H. Clark Barrett
How have interactions with predators and prey shaped human evolution? That they must have done so, at some point in the past, probably seems plausible to most of us. After all, we understand that the comforts and protections of contemporary urban and suburban environments are of relatively recent historical origin. The vast majority of us are no longer predators; when we do eat meat, it is delivered to us in shrink-wrapped packages. Nor will most of us end up becoming prey; only in some parts of the world does death by animal attack pose anything but the most negligible risk. And yet we all know that for most of our evolutionary history, such comforts did not exist. Our ancestors faced the risk of predator attack since well before they were human, stretching back to our most ancient mammalian ancestors. Pursuit of prey, too, stretches back to the earliest insectivorous primates and crescendos in the big game hunting of our own hominin lineage. Few things seem more Darwinian than predator-prey interactions, so it is hard to imagine such encounters not shaping our evolution.
And yet, most of us are probably unaware of just how deeply predator-prey interactions, looped over for vast stretches of evolutionary time, might have made us who we are. Indeed, even though there is good evidence for the influence of predator-prey interactions on many aspects of our bodies and minds, we still do not (and may never) know the full scope of that influence. It's possible, for example, that some of the most fundamental features of human nature—from big brains, to sociality, to long lifespans, to our heavy reliance on social learning and cultural transmission—were selected for, in part, because of their benefits for avoiding predation, and because of the increasing reliance of ancestral hominins on a hunting way of life. If predator-prey interactions were partly responsible for setting us on the evolutionary path that has brought us to where we are today, the scope and depth of their effects on human psychology and physiology might be hard to overestimate.
In this chapter I sketch the various ways in which predators and prey have shaped, or might have shaped, us. Although I focus on psychological adaptations, it is impossible to understand these adaptations without also understanding changes in our ecologies and in our bodies, because it is through interactions with the world that our psychological mechanisms evolve. Thus, where possible, I attempt to highlight how predator-prey interactions have shaped human bodies and minds as systems of interacting parts.
Evolution is a path-dependent process. The variation within a population that natural selection operates on at any given time is the product of all of the prior events and changes that have brought the population to that point. Because there is no start date for interactions with predators in the evolutionary lineage that has led to us, the effects of predators on our evolution are ancient, and might have shaped the evolution of many traits in ways that are not at all obvious now. For example, it's possible that vision itself—what many would consider to be a paradigm case of a domain-general ability—originally got its jump-start because of predator-prey interactions, which created an arms race to both see and not be seen, and possibly drove the diversification of animal species during the Cambrian, half a billion years ago (Parker, 2003). More recently, many paleontologists believe that we mammals owe our success to the extinction of the dinosaurs at the end of the Cretaceous period around 65 mya, removing the predators and competitors that had restricted our ancestors to a tree-dwelling, largely nocturnal niche (Meredith et al., 2011).
Interactions with predators and prey create a variety of adaptive problems that can have effects on nearly all aspects of organismal function and design, from life histories, to morphology, physiology, and cognition. Unlike other aspects of the environment, which are either static or changing independently of the species that inhabit them, predators and prey coevolve with each other via evolutionary feedback akin to an arms race (Van Valen, 1973). This feedback can create complex evolutionary dynamics with cascading effects on the taxa involved.
In our lineage, stretching back to the earliest primates and beyond, both predator avoidance and prey capture are ancient adaptive problems. The earliest primates were insectivorous. Predation on insects, as well as frugivory and an arboreal lifestyle, may have shaped some basic features of our clade, including vision and motor skills (Cartmill, 1992). However, the kinds of predation that humans engage in—hunting of large prey such as mammals and birds (in addition to insects)—is of more recent origin, probably originating within the ape clade and intensifying within our own genus, Homo. Increased reliance on hunting by humans is likely to have shaped us in profound ways, favoring increased reliance on cooperation, social learning, long lifespans, big brains, and cognitive mechanisms related to hunting and foraging (Kaplan, Hill, Lancaster, & Hurtado, 2000). In addition, we may owe our evolutionarily unique abilities as long-distance runners to hunting (Bramble & Lieberman, 2004; Carrier, 1984).
Predators, in turn, have had a profound influence on the biology and behavior of primates, including us. Primatologists have long thought that predation risk is one of the primary factors (though not the only one) favoring sociality (Hart & Sussman, 2005; Isbell, 1994; Kappeler & Van Schaik, 2002; Van Schaik & Van Hooff, 1983). Again, this is likely to be an ancient selection pressure in our lineage, as a high degree of sociality characterizes the primate clade in general (Isbell, 1994). More recently, predators might have had specific effects on evolution within the hominin lineage, especially given the relatively open, predator-dense habitats hominins are likely to have occupied for much of our evolution (Brain, 1981; Hart & Sussman, 2005; Kruuk, 2002).
First, I will review general features of the selection pressures that predation and hunting impose, and how they are likely to shape suites of mechanisms in body and mind. Then I will turn to specific aspects of human psychology that may have been shaped by predator-prey interactions.
Dangerous animals have coexisted with our ancestors since long before we were human. The archaeological record has permitted reconstructions of the array of predators in ancestral environments at various points in space and time (Blumenschine, 1987; Rose & Marshall, 1996). This array included fast-moving mammalian predators such as felids (cats) and hyaenids (hyenas), and the diversity of predators in past environments was even higher than today. Human encounters with predators occurred in several contexts, including hunting of humans by predators and competitive interactions between humans and predators over kills (Brain, 1981; Brantingham, 1998; Rose & Marshall, 1996; Stanford & Bunn, 2001). In modern environments, where the ranges of humans and predators such as large cats overlap and human activities such as hunting and foraging bring them into close proximity with predators, attacks occur regularly (Kruuk, 2002; Treves & Naughton-Treves, 1999). Together, these data suggest that cognitive mechanisms involved in predator detection and evasion would have been under selection in our lineage both before and after the origin of Homo sapiens.
Predator encounters are likely to have selected for a variety of traits in our lineage, both prior to and after the split with the chimpanzee-human common ancestor. Because predation is thought to be a major factor selecting for sociality in primates, many aspects of social cognition—in particular, mechanisms sustaining relationships with nonkin—may have been initially selectively favored because of predation. Reduction in predation risk could have provided the benefits that outweighed the various costs of social life, such as competition and increased exposure to pathogens. But in order to provide these benefits, other adaptive problems would have to have been surmounted: namely, all the problems involved with tolerating the presence of others. Although problems of cooperation might seem to be a different “domain” than predator avoidance—and indeed, many cooperation mechanisms might not process information about predators per se—they could have their origins, at least in part, in predators as a source of selection.
The role of predators in shaping our evolution is likely to have changed in various ways from our earliest primate ancestors to our most recent hominin ones. Predation as a source of mortality is known to be a factor shaping life histories, selecting for fast growth and early reproduction (Reznick & Endler, 1982). It also shapes, for example, decision-making about risk (Coss, 1999; Lima, 1998; Stankowich & Blumstein, 2005). However, increasing sociality in our lineage would have relaxed these selection pressures, allowing for longer lifespans and longer time horizons for decision-making. Still, one would still expect a variety of contingent mechanisms for inference and decision-making in contexts in which predation risk is high. These would include, for example, mechanisms for detection of predators and prediction of predator behavior (Barrett, 1999; Barrett, Todd, Miller, & Blythe, 2005; Coss & Goldthwaite, 1995; Frankenhuis & Barrett, 2013; Frankenhuis, House, Barrett, & Johnson, 2013; Gao, McCarthy, & Scholl, 2010; New, Cosmides, & Tooby, 2007; Thorpe, Gegenfurtner, le Fabre-Thorpe, & Bulthoff, 2001), specialized learning and memory processes having to do with danger and survival (Barrett & Broesch, 2012; Nairne, Thompson, & Pandeirada, 2007), as well as emotional and motivational mechanisms, including anxiety and fear, that modify behavior in light of predation risk (LoBue, Rakison, & DeLoache, 2010; Öhman & Mineka, 2001; Stankowich & Blumstein, 2005). These will be discussed in more detail later.
Additionally, humans have engaged in warfare and smaller-scale intraspecific conflict for a long time. To some extent, intraspecific conflict may make use of mechanisms originally involved in predator-prey contexts; however, there may be mechanisms evolved for human-on-human aggression, and defense against it (Duntley & Buss, 2005).
In addition to the role of prey, humans can adopt the role of predator. Hunting probably predates the origin of the hominin lineage because it is also practiced by our closest evolutionary relatives, chimpanzees. Moreover, the archaeological record suggests that meat has been an important part of hominin diets for millions of years. Increased reliance on meat, a risky, high-variance food source, may have played an important role in the evolution of human sociality and social cognition (Stanford, 1999). The earliest evidence for persistent, as opposed to occasional, carnivory dates back to approximately 2 million years ago (Ferraro et al., 2013). Hunting was and is a dangerous activity, not only because prey animals can themselves be dangerous, but also because of potential competition with other carnivores. There are many sources of archaeological evidence that humans could and did kill game animals, either with tools or by other means, and that they hunted a wide variety of prey, from large, fast ungulates to small rabbits and birds, which would have required diverse strategies and intuitive understanding of prey behavior (Mithen, 1996; Potts, 1989; Stanford & Bunn, 2001).
Hunting is likely to have shaped our lineage in diverse ways. As was the case for predation, the many ways in which a reliance on hunting might have shaped our bodies and minds is not intuitively obvious. According to Kaplan and colleagues, for example, accelerating reliance on hunting in the human lineage had multiple cascading effects on our life histories, social organization, and cognition (Kaplan et al., 2000). They suggest that the shift to meat, as well as other nutritionally dense and hard-to-process food sources, created a cascading, self-reinforcing coevolutionary process. Meat as a food source allows humans to grow large brains, which, in turn, improves our ability to hunt, as well as to cooperate. Increasingly long life histories are selected for because of the benefits of skill acquisition, as well as resource transfers across generations: up to three generations, in humans, which is rare among primates. According to Kaplan et al. (2000), hunting selected for both lengthened childhoods and increased adult lifespans, sociality, reliance on socially transmitted information and skills, and, perhaps, most importantly of all, intelligence. As was the case with predation, hunting may, therefore, be responsible for many aspects of our physiology and cognition that do not appear to be immediately tied to hunting: aspects of intelligence, cooperation, social learning, and paternal investment in offspring, for example.
An examination of the skills involved in hunting in preindustrial societies points to some of the ways that hunting might have shaped our cognition. The knowledge required for successful hunting is incredibly complex and increases into late adulthood. As a skill, hunting is in many ways more akin to mathematics or engineering than, for example, running a marathon (though it is that too; Bramble & Lieberman, 2004). Among the Ache and the Hiwi, for example, Kaplan and colleagues have documented that male hunting returns do not peak until around age 40. If hunting were more about strength and stamina than knowledge, one would expect hunting returns to peak earlier. Instead, hunting is clearly a knowledge-dependent skill at which older individuals can surpass younger ones, and anthropologists have documented elaborate and detailed knowledge related to hunting in adults in traditional societies (Blurton Jones & Konner, 1976; Kruuk, 2002; Liebenberg, 1990).
The cognitive skills involved in hunting span various domains of cognition, and the benefits of hunting may have been a major factor shaping their evolution. For example, virtually all hunting involves tool use, and up until relatively recently in human evolution most human-manufactured tools were probably related to hunting and subsistence in some way. This means that domain-specific skills of tool use in the brain may ultimately have their source in selection for hunting and foraging (Johnson-Frey, 2004). Moreover, mechanisms involved in social learning and teaching, or pedagogy, may have first evolved primarily for transmission of information about tools (Csibra & Gergely, 2009). More generally, foraging is likely to have selected for a variety of mechanisms, including mechanisms of spatial cognition, mechanisms for making decisions about resource distributions in the environment, mechanisms of search, and even the brain's dopamine-modulated risk-reward system (Hills, 2006; Hills, Todd, & Goldstone, 2008; Hutchinson, Wilke, & Todd, 2008; Wilke & Barrett, 2009).
Finally, direct interactions with prey, as well as predators, may have shaped mechanisms for detection of animate agents and predictive inferences about their behavior, including mechanisms of “mindreading,” or “theory of mind.” For example, from infancy humans, like many other animals, are attuned to interactions of chasing and pursuit, which capture attention and generate strong intuitions about goals and outcomes (Csibra, Bíró, Koós, & Gergely, 2003; Frankenhuis et al., 2013; Rochat, Striano, & Morgan, 2004). Indeed, it is possible that predator-prey interactions were an important, nonsocial source of selection on mechanisms of mindreading (Van Schaik & Van Hooff, 1983). Given the ancient origins of predator-prey interactions, which in vertebrates evolutionarily predate some forms of within-species social interaction such as parental care and cooperation, predation may be one of the oldest sources of selection for mindreading abilities.
Next I turn to the variety of psychological mechanisms in humans that might have been shaped by interactions with predators and prey, including mechanisms of perception, learning, inference, motivation, and decision-making, briefly reviewing the ways in which the designs of these mechanisms might have been shaped by predators and prey.
Predators and prey are likely to have shaped a variety of perceptual mechanisms, in diverse ways. Indeed, the origin of visual perception itself may be due, in part, to the benefits of detecting danger and finding food (Parker, 2003). Other senses, such as olfaction and audition, also play a major role in finding food and escaping predators. It is likely that every sense has been shaped in some way by interactions with predators and prey.
Perhaps the most important feature of both predators and prey is that they are agents: animate objects, capable of goal-directed motion (Leslie, 1994). In turn, the defining feature of agents is action: They move, in the service of goals (prey capture, escape). Across sensory modalities, but particularly in vision and audition, the detection and interpretation of motion is deeply embedded in the design of perception.
In vision, motion detection and processing is phylogenetically widespread, and makes use of a common set of computational principles that can be found at many levels of organization, from the level of single neurons to entire networks. For example, nervous systems often detect motion based on correlations in neural activity across parts of a network in time (Borst & Egelhaaf, 1989). Given that these design features are present in the nervous systems of distantly related, nonsocial animals, such as insects, they are likely to predate the evolutionary origins of sociality. It is plausible that the proper domain of our ability to detect motion—on which nearly all social perception and cognition depends—is predator-prey interactions, and that social-action processing evolved on top of these ancient mechanisms.
Beyond just detecting motion itself, there is the use of specific motion cues to detect and categorize agents and their goals. From a computational perspective, the problems of discriminating agents from nonagents and discriminating between different kinds of agents are enormously difficult. Consider how difficult it would be to write a computer program that could reliably pick out and identify animals from the churning confusion of information that reaches our eyes and ears and that could do so across the range of environments and conditions that humans encounter. Add to this a premium on speed and the possibility of extremely impoverished information (a brief movement in peripheral vision, ripples in the grass, something glimpsed through a gap in the leaves), and you have a task that is both extremely difficult and of utmost importance to survival. Yet there is evidence that selection has built just these kinds of things.
Motion can be used both to identify things that behave (agents) and to discriminate types of agent on the basis of how they behave. There is a substantial literature on the use of motion cues both to detect agents and to make inferences about behavior (Johnson, 2000; Rakison & Poulin-Dubois, 2001; Scholl & Tremoulet, 2000). Cues such as autonomous motion and change in trajectory are important cues to agency, and even very simple cues such as the direction of motion of a bilaterally symmetric object can be used to categorize it as an agent, with a front end (Hernik, Fearon, & Csibra, 2014, Tremoulet & Feldman, 2000). Beyond simple self-propulsion, there are a variety of cues that can be used to detect motion that is goal-directed, such as pursuit. Gergely, Csibra, and colleagues have demonstrated that motion that appears to be goal directed—for example, one object chasing another—triggers the intentional stance and specific expectations about how the objects will behave in infants as young as 9 months (Csibra, Gergely, Bíró, Koós, & Brockbank, 1999; Csibra et al., 2003; Gergely, Nádasdy, Csibra, & Bíró, 1995; Rochat et al., 2004). The motion signature of pursuit and evasion cannot only trigger the agency system but also be used to discriminate predation from other kinds of behavioral interaction and to activate inference systems and procedures specific to predators and prey. Several studies have shown that people are good at discriminating pursuit and evasion from other types of motion (Abell, Happé, & Frith, 2000; Barrett et al., 2005; Castelli, Happé, Frith, & Frith, 2000). This is true not just of dyadic interactions but interactions among multiple agents. For example, Gao et al. (2010) have demonstrated a “wolfpack effect” in perception, in which the simultaneous orientation of multiple agents toward oneself is immediately noticeable and compels escape responses.
Because of the cost–benefit asymmetries entailed by detection of predators and prey, error management logic applies (Haselton & Buss, 2000). Detection thresholds may be biased in favor of false positives, because failures to detect predators and prey may be more costly than false alarms—tempered, on the other hand, by the costs of excessive vigilance. One such bias appears to occur in responses to rapidly approaching objects. In the phenomenon known as visual looming, a rapidly expanding circular shadow (but not a rapidly shrinking shadow) was found to trigger defensive behaviors in rhesus monkeys, from ducking and flinching to alarm calling (Schiff, Caviness & Gibson, 1962). These reactions have been found in a variety of species, from fishes and frogs to human infants, and specialized neural circuits have been found that compute estimated time to contact for looming visual objects (Sun & Frost, 1998). In hearing there is a similar phenomenon of auditory looming, in which approaching sounds are perceived as starting and stopping closer than receding sounds the same distance away, possibly acting as an early-warning bias (Neuhoff, 2001).
In addition to whole body motion, the visual system is sensitive to the ways that body parts of animals move and can use such cues to discriminate types of animal, types of behavior, and other qualities such as size and formidability. A variety of studies have shown that people can distinguish animals from nonanimals and can even discriminate between kinds of animals (human, dog, horse, etc.) from point light displays in which illuminated points are placed on limbs or joints and the rest of the body is blacked out; this effect disappears when the displays are inverted (Johansson, 1973; Mather & West, 1993). Such “biological motion” perception is present from early infancy, and babies prefer right-side-up point-light displays (Simion, Regolin, & Bulf, 2008). Interestingly, the perception of biological motion is also present in chickens, suggesting that it may predate the origins of primate sociality (Vallortigara, Regolin, & Marconato, 2005).
In addition to motion cues, it is sometimes possible to detect predators through the use of static cues. Classic examples are the sinusoidal shape of snakes and the characteristic appearance of eyes. A variety of studies, for example, have shown that snakes “pop out” of perceptual arrays. Öhman, Flykt, and Esteves (2001) found that subjects could rapidly pick out pictures of snakes and spiders from arrays of fear-irrelevant objects (flowers and mushrooms) much faster than they could do the opposite task, suggesting that snake detection is a parallel process using a specialized detector, different from serial search for flowers. Similarly, the perceptual importance of eyes as cues to being seen is phylogenetically widespread, as evidenced by the commonness of eyespots and behavioral reactions to eyes.
Broken wing displays in plovers—a clear anti-predator response—are triggered by eyes (Ristau, 1991). Eye stimuli exacerbate the tonic immobility response, a last-ditch emergency response to capture by a predator, in restrained chickens (Gallup, 1998). Humans are known to be exquisitely sensitive to gaze direction (Baron-Cohen, 1995), and, although gaze as a triggering stimulus for antipredator mechanisms has not been specifically examined in humans, it is known to cause arousal in humans and perhaps the activation of antipredator responses (Coss & Goldthwaite, 1995).
Finally, perceptual detection experiments suggest that there may be additional cues that the visual system uses to detect and track predators, prey, and other animals. Thorpe et al. (2001) found that subjects can detect animals better than other classes of stimuli in far peripheral vision, where visual acuity is generally poor. This suggests another early-warning system for animals in the periphery. Using a change blindness paradigm, New, Cosmides, and Tooby (2007) found that subjects' abilities to detect changes in a scene that involved animals were significantly greater than their abilities to detect changes that involved nonliving objects, even when the object was much larger than an animal, such as a building. This provides evidence for an animate monitoring hypothesis: Mechanisms exist that are dedicated to tracking and updating locations of animates, but not inanimates, in visual scenes, because only animates are likely to move in the real world.
Just as the evolutionary roots of predator avoidance are ancient, so are the evolutionary roots of foraging. Not all foraging is for animate prey, but, in primates, at least some of it is. Although hunting as humans know it—hunting for comparatively large game with human-made tools—is likely to be relatively recent in origin, prey search and capture is ancient, dating back at least to the earliest insectivorous primates, and beyond.
In keeping with this, some of the effects of foraging on the design of our nervous systems are likely to be ancient. The “reward” system of our brain, the dopaminergic system, bears evidence of being designed for what is sometimes called area-restricted search, or search based on the clumpiness or patchiness of resources in the environment—causing us to perseverate when something good has been detected, and to become distracted when the rate of reward drops (Hills, 2006). Indeed, some pathologies, such as excessive perseveration on goals, may result from disruptions to this system.
Prey animals and other foraged resources are usually not evenly dispersed in the environment; they come in clumps. Our psychology of search and reward appears to be deeply organized around an expectation of patchiness (Hills, 2006; Hills et al., 2008; Hutchinson et al., 2008; Scheibehenne, Wilke, & Todd, 2011; Wilke & Barrett, 2009). This expectation extends across diverse domains, from foraging for food, to foraging for information on the Internet, to internal search in memory (Hills et al., 2008). Wilke and Barrett (2009) have suggested that the phenomenon known as “hot hand,” in which subjects overperceive clumps in distributions that are in fact random, reflects an adaptation to the clumpiness of ancestral resources. In a recent study, they found evidence that habitual gamblers are more likely to fall prey to the hot-hand illusion, suggesting that the evolved design of our foraging system may have significant real-world consequences (Wilke, Scheibehenne, Gaissmaier, McCanney, & Barrett, 2014).
The importance of the dopaminergic system in foraging reveals the deep and inseparable evolutionary link between cognition and motivation. Although these are sometimes treated as distinct in the psychological literature, the function of cognition is (ultimately) action, and action does not occur in the absence of motivation (Tooby, Cosmides, & Barrett, 2005). Motivation plays a role in all stages of cognition, shaping what we attend to, how we respond to it, and what we learn from it. In the case of foraging, motivations such as hunger and the satisfaction of getting a “hit” in prey search play a large role, and these subjective sensations shape our learning and future behavior in terms of where, how, and when we decide to look for things. In the case of predator avoidance, fear, anxiety, and risk assessment play a similarly large role.
Although fear and anxiety do not have to do exclusively with predation—there are, for example, social fears and anxieties—it may be, as was the case for perception, that predation is the most phylogenetically ancient selection pressure shaping the fear system (Cosmides & Tooby, 2000; LeDoux, 1996; Öhman & Mineka, 2001). Fear not only organizes escape and avoidance responses to dangers but also deactivates certain cognitive processes (e.g., mate search) and activates others (e.g., predator-prey routines) and may alter sensitivity thresholds of many systems. As an adaptive problem, predator avoidance shares some features with other danger-avoidance problems such as avoiding cliffs or sharp objects but also has some unique characteristics such as the fact that predators are mobile agents that seek to cause harm, unlike other sources of harm such as rotting food, which engenders its own danger-avoidance emotion, disgust.
The psychological literature on fear and fear learning is enormous, and there is not room to review it all here (for reviews, see LoBue, Rakison, et al., 2010; Öhman & Mineka, 2001). In this literature, there are debates about how “specific” fear learning is—that is, how tuned it is toward particular targets, such as snakes, predators, dangerous conspecifics, and artifacts—and about the nature of the learning mechanisms that lead to fear acquisition. Much research has examined the role of fear in conditioning, showing that fears of some types of objects and situations are easier to develop than others, and harder to extinguish. There are many demonstrations of such “content effects” or “biases” in fear learning.
For example, a variety of studies have compared conditioning stimuli such as snakes and spiders with fear-irrelevant controls such as houses, flowers, and mushrooms to demonstrate that conditioned associations between picture items and aversive conditioning stimuli, such as shocks, occur much more readily, and disappear more slowly, for dangerous than for nondangerous items, with fear generally being acquired more readily for ancestrally dangerous stimuli such as snakes (Hugdahl & Öhman, 1977; Mineka, Davidson, Cook, & Keir, 1984; Öhman & Mineka, 2001). In addition, studies with young children have provided evidence for a bias in attending to dangerous animals, such as snakes and spiders, and a tendency to associate these stimuli with fear (DeLoache & LoBue, 2009; LoBue & DeLoache, 2008, 2010; Rakison, 2009; Rakison & Derringer, 2008). Interestingly, Rakison (2009) found a sex difference in the association of snakes and spiders with fearful faces, with this association being found in 11-month-old girls, but not boys. And, a variety of studies have shown that preferential attention to, and learning about, danger stimuli need not be restricted to dangerous animals—they can include, for example, angry faces—nor to ancestral stimuli, as they can include stimuli such as syringes and guns (Blanchette, 2006; LoBue, 2010). This suggests a role for experience of current dangers in calibrating the fear learning system.
There are a variety of theories regarding the mechanisms underlying fear learning, varying in the type and degree of specificity they attribute to these mechanisms, and where they act in the flow of information processing in the brain. Öhman and Mineka (2001) offer a modular account of the fear learning system, proposing that the system possesses four distinguishing characteristics: stimulus-specificity; preferential activation by evolutionarily prepared danger stimuli, such as snakes, spiders, and falling objects; automatic triggering; impenetrability to conscious control; and dedicated neural circuitry, particularly in the amygdala. They suggest that in the most evolutionarily ancient fear systems, perceptual threat detectors were directly connected to motor reflexes designed to move the organism away from danger and that intervening control systems, though, in humans, higher-level processes can also mediate predator evasion strategies.
LoBue, Rakison, and DeLoache (2010) suggest a more minimal design that is rooted in perceptual biases to attend to dangerous stimuli, such as snakes, spiders, and human threats. On this view, the content effects seen in learning emerge from perceptual biases interacting with more general-purpose learning mechanisms, such as mechanisms of association and statistical learning.
For some kinds of dangerous animals, such as snakes, there appear to exist evolved perceptual templates or “prepared” cues, such as sinusoidal shapes or rapid looming, that allow response to the threat in the absence of learning. However, learning is clearly important in shaping responses to predators and prey, and one should expect it to be so, because learning is a useful adaptive tool.
There is some reason to expect that the most general-purpose forms of learning, such as classical conditioning, might not be ideally suited to learning about predators. For one, an equipotential learning system that had no evolved priors about the informativeness of predator-related cues might not learn as efficiently as a “prepared” learning system (Öhman & Mineka, 2001; Seligman, 1971). There is an even larger problem in learning about danger exclusively from one's own experience; some of the most informative learning opportunities can lead to injury or death.
For this reason, Barrett and Broesch (2012; Barrett, 2005) proposed that humans might possess a prepared social-learning system for learning about dangerous animals. Mineka and colleagues (1984) showed that juvenile rhesus macaques, reared in the lab with no prior experience with snakes, could acquire snake fear in a single trial if shown the face of an adult conspecific exhibiting fear toward a snake. The logic of prepared social learning makes sense here, because the costs of individual learning (e.g., a snake bite) might greatly outweigh the costs of learning from a knowledgeable conspecific (e.g., a possible false positive), and the benefits of social learning could be large. Thus, Barrett and Broesch proposed that humans might possess a similar mechanism, perhaps homologous to that of macaques, but potentially modified in the human case to admit not just facial expressions of fear but verbal statements of danger. They hypothesized a danger learning system with several features: a domain-specific preference for learning about danger as opposed to other types of information about animals; single-trial learning without feedback; retention of danger information in long-term memory; and similar memory effects across cultures. In an experimental memory task with American and Shuar children, they found evidence for all of these hypotheses (Barrett & Broesch, 2012). Follow-up work has found similar learning biases across the life course in Fiji (Broesch, Henrich, & Barrett, 2014).
Nairne and colleagues have demonstrated an additional facet of adaptive memory related in part to interactions with predators and prey: “survival processing” (Nairne et al., 2007; Nairne, Pandeirada, Gregory, & Van Arsdall, 2009). The survival-processing hypothesis posits that it is not just information “content” that determines whether and how it is stored, but also how the information is processed, that is, whether subjects process the survival value of the information as it is being encoded. They showed that merely asking subjects to rate the survival relevance of items (e.g., for securing food or water or avoiding predators) increases recall and recognition of these items in later surprise tests, compared to conditions in which the items are rated along some other dimension (e.g., pleasantness).
In addition, there are likely to be many other learning phenomena related to predator-prey interactions, such as disgust learning, triggered strongly by animal products as sources of disease transmission (Fessler, 2002), and learning about tool use, which may be rooted most anciently in learning to make and use tools for hunting (Csibra & Gergely, 2009).
Perhaps the most important way in which predators and prey differ from other obstacles or problems in the environment is that predators and prey are intentional agents: They are animate, sentient beings that process information and behave in the service of specific goals that they are well-adapted to achieve and that are in direct opposition to those of humans, either as prey or as hunters. This means that predators and prey are not passive, static components of the environment that simply need to be avoided or found. The biggest problem with predators is that, unlike other dangers such as cliffs or toxins, predators come to find you and are well designed to do so. The biggest problem with prey is that, unlike tubers or berries, they move, have the goal of avoiding capture, and possess adaptations such as camouflage and finely tuned sensory systems that help them achieve that goal.
These considerations suggest that predator avoidance and prey capture are likely to make use of mechanisms involved in understanding agency, from mechanisms for detecting the presence of agents in the environment to theory of mind mechanisms for reasoning about mental states (Baron-Cohen, 1995; Leslie, 1994). Because humans are intentional agents, too, many of the mechanisms that are brought to bear in social interaction—gaze direction detection mechanisms, for example—will also be brought to bear in predator-prey encounters. However, there are important elements of predator-prey interactions that have no analogy in human social interactions, because the goals of predators and prey are distinctly asocial. Barrett (1999, 2005) proposed that humans may possess a reliably developing “predator-prey schema,” a set of rules for predicting predator and prey behavior, embedded within (interacting with) the mindreading system (Barrett, 1999, 2005). One might think of this as an “island of competence” within the larger domain of mindreading (Frankenhuis & Barrett, 2013). What might appear to be a single, flat, undifferentiated domain such as mindreading might actually contain internal structure: Some types of interactions might be more easily conceptualized than others, and possibly earlier-developing, because they map onto ancestral forms of interaction whose understanding might yield survival benefits early in life. Predator-prey interactions—in addition to several others, such as kinship and dominance interactions—might be some of the most ancient, recurring, and fitness-relevant forms of interaction, and thereby might represent particularly important islands of competence within the domain of mindreading.
Consistent with this idea, there is evidence that young infants are particularly attentive to, and able to make predictive inferences about, interactions of pursuit and evasion, or chasing. Attention to and inferences about chasing develop similarly across cultures (Barrett et al., 2005), and early in infancy (Csibra et al., 2003; Frankenhuis et al., 2013; Rochat, Morgan, & Carpenter, 1997). Csibra et al. (2003) used a dishabituation paradigm to test 12-month-old infants' expectations regarding pursuit-evasion scenarios, presented using moving objects on a computer screen. These studies showed that infants not only encode the goals of a chaser and chasee (capture and escape, respectively), they form expectations about how each agent will most effectively pursue its goal: For example, a prey animal might attempt to escape through a hole that a predator is too large to fit through, and the predator might anticipate this escape route by going around an obstacle with a small hole in it to catch the prey on the other side. This work suggests that at least part of a predator-prey inference system is present by 12 months of age.
Several other studies support the idea of an evolved predator-prey schema. For example, Rochat et al. (1997) showed that 3- to 6-month-old infants, as well as adults, preferred to look at displays of contingent chasing, rather than displays of noncontingent motion with similar properties displayed side by side. Frankenhuis et al. (2013) later replicated this, and additionally decomposed the displays into their distinct cue components, showing that acceleration particularly drew infants' attention. In a follow-up to their original study, Rochat et al. (2004) showed that 8- to 10-month-old infants assigned distinct roles to chaser and chasee, as evidenced by surprise when these roles were reversed—suggesting distinct conceptual placeholders for the different roles within an early-developing predator-prey schema. At older ages (3–5 years), Barrett (1999) showed that both Shuar and German children are capable of producing realistic predictions about what happens in encounters between predators and prey, suggesting a schema organized around pursuit and killing that is not contaminated with information from friendly cartoon depictions of lions and other predators. A cross-cultural study with German and Shuar adults by Barrett et al. (2005) demonstrated that the chasing schema, as well as schemas for courtship, play, and several others, manifest reliably across cultures, leading to similar levels of discrimination on a perceptual categorization task.
Interestingly, it appears that these basic interaction schemas may involve attribution of distinct goals—as suggested by the role-reversal findings of Rochat and colleagues—but do not require attribution of beliefs. Castelli, Frith, Happé, and Frith (2002) found that autistic subjects were able to identify goal-directed sequences including pursuit and evasion but not sequences that required attribution of belief. Thus, the predator-prey or chasing schema may be an early-developing island of competence that involves goal attribution, but does not require attribution of knowledge and belief states, an ability that seems to be impaired, along with some other abilities, in autism (Baron-Cohen, 1995). However, this does not mean that calculation of beliefs and knowledge is never relevant for predator-prey interactions. It just means, minimally, that some action predictions can be done without the belief attribution system. In a series of studies, Keenan, Ellis, and colleagues (Ellis et al., 2014; Keenan & Ellis, 2003) have shown that predator-prey scenarios can influence children's judgments in a modified false belief task: When the correct answer to the task involves sending a prey animal to its death at the hands of a hidden predator, they are more likely to provide the incorrect answer.
One important consequence of predator-prey interactions is death. As prey, we face possible death from predators. As predators, we kill prey animals. Although the conventional wisdom in the developmental literature has been that children's understanding of death is poor, Barrett and Behne (2005) proposed that children might possess another early-developing island of competence within the larger domain of death understanding: in particular, understanding death as the cessation of agency. Children face an inferential problem when an animal dies: Unless they specifically remove the agency tag from the object—a now-dead piece of meat—they will continue to generate inferences such, as, for example, that the animal will react if touched. Barrett and Behne conjectured that the costs of this mistake might select for a mechanism that removes the agency tag from agents given certain cues of death, allowing them to cease monitoring it for change, and to be unafraid to approach or eat it. In a cross-cultural study with 3- to 5-year-old children, they found that by age 4 children were well above chance in disattributing agency properties to dead animals, compared to sleeping animals.
Given the importance of predators and prey in human evolution, it is likely that we have only begun to uncover the full array of predator-prey adaptations that the mind contains. Indeed, if one expands one's view to include phylogenetically ancient and widespread ways in which predator-prey interactions shape biology, there may be few aspects of our bodies and minds that have not been influenced in some way by the need to avoid predators and obtain food. This presents a challenge for a view of human nature as restricted to only those derived features that have evolved uniquely in us since our divergence from the other great apes. It also challenges a view of “domains” as cleanly separable, at least as sources of selection, because—for reasons outlined earlier—many aspects of our social cognition may initially have been selected for due to the benefits of cooperation for hunting and predator avoidance. This does not contradict the view that the mind is composed of many functionally specialized mechanisms; instead, it is consistent with a hierarchical specialization view, in which the mind's domains and mechanisms overlap on some levels of organization and design, and diverge on others (Barrett, 2012).
Until very recently, attack by predators was a real and constant possibility in everyday life. Selection to be aware of these creatures, of their thoughts, plans, and intentions, as well as a strategic intelligence to take advantage of this awareness, would have been strong. Here, we need to think in science-fiction terms. Imagine the human mind as an exquisitely designed computer, armed with state-of-the-art sensors, trackers, detectors, and inference engines all engineered for the purpose of predator defense and evasion. What would these look like? Without doubt, the best equipment designed by military science does not even come close. Yet, across psychology and neuroscience in general, relatively little attention has been paid to predator detection and evasion as adaptive problems that could shed light on the design of our minds.
On the other side of the coin, humans are predators by nature. We have been hunters of other animals for millions of years. Far from diminishing with time, selection for the skills necessary to stalk and kill animals has accelerated over the course of human evolution, as hunting has played an ever-increasing role in human subsistence. For those who have never hunted, the difficulty of the task is easy to underestimate. Dawkins (1976) coined the term the “life/dinner principle” to refer to the asymmetry in fitness payoffs to predators and prey for the two possible outcomes of a predation event: If the predation event is a success, the predator wins dinner, but the prey loses his life; vice versa, if it fails. There is another asymmetry, which might be called the “anywhere but here” principle: For a predator to succeed, the predator must manage to be in exactly the same place as the prey at exactly the same time; for the prey to succeed, it need only be anywhere else. Obviously, it is much easier to satisfy the latter condition than the former. This means that whereas prey can use a variety of “dumb” tactics to avoid predation, including hiding, crypsis, and living in holes or trees, predators must be designed to bring about a very unlikely and nonrandom physical state of the world, which prey are expressly designed to avoid. For tool-using predators, there is an added complication: We must either cause our own position to converge with that of the prey or cause the position of a projectile or trap to do so. This poses other adaptive problems such as the perceptual and motor problems involved in successfully aiming a projectile. Predation, then, may select for particular kinds of intelligence, and our evolutionary legacy as hunters is likely to have played an important role in the evolution of the human mind. Some aspects of our intelligence that we do not attribute to our history as hunters—from mindreading, to tool use, to strategic coordination—may nevertheless exist at least partly because of it. Additionally, our minds are likely to be full of many detection, tracking, and behavior anticipation mechanisms of which we might not be fully aware.
It is possible that investigating evolutionarily relevant problem domains such as predation, which are rarely considered by most contemporary cognitive and developmental psychologists, could lead to drastic reconsideration of how the domains of thought are organized. Rather than thinking of broad domains such as social cognition and theory of mind, we might realize that the mind is not organized around a few large problems but around many small ones such as agency detection, tracking objects, and inferring intention from motion, which do not map neatly onto the intuitive categories of contemporary psychology.
This “micro-modularity” view of mechanisms organized around rather specific adaptive tasks is, in many ways, more consistent with recent findings in cognitive neuroscience than with a view of domains as analogous to university departments, e.g., psychology, mathematics, and physics (see Boyer & Barrett, Chapter 5, this volume). Indeed, brain mapping studies are increasingly supporting a view of cognition that is both heavily “distributed” across many brain systems, and that involves the large-scale coordination of many smaller subsystems (Bullmore & Sporns, 2009). This is also consistent with a hierarchical modularity view, namely, that large-scale abilities such as “theory of mind” or “social cognition” involve the operation of mechanisms nested within larger assemblies, potentially in a flexible mix-and-match way (Barrett, 2012). Although evolutionary psychologists have argued for some time that the true domain map of the human mind is not likely to correspond to the way domains are carved in psychology textbooks, much work remains to be done in finding the mind's true joints. Adaptations to predators and prey provide a useful case study in how this might be done. There is probably not a cleanly delineated domain of predators and prey in the mind, but rather, a constellation of systems each shaped in unique ways by predators and prey—some exclusively so, and some not. For other potential domains of cognition, then—contagion, cooperation, sex—we might consider a broader set of possible models of how the underlying mechanisms are organized, including hierarchical, distributed models that involve the interaction of diverse mechanisms, some exclusive to the domain and some not.