r evolved disease-avoidance mechanisms and contemporary ...schaller/faulkneretal2004.pdf ·...

Evolved Disease-AvoidanceMechanisms andContemporary Xenophobic Attitudes

Jason Faulkner, Mark Schaller, Justin H. Park and Lesley A. DuncanUniversity of British Columbia

From evolutionary psychological reasoning, we derived the hypothesis that chronic andcontextually aroused feelings of vulnerability to disease motivate negative reactions to foreignpeoples. The hypothesis was tested and supported across four correlational studies: chronicdisease worries predicted implicit cognitions associating foreign outgroups with danger, andalso predicted less positive attitudes toward foreign (but not familiar) immigrant groups. Thehypothesis also received support in two experiments in which the salience of contagious diseasewas manipulated: participants under high disease-salience conditions expressed less positiveattitudes toward foreign (but not familiar) immigrants and were more likely to endorse policiesthat would favor the immigration of familiar rather than foreign peoples. These results reveal apreviously under-explored influence on xenophobic attitudes, and suggest interesting linkagesbetween evolved disease-avoidance mechanisms and contemporary social cognition.

keywords disease, immigration, intergroup attitudes, threat

Group Processes &Intergroup Relations

2004 Vol 7(4) 333–353

History teaches us that society has no shortage ofmeans available to dehumanize ‘undesirable’groups. The grave risks of this process are magni-fied when combined with the threat of infectiousdisease. (Markel, 1999)

HISTORICAL analyses of stereotypes, prejudiceand exclusionary intergroup behavior reveal atendency to associate foreign peoples withdisease. Ethnic outgroups are often blamed foroutbreaks of epidemic diseases, and these out-breaks can inspire violently xenophobic reac-tions to outsiders (Goldhagen, 1996; Markel,1999; Oldstone, 1998). Foreigners are alsoassociated with semantic concepts that connote

disease. This association is evident in xenopho-bic propaganda, in which ethnic outgroups arelikened to non-human vectors of disease, suchas rats, flies, and lice (Suedfeld & Schaller,2002). The associative link between foreignpeoples and disease shows up consistently inthe social science literature on immigration.

GPIR

Copyright © 2004 SAGE Publications(London, Thousand Oaks, CA and New Delhi)7:4; 333–353; DOI: 10.1177/1368430204046142

Author’s noteAddress correspondence to Jason Faulkner,Department of Psychology, University ofBritish Columbia, 2136 West Mall, Vancouver,BC V6T 1Z4, Canada [email: [email protected]]

04 046142 3/9/04 3:33 pm Page 333

www.sagepublications.com

Anti-immigrant literature in ancient Romelikened outsiders to detritus and scum (Noy,2000). Anti-immigrant attitudes in Englandhave been linked to beliefs that foreigners eatdisturbingly unclean foods (roots, frogs, haggisand other foods ‘fit only for dogs’; Stat, 1995,p. 192). And in the United States, Markel andStern (2002, p. 757) write, ‘Despite thedramatic changes in demography, the meaningof citizenship, and the ability to treat and cureacute and chronic diseases, foreigners were con-sistently associated with germs and contagion.’

Remarkably, this topic has received virtuallyno empirical attention in the psychologicalliterature on intergroup stereotypes and preju-dice. There are, of course, many additionalpsychological processes implicated in inter-group relations, and so there are many otherunrelated personality factors and contextualvariables that may contribute to xenophobicattitudes (see Fiske, 1998, for a review). In fact,there are many other kinds of disease-irrelevantthreats that can contribute to intergroup anxi-eties and prejudices (e.g. Stangor & Crandall,2000; Stephan, Ybarra, & Bachman, 1999). Still,given the apparent historical importance ofdisease in shaping societal-level reactions toforeign peoples, it seems fruitful to considerclosely its analogous role in the psychology ofindividuals. That is the purpose of this article.We employ an evolutionary framework toderive—and then empirically test—the hypoth-esis that people who feel most vulnerable tocontagious diseases are likely to react especiallynegatively toward subjectively foreign (but notfamiliar) outgroups.

Disease-avoidance mechanisms andintergroup attitudes

A number of theoretical models have suggestedways in which contemporary intergroup cogni-tion and behavior may reflect psychologicalmechanisms that evolved a long time ago (e.g.Kurzban & Leary, 2001; Neuberg, Smith, &Asher, 2000; Reynolds, Falger, & Vine, 1987).The common evolutionary logic underlyingthese perspectives is summarized by Kurzbanand Leary (2001): (a) The contemporary

human mind evolved in response to adaptiveproblems imposed by the environments inwhich ancestral populations lived; (b) for thepast several million years of this evolutionaryhistory, ancestral populations lived in socialgroups; (c) specific adaptive problems associ-ated with group life may have given rise tospecific psychological mechanisms that facili-tate negative perceptions of and reactions toindividuals associated with specific sorts ofgroups. In other words, certain prejudicial atti-tudes may have conferred adaptive benefitswithin ancestral environments, and now—eventhough contemporary environments are verydifferent in very many ways—the processesunderlying those prejudicial attitudes persist.

Within this broad evolutionary framework,different theoretical models identify linksbetween specific adaptive problems of the pastand specific psychological processes in thepresent. One model links the adaptive utility ofcooperative coalitions with contemporary ten-dencies to exclude and exploit members of out-groups (Kurzban & Leary, 2001). Anothermodel links the adaptive utility of status hierar-chies with contemporary tendencies to expressdomination over outgroups (Sidanius & Pratto,1999). Another model links the adaptive utilityof avoiding potentially dangerous intergroupencounters with contemporary tendencies toperceive outgroups in derogatory ways(Schaller, 2003). Yet another model links theadaptive utility of avoiding harmful pathogensand parasites with contemporary prejudicesagainst individuals who are perceived, at someheuristic level, to be potential carriers ofpathogens or parasites (Kurzban & Leary, 2001;Park, Faulkner, & Schaller, 2003). It is thisdisease-avoidance model specifically thatinforms the current investigation.

Disease-avoidance and prejudicial person perceptionPeople typically experience feelings of uneasi-ness and even disgust when they encounterindividuals perceived to be carriers of conta-gious disease. These psychological experi-ences—and the behavioral avoidance that theyprecipitate—may have evolutionary roots

Group Processes & Intergroup Relations 7(4)

334

04 046142 3/9/04 3:33 pm Page 334

(Kurzban & Leary, 2001; Rozin, Haidt,McCauley, & Imada, 1997). The logic is straight-forward. In ancestral environments, manycommunicable parasites and diseases were haz-ardous to individuals’ health. Behavioral ten-dencies that helped individuals to avoidinfection would have been adaptive. Oneadaptive tendency would have been the selectiveavoidance of individuals who were especiallylikely to be carriers of contagious diseases.

A prerequisite for behavioral avoidance issome set of psychological mechanisms thatfacilitate learning and detection of cues thatheuristically indicate the possible presence ofcontagious diseases. Thus, individuals may beperceptually sensitive to superficial physicalfeatures such as lesions, disfigurements, andother morphological abnormalities. (Cues suchas these are, of course, highly imperfect indi-cators of disease; however, evolution of a cue-recognition mechanism requires only a modestcorrelation between the presence of the cueand the actual presence of disease; see Kurzban& Leary, 2001.) In addition to these and othermorphological cues that heuristically connotedisease, it is also likely that individuals areespecially sensitive to superficial cues indicatingthat an individual is subjectively foreign toone’s own group. There are several reasons whysubjectively foreign peoples might have becomeheuristically associated with disease. First,contact with previously unknown populationsbrings with it an increased risk of encounteringcontagious diseases to which one has noacquired immunity (Diamond, 1999). Also,individuals from foreign cultures are morelikely to engage in subjectively strange customs(pertaining to hygiene, food preparation, etc.)that may violate local customs that inhibit trans-mission of disease. Consequently, individualsare likely to be especially adept at learning todetect a wide range of cues that—within theirlocal social environment—discriminate betweenfamiliar and foreign peoples.

Cue learning and perception is, by itself,insufficient to cause behavioral avoidance.Another prerequisite is the activation ofpsychological mechanisms that translate per-ception into action. Thus, in order to facilitate

behavioral avoidance, it is likely that there areevolved mechanisms through which the per-ception of disease-connoting cues elicits aspecific affective reaction, and activates cogni-tions in which the perceptual object is associa-tively linked to semantic concepts that connotedisease and more generally motivate behavioralavoidance. The affective reaction is likely to bethat of disgust. Disgust probably evolved pri-marily as a means to inhibit the ingestion ofpoisons and other harmful agents, but it playsan important role in interpersonal avoidance aswell; disgust is elicited by subjectively ‘offensive’people and motivates physical distancing fromthose people (Rozin, Haidt, & McCauley,2000). Schiefenhövel (1997) notes that peopleoften display disgust reactions when speakingabout ethnic outgroups, and Rozin et al. (1997,p. 73) suggest that ‘disgust in humans serves asan ethnic or outgroup marker’. Cognitiveresponses are likely to be varied. They mayinclude a wide variety of moralist judgmentsand beliefs (Haidt, 2001). They may alsoinclude stereotypical beliefs in which disease-relevant characteristics such as ‘dirty’ or ‘unhy-gienic’ are attributed to subjectively foreigngroups. Most obviously, perhaps, these cogni-tive responses are likely to precipitate xeno-phobic attitudes—attitudes indicating thatforeigners are to be avoided, excluded, andotherwise kept at a distance.

Moderating influences on disease-avoidantreactions to foreign peoplesResponses designed over evolutionary time toavoid recurrent threats not only confer poten-tial functional benefits; they entail potentialcosts as well (e.g., avoidant behavior consumescalories and distracts resources from otherfunctional activities). Therefore, these responsesare more functional in some situations thanothers. They are most functional under con-ditions in which the benefits most clearlyoutweigh the costs. As with most other evolvedpsychological mechanisms, evolved disease-avoidance mechanisms are likely to be engagedflexibly, depending on additional informationbearing on the relative costs and benefits ofresponse.

Faulkner et al. evolved disease-avoidance mechanisms

335

04 046142 3/9/04 3:33 pm 35

The actual benefits of a xenophobic responseto foreigners were likely to be very low (andoutweighed by costs) under conditions in whichforeign peoples are far away, or under con-ditions in which perceivers are invulnerable todisease. The benefits were likely to be higherunder conditions in which perceivers wereactually vulnerable to the interpersonal trans-mission of infectious agents. Therefore, just asother evolved psychological mechanisms aremoderated by a variety of functionally-relevantlocal variables (Conway & Schaller, 2002;Gangestad & Simpson, 2000), the disease-avoidance process that inspires xenophobicresponses to foreign peoples is likely to haveevolved in such a way that it is sensitive to, andmoderated by, variables influencing the extentto which perceivers feel personally vulnerableto disease.

Chronic perceptions of vulnerability to diseaseEvolved threat-avoidance mechanisms may bepan-human, but there is considerable variationacross individuals in the tendency for thesemechanisms to be triggered. For example,disgust is a universal human emotion, but somepeople tend to be more sensitive to disgustthan others (Haidt, McCauley, & Rozin, 1994).Similarly, for a variety of reasons bearing bothon genetic heritage and social-learning pro-cesses, some people are likely to feel morevulnerable to disease than others. Regardlessof the origins of individual differences inperceived vulnerability to disease, these indi-vidual differences are likely to moderate theactivation of evolved disease-avoidancemechanisms in the domain of person percep-tion. Previous research has, in various ways,revealed relations between individuals’ self-per-ceived vulnerability and negative reactions toothers who actually are diseased, as well as toothers who are stigmatized by characteristics—such as physical disability—that heuristicallyconnote disease (Bishop, Alva, Cantu, &Rittiman, 1991; Crandall, Glor, & Britt, 1997;Herek & Capitaniato, 1998; Park et al., 2003;Rozin, Markwith, & Nemeroff, 1992). Thisanalysis suggests a testable hypothesis that hasnot previously received empirical attention:

individual differences in perceived vulnerabil-ity to disease may predict xenophobic reactionsto members of subjectively foreign ethnicgroups. We tested this specific hypothesis inone set of four studies reported here.

Contextually aroused perceptions of vulnera-bility to disease Just as evolved mechanismsare responsive to chronic individual differ-ences, they are similarly responsive to inputfrom temporary contextual variables (Schaller,2003). Any contextual information that impliesincreased vulnerability to disease might alsoamplify negative reactions to categories ofpeople heuristically associated with disease.Disease-relevant stimuli in contemporaryenvironments are abundant and include, forexample, news stories about the latest diseaseepidemic, advertisements for hygienic productsthat make salient the omnipresence of germsand bacteria, and disgust-eliciting objects likesewers and odorous garbage cans. Even whenthese contextual stimuli do not explicitly impli-cate foreign outgroups as a source of disease,they may nevertheless trigger xenophobicresponses to foreign peoples. This specifichypothesis—that temporarily-heightened vul-nerability to disease amplifies negative reac-tions to foreigners—was tested in an additionalset of two studies reported here.

Overview of studies

We conducted six studies to examine the effectsof chronic and contextually-aroused percep-tions of vulnerability to disease on xenophobicreactions to foreign peoples. Four studies werecorrelational: we examined the extent to whichxenophobic attitudes were predicted by self-reported chronic perceptions of vulnerability todisease. In addition, we conducted two experi-ments, which included procedures designed toexperimentally manipulate perceptions of vul-nerability to disease. Across all studies, partici-pants’ attitudes toward subjectively foreign orfamiliar outgroups were assessed. Based on thereasoning outlined above, we expected thatparticipants who felt (either chronically ortemporarily) more vulnerable to disease would


336

04 046142 3/9/04 3:33 pm Page 336

react more negatively toward subjectivelyforeign outgroups, but not toward outgroupswho were subjectively familiar.

Our conceptual analysis implies that the pre-dicted effects should be observed primarily onmeasures that assess specific attitudes towardtarget groups marked by characteristics con-noting foreign-ness. These effects are lesslikely to occur on measures that assess non-target-specific attitudes. All studies includedtarget-specific attitude measures to test the pre-dictions. In addition, several studies includedadditional measures that were not target-specific (these measures assessed attitudestoward immigration in general); this allowed usto examine the extent that the predicted effectswere restricted to specific target groups.

The conceptual analysis also implies that thepredicted effects are likely to be observed onmeasures that assess attitudes more directlyrelevant to functional behavior, such as physicalavoidance or social exclusion. Consequently, allstudies included measures that seem directlyrelevant (e.g. implicit associations between thetarget group and danger, attitudes about theimmigration of target groups to participants’local area). These measures were the primaryfocus of our analyses.

It is less clear how strongly the underlyingprocess might compel perceivers to attributevery specific traits to target groups. It seemsplausible that, even in the absence of directsupporting information about unknownpeoples, perceivers might be more likely tojudge these peoples more negatively onspecific traits that seem logically relevant todisease (e.g. unclean). However, the affectivesystem (disgust) that presumably underlies thedisease-avoidance process tends to compel arather diffuse set of negative associations(Haidt, 2001). Moreover, in the absence ofactual information supporting specific attribu-tions, participants may be reluctant to offerspecific stereotypical judgments. To explorethe possibility of effects on trait attributions,several of the studies included measures assess-ing participants’ judgments of target groups ona variety of traits both relevant and irrelevantto disease.

Studies 1–4: perceived vulnerability to diseasecorrelates with xenophobic attitudesAssessment of perceived vulnerability to diseaseIn order to test the hypothesis, we developed aquestionnaire measure to assess individualdifferences in chronic perceptions of vulnera-bility to disease. This ‘perceived vulnerability todisease’ (PVD) measure included 14 items thatassessed general beliefs about personal suscep-tibility to illness (e.g. ‘I have a history of sus-ceptibility to infectious disease’, ‘My immunesystem protects me from most illnesses thatother people get’), and discomfort with behav-iors that connote a higher risk of germ trans-mission (e.g. ‘I prefer to wash my hands prettysoon after shaking someone’s hand’, ‘I’m com-fortable sharing a water bottle with a friend’).Each item was followed by a 7-point responsescale with the endpoints labeled ‘StronglyDisagree’ and ‘Strongly Agree.’ All items werescored such that higher values indicatedgreater perceived vulnerability (6 items werereverse scored). An overall PVD score for eachparticipant was computed by averagingresponses across all 14 items.

Preliminary psychometric evaluation indi-cated that the PVD scale had adequate internalvalidity (Cronbach’s alphas were typically inthe range from .70 to .80 across multiplesamples). Indicators of discriminant validitywere also encouraging: It was not meaningfullycorrelated with conceptually irrelevant con-structs (e.g. social desirability, self-esteem, locusof control), and was correlated—modestly—with only one of the Big Five personality dimen-sions, neuroticism ( John & Srivastava, 1999; rsof .20 and .22 across two samples). Convergentvalidity was indicated by positive correlationsbetween PVD and a measure of ‘belief in adangerous world’ (Altemeyer, 1988; rs typicallyranged between .30 and .40 across samples),and with sensitivity to disgust (Haidt et al.,1994; rs ranged between .22 and .53 across sub-scales assessing disgust in different domains).

The PVD measure was administered toparticipants in Studies 1–4. In each study, weexamined correlations between overall PVDscore and the various measures assessingintergroup attitudes. Studies 1 and 2 focused


337

04 046142 3/9/04 3:33 pm Page 337

on attitudes toward African outgroups. Africanswere expected to be perceived as culturallyforeign to participants (all participants werestudents at the University of British Columbia,which is a campus populated predominantly bypeoples of East Asian and European heritage,but with very few students of African heritage).Studies 3 and 4 examined attitudes towardseveral other outgroups that a separate sampleof UBC students rated as either subjectivelyforeign or familiar.

The hypothesized relation between PVD andintergroup attitudes is clearly directional; indi-viduals who perceive themselves to be morevulnerable to disease are expected to react morenegatively toward subjectively foreign ethnicgroups. Given this clear directional prediction,when correlations consistent with this hypothe-sis are observed, we report p values associatedwith directional (i.e. ‘one-tailed’) tests of thestatistical null hypothesis. When correlations areobserved that are not obviously predicted by thehypothesis, we report p values associated withmore conservative nondirectional tests of thenull hypothesis, and explicitly identify themas ‘two-tailed’. Exact p values are reportedthroughout, so that readers can employ alterna-tive strategies of statistical inference if desired.

Study 1An initial test of the hypothesis emerged froma study that was designed primarily to testhypotheses distinct from those under consider-ation here (additional results are reported inSchaller, Park, & Mueller, 2003; Study 2). Thestudy employed the ‘implicit association test’(IAT; Greenwald, McGhee, & Schwartz, 1998;Rudman, Greenwald, Mellott, & Schwartz,1999) to assess implicitly activated semanticassociations with the social category ‘African’.

Participants were 52 undergraduate studentsat UBC. All participants first completed a set ofquestionnaires, one of which was the 14-itemPVD measure. Participants then completed twoIAT tasks, presented in counterbalanced order,that were developed to assess implicit associ-ations between Africans and specific evaluativecharacteristics. One IAT was designed to assessassociations with ‘unpleasant’; the other IAT

was designed to assess associations with‘danger’. The IAT procedures are described indetail elsewhere (Greenwald et al., 1998;Schaller et al., 2003); therefore, these methodsare described in abbreviated form here.

Both IAT tasks were presented on IBM-com-patible computers. Participants were presentedwith a series of forced-choice categorizationtrials, and were instructed to respond as quicklyas possible. On each trial, a stimulus (i.e. a wordor a picture) was presented in the center of thecomputer screen, along with two responsecategories (e.g. ‘unpleasant’ and ‘pleasant’)presented on the upper left and upper rightsides of the screen. Participants responded bypressing one of two keyboard keys: either theE key, corresponding to the response categoryon the upper left side of the screen, or theI key, corresponding to the response categoryon the upper right side of the screen.

On an ‘unpleasant/pleasant’ IAT, partici-pants judged whether words were unpleasant orpleasant, and on a ‘danger/safety’ IAT, partici-pants judged whether words connoted dangeror safety. For both IAT tasks, participants alsojudged the ethnicity of faces: participants ofAsian backgrounds judged whether faces wereAfrican or Asian and all other participantsjudged whether faces were African or European.Each IAT task consisted of distinct blocks oftrials; in two critical blocks, participants com-pleted both word and face categorization trials.In the first critical block, the same keyboard keywas used to indicate that faces were African andwords were unpleasant (in the ‘unpleasant/pleasant’ IAT) or connoted danger (in the‘danger/safety’ IAT). Responses on trials in thisblock were expected to be relatively fast ifparticipants implicitly associated African faceswith ‘unpleasant’ or ‘danger’. In the secondcritical block, keyboard responses werechanged, so that different keys were used toindicate ‘African’ and ‘unpleasant’/’danger’.Responses on trials in this block were expectedto be relatively slow if participants implicitlyassociated African faces with ‘unpleasant’ or‘danger’. The difference in average responsetimes within each of the two critical blocks oftrials (mean rt in the 2nd critical block minus


338

04 046142 3/9/04 3:33 pm Page 338

mean rt in the 1st critical block) served as anindicator of the extent to which Africans weredifferentially associated with ‘unpleasant’ or‘danger’.

Results reported elsewhere (Schaller et al.,2003) revealed that there was a generaltendency among participants to differentiallyassociate Africans with both ‘unpleasant’ andwith ‘danger’. The key question here is whethervariability in these responses was associated withindividual differences in perceived vulnerabilityto disease. Correlations revealed that PVD hada negligible relation with the unpleasant/pleasant IAT index (r = .05, p = .351), but posi-tively predicted the danger/safety IAT index(r = .24, p = .045).

These results provide initial evidence thatindividual differences in PVD predict negativeimplicit attitudes about people of Africanheritage. Interestingly, PVD did not predict theextent to which Africans were associated withunpleasant characteristics in general; but it didpredict the extent to which Africans wereassociated with danger. This pattern of resultssuggests that chronic feelings of vulnerability todisease may be associated specifically withthreat-relevant beliefs, which are likely to stimu-late avoidance behaviors, but are not as stronglyassociated with other negative beliefs that areless functionally relevant to threat-avoidance.

Study 2The predictive effects of PVD were alsoexamined in a second study that was originallydesigned for a different primary purpose.Participants in this study were 82 undergradu-ate students at UBC. All participants completedseveral questionnaires, including the PVDmeasure. They were then presented with a one-page written passage that began, ‘There is acountry in Central Africa, which for thepurposes of this study we will refer to asKrasnee, that has been experiencing a greatdeal of civil unrest in recent years. A number ofpeople from this country are trying to leave.’The passage went on to describe a group ofKrasneeans who, for various reasons, wereseeking to immigrate to Vancouver, Canada,where participants in the study lived.

Open-ended and rating scale techniqueswere then used to assess attitudes toward Kras-neeans. Participants were first asked to ‘Write 5words that you feel describe Krasneeans’, andwere also asked to draw a picture representing‘what you would imagine a typical Krasneean tolook like’. One coder, working from a masterlist, counted the number of words indicated byeach participant that connoted some sort ofdanger or threat. Four additional coders lookedat each drawing and rated on a 10-point scalethe demeanor of the Krasneean(s) depicted inthe drawing (1 = most unfriendly; 10 = mostfriendly). The four sets of ratings were stan-dardized and then averaged to create an indexof the perceived friendliness of the typicalKrasneean (Cronbach’s alpha = .92).

Additional measures of participants’ atti-tudes toward Krasneeans employed rating-scaletechniques. Participants were asked to rate, onsix separate 9-point scales, the extent to whichKrasneeans were likeable, hostile, trustworthy, open-minded, ignorant, and poor. To assess attitudestoward immigration, participants were alsoasked to rate, on five point scales, the extent towhich they agreed with various statements. Oneitem specifically assessed attitudes toward Kras-neean immigration, and read, ‘None of theKrasneeans applying for citizenship should beallowed to immigrate to Canada’, and fouradditional items assessed attitudes towardCanadian immigration policies in general (e.g.agreement with the statement, ‘Canada’s immi-gration policies are too strict’). Participants’scores on these four items were combined tocreate a composite index of general anti-immigration attitudes.

Of primary interest as an indicator of xeno-phobia were responses to the item ‘None of theKrasneeans applying for citizenship should beallowed to immigrate to Canada.’ Resultsrevealed a modest positive correlation betweenPVD and endorsement of this statement (r = .21,p = .032): individuals with higher PVD scoresmore strongly opposed Krasneean immigration.

There was little relation between PVD andanti-immigration attitudes in general (r = .13,p = .117). On the measures assessing specificbeliefs about the characteristics of Krasneeans,


339

04 046142 3/9/04 3:33 pm Page 339

PVD scores correlated weakly with the numberof danger-relevant traits participants used todescribe Krasneeans (r = .14, p = .099), butcorrelated more strongly with the demeanor ofthe Krasneeans represented in participants’drawings (r = –.31, p = .002): Those withhigher PVD scores drew less friendly lookingKrasneeans. PVD correlated with ratings ofKrasneeans on two trait characteristics: likable(r = –.33, p = .002), and trustworthy (r = –.22,p = .025). Individuals with higher PVD scoresrated Krasneeans to be less likable and lesstrustworthy. There were negligible relationsbetween PVD and ratings of hostile, open-minded,ignorant, or poor (–.16 < rs < .06).

These results complement those of Study 1and reveal PVD predicted exclusionary atti-tudes toward a specific African immigrantgroup. There was some additional evidencethat PVD predicted some more specific trait-based beliefs about the immigrant group aswell. One could argue that the perceivedcharacteristics predicted most strongly by PVD(unfriendliness, unlikability, and untrustwor-thiness) are traits that, when observed inothers, typically motivate attempts to avoidclose contact. Characteristics that less clearlymotivate avoidance (e.g. ‘ignorant’ or ‘poor’)were not predicted by PVD. Nor did PVD mean-ingfully predict the perception of hostility;although perceived hostility certainly connotesa danger to be avoided, the danger it connotesis not functionally relevant to disease avoid-ance. It is interesting too that, although PVDpredicted opposition to immigration from thespecific target group, it did not as stronglypredict anti-immigration attitudes in general.This perhaps provides some initial evidencethat chronic feelings of vulnerability to diseaseunderlie attitudes reflecting disease-avoidanceconcerns with subjectively foreign, but notfamiliar outgroups. We conducted additionalstudies to more directly explore this possibility.

Study 3In a third study, we used methods similar to theprevious study to test the relation between PVDand attitudes toward potential immigrantgroups. Unlike Study 2—which assessed atti-

tudes only toward an immigrant group fromAfrica—in this study, the geographical origin ofthe immigrant group was experimentallymanipulated: the group was described as beingeither from eastern Africa, eastern Asia, oreastern Europe. A separate sample of 23 UBCstudents rated immigrants from eastern Africaas subjectively foreign in general, and ratedthem as likely to engage in unusual practices indisease-relevant domains (e.g. hygiene andfood preparation). In contrast, immigrantsfrom eastern Asia and eastern Europe wererated as subjectively familiar, and were per-ceived to have hygiene and food preparationpractices similar to those of participants them-selves. Therefore, PVD is expected to predictnegative attitudes toward African immigrants,but not toward East Asian and European immi-grants.

Participants were 86 undergraduate studentsat UBC. All participants completed severalquestionnaires including the PVD scale. Partici-pants then listened to a voice on audiotape,briefly describing an ethnic populationidentified as ‘Saznians’. As in the previousstudy, the description stated that Saznians wereseeking to immigrate to Vancouver, Canada.Participants were randomly assigned to hear, aspart of the audiotaped message, that theSaznian homeland was either in ‘easternAfrica’, ‘eastern Asia’, or ‘eastern Europe’.

Participants then rated the extent that theyagreed with various statements using 5-pointscales. The first item was of primary interest inthat it assessed attitudes specific to Saznianimmigration: participants rated the extent towhich they disagreed or agreed with the state-ment, ‘The Canadian government should allowthe Saznians from [Africa/Asia/Europe] toenter Canada.’ On an additional six items,participants rated the extent that they agreedwith other general pro- or anti-immigrationstatements. A composite index of general pro-immigration attitudes was created by reversescoring responses to anti-immigration state-ments and then averaging responses to allitems.

As expected, the relation between PVD andendorsement of Saznian immigration depended


340

04 046142 3/9/04 3:33 pm Page 340

importantly on the geographical origin ofSaznians (see Figure 1). When Saznians weredescribed as African, there was a substantialnegative correlation between PVD andendorsement of Saznian immigration (r = –.38,p = .016): individuals with higher PVD scoreswere less keen on the immigration of AfricanSaznians. In contrast, correlations were slightlypositive under conditions in which Saznianswere described as either eastern Asian orEuropean (rs were .15 and .16, respectively;both ps > .40, two-tailed). Further analyses (ztests comparing the magnitudes of these rs)substantiate the observation that the negativecorrelation observed in the Africa conditionwas distinct from the weak relations in the othertwo conditions, ps = .026 and .020, respectively.

PVD was also negatively related to pro-immigration attitudes in general (r = –.22, p =.029). This relation was not meaningfully mod-erated by geographical origin.

Overall, the results reveal some tendency forindividuals who perceive themselves to be

vulnerable to disease to have stronger anti-immigration attitudes in general. However,when expressing attitudes toward specificimmigrant groups, PVD does not predict anti-immigrant attitudes equally. In this sample(drawn from a population comprised primarilyof people of either European or East Asianethnicity), PVD did not predict more negativereactions toward immigration of easternEuropean or eastern Asian outgroups, but it didpredict more negative reactions toward immi-gration of an eastern African outgroup. Theseresults are consistent with the hypothesis thatPVD predicts negative reactions toward subjec-tively foreign, but not subjectively familiarethnic groups.

Study 4The results of the first three studies reveal apattern of findings linking PVD with negativereactions toward African outgroups. Theseresults are consistent with a psychologicalprocess in which ethnic foreign-ness serves as a


341

–0.40

–0.30

–0.20

–0.10

0.00

0.10

0.20

Origin of Immigrant Group

Eastern Africa Eastern Asia Eastern Europe

Figure 1. Correlations between PVD and pro-immigration attitudes as a function of the geographical origin ofpotential immigrants (Study 3).

04 046142 3/9/04 3:33 pm Page 341

cue that heuristically signals a threat of inter-personal disease transmission. These results arealso consistent with a different process in whichparticipants’ reactions might be influenced bysome rational assessment of the health risksassociated specifically with Africa. Certain high-profile infectious diseases, such as Ebola andHIV, have been more notably epidemic in partsof Africa than in other parts of the world. Con-sequently, it could be argued, PVD mightpredict negative attitudes uniquely towardAfrican peoples. This possibility is not uninter-esting; but the implications—both conceptualand practical—are more profound if PVDpredicts negative attitudes toward a broaderrange of subjectively foreign peoples. For thisreason, in a fourth study, we assessed therelation between PVD and attitudes towardthree non-African outgroups: immigrants fromPeru, Qatar, and Sri Lanka. A separate sample of23 UBC students (the same separate samplewho rated the foreign-ness of the immigrantgroups portrayed in Study 3) rated immigrantsfrom Peru, Qatar, and Sri Lanka to be subjec-tively foreign overall and in cultural practicespertaining to hygiene and food. Therefore,PVD was expected to predict xenophobic reac-tions to all three of these groups.

Participants were 47 undergraduate studentsat the University of British Columbia, primarilyof East Asian and European heritage (nonewere from Peru, Qatar, or Sri Lanka). Partici-pants were first presented with a writtendescription of a refugee group seeking to immi-grate to Vancouver, Canada. The passage wasidentical for all participants except for theapparent geographical origin of the refugees.Participants were randomly assigned to one ofthree conditions in which they read that therefugees were from Peru, Qatar, or Sri Lanka.After first identifying the geographical origin ofthe refugees, and describing their harsh livingconditions resulting from arid climate with lowrainfall, the passage proceeded as follows:

As a result of these living conditions, a largenumber of refugees from the plains of [Peru/Qatar/Sri Lanka] are seeking refugee status inCanada. They have told Canadian immigrationofficials that the standard of living in [Peru/Qatar/

Sri Lanka] is unacceptably low and that its healthand social systems do not meet the needs of[Peruvian/Qatar/Sri Lankan] people. The majorityof these refugees are applying to live in the greaterVancouver area.

To assess attitudes toward immigrants fromPeru, Qatar, or Sri Lanka, participants in eachcondition rated the extent that they agreed withthree statements that assessed attitudes andbeliefs pertaining specifically to the refugeegroup described. As in Study 3, one statementassessed whether participants felt the Canadiangovernment should allow the refugees to enterCanada. Two other statements assessed whetherparticipants felt the refugees might bringhealth problems or criminal problems intoCanada.

Three additional statements assessed attitudestoward immigration in general. Responses tothese last three items were combined to form asingle index assessing pro-immigration atti-tudes in general. Ratings on all statements weremade on 7-point scales.

Participants also rated the target immigrantgroup on a set of 13 specific characteristics. Fiveof these characteristics pertained to hygiene(sanitary, filthy, hygienic, clean, and dirty); therest tapped into a variety of other evaluativetraits (likeable, hostile, trustworthy, open-minded, ignorant, poor, lazy, and unintelli-gent). Ratings were made on 9-point scales.Following these procedures, participantscompleted various questionnaires, includingthe 14-item PVD measure.

A negative relation between PVD andendorsement of immigration of all threerefugee groups was expected, and so analysesfocused on the full set of data, collapsed acrossconditions. Results revealed a substantialnegative correlation between PVD and attitudestoward immigration of the specified group (r =–.40, p = .003). (A negative correlation wasobserved regardless of whether the refugeegroup was described as originating in Peru,Qatar, or Sri Lanka: The rs in each of these con-ditions were –.30, –.36, and –.53, respectively.)The higher individuals’ PVD scores were, theless willing they were to allow the group toimmigrate.


342

04 046142 3/9/04 3:33 pm Page 342

PVD also predicted beliefs linking the refugeegroups to health problems (r = .34, p = .012).There was a somewhat weaker correlationbetween PVD and beliefs linking the refugeegroups to crime (r = .25, p = .087, two-tailed).There was also a negative correlation betweenPVD and the 3-item measure of pro-immigrationattitudes in general (r = –.38, p = .004).

In contrast to these relations to the attitudemeasures, PVD did not strongly predictresponses on the 13-trait rating items. Onlythree correlations exceeded .2 in magnitude.Two of these correlations were with traits per-taining to hygiene: The correlation with ‘clean’was r = –.25 (p = .046) and the correlation with‘dirty’ was r = .22 (p = .073). The other notablecorrelation was that between PVD and ‘open-minded’ (r = –.29, p = .050, two-tailed).

The results of this study reveal a strongrelation between PVD and anti-immigrant atti-tudes toward three non-African foreign ethnicgroups. Individuals with higher PVD scoreswere less willing to allow foreign outgroups toimmigrate to Canada. They also held strongerbeliefs that outgroups might bring diseaseproblems to Canada and are less hygienic. Tosome extent, PVD also predicted beliefs aboutcriminality—a result suggesting that the under-lying process may be associated with a variety ofattitudes that are not logically disease-relevant,but which nonetheless motivate exclusionaryaction. PVD also clearly predicted anti-immi-grant attitudes in general.

It’s also worth noting that although PVDclearly predicted xenophobic attitudes, it didnot strongly predict the attribution of specifictraits to the immigrant groups. The reason forthis may lie partially in the nature of the affec-tive state—disgust—that is linked conceptuallyto the disease-avoidance process. Disgust typi-cally elicits crude evaluative attitudes (distaste,dislike) and behavioral attempts to reducecontact, but it may not immediately connotespecific trait information of the sort connotedby other emotions, such as fear.

Discussion of Studies 1–4These studies provide the first evidence thatxenophobic attitudes are moderated by indi-

vidual differences in perceived vulnerability todisease. Although the results on some ancillarymeasures (e.g. attribution of personal traits andcharacteristics, anti-immigrant attitudes ingeneral) were somewhat inconsistent acrossstudies, there was a consistent pattern of resultson the primary measures assessing xenophobicattitudes within each study. Canadian studentswho perceived themselves chronically to bevulnerable to diseases were more likely toassociate foreign peoples (Africans) withdanger, and were less likely to endorse theimmigration of foreigners (immigrants fromAfrica, Peru, Qatar, and Sri Lanka). Perceivedvulnerability to disease predicted xenophobicattitudes towards specific immigrant groupsthat were subjectively foreign, but not towardother groups (immigrants from eastern Asiaand eastern Europe) that were subjectivelyfamiliar.

These findings are consistent with an evolveddisease-avoidance process in which subjectiveforeign-ness heuristically triggers cognitionsthat motivate avoidance and exclusion,especially among individuals who personallyfeel vulnerable to disease. However, thesefindings are correlational in nature, and so it isnecessary to consider the possibility that therelation between PVD and xenophobic atti-tudes may be the product of shared variancewith other individual difference variables thathave nothing conceptually to do with disease.Several additional results address some of these‘third variable’ alternative explanations. Forinstance, it might be argued that PVD is merelyone indicator of a broader personality con-struct, such as neuroticism, that more directlypredicts xenophobic attitudes. This alternativeseems unlikely. Not only is PVD only weakly cor-related with neuroticism (see results summar-ized above), but neuroticism is not consistentlycorrelated with prejudicial attitudes. Somestudies have shown modest positive relationsbetween neuroticism and ethnic prejudice(Hebron & Ridley, 1965), other studies showlittle or no predictive effects of either neuroti-cism or anxiety (Maykovich, 1975; Orpen, 1973;Prentice, 1962) and at least one study reveals anegative correlation between neuroticism and


343

04 046142 3/9/04 3:33 pm Page 343

prejudice against ethnic immigrants (Ray &Kiefl, 1984). Potentially more plausible sourcesof spurious relations are other individual differ-ence constructs pertaining to other specificthreats and fears. For instance, PVD does cor-relate substantially with ‘belief in a dangerousworld’ (BDW), and prior research reveals thatBDW does predict prejudicial attitudes againstethnic outgroups (Altemeyer, 1988). We wereable to directly address this specific alternativebecause participants in Studies 1–4 also com-pleted Altemeyer’s (1988) measure of BDW.For each study, we conducted regressionanalyses that included both PVD and BDW aspredictors of the primary measure of xenopho-bic attitudes; results revealed that the predictiveeffects of PVD were not substantially reduced.We also created a composite dataset (N = 213),merging data from all four studies after firststandardizing responses on the primary depen-dent measure (we reverse scored the valuesfrom Studies 3 and 4 so that, across all fourstudies, higher standardized values reflectedmore negative attitudes). Within this compositedataset, PVD and BDW were entered simul-taneously as predictors. Results revealed amodest effect of BDW on xenophobic attitudes(beta = .13, p = .065, two-tailed) and—moreimportantly—a more substantial and uniqueeffect of PVD (beta = .23, p < .001).

Although these analyses help rule out severalspecific personality constructs as viable alterna-tive explanations for the effects of PVD, theycannot address all possible third-variable expla-nations. Consequently, we conducted twoadditional studies in which we attempted toexperimentally manipulate the extent thatpeople feel vulnerable to disease. Thisapproach allowed us to more rigorously test theproposition that disease-avoidance processescontribute to xenophobic attitudes.

Studies 5 and 6: temporary salience of diseaseamplifies xenophobic attitudesIn two experiments, we manipulated partici-pants’ temporary perceptions of vulnerabilityto disease, and then assessed attitudes towardthe immigration of subjectively foreign orfamiliar immigrant groups. In each study,

participants randomly assigned to one experi-mental condition viewed a series of picturesthat conveyed the ease with which bacteria andgerms are transmitted in everyday life. Theremaining participants (assigned to a controlcondition) viewed a series of pictures thatconveyed the ease with which accidents occurin everyday life. All participants then com-pleted measures assessing their attitudes towardone specific immigrant group, which—depend-ing on another experimental manipulation—was either subjectively foreign or familiar.Participants in Study 6 also completed anadditional measure in which they allocatedgovernment resources to attract immigrantsfrom a variety of both foreign and familiar loca-tions. In both studies, the conceptual hypothe-sis would be supported by evidence ofinteractions between the disease-saliencemanipulation and the foreign-ness of targetgroups: compared to participants in the controlcondition, participants in the disease-salientcondition were expected to respond morenegatively toward foreign—but not familiar—immigrant groups.

PretestingPrior to conducting these experiments, aseparate sample of 26 UBC students pre-rated aset of eight immigrant groups representingareas of the world with very different climaticand geographic characteristics. These groupswere immigrants from Taiwan, Peru, Poland,Nigeria, Scotland, Mongolia, Brazil, andIceland. Participants rated each immigrantgroup on scales assessing how different thegroup was from participants themselves (e.g.how culturally and visually different they are).Participants also rated each group on severalscales assessing how different they were inspecific disease-relevant domains (e.g. howdifferent their food preparation and hygienicpractices are). Ratings were made on 9-pointscales. On the basis of these responses, wecomputed an average foreign-ness score foreach group. Immigrants from Nigeria wererated as most foreign overall (M = 5.89),followed by immigrants from Mongolia (M =5.48), Brazil and Peru (Ms = 5.38), Iceland (M


344

04 046142 3/9/04 3:33 pm Page 344

= 4.62), Poland (M = 4.50), Taiwan (M = 4.33),and Scotland (M = 4.00).

These pretest results informed the methodsused in the two experiments. In one experi-ment, participants indicated attitudes towardpotential immigrants from either Nigeria(foreign) or Scotland (familiar). In a secondexperiment, participants indicated attitudestoward potential immigrants from eitherMongolia (foreign) or Taiwan (familiar), andalso completed an additional measure assessingrelative preferences among all eight pre-ratedimmigrant groups.

Study 5Participants were 57 undergraduate students atUBC. Participants first completed several ques-tionnaires, including the PVD measure. Theywere then randomly assigned to one of twoexperimental conditions. In one condition,participants viewed a slide show designed toremind them of the various ways that diseasesare transmitted in daily life, thus making thethreat of contagious diseases especially salient.This ‘Disease’ slide show included 11 picturesthat would ostensibly be used in a health edu-cation program (e.g. one slide pictures awoman unsuccessfully attempting to killcartoon germs in her kitchen, and is titled ‘TheHorrors of the Kitchen Sponge and FamilyPets’; another slide depicts a strand of hair sur-rounded by bacteria and is labeled ‘HairBacteria. A microscopic view of a strand of hairand some of the typical bacteria that surroundit’). The rest of the participants were assignedto a control condition, in which they viewed aslide show designed to make the threat ofphysical dangers especially salient. These slideswere intended to make participants feel sus-ceptible to a threat that was irrelevant todisease. This ‘Accidents’ slide show included11 pictures that would ostensibly be used ina safety education program. (Two examples are‘School Bus Hazards’ which depicts a girlbending over to pick up a book as a schoolbus is about to hit her, and ‘Electricity andWater Don’t Mix’ which portrays a bathingwoman surrounded by plugged-in electricalappliances.)

Participants viewed their assigned slide showtwice; the first time they rated the educationalinformativeness of each slide on a 10-pointrating scale, and after a second viewing theywrote a short paragraph describing their overallimpressions of the slide show.

Participants were then told that the initialstudy was over and were asked to complete aquestionnaire for an ostensibly unrelated study.The experimenter read aloud a brief descrip-tion of a group seeking to immigrate to Van-couver. Participants were randomly assigned tohear that this group was from either Nigeria orScotland. They then completed a questionnaireconsisting of rating scale items assessing theirattitudes toward the described immigrantgroup. These items were similar to those inStudy 4: three items assessed specific attitudestoward the potential immigrant group, threeitems assessed attitudes toward immigration ingeneral, and 13 items asked participants theextent that various traits are characteristic ofthe immigrant group.

In general, across all attitude and trait items,there was a general main effect for targetgroup, in that participants tended to rateNigerian immigrants less favorably thanScottish immigrants. On specific items, partici-pants believed Nigerians were more likely tobring costly health problems to Vancouver, andrated Nigerians to be less sanitary, less clean,less open-minded, and as poorer and dirtierthan Scottish immigrants (all ps < .05).

For the purpose of testing the hypothesisabout disease-salience and xenophobia, theattitude item of central interest was the itemassessing agreement with the statement: ‘TheCanadian government should allow [Nigerian/Scottish] immigrants to live in Vancouver.’ PVDscores, particularly for the average of PVD itemsassessing discomfort with behaviors thatconnote a risk of germ transmission, predictedresponses on this item; therefore the averagescore on this subset of PVD items was includedas a covariate in a 2 � 2 (slide show condition� target group condition) analysis of covari-ance. This analysis revealed the predicted inter-action, (F(1,52) = 6.00, p = .02). As depicted inFigure 2, participants in the control condition


345

04 046142 3/9/04 3:33 pm Page 345

were approximately equally likely to supportimmigration of Scots (M = 5.85) and Nigerians(M = 6.07), but those in the disease-saliencecondition were more likely to endorse theimmigration of Scots (M = 6.36) and less likelyto endorse the immigration of Nigerians (M =5.53).

In contrast to the clear effects of the disease-salience manipulation on this dependentmeasure, there were weaker and inconsistenteffects on other items. This failure to findeffects across a broad range of items is con-sistent with the general nature of the PVDeffects obtained in Studies 1–4. The effectsseem to show up primarily on attitude itemsthat bear most directly on physical contactand/or social exclusion. This may reflect theunderlying behavioral function of evolved

disease-avoidance processes. It is also possiblethat it reflects the operation of participants’self-presentational concerns: these studentsmay be somewhat willing to explicitly expressattitudes about governmental decisions bearingon desired and undesired immigrant groups,but may be less willing to overtly express stereo-typical judgments about those groups.

The results depicted in Figure 2 provideinitial support for the hypothesis that tem-porarily-heightened worries about vulnerabilityto disease can amplify xenophobic attitudestoward subjectively foreign (but not familiar)outgroups. Still, these results are limited. Thepurpose of the final study was to conceptuallyreplicate and extend this finding by examiningattitudes toward a broader range of potentialimmigrant groups.


346

5.00

5.25

5.50

5.75

6.00

6.25

6.50

Accidents Diseases

Salient Threat

End

orse

men

t

Nigerian Immigrants

Scottish Immigrants

Figure 2. Interaction between disease salience and geographical origin of immigrants on endorsement ofpotential immigration (Study 5).

04 046142 3/9/04 3:33 pm Page 346

Study 6The methods of Study 6 were similar to those ofStudy 5, with two main changes. One changepertained to the specific target immigrantgroups that were described to participants:Mongolians and Taiwanese. In pre-testing(described above), Mongolian and Taiwaneseimmigrants were rated, respectively, as thesecond-most and second-least foreign of theeight pre-rated groups. The second change wasthe inclusion of an additional measure, inwhich participants indicated the percentage ofa special government budget that should beallocated to help attract immigrants from eachof the eight different pre-rated nations. Thismeasure extends the analysis to include abroader range of subjectively foreign andfamiliar groups. In addition, it might also beexpected to be less reactive to participants’ self-presentation concerns.

Participants were 45 undergraduate studentsat UBC. Participants first completed severalquestionnaires, including the PVD measure,and then were randomly assigned to view eitherthe ‘Disease’ slide show or the ‘Accidents’ slideshow (this manipulation was identical to that inStudy 5). Following this, participants heard adescription about a specific immigrant groupseeking to settle in Vancouver; they wererandomly assigned to hear that the immigrantgroup was either from Mongolia (foreign) orTaiwan (familiar). Then, as in Study 5, theycompleted a set of rating-scale items assessingattitudes and beliefs about the specific targetimmigrant group.

Participants then completed an additionaldependent measure, labeled the ‘immigrationadvertising questionnaire’. At the beginning ofthis questionnaire, a short paragraph explainedthat the Canadian government was considering‘advertising’ Canada as a suitable destinationfor potential immigrants around the world.This paragraph also explained that Canada hada finite amount of resources for this kind ofadvertising, and so it sought input on the wayin which this advertising budget should beallocated to different countries from whichpotential immigrants might come. Participantswere asked to indicate the percent of the total

immigration advertising budget that should bespent in each of eight different countries:Taiwan, Peru, Poland, Nigeria, Scotland,Mongolia, Brazil, and Iceland. These were thesame eight countries that had been pre-testedfor perceived foreign-ness. Therefore, budgetpercentages that each participant allocated tothe four countries pre-rated to be most foreign(Nigeria, Mongolia, Brazil, and Peru) weresummed to compute a single index; and per-centages allocated to the four countries pre-rated to be most familiar (Scotland, Taiwan,Poland, and Iceland) were summed to computea second index. This allowed for a within-subjects comparison of budget allocations toattract foreign versus familiar immigrants.Participants in the disease-salient condition,compared to control participants, wereexpected to allocate relatively less of the budgetto subjectively foreign areas.

The results on the attitude and trait-ratingitems indicated that, in contrast to Study 5,participants did not generally rate the subjec-tively foreign immigrant group (Mongolians)any less favorably than the familiar immigrantgroup (Taiwanese). A main effect for targetgroup emerged on only one specific item:participants rated Mongolian immigrants to bepoorer than Taiwanese (t(43) = 2.69, p = .01).Also in contrast to the results of Study 5, thereemerged no interpretable effects of the disease-salience manipulation on these measures—including no evidence of the predictedinteraction on the item assessing endorsementof the target group’s immigration to Canada.The predicted interaction did emerge on thebudget allocation measure. A 2 � 2 � 2 (slideshow condition � target group condition �budget allocation index) repeated measuresanalysis of variance revealed the anticipatedtwo-way interaction between slide show con-dition and budget allocation index (F(1, 41) =4.58, p = .038). Whereas participants in thecontrol condition allocated roughly equalamounts of advertising resources to foreign andfamiliar countries (47.48% vs. 52.52%, respec-tively), participants in the disease-saliencecondition allocated less advertising resources toforeign countries and more to familiar countries


347

04 046142 3/9/04 3:33 pm Page 347

(37.57% vs. 62.43%, respectively). Figure 3depicts this interaction. A deeper examinationof the budget allocations revealed that alloca-tions to three of the four foreign countries(Nigeria, Mongolia, and Peru) were lower inthe disease-salient condition. Thus, the effectsof temporary disease salience extend beyondreactions merely to Nigerians (documented inStudy 5) to other foreign peoples as well. Theseresults reveal also that, although participants inthe disease-salient condition did not expressgreater xenophobic reactions to Mongolianimmigrants when they simply rated Mongoliansalone (in the attitude rating scales), they didappear to do so when these reactions wereembedded in a response context that allowed

them to appear especially welcoming to othergroups of immigrants—more familiar immi-grants, of course.

An effect of PVD also emerged on the budgetallocation measure: participants with higherPVD scores allocated less of the advertisingbudget to foreign areas (r = –.24, p = .05).

Discussion of Studies 5 and 6Across both experiments, there emerged someevidence that temporarily-activated worriesabout disease led to more negative reactions toforeign, but not familiar, immigrant groups.Study 5 revealed evidence of this effect on anattitude item pertaining to support for theimmigration of specific foreign and familiar


348

30

35

40

45

50

55

60

65

Accidents Diseases

Salient Threat

Per

cent

Allo

catio

n of

Fun

ds

Foreign ImmigrantsFamiliar Immigrants

Figure 3. Interaction between disease salience and foreign-ness of geographical areas on allocation ofimmigration advertising budget (Study 6).

04 046142 3/9/04 3:33 pm Page 348

target groups. Study 6 failed to replicate thiseffect exactly on the same kind of attitude itemrewritten to specify a different pair of targetgroups, but did conceptually replicate theeffect on a different measure (a budget allo-cation measure) that assessed interest in attract-ing a greater variety of foreign and familiarimmigrant groups. It is possible that the pre-dicted effect emerged more clearly on thislatter measure because, in contrast to the morestraightforward attitude measures, participantsmay have been less likely to worry that theirbudget allocation choices were indicative ofanti-immigrant prejudice. In any case, althoughthere was some inconsistency in the resultsacross the two experiments, both experimentsyielded evidence consistent with the same con-ceptual hypothesis.

It is worth noting also that the effects ofPVD documented in Studies 1–4 were also repli-cated on the budget allocation measure used inStudy 6. Thus, this experiment not only sub-stantiates the predicted effect of temporarydisease worries on xenophobia, it also furtherdocuments the effects of chronic diseaseworries on xenophobia as well.

General discussion

The six studies reported here provide the firstempirical evidence that chronic and temporar-ily-aroused feelings of vulnerability to diseasecontribute to negative attitudes toward foreignpeoples. These findings are consistent with anevolved disease-avoidance model of contem-porary stigma processes (Kurzban & Leary,2001; Park et al., 2003), and they extend thismodel to a new domain of application.

Subjective perceptions of familiarity andforeign-ness appear to be fundamental to theactivation of this underlying process. Individualdifferences in perceived vulnerability to diseasepredicted negative reactions only to subjectivelyforeign peoples, but not to subjectively familiarpeoples. Similarly, contextual information thatmakes disease especially salient led to negativereactions to subjectively foreign immigrants,but not to subjectively familiar immigrants. (Ifanything, greater worries about disease were

associated with even more favorable reactionsto familiar immigrants; see especially the resultsfrom Studies 3 and 5.) Thus, the process under-lying these effects appears to be conceptuallydistinct from other psychological processes thatprecipitate prejudicial attitudes toward a less-specified range of outgroups and socialcategories. This evidence of target-specificityand domain-specificity is consistent with an evo-lutionarily-informed meta-theoretical approachto prejudice processes in general. There aremany different kinds of threat that may precip-itate negative reactions to outgroups. Theseinclude symbolic threats at the level of self-concept and social identity, as well as perceivedthreats to political status, economic security,and physical well-being (Schaller, 2003;Sidanius & Pratto, 1999; Stephan, Diaz-Loving,& Duran, 2000). Different outgroups—definedby different features—precipitate differentkinds of threat, and these different kinds ofthreat are associated with qualitatively differentkinds of negative responses (Neuberg &Cottrell, 2002).

The results from several of these studiesrevealed some, albeit tentative, indication ofdomain-specificity in beliefs about subjectivelyforeign outgroups. For instance, in Study 1,PVD predicted implicit cognitive associationslinking Africans to danger, but not to unpleas-ant semantic concepts in general. In severalother studies, there was some evidence thatPVD predicted beliefs linking foreign groupswith specific kinds of traits that either connoteuncleanliness or that typically motivate physicalavoidance. However, these results were weakand inconsistent across studies, and bearfurther investigation. Many other evolved pro-cesses—such as those pertaining to fear—areassociated with fairly domain-specific judg-ments and reactions. For instance, variablesconnoting vulnerability to intentional physicalharm predict specific, functionally-relevant traitjudgments about ethnic outgroups (Schalleret al., 2003). It is possible that an evolveddisease-avoidance process (presumably linkedaffectively to disgust rather than fear) precipi-tates a less well-specified set of semanticassociations. Given the interest that issues of


349

04 046142 3/9/04 3:33 pm Page 349

domain-specificity inspire among evolutionarypsychologists (Kenrick, Sadalla, & Keefe, 1998),it will be useful for future research to investi-gate more thoroughly the domain-specificity ofdisease-based xenophobic reactions.

Empirical support for a disease-avoidancemechanism does not preclude the operation ofadditional adaptive mechanisms promotingxenophobic reactions to foreign peoples. Theviolation of local cultural norms not onlysignals the threat of contagion, but may alsosignal a threat to social efficiency and effectivegroup functioning (Gil-White, 2001). Just asspecific psychological reactions (such asdisgust) may be associated with the perceivedthreat of disease, other reactions (such ascontempt and moral disapproval) may beassociated with the perception of normviolation (Boyd, Gintis, Bowles, & Richerson,2003; Neuberg & Cottrell, 2002). A multi-faceted evolutionary approach can help predictthe circumstances under which subjectivelyforeign ethnic outgroups might most stronglyelicit disease-avoidant reactions rather thanreactions designed to maintain effective groupfunctioning. If an outgroup violates norms thatare particularly germane to disease (e.g.customs pertaining to personal hygiene, foodpreparation, or sex), then disease-avoidancemechanisms are especially likely to be engaged.However, if an outgroup violates other kinds ofnorms that are particularly pertinent to socialefficiency (e.g. language use, reciprocity), thenxenophobia might be predicted primarily byconcerns with effective group functioning. Inthe present studies, we did not assess the extentto which target groups were perceived asforeign in these other kinds of normativedomains; further research is needed to test pre-dictions about the relations between specificdomains of foreign-ness, the threats they imply,and the psychological reactions they elicit.

While the empirical results presented hereare consistent with the underlying evolutionarymodel from which these hypotheses werederived, these results alone cannot rule outalternative accounts for the contemporarytendency of xenophobic attitudes to be pro-nounced in individuals who feel particularly

vulnerable to disease. It might be argued, forinstance, that the effects documented here mayresult from a rational assessment of the disease-relevant risks associated with particular immi-grant groups (e.g. those from developingcountries), rather than from the heuristic acti-vation of evolved mechanisms. A rational risk-assessment process, however, implies thatexclusionary attitudes should be based on attri-butions of disease-relevant traits to foreignpeoples. While our findings occasionally indi-cated that PVD predicted such trait attribu-tions, these effects were weak when obtained(see Study 4) and were inconsistent acrossstudies; it seems unlikely that trait attributionsmediate the relation between PVD and exclu-sionary attitudes. People may occasionallyengage in a rational analysis when there aresalient reasons (e.g. the recent outbreak ofSARS) to associate specific groups with specificdisease threats. However, the present findingssuggest that there are additional non-rationalprocesses that contribute to similar xenophobicattitudes.

Additional evidence for a non-rationalprocess also emerges from statistical analysesthat attend to the two kinds of items from whichthe PVD measure is comprised. Recall thatsome items measure beliefs about personalvulnerability to disease, while other itemsmeasure discomfort with situations in whichgerms are likely to be transmitted. In a sense,the first set of items assesses thoughtful con-siderations about disease susceptibility, whilethe second set assesses more emotion-basedavoidant reactions. Across Studies 1–4, wefound that the emotion-based ‘germ-aversion’subscale typically exerted a stronger effect onxenophobic attitudes (mean r = .28 for thegerm-aversion subscale, compared to a meanr = .16 for the subscale measuring morethoughtful beliefs about disease susceptibility).These results are consistent with other findingson PVD and prejudice (Park et al., 2003), andprovide some indirect evidence that the xeno-phobia results are due less to a rational assess-ment of disease risk than they are to anautomatically-activated aversion to heuristicallythreatening social situations.


350

04 046142 3/9/04 3:33 pm Page 350

Indirect evidence for an automated non-rational mechanism does not, of course,provide verification for the hypothesized role ofevolutionary pressures in sculpting such amechanism. No such verification can ever beproduced by merely psychological researchmethods (Conway & Schaller, 2002). The valueof this evolutionary level of analysis lies not inits explanatory necessity, but in its meta-theoretical utility in generating new discoveriesand connecting different scholarly literatures—including qualitative analyses of anti-immigrantattitudes throughout human history (e.g.Markel & Stern, 2002), zoological studies ofxenophobia in other primate species (e.g.Holloway, 1974; Southwick, Siddiqi, Farooqui,& Pal, 1974), and the extensive psychologicalliterature on intergroup prejudice.

The utility of an evolutionary approach tocontemporary psychological processes is maxi-mized when it is considered in conjunction withmore ‘traditional’ meta-theoretical approachesin the social sciences. Consider the processes ofsocialization and cultural change. One mightargue that a process of cultural—not bio-logical—evolution over the past several millen-nia has contributed to beliefs about foreignersthat are functionally related to disease-avoid-ance concerns. As a result of contact withforeign populations and the inevitable co-occurence of disease outbreaks, cultural beliefsmay have arisen that associate foreigners withdisease and promote behavioral avoidance orsocial exclusion. These beliefs may be passedon inter-generationally, and so may become apermanent part of cultural knowledge. Ofcourse, processes of cultural evolution areentirely compatible—and often necessarilyinter-related—with processes of biological evol-ution (Boyd & Richerson, 1985; Kameda,Takezawa, & Hastie, 2003). For instance, theancient Greek belief that foreigners possess an‘evil eye’, which by itself was thought to causeillness (Nutton, 2000), may have emerged andpersisted because it motivates avoidance orexclusionary behaviors in the presence of for-eigners. Beliefs that are especially likely toemerge and persist within cultures are thosethat exploit fundamental emotional systems

such as disgust (Heath, Bell, & Sternberg,2001), and more generally are those that helpindividuals solve fundamental adaptive problems(Krebs & Janicki, 2004).

Concluding remarks

The research presented here provides a specificexample of the value of evolutionary logic inpsychological inquiry: reasoning about therecurrent adaptive problems imposed by ances-tral human environments, and the cognitivemechanisms that arose to solve them, can yieldnovel hypotheses about psychological processesin the here-and-now (Conway & Schaller,2002). We used this approach to generate thepreviously-untested hypothesis that individuals’feelings of vulnerability to disease have impli-cations for the seemingly unrelated domain ofintergroup cognition. Our findings providesupport for this hypothesis and indicate thatchronic and contextually aroused feelings ofvulnerability to disease should be added to thelist of psychological factors that contribute toxenophobic attitudes.

ReferencesAltemeyer, B. (1988). Enemies of freedom.

San Francisco: Jossey-Bass.Bishop, G. D., Alva, A. L., Cantu, L., & Rittiman,

T. K. (1991). Responses to persons with AIDS:Fear of contagion or stigma? Journal of AppliedSocial Psychology, 21, 1877–1888.

Boyd, R., Gintis, H., Bowles, S., & Richerson, P. J.(2003). The evolution of altruistic punishment.Proceedings of the National Academy of Sciences (USA),100, 3531–3535.

Boyd, R., & Richerson, P. J. (1985). Culture and theevolutionary process. Chicago: University of ChicagoPress.

Conway, L. G., III, & Schaller, M. (2002). On theverifiability of evolutionary psychological theories:An analysis of the psychology of scientificpersuasion. Personality and Social Psychology Review,6, 152–166.

Crandall, C. S., Glor, J., & Britt, T. W. (1997).Aids-related stigmatization: Instrumental andsymbolic attitudes. Journal of Applied SocialPsychology, 27, 95–123.

Diamond, J. (1999). Guns, germs, and steel. New York:Norton.


351

04 046142 3/9/04 3:33 pm Page 351

Fiske, S. T. (1998). Stereotyping, prejudice, anddiscrimination. In D. T. Gilbert, S. T. Fiske, &G. Lindzey (Eds.), The handbook of social psychology(4th edn, pp. 357–411). New York: McGraw-Hill.

Gangestad, S. W., & Simpson, J. A. (2000). Theevolution of human mating: Trade-offs andstrategic pluralism. Behavioral and Brain Sciences,23, 573–644.

Gil-White, F. J. (2001). Are ethnic groups biological‘species’ to the human brain? Essentialism in ourcognition of some social categories. CurrentAnthropology, 42, 515–554.

Goldhagen, D. J. (1996). Hitler’s willing executioners.New York: Random House.

Greenwald, A. G., McGhee, D. E., & Schwartz, J. L.K. (1998). Measuring individual differences inimplicit cognition: The Implicit Association Test.Journal of Personality and Social Psychology, 74,1464–1480.

Haidt, J. (2001). The emotional dog and its rationaltail: A social intuitionist approach to moraljudgment. Psychological Review, 108, 814–834.

Haidt, J., McCauley, C., & Rozin, P. (1994).Individual differences in sensitivity to disgust:A scale sampling seven domains of disgust elicitors.Personality and Individual Differences, 16, 701–713.

Heath, C., Bell, C., & Sternberg, E. (2001).Emotional selection in memes: the case of urbanlegends. Journal of Personality and Social Psychology,81, 1028–1041.

Hebron, M. E., & Ridley, F. (1965). Characteristicsassociated with racial prejudice in adolescent boys.British Journal of Social and Clinical Psychology, 4,92–97.

Herek, G. M., & Capitaniato, J. P. (1998). Symbolicprejudice or fear of infection? A functionalanalysis of AIDS-related stigma amongheterosexual adults. Basic and Applied SocialPsychology, 20, 230–241.

Holloway, R. (Ed.). (1974). Primate aggression,territoriality, and xenophobia. New York: AcademicPress.

John, O. P., & Srivastava, S. (1999). The big five traittaxonomy: History, measurement and theoreticalperspectives. In L. A. Pervin & O. P. John (Eds.),Handbook of personality: Theory and research (2ndedn, pp. 102–138). New York: Guilford Press.

Kameda, T., Takezawa, M., & Hastie, R. (2003). Thelogic of social sharing: An evolutionary gameanalysis of adaptive norm development. Personalityand Social Psychology Review, 7, 2–19.

Kenrick, D. T., Sadalla, E. K., & Keefe, R. C. (1998).Evolutionary cognitive psychology: The missingheart of modern cognitive science. In C. Crawford

& D. L. Krebs (Eds.), Handbook of evolutionarypsychology (pp. 485–514). Mahwah, NJ: Erlbaum.

Krebs, D., & Janicki, M. (2004). Biologicalfoundations of moral norms. In M. Schaller &C. S. Crandall (Eds.), The psychological foundationsof culture (pp. 125–148). Mahwah, NJ: Erlbaum.

Kurzban, R., & Leary, M. R. (2001). Evolutionaryorigins of stigmatization: The functions of socialexclusion. Psychological Bulletin, 127, 187–208.

Markel, H. (1999). When germs travel. The AmericanScholar, 68, 61–69.

Markel, H., & Stern, A. M. (2002). The foreignnessof germs: The persistent association of immigrantsand disease in American society. The MilbankQuarterly, 80, 757–788.

Maykovich, M. K. (1975). Correlates of racialprejudice. Journal of Personality and Social Psychology,32, 1014–1020.

Neuberg, S. L., & Cottrell, C. A. (2002). Intergroupemotions: A biocultural approach. InD. M. Mackie & E. R. Smith (Eds.), From prejudiceto intergroup relations: Differentiated reactions to socialgroups (pp. 265–283). New York: Psychology Press.

Neuberg, S. L., Smith, D. M., & Asher, T. (2000).Why people stigmatize: Toward a bioculturalframework. In T. Heatherton, R. Kleck, J. G. Hull,& M. Hebl (Eds.), The social psychology of stigma(pp. 31–61). New York: Guilford.

Noy, D. (2000). Foreigners at Rome: Citizens andstrangers. London: Duckworth.

Nutton, V. (2000). Did the Greeks have a word forit? In L. I. Conrad & D. Wujastyk (Eds.), Contagion:Perspectives from pre-modern societies (pp. 137–163).Burlington, VT: Ashgate.

Oldstone, M. B. A. (1998). Viruses, plagues, andhistory. New York: Oxford University Press.

Orpen, C. (1973). The cross-cultural validity of theEysenck Personality Inventory: A test of Afrikaans-speaking South Africa. British Journal ofSocial and Clinical Psychology, 11, 244–247.

Park, J. H., Faulkner, J., & Schaller, M. (2003).Evolved disease-avoidance processes andcontemporary anti-social behavior: Prejudicialattitudes and avoidance of people with physicaldisabilities. Journal of Nonverbal Behavior, 27, 65–87.

Prentice, N. M. (1962). Ethnic attitudes,neuroticism, and culture. Journal of SocialPsychology, 54, 75–82.

Ray, J. J., & Kiefl, W. (1984). Authoritarianism andachievement motivation in contemporary WestGermany. Journal of Social Psychology, 122, 3–19.

Reynolds, V., Falger, V., & Vine, I. (Eds.). (1987).The sociobiology of ethnocentrism. London: CroomHelm.


352

04 046142 3/9/04 3:33 pm Page 352

Rozin, P., Haidt, J., & McCauley, C.R. (2000).Disgust. In M. Lewis & J. M. Haviland-Jones (Eds.),Handbook of emotions (2nd edn, pp. 637–653). NewYork: Guilford Press.

Rozin, P., Haidt, J., McCauley, C., & Imada, S.(1997). Disgust: Preadaptation and the culturalevolution of a food-based emotion. In H. MacBeth(Ed.), Food preferences and taste (pp. 65–82).Providence, RI: Berghahn.

Rozin, P., Markwith, M., & Nemeroff, C. (1992).Magical contagion beliefs and fear of AIDS.Journal of Applied Social Psychology, 22, 1081–1092.

Rudman, L. A., Greenwald, A. G., Mellott, D. S., &Schwartz, J. L. K. (1999). Measuring the automaticcomponents of prejudice: Flexibility andgenerality of the implicit association test. SocialCognition, 17, 437–465.

Schaller, M. (2003). Ancestral environments andmotivated social perception: Goal-like blasts fromthe evolutionary past. In S. J. Spencer, S. Fein,M. P. Zanna, & J. M. Olson (Eds.), Motivated social perception: The Ontario Symposium(Vol. 9, pp. 215–231). Mahwah, NJ: Erlbaum.

Schaller, M., Park, J. H., & Mueller, A. (2003). Fear ofthe dark: Interactive effects of beliefs about dangerand ambient darkness on ethnic stereotypes.Personality and Social Psychology Bulletin, 29, 637–649.

Schiefenhövel, W. (1997). Good tastes and badtastes: Preferences and aversions as biologicalprinciples. In H. MacBeth (Ed.), Food preferencesand taste (pp. 55–64). Providence, RI: Berghahn.

Sidanius, J., & Pratto, F. (1999). Social dominance: Anintergroup theory of social hierarchy and oppression.New York: Cambridge University Press.

Southwick, C. H., Siddiqi, M. F., Farooqui, M. Y., &Pal, B. C. (1974). Xenophobia among free-rangingrhesus groups in India. In R. Holloway (Ed.),Primate aggression, territoriality, and xenophobia(pp. 185–209). New York: Academic Press.

Stangor, C., & Crandall, C. S. (2000). Threat and thesocial construction of stigma. In T. F. Heatherton,R. E. Kleck, M. R. Hebl, & J. G. Hull (Eds.), Thesocial psychology of stigma (pp. 62–87). New York,Guilford.

Stat, D. (1995). Foreigners and Englishmen: Thecontroversy over immigration and population,1660–1760. Newark: University of Delaware Press.

Stephan, W. G., Diaz-Loving, R., & Duran, A. (2000).Integrated threat theory and interculturalattitudes: Mexico and the United States. Journal ofCross-Cultural Psychology, 31, 240–249.

Stephan, W. G., Ybarra, O., & Bachman, G. (1999).Prejudice toward immigrants: An integrated threat

theory. Journal of Applied Social Psychology, 29,2221–2237.

Suedfeld, P., & Schaller, M. (2002).Authoritarianism and the Holocaust: Somecognitive and affective implications. InL. S. Newman & R. Erber (Eds.), What socialpsychology can tell us about the Holocaust:Understanding perpetrator behavior (pp. 68–90).Oxford, UK: Oxford University Press.

AcknowledgementsThis research was supported by a Natural Sciencesand Engineering Research Council of Canadafellowship awarded to Jason Faulkner and a Social Sciences and Humanities Research Council of Canada grant awarded to Mark Schaller. We are grateful to Chris Crandall, Steve Heine,Doug Kenrick, and Ara Norenzayan for their helpful comments on drafts of this article. We also thank Gurbreena Dhanoa, Chloe Kelly, David Loran, and Jennifer Scott for their assistancein data collection.

Paper received 29 June 2003; revision accepted 12 March2004.

Biographical notesJ A S O N FA U L K N E R is a PhD student at the

University of British Columbia. His researchexamines stereotyping, prejudice, anddiscrimination from an evolutionarypsychological perspective.

M A R K S C H A L L E R recently co-edited the bookThe Psychological Foundations of Culture (2004,Lawrence Erlbaum Associates), and is currentlyco-editing the book Evolution and Social Psychology(forthcoming, Psychology Press).

J U S T I N H. PA R K is currently a PhD student at theUniversity of British Columbia. He investigatespsychological mechanisms that contribute topeople’s construction of their social reality, inparticular, mechanisms that have been shapedby evolution.

L E S L E Y A. D U N C A N is an MA student at theUniversity of British Columbia. Her researchinvestigates disease-avoidance processes andstigmatization.


353

04 046142 3/9/04 3:33 pm Page 353

r evolved disease-avoidance mechanisms and contemporary ...schaller/faulkneretal2004.pdf ·...

Documents