WAyysSjn

1Survival of the Friendliest:Homo sapiensEvolved viaSelection for Prosociality
2
3AbstractThe challenge of studying human cognitive evolution is identifying uniquefeatures of our intelligence while explaining the processes by which theyarose. Comparisons with nonhuman apes point to our early-emergingcooperative-communicative abilities as crucial to the evolution of allforms of human cultural cognition, including language. The human self-domestication hypothesis proposes that these early-emerging social skillsevolved when natural selection favored increased in-group prosociality overaggression in late human evolution. As a by-product of this selection, hu-mans are predicted to show traits of the domestication syndrome observed inother domestic animals. In reviewing comparative, developmental, neurobi-ological, and paleoanthropological research, compelling evidence emergesfor the predicted relationship between unique human mentalizing abilities,tolerance, and the domestication syndrome in humans. This synthesis in-cludes a review of the first a priori test of the self-domestication hypothesisas well as predictions for future tests.
4
5INTRODUCTION
6
7Darwin viewed the evolution of human intelligence and morality as the greatest challenge forhis theory of evolution through natural selection (Darwin 1871). This remains the case today(MacLean et al. 2012). Our language and cultural cognition allows for cooperation and technologymore complex than anything seen in nonhumans, yet our neurobiology, psychology, and genomeare remarkably similar to other apes (Somel et al. 2013). A complete theory of human cognitiveevolution needs to explain how these shared traits evolved into new forms of human cognition. Tomeet Darwin’s challenge, we must identify derived cognitive features that evolved in our lineageand support our unique phenotype. Then we must identify the process by which these traits arose(Hare 2007, 2011). Adding to this challenge are discoveries suggesting that at least 10 differentspecies evolved within the genusHomo. Modern theories of human cognitive evolution must nowcontend with growing evidence thatHomo sapiensis just one among many human species thatevolved. It is no longer enough to point out what makes us human. We must also determine whatallowed our species to outlast as many as five other large-brained human species that shared theplanet with us, some perhaps until as recently as 27 kya (Wood & Boyle 2016).This article reviews the latest research suggesting that early-emerging cooperative-communicative skills are responsible for unique features of human cognition and that ourpsychology evolved in large part due to selection for prosociality (i.e., positive but potentiallyselfishly motivated acts as opposed to antisocial interactions; Eisenberg et al. 1983). Comparisonsof mentalizing skills between apes reveal that among apes, only human infants develop cooperative-communicative skills that facilitate human forms of cultural cognition; however, domestic dogs possess some social skills that resemble those seen in human infants. Research with experimentallydomesticated foxes and bonobos shows how selection for prosociality can lead to increases in thecooperative-communicative flexibility observed in dogs and infants. This comparative develop-mental work provides the basis for the self-domestication hypothesis, which proposes that uniquehuman psychology evolved as part of a larger domestication syndrome that converges with otherdomesticated animals.The human self-domestication hypothesis (HSD) draws on comparative, developmental, fossil,and neurobiological evidence to show that late human evolution was dominated by selection forintragroup prosociality over aggression. As a result, modern humans possess traits consistentwith the syndrome associated with domestication in other animals (Table 1). The HSD suggeststhis selective pressure also led to enhanced cooperation in intergroup conflicts. The hypothesisproposes that the reduced emotional reactivity that results from self-domestication and increasedself-control created a unique form of human tolerance allowing the expression of more flexiblesocial skills only observed in modern humans. Expanded developmental windows like those seen indomesticated animals allow this unique form of human tolerance and social cognition to developand leftH. sapiensas the last human standing (Figure 1).
8
9PANVERSUSHOMOSOCIAL UNDERSTANDING
10
11Our ability to mentalize, or attribute mental states to others, is the foundation of cultural cognition(Herrmann et al. 2007). Humans develop the ability to recognize what others perceive, feel, intend,and know. Children begin in infancy to recognize that others can have mental states and even falsebeliefs that can differ from their own or reality. As adults, we rely on our ability to infer others’unobservable thoughts based on observable social cues and context. Our ability to reason about theminds of others allows for everything from imitation to deception, group coordination, teaching,and language acquisition (Seyfarth & Cheney 2014, Tomasello 2009b). Given the central role ofthese skills in human social cognition, tremendous energy has been focused on testing whetherthey are shared with other animals or are a derived feature of human cognition (Hare 2011).
12
13Apes Take Others’ Perspectives and Cooperate Flexibly
14
15Beginning in the 1970s, chimpanzees (Pan troglodytes) quickly became central to studies of mentalattribution because they provided a powerful phylogenetic test. The skills we share with chim-panzees were probably present in our last common ancestor, but the skills found only in humanshelp explain our explosive evolutionary success.Initial research on chimpanzee theory of mind was as methodologically flawed as it was pio-neering. It was only after researchers began taking an ecological approach to cognition that majoradvances were made in our understanding of chimpanzee mentalizing (Hare 2001, Whiten 2013).Chimpanzees failed to solve visual perspective-taking problems by cooperating and communi-cating with humans. However, they solved similar problems when placed in competition againsteach other. A series of competitive experiments show that chimpanzees are skilled at taking theperspective of others based on their current and previous perceptual experience (Hare 2011).This led to more powerful methods using larger samples that also showed these skills in non-competitive contexts (MacLean & Hare 2012). These spontaneous measures of social cognitionruled out the potential for simple behavior reading or learning within the experiment (Hare 2011;although see Povinelli & Vonk 2004).Several studies have suggested that chimpanzees understand what others know, but there re-mains no compelling evidence for explicit false belief understanding in any animal, including great apes (Hare 2011, Martin & Santos 2016). Further studies have shown that, in many contexts, chim-panzees behave similarly to human children when they attribute intentions to others. For example,chimpanzees are more patient with an experimenter who is unable to share food than with onewho is clearly unwilling (Call et al. 2004).These social cognitive skills are also applied flexibly to a number of cooperative situations.Chimpanzees can solve novel instrumental problems that require cooperation and helping (Hare& Tan 2012, Melis et al. 2010). They know when they need to recruit help, which potential helperis most skilled, and what leverage they have to negotiate between equal and unequal payoffs (Meliset al. 2006a, 2008, 2009). Chimpanzees are also capable of taking different roles in collaborativetasks (Melis & Tomasello 2013).
16Taken together, chimpanzees are flexible in assessing what others perceive and intend. Al-though chimpanzees are skilled cooperators and take other’s perspectives in noncompetitive con-texts, it was initially easier to demonstrate many of their theory of mind skills in competitive tasks(Hare 2011).
17
18PS68CH07-Hare    ARI    17 November 2016    12:56Table 1   Evidence for domestication syndrome in modern humansDog versus wolfaExperimental foxversus control foxaBonobo versuschimpanzeeaHolocene humans versusLower Paleolithic humansAggressionLower-intensityinter- andintragroupaggression in feraldogsExperimental butnot control foxesare nonaggressivetoward humansLower intensity inter- andintragroup aggression inbonobosIntragroup tolerance allows andis favored due to demographicpressure in Holocene humans(Cieri et al. 2014, Henrich 2015)PhysiologyDogs show a mutedstress responsewhen interactingwith humansHigher basalserotonin and lowercorticosteroids inexperimental foxesBonobos have more apassive coping responseto social stressHolocene humans exhibitmorphologically inferredreductions in neonatalandrogens and pubertaltestosterone levels and increasedbrain serotonin and oxytocinavailability (Cieri et al. 2014,Nelson et al. 2011)MorphologyDogs show reducedcranial capacity anddepigmentation ofthe coatExperimental foxesshow a feminizedskull anddepigmentation ofthe coatBonobos show reducedcranial capacity,feminized faces, anddepigmentation of lipsand tail tuftsHolocene humans exhibit amodest reduction in cranialcapacity, feminized faces,globular cranial development,and depigmentation of the sclera(Cieri et al. 2014, Hublin et al.2015, Tomasello et al. 2007)ProsocialbehaviorDogs are moreattracted to humansthan to conspecificsExperimental foxesare more attractedto and interested inplaying andinteracting withhumans as adultsBonobos exhibit more playand sociosexual behavioras adults, voluntarilyshare food, and are morefood tolerantHolocene humans exhibitextreme levels of intragroupfood sharing, helping, and socialbonding (Kramer 2014,Warneken 2015)ExpandeddevelopmentalwindowPeriod ofsocialization withhumans beginsearlier and lastslonger in dogs; dogsretain the juvenilevocal repertoireinto adulthoodPeriod ofsocialization withhumans beginsearlier and lastslonger inexperimental foxes;experimental foxesretain the juvenilevocal repertoireinto adulthoodNonreproductivesociosexual behaviorsthat create toleranceemerge early and lastthroughout adulthood inbonobos; cognitive skillsrelated to spatial memoryand social inhibitionexhibit delayeddevelopment in bonobosHolocene humans exhibitearly-emerging social cognitionand graded brain developmentwith extreme delays in synapticpruning of cortical regions(Casey & Caudle 2013, Somelet al. 2009, Wobber et al. 2014)Social cognitionDogs are moresensitive to humansocial cuesExperimental foxesare more sensitiveto human socialcuesBonobos attend to eyesand exploit a human’sgaze more and showincreased cooperativeflexibilityHolocene humans exhibitincreases in cooperativecommunication, culturalratcheting, and coordinateddefense against out-groups, aswell as expanded social networks(Cieri et al. 2014, Hare 2011)aNonhuman comparisons are based on Hare et al. (2012
19
20Figure 1Homo sapiensevolved, in part, as a result of selection for increased in-group prosociality during thePaleolithic, leading to a variety of morphological, physiological, and cognitive changes also observed indomestic animals such asCanis familiaris.
21
22Only Human Apes Cooperatively Communicate
23
24Despite the social flexibility of chimpanzees, there are meaningful gaps in their understandingof and tolerance for others. Several studies show that, although chimpanzees can cooperate orcommunicate, they struggle to do both (Bullinger et al. 2014, Herrmann & Tomasello 2006,Melis et al. 2009). Central to this phenomenon is their inability to spontaneously and flexiblyuse gestures to find hidden food (Hare 2011). Human infants begin using other’s gestures andproducing them in their first year of life. Infants become part of the cultural world of adults bydeveloping an understanding of the intention behind novel and arbitrary gestures. In contrast,while nonhuman apes can slowly learn to use cooperative-communicative gestures, it is extremelydifficult for them to generalize what they learn to a new or arbitrary gestural signal (Call et al. 1998,Hare & Tomasello 2005b, Herrmann et al. 2007, MacLean & Hare 2015a; although see De Waalet al. 2008). This limitation is made worse by the inability of chimpanzees to remain tolerantduring cooperative activities. Although chimpanzees are skilled at using other chimpanzees associal tools, cooperation breaks down when the reward for joint effort becomes easily monopolized(MacLean&Hare2013).Chimpanzeesareconstrainedintheirabilitytoinhibitintolerancetowardpotential cooperative partners even when they know they cannot solve a problem alone. Only asmall minority of chimpanzee dyads within a group are tolerant enough to work for sharable foodand no dyad can cooperate once rewards require active sharing or turn taking (Hare et al. 2007,Melis et al. 2006b). This intolerance and inflexibility in cooperative-communicative contexts likelyprevent chimpanzees from sharing intentions and building on previous innovations in a processknown as cultural ratcheting (Hare 2011).
25
26THE DOMESTICATION OF DOG SOCIAL COGNITION
27
28Domestic dogs are more skillful at using human gestures than nonhuman apes (Hare et al. 2002).Dogs follow the direction of a human gaze or point to locate hidden food or toys. If a humanpoints to one of two locations, dogs are more likely to search where a human has indicated. Dogscan even spontaneously use novel and arbitrary gestures to help direct their search for objects orfood. Several controls rule out the possibility that these searches in response to human gesturesare reflexive or based on olfactory cues (Hare & Tomasello 2005b). Another similarity betweendogs and infants is that both commit the A not B search error when directed by a human in theclassic Piagetian task. Like human children in this task, dogs search in a hiding location that hasrepeatedly been baited in the past over a new location they observe being baited (Top ́al et al. 2009).Dogs are also the only nonhumans capable of fast mapping. In a way similar to the way youngchildren learn words, several border collies have acquired hundreds of object labels using theprinciple of exclusion after single interactions with each new toy (Kaminski et al. 2004). Dogs seemto understand the cooperative-communicative intent of human signals in ways resembling thoseof the youngest human infants. This raises the question of how such a distantly related species canshow skills that are crucial to human social cognitive development (Kaminski & Marshall-Pescini2014). Comparisons of canids have revealed an important process by which social cognitionevolves.
29
30Dogs Are Wolves Prepared to Cooperatively Communicate
31
32Dogs do not need intensive exposure to humans to begin using our gestures. Dog puppies also showskill at using basic human gestures (Hare et al. 2002). Although variance exists among individualdogs, as a species, dogs rely on human gestures even as puppies (Hare et al. 2010, Stewart et al.2015, Wobber et al. 2009). Dogs did not inherit this early-emerging pattern from wolves. Theskills of wolves at reading human gestures are more similar to those of nonhuman apes thanto those of dogs. To develop skills in reading human gestures, wolves require intensive humansocialization during a short critical period; as the window of socialization closes, they show littleskill at understanding humans without explicit training as adults (Gacsi et al. 2005, Hare et al.2002, Viranyi et al. 2008). Unlike dogs and infants, wolves do not commit the socially mediated Anot B error (Top ́al et al. 2009). The independence of wolves from humans also means that whenfaced with an impossible task, they continue to try to solve the problem without help, whereasdogs quickly look to a human for help (Mikl ́osi et al. 2003).
33
34Foxes Selected for Friendliness Cooperatively Communicate
35
36Because the early-emerging skill at reading gestures is not observed in wolves, the unusual socialskills of dogs probably appeared during domestication (Hare et al. 2010). This hypothesis wastested by examining the cooperative-communicative abilities of an experimental line of foxesthat had been intensively selected to be attracted to and nonaggressive toward people for 45generations (Trut 1999). The experimental line was compared to a control line bred randomly forhow they respond to humans. As a result of this selection, the experimental line is highly prosocialtoward humans and exhibits a suite of phenotypic traits known as the domestication syndrome(Table 1). In comparison to the control line, the experimental foxes show the expected increasesin approach toward humans. However, they also show a high frequency of traits not intentionallyselected, including physiological, morphological, developmental, and behavioral changes also seenin domestic animals. This domestication syndrome includes increased brain serotonin levels andreductions in hypothalamic-pituitary-adrenal axis reactivity. Morphologically, the experimentalfoxes show increased frequencies of piebald coats, star mutations (white spots on the forehead),shorter muzzles, feminized faces, floppy ears, and curly or shortened tails (Trut et al. 2009).Developmentally, most traits in the experimental foxes appear to relate to expanded developmentalwindows. Experimenters can socialize experimental foxes earlier in development and the periodof socialization remains open for much longer (i.e., similar to that of dogs) (Belyaev et al. 1985).Adult experimental foxes use juvenilized vocalizations and social behaviors such as tail waggingfar more frequently when approached by humans than adult control foxes do (Gogoleva et al.2008). When tested on the same cognitive measures of cooperative-communication used withapes, dogs, and wolves, fox kits from the experimental line are more skilled than same-age controlkits. The experimental foxes spontaneously use basic human gestures in two different paradigmsand perform similarly to dog puppies of the same age. Controls demonstrate that the experimentaland control foxes are similarly motivated even though only the experimental line responded tohuman gestures. Although the experimental line was never selected (or even evaluated) on thebasis of their cooperative-communicative abilities with humans, experimental kits perform likedog puppies when responding to human gestures (Hare et al. 2005).The foxes’ performance with human gestures supports the hypothesis that dogs’ social skillsevolved not only during domestication but also as a result of it. The fox experiment demonstratesthat selection on emotional reactivity changes not only temperament but also a suite of unrelatedphenotypictraitsasacorrelatedby-productofthisselection.Themoreflexibleskillsofcooperatingand communicating with humans probably represent another of the by-products of this selection for friendliness. Foxes, like most mammals, use the social cues of conspecifics, but due to selectionfor prosocial interactions with humans, these old cognitive skills are applied in a new interactionwith a new social partner (Hare & Tomasello 2005a).Less reactive temperament likely replaced fear with an attraction to humans as selection actedon developmental pathways. Shifts in development, especially early in fetal development, can alteremotional reactivity and are thought to create a cascade of unselected consequences throughoutthe phenotype (Price & Langen 1992, West-Eberhard 2003, Wilkins et al. 2014). For example,the neural crest hypothesis suggests that the domestication syndrome is a result of changes inthe migration pattern of melanocytes during neural crest formation, which simultaneously affectsneurohormone levels, pigmentation, and morphology early in development (Wilkins et al. 2014),although not in a universal pattern across domestic mammals (Sanchez-Villagra et al. 2016).
37
38Dog Self-Domestication
39
40Based on the fox findings, it might be that dog cognition also evolved as a by-product of selectionon emotional reactivity. However, in the case of dog evolution, natural selection acted on thetemperament of wolves. Wolves with a temperament allowing them to approach human settle-ments showed higher reproductive success and favored self-domestication (as do species adaptingto urban environments today; Ditchkoff et al. 2006). A population of wolves able to exploit thenew niche of scavenging food remains and feces became more tolerant of humans and could bediscriminated from other wolves due to morphological traits associated with the domesticationsyndrome. This new interaction with humans allowed old cognitive abilities to be expressed ina new context and resulted in more flexible social problem solving with humans. As the con-straint of temperament was lifted, heritable variance in these newly revealed social skills may havebeen targeted by selection (Wobber et al. 2009). Thus began the most successful interspecificcooperative-communicative relationship in mammalian evolutionary history (Hare et al. 2010,Hare & Woods 2013; although see Udell et al. 2010). This led to a bond so strong that exogenousadministration of oxytocin in dogs also modulates increases in mutual gaze, physical contact, andendogenous oxytocin expression in the humans with which they interact (Nagasawa et al. 2015).Not only have the emotional systems of dogs evolved but they have also hijacked our emotionalsystems for at least 15,000 years (MacLean & Hare 2015b).
41
42BONOBO SELF-DOMESTICATION
43
44The experimental foxes and dog self-domestication both imply that natural selection can lead toincreases in prosocial over aggressive behavior, which can in turn lead to domestication syndrome(Table 1). The self-domestication hypothesis predicts that natural selection would also haveshaped other species for increases in prosociality. Bonobos (Pan paniscus), one of our two closestliving relatives, have been identified as a likely candidate for self-domestication (Wrangham &Pilbeam 2002).
45
46Sexual Selection of Friendlier Male Apes
47
48Bonobos differ from chimpanzees in their aggressive behavior (Hare et al. 2012). No bonobo hasever been observed to kill another bonobo (although for a potential exception see Wilson et al.2014). Unlike chimpanzees, male bonobos do not coerce females, commit infanticide, or targettheir own mothers for aggression. Also unlike chimpanzees, male bonobos do not form coalitionswith one another within their groups but instead rely on their mother’s status to gain access to females (Surbeck et al. 2011). Finally, unlike chimpanzees, male bonobos do not participate inborder patrols or lethal raids into neighboring territories.Hare et al. (2012) proposed that male bonobos evolved to be more prosocial through sexualselection. Living in a richer and more predictable ecology, unrelated female bonobos formedbonds that allowed them to respond to male aggression in a way that female chimpanzees cannot.Although wild female chimpanzees rarely support other females when they are targeted by aggres-sive males, similar coercion by male bonobos is met by female coalitionary defense (Tokuyama &Furuichi 2016). Similarly, female bonobos do not tolerate male aggression toward juveniles andinfants (Hare & Yamamoto 2015, Walker & Hare 2016). According to the self-domestication hy-pothesis, bonobos evolved to be less aggressive because females were able to express a preferencefor less aggressive males. 
49
50A Priori Tests of Bonobo Self-Domestication
51
52After Wrangham & Pilbeam (2002) initially proposed that bonobos could be a candidate forself-domestication, a number of a priori tests examined whether derived features of the bonobophenotype fit the expected pattern of domestication syndrome (Table 1).Hare et al. (2012) reviewed the evidence that differences between bonobo and chim-panzee prosocial behavior, physiology, morphology, development, and cognition support self-domestication. Bonobos are more tolerant than chimpanzees when sharing food because bonobosuse sex and play to reduce social tension when conflicts arise (Hare et al. 2007; although see Jaeggiet al. 2010). Bonobos also voluntarily share food. When given the choice of either eating preferredfood before their morning meal or opening a one-way door to allow another bonobo to share thefood, bonobos prefer to eat together. When bonobos can choose to open a door for either a groupmember or a stranger, they prefer to share with another bonobo with which they have neverphysically interacted (Tan & Hare 2013). This level of xenophilia contrasts with the xenophobiathat chimpanzees show toward strangers (Wilson et al. 2014).The unusual sharing observed in bonobos is modulated by their physiological response to socialstress. Before a dyad of male bonobos is released into a room with food, they show increases incortisol, associated with a stress response, but not testosterone, typically associated with agonism.The increase of cortisol in male bonobos promotes tolerance by creating a passive coping stylethat encourages social contact to reduce anxiety through sociosexual behavior (likely by releasingthe anxiolytic oxytocin). Chimpanzee males show the exact opposite response. The testosteronereactivity seen in chimpanzee males reduces the potential for tolerance because it primes them forcompetition as they strive for higher status (Van Honk et al. 2010, Wobber et al. 2010a).Neurobiologically, bonobos resemble animals that show increased serotonin levels andconcomitant reductions in anxiety and aggression as the first sign of domestication (i.e., Agnvallet al. 2015, Plyusnina et al. 1991). Although levels of serotonin in the cerebrospinal fluid have notbeen directly measured due to the obvious ethical concerns, postmortem neuroanatomical studieshave found that bonobos have twice the density, relative to chimpanzees, of serotonergic axonsin the basal and central nuclei of the amygdala (Rilling et al. 2011, Stimpson et al. 2015). Thisis a pattern consistent with a species where social frustration is less likely to lead to aggression(Bernhardt 1997).Morphologically, bonobos show features associated with self-domestication, including reducedcranial size, canine dimorphism, and depigmentation of the lips and tail tufts. Bonobos also showevidence of an expanded window of social development for behaviors involved in generating tol-erance. Bonobos exhibit early-emerging sociosexual behaviors that are used in infancy to mitigatethe risk of social conflict while sharing food. They also retain nonreproductive sociosexual and play behaviors into adulthood that act to maintain tolerance. In contrast, chimpanzees show nosociosexual behavior in infancy, only demonstrating this behavior during reproductive bouts asadults (Wobber & Hare 2015, Woods & Hare 2011).Cognitively, bonobos illustrate the connections between temperament, tolerance, and socialcognition. Temperamentally, bonobos are more cautious and observant toward novelty than otherapes, with the exception of human children (Herrmann et al. 2011). Although bonobos do notdiffer from chimpanzees in their use of human cooperative-communicative gestures, they are moresensitive to human gaze direction (MacLean & Hare 2015a). Bonobos are more likely to co-orientin response to a shift in a human’s gaze direction or to respond appropriately to their intentions inaction (Herrmann et al. 2010). Eye tracking comparisons between bonobos and chimpanzees showthat bonobos focus more on the eyes of people whereas chimpanzees tend to focus on the mouthsof the same people (Kano et al. 2015). Bonobos also show more flexibility than chimpanzees ininstrumental cooperative tasks that require feeding tolerance. Whereas even the most experiencedchimpanzee cooperators fail to cooperate to obtain monopolizable food, experimentally na ̈ıvebonobos succeed. Even when small amounts of prized food (four apple pieces) are placed in oneeasily monopolizable location, bonobos cooperate and, on average, split the reward in half. Thetemperament of bonobos, which is conflict averse and uses social interaction to ease tension, allowsthem to solve more cooperative problems with a wider range of social partners than chimpanzees(Hare et al. 2007).Bonobos also show developmental delay across a number of cognitive domains related to forag-ing and feeding competition—they are particularly uninhibited in social contexts as a result (Rosati& Hare 2012, Wobber et al. 2010b). This lack of inhibition likely presents a major constraint onthe cooperative-communicative abilities of bonobos relative to humans. Hare et al. (2012) inter-pret these and other a priori tests as overwhelming support for bonobo self-domestication. Thisraises the possibility that other apes may have been affected by the same process during selectionfor increased prosociality.
53
54THE INFLUENCE OF HUMAN TEMPERAMENT ON MENTALIZING
55
56Studies of nonhuman apes suggest that cooperative-communicative forms of mentalizing evolvedin our genus and are important to our cognitive sophistication. These studies also show thattolerance is a constraint on cooperation and communication in nonhumans. In bonobos, dogs,and foxes, natural or artificial selection for prosociality has led to increases in tolerance and socialcognitive flexibility in association with the domestication syndrome.On the basis of this type of comparative work, Hare & Tomasello (2005a,b) proposed theemotional reactivity hypothesis. This hypothesis suggests that human levels of cooperative com-munication were a result of an increase in social tolerance generated by a decrease in emotionalreactivity. Without tolerance, advanced computational or social cognitive abilities would not beof much use because individuals could not share the benefits of joint effort. According to thishypothesis, an increase in tolerance in humans allowed inherited cognitive skills to be expressedin new social situations. Selection could then act directly on revealed variance in these newlyexpressed cognitive abilities.Taking advantage of individual differences in human responses to novel or startling situations(Kagan & Snidman 2009), Wellman and colleagues conducted an a priori test of the predictedrelationship between emotional reactivity and theory of mind development in human infants(Wellman et al. 2011). Children were observed for their interactions with others and then testedon false belief tasks. Consistent with the hypothesis, infants with the least aggressive and mostsocially reserved temperaments show the earliest expression of the false belief understandingthat supports cooperative forms of communication—including language (Lane et al. 2013, Mink et al. 2014, Wellman et al. 2011). Related findings come from an fMRI study of adults. Afterbeing provoked in a competition game, women who were highly reactive in a startle test showedthe least activity in the temporal parietal junction (TPJ), medial prefrontal cortex (mPFC), andprecuneus (PC) when deciding how to punish other women. These highly reactive women hadthe least activity in the cortical hubs of the brain’s mentalizing network. In contrast, even afterbeing provoked, women with low reactivity had high activity in their mentalizing network. Lowreactivity led to more tolerance of provocation and more mentalizing (Beyer et al. 2014).Individual differences in this temperament–social cognition axis supports the idea that changesin human social cognition relied on shifts in the hormonal and subcortical profiles (e.g., amygdalareactivity) linked to temperament. Evolutionary shifts in hormonal or neuropeptide expression orreceptivity are evolutionarily labile and, as demonstrated in domesticated animals, can dramaticallyalter prosociality and are believed to produce a cascade of correlated phenotypic effects.Because human and nonhuman social behavior are modulated by neurohormones, a number ofhormones and neuropeptides are potential targets for prosocial selection. Serotonin, testosterone,and oxytocin are among the most important interactants mediating aggressive behavior (Kuepperet al. 2010, Montoya et al. 2012). In experimental animal populations selected for friendlinesstoward humans, increases in brain levels of serotonin are the first physiological sign of reducedemotional reactivity and aggression (Agnvall et al. 2015, Plyusnina et al. 1991). Exogenous sero-tonin in people (i.e., citalopram) increases harm avoidance and cooperative behavior during moraldilemmas and cooperative economic games (Crockett et al. 2010, Wood et al. 2006). Low testos-terone is related to male prosociality and parental care (Burnham 2007). Exogenously administeredoxytocin in humans reduces aggression directed toward in-group members compared to individ-uals from out-groups (De Dreu & Kret 2016). Selection for increased prosociality could havereadily acted on any of these hormones to increase tolerance through decreased emotional reac-tivity. The changes in developmental pathways needed to alter emotional reactivity can then havewidespread effects throughout the phenotype, including effects on social cognition.
57
58THE HUMAN SELF-DOMESTICATION HYPOTHESIS
59
60The HSD builds on the work of Hare & Tomasello (2005b) and Hare et al. (2012) by proposingthat modern humans were selected for prosociality. This hypothesis is inspired by the unusuallevel of intragroup tolerance and cooperation in modern humans and the link between tempera-ment and social cognition demonstrated in animals and humans. The HSD predicts evidence for(a) selection for prosocial behavior linked to derived human cooperative-communicative abilitiesand (b) the domestication syndrome in our morphology, physiology, development, and cognition,as seen in other self-domesticated species (Table 1) (Wrangham 2014).Evolutionary shifts in hormonal profiles related to tolerance and cooperation should be iden-tifiable by using the morphological signatures of these changes found in the fossil record as wellas by using genetic comparisons (seeTable 2). The HSD goes beyond the emotional reactivityhypothesis proposed by Hare & Tomasello (2005b) in recognizing the likelihood that interac-tion between subcortical and cortical pathways led to unprecedented human tolerance. The HSDpredicts that humans have reduced reactivity that increases the reward for social interactions, butit also predicts that, unlike any other domestic species, human tolerance is also due to massiveincreases in inhibition. The HSD suggests that it is this self-control combined with reduced re-activity that creates the human-specific adaptation for more flexible tolerance and unique formsof human social cognition.The HSD may play a role in explaining three major moments in human cognitive evolution:(a) the initial appearance of the human adaptive package inHomo erectus,(b) increases in brain size between 2 million years ago and 80 kya, and (c) the lag between reaching the lower rangeof modern human brain size 500 kya and the expression of full-blown modern cultural behaviorapproximately 50 kya. Although all three moments are touched on in this review, the focus is largelyon HSD as an explanation for the paradox of behavioral modernity in late human evolution (i.e.,the temporal gap between the appearance of human morphology and the consistent expressionof modern behavior). This is where the HSD currently provides the most testable predictions.In the following sections, I review an initial a priori test of the core predictions of the HSD andthen examine morphological, physiological, and developmental evidence that can be interpreted in favor of the HSD being an explanation of the appearance of modern human behavior in theLater Stone Age/Upper Paleolithic.
61
62PS68CH07-Hare    ARI    17 November 2016    12:56Table 2   Morphological signals of increased tolerance and cooperative-communicative abilities inHomo sapiensMorphological traitImageReduction in browridge and facial length2nd digit to 4th digitlength ratio (2D4D)White scleraGlobular cranialdevelopmentChanges indicatedTime of appearanceReferenceCieri et al. 2014Nelson et al. 2011Tomasello et al. 2007Middle/UpperPaleolithicAfter the split fromNeanderthalsHublin et al. 2015Reduced pubertalandrogens and lessdespotic behaviorin malesReduced prenatalandrogens andincreased sensitivityto social cuesCasey 2015MacLean et al. 2014Predicted: after thesplit from NeanderthalsAfter the split fromNeanderthalsPredicted: after thesplit from NeanderthalsPredicted: beginningwith the appearanceof genus HomoIncreases in oxytocinand mutual gazeEarly emergence ofsocial cognition andthe brain’s socialnetworkLate onset of adultself-controlIncrease in self-controland social toleranceAbsolute brain sizeincreaseExtended synapticpruningNeurotransmitterSynapseDendriteA
63
64The First A Priori Test of Human Self-Domestication
65
66Largely based on artifacts in the fossil record, researchers have inferred that behavioral modernity(i.e., the persistent establishment of behaviors requiring extremely flexible forms of causal reason-ing, episodic memory, symbolic thought, etc.; McBrearty & Brooks 2000) and cultural diversitydid not occur for several hundred thousand years after human brain size reached within the lowerend of the current modern range (Holloway 2015; although see Schoenemann 2006). Cieri et al.(2014) explored the possibility that the explosion of cultural artifacts beginning around 80 kyaoccurred due to selection for temperament that allowed more cooperative communication andpromoted rapid transmission of innovations.Both theoretical models and ethnographic studies suggest that the modern toolkit was a productof demographic expansion (Henrich 2015). Before the Upper Paleolithic,H. sapienstransitionedfrom a low-density distribution to a larger, high-density distribution across a greater range ofecologies. This created a larger network of innovators and resulted in the technology revolutionof the Upper Paleolithic.The challenge for this hypothesis is explaining the sudden appearance of cultural ratchet-ing as competition for scarce resources increased with population density (Cieri et al. 2014).Without extreme levels of social tolerance, this type of competition would not only impede thesocial transmission of innovations but also prevent prosocial interactions (i.e., Aureli & De Waal1997, Horner 2010). Cieri et al. (2014) predicted that an increase in tolerance and demographicpressure allowed a wider network of demonstrators to interact and learn from each other. Thisallowed existing cognitive skills to be expressed in a wider range of contexts across a broader socialnetwork.Both testosterone and serotonin affect craniofacial morphology during development. Largepubertal spikes in testosterone are associated with an enlargement in the suborbital torus, or browridges, and elongated upper faces (Cieri et al. 2014). When facial features are manipulated inphotos, people judge the exaggerated facial features produced by pubertal testosterone as beingmore aggressive and less trustworthy (e.g., Wilson & Rule 2015). Serotonin also plays a role inthe early human fetal development of craniofacial morphology, although the mechanism is lesswell understood. Women taking selective serotonin reuptake inhibitors such as citalopram (anantidepressant medication) have an increased risk of giving birth to infants with reduced cranialsize (Alwan et al. 2007). These craniofacial changes in humans echo those in domesticated animalsand bonobos, in which an increase in serotonin and reduction in testosterone are associated withfacial feminization and reduced cranial capacity (Hare et al. 2012).Cieri et al. (2014) compared brow ridge size and facial width and length in fossil humans andmodern humans. They predicted increasingly hypoandrogenized facial features across late humanevolution. They also predicted reduced cranial capacity in Holocene humans, as observed in otherdomesticated animals.In 13 modern human fossil crania from the Middle Stone Age and Middle Paleolithic (prior to80 kya), 41 Later Stone Age/Upper Paleolithic (38–10 kya) crania, and more than 1,300 Holocene(less than 10 kya) crania, a temporal decrease in brow ridge size and the length of the face wasobserved. The brow ridge projection index of the Middle Stone Age sample was 1.5 standarddeviations above that of the Late Stone Age sample and as much as 3.0 standard deviations abovethe Holocene samples. This difference was present in both the hunter-gatherer and agriculturalsubsamples within the recent human sample.
67Facial length followed a similar pattern, with substantial facial shortening after 80 kya. Themost extreme facial shortening was in modern hunter-gatherers. Agriculturalists seem to regressand develop slightly longer faces than hunter-gatherers. Replicating previous analysis, modernhunter-gatherers and agriculturalists both had smaller cranial capacity than the Late Stone Agehumans (Leach 2003, Zollikofer & Ponce de Leon 2010).It is also important to note that while Cieri et al. (2014) confined their analysis to modernhumans, these results would have been far more extreme if they included similar comparisons toHomo neanderthalensisandHomo heidelbergensis. Although both archaic human species had similar-sized brains compared toH. sapiens, their faces are far more masculinized than the oldest modernhumans (Churchill 2014).Other biomarkers suggest reduced intrasexual competition and androgen exposure in modernhumans. Compared to Neanderthals and other hominins, the 2D4D ratio, the relative length ofthe second digit to the fourth digit (i.e., the index and ring fingers), of modern humans suggestslow prenatal androgen exposure in utero (McIntyre et al. 2009, Nelson et al. 2011). In mammaliantestes, comparisons of gene expression profiles across development also suggest pedomorphismin human testis development. Human gene expression is more similar to adolescent mice thanto chimpanzees that have gene expression profiles similar to other species with high levels ofintrasexual competition (Saglican et al. 2014).Fossil evidence suggests craniofacial feminization occurred just as cultural ratcheting was push-ing us toward behavioral modernity. When human populations became increasingly connectedand concentrated at high densities during the Holocene, human brains were reduced in size, whichin other species is associated with an increase in serotonin. Although brain size generally increasedduring the evolution ofHomo, there was a modest reduction in brain size late in human evolutionthat resembles reductions in domesticated animals. The 2D4D digit ratio and testis developmentsupport a shift in intrasexual competition and androgen exposure. The more masculinized 2D4Dpattern in Neanderthals suggests this shift occurred late in human evolution.The observed morphological changes in modern humans support the HSD. Selection likelymodulated tolerance with increased serotonin levels and a reduction in androgen exposure. Malesin particular must have been less aggressive in dominance interactions. The new level of toleranceresulted in old cognitive abilities being used more flexibly in new social contexts. Any heritablevariance in these new skills, revealed when the constraint of intolerance was lifted, could thenbecome the direct target of selection.
68
69The Eye of Cooperative Communication
70
71In the past two decades, another morphological signal of human cognitive evolution has beendiscovered. Of the dozens of primate species examined, only humans fail to produce eye-sclerapigment (Kobayashi & Kohshima 1997, 2001). All primates except humans pay a metabolic costto hide the social information contained in the direction of their gaze by producing melanin(Kobayashi & Kohshima 2001). Humans have white sclera and an elongated eye shape that con-trasts with the surrounding facial pigment and advertises the direction of our attention. Individualdifferences in levels of sclera melanin in nonhuman apes suggest heritable variability across homi-noids (Mayhew & G ́omez 2015).Humans show an early-emerging and subconscious preference for eyes with white sclera thatpersists into adulthood. Infants show a preference for looking at white sclera in their first weeks(Farroni et al. 2004). Infants prefer to look at faces with white sclera and dark pupils over eyes withan inverted pattern of coloration (i.e., black sclera, white pupils) (Farroni et al. 2005). Childrenprefer stuffed animals with white rather than darkened sclera. Adults share this preference but are unaware that their attraction is driven by the presence of white sclera (Segal et al. 2016, Whalenet al. 2004). By 7 months old, infants rely on white eye sclera to encode emotional cues using thesame cortical network observed in adults (Grossmann et al. 2008, Jessen & Grossmann 2014).Comparisons between human infants and nonhuman apes show that, although all species followgaze direction, only humans prioritize eye direction over facial direction when coorienting toanother’s gaze. The increased reliance of infants over other apes on eye gaze is probably facilitatedby our white sclera (Tomasello et al. 2007).Our visible eyes provide information about eye direction and cues of mutual gaze that seemcrucial to our unique forms of learning, cooperation, and communication (Grossmann 2016).Joint attention, which allows children to learn the association between labels and objects, relieson a child’s ability to follow an adult’s line of gaze (Tomasello 2009a, Tomasello & Farrar 1986).Mutual gaze is critical to the formation of shared intentions—a central psychological mechanismallowing for the development of unique human forms of cooperative communication (Tomaselloet al. 2005).White sclera may increase the potential for joint attention by enhancing eye-blink synchroniza-tion. When eye blinking is synchronized, humans improve the coordination of their subsequentmotor activities. This synchronization is thought to lead to a Hebbian association (i.e., neuralsynchronization) between the activity in the inferior frontal gyrus of two people locked in mutualgaze. This interneural synchronization may prime other regions of the brain’s social network andfacilitate shared intentions between individuals (Koike et al. 2016). White sclera may serve toincrease the likelihood and length of this synchronization.Visible eyes also promote cooperative behavior in humans. People donate more in a publicgoods game when they are watched by a robot with large eyes. Subjects whose computer screensdisplay a robot with oversized white sclera donate∼30% more to the public good than thosewithout the robot on their screens (Burnham & Hare 2007). When a picture of human eyes wasplaced above a public bicycle rack or on a paper leaflet, bike theft was eliminated and litteringreduced. In the controls without the eye manipulation, more bikes were stolen and there was morelittering (Bateson et al. 2015, Nettle et al. 2012).Eyes with white sclera favor expression of prosocial behavior. Visible human eyes facilitatecooperation by signaling the potential for social sanctions or reputational consequences. Thesesignals probably affect human cooperation subconsciously (Burnham & Hare 2007; although seeFehr & Schneider 2010, Jessen & Grossmann 2014).Depigmentation of human sclera may be a by-product of self-domestication because selectionagainst aggression alters melanin expression. Furthermore, scleral tissue originates from the neuralcrest (Seko et al. 2008, Wilkins et al. 2014). If shifts in melanocyte migration of scleral tissueproduced variance in white sclera as a result of self-domestication, selection could subsequentlyact directly on any heritable variance (West-Eberhard 2003).Given the role of neuropeptides and hormones in mediating human gaze behavior, prosocialityand white sclera likely coevolved. In particular, oxytocin is known to modulate mutual gaze andincrease attention to the eyes of others in humans (Gamer et al. 2010, Guastella et al. 2008). Inexperiments in which human males examined only the eye region of people making different emo-tional expressions, exogenous administration of oxytocin improved the ability of the participantsto infer the other person’s affective state (Domes et al. 2013, Meyer-Lindenberg et al. 2011).Effects of exogenous oxytocin seem specific to improving social memory of faces rather thanto improving memory more generally (Rimmele et al. 2009). Oxytocin also facilitates bonding ofparent and offspring through touch and mutual gaze (Baribeau & Anagnostou 2015, Carter 2014).This may explain increases in trust during cooperative games in subjects that have been givenintranasal oxytocin (Kosfeld et al. 2005). The role of mutual gaze and oxytocin in the human–dog bond demonstrates that this effect even extends to human–animal interactions (Nagasawa et al.2015, Waller et al. 2013).The HSD predicts that increasing oxytocin levels, receptor densities, or receptor responsive-ness in the brain was a crucial step in promoting the evolution of human prosociality (Baribeau &Anagnostou 2015). These increases would have promoted the use of eye gaze and favored the evo-lution of white sclera, allowing the close interactions that led to the social bonding and cooperativecommunication necessary for the expression of modern human behavior.
72
73Like-Me Psychology Drove Paleolithic Self-Domestication
74
75Humans are helpful or hurtful toward others based on perceived similarity to themselves. Asadults, this like-me psychology manifests itself as in-group favoritism across a variety of contextsand cultures (Mullen et al. 1992). This favoritism results in a high degree of tolerance towardin-group members that facilitates unique forms of collaboration and conformity (Burton-Chellew& West 2012, Kurzban et al. 2015). In contrast, ostracism and lethal aggression among hunter-gatherers primarily targets nonconformist or out-group members (Boehm et al. 1993, Wrangham1999). This type of antisocial or agonistic response is facilitated by the readiness of humans todehumanize out-group members or those that dehumanize their own in-group (Hodson et al.2014, Kteily et al. 2016).This in-group versus out-group preference appears early in development, which suggeststhat humans are prepared for social discrimination based on like-me preferences (Bloom 2013,Mahajan & Wynn 2012). The latest neurobiological evidence and evolutionary models suggestthat intragroup prosociality can explain our paradoxical kindness and cruelty toward others. Selec-tion for in-group prosociality drove late human self-domestication and, as a correlated by-product,is responsible for extreme forms of out-group aggression.Spontaneous sharing and helping develops early in infancy along with an in-group bias (Hamlin& Wynn 2011, Hamlin et al. 2007, Kinzler et al. 2011, Warneken 2015, Warneken et al. 2007).Nine-month-old infants prefer puppets that help another puppet who shares the child’s foodpreference but also prefer puppets that harm another puppet with a dissimilar food preference(Hamlin et al. 2013). In sharing games, children as young as 5 years old prefer to share within-group members, whereas 6-year-olds are more willing to pay a cost to punish selfish out-groupmembers than in-group members (Engelmann et al. 2013, Jordan et al. 2014). Children also seemselective in enforcing norms depending on a violator’s group affiliation (Schmidt et al. 2012). Thisearly-emerging expression of in-group bias supports the idea that we are biologically prepared tosupport in-group cooperation and communication.The neurobiology of this like-me psychology is also present in adults. A cortical network allowsfor the attribution of mental states and is built from regions specialized in social decision making(Adolphs 2009, Carter & Huettel 2013). Brain imaging using fMRI scans shows involvement ofthe mPFC, TPJ, and superior temporal sulcus (STS) in tasks requiring subjects to model theintentions, emotions, and beliefs of others (Amodio & Frith 2006, Harris & Fiske 2009). Ventralareas of the mPFC also show activity when people think about their own thoughts or emotionsabout others (Cikara et al. 2014b).This cortical network allows people to compare their own thoughts and feelings to those ofothers and is central to generating both compassionate and dehumanizing responses. When otherhumans are perceived as having good intentions (being warm) and having the capability to carrythem out (being competent), the mPFC is heavily involved in modulating empathic responses. Incontrast, the mPFC becomes less active when people are shown photos of individuals perceived as incompetent and cold (i.e., homeless people, drug addicts), and the amygdala and insula thatencode disgust become the most active regions (Harris & Fiske 2006, Rilling et al. 2008a).When subjects were allowed to punish in-group and out-group members for norm violations,punishment of in-group members was less likely and was associated with heightened activity andconnectivity between the mPFC and TPJ. Mentalizing regions become more active, suggestingthat people were justifying their groupmates’ infractions; the same regions were not as activewhen out-group members made the same transgression (Baumgartner et al. 2012). When there isdecreased activity in this mentalizing network, people are able to dehumanize others and are lesslikely to show empathy, tolerance, or prosociality (Baumgartner et al. 2012, Cikara et al. 2014a,Fiske et al. 2007, Harris & Fiske 2009, Waytz et al. 2012).The neuropeptide oxytocin is the strongest candidate for explaining the human-unique patternof empathy and dehumanization. Oxytocin not only increases eye contact and social bondingin humans but also exaggerates in-group favoritism. Adults given oxytocin are more likely tohumanize in-group rather than out-group members by attributing human-unique emotions to thein-group members and showing increased positive evaluations of them (De Dreu et al. 2011). Menwho are given oxytocin are three times more likely to donate money to their group rather than keepit for themselves (De Dreu et al. 2010). In economic games, intranasal oxytocin also reduces thelikelihood of men to cooperate with out-group members when they become a threat to in-groupmembers (De Dreu 2012, De Dreu et al. 2010, De Dreu & Kret 2016). Increased in-group bondsappear to drive larger defensive responses against potential threats from out-group members.These results probably occur because of oxytocin’s influence on the brain’s social network,which allows for mentalizing and empathy. Immunohistochemistry suggests the presence of oxy-tocin receptors in the cingulate cortex and amygdala and perhaps even the frontal cortex (Bocciaet al. 2013). Intranasal oxytocin also increases the resting-state connectivity between the amyg-dala and the mPFC (Sripada et al. 2013). This may cause a blunted mPFC response in individualscompeting on behalf of in-group members against an out-group. Consistent with this idea, lowermPFC reactivity has been observed during competitions that increased people’s willingness toharm out-group competitors (Cikara et al. 2014a). This is consistent with a subcortical systemheavily impacted by serotonin and oxytocin, which both mediate the strength of the response inthe brain’s social cortex.Selection for prosociality, which results in a reduced androgen profile and increases in serotonin(orreceptordensities)assuggestedbyCierietal.(2014),isexpectedtofacilitateoxytocinexpressionand binding (Baribeau & Anagnostou 2015). Sex steroids, including testosterone, affect the bindingof oxytocin and vasopressin. Whereas testosterone facilitates vasopressin binding, oxytocin andtestosterone are antagonistic. The production of oxytocin probably depends on serotonin receptoractivity and generates a positive feedback loop, whereas serotonin increases in the presence ofoxytocin.The effects of oxytocin are mediated through serotonin neurons (Baribeau & Anagnostou2015). This suggests that oxytocin’s dependence on serotonin and interactions with testosteronealtered its expression or reception over the past 80,000 years. The HSD predicts that the in-groupbonds of our species coevolved with out-group distrust due to changes in the serotonergic andandrogen systems that allowed oxytocin to have a greater impact on cortical regions related tosocial decision making.Although intergroup lethal aggression is likely a conserved trait (Gomez et al. 2016, Wilsonet al. 2014, Wrangham & Glowacki 2012, Wrangham et al. 2006), evolutionary modelingsupports the idea that any intensification of out-group aggression could be a by-product ofselection for intragroup prosociality late in human evolution. If each behavior evolves in isolation,the payoff is not as adaptive. If they emerge simultaneously, in-group favoritism in combination with out-group hostility is a highly successful strategy (Choi & Bowles 2007). The interactionof oxytocin, serotonin, and testosterone suggests a way in which enhanced in-group prosocialityand out-group aggression may have coevolved.Ethnographic analysis also supports the idea that humans show a shift in social structure con-sistent with self-domestication and the coevolution of parochialism. Hunter-gatherers’ intragroupinteractions are best described as a reverse dominance hierarchy. Group members work togetherto defend each other against any individual trying to monopolize power in the group. This sug-gests that the most aggressive group members would be at a selective disadvantage. Aggressionoccurs, but ostracism and lethal aggression are levied against those who do not conform to themore egalitarian social system (Boehm et al. 1993). These bonded egalitarian groups would havebeen more successful in outcompeting other hominin species or human out-groups.Humans became kinder and crueler as a result of selection for intragroup prosociality. Selectionacted on neurohormonal channels that tune subcortical regions to be more or less reactive andsubsequently influence the identity of those perceived as like us or not. Just as oxytocin bondsparent to offspring, making parents capable of extreme defensive aggression, our species becamesimilarly protective of our in-group members against out-group threats.
76
77Self-Control: The Unique Feature of Human Self-Domestication
78
79Increases in brain size are the defining characteristic of evolution in our genus and a morphologicalsignal of increased tolerance through self-control. Larger brains are associated with increased self-control. Self-control supports executive function and allows cortical regions to govern subcorticalregions. The effects of self-control thus include inhibiting aggressive responses in favor of prosocialreactions (MacLean 2016). This effect contrasts with high or low subcortical reactivity that maysacrifice inhibition, as in bonobos and some dog populations (Bray et al. 2015, Wobber et al.2010b).A large-scale phylogenetic study (MacLean et al. 2014) suggests the relationship between brainsize and self-control. An average of 15 individuals from 36 species of mammals and birds (N>550) were tested for their ability to spontaneously inhibit a prepotent response in two differenttasks. The first task presented subjects with food in a transparent tube. A correct response requiredinhibiting the urge to reach directly for the food and instead taking a detour by reaching throughone of the tube’s open ends. The second task gave subjects the Piagetian A not B task, whichrequires inhibiting perseverative search errors by choosing where food is hidden as opposed towhere it was repeatedly hidden in the past.Absolute brain size predicted performance across species. In more than 20 primate species,there was no link between performance and ecological variables such as social complexity orfrugivory. Instead, brain size was the best predictor of inhibition. Brain size explained up to 70%of the variance in self-control across primate species (MacLean et al. 2014). This indicates thatincreases in absolute brain size in humans were likely accompanied by increases in self-control.MacLean et al. (2014) suggest that the relationship between brain size and self-control inprimates exists due to the unique scaling relationship between neuron densities in primates. Un-like other taxa, which show reduced neuron densities in larger-brained animals, primate neuronnumbers scale isometrically with brain size. In primates, larger brains have the same densities ofneurons, leading to exponential growth in potential networks between them (Azevedo et al. 2009).As the total number of neurons increase, primate brains become more modular, which may createnew neural networks (Kaas 2000, Rilling et al. 2008b). Human brains take this primate trend to itsextreme (Herculano-Houzel 2012). Thus brain size, neural numbers, and self-control can evolveas by-products of selection for body size (although see Grabowski et al. 2016). Simply becoming larger to avoid predation or to promote thermoregulation would also increase self-control andtrigger a positive evolutionary feedback loop during early human evolution (e.g., perhaps as seenbetween the small- and large-brained representatives of earlyHomo; Wood & Boyle 2016). Initialincreases in self-control probably increased energetic productivity through more flexible solutionsusing ancient cognitive skills in new ways.Cooking is a strong candidate for the initial trigger of this evolutionary feedback loop.Nonhuman apes prefer cooked food and have many of the cognitive prerequisites for cooking(Warneken & Rosati 2015, Wobber et al. 2008). A slight increase in self-control would havebrought the energetic payoffs of cooking within reach (Wrangham 2009). Investing in expensivebrain tissue would increasingly produce benefits as human energetic productivity expanded withmore sophisticated cognitive abilities (Berbesque et al. 2016). These benefits would have led tothe modern human energy budget, which is beyond anything seen in other primates. As a result,foragers would have higher reproductive rates and larger brains than any other ape (Pontzer et al.2016). The cognitive by-product hypothesis is supported by fossil data interpreted to show brainsize in our lineage scaling allometrically with body size until approximately 600 kya. According tothis view, human brains became disproportionately large only in the last half million years (Hublinet al. 2015).The HSD predicts that increases in self-control as a result of an increase in brain size steadilydrove the evolution of tolerance and social cognitive skills. Late in human evolution, selection forin-group tolerance intensified and acted on our emotional axis, which, together with preexistingself-control, created unprecedented levels of social tolerance. Over the past 100,000 years, humansbegan to outcompete other hominins through the cooperation and communication that resultedfrom this increase in in-group bonding, tolerance, and cooperation.
80
81Self-Domestication Through Evolutionary Development
82
83It is important to consider the mechanism that selection might have targeted to produce thesechanges during human evolution. The HSD predicts that the evolution of developmental pathwaysis the unifying mechanism leading to the social cognition, temperament, and self-control thatcreate unique human intragroup tolerance.The widening of developmental windows is a common consequence of domestication. In do-mesticated animals, ancestral behavioral traits appear earlier and persist for longer (Trut et al.2009). This heterochrony suggests that a similar shift in human development provided the mech-anismforincreasesintoleranceandcooperativecommunication.Comparativeandneurobiologicalwork provides evidence of cognitive ontogeny consistent with the prediction that the human de-velopmental window extended both earlier and later for cognition related to increased toleranceand cooperative communication.Humans have unique early-emerging social cognition that facilitates participation in culturalforms of learning and is supported by a pattern of secondary altriciality and globular brain de-velopment. Longitudinal comparisons between age-matched human infants and nonhuman apesacross a range of cognitive skills have revealed shifts in human cognitive development.Wobber et al. (2014) published the first comparison that uses a longitudinal design to comparethe cognitive development of two dozen bonobos and chimpanzees to a similar sample of age-matched children. Each subject was tested on the same battery of cognitive tasks each year between2 and 4 years old. The battery was based on tasks used by Herrmann et al. (2007) and includedsocial and nonsocial problem-solving tasks.Contrary to models predicting slower cognitive development in humans relative to other apespecies (e.g., Charnov & Berrigan 1993), 2-year-old children are more skilled than nonhuman apes in social tasks requiring cooperation and communication. Performance in these tasks is nearceiling levels in humans by age 4, whereas the other apes show little development by this sameage. The same 2-year-old human infants perform at a similar level to the other apes in nonsocialtasks (i.e., tool properties, numerosity, spatial memory, etc.).This provides powerful evidence for specialized and early-emerging social cognition, whichbecomes the scaffolding for subsequent social learning in human infants (Herrmann et al. 2007).Early-emerging social skills allow human infants to cooperate and communicate with others andaccess all forms of cultural knowledge.Fossil evidence provides a clue to when this early-emerging social neurocognitive networkmight have evolved. The most prominent feature of human brain maturation is secondary altri-ciality or helplessness in human newborns. Human brains are born at 25% their adult volumecompared to other apes born with 45% of adult brain volume (Zollikofer & Ponce de Leon 2010).This extreme level of postpartum brain development gives unusual influence to social input duringbrain development (e.g., eye contact, motherese, etc.). Social interactions can influence the struc-ture and organization of brain development during postnatal brain growth and probably facilitatethe early-emerging social skills observed by Wobber et al. (2014).Morphologically, the most unique feature of the human skull is its globular shape (Zollikofer& Ponce de Leon 2010). Analysis of globularization shows that this shape change occurs early indevelopment and is largely a result of the maintenance of fetal brain growth rates until the eruptionof the deciduous teeth at around 30 months old. By 4 years of age, synaptic densities begin to peak,the brain approaches adult size, and more than 60% of a child’s metabolism is directed towardbrain growth (Hublin et al. 2015, Kuzawa et al. 2014). The globular shape that is produced bythe extension of exaggerated growth rates is largely driven by expansion in the brain’s parietalregion, which includes the PC and TPJ (Bruner et al. 2016, Gunz et al. 2012). This globularexpansion is facilitated by elevated levels of white matter development in the human infant PFCrelative to those of chimpanzees (Sakai et al. 2011). There is also evidence that the brain’s corticalsocial network (i.e., the TPJ, STS, and mPFC) becomes increasingly active in infants during thisperiod of globular brain development (Grossmann 2015). Globularization has also been linked tochanges in the development of the neural crest as well as to a set of candidate genes that showsigns of positive selection in humans (Benitez-Burraco et al. 2016).All of the brain regions leading to globular expansion are also involved in human social cogni-tion, including the attribution of mental states to others. This raises the possibility that globulargrowth in the fossil record signals the evolution of early-emerging social cognition.Fossil studies have provided estimates for the appearance of secondary altriciality and glob-ular brain development. In comparing the crania of infantHomo erectus,H. neanderthalensis,andH. sapiens, researchers have found thatH. erectusinfants were born with a more developed brainthan those of later hominins. Secondary altriciality is thought to have evolved in the commonancestor of Neanderthals and modern humans. In contrast, globular development was not seeninH. neanderthalensis,suggesting that it is a unique feature of our species (Hublin et al. 2015).These studies support the idea of an early window of social cognitive development associated witha derived strategy of brain maturation that evolved uniquely inH. sapiens.Although social cognitive skills develop early, synaptic pruning in cortical regions involved inexecutive function develops late. Human infants show similar levels of self-control to other apesuntil early childhood. It is not until around 6 years old that children show more inhibition thannonhuman apes (Herrmann et al. 2015, Vlamings et al. 2010).Even adolescent humans have lower self-control than adults. Adolescents engage in higher-risk behaviors while showing greater aversion to social punishment (Casey & Caudle 2013). Thesetrends are associated with the final stages of synaptic pruning in cortical regions thought to be PS68CH07-Hare    ARI    17 November 2016    12:56involved in executive control and inhibition (Casey 2015). Synaptic pruning in regions of the PFCrelated to self-control are only complete in our mid-20s. Only when these self-control networksare complete do adults become more risk averse and less sensitive to failure (Casey 2015). Asthe human brain increased in size, the process of synaptic pruning probably became increasinglypedomorphic. Brain myelination and white matter show a similar pattern of delayed development.Chimpanzees and macaques complete myelination at sexual maturity, but this same process is notcomplete in humans until our mid-20s (Somel et al. 2013).Genetic comparisons also support the theory of a widened developmental window in humans.Comparisons of gene expression in macaques, chimpanzees, and humans reveal that a host ofgenes in humans are expressed at either an accelerated or delayed rate (Somel et al. 2009). Thelargest changes in the developmental timing of brain gene expression was observed in the humanPFC as compared both to other areas of the human brain and to the chimpanzee PFC. Humansynaptic genes in the PFC do not show peak expression until up to 5 years of age, whereas thepeak for these same genes is reached after a few months in chimpanzees. This results in a late agefor peak synaptic densities in the human PFC (3.5–10 years of age); in contrast, peak densities inthe human auditory cortex occur between 6 months and 3.5 years of age, and synaptic eliminationbegins in the human visual cortex a few months after birth. This graded pattern contrasts withsimultaneous peaks in all brain tissue types in macaques. The window of synaptic development islonger in humans because elimination begins later but occurs at a slower pace (Somel et al. 2013).Research on self-control and synaptic pruning is consistent with an increased window of devel-opment in humans as predicted by self-domestication. The HSD predicts that both early-emergingsocial cognition and delayed adult inhibition will be linked to selection for prosociality and providea mechanistic explanation for features of HSD.
84
85CONCLUSION
86
87Researchers have frequently made use of the concept of domestication in explaining human evolu-tion (Boas 1911, Gould 1977, Leach 2003, Wrangham 2014). Darwin (1859) beganOn the Originof Specieswith a discussion of domestication through artificial selection and spent decades collect-ing examples of natural variation produced through domestication (Darwin 1868). Domesticationwas crucial to Darwin’s case for evolution through natural selection and led him to consider thepossibility of human domestication (Darwin 1871).It was not until the pioneering work of Dmitry Belyeav and colleagues that the HSD, or the linkbetween selection for prosociality and a wide variety of correlated by-products, was discovered (fora review, see Hare & Woods 2013). These by-products include morphological and physiologicalchanges,increasesincooperativecommunication,andexpandingdevelopmentalwindows(Belyaevet al. 1985, Hare et al. 2005). Belyaev’s experimental domestication of foxes clearly defined theselection pressure, its effects, and the potential developmental mechanisms targeted to producedifferent domesticated phenotypes. The fox experiments allowed further comparative work to testwhether natural selection produced similar results in dogs, bonobos, and humans.The experiments with foxes and comparisons between dogs and wolves reveal that changes incooperative-communicative abilities can occur as a by-product of selection for prosociality andagainst fear and aggression. A priori tests of bonobo self-domestication support the possibilitythat natural selection causes similar results.The HSD builds on this comparative work and suggests that selection for prosociality alsoplayed a large role in human evolution, especially during the Middle and Upper Paleolithic(Figure 1). The first a priori test of this hypothesis found evidence by the Upper Paleolithic for theexpected link between increases in cultural artifacts and craniofacial signals of increased tolerance.
88Future tests can evaluate the proposed link between the evolution of self-control and emotionalreactivity that allows flexible human tolerance and social cognition. Future paleoanthropological,neuroendocrine, and genetic research will be able to further examine links between intragroupcooperative communication, morphology (i.e., eye color), and heterochronic shifts related to early-emerging social cognitive development. Hopefully, the HSD will help energize efforts towardanswering the ultimate Darwinian challenge: how our minds evolved and allowedH. sapienstosurvive as Earth’s last remaining human.
89
90SUMMARY POINTS
911. Darwin’s greatest evolutionary challenge is identifying derived forms of human cognitionand the processes by which they evolved. Given recent evidence of the existence of manylarge-brained human species within the past 50,000 years, we must answer this challengefor both our genus and our species.
92
932. Comparisons between human and nonhuman ape infants suggest that the early emer-gence of cooperative communication provides the developmental foundation for humancultural cognition.
943. Domestic dogs converge with human infants in their ability to use human cooperative-communicative gestures. Experimentally domesticated fox kits show dog-like skills inusing human gestures even though these skills were not under selection. An increase incooperative-communicativeabilityappearstobeaby-productofselectionforprosocialityover aggression.
95
964. Comparisons between bonobos and chimpanzees support the hypothesis that natural se-lection favoring prosociality over aggression can lead to self-domestication. Bonobosshare many traits with domesticated animals, including increased flexibility in somecooperative-communicative contexts.
97
985. The HSD suggests that natural selection for prosociality and against aggression playeda large role in human evolution. Over the past 80,000 years, fossil humans show mor-phological evidence for selection against aggression that coincides with an increase incultural artifacts in the fossil record.
99
1006. Selection for in-group prosociality drove human self-domestication in the Paleolithic.Changes in oxytocin and eye sclera color provide two possible mechanisms to explainthe increases in cooperative communication, increases in in-group cooperation, and in-tensification of intergroup conflict that evolved as a result of this selection.
101
1027. Evolutionarily labile neurohormones and neuropeptides provide a ready target of selec-tion for prosociality over aggression. However, human tolerance is flexible beyond whatcan be accounted for by muted subcortical responses alone. Phylogenetic comparisonssuggesting a strong link between inhibition and absolute brain size point to the criticalrole of cortical regions in allowing for human levels of self-control and tolerance.
103
1048. Human self-domestication predicts increased developmental windows for traits relatingto increased tolerance and cooperative communication. Early-emerging social cognition,which develops despite secondary altriciality, together with graded synaptic pruningcontinuing into adulthood, played a central role in the evolution ofH. sapiens.
105
106FUTURE ISSUES
107
1081. The HSD predicts a strong link between individual differences in temperamentalprofile and mentalizing abilities. Future studies should continue to find evidence forthis relationship early in development, using studies of heritability and cross-culturalcomparisons.
109
1102. Self-domestication predicts the appearance of the domestication syndrome, but its ex-pression likely differs across taxa due to phylogenetic distance and different develop-mental, neurohormonal, or subcortical targets. Traits considered part of the syndromedo not present themselves universally (Sanchez-Villagra et al. 2016). This inconsistentpattern may be the result of multiple pathways to increased prosociality, each of whichmay generate different sets of correlated by-products. For example, humans and bono-bos are hypothesized to be self-domesticated, but humans exhibit lethal aggression andneither hominoid shows the high frequencies of piebalding seen in many other domesticmammals (Wilkins et al. 2014). Tools will be needed to discriminate between selectionagainst different forms of aggression (i.e., defensive, predatory, intragroup, intergroup,etc.) that affect different physiological or developmental mechanisms and may lead tothis differential expression of correlated traits (Hare et al. 2012). Future selection experi-ments that target different forms of aggression or prosociality will provide powerful tests(Sanchez-Villagra et al. 2016).
111
1123. The HSD suggests that the underlying evolutionary genetics behind human white scleracan potentially reveal the time of origin for human forms of cooperative communication(Tomasello et al. 2007). For example, if genetic disorders related to sclera color arediscovered, then this could provide a powerful test of the self-domestication hypothesisusing comparisons of the human and Neanderthal genomes. Individual variability inscleral whiteness in nonhuman apes may offer another route to genetic clues about theorigin of human sclera coloration (Mayhew & G ́omez 2015).
113
1144. The HSD predicts thatH. sapienshave an expanded developmental window. Morpho-logical comparisons of development inHomoare based on a few fossil specimens. Futurediscoveries of additional specimens should support extended fetal levels of brain growthand cranial globularization as well as a slower life history inH. sapiens, which wouldsupport the hypothesis of gradation in brain development across cortical levels (i.e., latemyelination and pruning of the cortex).
115
1165. Selection for prosociality is only associated with reduced rather than increased brainsize. Another major force, or multiple forces, must have been at play during early humanevolution to drive initial body and brain size increases. Increases in body size to escapepredation or better thermoregulate or as a result of the island syndrome may have initiallyproduced tolerance as a by-product of self-control increases occurring with concomitantchanges in brain size. Selection could then target any heritable variance in self-control.This alone may be the cognitive trait allowing for an initial shift in human energy pro-ductivity (Pontzer et al. 2016). Comparative behavioral ecological studies will likely helptest whether similar evolutionary scenarios have played out in other taxa.
117
1186. The cooperative breeding hypothesis might be considered an alternative hypothesis tothe HSD (Burkart & van Schaik 2010). Although the two hypotheses will likely provecomplementary, it will be important to outline and test their competing predictions(ideally using comparative phylogenetic techniques). For example, cooperative breed-ing does not make the heterochronic or morphological predictions made by the HSD,but both hypotheses predict increased prosociality. Therefore, the hypotheses might bereconciled. For example, it is conceivable that increases in cooperative breeding duringhuman evolution might have led to sexual selection. Females may have chosen to bondwith males who did not aggress toward them but rather toward threats to their jointoffspring—leading to human self-domestication.
119
1207. Self-domestication is predicted to play a role in shaping the phenotypes of both island-andurban-livingpopulations (Ditchkoff et al.2006,Raiaet al.2010).Carefulcomparisonsof island and mainland as well as wild and urban populations will help reveal the ecologicalconditions that favor self-domestication. These lessons will likely also allow for inferencesregarding the ecological conditions that favored human self-domestication.
121
1228. If selection acts on prosociality across a variety of species, shared genetics might producethe observed convergence across some species. Any common genetics discovered couldbe evaluated in extinct and living humans. This would require genotypic and phenotypiccomparisons of different pairs of wild and domestic animals, but initial attempts havenot revealed this type of commonality (Albert et al. 2012). Future research can explorenew genetic candidates related to neural crest development, cranial globularization, anddomestication that might be associated with human self-domestication (Benitez-Burracoet al. 2016). However, the human case may prove exceptionally challenging becausehuman behavioral traits are hyperpolygenic (Chabris et al. 2015).
123
124DISCLOSURE STATEMENT
125
126The author is not aware of any affiliations, memberships, funding, or financial holdings that mightbe perceived as affecting the objectivity of this review.
127
128ACKNOWLEDGMENTS
129
130I want to thank Richard Wrangham for making the connection between the dog domesticationliterature and hominoid evolution, both in the case of humans and bonobos. His ideas, discussions,and collaboration inspired this review. I also want to thank James Brooks for help with the referencesection and Steve Churchill, Evan MacLean, Jingzhi Tan, Michael Tomasello, Vanessa Woods,and Richard Wrangham for extremely valuable comments that resulted in a much-improved paper.
131
132
133LITERATURE CITED
134略