Discrimination and risk - taking , 1 Experiencing discrimination increases risk - taking

Prior research has revealed racial disparities in health outcomes and more health-compromising behaviors, such as smoking and drug abuse. It has been suggested that discrimination may contribute to such disparities, but the mechanisms through which this occurs are not well understood. Here, we examined whether the experience of discrimination affects acute physiological stress responses and increases risk-taking behavior. Black and White participants received rejecting feedback from partners who were either the same race (in-group rejection) or a different race (out-group rejection/discrimination). Physiological (cardiovascular and neuroendocrine) changes, cognitive processes (memory and attentional bias), and risk-taking behavior were assessed. Significant participant-race by partner-race interactions were observed. Cross-race, compared to same-race, rejection was associated with lower cortisol, increased cardiac output, decreased vascular resistance, greater anger, and more risk-taking behavior. These data suggest distinct profiles of physiological reactivity, cognitive processing, and risktaking in response to discrimination implicating direct and indirect pathways to health disparities. Discrimination and risk-taking, 3 Experiencing discrimination increases risk-taking Being rejected is a powerful aversive experience. In the short-term it affects emotions, thoughts, and behavior (Williams, 2001), and in the long-term can influence physical (Cacioppo, Hawkley, & Berntson, 2003) and mental health (Williams, 2001). However, not all types of social rejection affect us similarly. Rejection by out-group members (persons from different social categories) can be interpreted as discrimination, which may set in motion a distinct set of attributions, emotions, and behaviors compared to when rejection comes from in-group members (Crocker, Voelkl, Testa, & Major, 1991; Mendes, Major, McCoy, & Blascovich, 2008). We explored the effects of in-group compared to out-group social rejection with two specific goals. The first goal was to measure the physiological consequences of in-group compared to out-group rejection. The second goal was to examine how rejection influenced risk-taking behavior with the prediction that because discrimination typically evokes anger and approach-motivation, outgroup rejection would lead to more risk-taking relative to in-group rejection. Several lines of research have examined the physiological consequences of discrimination in an attempt to understand health disparities between Europeanand AfricanAmericans (Pascoe & Smart-Richman, 2009). In large scale epidemiological studies, for example, African-Americans tend to have higher resting blood pressure than their age-matched European-American counterparts (Krieger & Sidney, 1996), and Black and lower socioeconomic status adults, on average, have flatter (more dysregulated) diurnal cortisol cycles than Whites and higher socioeconomic status adults (Fuller-Rowell, Doan, & Eccles, 2011). Laboratory based studies have shown that participants who experience, view, or recall an episode of discrimination are angrier, exhibit a stronger cardiac reaction, and have a slower recovery profile than those not exposed to discrimination (e.g., Guyll, Matthews, & Bromberger, 2001; Mendes et al., 2008). Discrimination and risk-taking, 4 However, the data on acute responses to discrimination are not always straightforward, with some studies showing inconclusive results (e.g., Brondolo, Rieppi, Kelly, & Gerin, 2003). Even less consistent are the studies attempting to link experiences of discrimination with HPA activation, specifically cortisol responses. For example, in the Fuller-Rowell study cited above, African-Americans who perceived more discrimination had healthier diurnal cortisol cycles than those who perceived less discrimination. There is certainly reason to speculate that discrimination would acutely activate the HPA system – one of two primary stress systems. For example, non-human primate studies find that in stable hierarchies subordinate compared to dominant baboons exhibit greater HPA reactivity, higher basal cortisol levels, and impaired sensitivity of the HPA to negative feedback regulation (Sapolsky, 1982). In humans, HPA activation has been linked to loss of social standing (Mehta, Jones, & Josephs, 2008), negative social feedback (Koslov, Mendes, Patjas, & Pizzagalli, 2011), and with feelings of shame (Dickerson & Kemeny, 2004). Indeed, in a meta-analysis of cortisol reactivity the situational factors most consistently linked to cortisol increases were situations when “an important aspect of the self-identity is or could be negatively judged by others” (Dickerson & Kemeny, 2004, p. 358). Sapolsky’s baboon research and the cortisol meta-analysis suggest that perceptions and experiences of discrimination might activate the HPA, which may be implicated in the observed racial health disparities; however, we view this perspective as tenuous. For one, though rejection clearly engenders feelings of shame, discrimination based on uncontrollable factors (e.g., race) typically does not elicit shame but rather anger (Gibbons, Etcheverry, Stock, Gerrard, Weng, et al., 2010; Smart-Richman & Leary, 2009). At a neurobiological level, although shame might activate HPA responses (Dickerson & Kemeny, 2004) anger tends to elicit activation of the Discrimination and risk-taking, 5 sympathetic adrenal-medullary axis (SAM) (Stemmler, 2004). Though these two systems (HPA and SAM) are relatively independent, increases in cortisol tend to be associated with reduced sympathetic nervous system activity (Golczynska, Lenders, & Goldstein, 1995; Pavcovich & Valentino, 1997). Secondly, previous data suggest that an acute experience of out-group rejection (i.e., discrimination) is associated with increased cardiac activity (specifically cardiac output) and a decline in vasculature resistance (the primary determinant of blood pressure) – a profile of responses linked to challenge states (Mendes, et al., 2008). In contrast, same-race rejection was associated with lower cardiac output (less cardiac efficiency) and an increase in vasculature resistance – a profile of responses linked to threat states. Though these cardiovascular profiles are not perfect proxies for neuroendocrine activity, theoretically the threat pattern is more likely to be associated with cortisol increases than the challenge pattern. Therefore, we anticipated that social rejection from same-race compared to cross-race partners would activate the HPA, increase feelings of shame, and show cognitive consequences of HPA activation, like impaired short term memory. The second goal of this work was to examine how discrimination influences behavior, specifically risk-taking. A large corpus of research on racial health disparities suggests AfricanAmericans exhibit worse health outcomes and engage in riskier behaviors, such as substance abuse, overeating, and smoking, compared to their European-American counterparts (e.g., Gibbons, Gerrard, Cleveland, Wills, & Brody, 2004; Hertz, Unger, Cornell, & Saunders, 2005; Williams, Neighbors, & Jackson, 2008). For instance, in a longitudinal study African-American adolescents’ self-reported experiences of discrimination predicted their substance use over time, and this relation was mediated by anger (and reduced self-control) (Gibbons et al., 2010). In a Discrimination and risk-taking, 6 complementary lab study, imagining an experience of discrimination increased the accessibility of words associated with substance use. And again, anger mediated this effect. Anger may be an especially important emotion when examining behaviors associated with risk-taking. For example, dispositionally angry people and those made angry via writing vignettes express optimistic risk estimates and choose riskier options compared to fearful individuals (Lerner & Keltner, 2001). Furthermore, approach motivation resulting from the experience of anger can lead to performance improvements (Lerner & Tiedens, 2006). In the race-rejection study cited above (Mendes, et al., 2008), participants who experienced out-group rejection (i.e., discrimination) showed more anger and performed better on a word-finding task compared to participants receiving in-group rejection, indicating more approach-oriented behavior. In the current study, we examined the effects of in-group and out-group social rejection on physiological reactivity, cognitive and affective outcomes, and risk-taking. We anticipated that in-group compared to out-group rejection would be associated with larger cortisol increases, threat reactivity, and memory impairments, whereas out-group, relative to in-group, rejection would be associated with more approach/challenge responses, expressed anger, greater vigilance for danger, and more risk-taking behavior. To test predictions, we induced social rejection by giving participants negative feedback during a computer interaction task. We manipulated whether Black and White participants thought they were interacting with sameor different-race partners and all participants received rejecting social feedback. Immediately after the interaction, participants completed cognitive tasks and a risk-behavior measure. Cortisol and cardiovascular reactivity were measured throughout the study and we coded behavior for displays of anger and shame. Discrimination and risk-taking, 7 Method Participants Ninety-one participants (55% females; 49 Caucasian, 42 African-American; 50 students, 41 community members; race and student status were unrelated [p > .50]) were recruited from the Cambridge, MA area and were compensated two credit-hours or $30. Participants (Mage = 24.11 years, SD = 6.11; range 18-39) were excluded for hypertension, pacemakers, cardiac medications, and/or pregnancy. Procedure Upon arrival participants provided consent, completed an initial memory task, and then selected one of six avatars from their race/sex category to represent them during the study (Figure 1). The experimenter collected the first saliva sample, attached sensors and the participants relaxed for a 5-min baseline. After, the experimenter explained that the study examined “how the nature of communication has changed now that our social lives are increasingly moving online.” Participants were told that two other participants in different rooms were also completing the study and they would be communicating via a “chat” (Gmail) program. The partners’ responses were controlled by research assistants in an adjacent room. The “partners” were always the same-sex as the participant, but partner race was manipulated via their avatars, which were depicted as either Black or White individuals based on random assignment. At this point the experimenter explained that two of the individuals would give 5-min speeches about their strengths and weaknesses followed by four 2-min discussions about their opinions regarding various topics while the non-speech-giving participants listened and provided feedback via the chat program. The experimenter further indicated that the participant not chosen Discrimination and risk-taking, 8 to speak would be the moderator, whose role it was to choose who spoke first and to score performances. Fictitious Participant #3 was always “randomly selected” to be the moderator, and the moderator always chose Participant #2 – the actual participant – to speak first. The participants had 3-min to prepare and then were instructed to begin. Throughout the speech participants received rejecting feedback via the chat program. We developed a list of negative statements that were typed in real-time, but also in the interest of believability we tailored comments to be responsive to the content of participants’ speeches (e.g., “Someone’s a little high on themselves” when the participant said something positive about themselves; Figure 1). After the speech, participants completed attribution questionnaires. Next, the “partners” selected current event “hot topics” for the participant to discuss (e.g., “Are big box stores like Walmart good or bad for a community?”). The topics appeared in the chat program one at a time and the participant was instructed to discuss the topic identified for two-minutes until a new topic appeared. Again, negative feedback was typed in real time, presumably from the partners, which questioned the rationale of the argument, the quality of the participant’s speaking style, and the persuasiveness of the participant’s speech. When the speech and discussion-topics were completed, participants provided the second saliva sample (20-min after speech onset), and the experimenter explained that all participants would perform a set of cognitive tasks before Participant #1 began his/her speech. At this point, participants completed the delayed recall measure, emotional Stroop (to measure vigilance), and Columbia Card Task (in fixed order) followed by the final saliva sample (40-min after speech onset). The experimenter then informed participants that the study was over, probed for suspicion, and fully debriefed them. Measures Discrimination and risk-taking, 9 Additional detail can be found in the supplementary online material. Cardiovascular reactivity. Electrocardiography (ECG) and impedance cardiography (ICG) signals were collected, and reactivity scores were computed by subtracting responses from the final baseline minute (the “most relaxed" period) from those collected during the first minute of the speech and discussion tasks (the “most reactive” period). Analyses focused on two measures that best distinguish approach (challenge) and inhibitional (threat) states: cardiac output (CO) and total peripheral resistance (TPR) (Mendes, 2009). Neuroendocrine reactivity. Participants provided 1 ml saliva samples three times. Reactivity scores were computed by subtracting cortisol levels after the speech and recovery from those measured at baseline. Affective displays. Three female research assistants (two White, one bi-racial African-American/White) unaware of the partners’ race coded the videotaped speech and discussion topics. Coding categories were developed to determine the level of observable anger/approach and shame/withdrawal behavior. Inter-rater reliability was good to excellent (αs .75 to .89). Three items were used to compute the anger and shame composites. The anger composite included how hostile and tense participants appeared and “general anger behavior” (α =.72). The shame composite included measures of disengagement, how apologetic participants were, and “general shame behavior” (α =.81). Recall memory. Delayed recall was measured by having participants freely recall a story (Wechsler Memory Scale WMS-III) they were read at the beginning of the experiment after the rejection manipulation (timed to be 1-hour after the initial reading). Responses for details and themes were coded. Discrimination and risk-taking, 10 Attentional bias. An emotional Stroop task (MacLeod, Rutherford, Campbell, Ebsworthy, & Holker, 2002) measured vigilance for emotionally-negative information. Participants were asked to name the font color (red, green, or blue) of words as quickly and accurately as possible. Words were presented in two lists (a negative and a neutral list) of 100. An experimenter unaware of condition assignment recorded errors and how long it took participants to read each list. Interference scores were computed by subtracting the time it took participants to read the neutral list from their time on the negative list. Risk task. Participants completed the Columbia Card Task (CCT) (Figner, Mackinlay, Wilkening, & Weber, 2009), a computerized card game that has been validated as a measure of risk-taking. In this task, participants earned points by turning over as many gain cards as possible without turning over a loss card. The task manipulates probability of loss, amount of loss, and amount of gain by independently randomizing these pieces of information over 24 trials. If a loss card was chosen, the loss amount was subtracted from their score, and the trial ended. Because the loss cards represent an artificial ceiling on behavior, the number of cards turned over on nonloss trials was analyzed. Risk was operationalized as the greater number of cards turned over during non-loss trials. Data analysis Due to computer problems, two participants’ cardiovascular data and eight participants’ risk data were lost. Additionally, two colorblind participants did not complete the Stroop. Data were analyzed in 2 (Participant race: White or Black) x 2 (Partners’ Race: White or Black) ANCOVAs with the race variables as between-subjects factors and sex as a covariate. We predicted that participants’ and partners’ race would interact such that cross-race interactions would differ from same-race interactions. When we observed interactions without race main Discrimination and risk-taking, 11 effects, in the interest of space and interpretability, we present means by sameand cross-race conditions. Results Physiological reactivity Cardiovascular responses. ANCOVAs examining CO reactivity during the two tasks yielded no main effects, but significant interactions, Speech: F(1,84) = 4.32, p = .041, d = .45; Discussion: F(1,84) = 4.56, p = .036, d = .47. Participants rejected by different race partners (Whites rejected by Black partners; Blacks rejected by White partners) exhibited larger CO increases relative to participants rejected by same-race partners (Figure 2a). Analysis of TPR reactivity also produced significant interactions, Speech: F(1,84) = 6.26, p = .014, d = .55, Discussion: F(1,84) = 8.19, p = .005, d = .63. Again, cross-race rejection participants exhibited significantly lower TPR reactivity versus same-race rejection participants (Figure 2b). Taken together, out-group rejection was associated with lower TPR and greater CO reactivity than ingroup rejection indicating a challenge/approach pattern of physiological responding to discrimination. Neuroendocrine reactivity. We then examined cortisol changes as a function of participant and partners’ race. Again, there was a significant interaction, F(1,86) = 4.51, p = .037, d = .46. Participants rejected by in-group members had significantly greater increases in cortisol after the interaction tasks compared to participants receiving out-group rejection (Figure 2c). This effect persisted, albeit with smaller effect sizes and a non-significant difference, into the recovery period, F(1,86) = 2.72, p = .10, d = .36. Affective Displays Discrimination and risk-taking, 12 Displays of anger and shame were analyzed in a 2 (Emotion: anger vs. shame) x 2 (Participant Race) x 2 (Partners’ Race) mixed ANOVA with emotion type as a within-subjects variable. This analyses produced a significant 3-way interaction, F(1,86) = 7.30, p = .008, d = .58, that we deconstructed by emotion (Figure 3). As predicted, the anger index produced a significant interaction, F(1,86) = 5.64, p = .020, d = .51. Consistent with our predictions, cross-race rejection led to more observed anger than did same-race rejection. Analysis of shame displays produced a significant main effect for participant race, White participants displayed more shame than Black participants, F(1,86) = 6.45, p = .013, d = .55, and an interaction trend, F(1,86) = 2.61, p = .109, d = .35. We had predicted that same-race rejection would be associated with more shame behavior than cross-race rejection. Using simple contrasts, consistent with the prediction Black participants exhibited more displays of shame when rejected by same-race (M = .86, SD = .29) versus cross-race (M = .64, SD = .28) partners, F(1,86) = 4.10, p = .046, d = .44, but White participants exhibited no differences as a function of partners’ race, F < 1 (overall M = .97, SD = .43). WMS recall Analysis of story recall yielded no main effects but a marginal interaction, F(1,86) = 3.41, p = .07 d = .40. Participants rejected by same-race partners recalled marginally less story content (M = 16.49, SD = 5.37) than their cross-race rejection counterparts (M = 18.41, SD = 5.36). This pattern is consistent with previous findings showing that increases in cortisol affect low affinity receptors in the hippocampus impairing memory (e.g., Sapolsky, 1996; Lovallo & Thomas, 2000). Indeed, cortisol reactivity at time 3 was correlated with worse recall, β = -.21, p = .043. Attentional bias Discrimination and risk-taking, 13 Performance on the emotional Stroop task yielded a significant interaction, F(1,84) = 4.05, p = .047, d = .44. Consistent with the idea that discrimination may engender more vigilance, cross-race rejection was associated with greater attentional bias (M = 2.54, SD = 6.52) compared to those rejected by in-group members (M = -0.52, SD = 7.12). Thus, being rejected by out-group partners increased vigilance for emotionally-negative information as demonstrated by greater attentional capture for emotionally-negative words. Furthermore, this finding cannot be attributed to a speed-accuracy tradeoff as neither participant nor partner race impacted error rates on either list (negative list M = 0.80, SD = 0.93; neutral list M = 0.91, SD = 0.97), ps > .20. Risk taking We expected that cross-race rejection, which was associated with more approach/challenge responses, would be associated with increased risk-taking relative to samerace rejection. To test this hypothesis, we analyzed the mean number of cards participants turned over on the CCT. This analysis produced an interaction, F(1,78) = 4.93, p = .029, d = .50. Participants rejected by cross-race partners were riskier (i.e. turned over more cards) (M = 12.06, SD = 4.87) than those rejected by same-race partners (M = 10.18, SD = 3.37). To further explore the processes underlying risk-taking, we regressed the number of cards participants turned over on the cost/reward information (loss amount, loss number, and gain amount) nested within participants. We then examined which piece(s) of information influenced decisions. For ease, data are organized by same-race and cross-race rejection groups. Same-race rejection. Participants rejected by same-race partners turned over fewer cards when there were more loss cards in the array (3 vs. 1), β = -.42, p < .001, and when the loss amount (-750 vs. -250) was higher, β = -.18, p = .002. Differences in gain amount did not significantly influence decisions, β = .08, p = .23. Discrimination and risk-taking, 14 Cross-race rejection. Similar to those rejected by same-race partners, participants rejected by cross-race partners turned over fewer cards when loss probability, β = -.39, p < .001, and loss amount were higher, β = -.15, p = .012. Unlike the same-race rejection, however, the gain amount was associated with increased risk behavior, β = .15, p = .012. When cards were of higher value, participants who experienced cross-race rejection were riskier (i.e., turned over more cards). Compared to same-race rejection participants, those who were rejected by crossrace partners were more sensitive to rewards (gain amount), Z = 1.95, p = .051, but the type of rejection had no impact on attention to costs (loss amount or loss number), ps > .24 (Meng, Rosenthal, & Rubin, 1989). Thus, cross-race rejection participants took more risks than samerace rejection participants and also exhibited increased reward sensitivity. Discussion There are three noteworthy findings from this work. First, cross-race compared to samerace rejection elicited a distinct profile of physiological reactivity, specifically lower cortisol reactivity, larger increases in cardiac output, and decreased vascular resistance. Observers unaware of partners’ race coded more anger behavior following cross-race rejection compared to same-race rejection. These findings are consistent with intergroup and discrimination research highlighting that anger, not shame, is the dominant emotional response following perceptions and experiences of racial bias. Second, we show that reactions to cross-race rejection, compared to same-race, extend to cognitive processes like attentional bias and better memory. The latter finding is consistent with decades of research showing that cortisol increases can impair memory, and we observed significantly greater cortisol increases in same-race compared to cross-race rejection. Finally, cross-race versus same-race rejection led to increased risk-taking behavior. As an ancillary analysis, cross-race rejection was associated with greater reward Discrimination and risk-taking, 15 sensitivity, and these effects were at least partially mediated by the changes in CV reactivity (see supplemental online material for details). Consistent with previous work (e.g., Gibbons et al., 2010), this research shows that experiences of discrimination not only increase individuals’ willingness to take risks but also can directly lead to risky behavior. Cross-race rejection increased vigilance for emotionally-negative information. Functionally, vigilance facilitates the detection of danger and helps individuals respond effectively to stressors. However, attentional vigilance increases the likelihood that once discrimination is perceived, individuals may then experience increased sensitivity to threat cues, which in some cases may be adaptive but in others may lead to “false alarms” or detection of bias during ambiguous situations (Wang, Leu, & Shoda, in press). Considering that a bias for emotionally-negative information maintains feelings of anxiety and is linked to a host of clinical conditions (e.g., Mathews & MacLeod, 2002; McNally, et al., 1994), future research might consider the mental health consequences of vigilance resulting from perceptions of discrimination. Taken together, the data are consistent with the idea that out-group social rejection may influence physiological responses, behavior, and cognition. It might seem surprising that social rejection affected physiological responses without any face-to-face interaction but rather with only minimal computer interaction. These findings are potentially important as our social lives are increasingly moving online with the popularity of social networking outlets like Facebook, Twitter, and Google+. One might think that instances of social rejection online (e.g. cyber-bullying) might be less potent than receiving the same feedback face-to-face. However, in this research rejection over a chat program produced similar patterns of physiological reactivity as are observed in face-to-face situations of negative evaluation (e.g., Mendes, et al., 2008). Because computers provide users with a degree of Discrimination and risk-taking, 16 anonymity, online negative feedback might be more frequent. This typically-anonymous feedback might be more vitriolic than what would be expected if the commentators were accountable for their feedback, as would be the case in a direct, face-to-face interaction. Thus, seemingly innocuous negative comments can potentially have deleterious effects for targets whether the feedback is given over a computer or during “real world” interactions. It is important to not overlook the findings of same-race rejection, which were associated with distinct physiological and cognitive profiles. Participants rejected by members of their own race exhibited greater cortisol increases, less efficient cardiac output, increased vascular resistance, and impaired memory recall. From a physiological perspective this pattern of reactivity has been linked to accelerated “brain aging,” cognitive decline, and early risk of Alzheimer’s disease (Jefferson, et al, 2010). Thus, this research suggests that the attributions one makes for social rejection (self-blame or other-blame) might trigger different physiological pathways and possibly, over time, different diseases (Leventhal & Patrick-Miller, 2000). An important avenue for future research might be to examine the contexts, stigmas, and populations that are more likely to experience in-group compared to out-group rejection and how these responses differentially influence pathways to behaviors and mental and physical health outcomes. Discrimination and risk-taking, 17