Why do we predict someone’s political beliefs from their beard style? Or their food preferences from how they pronounce key vowels?

Stereotyping is the process of predicting an individual’s traits on the basis of their belonging to a category[i]. The fact that humans form stereotypes about social groups so readily makes sense if we consider our evolution as a cultural species.

Imagine you hear someone trying to order a “hero” at a sandwich shop. Either: 1) you’re a New Yorker, think nothing of it and keep pondering which fillings will go in your hero, 2) you think this person is not from around here and wait for more clues about their origins, or 3) you expect that this person is more socially liberal, more impatient, more likely to be brusque, and more likely to be a Yankees fan than the average American.

Contrast this with a chimpanzee’s response when she encounters an unknown male while searching for fruit on her own. She becomes tense, recalling the one important stereotype she has about outgroup males: they are a threat. If she has recently given birth, this male is more likely to kill her infant than a known male would be. If she is outnumbered by outgroup males she risks being battered to death.[ii] But beyond knowing that this unknown individual is threatening, and of a certain age and sex, this wandering chimpanzee can make relatively few predictions about the stranger’s behavior.

First we should note that in contrast to many primates, humans are remarkably tolerant of strangers.  We should take a moment to appreciate the fact that we do not fear for our lives when meeting someone new. And yet, we hold many more stereotypes about members of different social groups than chimpanzees seem to. In fact, the willingness to interact across cultural group boundaries may be part of the reason humans develop as many social group stereotypes as they do. For the rest of this article, we will try to understand why humans became such promiscuous stereotypers. First we explain why other species are unlikely to have many stereotypes about social groups. Then we turn to the main two conditions that would promote stereotyping; 1) marking of group membership, for example with dialect or clothing differences, and 2) having multiple cultural traits cluster with each other[iii]. Finally, we describe some ways that stereotypes can become inaccurate.

Why chimps have only one social group stereotype
Why couldn’t the lone chimpanzee in our intergroup scenario predict much about the stranger she encountered? Well first she would have a hard time telling what group the chimpanzee belonged to. While chimpanzee groups have distinct cultural traditions, for example in their feeding technology,[iv] to our knowledge they do not use auditory or visual features to signal group membership.

Second, even if our lone chimpanzee forager identified which group the stranger belonged to, this would not help her predict how to interact with him, or whether she should migrate to his group upon reaching reproductive maturity. Group membership is hard to detect for chimpanzees and the culturally transmitted differences between groups do not meaningfully impact interactions: in short, chimpanzees have little use for stereotypes about social groups beyond fearing unknown males.

The chimpanzee example reveals the two central conditions for having and using group stereotypes: the ability to easily detect group membership and having multiple useful predictions one could make on its basis. This is not to say that no other species has group markings and the potential to use these to predict others’ behaviors. In several species of whales, vocalizations indicate which clan an individual belongs to, which is particularly useful when several clans range in the same areas[v]. These markers of clan membership are also correlated with several important behaviors such as foraging techniques, types of social relations, migration and movement patterns. It is unclear whether whales use stereotypes about these behaviors upon hearing an individual with a different clicking sound sequence. But we do know that whales preferentially associate with others who use the same clicks. This indicates that whales can identify outgroup members on the basis of a marker.

Since whales can identify individuals as belonging to outgroups, might even identify the specific outgroup on the basis of the vocalization, and would potentially benefit from predicting their cultural traits, I expect they are more likely than chimps to be stereotypers. However, a definitive answer awaits empirical research.

To wear a fleece or a sports coat or Why people mark group identities
Like the whale clans, and unlike chimpanzee groups, many human social groups are marked either by readily detectable linguistic or stylistic differences. For example, where I work in southern Peru, women from the lake area wear sashes with woven animal figures, while nearby women further inland wear black clothes with brightly colored embroidery. This satisfies one of the conditions for stereotype use: being able to detect an individual’s category membership even if you do not know them. Whether these group markers are intentional or not doesn’t matter for others who will readily use them to predict things about you. For example, it’s unlikely that New Yorkers coined the term heroes as a marker of their identity, but others can still use this speech pattern to guess their regional origins.

There are, however, several reasons why humans mark their group identities. One motivation is communicating to others that you have certain preferences, values, and expectations about how social interactions should proceed.[vi] For example, wearing an outdoorsy fleece can signal that one likes hiking, is informal, cares about environmental policy, etc., and will thus attract like-minded friends who will not judge you for having different norms. This same marking strategy can be used to identify people with complementary and different traits, such as trading partners with different skills or resources. A different kind of marker can serve to show commitment to a group by signaling an allegiance and burning bridges with alternative groups. For example, gangs often expect their members to visibly display tattoos that show their affiliation. Permanently displaying that you were affiliated with a gang will limit your ability to be trusted by rival gangs

The mystery of the latte-drinking Volvo driver or Why cultural traits cluster together
The second important ingredient for fostering stereotyping is correlations between markers of belonging to a category and several other traits. It is true that the chimpanzee forager in our example invoked a stereotype to predict that the outgroup male was a potential threat, even though she had never met him before. However, the predictions she could make were minimal. Paying attention to someone’s category membership is more useful if there are multiple predictions one could make on its basis, and if the best responses are going to differ according to these stereotypes. This brings us to the mystery of the latte-drinking Volvo driver.

During the 2004 Democratic presidential primary a particularly memorable conservative TV ad lambasted then front-runner Howard Dean invoking images of his being a “…tax-hiking, government expanding, latte-drinking, sushi-eating, Volvo-driving…” candidate.[vii]  Such caricatures were clearly exaggerated for effect, but why are the correlation between government spending, food and car preferences so plausible to invoke when on the surface these traits seem causally independent of each other? Why are politically conservative college students more likely to have sports decor, laundry baskets and postage stamps in their dorm rooms, while liberal college students are more likely to have maps as decoration, art supplies and many books?[viii] We may be able to find environmental differences underlying some of these correlations – for example, sushi and liberals are easier to find near water. To understand many of the other correlations and more generally the emergence of these cultural clusters it’s useful to delve into how, and why, humans learn from others.

Social learning is good for efficiently acquiring a lot of information, but some of the mechanisms that humans use to learn from others are also prone to produce cultural clusters. Here we describe three ways this can happen.

  • We pay attention to, and copy, people that we consider prestigious, even in domains that may be unrelated to their success. Generally this is a good heuristic for getting information, but it might lead us to copy the medicinal plant and fishing techniques of the best yam grower in town, even if he knows little about the other activities[ix].
  • Humans are also faithful imitators of others’ actions. If shown a video of someone getting candy out of a box in an inefficient way, people are much more likely to copy irrelevant actions, such as tapping the box with a feather, than a chimpanzee is.[x] Faithful copying is more common when engaging with boxes that are difficult to understand. This suggests that such blind imitation may be favored in part because the technologies and institutions that humans have created are complex, and often function because of mutually dependent elements that are hard to understand.
  • There is also evidence that we copy others more faithfully when we wish to appease, or affiliate with, group members.[xi] This motivation likely explains much of the clustering along political ideological lines. Why others will like me more if I dress like them, talk like them, and hold the same beliefs about GMOs as they do is still somewhat of a puzzle. The previous section suggested that coordination and commitment may help explain some of this phenomenon, but research is ongoing in this domain.

These forces add to the environmental differences that influence regionally defined groups to produce landscapes of cultural clusters that promote the use of simple heuristics and stereotypes to make multiple predictions about others.

At this point you may be wondering if this kind of cultural clustering of traits is a recent historical phenomenon, only relevant in large post-industrial nation states where subcultures can specialize and abound. In fact, the ethnographic record is full of examples of cultural group boundaries with similarly clustered norms. For example, among speakers of Batak in Sumatra women traditionally went to live near their husband’s family after the groom’s family paid her family for the marriage, they were not allowed to instigate a divorce or inherit, and the society is predominantly Christian. In contrast, the neighboring Minangkabau speakers were more likely to move near the wife’s family after marriage, allowed women to inherit and divorce, and are predominantly Muslim[xii]. In fact linguistic boundaries may have been such a common indicator of cultural clustering in human social groups that children use it as a way of making predictions about new people, even in cultural contexts where adults do not[xiii]. The archaeological record also shows that early Homo sapiens acquired resources from far away. For example, our ancestors used obsidian from 200 km away by the Middle Stone Age. These patterns are consistent with long distance trade that likely involved interacting across cultural groups.

The mystery of the lazy, but industrious, migrant or How inaccuracies creep in
If stereotyping is meant to take advantage of the real correlations between group markers and cultural traits to make predictions about strangers, why do they often seem to bear little resemblance to reality, or become exaggerated claims? For example, two seemingly contradictory stereotypes about Mexicans being lazy, and hard-working persist in the United States – try searching the stereotypes online yourself; you’ll get nearly the same number of hits for each. Clearly, in their simplest form they cannot both be accurate[xiv].

There are many causes of stereotype inaccuracy, but let’s highlight two here. First, the same social learning mechanisms that can give rise to clustered distributions of traits can also be vulnerable to acquiring inaccurate or outdated information from others. This is particularly problematic when people who are in better positions to be heard have inaccurate information, or incentives to be misleading. For example, a majority of Americans consistently believed that crime rates were worse now than in the past year, even during long periods of dropping crime rates[xv]. News reports driven by profit rather than accuracy have been implicated in such inaccurate perceptions about crime, and influence social stereotypes about groups[xvi].

Second, we may have genetically evolved heuristics for privileging certain cues, like language use, as indicators of cultural group membership. In most environments during the course of human evolution these heuristics likely produced accurate stereotypes on average. However, today we may systematically misapply them in contexts where language is not associated with cultural clusters. Furthermore, we live in more complex, dense, and large-scale societies, with more cross-cutting group identities today. This means that social group boundaries no longer simply reflect neighboring regional groups. Getting accurate information about such groups may pose more of a challenge today, than it did in the past.

The good news is that neither of these explanations can sustain inaccurate stereotypes over the long run if people update their beliefs by careful analysis of socially and individually learned information.

References and Footnotes
[i] We will use the term stereotype whether the belief is accurate or not, whether the belief is fixed or malleable, and whether it is widely held or a minority belief. See Jussim, Lee, Jarret T. Crawford, and Rachel S. Rubinstein. “Stereotype (in) accuracy in perceptions of groups and individuals.” Current Directions in Psychological Science 24.6 (2015): 490-497 for further justification for this use of the term.

[ii] Wrangham, Richard W., Michael L. Wilson, and Martin N. Muller. “Comparative rates of violence in chimpanzees and humans.” Primates 47.1 (2006): 14-26.

[iii] Moya, Cristina, and Robert Boyd. “The evolution and development of inferential reasoning about ethnic markers: comparisons between urban United States and rural Highland Peru.” Current Anthropology 57.S13 (2016)

[iv] Whiten, Andrew, et al. “Cultures in chimpanzees.” Nature 399.6737 (1999): 682-685.

[v] Rendell, Luke, and Hal Whitehead. “Culture in whales and dolphins.”Behavioral and Brain Sciences 24.02 (2001): 309-324.

[vi] McElreath, Richard, Robert Boyd, and PeterJ Richerson. “Shared Norms and the Evolution of Ethnic Markers.” Current Anthropology 44.1 (2003): 122-130.

[vii] The list of descriptors was longer in the original context. The political use of such stereotypes has been the subject of some research e.g. Nunberg, Geoffrey. Talking right: How conservatives turned liberalism into a tax-raising, latte-drinking, sushi-eating, Volvo-driving. Publicaffairs, 2007.

[viii] Carney, Dana R., et al. “The secret lives of liberals and conservatives: Personality profiles, interaction styles, and the things they leave behind.”Political Psychology 29.6 (2008): 807-840.

[ix] Henrich, Joseph, and James Broesch. “On the nature of cultural transmission networks: evidence from Fijian villages for adaptive learning biases.”Philosophical Transactions of the Royal Society of London B: Biological Sciences 366.1567 (2011): 1139-1148.

[x] Whiten, Andrew, et al. “Emulation, imitation, over-imitation and the scope of culture for child and chimpanzee.” Philosophical Transactions of the Royal Society of London B: Biological Sciences 364.1528 (2009): 2417-2428.

[xi] Watson-Jones, Rachel E., et al. “Task-specific effects of ostracism on imitative fidelity in early childhood.” Evolution and Human Behavior 35.3 (2014): 204-210.

[xii] Loeb, Edwin Meyer. “Patrilineal and matrilineal organization in Sumatra: the Batak and the Minangkabau.” American Anthropologist 35.1 (1933): 16-50.

[xiii] Moya, Cristina. “Evolved priors for ethnolinguistic categorization: A case study from the Quechua–Aymara boundary in the Peruvian Altiplano.”Evolution and Human Behavior 34.4 (2013): 265-272.

[xiv] In fact Mexico consistently ranks among the OECD countries with the highest average number of hours worked per year, per worker. http://stats.oecd.org/Index.aspx?DataSetCode=ANHRS

[xv] http://www.gallup.com/poll/150464/americans-believe-crime-worsening.aspx

[xvi] Gilliam Jr, Franklin D., and Shanto Iyengar. “Prime suspects: The influence of local television news on the viewing public.” American Journal of Political Science (2000): 560-573.


Published On: August 24, 2016

Cristina Moya

Cristina Moya

Cristina Moya is a Postdoctoral Research Fellow in the Human Evolutionary Biology Department at Harvard University. Her research focuses on how humans adapt to culturally-structured social worlds. Her primary research areas focus on how people reason about cultural group boundaries and make reproductive decisions. She uses ethnographic, developmental psychological, and cross-cultural comparative methods to address these questions. She has conducted extensive fieldwork along the Quechua-Aymara language boundary in the Peruvian Altiplano to understand the nature of reasoning about linguistic social groups.


  • Matt Zefferman says:

    Do we know if chimpanzees stereotype less than humans in non-social contexts (e.g., generalizing about catagories of inanimate objects)?

    • Cristina Moya says:

      Good point. There’s lots of research showing competent categorization, including of novel kinds in non-humans (primarily in pigeons and monkeys). It also shows differences between them and humans in certain generalization abilities. However, it’s hard to determine what would be a comparable set of boundaries to test across species. Should we really expect primates and pigeons to care as much about chairs as humans do? Arbitrary rotated 3D objects are often used as “neutral” stimuli for comparative categorization studies, but even these may be more important in human’s lives than in caged pigeon’s lives given our built environment and tool-use. In short, I wouldn’t be surprised if we extrapolated more on the basis of inanimate category kind than other species did, but it wouldn’t be a very functional hypothesis to predict otherwise. It would be interesting to see if there were category kinds that were more relevant to other species and had lots of information structure for which they extrapolated more.

      • Joe Brewer says:

        Hey Cristina,

        Your comment got me thinking about an example I use when teaching about frame semantics. I hold up a glass of water and explain how it is understandable as a socially constructed mental object suited for grasping by the human hand and prototypically used for drinking liquids. I then go on to describe the semantic frame for the glass as a container with various roles, relationships, and inferential logics — all of which are socially constructed for motivated reasoning by human beings as they interact with containers that hold fluids in their daily lives.

        None of these social constructions would necessarily be present in the mind of a chimpanzee, though they may have their own semantic frames for objects that are meaningful in their social worlds. Gets me thinking about how challenging it will be to conduct studies on frame semantics and conceptualization in non-human species.

        Important stuff!

        • Cristina Moya says:

          Hey Joe,
          Thanks for the comment. Sounds like a different disciplinary language for talking about a lot of related concepts. So yes, I wonder how much this calls into question the interpretation of the comparative cross-species categorization work. It may be that for some basic boundaries related to the physical world, species would have similar semantic frames, but the parameter space for those kinds of similarities seems quite limited.

          • Joe Brewer says:

            This is aligned with my thinking. What I see is that there will be basic “body logic” structures for choreographed movements, object detection, intuitive physics, image tracking (for predators to direct their actions toward prey, for example), and so forth.

            These fit into the rubric of basic-level categories for the research on human conceptualization and that of primary metaphors in metaphor theory.

            The common ground between humans and other animals comes through our shared embodiment. Our differences in conceptual abilities are very important to map out, as are the details of the similarities. I see this as a strong supporting argument for cross-cultural and cross-species research in the future. 😉

          • Cristina Moya says:


  • Brian McInnis says:

    You just stereotyped chimpanzees, humans, and possibly whales.

  • Andrew O'Keeffe says:

    Hi Cristina, read your article with interest – anything mentioning chimps catches my attention! Mine is a comment that questions the premise. There are studies reported in The Chimpanzees of Budongo Forest (page 136) that accents of the vocalisations of different chimp communities vary – with the researchers proposing that that is handy cultural learning because if the sounds are from likely threatening strangers you can then head the other way out of potential danger. Also at P135 the pant hoots are modified to resemble those of the reigning alpha male.

  • Doug Martin says:

    Dear Cristina,
    I really enjoyed your article! I was struck by how many of the themes are related to work from our lab examining how stereotypes form via cumulative cultural evolution (e.g., Martin et al., (2014). The spontaneous formation of stereotypes via cumulative cultural evolution. Psychological Science, 25, 1777-1786, doi:10.1177/0956797614541129).

    • Cristina Moya says:

      oh cool. thanks for the citation. I tried to do something like your transmission chain in the Peruvian Altiplano where I work along the ethnolinguistic boundary. I should pull that out.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.