Meat Is Good to Taboo ¤ Dietary Proscriptions as aProduct ofthe Interaction of Psychological Mechanisms and Social Processes DANIEL M.T. FESSLER¤¤ and CARLOS DAVID NAVARRETE¤¤ ABSTRACT Comparingfood taboos across 78 cultures,this paperdemonstrates that meat,though aprizedfood, is also the principaltarget ofproscriptions. Reviewing existing explanations oftaboos, we nd that bothfunctionalist and symbolic approaches fail to account for meat’s cross-cultural centrality anddo not reect experience-near aspectsof food taboos, principalamong which isdisgust. Adopting an evolutionaryapproach to the mind,this paperpresents an alternative toexisting explanations of food taboos. Consistent with the attendant risk ofpathogentransmission, meat has special salience as a stimulusfor humans, asanimal products are stronger elicitorsof disgust and aversion than plant products.We identifythree psychosocialprocesses, socially-mediatedingestive conditioning, egocentric empathy , and normativemoralization ,each ofwhich likelyplays a rolein transforming individualdisgust responsesand conditioned food aversions into institutionalized food taboos. Introduction Culturalunderstandings concerning food, edibility, contamination, and re- latedtopics exhibit enormous variation across groups (Barer-Stein 1999; Rozin2000; Simoons 1994). However, despite such evident heterogeneity, investigators(e.g., Rozin 1987; Haidt et al.1997; Simoons 1994; Tambiah 1969)have offhandedlysuggested that animals and animal products seem especially likely tobe the focusof foodtaboos. The possibilityof uniformity ¤Harriet Whiteheadand Robert Aunger kindly suppliedus with draftsof their respective works.Paul Arguello assisted with research. FranciscoGil-White, RoyD’ Andrade,Robert Aunger,and Robert Sussman provided useful suggestions. We thank the many investigators whoshared taboo data and references. ¤¤Center for Behavior,Evolution, and Culture andDepartment ofAnthropology, UCLA, LosAngeles, CA 90095-1553. c KoninklijkeBrill NV, Leiden, 2003 Journal of Cognition and Culture 3.1 ° 2 DANIEL M.T.FESSLERANDCARLOSDAVIDNA VARRETE ina domainsubject to substantialvariation is ofgreatinterest, for such pat- ternscry out for explanation (Brown 1991). We have thereforeconducted the rstsystematic ethnological investigation of food prohibitions designed toevaluate the relative prominenceof animalproducts as a focusof taboos. Afterdescribing the methodsused in our investigation we present results demonstratingthat meat is indeed disproportionately represented in food taboos.Reviewing the twoprincipal existing explanationsof food taboos, wendthat neither functionalistnor symbolic approaches provides a com- pelling account.We thereforeconsider a variety ofpsychosocial processes thatmay contributeto taboo formation, highlighting the roleof emotions ineach process.Arguing that meat, while nutritious, is also potentially dangerous,we propose that natural selection has producedan ambiva- lence towardmeat such that, compared to other foods, meat is morelikely tobecome the targetof disgust. We then showhow this predisposition can articulatewith the psychosocialprocesses that generate taboos,thereby accountingfor the prominenceof meat infoodproscriptions. ACross-Cultural Studyof theTargets of Food Taboos The Sample ofFood Taboos and the Problem ofNon-independence Anthropologistsoften subsume proscriptions of markedly differingtypes underthe rubricof ‘ taboo’(Valeri 2000:43-6).While thispractice elimi- nates distinctionsthat are vital forunderstanding the detailsof any given culturalsystem, becauseour goal is toinvestigate patternsof human belief attheir broadest, we adopt an extreme versionof this approach, treating all foodproscriptions as equivalentregardless of whether they applyto all orpart of society, all orpartof aspeciesor fooditem, or all orpartof the calendaror life span.To collectdata on tabooswe conductedan extended survey ofprint and electronic ethnographies and studies in related elds. We alsocontacted investigators known to have workedon taboos, and em- ployeda snowballstrategy to learn ofothers possessing relevant data.So asnot to weightthe studytoward heavily-investigated societies,we avoided usingmany sourceson the same culture.While oursearch was, to our knowledge,the mostextensive ever undertaken,we make noclaim that it wasexhaustive. However,although there aredoubtlessly numerous sources thatwe did not encounter, there isnoreason to believe thatour methods biasedthe natureof the datacollected. Our search produced information MEAT ISGOODTOTAB OO 3 onfood prohibitions (henceforth used interchangeably with ‘ taboos’) in seventy-eight cultures.For each culture,the rst ve taboosencountered inthe text(s) werenoted (if fewer than ve tabooswere listed in the source, all werenoted). Limiting the numberof taboos per culture provided an initialconstraint on the disproportionatein uence ofsocieties having a largenumber of taboos; we selected ve asthe cut-offbecause inspection suggestedthat this was slightly abovethe average numberof taboos re- ported.Because wehad previously demonstrated that taboos imposed on pregnantwomen focus primarily on meat(Fessler 2002),pregnancy taboos wereexcluded in orderto avoidprejudicing the results.Using foodgroups commonlyemployed in the nutritionliterature, the numberof taboos was talliedfor the followingcategories: meat, vegetables, fruit,sweets, dairy products,and starches. Because ourwork on pregnancy taboos suggested thatspicy foods have specialsalience asthe targetof taboos, we included thisas a seventh category.Results aredisplayed in the Appendix.For illustrativepurposes, Table 1 presentsthe average numberof taboos on meatversus taboos on all non-meatfoods. Note that, in each region,meat taboosare more frequent than non-meattaboos, and, with the exception ofSoutheast Asia, this difference is quitemarked. Ifall ethnographieswere independent sources of data,simple statistical tests couldbe performed on the Appendix.However, such an approach riskscommitting Galton’ s error,the assumptionof independence among societiesthat do not constitute independent data points. The possibilityof i) the diffusionof cultural traits between neighboringsocieties, and ii) the recentsplintering of once homogeneous groups makes any suchtest prob- lematic.Previous attempts to solve thisproblem employ autocorrelation techniquesto recalibrate in ated sample sizes (Naroll 1976) or contin- gency tablesthat collapse matrices by the amountof excess tobalance the numberof cells ineach independentcategory (Strauss & Orans1975). These techniquesunderestimate the varianceof data points within inde- pendentsamples and may skew condence intervals infavor of the null hypothesis.However, these problemscan be managedusing bootstrapping techniques.Standard bootstrapping involves the samplingof the original data(with replacement), and the executionof some formula of interest to createa samplingdistribution of the statisticof interest on which hypoth- esis tests canbe performed (Fox 1997).Although this method solves the 4 DANIEL M.T.FESSLERANDCARLOSDAVIDNA VARRETE Table 1 Average number ofmeat and non-meat taboosper societyby geographic region Region Meat non-Meat Australia 4.44 0.33 C. America 2 0 East Asia 3 1.33 Europe 5 0 Mid. East 5 0 N. Africa 2.4 1.4 N. America 3.2 0.1 Oceania 2 0.85 S. Africa 3.33 0.58 S. America 3.5 0.1 S. Asia 2.6 0.4 S. E. Asia 1.88 1.63 Totals 38.35 6.72 Total Mean 3.2 0.56 S.D. 1.59 0.13 problemof non-normally distributed error, it does not solve the problem ofnon-independence, as bootstrapping too assumes that the originalset iscomposed of independent data points. We thereforedevised a novel bootstrappingtechnique that corresponds to a z-test forproportions, uses all available data,does not underestimate variances by averaging out data pointsinto contingency tables, and minimizes Galton’ s problemof non- independence. Methods ofStatistical Analysis We organizedall ofthe societiesin our sample into twelve geo- graphic/culturalregions that we treat as roughly independent areas. This approachis conservative in that, by clustering societies in categorieslarger thancommonly-recognized culture areas, we decreasedthe likelihoodthat, ifwe compared two societies from different categories, they wouldshare substantivehistorical links. We wrotesoftware that randomly chose a geo- graphicalregion and then randomlyselected oneculture from within that MEAT ISGOODTOTAB OO 5 region.1 The programrepeated this process with replacement until the numberof samples equaled the numberof geographic regions ( n 12). D The frequenciesof taboos for each foodcategory (e.g. meat, fruit, etc.) werethen convertedto proportions that describe the relative occurrence oftaboos in the meat categoryversus the relative occurrenceof taboos ineach non-meatcategory, resulting in a totalof seven tests ofdiffer- ences ofproportions. The processwas repeated 10,000 times to create, foreach comparison,a distributionof differences. p-values wereestab- lishedby calculating the percentageof cases (outof 10,000) in which the value differencesfor each comparisonwas less thanor equal to zero. One mightargue that each non-meattaboo does not represent a categoryfor a specic tabooedfood, as some societies might de ne food taboossimply with regard to meat versus non-meat. A test ofdifferences inproportions was therefore also performed for the differencebetween meattaboos and all otherfood taboos combined. The argumentcould also bemade that multiple meat taboosshould not be treated as independent entities,as