Richard Wrangham’s The Goodness Paradox is an excellent book. Like his previous books, Demonic Males (; with Dale Peterson) and Catching Fire (), it is bold, innovative, clearly written, and often persuasive. The Goodness Paradox covers an impressive array of topics, ranging from the neurobiology and genetics of aggression, to the origins of our moral emotions, to the evolutionary underpinnings of militaristic psychology. Scientists and empirically oriented philosophers alike will find much of value in the book. Here I summarize and then briefly evaluate the book’s overall argument.
The paradox as Wrangham sees it is this (Introduction, Chapter 1): Humans are an unusually docile species compared with mammals in general and primates in particular. Our within-group interactions are marked by a high degree of social tolerance, which is foundational to our unique forms of cooperation and cultural learning. At the same time, we are an unusually violent species. While infanticide is relatively widespread in the animal world, few other species intentionally kill adult conspecifics. Humans collectively organize and carry out such killings, often with great efficiency. How can these opposing tendencies reside within one and the same psychology, and how and why did we evolve to be this way?
In solving this paradox, Wrangham’s first step is to distinguish two types of aggression: reactive and proactive (Chapter 2). Reactive aggression is ‘hot’—its expression is accompanied by anger, fear, or both; proactive aggression is ‘cold’—it is planned, controlled, and often carried out in the absence of strong emotion. The sense in which humans are unique, Wrangham explains, is that we show a reduced tendency for reactive aggression and an increased tendency for proactive aggression. The two types of aggression are associated with distinct neurological pathways and structures. Heritability and behavioural genetics studies show that these pathways and structures are shaped to a considerable extent by genes.
This neatly explains how both tendencies can be simultaneously present in humans. The remainder of the book tackles the rest of the paradox: how and why we evolved to be this way. This material naturally divides into two parts. The first addresses our reduced tendency for reactive aggression, and revolves around the self-domestication hypothesis (Chapters 3–6). This hypothesis has recently been championed by Wrangham and colleagues, but has a much longer history (nicely summarized here). The hallmark of domestication is docility, but that is just one of many traits that domesticated animals share. Relative to their wild ancestors, domesticates show reduced skull (and hence brain), jaw, and tooth size, a more gracile skeleton, floppy ears, curly tails, white patches on the head, white feet, higher reproductive turnover, decreased sexual dimorphism, increased playfulness, increased sexuality, and more. For a long time, this syndrome of traits was explained in terms of these animals adapting to life with humans (domesticates evolved to have white patches on their foreheads, for example, because that allowed human keepers to more readily identify them). Then the work of Alexander Belyaev and colleagues showed that the entire domestication syndrome can be produced by selection for reduced reactive aggression alone. Wrangham terms this result ‘Belyaev’s rule’. The common thread running through these traits is that they are paedomorphic: they are associated with earlier life stages in the ancestor species, now retained later into life. This process of juvenilization is evidently caused by changes to the neural crest cell system.
Armed with Belyaev’s rule—or more strictly speaking, its inverse—Wrangham first looks at bonobos and then humans (Homo sapiens). Unfortunately, we have no fossil evidence of the immediate ancestor of bonobos. However, the many similarities between chimps and gorillas strongly suggests that this ancestor was chimp-like, and hence Wrangham uses chimps as a stand-in for this ancestor. Wrangham then argues—very persuasively in my view—that many bonobo traits are paedomorphic, and thus concludes that bonobos have undergone domestication. This is a critical proof of principle for Wrangham, as it shows that domestication can occur ‘in the wild’. If bonobos have domesticated, they have self-domesticated.
Turning to humans, Wrangham employs a similar strategy. The goal is to show that humans are juvenilized relative to their immediate ancestor. Following Chris Stringer, Wrangham refers to this ancestor as ‘Mid-Pleistocene Homo’, wishing to remain neutral on some contentious phylogenetic issues. In Wrangham’s view, we lack sufficient fossil evidence to directly compare humans and Mid-Pleistocene Homo. Hence, he takes Neanderthals as a model for the latter. This methodological assumption is not implausible, but it can be questioned. (In particular, one might worry that Neanderthal traits show peramorphism, the opposite of paedomorphism.) Wrangham attempts to buttress this assumption in a few ways. The most persuasive consideration in my view is evidence of recent selection in humans, but not in Neanderthals or Denisovans, for a few genes known to be involved in domestication. Wrangham concludes that the rise of Homo sapiens is explained by self-domestication, an event he dates to around 300 kya based on fossil evidence.
Wrangham then moves on to humans’ propensity for reactive aggression (Chapters 7–12). At this point, the execution hypothesis becomes the central organizing theme. First sketched by Darwin, and recently much developed by Christopher Boehm (, ), this hypothesis holds that humans’ prosocial qualities are the result of an evolutionary history of lethal within-group violence directed towards antisocial individuals. This violence is assumed to have been carried out primarily by groups of cooperating males. By working together, six or seven adult males could dispatch an antisocial individual with little or no risk to themselves. Wrangham refers to such aggression as ‘coalitionary proactive aggression’.
The execution hypothesis is attractive. For starters, it makes no appeal to Pleistocene-era group selection. It is an individual-selection account of human prosocial tendencies. Second, it fills in what is arguably a gap in reputational and partner-choice accounts. Bullies must have been a problem early in human evolution. A physically imposing bully need not care what his groupmates think, and may simply take the hard-won spoils of their cooperation. Bullies—at least the most extreme ones—are a job for coalitionary proactive aggression. Third, the hypothesis plausibly explains certain core features of our moral psychology. We are equipped with biases (inaction bias, side effect bias, noncontact bias) that work to maintain plausible deniability in the face of accusations of immoral conduct from our peers. There are still other attractive qualities.
Wrangham ties coalitionary proactive aggression to human language. On his view, language was necessary before male coalitions could target antisocial group members with enough regularity to select for reduced reactive aggression. An antisocial male may well have allies in the group who are prepared to fight by his side. Hence, it is critical that conspirators can check the intentions of others, and ideally promote a consensus before launching an attack. This is a job for language. Relatedly, language is presumably necessary for a coalition of individuals to collectively plan an attack on a particular individual in the first place (as opposed to opportunistically attacking him or her).
In short, then, language enhanced our capacity for proactive aggression, which in turn led to selection against reactive aggression, which in turn led to the emergence of self-domestication with its associated syndrome of traits. We are so docile (lacking in reactive aggression) because we are so aggressive (in the proactive sense).
There is much in this picture that strikes me as plausible, but I see two interrelated issues. As I understand him, Wrangham thinks cooperation and cultural learning would have been significantly constrained in our line until there was a marked reduction in reactive aggression. This is indeed plausible. However, Wrangham also claims that the trend towards reduced reactive aggression only got underway around 300 kya. This makes a strong prediction: we should not find evidence of extensive cooperation and cultural learning prior to this period. The archaeological record tells a different story.
Wrangham himself relates the complexity of forager life in the Middle Pleistocene. For example, he says that Gesher Benot Ya’aqov—an open-air site in Israel dating to around 780 kya—‘indicates a sophisticated system of hunting and gathering’ (p. 117). He continues:
[…] these Homo ate dozens of different kinds of plants depending on the season, including seeds, fruits, nuts, vegetables, and water plants. They used fire so continuously throughout the period of habitation that they could apparently make it at will. Combined with evidence of butchery, fire use suggests that the camp would have routinely smelled of delicious roasting meat, often from deer but including animals as large as elephants. People also knapped rocks for various uses. Among various tools were sharp-edged cleavers, scrapers, and small flint flakes that they likely hafted onto spears. They brought thin slabs of basalt to camp to use as stone ‘boards’, apparently for cracking nuts or pounding meat. The preparation of some foods, such as prickly water-lily nuts, would have been demanding. Based on the way people harvest and prepare the same species today, the Mid-Pleistocene Homo would have had to dive under water to gather the nuts, dry them, roast them, and perhaps pop them. These were well-organized foragers. (p. 117)
This is not a portrait of life plagued by chimp-like reactive aggression. The practice of central-place foraging, a wide resource portfolio, continual use of fire (and hence continual sourcing of fuel), ignition technology, big-game hunting, specialized kit, composite tool technology, multi-step food-processing procedures—all of this implies a regime of complex cooperation, coordination, and social learning. Not as complex as that practiced by Homo sapiens, presumably, but still far more complex than anything we find in other great apes. This is very hard to square with the idea that these Homo had not yet found a way to escape the tyranny of antisocial group members. In addition, the activity of these Homo strongly suggests the possession of a much-upgraded system of communication. One could not manage the cooperation and coordination problems solved by these individuals with great ape-like communicative capacities, nor could one transmit the extensive natural history knowledge they must have possessed.
This brings us to the other issue: the omission of any serious discussion of protolanguage. In fairness, Wrangham does sometimes speak of language as admitting of degrees. For example, in discussing Neanderthals, he suggests that they ‘did not have language as we know it’ (p. 164) and that ‘linguistic ability improved substantially in Homosapiens compared to all other Homo’ (p. 164). But at the same time, he approvingly cites the work of Ian Tattersall, who holds that language arose suddenly in Homo sapiens: ‘As the paleoanthropologist Ian Tattersall noted, the algorithmic basis of language appears to be rather simple, which suggests that its invention was “more or less instantaneous”’ (p. 167). In any case, the important point is this: if, as many theorists think, language evolved gradually over the last 1.5 mya or so, then communication about antisocial individuals is likely much older than Wrangham supposes. (As an ardent defender of protolanguage as an important precursor to language, that would be my view.) Of course, one might hold this and still think that such communicative abilities reached unprecedented levels of sophistication in Homo sapiens.
Here, then, are two alternative scenarios to consider: One is that self-domestication was a much more gradual process. More specifically, one might think that selection against reactive aggression began in the Late Early Pleistocene. This would predict paedomorphic traits in Mid-Pleistocene Homo relative to earlier species of hominin (australopiths, habilines). The reduced craniofacial anatomy, tooth size, and sexual dimorphism of erectines and heidelbergensians relative to older hominins fits this pattern, as do their longer childhoods. (On the other hand, there is a dramatic increase in hominin brain size over this period.) This trend reached its pinnacle in Homosapiens, driven by increases in linguistic sophistication and likely other factors (for example, increasing social complexity).
Alternatively, one might imagine that enhanced self-control was an important waystation on the road to self-domestication. The hunting, foraging, and technological practices of Mid-Pleistocene Homo strongly suggest an upgrade in executive control. Enhanced self-control would have been part of that. To give just one example: as Travis Rayne Pickering () has emphasized, success at ambush hunting using stabbing spears would have required considerable emotional control. In particular, hunters would have had to sit quietly for extended periods of time at very close proximity (Sterelny [forthcoming]). Thus, we might imagine an initial stage in which Mid-Pleistocene Homo became better at controlling the ‘hot’ emotions that accompany reactive aggression. Then, as coalitionary proactive aggression increased in scope and efficiency, there was selection for decreased emotional sensitivity (for not having these ‘hot’ emotions so easily provoked in the first place), thereby easing the burden on control. Only with the latter did a positive trend towards self-domestication begin. In Homosapiens, coalitionary proactive aggression became so powerful that members of the species became pronouncedly juvenilized.
There is much more I would like to say on these and other topics related to the book, but I will have to leave at that. The Goodness Paradox is a wonderful book, and is highly recommended for anyone interested in human evolution. It is persuasive on many points, and where it is not, it serves as a productive catalyst for further theorizing.
Ronald J. Planner
Australian National University
Boehm, C. and Boehm, C. : Hierarchy in the Forest: The Evolution of Egalitarian Behavior, Cambridge, MA: Harvard University Press.
Boehm, C. : Moral Origins: The Evolution of Virtue, Altruism, and Shame, New York: Basic Books.
Pickering, T. R. : Rough and Tumble: Aggression, Hunting, and Human Evolution, Berkeley, CA: University of California Press.
Sterelny, K. [forthcoming]: Culture and Cooperation in the Human Lineage, Oxford: Oxford University Press.
Wrangham, R. W. and Peterson, D. : Demonic Males: Apes and the Origins of Human Violence, Boston, MA: Houghton Mifflin Harcourt.
Wrangham, R. : Catching Fire: How Cooking Made Us Human, New York: Basic Books.
 Neural crest cells are one of the four main types of cell that form early in development in chordates. They migrate throughout the body, forming a variety of morphological structures.
 Since chimps are more closely related to bonobos than they are to gorillas, any similarity between chimps and gorillas that is absent in bonobos either represents an ancestral feature that has been lost in bonobos or a new trait that has evolved independently in both chimps and gorillas. The former is taken as more plausible on grounds of parsimony, though this is not the only reason for thinking that bonobos are the outlier.
 Though here reputation and partner-choice accounts are also explanatory. They might all form an explanatory package.
 Of course, this can be important in between-group violence as well. But in general males will have little to no incentive to cooperate against a male from their own group. A male may lose his nerve in a between-group contest, but that is not the same thing as ‘taking up arms’ against a groupmate.
 This refers to bringing food back to a ‘home base’ before consuming it, and suggests sharing. Apes, by contrast, tend to consume their food on the go.