisol; Hypocortisolism; Acute stress; Social stress; Stress reactivity; Social evaluation; Social status; Shame; Social self-esteem

1. Introduction Social stressors are among the most provocative types of threat that humans and other animals face, with implications for behavior, physiology and health (see Buwalda et al., 2005; Fuchs and Flügge, 2003; Henry, 1992; Tamashiro et al., 2005). We have argued that threats to 夽 This research was supported by a National Research Service Award (MH12703) from the National Institute of Mental Health, an American Psychological Association Dissertation Research Award, a Society for the Psychological Study of Social Issues Grants-in-Aid Award, a UCLA Graduate Division Dissertation Research Award, and by grant AG10415 from the National Institute on Aging, UCLA Claude D. Pepper Older Americans Independence Center, supporting the Wrst author. * Corresponding author. Fax: +1 310 794 2199. E-mail address: [email protected] (T.L. Gruenewald).

0889-1591/$ - see front matter © 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.bbi.2005.11.005

the social self, which we conceptualize as threats to social esteem or status, can be potent elicitors of changes in cognitive, emotional and physiological states, that may aVect mental and physical well-being. In humans, social self threats encompass situations or factors that threaten one’s social esteem or status, including social rejection, ostracism, exclusion, scorn, or contexts in which one’s competencies, abilities, or characteristics upon which a positive social image is based are called into question (e.g., poor performance in social-evaluative contexts). Social self threats in humans may serve as an analog of threats to social status position or social alliances in other social animals (e.g., chimpanzees, baboons). In a theoretical conceptualization which we have entitled the Social Self Preservation Theory (Dickerson et al., 2004; Gruenewald et al., 2004b; Kemeny et al., 2004), we assert that threats

T.L. Gruenewald et al. / Brain, Behavior, and Immunity 20 (2006) 410–419

to social status or esteem elicit the experience of negative self-evaluative emotions (i.e., shame, humiliation) and cognitions (i.e., lowered social self-esteem), as well as activation of physiological systems that would support adaptive behavioral changes in this context, e.g., activation of the hypothalamic–pituitary–adrenal (HPA) axis and proinXammatory immune mediators (see Dickerson et al.). This theory is based on the premise that psychological and physiological responses to threat to the social self evolved and have been maintained because of the centrality of social esteem and status in reproductive and survival success (see Barkow, 1989; Gilbert, 1998; Price et al., 1994). Our assertion that threats to the social self should elicit the experience of shame and diminish social selfesteem is based upon theoretical perspectives which assert that these emotional and cognitive responses occur in situations in which individuals experience an actual or perceived loss of social value, status or esteem (Cooley, 1902; Gilbert, 1997; James, 1890; Leary, 1995). We focus on activation of the HPA axis and increases in proinXammatory immune mediators as physiological responses to social self threat based, in part, on animal research which documents that the level and functioning of biomarkers of these systems are associated with social status level and change in response to social status threats (e.g., Avitsur et al., 2001; Haller et al., 1996; Holst, 1997; Kollack-Walker et al., 1997; Kimura et al., 2000; Pich et al., 1993; Quan et al., 2001; Sapolsky, 1993; Skutella et al., 1994; Shively et al., 1997; Stefanski and Engler, 1998, 1999; Stefanski et al., 2001). In research with human participants, we have documented that laboratory stress paradigms which may threaten the social self are more likely to lead to increases in shame and decreases in social self-esteem (Gruenewald et al., 2004b), increases in cortisol (Dickerson and Kemeny, 2004; Gruenewald et al., 2004a), and increases in proinXammatory cytokines (Dickerson et al., 2005), than paradigms in which this threat is absent. While we believe that the psychological and physiological responses to social threat we outline are prototypical and possibly adaptive reactions (e.g., these responses may be associated with submissive behavior which is adaptive under conditions of social threat), individual diVerence factors associated with concerns about the self within social contexts, such as social anxiety or rejection sensitivity, may aVect reactions to such threats. We have been particularly interested in individuals’ subjective perceptions of their social status in social groups as an individual diVerence factor that may moderate responses to acute social self threats. We have found that individuals with lower subjective perceptions of status within important social groups, such as at school, have lower scores on indicators of mental health, including lower self-esteem, more shame-related emotions and cognitions, and more depressed and anxious mood (Gruenewald et al., 2001), and are also more likely to report feeling shame and

411

behaving submissively in social situations. These tendencies may also render these individuals to be more psychologically and physiologically reactive to situations which threaten to further diminish their social esteem and status. There has been little empirical examination of whether individuals with low subjective social status (SSS) are more psychologically and physiologically reactive to acute social stressors than those who perceive themselves to be of high social status. Adler and her associates (2000) demonstrated that individuals who perceived that they were of low status in society failed to show habituation in their cortisol responses to repeated exposure to a laboratory stressor paradigm (speech and mental arithmetic tasks with social evaluation) over a three-day period, as compared to individuals who perceived that they had higher societal status. In contrast, subjective perceptions of status in society did not moderate cortisol responses to a single exposure of performance of a set of challenging cognitive tasks in older adults in an investigation by Steptoe and colleagues (2005). However, performance of the cognitive stressor tasks did not occur in a context of explicit social evaluation as was characteristic of the study by Adler et al.; thus, it is possible that individuals high and low in SSS may be more likely to show divergent cortisol responses to a stressor which may threaten the social self. Our goal in the present investigation was to further explore the role of subjective perceptions of social status in moderating individuals’ cortisol responses to an acute stressor in the laboratory, as well as other psychological (emotional, cognitive) and physiological (blood pressure, heart rate) responses, and to determine whether SSS moderation of these responses was more likely to occur in response to experience of a stressor characterized by threat to the social self or its absence. Participants for this investigation were drawn from a larger study examining the association between social relationships within residential college dormitories and mental health. As part of this larger study, all participants completed measures of their SSS within their dormitories. Our examination of SSS as a moderator of acute stress reactivity draws upon Wndings from a previous study which documented that cortisol and shame increases, and social self-esteem decreases, were more likely to occur when challenging speech and mental arithmetic tasks (Trier Social Stress Test) were performed in front of an evaluative social audience, but not when the same tasks were performed in the absence of social-evaluative threat (SET, i.e., without an audience; Gruenewald et al., 2004b). However, the performance of the tasks under both socialevaluative and non-evaluative conditions led to similar cardiovascular responses, indicating that these responses were not as sensitive to the social-evaluative nature of the tasks. Given our previous Wndings, we expected that the hypothesized moderation of SSS on psychological and cortisol stress responses would be speciWc to conditions of

412

T.L. Gruenewald et al. / Brain, Behavior, and Immunity 20 (2006) 410–419

social self threat; speciWcally, we expected that individuals who perceived themselves to be of low status on their dorm Xoor would show greater decrements in social selfesteem and greater increases in shame and cortisol in the SET condition, but that such responses in the non-evaluative stressor condition would be similar in those of low and high SSS. We chose to assess social status within dorm Xoors because we wanted to gauge individuals’ subjective perceptions in a social group in which they had a fair amount of interaction with group members. As students who live in these dormitories spend a large portion of their daily life with each other—i.e., they share living space, eat together and engage in social activities together—we assumed that individuals’ perceptions of their status in these groups may be meaningful to participants. Individuals may also draw upon their attained status in their dorm in assessing their ability to attain high social status in other peer groups. We believe that individuals who perceive that they have low social status in an important social group are subject to a chronic threat to the social self—in other words, these individuals persistently perceive that their social self is devalued in an important social context. Such a perception may operate as an individual diVerence factor that may aVect individuals’ psychological and physiological responses to an acute threat to the social self, as such a threat may provide further conWrmation of the individual’s low social value and esteem. The assessment of status in a cohabitating group of peers may also provide a closer analog to the status typically assessed in social groups in non-human animal (e.g., monkeys, baboons) studies, than would other forms of status in humans, such as status in the larger society or socioeconomic status. 2. Method 2.1. Participants Experiment participants were recruited from a larger survey study (N D 363) designed to examine associations between characteristics of social life in college dormitories and mental health (results of the survey study not reported here). Participants were recruited to participate in the experimental investigation approximately two to six weeks following their participation in the larger survey study. As described elsewhere (Gruenewald et al., 2004b), 81 undergraduate college students (34 male, 47 female) ranging in age from 17 to 22 years (M D 19) from the larger investigation participated in the experiment. The majority (71.3%) of the participants were college freshmen (71.3%), the remainder were sophomores (23.8%) or above (5.1%). The sample was predominantly Asian-American (44.9%) or Caucasian (32.1%), with smaller numbers of other self-identiWed ethnic groups Black/African American (1.3%), Latino/a, Chicano/a (6.4%), Filipino/a (3.8%) or ‘Other’ (11.5%). All potential subjects participated in a telephone survey during which they were queried about the presence of health conditions or behaviors which might aVect salivary cortisol or cardiovascular measurements, including a number of health conditions (e.g., Cushing’s disease, hypertension) and use of non-prescription (e.g., regular tobacco use, marijuana) and prescription (e.g., oral contraceptives) drugs. Any individual self-reporting the presence of one of these health conditions or engagement in one of these behaviors was excluded from participa-

tion in the study. Potential subjects were not screened for participation on the basis of body weight, but the majority of participants fell within the normal range for body mass index (BMI; 75% had a BMI from 18.5 to 24.9). BMI ranged from 17 to 31.17 (M D 22.16, SD D 3.1); 10% had a low/underweight BMI (17–18.5), and 15% had an overweight/obese BMI (25+), with only 3 of these subjects having a BMI slightly over 30. On the day of the experiment appointment, subjects were asked not to engage in strenuous exercise, drink alcohol, or smoke, and not to eat or consume caVeine or dairy products at least 1 h prior to the session. Compliance with these requests was conWrmed during the session. Participants were given Psychology course experiment participation credit or paid $20, whichever they preferred, for their participation in the study.

2.2. Measures 2.2.1. Subjective social status A modiWed version of the MacArthur Subjective Social Status Scale (Adler et al., 2000) was used to assess SSS within participants’ dorm Xoors. This scale measures SSS by presenting respondents with a picture of a ladder with 10 rungs that is supposed to represent a particular social group (e.g., society, school, dorm Xoor) and asking participants to select the rung of the ladder that represents where they believe they fall in the social hierarchy of their group. Ladder rung selections are given numerical codes ranging from 1 D bottom rung of ladder to 10 D top rung of ladder, with higher scores indicating higher self-perceived status. Respondents were asked to rank their status with regard to how respected, esteemed and admired they felt they were within their dorm Xoor group. 2.2.2. State emotion and cognition measures As described elsewhere (Gruenewald et al., 2004b), we also examined anxiety (nervous, timid, tense, anxious, afraid) and shame-related (ashamed, humiliated, self-conscious, embarrassed) emotions before and after the stress tasks using modiWed emotion subscales from The AVects Balance Scale (ABS; Derogatis, 1975). The shame subscale from the State Shame and Guilt Scale (SSGS; Marschall et al., 1994) was used as a second measure of shame. The social and performance subscales of The State Self-Esteem Scale (SSES; Heatherton and Polivy, 1991) were used to assess participants’ thoughts about themselves in each of these domains before and after the stress tasks. Participants were also given Wve additional questions that assessed their overall rating of their performance on the speech and the math stress tasks, as well as feelings of diYculty, threat, and challenge during the activities. In addition, participants in the SET condition were asked to rate how much they liked the evaluators, how much they thought the evaluators liked and accepted them, how eVective they believed they were in convincing the panel of evaluators that they should get the job during the speech portion of the activity, and how well they thought the panelists would rate their performance overall. 2.2.3. Physiological measures Measures of heart rate (HR), systolic blood pressure (SBP) and diastolic blood pressure (DBP) were taken with a Critikon automatic sphygmomanometer (Dinamap Model 8146, Critikon, Tampa, Florida). Measurements were taken oscillometrically by placing an occluding cuV over the brachial artery of the subject’s non-dominant arm. Readings were taken every 2 min during six phases of the experimental session (see below for further description). Readings within each time period were averaged to produce a single value for each experimental phase. Salivary cortisol was assessed at four points throughout the experimental session (see below for detailed description). Salivary samples were collected with the Salivette device (Sartstedt, Rommelsdorf, Germany), which is essentially a cotton roll which participants were asked to place under their tongue for 3 min. The saturated roll was placed into a sealed, plastic container, refrigerated for no longer than 16 h, and then centrifuged to extract saliva. Samples were frozen at ¡80 °C until assayed with a

T.L. Gruenewald et al. / Brain, Behavior, and Immunity 20 (2006) 410–419

0

413

5 baseline cortisol

anticipation cortisol

post-stress cortisol recovery cortisol

Fig. 1. Timeline of laboratory session activities. ‘Qs’ indicates questionnaires.

commercially-available high sensitivity competitive enzyme immunoassay (EIA) (Salimetrics, State College, PA, USA, sensitivity <.004 g/dL, intraassay variation is 3%, and inter-assay variation is 10%). Values are reported in microgram per deciliter.

2.3. Design and procedure A detailed description of the experimental procedure is available elsewhere (Gruenewald et al., 2004b). BrieXy, participants in this investigation participated in a 90-min laboratory session, scheduled between the hours of 2:30 and 7:00 p.m. (scheduled timeslots: 2:30–4:00, 4:00–5:30, 5:30– 7:00), in which they were asked to perform a set of challenging speech and mental arithmetic activities in the presence of SET or in the absence of social evaluation (non-SET). All participants provided informed consent for their participation in the experiment. 2.3.1. Overview of laboratory activities A general outline of activities during the laboratory visit is depicted in Fig. 1. The Wrst 20 min of the session was a baseline resting period during which participants completed a set of pre-stress questionnaires and rested quietly. Baseline heart rate and blood pressure readings were taken during the last 10 min of this initial rest period and the Wrst saliva sample was taken immediately following (‘baseline’ measures). Participants were then given tape-recorded instructions for the speech and math stressor tasks, and then were given 10 min to mentally prepare for their speech. Heart rate and blood pressure readings were taken while the participants listened to the tape-recorded instructions for the tasks (‘task instruction’ measure) and during the 10-min preparation period (‘task preparation’ measure). A second saliva sample was taken at the end of the 10-min preparation period (‘stress anticipation’ measure). Participants then gave a 5-min speech and participated in a 5-min mental arithmetic task. Heart rate and blood pressure were monitored during these activities (‘speech’ and ‘math’ measurements). Following the stressor tasks, participants completed post-stress questionnaire measures and 5 min following the cessation of the stressor, provided another saliva sample (‘post-stress’ measure). Participants then rested quietly for 5 min. Heart rate and blood pressure were monitored during this resting period and the last saliva sample was taken at the end of the rest period (‘recovery’ measures). Finally participants were fully debriefed and paid or awarded Psychology experimental credit for their participation. 2.3.2. Overview of stressor tasks A modiWed version of the Trier Social Stress Test (TSST; Kirschbaum et al., 1993) was used as the stressor activity. The TSST, which consists of a challenging speech and mental arithmetic task, has been used extensively in acute laboratory stress investigations and has been shown to reliably elicit cortisol increases (Kirschbaum et al., 1993). In this investigation, participants were asked to explain why they would be good candidates for a job as a member of a focus group in which they would have to work eVectively with other students to devise solutions to problems typically faced by college students during their speech. The mental arithmetic task consisted of solving demanding arithmetic problems under time pressure on a computer for 5 min, using a procedure borrowed from Pruessner et al. (1999). In one condition of this study, participants were informed that they would perform the speech and math activities in front of a panel of evaluators who would judge their performance (‘social-evaluative threat (SET) condition’), while in the second condition of the study participants were

informed that they would do these activities while alone in the room (‘non-evaluative (non-SET) condition’).1 In the SET condition, evaluators behaved in a critical and non-accepting manner (e.g., did not smile, displayed a stern expression) during delivery of the participant’s speech, and sat behind the participant to watch his performance while he completed the arithmetic task. Further details on the design and characteristics of the stressor activities and conditions are available in Gruenewald et al. (2004b).

2.4. Analytic strategy Analyses presented below examine SSS as a moderator of psychological and physiological responses to SET and non-SET stressor conditions. Main eVects of SET and non-SET conditions on psychological and physiological stress responses were documented in a previous investigation (see Gruenewald et al., 2004b); these results will not be repeated in detail here. To examine SSS as a moderator of psychological and physiological responses to the social-evaluative and non-evaluative stressor conditions, repeated measures ANCOVAs were utilized with stressor condition (SET and non-SET) and SSS (high and low) as between-subjects variables and measurement timepoint (for emotion and cognition measures: pre- and post-stress; for cortisol: baseline, anticipation, post-stress, recovery; for heart rate and blood pressure: baseline, task instructions, task preparation, speech, math, post-stress, recovery) as a within-subjects variable. A signiWcant 3-way interaction of stressor condition, status, and time would indicate status moderation of responses to the social-evaluative versus non-evaluative stressor tasks. As discovered in our previous analysis, SET and non-SET condition participants did not diVer on baseline levels of any psychological or physiological variables except anxiety, thus baseline anxiety was used as a covariate in all analyses of emotion, self-esteem and physiological variables.

3. Results 3.1. Subjective social status Subjective ratings of social status within the dorm Xoor group were fairly positive, with the mean score for all experiment participants averaging above the midpoint of the 10-point scale (M D 6.8, SD D 2.0, Mdn D 7). Mean scores of participants in the experiment did not signiWcantly diVer from those of participants in the larger survey 1 A variety of experimental controls were used to ensure that participants in both stressor conditions expended similar activity and eVort. Sound (but not actual spoken content) was monitored in both conditions from an adjoining room and if participants in either condition ceased talking for 30 s or longer they received an alerting response to urge them to continue talking for the duration of the speech period. The speeches were also audiotaped in both conditions to ensure that participants talked for the entire time about the assigned speech topic. Participants were aware of this audiotaping and were informed that the speeches would be reviewed to conWrm that they stayed on topic. Review of all speeches conWrmed that participants in each condition spoke on the assigned topic. Participants in the non-SET condition (i.e., without an evaluative audience) were asked to give their speech in the same manner as participants in the SET condition (i.e., standing while facing a desk and two chairs that evaluators sat in for the SET condition).

414

T.L. Gruenewald et al. / Brain, Behavior, and Immunity 20 (2006) 410–419

study who did not participate in the experiment. Subjects were classiWed into high and low SSS using a median split; those with scores below 7 were placed in the low SSS group (n D 34) and those with scores at 7 or above were placed in the high SSS group (n D 47).2

Subjective social status did not signiWcantly moderate perceptions of task threat, challenge or diYculty, nor did high and low SSS participants diVer on ratings of the panel of evaluators in the SET condition. As previously reported, threat ratings were signiWcantly higher in the SET as compared to the non-SET stressor, challenge ratings were marginally higher, and ratings of task diYculty did not diVer. Mean ratings of perceived judgment, liking, and acceptance regarding the panel of evaluators were lower than the scale midpoint for all rating questions, indicating that participants in the SET condition did not like the evaluators or think that the evaluators liked them or judged them favorably.

SBP

130 120

mm/Hg

110 100 90

DBP

80 70 60 50

140 130 120 110

bpm

3.2. Status moderation of task and evaluator ratings

140

100

HR

90 80 70

3.3. Status moderation of emotional and cognitive responses

60

3.4. Status moderation of cardiovascular responses Subjective status level also did not moderate cardiovascular responses to the social-evaluative versus non-evaluative stressor (F values for 3-way interaction of status £ stress condition £ time: HR F (5,355) D 1.04, p D .40; SBP F (5,355) D .53, p D .75; DBP F (5,355) D .44, p D .82). As previously reported, all cardiovascular parameters increased signiWcantly in response to performance of stressor tasks in both stress conditions. This pattern did not vary for high and low SSS participants (see Fig. 2). There 2 A large number of participants had a SSS score at the median of 7 (n D 17); these subjects were classiWed into the high rather than low SSS group because a score of 7 is well above the midpoint of the scale and above the mean score for the sample.

ry

m at h

re co ve

pr ep a

sp e

ra t

ec

h

io n

n io at ip an tic

ba s

el

in e

50

Subjective social status did not signiWcantly moderate changes in anxiety, shame, social or performance selfesteem in the SET versus the non-SET condition (F values for 3-way interactions of status £ stress condition £ time: anxiety F (1,77) D .43, p D .51; ABS shame F (1,71) D .07, p D .80; SSGS shame F (1,76) D .38, p D .54; social selfesteem F (1,76) D .97, p D .33; performance self-esteem F (1,76) D .15, p D .71). As reported in a previous analysis (Gruenewald et al., 2004b), anxiety and shame increased and social and performance self-esteem decreased as a result of performance of the stressor tasks in both stress conditions, with signiWcantly larger increases in shame and decreases in social self-esteem occurring in the SET condition (there was no stress condition eVect for anxiety or performance self-esteem). Both high and low status participants showed this pattern of responses.

time point SET/low SSS

SET/high SSS

non-SET/low SSS

non-SET/high SSS

Fig. 2. Change in systolic blood pressure (SBP), diastolic blood pressure (DBP), and heart rate (HR) in high and low subjective social status (SSS) participants in the social-evaluative threat (SET) and non-evaluative (non-SET) stressor conditions.

was a signiWcant between-subjects eVect of SSS for diastolic blood pressure levels (F (1,71) D 5.56, p D .02), with those of higher status having higher levels across the session. 3.5. Status moderation of cortisol responses As hypothesized, SSS moderated cortisol responses to the social-evaluative stressor (3-way interaction of status £ stress condition £ time, F (3,219) D 6.21, p D .000, Greenhouse–Geisser adjusted F (1.4,101.3) D 6.21, p D .008) in the entire sample.3 Cortisol values for each measurement timepoint for high and low status participants in each stress 3 A linear regression analysis using the continuous SSS variable and the change in cortisol from baseline to the post-stressor measurement yielded comparable results to analyses using repeated measures ANCOVA. Stressor condition was a signiWcant predictor (p D .000;  for dummy-coded stressor condition variable representing SET condition D .40), as was the interaction between stressor condition and SSS (p D .038;  for interaction term representing SSS £ stressor dummy-coded variable representing the SET condition D .31), while SSS level alone was not a signiWcant predictor of stressor-induced change in cortisol (p D .83). Inspection of the parameter estimate for the interaction term indicates that there is an approximate .04 g/dL change in cortisol (baseline to post-stressor change) for each unit change in SSS in the SET condition.

T.L. Gruenewald et al. / Brain, Behavior, and Immunity 20 (2006) 410–419 0.4 0.35

cortisol ug/dL

0.3 0.25 0.2 0.15 0.1 0.05 0 baseline

anticipation

post-stress

recovery

timepoint SET/low SSS

SET/high SSS

non-SET/low SSS

non-SET/high SSS

Fig. 3. Change in cortisol levels throughout the session in high and low subjective social status (SSS) participants in the social-evaluative threat (SET) and non-evaluative (non-SET) stressor conditions.

condition are displayed in Fig. 3. Contrary to our expectations, however, individuals with high SSS had higher poststress cortisol levels than low SSS participants in the SET condition (post-stress cortisol MdiVerence D .14 g/dL, SEM D .05, p D .006, recovery cortisol MdiVerence D .12 g/dL, SEM D .04, p D .001). The cortisol increases of those with low SSS were not even signiWcantly diVerent from their baseline (pre-stress) cortisol levels (MdiVerence D .03 g/dL, SEM D .04, p D .34). Post-stressor cortisol levels of those with high SSS were signiWcantly greater than their baseline levels at both post-stressor measurements (baseline-poststress MdiVerence D .20 g/dL, SEM D .03, p D .00; baselinerecovery MdiVerence D .12 g/dL, SEM D .02, p D .00). Post hoc analyses indicated that the inclusion of gender as a covariate or as a between-subjects predictor in repeated measures ANCOVAs for cortisol did not aVect the pattern of results reported above. Gender did not interact with SSS to predict cortisol values across the session, nor did males and females diVer in their baseline levels of cortisol. Body mass index was also examined as a covariate in cortisol analyses; the inclusion of this variable also did not aVect the results reported above. Although participants who attended the last session of the day (5:30–7:00 p.m.) had marginally lower baseline levels of cortisol (p D .08) as compared to those in the preceding timeslots (2:30–4:00 and 4:00–5:30 p.m.), time of session appointment did not interact with experimental condition or SSS level to aVect changes in cortisol across the session. 3.6. Association of SET-induced psychological and cortisol responses Contrary to our expectations, subjective status did not moderate cognitive and emotional responses to the socialevaluative versus non-evaluative stressor condition. Participants with both high and low SSS showed similar responses, with increases in anxiety and decreases in performance selfesteem in both stress conditions, and greater increases in shame and greater decrements in social self-esteem in the

415

SET condition. As hypothesized, SSS did moderate cortisol responses to the social-evaluative stressor, but participants high in SSS showed robust cortisol increases while subjects low in SSS did not produce a signiWcant cortisol increase in the SET condition. Given that both groups showed similar psychological responses in the SET condition, it is interesting that they showed divergent cortisol responses. An examination of associations between stress-induced changes in cortisol and the psychological measures indicated diVerent patterns of association for some of the psychological measures in low and high SSS participants. High SSS subjects showed positive associations between SET-induced increases in cortisol and SSGS shame increases (r D .40, p D .06), perceived threat ratings (r D .59, p D .00), perceived task diYculty ratings (r D .42, p D .04), and a negative association between cortisol increases and perceived acceptance from panel evaluators (r D ¡.41, p D .05). Ratings or changes in other psychological measures were not signiWcantly associated with cortisol increases in high SSS participants. In general, low SSS subjects did not show a signiWcant pattern of association between SET-induced changes in cortisol and psychological variables. Low SSS participants did show a signiWcant negative correlation between SET-induced changes in cortisol and threat ratings (r D ¡.66, p D .00). Given that relatively few low status participants showed a sizable change in pre- to poststressor cortisol levels this signiWcant negative correlation was surprising. Further inspection indicated that this correlation was driven by a few low status individuals with low threat ratings and small cortisol increases and a few individuals with high threat ratings and slight cortisol decreases; the majority of low status participants showed no change in cortisol levels across the range of threat ratings. 3.7. Ancillary analyses—other potential psychological moderators As previously reviewed, we have found that subjective perceptions of status in speciWc social groups, such as at school, are associated with indicators of mental health, with those of higher status perceptions having higher levels of selfesteem, and lower levels of depressed and anxious mood. In this small group of individuals participating in the experiment, we found that perceptions of status within one’s dorm Xoor group were signiWcantly associated with self-esteem (r D .37, p D .00) as measured by the Rosenberg Self-Esteem Scale (Rosenberg, 1965) and depressed mood (r D ¡.30, p D .00) as measured with the Center for Epidemiologic Studies of Depression (CESD) Scale (RadloV, 1977), and showed a marginally signiWcant association with anxiety (r D ¡.19, p D .09) as measured by the Spielberger Trait Anxiety Scale (Spielberger et al., 1970). It is possible that the SSS moderation of cortisol responses to the social-evaluative stressor might just be a proxy for eVects of one of these other psychological constructs. However, none of these variables showed a signiWcant association with cortisol change in the SET stressor condition, or interacted with stressor condition to aVect cortisol levels across the session (all p’s > .25). These

416

T.L. Gruenewald et al. / Brain, Behavior, and Immunity 20 (2006) 410–419

results are consistent with our theoretical perspective which asserts that individual diVerence factors relating to concerns of the self within a social context are more likely to moderate responses to social self threats than individual diVerence factors which are less focused on the social self. 4. Discussion The aim of the present investigation was to examine SSS as a moderator of cognitive, emotional and physiological responses to SET, one type of threat to the social self. Subjective status as measured in this investigation assessed participants’ perceptions of their own status in a relatively small, intimate social group (college students residing on the same dorm Xoor). Contrary to expectations, SSS did not moderate cognitive or emotional responses to SET versus non-SET threat in the laboratory. High and low status participants had similar task ratings, showed an increase in anxiety and a decrease in performance self-esteem in both stressor conditions, and showed greater increases in shame and decreases in social self-esteem in the social-evaluative stressor condition. The failure to Wnd moderation of shame and social self-esteem responses by perceived social status as hypothesized may either indicate that status perceptions do not diVerentially aVect conscious responses (i.e., those captured on our paper and pencil measures) on these variables as a result of social self threat experience, or that our measures of these constructs were not sensitive enough to capture diVerences in response between the two groups. Cardiovascular responses also did not signiWcantly diVer between high and low status participants in response to stressor task performance, although individuals with high levels of SSS did have greater diastolic blood pressure levels overall. Given our previous Wndings of similar cardiovascular responses to social-evaluative and non-evaluative stressor experience, and less consistent Wndings of cardiovascular diVerences in response to social threat in nonhuman animals of varying levels of social status, we had not expected that SSS would moderate the cardiovascular responses to SET examined in this study. As hypothesized, however, participants high and low in SSS did show signiWcantly diVerent cortisol responses to stressor task performance. This divergence only occurred within the SET condition indicating that diVerences in stress reactivity among individuals high and low in SSS emerged in a situation characterized by threat to social esteem or status. Contrary to our expectations, individuals with high SSS, not low, showed the greatest cortisol response to SET. In fact, individuals low in SSS showed only a very small, and non-signiWcant cortisol increase in response to the social-evaluative stressor. It is possible that the non-signiWcant cortisol response to SET of individuals with low SSS may simply indicate that they did not Wnd the tasks to be ‘stressful’ or ‘threatening.’ However, this explanation stands in contrast to their task ratings and cognitive and emotional responses, which indicated a pattern of negative self-evaluative emotions and

cognitions and perceived threat in response to performance of stressor tasks under SET. This pattern of responses was similar to that of individuals with high SSS. Associations between SET-induced changes in cortisol and these psychological responses was dissimilar for participants high and low in SSS, however. Individuals with high SSS showed signiWcant associations between cortisol increases and increases in shame and ratings of task diYculty, threat, and lack of acceptance from evaluators; participants with low SSS did not show this signiWcant pattern of associations between stressor-induced cortisol change and psychological responses. These Wndings appear to indicate a disconnect between psychological responses and bodily responses in individuals with low SSS. Another potential explanation for the lack of a signiWcant cortisol response in participants with low SSS is that these participants had lower levels of ego involvement or less concern about social evaluation during performance of the speech and math stressor tasks. On the other hand, high status participants may have felt more threatened by negative social evaluation since they had more to lose in terms of a change in relative social status, resulting in greater cortisol reactivity in the social-evaluative condition. It is diYcult to objectively assess these factors, thus these hypotheses cannot be tested in the current investigation. However, analysis of indicators of participant performance do not suggest diVerential engagement of high and low SSS subjects. The two groups of participants performed similarly on the math stressor task (i.e., similar number of correct responses, errors, and average time to answer; data not shown), and evaluators present in the SET condition who were blind to the SSS level of participants rated the performance of high and low SSS subjects as similar on the speech and the math task, as did an additional rater who listened to an audiotape recording of the speech and provided an overall rating on the basis of the quality of arguments made, conWdent delivery, etc. (data not shown). Thus, engagement with the tasks and performance as perceived by others appear to be similar for high and low SSS participants. The similar declines in social self-esteem and increases in shame in high and low SSS subjects also suggest that high and low SSS participants showed a similar level of concern with negative evaluation Xowing from the evaluators in the SET condition. Our results diVer from those of Adler and her associates (2000) which found evidence for greater cortisol reactivity (as indexed by the failure to habituate to repeated stressor exposure) to a similar laboratory stressor for individuals with low SSS in society. It is diYcult to compare our Wndings with their Wndings, as we examined cortisol responses to only a single stressor exposure and measured subjective status within a smaller social group. Our Wnding of hyporeactive cortisol responses in low status participants are similar to those observed in an investigation by Hellhammer and colleagues (1997), which examined cortisol stress reactivity as a function of social status level in small groups of army recruits. Similar to the results of the present investiga-

T.L. Gruenewald et al. / Brain, Behavior, and Immunity 20 (2006) 410–419

tion, their study found that high status recruits showed a sizable and signiWcant increase in cortisol following stressor task performance while low status recruits showed a smaller, non-signiWcant increase in cortisol levels. Both studies evaluated social status in small groups. However, social status was measured with a more objective peer-rating measure in their investigation. Based on our previous research, we had expected that both individuals with high and low levels of subjective status would show cortisol increases in response to performance of stressor tasks under SET, but that the magnitude of responses would be signiWcantly greater in individuals who already perceived themselves to be of low status. The lack of a signiWcant cortisol response to SET in low SSS participants was unexpected. Investigations of cortisol responses to acute, novel stressors (e.g., restraint stress, social stress) in animals often indicate that low status animals show greater cortisol increases than their dominant counterparts, however, a failure to Wnd signiWcant cortisol responses to stressors have also been found among low status animals or subgroups of low status animals in a number of studies (see Tamashiro et al., 2005, for review). It is possible that the lack of a signiWcant cortisol response in the low perceived status group despite a signiWcant emotional response is a result of some form of dysfunction in the HPA axis in this group, e.g., a downregulation of pituitary CRH receptors, alterations in negative feedback sensitivity, or an alteration in hypothalamic coordination of HPA activity following emotional input. This type of alteration could result from an increased sensitivity to SETs in the environment or increased exposure to such threats (associated with early environmental conditions, for example) resulting in a hyporeactivity of the system over time. This kind of explanation has been oVered for the low basal levels of cortisol sometimes found among persistently psychologically reactive individuals, such as those with PTSD. However, it was impossible to test the validity of this explanation in the current study. Whether the pattern of psychological and physiological responses to SET witnessed in high or low status individuals is of greater concern for mental or physical health is unclear. From a psychological perspective, both individuals with high and low SSS are vulnerable to experience negative psychological states as a consequence of social self threats. From the vantage point of cortisol reactivity, individuals with high SSS appear to be more reactive to such threats, as they show increased activity while those with low SSS do not. High status participants also showed greater concordance between SET-induced psychological changes and cortisol changes. These Wndings indicate that high status individuals are more likely to experience psychobiological responses to social self threat that, if exposure was repeated or prolonged, might render individuals vulnerable to adverse mental and physical health states. On the other hand, cortisol hyporeactivity to SET in low SSS participants, especially given their signiWcant psycho-

417

logical responses, may also indicate a psychobiological vulnerability for negative mental and physical health states. Blunted cortisol responses to acute laboratory stressors have been found in other psychosocial conditions, such as chronic stress experience, depression, and neuroticism (Burke et al., 2005; Matthews et al., 2001; Phillips et al., 2005; see Heim et al., 2000). Heim and colleagues have also noted that hypocortisolism, assessed under basal and stimulated (pharmacological and psychosocial challenge) conditions, is characteristic of some disease conditions, such as post-traumatic stress disorder, chronic fatigue syndrome, rheumatoid arthritis, asthma and Wbromyalgia. 4.1. Limitations and future directions One limitation of the present investigation is that the social-evaluative threat faced by participants in the laboratory did not come directly from individuals in the social group in which SSS was measured. Logistical constraints and the need to standardize the stressor experience prevented us from using peers from each participant’s own group as the source of social evaluation during performance of stressor tasks. However, evaluators were college age peers, thus they were similar to individuals that would have composed each participant’s social group in which status was assessed. Another potential limitation of our investigation is the generalizability of our Wndings. It is unclear whether subjective perceptions of status measured in other social groups, such as the larger school environment, community or society, might operate in a similar fashion to aVect psychological and biological responses to an acute social self threat. We chose to assess perceived status in a small group of cohabitating individuals because we wanted to assess status perceptions in a group that played a role in the day-to-day lives of our participants and that could be viewed as an analog to social living groups examined in studies of social status, behavior and physiology in non-human animals. Links between perceptions of social status, psychological states and biological variables may be weaker as the social group in which one assesses their relative status grows in size and has less deWned boundaries of membership. It is also possible that links between these variables may be stronger in younger as compared to older adults. Young adults may focus more or care more about their relative levels of social status within peer-age social groups as many are still in pursuit of Wnding partners, choosing occupational paths, and establishing their adult social identities. Thus, perceptions of low or high social status may have a greater impact on mental well-being and physiological states in young adults as compared to older adults whose attention is directed to other pursuits that are less impacted by levels of peer status (e.g., childcare). However, these are hypotheses that will need to be clariWed by future research. Future research will also be needed to more clearly elucidate the relationships between subjective perceptions of social status and cortisol responses to acute threat. Our

418

T.L. Gruenewald et al. / Brain, Behavior, and Immunity 20 (2006) 410–419

Wnding of a hyporeactive cortisol response to acute stress in participants with low SSS conXicts with previous investigations which found evidence of hyperreactivity (as assessed by lack of habituation to repeated stressor exposure) or no moderation of cortisol reactivity by SSS perceptions. It will be important to not only discern the pattern of cortisol responses characteristic of high and low status individuals but also characteristics of stressful situations that elicit divergent responses (e.g., social versus non-social threat). Whether exposure to an acute social threat could also cause a shift in perceptions of social status within a speciWc group, and the associated psychobiological correlates of such a change in perceived status should also be an aim of future research. To date, there is a limited body of research on the psychobiological correlates of SSS perceptions. The results of this investigation indicate that subjective perceptions of social status within a small social group are associated with physiological reactivity to social threat. More research will be needed to identify other psychobiological factors associated with status perceptions, social environment and individual diVerence factors which aVect these perceptions, and the social groups (e.g., society, school, more intimate social groups) for which perceptions of status show an association with psychological and biological states that may have implications for health. References Adler, N.E., Epel, E.S., Castellazzo, G., Ickovics, J.R., 2000. Relationship of subjective and objective social status with psychological and physiological functioning: preliminary data in healthy white women. Health Psychol. 19 (6), 586–592. Avitsur, R., Stark, J.L., Sheridan, J.F., 2001. Social stress induces glucocorticoid resistance in subordinate animals. Horm. Behav. 39, 1–11. Barkow, J.H., 1989. Darwin, Sex, and Status. University of Toronto Press, Toronto. Burke, H.M., Davis, M.C., Otte, C., Mohr, D.C., 2005. Depression and cortisol responses to psychological stress: a meta-analysis. Psychoneuroendocrinology 30 (9), 846–856. Buwalda, B., Kole, M.H., Veenema, A.H., Huininga, M., de Boer, S.F., Korte, S.M., Koolhaas, J.M., 2005. Long-term eVects of social stress on brain and behavior: a focus on hippocampal functioning. Neurosci. Biobehav. Rev. 29 (1), 83–97. Cooley, C.H., 1902/1983. Human Nature and the Social Order. Transaction Books, New Brunswick. Derogatis, L.R., 1975. AVects Balance Scale. Clinical Psychometric Research, Riderwood, MD. Dickerson, S.S., Kemeny, M.E., 2004. Acute stressors and cortisol responses: a theoretical integration and synthesis of laboratory research. Psychol. Bull. 130 (3), 355–391. Dickerson, S.S., Gable, S.L., Kemeny, M.E., Aziz, N., Irwin, M.R., 2005. Social-evaluative threat and proinXammatory cytokine activity: an experimental laboratory investigation. In: Presented at the Annual Meeting of the American Psychosomatic Society. Vancouver, CA. Dickerson, S.S., Gruenewald, T.L., Kemeny, M.E., 2004. When the social self is threatened: shame, physiology, and health. J. Pers. 72 (6), 1191–1216. Fuchs, E., Flügge, G., 2003. Chronic social stress: eVects on limbic brain structures. Physiol. Behav. 79, 417–427. Gilbert, P., 1997. The evolution of social attractiveness and its role in shame, humiliation, guilt and therapy. Brit. J. Med. Psychol. 70 (Pt 2), 113–147.

Gilbert, P., 1998. Shame, status, and social roles. Psychobiology and evolution. In: Gilbert, P., Andrews, B. (Eds.), Shame: Interpersonal Behavior, Psychopathology, and Culture. Oxford University Press, New York, pp. 23–33. Gruenewald, T.L., Dickerson, S.S., Kemeny, M.E., 2004a. A social function for shame: The social self-preservation theory. In: Tracy, J.L., Robins, R.W., Tangney, J. (Eds.), The Self-Conscious Emotions. Gruenewald, T.L., Kemeny, M.E., Aziz, N., Fahey, J.L., 2004b. Acute threat to the social self: shame, social self-esteem, and cortisol activity. Psychosom. Med. 66 (6), 915–924. Gruenewald, T.L., Kemeny, M.E., Adler, N., 2001. Subjective social status, social hierarchy/rank and mental health. Poster Session Presented at the Annual Meeting of the American Psychosomatic Society. Monterey, California. Haller, J., Kiem, D.T., Makara, G.B., 1996. The physiology of social conXict in rats: what is particularly stressful? Behav. Neurosci. 110 (2), 353–359. Heatherton, T.F., Polivy, J., 1991. Development and validation of a scale for measuring state self-esteem. J. Pers. Soc. Psychol. 60 (6), 895–910. Heim, C., Ehlert, U., Hellhammer, D.H., 2000. The potential role of hypocortisolism in the pathophysiology of stress-related bodily disorders. Psychoneuroendocrinology 25 (1), 1–35. Hellhammer, D.H., Buchtal, J., Gutberlet, I., Kirschbaum, C., 1997. Social hierarchy and adrenocortical stress reactivity in men. Psychoneuroendocrinology 22 (8), 643–650. Henry, J.P., 1992. Biological basis of the stress response. Integr. Physiol. Behav. Sci. 27 (1), 66–83. Holst, D., 1997. Social relations and their health impact in tree shrews. Acta Physiol. Scand. Suppl. 640, 77–82. James, W., 1890. The Principles of Psychology. Encyclopaedia Britannica, Chicago. Kemeny, M.E., Gruenewald, T.L., Dickerson, S.S., 2004. Shame as the emotional response to threat to the social self: implications for behavior, physiology, and health. Psychol. Inquiry 15 (2), 153–160. Kimura, K., Shimizu, K., Hayashi, M., Iskikawa, T., Ago, Y., 2000. Pituitary– adrenocortical responses to the Wrst dyadic encounters in male rhesus monkeys: eVect of dominance relationship. Am. J. Primatol. 50, 247–256. Kirschbaum, C., Pirke, K.M., Hellhammer, D.H., 1993. The ‘Trier Social Stress Test’—A tool for investigating psychobiological stress responses in a laboratory setting. Neuropsychobiology 28 (1-2), 76–81. Kollack-Walker, S., Watson, S.J., Akil, H., 1997. Social stress in hamsters: defeat activates speciWc neurocircuits within the brain. J. Neurosci. 17 (22), 8842–8855. Leary, M.R., 1995. Self-Presentation: Impression Management and Interpersonal Behavior. Brown & Benchmark Publishers, Madison, WI. Marschall, D.E., Sanftner, J.L., Tangney, J.P., 1994. The State Shame and Guilt Scale. George Mason University, Fairfax, VA. Matthews, K.A., Gump, B.B., Owens, J.F., 2001. Chronic stress inXuences cardiovascular and neuroendocrine responses during acute stress and recovery, especially in men. Health Psychol. 20 (6), 403–410. Phillips, A.C., Carroll, D., Burns, V.E., Drayson, M., 2005. Neuroticism, cortisol reactivity, and antibody response to vaccination. Psychophysiology 42 (2), 232–238. Pich, E.M., Heinrichs, S.C., Rivier, C., Miczek, K.A., Fisher, D.A., Koob, G.F., 1993. Blockade of pituitary–adrenal axis activation induced by peripheral immunoneutralization of corticotropin-releasing factor does not aVect the behavioral response to social defeat stress in rats. Psychoneuroendocrinology 18 (7), 495–507. Price, J., Sloman, L., Gardner, R., Gilbert, P., Rohde, P., 1994. The social competition hypothesis of depression. Brit. J. Psychiat. 164 (3), 309–315. Pruessner, J.C., Hellhammer, D.H., Kirschbaum, C., 1999. Low self-esteem, induced failure and the adrenocortical stress response. Pers. Individ. DiVer. 27 (3), 477–489. Quan, N., Avitsur, R., Stark, J.L., He, L., Shah, M., Caligiuri, M., Padgett, D.A., Marucha, P.T., Sheridan, J.F., 2001. Social stress increases the susceptibility to endotoxic shock. J. Neuroimmunol. 115 (1-2), 36–45. RadloV, L.S., 1977. The CES-D Scale: a self-report depression scale for research in the general population. Appl. Psychol. Measur. 1 (3), 385– 401.

T.L. Gruenewald et al. / Brain, Behavior, and Immunity 20 (2006) 410–419 Rosenberg, M., 1965. Society and the Adolescent Self-image. Princeton University Press, Princeton, NJ. Sapolsky, R.M., 1993. Endocrinology alfresco: psychoneuroendocrine studies of wild baboons. Recent Prog. Horm. Res. 48, 437–468. Shively, C.A., Laber-Laird, K., Anton, R.F., 1997. Behavior and physiology of social stress and depression in female cynomolgus monkeys. Biol. Psychiat. 758 (1–2), 871–882. Skutella, T., Montkowski, A., Stohr, T., Probst, J.C., Landgraf, R., Holsboer, F., Jirikowski, G.F., 1994. Corticotropin-releasing hormone (CRH) antisense oligodeoxynucleotide treatment attenuates social defeat-induced anxiety in rats. Cell. Mol. Neurobiol. 14 (5), 579–588. Spielberger, C.D., Gorsuch, R.L., Lushene, R.E., 1970. STAI Manual for the State/Trait Anxiety Inventory. Consulting Psychologists Press, Palo Alto, CA.

419

Stefanski, V., Engler, H., 1998. EVects of acute and chronic social stress on blood cellular immunity in rats. Physiol. Behav. 64 (5), 733–741. Stefanski, V., Engler, H., 1999. Social stress, dominance and blood cellular immunity. J. Neuroimmunol. 94 (1-2), 144–152. Stefanski, V., Knopf, G., Schulz, S., 2001. Long-term colony housing in long evans rats: immunological, hormonal, and behavioral consequences. J. Neuroimmunol. 114 (1-2), 122–130. Steptoe, A., Kunz-Ebrecht, S.R., Wright, C., Feldman, P.J., 2005. Socioeconomic position and cardiovascular and neuroendocrine responses following cognitive challenge in old age. Biol. Psychol. 69 (2), 149–166. Tamashiro, K.L., Nguyen, M.M., Sakai, R.R., 2005. Social stress: from rodents to primates. Front. Neuroendocrinol. 26 (1), 27–40.