Our recent paper on lethal aggression (Wilson et al., 2014) has attracted lots of attention, most of it positive. Not surprisingly, though, some critical responses have also emerged. Copied below is our response to one of these critiques.

John Horgan at Scientific American devoted a blog post to critiquing the paper — though this critique is not really focused on the paper, but on whether the paper’s findings support what Horgan calls the “deep roots theory of war.” Horgan later published a critique by Brian Ferguson, the main advocate of the Human Impacts Hypothesis. In introducing this critique, Horgan offered to post responses from any of the original paper’s authors. We therefore have written a response, which is now posted here. In case it might be useful, I’ve copied it here as well:

Human impacts are neither necessary nor sufficient to explain chimpanzee violence (or bonobo non-violence)

Michael L. Wilson, Christopher Boesch, Takeshi Furuichi, Ian C. Gilby, Chie Hashimoto, Catherine Hobaiter, Gottfried Hohmann, Kathelijne Koops, Tetsuro Matsuzawa, John C. Mitani, David Morgan, Martin N. Muller, Roger Mundry, Anne E. Pusey, Julia Riedel, Crickette Sanz, Anne M. Schel, Michel Waller, David P. Watts, Frances White, Roman M. Wittig, and Richard W. Wrangham[1]

In response to our recent paper (Wilson et al., 2014), Brian Ferguson (2014) critiques the methods we used to test whether chimpanzee violence is the result of human impacts. As Joan Silk notes in her commentary on our paper, “These results should finally put an end to the idea that lethal aggression in chimpanzees is a non-adaptive by-product of anthropogenic influences — but they will probably not be enough to convince everyone” (Silk, 2014: 321).

We expect that for the majority of primatologists, and among the wider community of animal behavior researchers, the results of our study are neither surprising nor controversial. But for those hostile to the idea that human violence relates in any way to biology or adaptive behavior, the Human Impacts Hypothesis (HIH) offers an out. Violence among our ape cousins is, in this view, the result of human contact, not the result of evolution favoring aggression as a strategy. The argument closely parallels Ferguson’s earlier argument that violence in tribal societies is mainly the result of contact with outsiders, especially European imperialists (Ferguson, 1990).

In contrast to some critics, Ferguson recognizes that “there is no question that chimpanzees have the capability to make war and have done so on occasion” (Ferguson, 2011: 249). Ferguson’s critiques thus represent a departure from what we might call the “strong anti-adaptationism” of previous proponents of the HIH. For example, Power (1991) argues that chimpanzee violence is a non-adaptive response to frustration caused by restrictive feeding methods at the first long-term study sites of chimpanzees, Gombe and Mahale. Ferguson embraces Power’s hypothesis that feeding chimpanzees made them more violent, but in contrast to Power, he argues that chimpanzee violence is mainly the result of resource competition, which is exacerbated by human activities such as feeding and deforestation. This argument differs little from arguments that behavioral ecologists regularly make to explain chimpanzee violence. For example, intercommunity violence in chimpanzees is strongly associated with competition over food resources. At Kanyawara, intercommunity encounters occurred most frequently when key food species were abundant in areas bordering neighboring communities (Wilson et al., 2012). Lethal aggression is strongly associated with territorial expansion at Ngogo (Mitani et al., 2010) and Gombe (Goodall, 1986), and by expanding territory chimpanzees increase the amount of food available to themselves, their mates and offspring (Williams et al., 2004; Pusey et al., 2005).

However, in his critique, Ferguson seems to conflate “resource competition” with “disturbance brought about by the actions of people.” Resource competition is not necessarily a “disturbance”, nor does it occur only as a consequence of “disturbance.” Instead, resource competition is a routine part of existence for living things. This is one of the central premises of evolution by natural selection, and evolutionary disciplines such as behavioral ecology. Ferguson therefore appears to agree with us that violence is an adaptive strategy for resource competition. And we agree that actions of people can sometimes affect resource competition in other animals, for example by adding a valuable, concentrated resource (Wrangham, 1974), or by removing key resources such as fruit trees, increasing competition for available land (Goodall, 1977). What we disagree on is whether the evidence indicates that human impacts are the main cause of chimpanzee violence.

Ferguson argues that “human impact must be approached in historical detail.” The 30 co-authors of our paper are deeply familiar with the historical details, with many of them having been involved in these long-term sites for decades. But historical awareness does not preclude a scientific approach. Clear predictions emerge from the hypothesis that human impacts cause chimpanzees to kill. We have sought to test those predictions systematically. We also appreciate that our study sites and study populations, and the human populations that now surround those sites, have complex histories that long pre-date our research projects, and that we should be aware of this fact despite the difficulties of reconstructing these histories. However, we are also concerned with history in another sense: the evolutionary history of chimpanzees. For the great majority of this history, chimpanzees and their immediate ancestors could not have been subject to “disturbance” of the kinds that Ferguson and other proponents of the HIH invoke.

When introducing Ferguson’s critique, John Horgan writes of confirmation bias, which is indeed a concern in any scientific endeavor. To counter this bias, scientists use tools such as collecting quantitative data and using statistical methods to test which models best explain the observed data. In our study, we sought to use methods that are objective and transparent, and which would provide well-substantiated answers, whether they agreed with our prior opinions or not.

In contrast, the “holistic” approach promoted by Ferguson is vulnerable to confirmation bias. Without clearly defining variables such as disturbance, and without using quantitative data and statistical methods designed to test whether a given set of results is likely given the available sample size, efforts to compare contrasting interpretations of a given set of data risk degenerating into cherry picking and special pleading.

Ferguson argues that “[t]he three measures Wilson et al. created to test for human impact are questionable.” However, he agrees in principle with each of these measures, and offers no quantifiable alternatives.

He states that our first measure, “artificial provisioning of food, is good, where it applies, though the impact of provisioning varies by how it is carried out and other conditions of food availability.” By this Ferguson presumably refers to the argument of Power (1991) that the “restricted” provisioning at Gombe and Mahale is what frustrated chimpanzees and (in Power’s view) fundamentally changed their behavior.

We do not dispute that providing food can change the behavior of chimpanzees and other animals. Wrangham (1974) found that at Gombe, chimpanzees and baboons were more aggressive in the feeding area than when foraging for natural foods away from the feeding area. What is clear from our study, however, is that provisioning is neither necessary nor sufficient for chimpanzees to kill.

We examined data on both chimpanzees (18 communities at 10 study sites) and bonobos (4 communities at 3 study sites). Of these sites, provisioning occurred at two chimpanzee sites (Gombe and Mahale) and one bonobo site (Wamba).

Evidence for lethal aggression has been found in seven of the of eight never-provisioned study sites (Budongo, Fongoli, Goualougo, Kalinzu, Kibale, Kyambura, Taï); the only exception is the small, isolated population at Bossou. Ferguson notes that “sites marked P for provisioning cluster (4 out of 7 cases) toward the high end of the killing distribution.” Fair enough, but the two highest rates of killing are for sites where chimpanzees were never provisioned. Moreover, at Mahale and Gombe, killings have continued to be observed long after provisioning ended (in 1987 and 2000, respectively). Killings by never-provisioned chimpanzees have also been reported at shorter-term study sites not included in our study, including Loango in Gabon (Boesch et al., 2007), as well as Conkouati-Douli National Park, Congo, where male wild-born, captive chimpanzees released into the wild were attacked by resident chimpanzees (Goossens et al., 2005).

Ferguson claims that “Provisioning’s statistical association with killing, however, is diluted by two other sites,” Mahale’s K-group and Wamba. “Diluted” is an odd choice of words here; if provisioning causes chimpanzees to become violent, then presumably every community provides relevant data, rather than dilution. Excluding K-group from the analysis would involve picking and choosing, excluding any data that doesn’t fit the expectation (in other words, guaranteeing confirmation bias).

Provisioning is thus clearly not necessary for chimpanzees to kill. Nor is it sufficient for killing to occur. Killings have been observed at both of the historically provisioned chimpanzee sites, but the observed and inferred killings by the Mitumba community at Gombe occurred only after provisioning ended there (Wilson et al., 2004). (We note, though, that one infanticide is suspected to have occurred at Mitumba during the time that Mitumba chimpanzees were provisioned (Pusey et al., 2008)).

Killings have not been observed at the provisioned bonobo site (Wamba). The one suspected bonobo killing took place at a non-provisioned site (Lomako). Ferguson notes that bonobos are a different species. Of course, this is true. But if provisioning causes chimpanzees to kill, why should it not cause other species to kill, especially closely related species?

Ferguson argues that Wamba should not be included in the analysis because bonobos are a different species. Fair enough; and indeed, in Table 3, we present results focused on just chimpanzees, excluding bonobos, and found that provisioning history did not explain variation in rates of killing among chimpanzee communities. But we also note that Ferguson has previously written that violence in chimpanzees is the result of social learning, proposing that bonobos would behave like chimpanzees if they experienced similar conditions (“What would happen if a bonobo were raised among chimpanzees or vice versa? I expect their behaviors would reflect the local custom” (Ferguson 2011: 255)). Following this line of logic seems to suggest to us that exposing bonobos to the same stimulus as chimpanzees (provisioned food) should result in a similar increase in aggressive behavior. But the one suspected case of killing by bonobos occurred at the never-provisioned site of Lomako, rather than the provisioned site of Wamba.

In our view, the much less frequent occurrence of violent aggression in bonobos compared to chimpanzees raises interesting questions about the evolution of non-violence as well as violence.

We note with interest, though, that Ferguson’s argument for provisioning is profoundly different from Power’s argument. Power argued that restricted provisioning fundamentally changed the behavior of chimpanzees at Gombe and Mahale, and that a wide range of chimpanzee behaviors reported by researchers there were the result of provisioning: male dominance hierarchies, despotic alpha males, possessive sexual behavior, closed membership of social groups, territorial behavior, female dispersal, hunting of monkeys, and intergroup killings. Studies of never-provisioned chimpanzees have found that all of these behaviors occur in the absence of provisioning. We can therefore reject the idea that chimpanzee behavior is fundamentally altered by provisioning. And indeed, instead of following Power’s argument that chimpanzee violence is maladaptive, Ferguson accepts that chimpanzee violence is an adaptive component of resource competition.

As a second measure of human impacts, we examined size of protected area. Ferguson notes that “some chimpanzee groups living within large protected areas have been heavily impacted.” Fair enough. That is why we conducted multivariate analyses considering several different variables. However, size of protected area is a measure that is readily quantified, and is likely important both for chimpanzee conservation and as a general measure of the degree to which chimpanzees have been affected by human activities. Like our other measures of human disturbance, however, size of the protected area did not have a consistent effect on rates of violence.

Our third measure of human impacts was an index of disturbance, based on a method developed by Naomi Bishop and colleagues for assessing the impacts of human activities on Hanuman langur monkeys in India (Bishop et al., 1981). Ferguson agrees that these measures “work well as a general index of over-all human impact,” but complains that “they do not work as predictors of intensified violence.” Similar complaints could be made of any effort to quantify human disturbance, because what exactly qualifies as disturbance is never clearly stated. However, it is worth pointing out that the disturbance index developed we used was originally developed to address a controversy over causes of infanticide (Bishop et al., 1981).

Importantly, for our disturbance rankings, each site director ranked their own site without prior knowledge of the rankings of other researchers. These rankings thus provide an independent estimate of disturbance made by the people who best know each of these sites.

Ferguson notes that Budongo has been exposed to various forms of lumber extraction, and other chimpanzee sites have experienced “islandization.” Fair enough. As primatologists actively involved in conservation efforts, we are deeply familiar with such issues at our sites. Capturing all of these different effects in a single variable, or a single index that combines several measures of disturbance (the approach we used), can never be wholly satisfactory. But we believe that the rankings that we did for our study do correlate reasonably well with the degree to which these different sites have been affected by humans. Goualougo is clearly the least affected chimpanzee site, and Bossou the most. Kanyawara, at the edge of Kibale National Park, has a higher disturbance rating than Ngogo, at the center of the park. We do not argue that our index is perfect, but we are not aware of a better one.

Ferguson points to many different possible impacts: provisioning, habitat clearance, timber extraction, hunting, and so on. Some of these could well have an effect on chimpanzee violence, by (for example) increasing the intensity of competition for resources. This is an adaptationist argument following standard theory in behavioral ecology, and as such, is an approach that we find reasonable. It is the fact that these are reasonable questions that motivated us to conduct our study. Given the possibility that human activities can affect rates of violence in chimpanzees, it is important to investigate the extent to which the observed patterns of violence reflect human impacts. This is what we have attempted to do, using data that are quantifiable, systematic, defined using the same criteria across study sites, and using a statistical approach that allows us to test predictions from multiple contrasting models. We found that human impacts did not explain the variation in rates of lethal aggression as well as other factors. Eastern chimpanzees killed more often than western chimpanzees, which killed more often than bonobos. Communities with more males and communities living in denser populations killed more frequently than communities with fewer males and sparser populations.

It is not at all clear why chimpanzees should react this way when exposed to humans, but not bonobos, baboons, or other species. For example, baboons have been studied at Gombe almost as long as chimpanzees and have been influenced by human activity. But they have not been observed to participate in coalitionary killing. What accounts for such differences between species when exposed to precisely the same human impacts? The absence of coalitionary killing in baboons makes sense from an evolutionary perspective (Wrangham, 1999), but is inexplicable from an anti-adaptationist perspective.

Moreover, chimpanzees and humans are far from the only species to engage in lethal aggression. Fatal fighting occurs widely among animals, and includes a broad range of examples, such as fatal fights among male spiders competing for mating opportunities (Leimar et al., 1991) and the killing and consumption of male spiders by female spiders, after (or sometimes during or even before) the males have mated (e.g., Andrade, 1996). Most such fatal fighting involves fights between individuals over a highly valuable resource. In contrast, coalitionary killing occurs in fewer species, and seems mainly limited to certain social insects, social carnivores such as wolves, lions and spotted hyenas, and a few primates, including humans and chimpanzees (Wrangham, 1999). Why coalitionary killing occurs in these species but not others is explicable, in principle, using the comparative method to develop testable hypotheses. An anti-adaptationist approach promises no such explanatory power.

Whether chimpanzee violence is adaptive or not, is a question for which we do not yet have a definitive answer. Answering this question in full requires information on reproduction and information on individual participation in violence, which is available for only a few sites and which has not yet been analyzed. Additionally, chimpanzees (like humans and other animals) may sometimes make mistakes, participating in killings that result in fitness (i.e. reproductive) costs. Whether a given behavioral strategy is adaptive depends on average effects of traits. Given these caveats, previous studies provide evidence in support of the view that chimpanzee violence provides fitness benefits to the attackers. Mitani et al. (2010) found that the intergroup killings by the Ngogo community were associated with substantial territorial expansion in the area where disproportionately many of the killings had taken place. Studies at Gombe provide evidence that larger territories provide important fitness benefits, including more food, as indicated by heavier individual body weights, controlling for age and reproductive condition (Pusey et al., 2005) and shorter inter-birth intervals for females (Williams et al., 2004). Males who enlarge their territory thus provide more food for their mates and offspring, enabling faster reproduction, and thus greater reproductive success for the aggressors.

In his critique, Ferguson mostly ignores the second focus of our paper: the pattern of attackers and victims. We found that most participants (92%) were male, as were most victims (73%). Most victims were members of other communities (63%). Intercommunity killings generally involved gang attacks, in which attackers outnumbered victims by a factor of 8:1. These patterns make sense when seen as adaptive strategies. Male chimpanzees defend group territories; eliminating members of rival communities enables males to increase the amount of food available to themselves, their mates, and offspring (Williams et al, 2004; Pusey et al., 2005; Mitani et al., 2010). Chimpanzees prefer to attack when the odds are in their favor (Wrangham 1999; Wilson et al., 2012). Viewing these behaviors solely as a non-adaptive response to human disturbance provides no insights into why attacks mainly involve males attacking members of other groups when the odds are in their favor.

The question of what, if anything, chimpanzee violence has to do with human warfare is one we did not address in our paper. We expect that among our 30 co-authors some diversity of opinion exists on this topic. We would all agree, though, that definitive claims about human behavior need to be based on data from humans.

Nonetheless, there are some important things we can learn from chimpanzee studies. Our study examines lethal aggression broadly, including infanticide and within-community violence. Despite this, the criticisms of Ferguson (as well as John Horgan’s earlier post) focus mainly on intercommunity killing, and its relevance to studies of human warfare. Ferguson, Horgan, and many others argue that warfare has a relatively recent origin (within the past 10,000 years (Ferguson, 2003)), due to some relatively new phenomenon, such as agriculture, or settled societies, or food storage, or property rights, or ideology, or new kinds of weapons. Chimpanzees have none of these things. They do sometimes use weapons (sticks and stones) but they don’t generally use them to kill each other. So the documentation of warlike behavior in chimpanzees shows that similar behavior could have occurred in humans long before the origin of agriculture and other evolutionarily recent innovations. It also raises the intriguing possibility that humans and chimpanzees share similar patterns of violence due to our shared evolutionary history; we may have inherited these patterns of behavior from our common ancestor.

As many have noted, however, and as we fully recognize, the existence of bonobos, with their much less violent societies, highlights the need to be cautious in how much we infer along these lines. It is possible that the lineages leading to humans and chimpanzees have both become more violent, or that the lineage leading to bonobos has become more peaceful over evolutionary time. We don’t yet know the answer to this question.

We heartily agree with Silk’s point that “Humans are not destined to be warlike because chimpanzees sometimes kill their neighbours” (Silk, 2014: 322). Variation in rates of warfare among countries today, and across historical time, clearly show that people can develop institutions and mechanisms that reduce the frequency and severity of warfare (Gat, 2006; Pinker, 2011). Chimpanzee communities also vary considerably in their rates of intercommunity violence. As we have found, this variation is better explained by differences among species, populations, and demography than by human impacts (Wilson et al., 2014).

References

Andrade, M. C. B. (1996). “Sexual selection for male sacrifice in the Australian redback spider.” Science 271(5245): 70-72.

Bishop, N., S. B. Hrdy, J. Teas and J. Moore (1981). “Measures of human influence in habitats of South Asian monkeys.” International Journal of Primatology 2(2): 153-167.

Boesch, C., J. Head, N. Tagg, M. Arandjelovic, L. Vigilant and M. M. Robbins (2007). “Fatal chimpanzee attack in Loango National Park, Gabon.” International Journal of Primatology 28: 1025-1034.

Ferguson, R. B. (1990). “Blood of the Leviathan: Western contact and warfare in Amazonia.” American Ethnologist 17(2): 237-257.

Ferguson, R. B. (2003). “The birth of war.” Natural History 112(6): 28-35.

Ferguson, R. B. (2011). Born to Live: Challenging Killer Myths. Origins of Altruism and Cooperation. R. W. Sussman and C. R. Cloninger: 249-270.

Ferguson, R. B. (2014). “Anthropologist Brian Ferguson challenges claim that chimp violence is adaptive.” Retrieved 19 September 2014, from http://blogs.scientificamerican.com/cross-check/2014/09/18/anthropologist-brian-ferguson-challenges-claim-that-chimp-violence-is-adaptive/.

Gat, A. (2006). War in Human Civilization. Oxford, Oxford University Press.

Goodall, J. (1977). “Infant killing and cannibalism in free-living chimpanzees.” Folia Primatol 22: 259-282.

Goodall, J. (1986). The Chimpanzees of Gombe: Patterns of Behavior. Cambridge, Massachusetts, Belknap Press.

Goossens, B., J. M. Setchell, E. Tchidongo, E. Dilambaka, C. Vidal, M. Ancrenaz and A. Jamart (2005). “Survival, interactions with wild conspecifics and reproduction in wild-born orphan chimpanzees following release into Conkouati-Douli National Park, Republic of Congo.” Biological Conservation 123: 461-475.

Leimar, O., S. Austad and M. Enquist (1991). “A test of the sequential assessment game: fighting in the bowl and doily spider Frontinella pyramitela.” Evolution 45(4): 862-874.

Mitani, J. C., D. P. Watts and S. J. Amsler (2010). “Lethal intergroup aggression leads to territorial expansion in wild chimpanzees.” Current Biology 20(12): R507-R508.

Pinker, S. (2011). The Better Angels of Our Nature: Why Violence Has Declined, Viking.

Power, M. (1991). The Egalitarians—Human and Chimpanzee: An Anthropological View of Social Organization. Cambridge, Cambridge University Press.

Pusey, A. E., G. W. Oehlert, J. M. Williams and J. Goodall (2005). “The influence of ecological and social factors on body mass of wild chimpanzees.” International Journal of Primatology 26: 3-31.

Pusey, A. E., C. Murray, W. R. Wallauer, M. L. Wilson, E. Wroblewski and J. Goodall (2008). “Severe aggression among female chimpanzees at Gombe National Park, Tanzania.” International Journal of Primatology 29(4): 949-973. get pdf

Silk, J. B. (2014). “Animal behaviour: The evolutionary roots of lethal conflict.” Nature 513(7518): 321-322.

Williams, J. M., G. Oehlert, J. Carlis and A. E. Pusey (2004). “Why do male chimpanzees defend a group range? Reassessing male territoriality.” Animal Behaviour 68(3): 523-532.

Wilson, M. L. (2012). Long-term studies of the Gombe chimpanzees. Long-term Field Studies of Primates. P. Kappeler and D. P. Watts. Heidelberg, Springer-Verlag: 357-384. get pdf

Wilson, M. L., C. Boesch, B. Fruth, T. Furuichi, I. C. Gilby, C. Hashimoto, C. Hobaiter, G. Hohman, N. Itoh, K. Koops, J. Lloyd, T. Matsuzawa, J. C. Mitani, D. C. Mjungu, D. Morgan, R. Mundry, M. N. Muller, M. Nakamura, J. D. Pruetz, A. E. Pusey, J. Riedel, C. Sanz, A. M. Schel, N. Simmons, M. Waller, D. P. Watts, F. J. White, R. M. Wittig, K. Zuberbühler and R. W. Wrangham (2014). “Lethal aggression in Pan is better explained by adaptive strategies than human impacts.” Nature 513: 414-417.

Wilson, M. L., S. M. Kahlenberg, M. T. Wells and R. W. Wrangham (2012). “Ecological and social factors affect the occurrence and outcomes of intergroup encounters in chimpanzees.” Animal Behaviour 83(1): 277-291. get pdf

Wilson, M. L., W. Wallauer and A. E. Pusey (2004). “New cases of intergroup violence among chimpanzees in Gombe National Park, Tanzania.” International Journal of Primatology 25(3): 523-549. get pdf

Wrangham, R. (1974). “Artificial feeding of chimpanzees and baboons in their natural habitat.” Animal Behaviour 22: 83-93.

Wrangham, R. W. (1999). “The evolution of coalitionary killing.” Yearbook of Physical Anthropology 42: 1-30.

[1] We invited all co-authors of our original paper to contribute to this response. All of those who responded to our invitation provided feedback and asked to be included as co-authors. Most (or perhaps all) of those who have not yet responded to this request are currently in the field with limited access to email. Not being included on this list, therefore, does not necessarily imply any disagreement with the contents of this response.

Are chimpanzees naturally violent? Or is chimpanzee violence the result of human interference, such as artificial feeding or habitat loss? Along with 29 co-authors, I  examine this question in a paper published this week in Nature.

Jane Goodall writes in her magnum opus, The Chimpanzees of Gombe:

Early field studies of chimpanzees (including my own) gave rise to the myth of the gentle, peace-loving ape. As more data on chimpanzee behavior have been collected over the years, this myth has gradually been dispelled. (Goodall 1986: 313)

Despite observations of violence in chimpanzees by Goodall and many others, a few  people still cling to the myth that chimpanzees are peace-loving apes at heart, moved to violence only by human impacts. This view was most fully developed by Margaret Power in her 1991 book, The Egalitarians—Human and Chimpanzee.

As far as I know, Power never studied chimpanzees in the wild. Instead, she based her arguments on reading the literature, especially Goodall’s early work, and also work by other people who had conducted shorter studies of chimpanzees at other sites, such as Vernon and Frankie Reynolds’ study of chimpanzees at Budongo Forest, Uganda, and Michael Ghiglieri’s study of chimpanzees at Ngogo, in Kibale Forest, also in Uganda.

Power distinguished “naturalistic” studies of chimpanzees from “provisioning” studies, in which chimpanzees were given food by researchers. This distinction follows a long tradition of researchers who have argued that Gombe chimpanzees were no longer truly “wild” after Goodall started feeding them bananas (e.g., Reynolds, 1975).

Power noted that in a 1974 paper,  Richard Wrangham showed that Gombe chimpanzees behaved more aggressively in “camp,” where researches fed them bananas,  than in the rest of their range, where they fed on naturally occurring foods. While Wrangham explained this change in behavior as a natural response to competing over an especially rich and concentrated source of food, Power developed a more elaborate argument based on psychological frustration theory, arguing that chimpanzee behavior was fundamentally changed by frustrations encountered at the feeding station.

Power argued that all sorts of chimpanzee behavior described by Goodall and colleagues, such as territorial behavior, dominance hierarchies, intense competition for mating opportunities, bullying by the alpha male, hunting of monkeys, and lethal aggression, were not natural behavior, but were instead the result of frustration caused by restrictive feeding of chimpanzees. She made the same arguments for Mahale, the study site established by Toshisada Nishida shortly after Goodall began her studies at Gombe. Nishida and his team used sugar cane to attract chimpanzees to an observation area. Power argued that this fundamentally changed their behavior, just as it had for Gombe chimpanzees. Only the earlier, “naturalistic” observations of  chimpanzees could be trusted. Power discounts all later observations from these sites, even though at both Gombe and Mahale, as chimpanzees became better habituated, researchers increasingly followed chimpanzees throughout their forest range, rather than focusing on observations at the feeding stations.

When Power published her book in 1991, chimpanzees had been studied at many different sites across Africa, but understanding of chimpanzee behavior in the wild was still very much dominated by studies from Gombe and Mahale. Nonetheless, new long-term studies were already underway at places including Taï Forest in Côte d’Ivoire and Kibale Forest in Uganda. These studies have continued, and new studies have been established at other sites. Researchers eventually stopped feeding chimpanzees at both Mahale and Gombe, and none of the newer study sites used artificial feeding to observe chimpanzees. Many of the new sites were in large, relatively undisturbed protected areas. And yet chimpanzees at all these sites demonstrated patterns of behavior that Power argued were the result of provisioning, including dominance hierarchies, bullying by alpha males, intense competition among males for mating opportunities, hunting of monkeys, and territorial behavior. These observations soundly refuted Power’s hypothesis that the behavior of provisioned chimpanzees was fundamentally different from that of unprovisioned chimpanzees.

Nonetheless, when I started studying chimpanzees in the mid-1990s, the number of detailed observations of lethal aggression in chimpanzees was still small. The most detailed accounts of killing were those from Gombe and Mahale. It seemed at the time reasonable to wonder whether those killings were the result of something unusual about those sites, such as the artificial feeding that occurred there. Or perhaps something else was responsible, such as the ecology of these sites, both located near the southeastern limits of the range of the species.

When I started graduate school, I was mainly interested  in language evolution. I wanted to do playback experiments with chimpanzees in order to test whether they had symbolic communication, like Dorothy Cheney and Robert Seyfarth had found with vervet monkeys. I went to Harvard to work with Marc Hauser, who had been a student of Cheney and Seyfarth, and Richard Wrangham, who had established a new long-term study of the Kanyawara community in Kibale.

Although my main focus was language evolution and communication, I was also inspired by two papers published in Current Anthropology:  “The human community as a primate society,” by Lars Rodseth, Richard Wrangham, and Barb Smuts, and “Intergroup aggression in chimpanzees and humans,” by Joe Manson and Richard Wrangham. The approach of these papers, seeking to explain human behavior through comparative study of other primates, seemed exactly the sort of thing we should be doing to gain a proper understanding of our species.

In 1996, I started my first field season at Kanyawara, doing a pilot study of playback experiments. These turned out to be the first successful playback experiments to wild chimpanzees — and they also ended up shifting my focus from language evolution to intergroup aggression.

We knew it would be hard to do playback experiments with chimpanzees. They are smart, and fast. We worried that playing soft calls at close range wouldn’t work easily, as they would find the speaker and catch on that something wasn’t right. So we needed to work with loud calls that would let us set up the speaker far away from the chimps. We also needed to focus on an experimental question that could be answered with relatively few trials, since we figured chimpanzees would habituate quickly to the experimental situation if we did the same thing over and over. So we settled on simulating intergroup events, playing a single pant-hoot call from a male stranger, using calls that John Mitani had recorded from chimpanzees at Mahale. This would enable us to test whether chimpanzees could assess the relative numbers of their opponents, much as Karen McComb, Craig Packer and Anne Pusey had recently shown with lions (McComb et al., 1994). As a result of choosing to do these particular experiments, I ended up shaping the rest of my dissertation research around questions of intergroup aggression.

In the fall after I had finished my first round of experiments at Kibale, Richard Wrangham and Dale Peterson published Demonic Males. This book inspired a closer look at intergroup aggression in chimpanzees. It also attracted criticism, particularly I think from people responding to the title of the book, rather than its contents. Among the more vocal critics have been Bob Sussman (1999, 2013) and Brian Ferguson (2011). Both Sussman and Ferguson have resurrected Margaret Power’s arguments that chimpanzee violence is not natural, but somehow the fault of humans.

At the same time, the evidence for chimpanzee violence continued to accumulate, not just at Gombe, but at sites across Africa. Many cases of violence have been reported from sites that were never provisioned. Critics such as Sussman and Ferguson have therefore shifted the focus away from provisioning and more towards other forms of disturbance: habitat loss, deaths from poaching and disease, and so forth.

In 2001, I finished my PhD and started working as a post-doc with Anne Pusey at the University of Minnesota. My main goal as a post-doc with Anne was to look at intergroup aggression data from Gombe, one of the very few sites where neighboring habituated communities could be studied. This provided a rare opportunity to examine intergroup aggression from both sides of the interaction.

Working at Gombe also increased my awareness of issues related to human impacts.  I ended up spending three years based at Gombe full-time, working for the Jane Goodall Institute. Conservation issues are important at every ape research site, but are particularly prominent at Gombe, given that it is a relatively small park exposed to substantial human impacts, especially deforestation outside the park. And throughout the time I have been working at Gombe, chimpanzees have, from time to time, attacked and killed one another.

In thinking about human impacts and chimpanzee violence, there are really two major issues to consider. One is whether human impacts, such as provisioning and habitat destruction, affect rates of violence. The other question is whether violence is mainly adaptive behavior or not. For example, Wrangham (1974) clearly showed that rates of aggression were higher at the feeding station than in the forest. But he argued this was a natural, adaptive response to a highly concentrated, high quality food source. Individuals who competed aggressively for bananas would get to feast on soft, easily digested fruits, rich in sugar and starch. Individuals who stayed out of the fray would go hungry, or have to go searching long distances in the forest for natural foods.

Likewise, during the decades when forests adjacent to Gombe were being cleared for farmland, it seems entirely plausible that this could lead to higher rates of violence, as chimpanzees retreated into the remaining protected area inside the park. If the number of chimpanzees in the area stayed the same, but the available habitat shrank, this could lead to increased aggression. In this case, though, increased aggression might well be a strategy by which individuals increased their reproductive success. Individuals that simply retreated, rather than defending their land, would be forced out of the good areas, while the aggressive victors would enjoy the spoils.

Critics such as Sussman and Ferguson seem mainly interested in arguing that aggressive behavior is maladaptive. They don’t like the idea that aggressors might benefit from violent behavior, and seem mainly worried about the consequences of such arguments. For example, if we argue that violent behavior is favored by natural selection, does that mean that we must then excuse violent behavior, and accept it as natural?

I think that such concerns are unjustified, however. Just because something is “natural” doesn’t mean it is desirable, or inevitable. Smallpox virus is natural, but deeply undesirable, and humans have intentionally and with great effort eradicated this virus. We have likewise made great strides towards reducing rates of warfare and other violence.

Nonetheless, whether chimpanzee violence is natural, or the result of human impacts, is an important question to get right. I have spent a number of years trying to answer it. I first presented a version of this study in 2004, in a talk at the International Society for Research on Aggression meetings in Santorini, Greece, titled: “Is chimpanzee intergroup violence the result of human disturbance?”

In this earlier effort, based mainly on the published literature, I didn’t find any strong link between human impacts and chimpanzee violence. By this time killings had been reported from several sites that had never been provisioned, including Ngogo and Kanyawara in Kibale Forest, Uganda, and Budongo, also in Uganda. But I had other things on my plate, and ended up putting this study on the back burner for several years.

It wasn’t until 2011 that I started working on the project again in earnest. Once I got started I realized that to really get a handle  on this question, published data wouldn’t be enough. I wanted to make sure that the data from each site was accurate, and that variables demographic and ranging data were all coded the same way. I wanted independent ratings of disturbance from each site. And I knew from my own experience that it can take years from the time that killings are observed to when detailed descriptions are written up and published. A study using just the published cases would result in a potentially severe underestimation of rates of violence. So I started contacting researchers at other sites and asked if they would be interested in participating. In the end the study included all the main long-term studies of chimpanzees and bonobos, and a long list of co-authors.

The main take home message of our study is that chimpanzee violence is natural behavior, not the result of human impacts. We have two main lines of evidence for this.

First, if we look across study sites, the degree of human impacts doesn’t explain the variation in rates of killing that we see. The site with the highest rate of killing, Ngogo, is in a forest with relatively low human impacts, and these chimpanzees were never artificially fed by researchers. In contrast, the site with the highest human disturbance rating, Bossou, is a site where killing has never been observed, despite many years of observation. Instead, overall killing rates are better explained by differences among species (chimpanzees kill more often than bonobos), and differences in demography (groups with more males, and that live at higher population densities, have more killing). Moreover, high population density appears to reflect good habitat quality, rather than human disturbance.

Second, if we look at the detailed patterns of who is killing whom, we see patterns that make sense from an evolutionary viewpoint, but which are hard to explain otherwise. Attackers did not kill at random. Instead, they mainly killed members of other groups (63% of killings). They mainly killed when they had an overwhelming numerical advantage (median 8:1 ratio of attackers to victims in intergroup killings). Attackers were much more often male than female (92% of participants in attacks) and they mainly killed males (73% of victims). They mainly killed when it was easy to kill victims, either because of a strong numerical advantage, or because the victim was weak (such as infants).

People often ask what the implications of this study are for human
behavior. I would say that definitive claims about human behavior need to be based on data from humans. But there are some important things we can learn from chimpanzee studies. One is that we can get much more detail on the contexts of killings in chimpanzees than is normally possible in human studies. We can watch them do everything in their daily lives, including killing — something we can’t do easily, ethically, or legally with humans. So we can collect lots of data that is useful for testing hypotheses about the biology of violence that apply to humans as well as other species. Another thing we have learned relates especially to the origins of warfare. Some people argue that warfare has a recent origin, due to some relatively new phenomenon, such as agriculture, or settled societies, or food storage, or property rights, or ideology, or new kinds of weapons, and so on. Chimpanzees have none of these things. They do sometimes use weapons (sticks and stones) but they don’t generally use them to kill each other. So the documentation of warlike behavior in chimpanzees shows that similar behavior could have occurred in humans long before the origin of agriculture and other evolutionarily recent innovations. It also raises the intriguing possibility that humans and chimpanzees share similar patterns of violence due to our shared evolutionary history; we may have inherited these patterns of behavior from our common ancestor.

The existence of bonobos, however, with their much less violent societies, highlights the need to be cautious in how much we infer along these lines. It is possible that the lineages leading to humans and chimpanzees have both become more violent, or that the lineage leading to bonobos has become more peaceful over evolutionary time. We don’t yet know the answer to this question.

References

Ferguson, R. B. (2011). “Born to Live: Challenging Killer Myths.” Origins of Altruism and Cooperation. R. W. Sussman and C. R. Cloninger, Eds. Springer New York. 36: 249-270.

Goodall, J. (1986). The Chimpanzees of Gombe: Patterns of Behavior. Cambridge, Massachusetts, Belknap Press.

Manson, J. H. and R. W. Wrangham (1991). “Intergroup aggression in chimpanzees and humans.” Current Anthropology 32(4): 369-390.

McComb, K., C. Packer and A. Pusey (1994). “Roaring and numerical assessment in contests between groups of female lions, Panthera leo.” Animal Behaviour 47: 379-387.

Power, M. (1991). The Egalitarians—Human and Chimpanzee: An Anthropological View of Social Organization. Cambridge, Cambridge University Press.

Reynolds, V. (1975). “How wild are Gombe chimpanzees.” Man 10(1): 123-125.

Rodseth, L., R. W. Wrangham, A. M. Harrigan and B. B. Smuts (1991). “The human community as a primate society.” Current Anthropology 32(3): 221-254.

Sussman, R. W. (1999). “The myth of man the hunter, man the killer and the evolution of human morality (evolutionary and religious perspectives on morality).” Zygon 34(3): 453-472.

Sussman, R. W. (2013). Why the legend of the killer ape never dies: The enduring power of cultural beliefs to distort our view of human nature. War, Peace, and Human Nature: The Convergence of Evolutionary and Cultural Views. D. P. Fry. Oxford, England, Oxford University Press: 97-111.

Wilson, M. L., R. W. Wrangham and A. E. Pusey (2004). “Is chimpanzee intergroup violence the result of human disturbance?” XVI World Meeting of the International Society for Research on Aggression, Fira, Santorini, Greece.

Wilson, M. L., C. Boesch, B. Fruth, T. Furuichi, I. C. Gilby, C. Hashimoto, C. Hobaiter, G. Hohman, N. Itoh, K. Koops, J. Lloyd, T. Matsuzawa, J. C. Mitani, D. C. Mjungu, D. Morgan, R. Mundry, M. N. Muller, M. Nakamura, J. D. Pruetz, A. E. Pusey, J. Riedel, C. Sanz, A. M. Schel, N. Simmons, M. Waller, D. P. Watts, F. J. White, R. M. Wittig, K. Zuberbühler and R. W. Wrangham (2014). “Lethal aggression in Pan is better explained by adaptive strategies than human impacts.” Nature 513: 414-417.

Wrangham, R. (1974). “Artificial feeding of chimpanzees and baboons in their natural habitat.” Animal Behaviour 22: 83-93.

Wrangham, R. W. and D. Peterson (1996). Demonic Males: Apes and the Origins of Human Violence. Boston, Houghton Mifflin.

Do playing violent video games and watching violent television shows make people act violently? That was one of many questions discussed at the XXIst World Meeting of the International Society for Research on Aggression (ISRA), which met this July in Atlanta, Georgia.

ISRA was founded in 1972, “In the shadow of the Cold War, the Vietnam War, and social unrest throughout the United States.” Since that beginning, a goal of the society has been to use research to find ways to “reduce harmful aggression.” This is a goal that I share. But during the course of the meetings, I couldn’t help but wonder: after all these years of research, what have aggression researchers learned that has been of practical use to help people reduce harmful aggression in the real world?

Some of the ISRA talks focused on the neurobiology of aggression. This is a scientifically important topic, definitely worth pursuing. But I suppose that the practical benefits of such research are largely limited to a fairly narrow range of specific circumstances, such as developing drugs to help reduce violent behavior in people with certain mental illnesses. Will a better understanding of neurobiology really help to reduce rates of warfare, or civil strife, or violent crime?

Many of the other talks and posters focused on the social psychology of aggression. A particularly popular topic was the effect of violent media on aggressive behavior. The most prominent of these was the John Paul Scott Award talk by Professor Rowell Huesmann. This award “recognizes a lifetime or substantial contribution to aggression research,” so this is clearly a topic close to the heart of the society.

Professor Huesmann spoke on aggression and the media. He described aggression as a contagious disease, which is contracted through exposure to violence, whether real or fictional. Watching violence on television, or playing violent video games, gives children “scripts” about how they should act. Because people imitate, they imitate the violence they see on TV and experience in videogames and act aggressively in real life.

Professor Huesmann is an esteemed scholar with a long list of publications, honors and grants. His findings are based on studies with large sample sizes and careful statistics that show consistent, positive correlations between exposure to violent media and violent real life behavior. He also seemed like a very nice guy, earnestly devoted to reducing problems of violence in the real world.

Despite all this, I think the Media Violence Hypothesis doesn’t explain a great deal of violent behavior, even though it is intuitively appealing on several levels.

The argument that exposure to media violence causes real world violence has a certain plausibility to it. Kids imitate things they see on TV. In my high school, certain guys were constantly imitating pro wrestling moves with their buddies (though they always did so in comic slow motion). Teachers tell me they see their kids imitating moves they see on TV, such as Teenage Mutant Ninja Turtle attacks.

Another appealing part of the argument is that as video games become increasingly realistic, the violence becomes increasingly graphic and disturbing. Back in the 1990s, I played a game on a friend’s computer that must have been some version of Wolfenstein — a first person shooter that involved walking around a castle and shooting at Nazis. After playing for a while, I found the whole first person shooter aspect rather disturbing, particularly the blood pouring from the bullet wounds of the dead computer Nazis. I wouldn’t want my kids playing this game, and can imagine how spending many hours playing such games might affect a person.

So I can see how the Violent Media Hypothesis is appealing, especially to parents and educators. This hypothesis also suggests an obvious intervention: increased control over the media. If the hypothesis is true, then limiting violence in the media should provide strong practical benefits for reducing rates of violence in the real world — though of course this raises all sorts of problems for anyone interested in protecting First Amendment rights.

Despite the intuitive appeal of the hypothesis, though, and quite apart from concerns about media censorship, there are a number of reasons to think that that Media Violence Hypothesis doesn’t really hold up.

First of all, exposure to violent media isn’t a necessary precursor to violence. Neither ants nor chimpanzees nor any other animals require exposure to violent media to commit acts of violence. Numerous acts of violence occurred in human history, and prehistory, long before the invention of Pac-Man and Space Invaders.

This is, admittedly, a weak objection, given that the argument isn’t just about the media; it’s about the violence depicted in the media. Movies, TV and video games just provide novel means of displaying violence; new ways of spreading the contagion of violence. Long before violent media existed, people saw acts of violence being committed, which according to this hypothesis, would have resulted in further imitative acts of violence, which in turn would spawn yet more acts of violence.

There are, however, other reasons to think this hypothesis is implausible. One of these is my own personal experience. I don’t particularly seek out violent media. I have never played Grand Theft Auto or similar video games, and I don’t seek out slasher films. All the same, over a lifetime of watching TV and movies, and playing various kinds of games, I have been exposed to an enormous amount of simulated violence: murder, mayhem, rape, pillaging, decapitation, dismemberment, and all manner of horrible things. I’ve watched the entire population of the planet of Alderaan being destroyed by the Death Star. As a teenager playing Dungeons and Dragons, I fought countless virtual battles, throwing spears and whiskey bombs at orcs, hobgoblins and various other enemies. Playing Risk, I have commanded armies that slew many thousands of soldiers (though these were represented in the innocuous, non-bloody form of Roman numerals). I watched all six seasons of the Sopranos, in which many horrible murders were graphically depicted.

One would think that, according to the Media Violence Hypothesis, I would be deeply infected with the contagion of violence, and would have acted out these violent scripts repeatedly in my life. And yet, I have enjoyed a quite peaceful existence,  apart from a few rough wrestling bouts and fist fights in grade school.

I don’t think I’m unusual in this respect. Lots of people watch lots of violence on TV and movies, and experience simulated violence in video games, fantasy role playing games, and other forms, and yet have never committed a single act of violence.

Given all this, it would seem that violent media are neither necessary nor sufficient for violence to occur. Nor does it explain one of the most consistent findings in aggressive studies: men are consistently more violent than women. Men commit the great majority of homicides, and commit far more atrocities of war than women. If both sexes are routinely exposed to the same violent media, why should men be more violent than women?

Additionally, the Violent Media Hypothesis doesn’t  fit with patterns of violence over time, or across geographic space.

As Manuel Eisner described in his excellent plenary talk, rates of homicide have declined over the past thousand years or so in Europe. If the Media Violence Hypothesis were correct, we should see an enormous increase in violence following the invention and widespread dissemination of movies, television, and video games in the 20th Century. But what we see instead is a marked decrease in many different kinds of violence, including warfare and homicide. The horrors of WWI occurred when the film industry was still in its infancy. Television barely existed at the start of WWII. Since these cataclysmic events, the 20th Century has seen a steady decline in per capita rates of death from war, and decreases in homicide rates across Europe and North America. According to Eisner, there was an uptick in homicide in the 1960s through the early 1990s, but homicide rates declined thereafter, despite the release of a series of increasingly violent video games. (Grand Theft Auto, for example, was first released in 1997, when homicide rates in the US were on their way down.)

Considering geography, one would predict that the parts of the world most plagued by violence would be those most intensively saturated by television, films, internet and video games, such as the United States, Europe, and Japan. But while the US has higher rates of violence than Europe and Japan, its homicide rates are far below those of many countries where access to violent media is much more restricted. In many war-torn countries of Africa and the Middle East, per capita incomes are much lower and many people lack electricity, much less Xboxes.

So on a broad scale, the data don’t seem to support the Media Violence Hypothesis. It cannot explain why the latter half of the 20th Century was more peaceful than the first half, or why Japan is more peaceful than, say, Afghanistan.

The hypothesis seems even more doubtful from a theoretical perspective, especially when considered from an evolutionary perspective. In particular, why should evolution design people, or any other organism, to imitate acts of violence?

In evolutionary game theory, violence is a strategy that animals use to obtain key resources for survival and reproduction. If I am attacked by a predator, or a hostile member of my own species, I will fight back. Male chimpanzees fight for status, and for opportunities to mate with fertile females, and to defend and expand their territory.

Under many circumstances, imitating violent behavior would be maladaptive. For example, in chimpanzee societies, the alpha male frequently conducts charge displays, charging at, chasing, and sometimes hitting other community members. If a young, low-ranking male started imitating the alpha male by charging around and attacking other chimpanzees, he would get beat up in a hurry.

Professor Huesmann described studies showing that both Palestinian and Israeli children exposed to more media violence were more aggressive. Fair enough. But does anyone really think that the ongoing conflict between Israelis and Palestinians is the result of too much violent television? Or that Russia’s invasion of Crimea and ongoing support for separatists in Eastern Ukraine is the result of, say, Vladimir Putin watching too many Steven Seagal movies?

I would like to think that research on aggression can, and should, contribute to reducing harmful aggression. But I think that to find practical and effective solutions, we need to focus on the right level of analysis, guided by theoretically plausible hypotheses. Evolutionary theory provides critically important guidelines for what sorts of hypotheses are likely to be plausible. But a complete understanding of violent behavior also needs to look at a broad range of factors, including history, politics, and demography. A few disturbed people may be inspired to violent acts by violent video games, but it seems likely that the great majority of violent acts are best explained by conflicts over limited resources,  such as land, oil, and status.