When ostracism leads to aggression: The moderating effects of control deprivation

Journal of Experimental Social Psychology 42 (2006) 213–220 www.elsevier.com/locate/jesp

When ostracism leads to aggression: The moderating eVects of control deprivation 夽 Wayne A. Warburton a,¤, Kipling D. Williams b, David R. Cairns a a

b

Department of Psychology, Macquarie University, Sydney, NSW 2109, Australia Department of Psychological Sciences, Purdue University, West Lafayette, IN 47907, USA Received 7 March 2004; revised 27 February 2005 Available online 3 June 2005

Abstract We hypothesized that increasing or decreasing levels of control in an ostracized individual could moderate aggressive responding to ostracism. Participants were either ostracized or included in a spontaneous game of toss, and then exposed to a series of blasts of aversive noise, the onsets over which they had either control or no control. Aggression was deWned as the amount of hot sauce participants allocated to a stranger, knowing the stranger did not like hot foods, but would have to consume the entire sample. Ostracized participants without control allocated more than four times as much sauce as any other group; ostracized participants who experienced restored control were no more aggressive than either of the groups who were included. Aggressive responding to ostracism may depend on the degree to which control needs are threatened in the target, and is discussed in terms of Williams’s (2001) needs threat model of ostracism.  2005 Elsevier Inc. All rights reserved. Keywords: Ostracism; Aggression; Social exclusion; Control

Introduction What is the relationship between ostracism and aggression? Recent laboratory Wndings suggest that social exclusion can lead to increased aggression (Twenge, Baumeister, Tice, & Stucke, 2001), and an examination of 15 recent school shootings in the US, implicates “acute or chronic rejectionƒ in the form of ostracism, bullying and/or romantic rejection” as a possible causal factor in 87% of cases (Leary, Kowalski, Smith, & Phillips, 2003, p. 202). Conversely, aggressive behavior may also result in ostracism (see McDougall, 夽 An Australian Research Grant to the second author provided part of the funding for this research. This research fulWlled partial requirements for the Wrst author’s honour thesis. We would like to thank Doris McIlwain, Trevor Case, Julie Fitness, and Cassandra Govan, for their comments on the design and previous drafts. * Corresponding author. Fax: +61 2 9850 8062. E-mail address: [email protected] (W.A. Warburton).

0022-1031/$ - see front matter  2005 Elsevier Inc. All rights reserved. doi:10.1016/j.jesp.2005.03.005

Hymel, Vaillancourt, & Mercer, 2001, for a review), and so it seems feasible that for some individuals, ostracism and aggression may be mutually causative, with each feeding the other and causing a spiraling escalation in both. In a longitudinal study by Kupersmidt, Burchinal, and Patterson (1995), not only did rejection predict aggression in elementary and middle school students, but as rejection increased over time, so did aggression. If such a cycle does exist, then factors that can moderate aggressive responding to ostracism (we call this the outcast-lash-out eVect), may prevent ostracism-related aggression from escalating into outright violence. Identifying outcast-lash-out moderators is complicated, however, because of inconsistencies in the research Wndings related to rejection, social exclusion, and ostracism (see Warburton & Williams, 2005). Some studies have found that ostracism elicits pro-social responses, whereas other research using rejection and social exclusion paradigms have found anti-social

214

W.A. Warburton et al. / Journal of Experimental Social Psychology 42 (2006) 213–220

responses. It might be tempting to dismiss these diVerences as an artifact of the diVerent deWnitions and methods used. After all, ostracism refers to being ignored and excluded, and is usually manipulated experimentally in vivo, but research into both rejection (which implies initial or anticipated acceptance) and social exclusion (which implies not being included) generally uses paradigms in which excluded/rejected participants receive news of future exclusion/rejection through a third party or via feedback from a personality test. We can Wnd no reason, however, why these operationalizations should be so fundamentally dissimilar as to produce opposite results. Rather, in the absence of a better explanation, we assume that all three paradigms reliably produce perceptions of social exclusion and could produce either proor anti-social responses, but that, as yet, studies have not manipulated or assessed the types of moderating factors that can determine the direction of the sociality of responses. To address this gap, the focus of this paper will be to determine whether we can steer ostracismrelated behaviors toward or away from aggressiveness by manipulating a relevant moderating factor that underlies the direction of the response. Why should ostracism lead to aggression? Twenge et al. (2001) suggested that social exclusion might weaken normal social restraints on selWsh and aggressive behavior, thus releasing an instinctual drive to aggress. This explanation appears to predict an increase in aggressive responding to all instances of social exclusion and does not adequately explain the many Wndings in which ostracism, despite being painful (Eisenberger, Lieberman, & Williams, 2003), has elicited pro-social rather than aggressive behaviors, such as conformity (Williams, Cheung, & Choi, 2000) social compensation (Williams & Sommer, 1997); unconscious mimicry (Lakin & Chartrand, in press), and increased attention to and processing of social information (Gardner, Pickett, & Brewer, 2000; Pickett, Gardner, & Knowles, 2004). Similarly, Baumeister, Twenge, and Ciarocco (2003) suggested that because social exclusion may lead to emotional numbness, cognitive overload and self-regulatory deWcits (Twenge, Catanese, & Baumeister, 2002, 2003), it may also lead to aggression through a state of passivity in which the excluded individual’s resources are too depleted to either restrain aggressive impulses or resist group pressures. Although this hypothesis oVers an explanation for both pro-social and anti-social responses to ostracism, it cannot adequately account for Wndings that ostracism can elicit various eVortful pro-social behaviors, such as social compensation and increased processing of social information. It is possible, however, that another factor may be instrumental in the outcast-lash-out eVect. According to Williams’s (1997, 2001) model of ostracism, an

ostracized individual will experience an immediate threat to four basic human needs—belonging, control, self-esteem, and meaningful existence—and will consequently be motivated to restore those needs. Of these factors, control has considerable potential as an outcastlash-out moderator, because there are demonstrated links between control and aggression. It should be noted here that belonging may best be fortiWed by pro-social behaviors, and can also achieve a sense of control. However, when control is suYciently thwarted, desires to fortify control can outweigh desires to be liked, thus allowing aggression to be a functional response (Warburton & Williams, 2005). Control and aggression Control has been linked to aggressive behavior in various ways. Mueller (1983) theorized that a loss of personal control leads to aggression either as a reactant attempt to restore a lost freedom or as an angry response to frustration. Tedeschi and Felson’s (1994) social interaction theory portrayed aggression as a coercive tool used to control others’ behavior. Other theorists have suggested that individuals may aggress as a way of restoring a generalized sense of personal power or control over others (see Depret & Fiske, 1993; Frieze & Boneva, 2001). It also seems feasible that some acts of indirect (Richardson & Green, 1997), relational (Crick & Grotpeter, 1995), or displaced aggression (Marcus-Newhall, Pedersen, Carlson, & Miller, 2000), may be used as a means to restore a sense of personal control or power (e.g., Crick, Casus, & Mosher, 1997), perhaps through the device of symbolically asserting superiority over another (see Baumeister, Smart, & Boden, 1996; Williams & Warburton, 2003). There appears to be little experimental literature about the processes underlying control–aggression eVects, but it seems reasonable to suggest that they may involve schemas or scripts in which an aggressive response is believed to be an eVective remedy to a loss of control or power. According to script theories of aggression (e.g., Huesmann, 1986, 1998), the activation of such knowledge structures by a relevant trigger (such as a signiWcant loss of control), would set in motion linked goals and action plans, and these, in turn, may increase the likelihood of an aggressive response. Further, if an individual has control–aggression scripts that are chronically accessible, they might automatically respond to a loss of control trigger with aggression, and with little conscious awareness of the processes underlying their behavior (see Todorov & Bargh, 2002). Consistent with such a mechanism, recent Wndings in our laboratory show that individuals vary in the degree to which they hold control–aggression beliefs, and that higher levels of such beliefs predict greater aggressive responding to a control loss.

W.A. Warburton et al. / Journal of Experimental Social Psychology 42 (2006) 213–220

If control–aggression eVects are underpinned by such processes, they may be best mapped onto Anderson and Bushman’s (2002) General Aggression Model (GAM). The GAM proposes that inputs from the person (e.g., control–aggression beliefs, other relevant traits) and the situation (such as a perceived loss of control) combine to inXuence the individual’s present internal state, by activating relevant cognitions (e.g., control–aggression schemas or scripts) or aVects, or by increasing levels of arousal. These three components (which may aVect each other) inXuence the subsequent appraisal and decision-making processes that may ultimately lead to aggressive behaviors. The GAM diVers from many previous models of aggression (e.g., Berkowitz, 1993; Geen, 1990), by allowing for cognitions to mediate aggression without a preceding increase in negative aVect. Thus, it is possible within the GAM for a loss of control trigger to lead to aggressive behavior without necessarily aVecting an individual’s mood. This is important for the purposes of this paper, because the exclusion-aggression eVects found by Twenge et al. (2001) were not accompanied by a lowering of mood. Ostracism, control, and aggression Given the links between control and aggression, we propose that the extent to which ostracism elicits aggression may be partly determined by the level of perceived personal control an individual has after an ostracism episode. Although ostracism alone threatens personal control, this threat may not be enough to override any barriers to aggression. For instance, the paradigms used by Williams and co-workers involve a short-term episode of being ignored and excluded in a (virtual or real) ball toss game. Such ostracism deprives control during the four minutes of ostracism, but does not imply a loss of control in the future. If, however, subsequent events further undermine the ostracized individual’s sense of control, the need to restore control may become a strong enough trigger to elicit an aggressive response. Interestingly, the Wndings of Twenge et al. (2001) seem to be somewhat consistent with such an explanation. Twenge et al., used two paradigms; the life-alone paradigm in which participants were given a prognosis of a future in which they would have no permanent close bonds with others, and the group-interaction paradigm in which a group of real participants began the experiment with a get-acquainted task, followed by feedback that none of the others wanted to work with the participant. We believe that both of the paradigms involve a substantial threat to control beyond simple exclusion. The life-alone paradigm, if believed by the participants, suggests that there is nothing they can do to change this outcome, thus greatly restricting their sense of control. Group-interaction participants probably experience the getting-acquainted task as pleasant and enjoyable, receiving positive feedback from the other real partici-

215

pants. Thus, when they receive rejection feedback following this activity it is likely to be both surprising and unexpected. This robs participants of two types of control: (1) direct control (i.e., each participant’s fate was determined by the choices of the other participants in the task) and (2) predictability (i.e., participants’ perceptions of the getting-acquainted activity were unrelated to the feedback that they ultimately received). It may even undermine their conWdence in their ability to read the verbal and nonverbal signs of others. Testing control motivation as an outcast-lash-out moderator The primary focus of the current experiment was to test whether subsequent loss or restoration of control would moderate aggressive responses to ostracism. Participants were tested for aggressive responses after they had experienced either ostracism or inclusion in an apparently impromptu game of toss, and had experienced either restored control or diminished control over an aversive noise. Process measures for mood and arousal were also collected, but it was decided not to administer substantial measures for any constructs that may have primed participants to the true purpose of the experiment (e.g., control–aggression beliefs, desire for control, etc.), because of the risk of contaminating the more important behavioral manipulations and measures. We predicted that ostracized participants who experienced diminished control would have a strong urge to aggress, but that ostracized participants who had experienced restored control would have satisWed their control deprivation, and hence, would have little inclination to aggress. Because the underlying processes may be cognitive in nature, we further predicted that these eVects would not be substantially due to lowered mood or increased levels of arousal.

Method Participants and design Forty undergraduate psychology students (23 females, 17 males; mean age D 21.58, SD D 5.00) participated in the study in return for course credit. Participants were randomly assigned to one condition of a 2 (inclusion or ostracism) £ 2 (restored or diminished control) between-S design. Because gender eVects are well demonstrated in aggression research (e.g., Bettencourt & Miller, 1996), participants were randomly allocated to groups within gender. Procedure Participants were informed that they were taking part in a taste preferences experiment, and were seated in a

216

W.A. Warburton et al. / Journal of Experimental Social Psychology 42 (2006) 213–220

triangular formation with two other ostensible participants, who were actually confederates. They were asked to provide details about their age, sex and taste preferences and then Wlled out a mood and arousal questionnaire which asked participants to rate on a 7-point bipolar scale how they were feeling “at this very moment” in terms of being happy–sad, pleased–annoyed, unfriendly–friendly, afraid–unafraid, contented–discontented (mood items), bothered–untroubled, worried– indiVerent, relaxed–tense, agitated–calm (mood items, also reverse scored to measure stress arousal), active– inactive, drowsy–alert, and lethargic–energetic (physiological arousal items). Participants were then asked to wait quietly while the experimenter made sampling arrangements with three other participants in another part of the building. During the experimenter’s absence, the naïve participant experienced either ostracism or inclusion in an apparently impromptu game of toss. The ostracism-inclusion manipulation In a pre-arranged script, one confederate removed a toy from a nearby basket marked Children’s Play Study, and began a game in which the toy was tossed from one person to the other, with the vigor and hilarity of the game increasing over time. In the ostracized condition, participants received three tosses, but then found themselves shut out of the game as the two confederates turned their shoulders toward each other, stopped making eye-contact with the participant, and continued the game without them. Included participants, however, received about a third of the tosses for the remainder of the 4-min period. After a total of 4 min the experimenter returned to the room. The same confederates (one male, one female) were used in all trials. The confederates were then told they would move to other rooms in order to complete the study so as to remove biasing inXuences during the taste-perception phase of the experiment. After this, participants were given the cover story that their sensory modality of hearing would be Xooded prior to the actual taste-perception phase of the experiment. The diminished or restored control manipulation Participants listened to an amalgam of unpleasant sounds (i.e., chalk squeaking on a blackboard, metal scraping on metal, and several high-pitched human screams) through headphones, and played on a portable CD player. Peak volume was 108 dBs. Participants in the diminished control condition heard 10 blasts of the noise, which came at random intervals ranging from 3 to 33 s (and totaling 150 s), and in random lengths ranging from 1 to 11 s (and totaling 50 s). Their instructions were as follows: “You are now going to listen to some sounds that will be used to saturate your sensory modality of hearing. The sound that you listen to may be unpleasant. You

will be asked to close your eyes and attend to each sound, whenever it comes, as closely as possible and for as long as it lasts. You will have no control over any aspects of these sounds—their type, length or timing. Nevertheless, I want you to give your closest attention to each sound as it comes, and to keep your eyes closed until I ask you to open them again. Now please put on the headphones, close your eyes and wait for the Wrst sound.” Those in the restored control condition were asked to listen to a 5-s track of the noise 10 times (50 s in total), but were given control over its administration and timing via the CD player’s remote control unit. They were told that they could play the tracks whenever they felt ready, but were asked to not allow the gaps between administrations to exceed 15 s (a total of 150 s). Their instructions were as follows: “You are now going to listen to some sounds which will be used to saturate your sensory modality of hearing. The sound that you listen to may be unpleasant. You will be asked to attend as closely as possible to one sound, 5 s long, 10 times. You will have complete control over when each sound is administered. I would ask you, however, to keep the breaks between sounds to no more than 15 s or so. When you are ready to hear each sound, just press play on the remote control and attend to the sound till it stops. Then press the stop button. Please repeat this pattern until all 10 trials are over. You may ask at any time how many trials are left, and for the last 3 trials I will hold up Wngers to indicate how many trials there are to go. Please put on your headphones and we will begin. When you are ready, you may begin your trials at any time from now.” Manipulation checks Following this, participants completed the mood/ arousal questionnaire again. Because we wished to avoid contaminating the behavioral responses by priming participants with ideas related to the true purposes of the experiment, manipulation checks were carried out retrospectively at this point, among a number of Wller items. Participants were asked to indicate the extent to which they currently felt in control–not in control on a 7-point, bipolar scale, and to mark on a 7-point scale (where 1 D not at all true and 7 D very much true) the extent to which they felt excluded and noticed “during the waiting period prior to the noise task” (i.e., during the time when the ostracism manipulation occurred). The aggression measure Using an adaptation of the procedure used by Lieberman, Solomon, Greenberg, and McGregor (1999), aggression was measured as the amount of hot sauce participants would allocate to a stranger knowing that (a) the target strongly disliked hot and spicy foods,

W.A. Warburton et al. / Journal of Experimental Social Psychology 42 (2006) 213–220

(b) the target would have to consume all of the sample that they packed, and (c) the sauce was very hot. Our cover story continued by explaining to participants that because of human sensitivity to taste cues, and the diYculty this posed for getting unbiased responses in taste experiments, we needed them to package the taste samples for each other with complete anonymity. All participants were asked whether they would be willing to help out in this way (none refused). A hat was then produced, ostensibly for the purpose of randomly selecting the sample type that they would be packaging. Participants withdrew a folded slip which always revealed the words, “Category 5: Hot and Spicy”. Participants were then given a taste sample in a sealed Styrofoam cup (three plain crackers), which they were asked to consume entirely and evaluate once the experimenter had left the room (all participants complied). Following this, participants were presented with a stainless steel bowl half-Wlled with the hot sauce, a sealed envelope, empty Styrofoam cups, tightly Wtting lids, napkins, and a teaspoon. Because the experimenter would be absent during the packaging of the food sample, participants were asked to allocate the hot sauce according to the instructions on a written sheet. This sheet made it clear that the taster would be required to consume the entire sample that they packed, and that participants should read the taster’s taste preferences from the sheet in the sealed envelope, to ascertain whether or not that person was allergic to the sample they would give. To do this, participants needed to read the hot and spicy sauce item, which revealed that the target had rated hot and spicy foods at “3” on a 21-point scale (1 D extreme dislike, 21 D extreme liking). The sauce allocation instructions were similar to those used by Lieberman et al. (1999, p. 339): “All quantities of the sample food are useful. From the larger sample provided, put into the cup as much or as little of the food sample as you want to.” Participants were also asked to place the lid Wrmly on the cup for the purpose of anonymity. After hot sauce allocation, participants were asked to specify on a 21-point scale the rating that the taster had given for hot and spicy foods. The cup of sauce was weighed in an adjoining room on a scale, which provided a digital readout of weights in one-gram increments. Finally, participants were asked what they had thought the experiment was about (none guessed its true purpose), and were thoroughly debriefed and thanked.

Results Manipulation check The two ostracism items have been reliably combined in previous ostracism research (e.g., Zadro, Williams, & Richardson, 2004), and were thus combined into a single

217

measure. Ostracized participants reported feeling less included and noticed (M D 3.95, SD D 1.13) than did included participants (M D 5.23, SD D 1.27, F (1, 36)D 10.81, p D .001), demonstrating that the ostracism manipulation was successful. Means for the control manipulation check were in the expected direction, with the feelings of control reported by participants in the restored control condition (M D 4.05; SD D 1.57) being somewhat but not signiWcantly higher than those in the diminished control condition (M D 3.15; SD D 1.85, F (1, 36) D 2.67, p D .11). There were no other signiWcant main eVects or interactions for these variables. In order for the hot sauce measure to be considered valid, participants needed to understand that the taster did not like hot and spicy foods, and to believe that the sauce was hot enough to harm another in quantity. Participants had clearly known that the target did not like hot foods, as they gave an average rating of 3.13 out of 21 (SD D 1.84) for the target’s hot food preference, and this did not diVer from the true answer, 3.0, t (39) D .43, ns. The hotness and damage potential of the hot sauce was tested in an addendum to this experiment. Fifteen new participants sampled the sauce and rated it as being very hot (M D 7.13, SD D 1.46, on a 9-point scale). Participants were then shown two cups, containing 27 and 6 g of the same sauce, the rounded mean allocation of the most aggressive group, and the rounded mean allocation for the other three groups, respectively (see Fig. 1 for the stimuli used). When asked the extent to which “consuming that amount of sauce would cause pain and discomfort,” on a scale where 1 D not much discomfort and 9 D extreme discomfort, they indicated that 27 g would cause a great deal of pain and discomfort (M D 7.67, SD D 1.18), but that 6 g would cause comparatively little discomfort (M D 3.13, SD D 1.77; t (14) D 14.09, p < .0005). Did subsequent restored or diminished control moderate aggression? As predicted, the sauce allocation of the ostracism diminished control group (M D 27.50, SD D 25.98) was far greater than that of any other group (Ms D 7.30, 6.80, 4.80; SDs D 10.03, 6.88, 2.44, respectively). The obtained data distribution was skewed however, and so, in line with the method used by Lieberman et al. (1999), a log 10 data transformation was made. The new data set was normally distributed and was used as the dependent measure for aggression in all further calculations. As shown in Fig. 2, ostracized participants were more aggressive than included participants, F (1, 36) D 7.09, p D .012, and participants with diminished control were more aggressive than those with restored control, F (1, 36) D 6.07, p D .019. As predicted, there was a signiWcant interaction between the ostracism and control conditions, F (1, 36) D 6.73, p D .014 (d D 1.37), such that participants who were ostracized and experienced

218

W.A. Warburton et al. / Journal of Experimental Social Psychology 42 (2006) 213–220

Fig. 1. Average hot sauce allocated across the inclusion conditions and the ostracized restored control condition (on the left) compared with hot sauce allocated in the ostracism diminished control condition (on the right).

Fig. 2. Allocation of hot sauce by ostracism group and control group after log 10 transformation of the data. Standard deviations are in brackets.

manipulations, new scales were created by subtracting the pre-manipulation scores from the post-manipulation scores, so that a negative score would indicate that levels of the relevant variable had dropped and a positive score the reverse. Over the course of the experimental manipulations participants experienced a drop in mood (M change D ¡1.06, SD D .96, t (39) D ¡8.69, p < .0005), an increase in stress arousal (M change D 1.31, SD D 1.07, t (39) D 7.74, p < .0005) but no change in physiological arousal (M change D .28, SD D 1.42, t (39) D 1.22, ns). However, none of these changes diVered across groups, with no signiWcant main eVects or interactions for any variables (all Fs (1, 36) < 2.48, all ps ns). When the primary analysis was re-run with all three change measures held constant as covariates, the interaction was still signiWcant, F (1, 36) D 5.16, p D .030, providing no evidence for mood or arousal moderation.

Discussion diminished control allocated far more hot sauce than any other group. Did aggressive participants report a bigger drop in mood or higher levels of arousal? The nine mood items, four stress arousal items, and three physiological arousal items were combined (within construct) to create three single scales which all had adequate internal consistency (coeYcient
D .80, .70, and .75, respectively, for pre-manipulation data and .88, .76, and .76 for post-manipulation data). To determine whether levels of these variables changed during the

The primary aim of this experiment was to test whether restoring or diminishing control following ostracism would inXuence subsequent aggression. This hypothesis received strong support. Those who were ostracized and also experienced a further loss of control over aversive noise were (using the raw data averages) more than four times as aggressive as any other group. In contrast, those who were ostracized but were allowed to take control over the same noise were no more aggressive than participants who had been included. These Wndings suggest that by providing individuals with a means to restore their sense of personal control after

W.A. Warburton et al. / Journal of Experimental Social Psychology 42 (2006) 213–220

ostracism, they will show no obvious tendency to aggress. On the other hand, if after being ostracized individuals experience a further loss of control, then they will be quite aggressive. These results are consistent with previous research which has found that ostracized individuals experience a signiWcantly lowered sense of control (see Williams, 2001; for a summary of this research) and demonstrates that not only can ostracism lead to aggression, as Twenge et al. (2001) found, but that this eVect can be moderated by restoring personal control. It may seem surprising that even though the behavioral eVect for control was robust, the scores on the self-report measure for levels of perceived control (i.e., the control manipulation check), while in the expected direction, failed to reach signiWcance. It is important to note here that there are well-documented diYculties with overtly measuring personal control constructs, especially when using the generic word “control” (Skinner, 1996). Thus, we are inclined to think it is likely that the self-report measure under-estimated the experience of control. These Wndings cannot be adequately explained by the social disinhibition hypothesis (Twenge et al., 2001), because both ostracism groups experienced the same degree of social exclusion, and yet only those who also experienced diminished control displayed heightened aggression. This dichotomy in responding also undermines to some extent the passivity hypothesis and the self-control deWcits hypothesis (Baumeister et al., 2003). The ostracized-restored control group experienced the same degree of social exclusion and the same amount of noise as did the ostracized-diminished control group, and, importantly, the two groups did not diVer in the degree to which they experienced negative emotions or heightened arousal over the course of the experiment. Thus, while it cannot be ruled out that that increased cognitive and emotional load contributed to the eVects, it seems unlikely that these factors could fully explain the large diVerence in aggression found between these two groups. The results seem to be more consistent with a needsthreat explanation (Williams, 1997, 2001). The eVect of the ostracism episode may have been to predispose participants to regain control. In previous research in our labs however, ostracism alone has led to relatively prosocial responses. We think that the barriers preventing aggression were toppled when ostracized individuals endured yet another loss of control, and that in response to this enhanced control threat, participants produced the surprisingly strong display of aggression that was found. The processes underlying these aggressive responses appear not to have involved an increase in negative aVect, stress arousal or physiological arousal, a Wnding that is consistent with the Twenge et al. (2001) null-Wnding for a mood mediation of the outcast-lash-out eVect. In terms of the three routes to aggression proposed in

219

Anderson and Bushman’s (2002) GAM, these Wndings suggest that the aggressive behaviors demonstrated by the ostracized-diminished control group were primarily driven by cognitions of some sort, and are consistent with recent revisions to the GAM to allow for the possibility that aggressive cognitions may be activated without a preceding increase in negative aVect. Because cognitive process measures could not be collected without the risk of contaminating the behavioral data, we can only speculate about the nature of any beliefs, schemas or scripts that may have elicited the aggressive responses found, but it seems reasonable to suggest that future research examining the role of control–aggression schemas may be fruitful. One curious Wnding was that individuals who were included, yet were subjected to uncontrollable noise blasts, showed no evidence of aggression. Research by others (e.g., Donnerstein & Wilson, 1976) has shown that responses to uncontrollable noise per se are suYcient to produce aggression. Our conjecture is that in our study, included individuals had a relatively enjoyable experience prior to the loss of control over the aversive stimulus, and this may have been suYcient to oVset its eVects. Limitations As with most experimental tests of aggression, the Wndings from the study do not necessarily generalize to situations outside the laboratory (see Tedeschi & Bond, 2001). It is especially important to note the contextual diVerences between the aggressive responses found here, which were made in a respectable university laboratory, and the unsanctioned acts of violence of the school shooters. Implications and conclusion Nevertheless, these Wndings may have an important application in terms of those circumstances in which an individual enters into the hypothesized cycle of social isolation and aggressive or violent behavior. If ostracized individuals could be identiWed using criteria from Williams’s (2001) model (e.g., children observed to be frequently and clearly excluded in schools), it is feasible that by assisting them to regain control in some salient area of their lives, it may be possible to decrease the likelihood that they will aggress when ostracized, and thus reduce the potential for an ostracism–aggression based cycle of violence, or, if an existing cycle exists, to interrupt it. Ironically, the most frequent method of disciplining such children is through further isolation (time out or expulsion), both of which may diminish control enough to encourage aggressive responses. In summary, these Wndings have potential ramiWcations for important social phenomena, such as cycles of ostracism and aggression, time-out, and some acts of

220

W.A. Warburton et al. / Journal of Experimental Social Psychology 42 (2006) 213–220

mass violence. Further research providing insights into the processes underlying the outcast-lash-out eVect, or detailing the boundary conditions under which it will occur, may thus prove to be valuable in developing strategies to moderate ostracism-related aggression.

References Anderson, C. A., & Bushman, B. J. (2002). Human aggression. Annual Review of Psychology, 53, 27–51. Baumeister, R. F., Smart, L., & Boden, J. M. (1996). Relation of threatened egotism to violence and aggression. Psychological Review, 103, 5–33. Baumeister, R. F., Twenge, J. M., & Ciarocco, N. (2003). The inner world of rejection: EVects of social exclusion on emotion, cognition, and self-regulation. In J. P. Forgas & K. D. Williams (Eds.), The social self: Cognitive, interpersonal and intergroup processes (pp. 161–174). New York: Psychology Press. Berkowitz, L. (1993). Pain and aggression: some Wndings and implications. Motivation and Emotion, 17, 277–293. Bettencourt, B. A., & Miller, N. (1996). Gender diVerences in aggression as a function of provocation: A meta-analysis. Psychological Bulletin, 119, 422–447. Crick, N. R., Casus, J. F., & Mosher, M. (1997). Relational and physical aggression in preschool. Developmental Psychology, 33, 579–588. Crick, N. R., & Grotpeter, J. K. (1995). Relational aggression, gender, and social-psychological adjustment. Child Development, 66, 710– 722. Depret, E., & Fiske, S. T. (1993). Social cognition and power: Some cognitive consequences of social structure as a source of control deprivation. In G. Weary, F. Gleicher, & K. L. Marsh (Eds.), Control motivation and social cognition (pp. 176–202). New York: Springer-Verlag. Donnerstein, E., & Wilson, D. W. (1976). The eVects of noise and perceived control upon ongoing and subsequent aggressive behavior. Journal of Personality and Social Psychology, 34, 774–783. Eisenberger, N. I., Lieberman, M. D., & Williams, K. D. (2003). Does rejection hurt? An fMRI study of social exclusion. Science, 302, 290–292. Frieze, I. H., & Boneva, B. S. (2001). Power motivation and motivation to help others. In A. Y. Lee-Chai & J. A. Bargh (Eds.), The use and abuse of power (pp. 75–89). Philadelphia, PA: Psychology Press. Gardner, W. L., Pickett, C. L., & Brewer, M. B. (2000). Social exclusion and selective memory: How the need to belong inXuences memory for social events. Personality and Social Psychology Bulletin, 26, 486–496. Geen, R. G. (1990). Human aggression. PaciWc Grove, CA: Brooks Cole. Huesmann, L. R. (1986). Psychological processes promoting the relation between exposure to media violence and aggressive behavior by the viewer. Journal of Social Issues, 42, 125–140. Huesmann, L. R. (1998). The role of social information processing and cognitive schema in the acquisition and maintenance of habitual aggressive behavior. In R. G. Geen & E. Donnerstein (Eds.), Human aggression: Theories, research and implications for social policy (pp. 73–109). San Diego, CA: Academic Press. Kupersmidt, J., Burchinal, M., & Patterson, C. J. (1995). Developmental patterns of childhood peer relations as predictors of externalizing behavior problems. Development and Psychopathology, 7, 825– 843. Lakin, J. L., & Chartrand, T. L. (in press). Exclusion and nonconscious behavioral mimicry. In K. D. Williams, J. P. Forgas, W. von Hippel (Eds.), The social outcast: ostracism, social exclusion, rejection, and bullying. Psychology Press, New York.

Leary, M., Kowalski, R. M., Smith, L., & Phillips, S. (2003). Teasing, rejection, and violence: Case studies of the school shootings. Aggressive Behavior, 29, 202–214. Lieberman, J. D., Solomon, S., Greenberg, J., & McGregor, H. A. (1999). A hot new way to measure aggression: Hot sauce allocation. Aggressive Behavior, 25, 331–348. Marcus-Newhall, A., Pedersen, W. C., Carlson, M., & Miller, N. (2000). Displaced aggression is alive and well: A meta-analytic review. Journal of Personality and Social Psychology, 78, 670–689. McDougall, P., Hymel, S., Vaillancourt, T., & Mercer, I. (2001). The consequences of childhood peer rejection. In M. Leary (Ed.), Interpersonal rejection (pp. 213–247). New York: Oxford University Press. Mueller, C. W. (1983). Environmental stressors and aggressive behavior. In R. G. Geen & E. I. Donnerstein (Eds.), Aggression: theoretical and empirical reviews, Vol. 2: issues in research (pp. 51–76). New York: Academic Press. Pickett, C. L., Gardner, W. L., & Knowles, M. (2004). Getting a cue: The need to belong and enhanced sensitivity to social cues. Personality and Social Psychology Bulletin, 30, 1095–1107. Richardson, D. R., & Green, L. R. (1997). Circuitous harm: Determinants and consequences of nondirect aggression. In R. M. Kowalski (Ed.), Aversive interpersonal behaviors (pp. 171–189). New York: Plenum Press. Skinner, E. A. (1996). A guide to constructs of control. Journal of Personality and Social Psychology, 71, 549–570. Tedeschi, J. T., & Bond, M. H. (2001). Aversive behavior and aggression in cultural perspective. In R. Kowalski (Ed.), Behaving badly: Aversive behaviors in interpersonal relationships (pp. 257–293). Washington, DC: American Psychological Association. Tedeschi, J. T., & Felson, R. B. (1994). Violence, aggression and coercive actions. Washington, DC: American Psychological Association. Todorov, A., & Bargh, J. A. (2002). Automatic sources of aggression. Aggression and Violent Behavior, 7, 53–68. Twenge, J. M., Baumeister, R. F., Tice, D. M., & Stucke, T. S. (2001). If you can’t join them, beat them: EVects of social exclusion on aggressive behavior. Journal of Personality and Social Psychology, 81, 1058–1069. Twenge, J. M., Catanese, K. R., & Baumeister, R. F. (2002). Social exclusion causes self-defeating behavior. Journal of Personality and Social Psychology, 83, 606–615. Twenge, J. M., Catanese, K. R., & Baumeister, R. F. (2003). Social exclusion and the deconstructed state: Time perception, meaninglessness, lethargy, lack of emotion and self-awareness. Journal of Personality and Social Psychology, 85, 409–423. Warburton, W. A., & Williams, K. D. (2005). Ostracism: When competing motivations collide. In J. P. Forgas, K. D. Williams, & S. Laham (Eds.), Social motivation: Conscious and unconscious processes (pp. 294–313). New York: Cambridge University Press. Williams, K. D. (1997). Social ostracism. In R. M. Kowalski (Ed.), Aversive interpersonal behaviors (pp. 133–170). New York: Plenum Press. Williams, K. D. (2001). Ostracism: The power of silence. New York: The Guilford Press. Williams, K. D., Cheung, C. K. T., & Choi, W. (2000). Cyberostracism: EVects of being ignored over the Internet. Journal of Personality and Social Psychology, 79, 748–762. Williams, K. D., & Sommer, K. L. (1997). Social ostracism by one’s coworkers: Does rejection lead to social loaWng or compensation. Personality and Social Psychology Bulletin, 23, 693–706. Williams, K. D., & Warburton, W. A. (2003). Ostracism: A form of indirect aggression that can result in aggression. International Review of Social Psychology, 16, 101–126. Zadro, L., Williams, K. D., & Richardson, R. (2004). How low can you go? Ostracism by a computer lowers belonging, control, self-esteem, and meaningful existence. Journal of Experimental Social Psychology, 40, 560–567.