Schaller, Park, and Kenrick
Human Evolution and Social Cognition
University of British Columbia
Justin H. Park
University of Groningen
Douglas T. Kenrick
Arizona State University
R.I.M. Dunbar and L. Barrett (Eds.),
Oxford Handbook of Evolutionary Psychology (pp. 491-504).
Oxford UK: Oxford University Press.
The massive and complex information-crunching capacities of the human brain were designed to help our ancestors make functional decisions in an environment that included other people as a prominent feature. Some of those people were relatives; some were strangers. Some were socially dominant; some were meek. Some were potential allies; others were potential enemies. Some were potential mates; others were potential competitors for those mates. Many aspects of human cognition – especially the processes that define the conceptual territory of social cognition – are adapted to the recurrent problems and opportunities posed by these other members of ancestral human populations.
So, if we are to understand social cognition fully and deeply, it is useful – perhaps even essential – to employ the following scientific strategy: First, identify the set of fitness-relevant "problems" recurrently posed by human social environments (what opportunities and dangers have other people traditionally posed?). Second, employ an evolutionary cost–benefit analysis to deduce plausible cognitive adaptations that would have helped "solve" those problems. Third, deduce the specific implications of these adaptations for human cognition in contemporary social environments. And, fourth, test those hypothesized implications rigorously with empirical data.
This evolutionarily informed approach to the study of human social cognition produces at least two substantial scientific benefits. First, this approach can yield a deeper understanding of many well-documented social cognitive phenomena – an appreciation not only for the proximate triggers of those phenomena in the contemporary workings of the human mind, but also for the ultimate causes of these phenomena within the history of the species. When considered in an evolutionary light, human social cognition is not merely one domain of inquiry within the small scientific province of social psychology; it is instead a topic of relevance to any scientist who cares about the evolution and behavioral ecology of mammalian species in general. Beyond connecting social cognition to these broader questions, the other benefit is its powerful heuristic potential. To those whose primary goal is simply to predict human social cognition and behavior, the evolutionary approach to social cognition yields novel and important discoveries about the contemporary workings of the human mind.
1. The problem set: perils and prospects of social life
So just what were the enduring social problems that imposed selection pressures on ancestral populations? A number of different social scientists have attempted to answer this question. Some answers focus on fundamental domains of sociality – whether defined in terms of elementary forms of social relationships (A Fiske 1992), algorithms of social life (Bugental 2000), or social geometries that govern interpersonal interactions (Kenrick et al. 2002, 2003). Other answers focus on fundamental human motives that are aroused by and govern behavior within different kinds of social interactions (S Fiske 2004; Kenrick et al. 1999). Across these various conceptualizations, there emerges a set of enduring problems that likely exerted substantial influence on the evolution of human populations. This set of problems can be broken down into two subsets: (a) a set of social prospects or opportunities, the successful obtainment of which would have had a positive impact on inclusive fitness; and (b) a set of social perils, the successful avoidance of which would have had a positive impact on inclusive fitness.
Table 1 lists some examples of specific prospects and perils pertaining to specific domains of social life. We will elaborate just a bit on several of them for illustrative purposes.
Consider first the positive impact of other people. Affiliating with others offers the potential for interpersonal bonds and social support, and the successful attainment and maintenance of these interpersonal relationships can have important positive consequences for fitness (Dunbar 1997; Taylor and Gonzaga in press). Social interactions also provide the necessary means for selectively distributing resources to one's offspring and other kin, and more generally provide the opportunity to help ensure the reproductive success of those kin. And, of course, it is the act of reproduction itself that is the preeminent prospect offered by social interaction. Mammals don't reproduce alone; reproductive fitness has depended crucially on successful mating. This requires that individuals not only successfully solve the problem of attracting a mate, but also the problem of selecting a mate (or mates) bearing characteristics optimal to one's own inclusive fitness. Of course, given differences in parental investment that have characterized so much of human evolutionary history, different mating tendencies have had different fitness-relevant costs and benefits for males and females. Thus, when faced with the prospect of selecting a mate who optimizes one's own inclusive fitness, one expects to witness sex differences in the behavioral strategies employed by men and women, and in the goals that they look to satisfy (Buss and Schmitt 1993; Kenrick et al. 1990).
Social interactions are not only a source of potential benefits; they are the source of many fitness-relevant perils as well. The set of perils includes threats to health and well-being (e.g., Kurzban and Leary 2001; Neuberg and Cottrell in press; Schaller et al. 2003). Such threats may result from another's intention to do harm, or they may be unintentional, such as the threat of contracting parasites or pathogens from someone who is already infected. A rather different sort of peril arises in the guise of cheating, stealing, or other forms of social contract violation, such as when another individual fails to reciprocate a resource-consuming prosocial act (Cosmides and Tooby 1992). Even if one's own health or welfare is not at stake, any such threat to one's kin would also have consequences on inclusive fitness. The set of perils is not merely limited to other individuals who engage in behavior that affects oneself (or one's kin) directly. To the extent that one's fitness outcomes are dependent on the presence of a social group and the efficient functioning of that group, then any individual who engages in behavior detrimental to the functioning of the group can also be viewed as a source of peril. In addition, given that many fitness outcomes have historically depended upon group living, a fundamental form of social peril lies in the potential to be cast out or rejected from one's social group (Baumeister and Leary 1995).
These and other social problems – prospects to be achieved and perils to be avoided – have endured for countless generations in human evolutionary history. These problems are likely to have exerted nontrivial selection pressures on the evolution of human social cognition.
2. The solution set: evolved features of social cognition
With this quick review of fitness-relevant problems in mind, we can now address the central question: Just what evolutionarily plausible cognitive adaptations might have arisen to help solve one or more of these problems? Table 2 provides an illustrative list of examples, most of which pertain to specific attentional hypersensitivities or information-processing biases. To understand the evolutionary origins of these cognitive adaptations, it is useful to first deconstruct each fitness-relevant problem into a set of smaller subproblems. After all, although each problem is defined in terms of behavioral outcomes, the solution may require a cascade of cognitive events that precede and promote specific kinds of behavior.
At the very least, there are two kinds of cognitive steps implicated in any functionally useful behavioral response to social stimuli. One must first attend to any set of social stimuli so as to identify and differentiate between individuals with different implications for one's inclusive fitness. And, after fitness-diagnostic clues have been perceived, one must have some means for efficiently facilitating a functionally beneficial behavioral response. Therefore, in discussing the evolved features of social cognition, we begin with these two subproblems – one that implicates attentional processes and the other that implicates a variety of higher-order cognitive processes – and review possible cognitive adaptations that help to solve them.
2.1. Hypervigilance and selective allocation of attentional resources
In order to solve any of the fitness-relevant problems of social life, one must identify those individuals who pose specific kinds of perils or prospects. In order to avoid contracting contagious diseases, for instance, one must identify individuals who are already infected and discriminate them from those who are not. In order to choose an optimal mate, one must identify those individuals who have desirable characteristics and discriminate them from those who do not. Successful social identification and discrimination requires the allocation of attention to features that are actually diagnostic of those specific dangers or opportunities.
Attention is a limited resource. To the extent that attention is allocated to specific kinds of features or to specific individuals in the social environment, one is less able to allocate attention to other features or individuals. It would have been adaptive for individual animals to selectively allocate attentional resources to particular pieces of information in the social environment that are especially relevant to recurrent problems of social life and that most readily compel fitness-optimizing solutions to those problems.
Plenty of evidence in the behavioral ecology literature indicates that animals selectively acquire information that is relevant to survival and reproduction (Dukas 2002). Conceptually similar findings are well documented in the literature on human perception and cognition. Most of this research focuses on visual attention. For instance, compared with other less threatening kinds of visual stimuli, people are especially quick to visually detect the presence of snakes and spiders (Öhman et al. 2001). This finding is buttressed by neural correlates of attention: Studies assessing event-related potentials (ERPs) in the human brain indicate a faster response to emotionally negative stimuli than to either emotionally positive or neutral stimuli (Carretié et al. 2004). It appears that visual attention is selectively allocated to the detection of threats in the natural environment.
Does this conclusion apply also to threats unique to the social environment? Yes. There is a burgeoning literature on the effects of human faces and facial features on visual attention. People are uniquely attentive to the features of human faces, especially those features – such as the eyes, eyebrows, and mouth – that are most strongly diagnostic of facial emotions (e.g., Lunqvist and Öhman 2005; Ristic et al. 2002). People seem to be particularly attentive to facial expressions that connote threat. Compared with other kinds of social stimuli – including more emotionally positive facial expressions – angry faces are especially quick to grab and/or hold attention (Fox et al. 2001). As with nonsocial stimuli, these effects are buttressed by ERP results indicating a more immediate neural response to angry faces (Schupp et al. 2004).
People selectively allocate attention not only to potential sources of threat, but also to potential sources of reproductive reward. In a study that assessed the temporal duration of eye-fixations on male and female faces of varying physical attractiveness, Maner et al. (2003) found that men allocated substantially more time looking at attractive (relative to unattractive) female faces. Given that facial attractiveness serves as a cue indicating fitness and fertility (Fink and Penton-Voak 2002; Thornhill and Gangestad 1999), this finding is consistent with the hypothesis that men selectively allocate attention to individuals who offer the greatest promise of reproductive reward. (Men did not show any such attentional bias toward attractive male faces – a context in which physical attractiveness would not serve as a cue to reproductive fitness. Further, women showed a qualitatively different pattern of results – consistent with logic derived from the theory of differential parental investment, indicating that physical attractiveness serves a somewhat different function in the mating strategies of men and women.)
In many circumstances, there is a positive relationship between the attention allocated to a target individual and later memory for that individual's identifying features. Consequently, allocation of attention can sometimes be indicated indirectly by memory measures. Using such an approach, various lines of research suggest that attentional resources are selectively allocated to those individuals who appear to be potential sources of reproductive prospect or peril. For instance, several studies have examined whether "cheaters" – individuals who violate social contracts – are especially memorable. It appears that they are (Mealey et al. 1996; Oda 1997; Yamagishi et al. 2003).
In sum, the human mind seems to be hypervigilant to cues connoting fitness-relevant perils and prospects. The evolutionarily enduring problems of social life have left their mark on the highly automatized processes of social attention.
2.2. Activation and manipulation of social knowledge structures
Selective attention alone is insufficient to solve the recurrent problems of social life. Animals must not only gather fitness-relevant information about the world around them; their minds must do something with that information. Therefore, animals likely evolved specific kinds of higher-order cognitive processes that provide quick, efficient means of facilitating adaptive behavioral responses whenever fitness-relevant information is detected.
Theory and research in this broad domain of inquiry can be loosely lumped into two categories: (a) research that focuses on reasoning and human decision processes, and (b) research that focuses more simply on the activation of knowledge structures into working memory.
Within the realm of reasoning, it has been argued that there evolved cognitive algorithms of information integration that facilitated accurate diagnosis of those individuals who violate social contracts (Cosmides and Tooby 1992). The plausibility of a special "cheater-detection" mode of reasoning has been the focus of an extensive line of research. Abundant evidence suggests that people show enhanced facility for a specific form of propositional reasoning under conditions in which the reasoning task is clearly relevant to social contract violations (e.g., Cosmides 1989; Fiddick et al. 2000; Sugiyama et al. 2002). Indeed, neuroscience data have indicated that a somewhat different set of brain structures is involved in fitness-relevant versus fitness-irrelevant forms of the same logical reasoning task (Adolphs 1999; Stone et al. 2002).
Other lines of research on reasoning and decision-making have focused on the evolutionary implications of other kinds of social problems, including problems related to the allocation of resources to kin versus nonkin (Burnstein et al. 1994) and problems pertaining to the navigation of social hierarchies (Cummins 1999). There are several lines of research that focus specifically on predictable biases in social judgment and social decision-making (e.g., Nesse 2005; Haselton and Nettle in press). For instance, within the realm of mating, there is a predictable bias such that men misjudge women to be more desirous of sexual relations than they actually are – a bias that can be readily predicted from an evolutionarily informed cost–benefit analysis indicating that for men (compared with women), ignoring a willing mate incurs heavier fitness costs than approaching an unwilling one (Haselton and Buss 2000). Conceptually similar analyses have been applied to many other domains of social interaction, yielding hypotheses specifying adaptive "errors" and biases across a broad range of judgment and decision-making. The empirical database supports these evolutionarily informed hypotheses (Haselton and Funder in press; Haselton and Nettle in press).
Research on reasoning yields conclusions that pertain primarily to processes through which information, already in working memory, is manipulated and integrated. But how does that information get into working memory in the first place? In some cases, the information is perceived directly and concurrently from the external environment – thanks in part to the selective allocation of attentional resources, discussed above. In addition, other potentially useful information may have been acquired previously (e.g., learned associations) and archived in long-term memory. It would have been adaptive for individuals to have selective access to whatever archived information is especially pertinent to the adaptive problems of social life – information that most readily compels fitness-optimizing solutions to those problems. Thus, there is another class of evolved cognitive algorithms. These are simple stimulus–response algorithms in which some perceived cue acts as a stimulus, automatically activating into working memory specific cognitions that dispose individuals to respond in ways that confer fitness benefits.
When one perceives a potentially threatening individual, for instance, adaptive behavior (e.g., avoidance or the adoption of a defensive posture) is facilitated by the automatic activation of cognitions characterizing the individual as a threat. This stimulus–response algorithm not only influences cognitive responses to obvious sources of social peril (e.g., individuals with angry facial expressions), it also leads to predictable biases in the stereotypes that are activated when people encounter members of racial or ethnic outgroups. Ethnic group membership represents a contemporary analog to the sorts of coalitional group memberships that have played a substantial role in social life throughout human evolutionary history (Kurzban et al. 2001). Throughout that history, coalitional ingroups were sources of support and safety, whereas encounters with outgroup members (perhaps especially unexpected encounters with outgroup males) represented potential threats to personal welfare. Consequently, perceptual encounters with unknown outgroup members may automatically activate cognitions connoting threat. This is evident not only in the semantic contents of cognitively accessible stereotypes about ethnic outgroups (Schaller et al. 2003), but also in the fearful emotional responses to these outgroups (e.g., Cottrell and Neuberg 2005). These patterns of cognitive response are bolstered by data indicating that the perception of ethnic outgroup members stimulates greater activity in brain structures associated with fear and triggers a physiological threat response (Blascovich et al. 2001; Phelps et al. 2000).
Other perceptual cues – associated with other kinds of potential peril – automatically activate qualitatively different kinds of threat-relevant cognitions into working memory. Morphological anomalies, such as physical disabilities and facial disfigurements, are likely to have historically served as heuristic cues indicating parasitic infection; consequently, humans and other primates respond aversively to individuals bearing such anomalies (Goodall 1986; Kurzban and Leary 2001; Park et al. 2003). Underlying these behavioral reactions, it appears that the very perception of these morphological anomalies automatically activates disease-relevant cognitions into working memory – even under conditions in which perceivers explicitly know that the disfigured individual poses no health risk at all (Park et al. 2003).
Cues connoting potential fitness-relevant opportunities have similar algorithmic consequences, activating specific kinds of cognitions that promote adaptive behavior. In the realm of mating, men erroneously detect exaggerated levels of sexual arousal from the objectively neutral facial expressions of physically attractive women, compared with unattractive women or with attractive men; no such effect occurs among women (Maner et al. 2005). These results suggest that when men perceptually encounter physically attractive opposite-sex others (a constellation of cues that connotes a fitness-enhancing opportunity), this activates a set of optimistic attributions about additional characteristics of those women. It is exactly this sort of attribution that increases the likelihood of actually pursuing a mating opportunity.
Finally, there is plenty of evidence indicating that the perception of kinship cues triggers highly automatized cognitive responses. Animals rely on crude heuristic cues to infer the extent to which a conspecific is genetically related; and like many other animal species, humans use cues pertaining to familiarity and phenotypic similarity (Hepper 1991; Lieberman et al. 2003; Rendall 2004). If indeed these cues serve as stimuli triggering an evolved stimulus–response mechanism, then the perception of such cues in another person – even if that other person is known to be genetically unrelated – should immediately trigger emotional and cognitive responses that (a) inhibit sexual intercourse and (b) facilitate prosocial allocation of resources. Empirical evidence is consistent with these predictions. For instance, perceived facial similarity triggers attributions indicating both greater trustworthiness and decreased sexual attractiveness (DeBruine 2002, 2005). Greater attitudinal similarity – even in a total stranger – automatically activates semantic cognitions connoting kinship and is associated with a variety of prosocial intentions (Park and Schaller 2005).
In sum, just as the enduring problems of social life have left their mark on low-level attentional processes, they have left their mark on a variety of higher-order cognitive processes as well. And in doing so, they exert a predictable influence on social cognition in contemporary environments.
2.3. Costs, benefits, and the functional flexibility of evolved social cognition
Evolved psychological mechanisms may operate automatically, but that doesn't mean that they are invariant in their operation. Quite the contrary: These mechanisms are highly flexible and predictably influenced by regulatory cues in the immediate environment. This point – easily deduced from an evolutionary cost–benefit analysis – has enormous implications for social cognitive phenomena.
Evolved psychological mechanisms are associated with specific benefits (animals that had these capacities had greater reproductive fitness than those that didn't) but their actual operation typically entails some potential costs as well. Attentional hypervigilance consumes metabolic resources (or reduces the time available for acquiring those resources), as does the activation of knowledge structures. This is the case especially when these cognitive processes are accompanied by specific affective responses as well, as they often are. Moreover, because of the finite metabolic resources available to an organism at any moment, the engagement of any one specific adaptive mechanism limits the extent to which other adaptive mechanisms might be engaged. Thus, to be more optimally adaptive, these evolved cognitive mechanisms should have evolved in such a way to be functionally flexible; they are especially likely to be engaged when additional information in the immediate environment indicates that the functional benefits are especially likely to outweigh the functional costs (but are less likely to be engaged when additional information indicates either lower benefits or higher costs). For example, the perception of a sudden loud noise automatically triggers a startle response. This acoustic startle reflex is surely adaptive in the promotion of self-protection. But it is also variable: The response is stronger under conditions – such as ambient darkness – in which people feel especially vulnerable to harm (Grillon et al. 1997).
Information signaling potential costs and benefits, and thus potentially moderating the engagement of evolved social cognition processes, may lie not only in individuals' external environments (physical and social contexts that may change from moment to moment), but also in the cognitive environments that individuals carry with them chronically (acquired attitudes, personality traits, and other dispositions). Thus, for instance, attentional hypervigilance to spiders occurs more strongly among arachnophobes, and hypervigilance to snakes occurs more strongly among individuals who are more chronically fearful of snakes (Öhman et al. 2001).
The same point applies to social cognition, too. Regardless of whether cost–benefit information is implied by chronic individual differences (e.g., individual differences in mating motives; Simpson and Gangestad 1991) or by transitory aspects of one's immediate social context, this information leads the human mind to prioritize implicitly the adaptive problems that need to be solved at any given moment. Because of this adaptive functional flexibility, evolved social cognitive phenomena vary predictably across individuals and in response to specific kinds of contextual cues.
In the realm of social attention, hypervigilance to social threat varies as a function of the perceiver's chronic feelings of vulnerability: Angry faces capture attention more quickly – and hold attention for longer periods of time – among more highly anxious individuals (Bradley et al. 2000; Fox et al. 2001). Conceptually similar effects have been found in the domain of mating. People who chronically adopt an "unrestrictive" approach to mating (and who thus chronically seek mates) are especially likely to allocate visual attention to physically attractive members of the opposite sex (Maner et al. 2003).
Evolved mechanisms of reasoning, judgment, and decision-making are also engaged flexibly in response to heuristic cost–benefit information. For instance, in altruistic judgment tasks, people predictably discriminate in favor of closer kin; but this effect itself shows up more strongly in life-and-death situations than in other situations in which the costs of not helping are less profound (Burnstein et al. 1994). Even the paradigmatic example of an evolved reasoning algorithm – the form of propositional reasoning that serves a "cheater-detection" goal – is moderated by the perceiver's social status (Cummins 1999). Compared with low-ranking individuals, high-ranking individuals (who have more resources and so are more likely to suffer costs from the presence of undetected cheaters) are more likely to demonstrate error-free propositional logic when reasoning about social contracts.
There is also abundant evidence of functional flexibility in the activation of adaptive knowledge structures into working memory. People who are chronically worried about the dangers posed by other people or have been made to feel temporarily vulnerable to harm are especially prone to the automatic activation of threat-relevant stereotypes when they perceive members of an ethnic outgroup (Schaller et al. 2003). This phenomenon not only affects stereotypic judgments about the entire outgroup, but also influences inferences about the characteristics of individuals. For instance, people who are in a temporarily fearful state are especially likely to erroneously perceive anger – but not fear or other negative emotions – in the face of an outgroup member (Maner et al. 2005). These effects occur even if the actual outgroup poses no realistic threat whatsoever, and even if the sense of vulnerability is the result of a blatantly artificial manipulation (e.g., a temporary lack of ambient light in a psychology laboratory or a few minutes viewing of a Hollywood movie). Conceptually similar evidence of functional flexibility emerges in the automatic responses to cues that connote parasite infection and transmission (Faulkner et al. 2004; Park et al. 2003). For example, Faulkner et al. (2004) hypothesized that the subjective foreignness of an ethnic group might serve as a heuristic cue connoting the threat of parasite transmission. Consistent with this hypothesis, xenophobic attitudes toward foreign peoples was especially pronounced among individuals who felt chronically vulnerable to disease and among individuals for whom the risk of parasite transmission was temporarily salient. Further support is provided by evidence that newly pregnant women (e.g., in the first 10 weeks of pregnancy, a time during which the developing fetus is especially vulnerable) show an enhanced sensitivity to disgust and stronger xenophobic attitudes toward subjectively foreign groups (Navarrete et al. 2005, unpublished manuscript, University of California at Los Angeles).
Finally, in the domain of mating, the tendency for men to optimistically perceive sexual arousal in the faces of attractive women is stronger among men for whom a mating motive has been artificially induced, and it is also stronger among men who are sexually unrestricted and thus have a chronically more active mating motive (Maner et al. 2005). Meanwhile, among women, the menstrual cycle moderates the strength of adaptive sexual responses to specific stimulus features in men. During the most fertile phase of the cycle, women are especially likely to respond positively to men who are symmetrical, masculine, and have various other characteristics indicative of high reproductive fitness (Gangestad et al. 2004; Gangestad and Thornhill 1998; Penton-Voak and Perrett 2000).
The picture that emerges from these and other results is a portrait of a human mind that evolved in response not only to the perils and prospects of the social environment, but also to the fact that specific perils and prospects matter more under some circumstances than others. Functional flexibility is an adaptation of profound importance, and implications of this adaptive functional flexibility can be found everywhere within the social cognition literature.