Research paper
The Evolution and Development of Inferential Reasoning about Ethnic Markers
Comparisons between Urban United States and Rural Highland Peru
by Cristina Moya and Robert Boyd
CA1 Online-Only Material: Supplement A
Social scientists have long argued about the relationship between ethnic phenomena, symbolic markers, and cultural traits. In this paper, we illustrate the potential of functionalist cultural and genetic evolutionary models to reconcile these debates. Specifically, we argue that we must take seriously the role of cultural similarity in delineating certain category boundaries if we are to understand the origins and development of ethnic stereotyping. We examine whether symbolic markers—namely, sartorial ones—are privileged in the development of social stereotypes by comparing how children and adults in the urban United States and rural highland Peru perform a categorization task. We find that arbitrary sartorial markers motivate generalizations about novel traits in all samples except among US children, even when they crosscut body morphology, emotional expression, and socioeconomic cues. Unlike children in the United States, children in the Peruvian sample demonstrate an even stronger reliance on sartorial and work site cues than do adults of the same community. This suggests a role for early-developing evolved biases that guide learning and require appropriate cultural inputs or different niches for adults and children. We document further cross-cultural variation, in that US participants privilege socioeconomic cues to occupational status more than other cues, whereas Peruvian participants rely on sartorial cues more than other cues, indicating the importance of cognitive rules for learning locally relevant social taxonomies.
Unlike other primates, humans symbolically mark group iden- tities and categorize others according to these markers. The use of personal ornaments, like ochre and shell beads, dates to the Middle Paleolithic (d’Errico and Vanhaeren 2009; Henshil- wood et al. 2011), although it is only in the more recent ar- chaeological record that we get clear evidence that markers mapped onto important social identities. Ethnoarchaeologists have devoted much effort to studying how to use ethnographic analogy to infer identities and patterns of behavior from sty- listic variation in the material record (David and Kramer 2001; Jones 1997). The enterprise revealed much variation in the ways boundaries were constituted, the extent to which symbols car- ried information, and how markers were associated with other norms. This work has necessarily been observational and fo- cused on the production side of stylistic variation. At the same time, psychologists have been using experimental methods to
explore receiver-side perceptions of arbitrarily marked social groups (Baron et al. 2014; Brase 2001; Rhodes and Gelman 2008). However, this work is rarely informed by the range of ethnographic variation in social group boundaries and tends to conflate functionally distinct social categories.
In this paper, we investigate the cognitive mechanisms hu- mans use to learn about the culturally structured social worlds they inhabit. If humans have been using symbolic markers for much of their evolutionary history—for example, to commu- nicate about interaction norms (Barth 1969; Wobst 1977)—it is plausible that they have evolved expectations that such mark- ers will be informative. We use ethnographic and experimen- tal data collected from children and adults in Huatasani, Peru, and Los Angeles, California, to determine whether heuristics for predicting strangers’ behaviors on the basis of their sym- bolic markings develop reliably in these two cultural environ- ments. The results shed light on the likely genetic and cultural evolutionary processes responsible for our stereotyping of sym- bolically marked groups. First, we describe a coevolutionary framework for understanding ethnic stereotyping and how it can help reconcile primordialist and constructivist debates about ethnicity in anthropology. Second, we outline competing hy- potheses about why stereotypes based on sartorial markers are common and how they develop. Third, we describe the meth- ods and ethnographic contexts where we conducted the stud-
Cristina Moya is a Postdoctoral Research Fellow in the Department of Human Evolutionary Biology at Harvard University (Peabody Mu- seum, 11 Divinity Avenue, Cambridge, Massachusetts 02138, U.S.A. [moya@g.harvard.edu]). Robert Boyd is Professor in the School of Hu- man Evolution and Social Change at Arizona State University (P.O. Box 872492, Tempe, Arizona 85287, U.S.A. [robert.boyd.1@asu.edu]). This paper was submitted 18 IV 15, accepted 26 I 16, and electronically published 26 V 16.
q 2016 by The Wenner-Gren Foundation for Anthropological Research. All rights reserved. 0011-3204/2016/57S13-0012$10.00. DOI: 10.1086/685939
Current Anthropology Volume 57, Supplement 13, June 2016 S131
ies. Finally, we discuss the inductive reasoning experiments and how they speak to the evolutionary and developmental origins of social category–based stereotyping.
A Brief Coevolutionary Account of Ethnic Stereotyping
Minimally, social scientists define ethnicity as identities that are self-ascribed. However, this does not exclude categories, such as gender or age set, that may require different analytical tools, given that they crosscut residential groups. A commonly implied additional entailment for a category to be ethnic is that membership is descent based or, more minimally, that it depends on having an attribute that is perceived as inheritedby descent (Chandra 2012). Such definitional limitations are useful but presuppose the answer to the question of how people de- lineate boundaries, and they do not seem to capture identities that rely more on performative attributes (e.g., Astuti 1995). For the purpose of this paper, we will define the set of cate- gories of interest as those that are symbolically marked but are not defined by demographic attributes that tend to vary within society, like age and gender. This working definition helps us frame the question around a function—namely, how humans learn and form stereotypes about symbolically marked clusters of individuals when category membership cannot be deter- mined from other visible features, like age and gender.
Humans have culturally evolved niches where symbolically marked clusters of individuals play a prominent role and can change the developmental, cultural, and genetic selection pres- sures (Odling-Smee, Laland, and Feldman 2003). This means that any complete account of ethnic stereotyping is likely to be a coevolutionary one (Moya and Henrich 2016). In response to these new social groups, further cultural evolutionary pro- cesses likely gave rise to shared beliefs about the ethnic bound- aries, children’s learning strategies likely changed, and natural selection may have acted on human cognition to help children learn these taxonomies. We explain such an account in mul- tiple steps, reviewing (1) a functionalist approach to stereo- typing, (2) how clusters of cultural traits and intentional mark- ing can culturally evolve in ways that foster stereotyping about symbolically marked groups, and (3) how cognitive systems must develop to accommodate cultural variation. We then out- line several predictions about (4) how categorization should develop across cultures and (5) the mechanisms that can be used for detecting meaningful symbolic boundaries. We end the introduction by describing the previous research on the cognitive development of symbolic group categorization.
Why Stereotype?
Stereotypes are the result of categorization systems that sim- plify and reduce real-world variation. This makes it easier to respond quickly to environmental variability, especially when one does not have complete information. However, this nec-
essarily means placing different people in the same category, increasing the risk of errors from overgeneralization. An individual-level functionalist approach predicts that social categorization systems carve up the world in ways that fa- cilitate useful predictions about strangers. Note that these are often, but need not be, accurate (see CA1 online supplement A). In contrast, most psychological theories regarding social categorization focus only on proximate motivations that shape stereotypes—for example, wanting to maximize the distinc- tiveness of one’s own group or have positive self-esteem by thinking highly of one’s group (Brewer 1991; Greenwald et al. 2002; Turner et al. 1987). Importantly, these motivations have not been robustly documented across cultures and may be at odds with ultimate adaptive (i.e., fitness-relevant) functions. For example, it is unclear that overestimating one’s group’s ability to defeat others in combat would be favored by natural se- lection, compared with an accurate cautious perception.
Useful concepts should have at least the following three features: (1) They should balance the benefit of simplification against the errors this creates (e.g., optimize the trade-off be- tween being able to make quick predictions about new people and coming to incorrect conclusions because of overgenerali- zation; Coley, Medin, and Atran 1997), (2) they should lead to predictions that allow individuals to meet adaptive goals (e.g., choosing reliable interaction partners or avoiding hostile strangers; Cottrell and Neuberg 2005; Hunn 1982), and (3) they should rely on easily detectable cues to group membership (e.g., visually or aurally salient cues). Next we discuss how cultural evolutionary processes tend to give rise to group boundaries that meet these criteria for promoting useful concepts.
Evolution of Cultural Clustering and Markers
A number of cultural evolutionary processes can produce clus- tered distributions of cultural traits across geographic and social landscapes.1 For example, copying others who are similar to oneself, imitating high-status locals, or doing whatever the ma- jority of others do can evolve under a wide range of circum- stances (Boyd and Richerson 1985; Henrich and McElreath 2003; Perreault, Moya, and Boyd 2012). These learning rules increase within-group homogeneity. Furthermore, unlike social learning in nonhumans, our faithful imitation (Lyons, Young, and Keil 2007; Tennie, Call, and Tomasello 2009), the intergen- erational accumulation of cultural knowledge (Boyd, Richer- son, and Henrich 2011), and the functional interdependence of cultural features can increase the set of traits that covary along the same cluster boundaries.
These mechanisms produce information-rich social bound- aries satisfying the first condition for making concepts useful. The fact that the boundaries can map onto coalitions, social
1. By clustered distributions, we mean nonuniform ones where multi- ple traits covary. Cultural traits can be behaviors, skills, norms, or institu- tions that are primarily acquired via social learning.
S132 Current Anthropology Volume 57, Supplement 13, June 2016
networks, or people with different interaction norms and skills means that they may promote fitness-relevant predictions about with whom, when, and how to interact. This satisfies the second requirement of useful concepts. Finally, signaling ethnic iden- tities can evolve, for example, to prevent interacting with oth- ers with different norms in coordination games (McElreath, Boyd, and Richerson 2003). Many ethnic markers seem to have culturally evolved to signal group membership in larger anon- ymous societies (Moffett 2013). These include sartorial cues, tattoos, dialectical variation, and even traits previously consid- ered to be only functional, like point shape (Wiessner 1983). People may categorize others even if no one signals their group membership (see CA1 supplement A), but selection for signal- ing produces clearer signs, thus making it easier to identify a stranger’s group membership. This ease of identification meets the third condition of boundaries that would promote useful concepts.
The empirical evidence that clusters of cultural traits map onto symbolic boundaries is more mixed. In examining the social significance of bead headband styles, Wiessner (1984) found little evidence that these tracked linguistic boundaries. Instead, styles diverged most at boundaries of interaction, where com- municating ones norms would be particularly important for coordination (McElreath, Boyd, and Richerson 2003). Simi- larly, stylistic variation maps onto only some ethnic bound- aries in northern Kenya—that is, those with higher between- group competition over limited resources, often despite frequent social interactions across the border (Hodder 1982). Other broader-scale work suggests that language phylogenies explain much variation in cultural traits, particularly in domains related to social organization, suggesting some linguistic boundaries would predict social norm variation (Mathew and Perreault 2015; but see Towner et al. 2012).
This evidence that stylistic markers are contingently asso- ciated with social and cultural behavior parallels debates about the nature of ethnic identities. Constructivist theories empha- sizing the contingent, political, and constructed nature of ethnic identity have supplanted more primordialist approaches stress- ing the deep history of, bounded nature of, and cultural content associated with ethnic groups (Gil-White [1999] reviews these positions). While strong primordialism is rare, several practi- tioners of a culture area approach to anthropology in the early twentieth century proceeded with empirical work as if ethnic groups were unproblematic, discrete cultural units of analysis (Wissler 1927), and more recently, some archaeologists have been critiqued for similarly static understandings (Sackett 1990). The fact that, in several social contexts, strategic ethnic shifts decouple identities from norms, values, and other in- stitutions (Barth 1969; Moerman 1965; Wimmer 2013) and that group membership signaling is often strategic (Wiessner 1983) led many social scientists since the mid-twentieth century to reject notions of ethnicity that rely on cultural similarity.
We believe unlinking our understandings of ethnic categori- zation from culture is premature, as it is the cultural nature of
ethnic phenomena that differentiates them from groups in other primates and social animals.2 Furthermore, a functionalist ap- proach to cognition requires an understanding of the mate- rial, real-world patterns that concepts reflect. This means that people’s perceptions of ethnic categories are unlikely to be completely arbitrary or divorced from some cultural content, even as they constrain the real-world variation that is deemed relevant.3 An outside (or etic) perspective on these phenom- ena still benefits from an inside (or emic) perspective for un- derstanding which elements of cultural repertoires are deemed important for delineating different social boundaries and en- gaging in predictable, fruitful intergroup interactions. Recog- nizing that ethnic phenomena are unlikely to emerge from a unitary cognitive architecture (Moya and Boyd 2015) also sug- gests that cultural content may be functionally relevant for some ethnic processes but not for others. For example, only a few cultural norms may be relevant to coordinating inter- group interactions or in-group cooperation. This may help ex- plain some of the discrepancies in the literature and why only some traits covary with symbolic markers that are meant to signal adherence to a subset of relevant norms.
Categorization Systems in Culturally Variable Worlds
If social landscapes have symbolically marked cultural clus- ters, and if predicting others’ behavior is beneficial, children should be adept at learning about these. The diversity of eth- nic boundaries throughout space and time and the fact that intraethnic social categories delineating roles (Bloch 2016), like age sets and genders, are often symbolically marked means that learning mechanisms must be capable of reliably acquiring a range of possible associations. Much as with language learn- ing, a developing child has to be equipped with the capacity to learn any number of social taxonomies that human societies have culturally evolved (Moya 2013).
Even plastic learning systems require rules to guide devel- opment. Heuristics biasing attention or expectations toward common indicators of cultural-cluster boundaries (e.g., lan- guage, dialect, and intentional sartorial choices) may facilitate the acquisition of useful stereotypes. However, to accom- modate diverse social worlds, these mechanisms require inputs from one’s social environment to develop properly. For ex-
2. Concepts of culture do not even figure in psychological theorizing on stereotyping. In the constructivist literature, it is more common to de- emphasize the relevance of cultural differences. For example, in her attempt at defining ethnicity, Chandra claims that “features such as a common cul- ture, common territory, common history or a common language are variables that sometimes distinguish ethnic identities rather than the constants that define them” (2012:10).
3. As an analogy, consider that color categories are a function of the physical properties of light and that our perceptual systems dismiss much of this real-world information, making us think that only those light waves with frequencies between the red and blue parts of the visible spectrum exist.
Moya and Boyd Inferential Reasoning about Ethnic Markers S133
ample, even if certain heuristics privilege categorizing others on the basis of sartorial markers, a developing child would still have to determine which kinds of clothing differences matter. Are ornaments or hairstyle part of the marker? This suggests mech- anisms akin to prepared-learning adaptations may be impor- tant, making it easier to learn certain associations than others. For example, children socially learn which animals are dan- gerous more readily than which are carnivorous (Barrett and Broesch 2012), humans and monkeys readily learn to fear snakes (Öhman and Mineka 2001), and US adults readily learn associations between aversive stimuli and novel minimal out- groups created in the laboratory and marked by shirt color (Navarrete et al. 2012). If this account is accurate, some, but relatively little, experience with real social groups that are sartorially marked would be necessary to trigger expectations that other clothing markers will be socially meaningful and worth stereotyping.
Children must also be capable of updating such readily learned associations. For example, in contexts where linguis- tic or sartorial cues are not indicative of important cultural norms, but religious affiliation is, children must be capable of updating their concepts to reflect the latter boundary. Not only is human cognition designed for massive cultural learning (Henrich 2015), but humans have also likely evolved to be ef- fective at teaching their children norms (Csibra and Gergely 2009; Kline 2015). Furthermore, we culturally evolve devel- opmental environments that facilitate the acquisition of locally appropriate concepts (Flynn et al. 2013).
Deriving Predictions from a Coevolutionary Account
Predictions about Developmental Trajectories across Cultures
For the above reasons, it is useful to study the development of social concepts across cultures, to reveal both the geneti- cally evolved heuristics that children use in forming stereo- types and how they learn culturally variable beliefs. Differ- ent theoretical accounts make different predictions about the cross-cultural and developmental patterns of clothing-based stereotyping. (1) If genetically evolved biases to privilege sar- torial cues for stereotype formation are at play, we expect greater cross-cultural similarity between children’s reasoning and more divergence among adults. Additionally, these biases may make children rely on clothing-based generalizations more than adults do, particularly in cultural contexts where adults have learned that sartorial markers are not the primary markers of socially meaningful boundaries. (2) Alternatively, humans may use structured learning rules that evolved for forming broader sets of associations to learn about the relationship between cloth- ing markers and other traits. These mechanisms might paral- lel those that elephants use to learn that certain human pop- ulations—that is, those with Masai ethnic markers—threaten them (Bates et al. 2007; McComb et al. 2014), despite no evo- lutionary history of sartorial marking in their species. This ac- count predicts that children would rely on clothing stereotypes
less than adults do, since they would have toacquire suchbeliefs. It is also possible that humans use cognitive mechanisms that evolved specifically for reasoning about symbolic markers in addition to a broader set of individual and cultural learning mechanisms.
Recognizing that children inhabit different social develop- mental niches and face different adaptive challenges than adults (Flynn et al. 2013) suggests additional functionalist predic- tions. For example, while younger children are dependent on adults to make social decisions for them, they do not need sophisticated categorization rules. Only as children expand their social networks do they start facing the problem of quickly predicting the behavior of potential interaction partners. This suggests a possible curvilinear relationship between age and clothing-based stereotyping; young children may have weak stereotypes about sartorial markers but start using them even more than adults as they start interacting with others inde- pendently. A curvilinear pattern could be consistent either with a prepared-learning account or with children having devel- opmental niches where clothing markers matter more for them than for adults.
We wish to be clear that children are socialized agents and members of their respective cultural worlds. We do not intend to interpret their responses as the output of asocial evolved cognitive processes alone. Rather, in the course of learning about their respective local social taxonomies, we expect children to bring to bear both the information that they have socially and individually learned, and evolved conceptual structures that fa- cilitate this learning.
Predictions about Mechanisms for Distinguishing Ethnic Symbolic Markers
Assuming people do form stereotypes about others on the basis of their sartorial markers, what is it about this cue that distinguishes it from other dimensions along which a stranger could be categorized? If the fact that the sartorial marker is a signal—that is, that it evolved to communicate information (Maynard Smith 2004)—motivates people to infer that others with the same clothing are similar, then they should also infer similarity on the basis of individuals’ emotional facial expres- sions. If people infer resemblance on the basis of any low-level feature of a figure, then they should be equally likely to make clothes-based and body shape–based inferences, because these characteristics cover nearly the same surface area of a person. If people infer that others are similar on the basis of the social relevance of their shared cue, then they should be equally likely to predict others’ behavior on the basis of their sartorial mark- ers and their having the same intraethnic role (e.g., occupa- tion). Finally, it is possible that sartorial markers promote more similarity inferences if these are cognitively privileged as eth- nic markers that convey information about cultural-cluster mem- bership.
For this study, we investigate these four cues (i.e., emotional expression, body morphology, occupational role, and clothing
S134 Current Anthropology Volume 57, Supplement 13, June 2016
style) to differentiate between these hypotheses (see table 1 for summary of cue affordances and fig. 1 for sample stimuli). Incidentally, comparing how people reason about these cues allows us to test two other hypotheses. First, markers that are stable and difficult to fake may be privileged indicators of eth- nic groupmembership (Cohen 2012; Nettle and Dunbar 1997; Sosis, Kress, and Boster 2007). While none of the cues we chose are costly signals, body morphology would be the most difficult one to change in a short period of time. Furthermore, morphological traits are often perceived to be more inter- generationally stable (Moya, Boyd, and Henrich 2015). This means that, if temporally stable cues promote more inductive inferences, participants should rely most on body morphol- ogy cues to categorize other individuals. Second, if very low- level visual salience is important to categorization, then our participants should predict similarities primarily on the basis of the occupational cue, because it represents a larger surface area (the background) of the image.
Previous Research on Reasoning about Symbolic Boundaries
While some evidence suggests that there are evolved biases for language-based stereotyping (see CA1 supplement A), most data on how people reason about sartorial cues has been conducted while addressing broader questions about social group reasoning. For example, Hirschfeld (1995) finds that 3-year-old children infer that characters with similar clothes would be similarly clad throughout development and would resemble each other, but only when clothing indicated the character’s occupation. When occupation was pitted against clothing color, children responded that the former was more stable through the life course (Hirschfeld 1995). More com- monly, sartorial indicators such as color are used as minimal markers for novel groups that are assigned in the laboratory (Dunham, Baron, and Carey 2011; Mahajan and Wynn 2011). However, these data do not show that sartorial cues are priv- ileged for motivating inferences and in-group preference, since similar research on the minimal group paradigm with adults reveals that other arbitrary bases of categorization, such as overestimating dots on a screen, may engender similar in- group preferences (Tajfel et al. 1971).
Furthermore, demonstrating mild preferences toward arbi- trary in-groups does not directly speak to how much children are willing to generalize on the basis of these minimal cat- egories. Other experiments show that children are willing to draw novel inferences about individuals on the basis of their
Table 1. Summary of functional features of social cues
Functional affordance
Social cue Signal Covers figure’s surface Socially salient More stable Evolutionary history as ethnic marker
Emotional expression Yes . . . . . . . . . . . . Body shape . . . Yes . . . Yes . . . Clothing style Yes Yes Yes . . . Yes Job location . . . . . . Yes . . . . . .
Note. Each row represents a social cue used in the study stimuli. The columns represent functional affordances associated with each cue and correspond to various hypothesized cognitive rules motivating category-based predictions. If a cue has one of the features hy- pothesized to be important, the relevant cell is marked “Yes.”
Figure 1. Examples of stimuli used in Huatasani. The character represented in a has the following cue values: sad facial expres- sion, fatter body morphology, geometric shirt and cap clothing style, and low-wealth rural job. The character represented in b has the opposite features: happy facial expression, skinnier body mor- phology, gray striped clothing, and high-wealth urban job. Across characters, the category cues were, in fact, crossed with each other, meaning that being happy, skinny, in a high-wealth work site, or dressed in gray striped clothing were not correlated at all with each other (see CA1 supplement A for Los Angeles stimuli and wider variation). A color version of this figure is available online.
Moya and Boyd Inferential Reasoning about Ethnic Markers S135
wearing similar shirt colors and that they do so to the same extent as they use race and gender (Diesendruck and Weiss 2014). However, these inductive inferences are trumped by information about similarity on other cultural traits (e.g., food preferences) when these are at odds with the visual markers. This suggests that shirt color may act as a placeholder for in- formation about cultural-cluster membership until a better predictor is offered. Furthermore, other developmental psy- chologists have failed to show generalizations based on ap- pearancealone inyoungchildren(Baronet al. 2014;Rhodes and Gelman 2008). The study by Baron et al. (2014) did find that adding labels to the visual markers promoted inferences, sup- porting other research showing that labeling can efficiently transmit information to children about the important ethnic markers in their environments (Heyman and Gelman 2000; Rhodes, Leslie, andTworek2012).Thesemixed resultsmay stem from the fact that shirt color is not clearly interpretable as a signal of group membership on its own.
The fact that the sartorial cues used in much of this liter- ature are intentionally simple (e.g., shirt color) and often re- flect within-population variation limits our ability to extrap- olate from these data to reasoning about more complex ethnic markers. In contrast, Hirschfeld and Gelman (1997) used more complex stimuli and showed that midwestern US children and adults do infer that more foreign clothing styles predict foreign language use. While children in this sample predicted language use most on the basis of race and dwelling style, they still used sartorial stereotypes as well, and adults used clothing style the most in their predictions. US adults also reason as if culturally acquired and intentional markers (e.g., a serpent tattoo) have high cue validity, meaning that people expect others to have the marker if, and only if, they are members of a group (Brase 2001). However, they do not commit the same logical falla- cies on the basis of less clearly intentional and culturally ac- quired cues, such as having attached earlobes. In most so- cieties where shirt color has been used as a sartorial marker in social cognition experiments, children probably have more experience with shirt color varying within populations (much as earlobe morphology does) than they do with it being used as a signal of group membership. For this reason, we use more complex sartorial cues that combine geometric design, shape, and color in our experiments. This allows us to investigate the development of social category inferences on the basis of a plausible cue to cultural-cluster membership and ethnically structured social relations.
Methods
Sites and Participants
The experiments were performed in two sites with very dif- ferent social taxonomies: Huatasani, Peru, and Los Angeles, California. The US participants were affiliated with the Uni- versity of California at Los Angeles (UCLA) and represent an urban and cosmopolitan demographic. Huatasani is a rural
village in the Peruvian highlands situated along the Quechua- Aymara linguistic boundary.4 Most people at this site engage in subsistence agro-pastoralism, but families supplement this activity to varying degrees with labor migration to mines or cities and as merchants.
Social boundary concepts in Huatasani differ in several ways from those more often studied by US psychologists. In con- trast to the racialized and essentialized ethnic taxonomies of the United States, Peruvian adults along the Quechua-Aymara linguistic boundary do not express primordialist sentiments about any of the local social groups. Furthermore, differences in market integration represent an important cultural bound- ary in the Peruvian fieldsite and are decoupled from mark- ers that commonly denote ethnicity elsewhere, like language or religion. Sartorial markers, on the other hand, are com- monplace to demarcate members of large categories in the Peruvian highlands, not just to delineate minority subgroups (e.g., punks or Mennonites), as is more common in the United States.
Different sampling strategies were used at each site. In Los Angeles, all interviews took place on the UCLA campus. Thirty-three second graders (mean age: 7 years) were recruited from the UCLA Lab School, and 52 undergraduate stu- dents (mean age: 20 years) signed up via the California Social Science Experimental Laboratory (CASSEL). In Huatasani, 167 people of more variable ages were recruited (see table 2 for age and gender distribution). The 34 children less than 9 years of age are used for comparison to the Los Angeles child sample, while the 86 adults over 18 years of age (mean age: 42 years) are compared with the UCLA undergraduate sam- ple. See CA1 supplement A for full developmental analysis using the Peruvian sample. In Huatasani, adult community members were recruited through snowball sampling with the help of a local research assistant, while children were recruited via schools and along with participating parents. Most inter- views were conducted in Spanish, and a few interviews were conducted in Quechua, but we do not expect this to have posed a barrier to participation (see CA1 supplement A for recruit- ment details).
Experimental Design
Participants were told that they would be introduced to fic- tional characters and asked to make predictions about them. A single character’s image (the exemplar) appeared on a com- puter screen with text describing a rare trait he had (e.g., liking banana tea; see CA1 supplement A for full list). On the next screen, participants were shown a grid with images of 16 men and asked to predict who else among them would have the same trait (e.g., would also like drinking banana tea). Once participants had chosen as many characters as they thought
4. C. Moya has conducted ethnographic fieldwork since 2007 in Huatasani and had lived there for a year at the time of this interview.
S136 Current Anthropology Volume 57, Supplement 13, June 2016
would share the trait, they proceeded to the next screen, where they were introduced to a new character with a new rare trait, and the process was repeated. Adult participants in Los Angeles made predictions about 16 traits, whereas all other samples were asked about a randomly assigned subset of nine traits (see CA1 supplement A for other protocol differ- ences across sites). These were pretested in Huatasani with two local assistants to ensure that the traits were understood but were not associated with any known social boundary. The order in which the traits were presented and which pictures were used as exemplars were randomized for each partici- pant, while the position of images in the grid was randomized within participants for each screen.
The characters’ images were generated using FaceGen software and varied along four dimensions: emotional, bodily, sartorial, and occupational cues (see fig. 1).5 Each character could have a happy or sad facial expression (emotional), be fatter or skinnier (body shape), wear a gray striped shirt or a shirt and cap with green geometric shapes (sartorial), and stand in front of an office environment or a tractor (occu- pational). To ensure cultural relevance, the occupational cue was changed for Peruvian participants to either a recogniz- able background of the largest mining town in the region or a field with sheep, like those outside the village. This mining versus agro-pastoralist divide corresponds to important dif- ferences in market integration and wealth, much as the office and tractors in the US stimuli correspond to socioeconomic differences between white- and blue-collar occupations. All of these features were varied independently so that there was no association between the cues. This resulted in 16 possible target characters for the grid (i.e., one character who shared all the traits with the exemplar, one who shared none, four who shared a single cue, and so on; see CA1 supplement A for sample grid).
Analysis
We ran multilevel logistic regression models predicting the probability of choosing that two characters are similar given that they share each of the four possible cues (coded as four independent dummy variables). Each model included inter- action terms between shared cue and participant age category
(as parsed in table 2) to test whether the effect of a shared cue changes with development. Models for each site were run separately. Because we have multiple observations per par- ticipant (i.e., they were asked about several traits), we include a random effect for participant in our models to deal with their nonindependence. We have a wider age range of par- ticipants in Huatasani, so we examine the developmental tra- jectory in more detail for that site, considering both finer gradations of intermediate age categories and continuous age models in CA1 supplement A.
Ethnographic Context of Sartorial Marker Use
Sartorial markers are important indicators of various social identities in the Huatasani region, both of ethnic-like bound- aries and intraethnic roles. Participants claim that Aymara- speaking women prefer brighter colors (e.g., fuchsias) than do Quechua-speaking women. However, these distributions are largely overlapping, and participants believe there is no dis- crepancy in the clothing of Aymara- and Quechua-speaking men. Furthermore, this linguistic boundary does not moti- vate many ethnic phenomena in this region (Moya and Boyd 2015). Huatasaneño adults form few stereotypes about this language boundary compared with market-integration bound- aries (Moya 2013). Nor do they treat linguistic identities as biologically transmitted or stable (Moya, Boyd, and Henrich 2015), and cooperation is not organized at the scale of lan- guage categories (Moya and Boyd 2015).
At the regional provincial level, there are some traditional stylistic differences in handmade clothing items and textiles that mark individuals’ (usually women’s) geographic origin. These can still be used to identify strangers from specific re- gions during market days in a neighboring regional hub. How- ever, such markers do not distinguish Huatasaneños and are increasingly rare among the younger generation outside of ritual celebration days.
Sartorial markers feature prominently during ritual dance and music competitions. On these occasions, musical groups wear matching costumes that distinguish them as a cooper- ative unit. These can range in elaboration from idealized tra- ditional garments—ostensibly from different regions—to color- coordinated everyday wear. The musical groups are organized along community or neighborhood lines, but they often have “collaborators” from distant towns in the Altiplano.
Most importantly, sartorial differences are clear indicators of degrees of market integration and can signal that one en- gages in cultural practices associated with more rural or more urban lifestyles. Market integration differences are more clearly marked in women. Rubber-tire shoes (ojotas), large skirts (polleras), certain hat styles (e.g., bowler hats), and woven carrying cloths (q’epis or awayus) are associated with indig- enousness and rural residence. Several younger women who had worked in larger cities spoke of switching clothing styles depending on their residence. Some of them preferred to wear pants and caps but, while living in Huatasani, conformed to5. The tool can be found at http://www.facegen.com.
Table 2. Participant age and gender distribution across sites
No. participants (% male)
Age (years) Huatasani Los Angeles
7–8 34 (62) 33 (52) 9–10 27 (74) . . . 11–18 20 (40) . . . 118 86 (33) 52 (52)
Total 167 (46) 85 (52)
Moya and Boyd Inferential Reasoning about Ethnic Markers S137
rural norms of wearing skirts and larger hats after being reproached for “thinking themselves men.” These bound- aries marking differential market integration in the region blur the distinction between ethnicity and roles. One may argue that such boundaries are becoming more ethnic in nature in- sofar as they correspond to regional cultural clusters (Moya and Boyd 2015) and delineate the set of norms and rules that guide individual decisions and interactions (Barth 1969).
It is difficult to compare the extent and significance of sartorial variation in the Peruvian highlands with that among our participants in Los Angeles. While clothing options abound in the United States, few of these are associated with social identities, and few ethnic or immigrant communities use dis- tinct sartorial markers. However, the number of minority sub- groups that use stylistic differences to mark cultural norm adherence, even on a college campus (e.g., goths, jocks, and hippies), may rival the variation and social significance of sar- torial cues in the Huatasani region. Experimental research in the United States suggests that clothing choices affect first impressions (Davis and Lennon 1988), although a study per- formed on our same Los Angeles student population suggests that natural variation in sartorial style similarity does not affect cooperation (Gervais et al. 2013).
Results
Figure 2 shows how sharing a social cue (e.g., emotional ex- pression, body shape, clothing style, and job site) affects peo- ple’s beliefs that two characters will share another trait. We report odds ratios (ORs) from regression models (i.e., the odds of believing the characters share a trait given that they share a cue relative to the odds of believing that they share a trait if they do not share a cue) along with their 95% confi- dence intervals (CIs). This means that the larger the values, the more heavily participants rely on this social cue to make predictions about new characters.6 In CA1 supplement A, we examine whether people made more assessments of similarity about some traits and if there were interactions between trait and cue type. There are few consistent interactions, so we treat all traits the same as repeated observations for the remaining analyses.
Developmental Patterns in Huatasani
Huatasani children broadly resemble adults in the extent to which they rely on different cues. Both adults and children in Huatasani rely most heavily on the characters’ clothing to make inductive inferences (fig. 2a). Both discard emotional expression information as a basis for making predictions. Across development, Huatasaneños also rely to a similarly
moderate extent on characters’ body morphology to draw similarity inferences. However, children rely on sartorial and occupational information more than do adults. Children under 9 years of age show 1.31 times the odds of making a clothing- based inference compared with adults (95% CI: 1.08–1.59) and 1.45 times the odds of making an occupation-based inference (95% CI: 1.21–1.77).
A more detailed examination of the developmental trajec- tories in the Huatasani sample reveals that the greater reliance on the occupation cue in childhood is restricted to children under 9 years of age (fig. 2a). On the other hand, the greater use of the sartorial cue increases in 9–10-year-old children before decreasing in teenagers. Continuous age models of socialization processes confirm an earlier and steeper reduc- tion in the use of occupational cues than in the use of sar- torial cues to make similarity inferences (see CA1 supple- ment A).
Developmental Patterns in Los Angeles
The effect sizes are all significantly lower among Angeleno children than among adults, meaning that children were less sensitive to all cues when making their decisions (fig. 2b).7
There may be methodological and motivational reasons why US children generally show smaller effect sizes than any of our other samples (see protocol differences in CA1 supple- ment A). As evidence of this, we see that children in Los Angeles are the only ones who believe that the characters are less likely to share a novel clothing trait if they are depicted wearing the same sartorial cue (see CA1 supplement A). Therefore, for the US site, it may be more insightful to com- pare the relative effects that the different visual cues have on predictions of similarity within each age category separately.
Even focusing on the within-age-category patterns confirms that the US samples less clearly resemble each other. How- ever, in the United States, both children and adults rely on the occupational cues more than other cues to make inferences. Adults rely nearly as heavily on sartorial cues as they do on occupational ones (OR: 2.18 and 2.29, respectively), whereas children rely on clothing only half as much as they rely on occupational cues (OR: 1.11 and 1.20, respectively). Instead, US children believe body shape to be nearly as predictive of novel trait similarity (OR: 1.19) as occupational cues are. Much like both Peruvian samples, US adults perceive body shape as having low (though significantly greater than zero) predictive power (OR: 1.42). Finally, US adults are the only sample to use emotional expression at all in their inferences,
6. Reassuringly, none of the effects are significantly below one, in- dicating that participants never thought that sharing a visual cue de- creased the characters’ likelihood of being similar on a novel trait.
7. Los Angeles children were not shy in making inferences about similarity generally. On average, they responded that 55% of targets shared the novel trait with the exemplar. This number was 45% for US adults and 47% and 40% for children and adults, respectively, in Huatasani.
S138 Current Anthropology Volume 57, Supplement 13, June 2016
although they rely on it significantly less than they do on either occupational or sartorial cues.
Discussion
Our results indicate that people primarily base inferences about others on the socially meaningful cues that they share— namely, in this study, clothing style and occupational back- ground. The developmental trajectories do not show evi- dence of cross-cultural diversification in adulthood from a more similar set of categorization rules in childhood, as one would expect if children’s responses reflected universal prior expectations. Nor do adults’ responses converge, as would be expected if the people were learning about similar niches in the two cultural contexts. This suggests we have to look at each site’s developmental process in more detail. These pat- terns do provide two important insights, one theoretical and one methodological. First, the developmental trajectory we see in the Huatasani sample shows that children prioritize occupational and sartorial categories more than adults do when predicting who will resemble each other, unlike in the United States (Hirschfeld and Gelman 1997). This suggests either that children use different heuristics than adults do for the task or that, in Huatasani, children and adults have dif- ferent developmental niches. Second, standard developmen-
tal psychology recruitment strategies that target undergradu- ate populations for adult samples and local elementary schools or neighborhood parents for child samples are likely sampling from different populations and thus underestimating cultural continuity across generations. Below, we discuss what the pat- terns with respect to each cue reveal about how humans parse their social worlds and then return to this last methodological point.
Emotional and Body Morphology Cues
Emotional facial expressions that evolved to communicate information about transient affective states were seldom used as a basis of inference. This means that a cue being a signal is not sufficient to promote stereotyping.
The fact that people relied little on body morphology for making inferences about the characters suggests that partic- ipants do not simply use any stable property of a person or any feature that covers the person’s body as a basis for in- ferring similarity.8 This casts doubt on accounts that expect
8. US children provide the one exception to this pattern in relying nearly as much on body morphology as they do on their highest ranked cue (occupation). We discuss possible interpretations of this in CA1 supplement A.
Figure 2. Developmental trajectory of cue reliance in Huatasani (a) and Los Angeles (b). Effects of characters’ sharing a cue (i.e., body morphology, emotional expression, job location, and clothing style) on the odds that participants believe them to share another trait are plotted against age in years. The Los Angeles sample did not include children or teenagers of intermediate ages. Odds ratios greater than 1 indicate that participants are more likely to expect characters to have the same trait if they share a social category cue. A color version of this figure is available online.
Moya and Boyd Inferential Reasoning about Ethnic Markers S139
stronger effects for social boundaries that are marked with genetically inherited or stable cues. It also supports previous research showing that visual salience alone is not an impor- tant contributor to children’s memory for social categories (Hirschfeld, 1993), to generalizations based on them (Baron et al. 2014), or to inferring traits are stable across time (Rhodes and Gelman 2008). While morphological cues are not developmentally privileged bases for categorizing others, they clearly become racialized in several cultural contexts (Kinzler et al. 2009; Kurzban, Tooby, and Cosmides 2001; Pietraszewski et al. 2015). This suggests that visually salient features of others likely require social or individual learning for children to map their relevance to an ethnic boundary or social role.
Occupational Cues
The strong reliance on occupational cues associated with socioeconomic status in most of our samples suggests that people are prone to make generalizations on the basis of so- cially meaningful categories, even if such social boundaries are recent historical innovations and vary within groups. The occupational cues at each site were different but locally rel- evant, and they had significant status connotations. In the United States, they denoted class, and in Peru, they denoted market integration differences. These socioeconomic cate- gories in complex societies have acquired many features of ethnic categories. For example, in the United States, educa- tional endogamy has been increasing (Mare 1991), and in the Peruvian highlands, market integration differentials imply a suite of other cultural characteristics and map onto regional origins (Orlove 1998).
Two additional pieces of evidence suggest that the occu- pational cues we used were developmentally privileged as a basis of generalization. First, Huatasaneño children relied on these cues significantly more than did adults at the same site. Second, both the US and Peruvian children weighed the occupational information highly relative to most other cues despite the beliefs diverging in adulthood across cultures. This is consistent with children sharing panhuman biases for priv- ileging occupational or status cues as a basis for making gen- eralizations about others. However, it is not clear why biases for stereotyping occupational categories specifically would have evolved. While such socioeconomic specialization within a society is relatively recent in human history, it is possible that socioeconomic or occupational differences are processed by cognitive mechanisms for reasoning about status hierar- chies or within-group roles. American infants show some understanding of dominance relations (Thomsen et al. 2011) and by age 5 exhibit sensitivity to novel social groups’ status (Horwitz, Shutts, and Olson 2014), suggesting these might be reliably developing intuitions. Another possibility is that the location backgrounds we used as occupational cues connoted geographic, and therefore ethnic, origin, especially for chil- dren in Peru who were more likely to be unfamiliar with the
mining context. The same visual stimuli can have different social interpretations at different developmental stages in Huatasani as children learn about their local social geography.
We were also surprised that adults in Huatasani relied relatively little on the occupational information, given the importance of market integration in shaping social identities in the Andes and at this site specifically (Moya 2013).9 This result, and the contrasting high levels of occupation-based categorization among Angelenos, may be due to differing perceptions about the mutual exclusivity of the activities. The majority of people who engage in mining in the Huatasani area also engage in agro-pastoralism. This means that adult participants could have interpreted our occupational cues as temporary indicators of where the character happened to be. In contrast to the Andean context, few people in the US labor market can simultaneously work in both white- and blue- collar jobs. This means the visual cues to occupation could more safely be used to assess characters’ socioeconomic roles in the United States than in Peru.
Sartorial Cues
While Angelenos privileged occupational cues, Huatasaneños in both age groups relied most on clothing cues to make predictions. Low-level differences in image processing are unlikely to account for this cultural difference,10 so we focus on possible explanations for the intersite differences that are specific to social cognition.
The fact that our Peruvian participants and adults in Los Angeles treated a fictional clothing marker as highly predic-
9. The discrepancy between these studies may be due to several meth- odological differences. First, the earlier study forced participants to make a language- or occupation-based prediction, meaning that the higher rates of occupation-based stereotyping may have reflected the relative unimportance of the Quechua-Aymara boundary to locals. Second, oc- cupational cues in this study reflected real-world locations and subsis- tence strategies, whereas the sartorial cues were fictional. We have found that making social category labels fictional increases reliance on eth- nolinguistic inferences relative to occupational ones at this site (Moya 2013). This means the fictional nature of the clothing marker could have swamped people’s reliance on the occupational information. Third, the occupational information was presented via verbal labels in the earlier study, whereas in this study it was presented via visual reference to the locations associated with agro-pastoralism and mining. This visual cue may have been too oblique a reference to market integration for partici- pants in Huatasani.
10. It may be that participants in different sites attend to different parts of images. For example, researchers have shown that people from the United States attend to features of focal objects relatively more than to image backgrounds compared with Chinese participants (Chua, Boland, and Nisbett 2005). We lack direct comparisons between US and Peruvian highlanders on holistic image processing. However, given the previous evi- dence that US participants tend to focus on objects in the foreground, this low-level explanation would predict that our US participants would rely more on the clothing and body morphology cues than the occupational cue provided in the background. However, we find the opposite result.
S140 Current Anthropology Volume 57, Supplement 13, June 2016
tive of other behavioral traits suggests the readiness with which humans imbue sartorial cues with social significance. The novel nature of the clothing marker could have moti- vated people to reason that this cue was more likely to be associated with rare traits (Risen, Gilovich, and Dunning 2007), if participants knew that the real-world occupations, body builds, and facial expressions were not associated with the novel traits. This may be part of the mechanism people use for parsing ethnic boundaries if they expect novel ethnic markers to be more common than novel body types or fa- cial expressions, for example. This is consistent with the fact that novel category labels increase ethnolinguistic generaliza- tions more than occupational ones at this site (Moya 2013). It may also explain why known cultural attributes that are not clearly signals do not motivate much essentialism in chil- dren (Baron et al. 2014; Rhodes and Gelman 2008). However, this proximate explanation alone cannot account for the cross- cultural and developmental variation in reliance on the sar- torial cue.
Huatasaneño children rely on clothing-based predictions even more than do adults in the same site. This bias is exag- gerated in middle childhood and starts attenuating in the teenage years. This early-developing bias is consistent with a role for evolved mechanisms that privilege intentional cul- tural markers, such as clothing style, in early development when forming and learning stereotypes. However, it is worth noting that US children in our study are relatively indiffer- ent to sartorial information. Previous research on Canadian children similarly found that 4-year-old children failed to generalize traits on the basis of shared visual features (hats or body color) unless the fictional groups were labeled, although 7-year-old children and adults did so on the basis of the visual stimuli alone (Baron et al. 2014). This means that, if evolved biases for forming stereotypes about sartorially marked bound- aries exist, they do not develop reliably in young children across all contexts.
The curvilinear developmental trajectory for reliance on clothing-based stereotyping in Huatasani (fig. 2) is consistent with either the importance of age-specific developmental niches or with a prepared-learning process that requires triggering with some social experience. The most plausible difference in developmental niches is the fact that childrens’ teachers dress in a more market-integrated fashion than do their par- ents.11 While this feature of the educational system persists in high school and is visible to adults, by adolescence people have broader social networks and can more accurately es- timate the importance of clothing style in the region. Alter- natively, a prepared-learning mechanism that requires appro- priate inputs to develop expectations that sartorial markers denote information-rich boundaries can also produce such a
curvilinear developmental trajectory; early experiences can pro- duce strong associations that then attenuate with further learn- ing. The US children in our sample may be missing such a triggering experience (e.g., if individuals in their social worlds are fairly homogenously clad).
The cross-cultural differences in children’s reasoning are worth further analysis. Mechanisms for privileging sartorial markers should not be particularly important until children start expanding their social networks and meeting strangers. For Angeleno children, this developmental phase probably starts later, given their few opportunities for independent movement and partner choice. This means that quick ste- reotype formation is a less critical task for them, given their coddled developmental niche. Parents in Los Angeles engage in more supervision of their children and control their social networks more than do parents in Huatasani due to the heavy reliance on cars in the city and differences in parenting norms.12
This implies that children in Huatasani need to start making adaptive social inferences earlier in their lives than do children in Los Angeles.
Furthermore, educational institutions in each site promote different kinds of associations with sartorial markers. The in- stitutions in Peru homogenize sartorial variation among their pupils by mandating uniforms. This means that the Peruvian children should be less likely than the US ones to believe clothing differences among peers are meaningful. On the other hand, sartorial differences between teachers and the pupil’s parents are much more marked in the Peruvian con- text than in the United States. These stylistic differences be- tween parents and teachers in Huatasani represent important distinctions in market integration and regional origin.13 This means that children in our Peruvian sample may be more attuned to the importance of sartorial differences as social group markers among adults because of their educational institutional niche. Future longitudinal or experimental learn- ing studies can help disambiguate between prepared-learning accounts and one that relies exclusively on broader learning rules responding to differences in socioecological niches.
Methodological Considerations for Developmental Work
The general resemblance between the Huatasaneño adults’ and children’s responses speaks to their shared cultural con- text. CA1 supplement A further shows the positive relation- ship between child and adult responses across specific traits. For example, Peruvian adults and children thought that char- acters who wore the same clothing would also share a ritual, whereas both adults and children thought those who wore
11. School uniforms also differentiate pupils of different schools, but this is not associated with meaningful cultural differences, since the three elementary schools in town have different uniforms but correspond to very similar neighborhoods.
12. This is consistent with previous findings that US adults believe parents socially influence their children, but Huatasaneños do not (Moya, Boyd, and Henrich 2015).
13. While some of the teachers are from Huatasani, the vast majority are from larger cities in the region and only stay in Huatasani during the school week.
Moya and Boyd Inferential Reasoning about Ethnic Markers S141
the same clothing would not necessarily have the same plant knowledge.
In contrast, it is troubling that the US samples of adults and children do not resemble each other much beyond their relying most on the occupational cues. In fact, there is no association between how much children and adults in Los Angeles made predictions about various traits given shared social cues (see CA1 supplement A). This might reflect our having sampled US children and adults from different cul- tural populations.14
Sampling different populations of children and adults also affects the kinds of social network structures that each are exposed to. Across cultures, children likely have access to fewer sources of information about social categories outside their kin networks than adults do. This generational differ- ence in network breadth is likely to be larger in the United States than in the Peruvian samples and may partly account for the noncorrespondence in adults’ and children’s re- sponses. This is because both (1) the US sample of children is particularly self-selected and (2) the adult US sample reflects a university environment that facilitates expanding and di- versifying social networks.
While we followed a standard sampling strategy in devel- opmental psychology for recruiting participants in Los Ange- les, our results should serve as a warning against such prac- tices and as motivation for doing ethnographically informed community-based research, particularly when intergenerational socialization processes are relevant to the research question. The two potential sources of bias described above suggest that much of developmental psychology might be underestimat- ing the importance of social influence when adopting similar university-based sampling strategies.
Conclusion
Symbolically marked social categories are pervasive and ap- pear to be an important part of human evolutionary history. Therefore, it is worth considering whether human social cognition bears evidence of design for reasoning about these complex cultural landscapes with variable ethnic bound- aries and intraethnic roles. We focused on the possibility that evolved biases, such as prepared-learning rules, favor the use of sartorial markers as the basis of novel stereotypes.
First, we show that our participants in Huatasani, Peru, and adults in Los Angeles, California, relied heavily on novel sartorial markers to make predictions about strangers. This
suggests a cross-cultural tendency to develop stereotypes on the basis of unfamiliar, but intentional, clothing cues. Review- ing the predictions from table 1, this is not simply because the clothing serves as a signal, is stable, or visually covers the characters’ surfaces, since facial expressions and body morphology did not promote as many inferences about sim- ilarity.
Despite these similarities across the adult populations, the developmental pathways in reasoning about clothing style differ in the two societies. This means that adaptations for social categorization fundamentally rely on cultural inputs to develop. The developmental patterns we document in the Peruvian highland site suggest that children are even more biased in their use of sartorial markers than are adults. This may reflect evolved biases, such as prepared-learning rules, that privilege symbolic markers when forming associations, the particular significance of clothing markers in Huatasani’s elementary schools, or both. However, Angeleno children rely relatively little on sartorial markers when deciding which characters will resemble each other. This implies that any evolved mechanisms at play rely on cultural inputs, and are supported by adaptations for social learning. Social institu- tions and developmental niches shaped by pedagogists may also facilitate the acquisition of these beliefs.
We also find that occupational information associated with socioeconomic status most promoted beliefs about behavioral similarity in both US samples. This suggests that class cate- gories may be acquiring more ethnic features in the US con- text insofar as people perceive them to be information-rich and socially meaningful boundaries or that they trigger psy- chological expectations about intraethnic roles.
The relatively greater importance of occupational cues in Los Angeles and sartorial ones in Huatasani suggests that evolved learning mechanisms may more generally promote stereotype formation about socially meaningful categories ac- cording to local context. However, the fact that, across such different contexts as Huatasani and Los Angeles, most par- ticipants relied heavily on novel sartorial cues to make in- ferences about rare traits, and that children in Peru did so even more than adults, points to people’s willingness to im- bue arbitrary symbolic markers with meaning and gener- alize on their basis. Any such biases only make sense in light of a coevolutionary history; living in culturally evolved social groups produced new selection pressures for acquiring ap- propriate and helpful ethnic concepts.
Acknowledgments
We thank participants in Huatasani and Los Angeles who made this work possible. The teachers at both sites and the staff at CASSEL further facilitated the smooth running of these experiments. Research assistants Mesa Dobek, Saida Calancho, Elise Waln, and Alex Arndt helped with stimuli production and administering the experiment. Polly Wiessner, Agustin Fuentes, participants in the Integrating Anthropology
14. UCLA undergraduates come from around the world, although primarily from California, and have quite a range of family backgrounds: 69% of participants identified as being of East Asian origin, and only 35% were English monolinguals. While we do not have the comparable statistics for the child population from Los Angeles, the children were predominantly of European American origin. These children also rep- resent a more homogenous and privileged segment of the Los Angeles community, attending a private elementary school.
S142 Current Anthropology Volume 57, Supplement 13, June 2016
Wenner-Gren workshop, Nicholas Davis, Chris Grassa, and an anonymous reviewer gave helpful feedback on earlier drafts of this paper. This research was funded by National Institutes of Health grant 1RC1TW008631 and an International Cogni- tion and Culture Institute minigrant.
References Cited
Astuti, Rita. 1995. The Vezo are not a kind of people: identity, difference, and ethnicity among a fishing people of western Madagascar. American Ethnologist 22(3):464–482.
Baron, A., Y. Dunham, M. Banaji, and S. Carey. 2014. Constraints on the acquisition of social category concepts. Journal of Cognition and Develop- ment 15(2):238–268. doi:10.1080/15248372.2012.742902.
Barrett, H. C., and J. Broesch. 2012. Prepared social learning about danger- ous animals in children. Evolution and Human Behavior 33:499–508.
Barth, Frederick. 1969. Ethnic groups and boundaries: the social organization of cultural difference. Boston: Little, Brown.
Bates, Lucy A., Katito N. Sayialel, Norah W. Njiraini, Cynthia J. Moss, Joyce H. Poole, and Richard W. Byrne. 2007. Elephants classify human ethnic groups by odor and garment color. Current Biology 17(22):1938–1942. doi:10.1016/j.cub.2007.09.060.
Bloch, Maurice. 2016. Imagination from the outside and from the inside. Current Anthropology 57(suppl. 13):S80–S87.
Boyd, R., and P. Richerson. 1985. Culture and the evolutionary process. Chicago: University of Chicago Press.
Boyd, R., P. Richerson, and J. Henrich. 2011. The cultural niche: why so- cial learning is essential for human adaptation. Proceedings of the National Academy of Sciences of the USA 108(suppl. 2):10918–10925. doi:10.1073 /pnas.1100290108.
Brase, Gary. 2001. Markers of social group membership as probabilistic cues in reasoning tasks. Thinking and Reasoning 7(4):313–346.
Brewer, M. B. 1991. The social self: on being the same and different at the same time. Personality and Social Psychology Bulletin 17:475–482.
Chandra, K. 2012. What is ethnic identity? a minimalist definition. In Con- structivist theories of ethnic politics. K. Chandra, ed. Pp. 51–96. Oxford: Oxford University Press.
Chua, Hannah Faye, Julie E. Boland, and Richard E. Nisbett. 2005. Cultural variation in eye movements during scene perception. Proceedings of the National Academy of Sciences of the USA 102(35):12629–12633. doi:10.1073 /pnas.0506162102.
Cohen, Emma. 2012. The evolution of tag-based cooperation in humans. Current Anthropology 53(5):588–616. doi:10.1086/667654.
Coley, J. D., D. L. Medin, and S. Atran. 1997. Does rank have its privilege? inductive inferences within folkbiological taxonomies. Cognition 64:73– 112.
Cottrell, Catherine A., and Steven L. Neuberg. 2005. Different emotional re- actions to different groups: a sociofunctional threat-based approach to “prejudice.” Journal of Personality and Social Psychology 88(5):770–789.
Csibra, Gergely, and György Gergely. 2009. Natural pedagogy. Trends in Cognitive Sciences 13(4):148–153. doi:10.1016/j.tics.2009.01.005.
d’Errico, F., and M. Vanhaeren. 2009. Additional evidence on the use of personal ornaments in the Middle Paleolithic of North Africa. Proceedings of the National Academy of Sciences of the USA 106(38):16051–16056. doi:10.1073/pnas.0903532106.
David, N., and C. Kramer. 2001. Ethnoarchaeology in action. Cambridge: Cambridge University Press.
Davis, L. L., and S. J. Lennon. 1988. Social cognition and the study of clothing and human behavior. Social Behavior and Personality 16(2):175–186.
Diesendruck, G., and E. Weiss. 2014. Children’s differential weighting of cues to social categories. Cognitive Development 33:56–72.
Dunham, Yarrow, Andrew Scott Baron, and Susan Carey. 2011. Conse- quences of “minimal” group affiliations in children. Child Development 82 (3):793–811. doi:10.1111/j.1467–8624.2011.01577.x.
Flynn, Emma G., Kevin N. Laland, Rachel L. Kendal, and Jeremy R. Kendal. 2013. Target article with commentaries: developmental niche construction. Developmental Science 16(2):296–313. doi:10.1111/desc.12030.
Gervais, Matthew M., Michelle Kline, Mara Ludmer, Rachel George, and Joseph H. Manson. 2013. The strategy of psychopathy: primary psycho-
pathic traits predict defection on low-value relationships. Proceedings of the Royal Society B: Biological Sciences 280(1757):20122773. doi:10.1098/rspb .2012.2773.
Gil-White, F. J. 1999. How thick is blood? the plot thickens: if ethnic actors are primordialists, what remains of the circumstantialist/primordialist controversy? Ethnic and Racial Studies 22(5):789–820.
Greenwald, A. G., M. R. Banaji, L. Rudman, S. Farnham, B. Nosek, and D. Mellot. 2002. A unified theory of implicit attitudes, stereotypes, self- esteem, and self-concept. Psychological Review 109(1):3–25.
Henrich, J. 2015. The secret of our success. Princeton, NJ: Princeton Uni- versity Press.
Henrich, J., and R. McElreath. 2003. The evolution of cultural evolution. Evolutionary Anthropology 12(3):123–135. doi:10.1002/evan.10110.
Henshilwood, Christopher S., Francesco D’Errico, Karen L. van Niekerk, Yvan Coquinot, Zenobia Jacobs, S. E. Lauritzen, Michel Menu, and Renata García-Moreno. 2011. A 100,000-year-old ochre-processing workshop at Blombos Cave, South Africa. Science 334(6053):219–222. doi:10.1126/science .1211535.
Heyman, G., and S. Gelman. 2000. Preschool children’s use of novel pred- icates to make inductive inferences about people. Cognitive Development 15 (3):263–280. doi:10.1016/S0885-2014(00)00028-9.
Hirschfeld, L. 1993. Discovering social difference: the role of appearance in the development of racial awareness. Cognitive Psychology 25:317–350.
———. 1995. Do children have a theory of race? Cognition 54:209–252. Hirschfeld, L., and S. Gelman. 1997. What young children think about the
relationship between language variation and social difference. Cognitive De- velopment 12(2):213–238. doi:10.1016/S0885-2014(97)90014-9.
Hodder, Ian. 1982. Symbols in action: ethnoarchaeological studies of material culture. Cambridge, MA: Cambridge University Press.
Horwitz, Suzanne R., Kristin Shutts, and Kristina R. Olson. 2014. Social class differences produce social group preferences. Developmental Science 17 (6):991–1002.
Hunn, Eugene. 1982. The utilitarian factor in folk biological classification. American Anthropologist 84(4):830–847.
Jones, Siân. 1997. The archaeology of ethnicity: constructing identities in the past and present. London: Routledge.
Kinzler, Katherine, Kristin Shutts, J. DeJesus, and Elizabeth Spelke. 2009. Accent trumps race in children’s social preferences. Social Cognition 27 (4):623–634.
Kline, M. 2015. How to learn about teaching: an evolutionary framework for the study of teaching behavior in humans and other animals. Behavioral and Brain Sciences 38(e31). doi:10.1017/S0140525X14000090.
Kurzban, R., J. Tooby, and L. Cosmides. 2001. Can race be erased? coalitional computation and social categorization. Proceedings of the National Acad- emy of Sciences of the USA 98(26):15387–15392.
Lyons, Derek E., Andrew G. Young, and Frank C. Keil. 2007. The hidden structure of overimitation. Proceedings of the National Academy of Sciences of the USA 104(50):19751–19756. doi:10.1073/pnas.0704452104.
Mahajan, N., and K. Wynn. 2011. Origins of “us” vs. “them.” Cognition 124 (2):227–233.
Mare, Robert D. 1991. Five decades of educational assortative mating. American Sociological Review 56(1):15–32. doi:10.2307/2095670.
Mathew, S., and C. Perreault. 2015. Behavioural variation in 172 small-scale societies indicates that social learning is the main mode of human adap- tation. Proceedings of the Royal Society B: Biological Sciences 282(1810): 20150061–20150061. doi:10.1098/rspb.2015.0061.
Maynard Smith, J. 2004. Animal signals. Oxford: Oxford University Press. McComb, Karen, Graeme Shannon, Katito N. Sayialel, and Cynthia Moss.
2014. Elephants can determine ethnicity, gender, and age from acoustic cues in human voices. Proceedings of the National Academy of Sciences of the USA 111(14):5433–5438. doi:10.1073/pnas.1321543111.
McElreath, R., R. Boyd, and P. Richerson. 2003. Shared norms and the evo- lution of ethnic markers. Current Anthropology 44(1):122–129.
Moerman, M. 1965. Ethnic identification in a complex civilization: who are the Lue? American Anthropologist 67(5):1215–1230.
Moffett, Mark W. 2013. Human identity and the evolution of societies. Hu- man Nature 24:219–267. doi:10.1007/s12110-013-9170-3.
Moya, C. 2013. Evolved priors for ethnolinguistic categorization: a case study from the Quechua-Aymara boundary in the Peruvian Altiplano. Evolution and Human Behavior 34(4):265–272. doi:10.1016/j.evolhumbehav .2013.03.004.
Moya, C., and R. Boyd. 2015. Different selection pressures give rise to dis- tinct ethnic phenomena: a functionalist framework with illustrations from
Moya and Boyd Inferential Reasoning about Ethnic Markers S143
the Peruvian Altiplano. Human Nature 26(1):1–27. doi:10.1007/s12110-015 -9224-9.
Moya, C., R. Boyd, and J. Henrich. 2015. Reasoning about cultural and genetic transmission: developmental and cross-cultural evidence from Peru, Fiji and the US on how people make inferences about trait transmission. Topics in Cognitive Science 7(4):595–610. doi:10.1111/tops.12163.
Moya, C., and J. Henrich. 2016. Culture-gene coevolutionary psychology: cultural learning, language, and ethnic psychology. Current Opinion in Psychology 8(1):112–118. doi:10.1016/j.copsyc.2015.10.001.
Navarrete, Carlos David, Melissa M. McDonald, Benjamin D. Asher, Norbert L. Kerr, Kunihiro Yokota, Andreas Olsson, and Jim Sidanius. 2012. Fear is readily associated with an out-group face in a minimal group context. Evolution and Human Behavior 33(5):590–593. doi:10.1016/j.evolhumbehav .2012.02.007.
Nettle, D., and R. Dunbar. 1997. Social markers and the evolution of recip- rocal exchange. Current Anthropology 38(1):93–99.
Odling-Smee, F. John, Kevin N. Laland, and Marcus W. Feldman. 2003. Niche construction: the neglected process in evolution. Princeton, NJ: Princeton University Press.
Öhman, A., and S. Mineka. 2001. Fears, phobias, and preparedness: toward an evolved module of fear and fear learning. Psychological Review 108(3):483–522.
Orlove, B. 1998. Down to earth: race and substance in the Andes. Bulletin of Latin American Research 17(2):207–222.
Perreault, Charles, Cristina Moya, and Robert Boyd. 2012. A Bayesian ap- proach to the evolution of social learning. Evolution and Human Behavior 33(5):449–459. doi:10.1016/j.evolhumbehav.2011.12.007.
Pietraszewski, D., O. Curry, M. Petersen, L. Cosmides, and J. Tooby. 2015. Constituents of political cognition: race, party politics, and the alliance de- tection system. Cognition 140:24–39.
Rhodes, Marjorie, and S. Gelman. 2008. Categories influence predictions about individual consistency. Child Development 79(5):1270–1287.
Rhodes, Marjorie, S. J. Leslie, and C. M. Tworek. 2012. Cultural transmission of social essentialism. Proceedings of the National Academy of Sciences of the USA 109(34):13526–13531. doi:10.1073/pnas.1208951109.
Risen, Jane L., Thomas Gilovich, and David Dunning. 2007. One-shot illusory correlations and stereotype formation. Personality and Social Psychology Bulletin 33(11):1492–1502. doi:10.1177/0146167207305862.
Sackett, J. R. 1990. Style and ethnicity in archaeology: the case for isochrestism. In The uses of style in archaeology. Cambridge: Cambridge University Press.
Sosis, R., H. Kress, and J. Boster. 2007. Scars for war: evaluating alternative signaling explanations for cross-cultural variance in ritual cost. Evolution and Human Behavior 28:234–247.
Tajfel, Henri, M. G. Billig, R. P. Bundy, and Claude Flament. 1971. Social categorization and intergroup behaviour. European Journal of Social Psy- chology 1(2):149–178. doi:10.1002/ejsp.2420010202.
Tennie, Claudio, Josep Call, and Michael Tomasello. 2009. Ratcheting up the ratchet: on the evolution of cumulative culture. Philosophical Trans- actions of the Royal Society B: Biological Sciences 364 (1528):2405–2415. doi:10.1098/rstb.2009.0052.
Thomsen, Lotte, Willem E. Frankenhuis, McCaila Ingold-Smith, and Susan Carey. 2011. Big and mighty: preverbal infants mentally represent social dominance. Science 331(no. 6016):477–480.
Towner, M. C., M. N. Grote, J. Venti, and M. B. Mulder. 2012. Cultural macroevolution on neighbor graphs. Human Nature 23(3):283–305.
Turner, J. C., M. A. Hogg, P. J. Oakes, S. D. Reicher, and M. S. Wetherell. 1987. Rediscovering the social group: a self-categorization theory. Oxford: Blackwell.
Wiessner, P. 1983. Style and social information in Kalahari San projectile points. American Antiquity 48(2):253–276.
———. 1984. Reconsidering the behavioral basis for style: a case study among the Kalahari San. Journal of Anthropological Archaeology 3(3):190– 234. doi:10.1016/0278-4165(84)90002-3.
Wimmer, Andreas. 2013. Ethnic boundary making. Oxford: Oxford Univer- sity Press.
Wissler, Clark. 1927. The culture-area concept in social anthropology. Ameri- can Journal of Sociology 32(6):881–891.
Wobst, H. Martin. 1977. Stylistic behavior and information exchange. In For the director: research essays in honor of James B. Griffin. Pp. 317–342. Ann Arbor: Museum of Anthropology, University of Michigan.
S144 Current Anthropology Volume 57, Supplement 13, June 2016
Copyright of Current Anthropology is the property of Wenner-Gren Foundation and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.