A Proposal for the Evolution and Development of Free Will
Eddy Nahmias 1. Introduction This dissertation has argued that free will is constituted by a set of cognitive abilities, which allow us to introspect on our motivational states and evaluate them—to know our own minds—and to act on this knowledge. Rather than requiring a unique metaphysical ability (such as contra-causal powers), this view sees free will as coming in various degrees. And in normal adult humans these abilities are particularly well-developed (in a way that supports reactive attitudes and moral responsibility, allowing for our social and legal practices). We want to know, then, how it is that these abilities have become relatively advanced in humans, and how they become more advanced as humans grow up. We should, therefore, look for naturalistic explanations for the existence of free will: How, as a species, did we evolve free will? And how, as individuals, do we develop free will? From our vantage point, it may seem obvious that possessing the sorts of knowledge about oneself and the world, which I have associated with free will, would provide a clear selective advantage. An animal with the ability to predict its own behavior by being aware of its various desires and their possible outcomes in the world would surely increase the animal’s rate of survival and reproduction relative to conspecifics that lacked this ability. The story would go like this: Most animals react differentially to their environment, moving towards what has helped their ancestors survive and reproduce, or what has positively reinforced them in the past, and moving away from what has killed their ancestors, or what has negatively reinforced them. These motivations—genetically derived or behavioristically learned—are crucial to the animal’s survival and reproduction. However, they are based on mechanisms which drive the animal to act on whichever motivation is strongest at a given time (like Frankfurt’s wanton). But the strongest motivation is not always the most advantageous for the animal, given other factors in its environment and given its longer-term prospects (e.g. a subordinate monkey acting on his motivation to mate may never reproduce if the dominant kills him or expels him from the group). Hence, there should be selective pressures for the ability to inhibit one’s immediate desires and allow for increased flexibility of behavior—just as behavioristic learning allows more flexible responses to variable environments than genetic mechanisms. If only an animal could consider whether or not to act on its strongest motivation. One way this might occur is if the animal could represent itself in its environment, represent various actions (based on its various motivations), and represent likely outcomes of those actions. That is, an animal that could know its various motivations, imagine their likely outcomes in the world, and act accordingly would do well in comparison to an animal that just acted on its immediate desires. Basically, there should be selective pressures for the abilities I have associated with free will; persons should be able to survive and reproduce better than wantons. That’s a plausible story. But it leaves way too many questions unanswered. For instance, what particular environmental features (social and physical) would select for these representational cognitive abilities? And from what prior traits (e.g. cognitive structures) might these abilities arise? Not every potentially advantageous trait evolves (not even close); it takes the right mutations in the right selective environment. And you can’t get something from nothing; complex traits do not evolve without precursors (any more than adult human beings arise from
Theory of mind. As Darwin stated: “If no organic being excepting man had possessed any mental power. our apparent uniqueness can be explained. But it can be shewn that there is no fundamental difference of this kind” (1871: 445).1 We need to know. the ability to imitate others and respond to what they see. However. I will not offer evidence of heritability or patterns of gene flow in populations. when dealing with cognitive abilities. we need a lot more information. My proposal deals with the evolution of theory of mind in primates and its development in very young children. including developmental and comparative psychology. As with other traits. Indeed. 2.swamps). which may have possessed intermediate cognitive abilities between those displayed by modern humans and by our nearest living relatives. then we should never have been able to convince ourselves that our high faculties had been gradually developed. Agent causation seems to attribute to humans “a wholly different nature” from any other animal (not surprising given its historical connection to immaterial souls). or if his powers had been of a wholly different nature from those of the lower animals.2 But even the abilities to introspect on. offers a foundation for the introspective abilities required for free will. and the ability to predict one’s own and others’ behavior based on internal states. are extinct. 2
. We have no reason to suppose that any of the lower animals have this capacity” (1871: 88-89). I will suggest. Hence. By surveying theories and experiments from several fields. to represent motivational and perceptual states. for instance. I will provide evidence for some elements of such an explanation—primarily for an evolutionary trend suggesting selection has occurred and a description of the selective environment in which the trait is adaptive. is sometimes referred to as theory of mind. seems to contradict the claim I quote above when he writes: “A moral being is one who is capable of comparing his past and future actions or motives. An evolutionary account of free will must locate it in abilities (or “powers”) that can be traced to ones possessed by other animals. in part. there should be no inexplicable differences in kind between related species’ abilities. To provide even the beginnings of a plausible evolutionary account for a particular trait. And we need to know what primitive traits might have served as the building blocks for the derived (evolved) trait. and which features of the environment in which they evolved might have made the difference. including perhaps the ability to recognize others as self-moving agents rather than objects. the ability to inhibit one’s immediate impulses.
See Waller (1998). we will need to look for lower-level building blocks of the cognitive abilities required for full-fledged free will. which creatures have the trait in question and which of their nearest relatives don’t. The Basic Idea
See Brandon (1990: 165-176) for the criteria required for a complete adaptation explanation. Darwin. I believe we will find them in more basic representational abilities. This last ability. In the proposal below. identify with. I offer a proposal for the evolution and development of these representational cognitive abilities. because many human ancestors. I leave it to others to advance proposals for how these foundational ablities develop into full-fledged free will as children grown up in an appropriate social setting. and influence our motivations in the way I have described seem to involve significantly different powers than those possessed by other animals. recognizing the significance of such abilities. and of approving or disapproving of them.
and second. where he suggests that reciprocal altruism may have led to the evolution of cognitive abilities and increases in brain size: Given the psychological and cognitive complexity the system [of reciprocal altruism] rapidly acquires. It begins with Robert Trivers’ model for reciprocal altruism. Furthermore. do not need to be “the same” as another’s for introspection to be advantageous. and is effective in making accurate predictions about those behaviors. Astington (1993. See. For instance. my proposal turns to the growing body of research on theory of mind and deception in animals and children. Robert Premack and Guy Woodruff introduced the concept of theory of mind to describe a chimpanzee’s ability to predict how another agent would behave. its introspected conscious experience. so that the philosophical question of eliminativism can be avoided. as Humphrey points out (1983. (1971: 223) While Trivers suggests that this may be the case.”3 However. relative to body size. and cheating selects for catching cheats. specifically about the behavior of other organisms. The evolution of these cognitive abilities may also account for at least some of the increase in hominid brain size that Trivers refers to: modern humans have brains that. “there can be no morality without reciprocity. such as desires and beliefs. an animal’s own mental states. a system in which individuals exchange benefits over time. because such states are not directly observable. 2). have increased in size roughly threefold since we shared a common ancestor with chimpanzees about 5-6 million years ago. because the system can be used to make predictions. given his apparent goals: In saying that an individual has a theory of mind. Frans de Waal writes. (1978: 515) An animal that can ascribe to another agent mental states. This model has been used to support theories about the evolution of human ethical systems and moral emotions. can better predict the agent’s future behavior. I propose that a complex social environment involving reciprocal altruism selects for deception and detection of deception. few researchers have latched onto the final sentence of Triver’s original paper. for instance. we mean that the individual imputes mental states to himself and others. creating a sort of feedback loop selecting for increasingly advanced abilities to understand how one’s own and others’ mental states are predictive of future behavior. chap. This is true regardless of whether we want to say that the other animal actually has these states. one may wonder to what extent the importance of altruism in human evolution set up a selective pressure for psychological and cognitive powers which partly contributed to the large increase in hominid brain size during the Pleistocene. 36). A system of inferences of this kind is properly called a theory. and evolutionary ethics and evolutionary psychology draw heavily on these theories. which unify broad classes of behavior into patterns of beliefs and desires. in turn.4 Put simply: sharing selects for cheating. first.” Here. and both of these select for “mind-reading. it only needs to function to predict the other’s behavior. I want to suggest how it may have occurred.5 While
de Waal (1996: 136).My proposal seeks to link together ideas from several areas of recent research in primate social behavior and cognition and in child development. All that is necessary is that the theory actually serves to unify behaviors under unobservable mental attitudes. 3
5 4 3
. Sociobiology is largely based on kin altruism and reciprocal altruism.
Such deception can be contrasted with younger children’s “lies. social primates are required by the very nature of the system they create and maintain to be calculating beings. who pass falsebelief theory of mind tests. As he states: There are benefits to be gained for each individual member [of a social group] from preserving the overall structure of the group. Thus. and have found that apes deceive in more complex ways than monkeys (while prosimians do not seem to deceive at all). which should not be surprising.6 Researchers such as Richard Byrne and Andrew Whiten have gathered evidence of deception in primates. since the more complex forms of deception seem to require false-belief theory of mind. Four year olds. reflective consciousness. children in their third year develop the ability to explain and predict other’s behavior based on their motivational states (desires). such as a child’s saying “I’m tired” to avoid doing what she doesn’t want to do (the simplicity of such misrepresentation becomes evident when the child says “I’m tired” to avoid being put to bed). and at the same time from exploiting and outmanoeuvring others within it. 4
.cognitive ethologists continue theory of mind research on monkeys and apes. but it also creates opportunities to exploit conspecifics. also begin to employ intentional deception. and that this is a relatively recent step in the sequence of phylogenetic elaboration in social intellect which is reflected not only in human superiority over other contemporary apes. (1976: 19) Humphrey believes social complexity will lead to cooperation. which is essentially theory of mind. Finally.” which are based simply on what has (sometimes) worked in the past. They undertook this project to advance their theory that the complexities involved in manoeuvring in one’s social environment (rather than just dealing with one’s physical environment) created selection pressures for particular aspects of primate and human intelligence: It is suggested that it is the evolution of social intelligence which explains human brain power. children sometime after the age of four come to recognize that others’ beliefs (and their own) can be distinct from the way the world actually is and thus can be false. it has also blossomed in developmental psychology. The development of increasingly complex deception parallels these stages.
See Perner (1991: 189-200). But not until their fourth year do children demonstrate an understanding of other’s informational states (beliefs) as well. (1988: 2) Nicholas Humphrey was one of the first researchers to suggest social complexity as the most significant factor driving increased primate intelligence. to introspect our own mental states so as to predict other’s behavior. since they require understanding that others’ beliefs can be different than one’s own and hence can be made different than what one knows to be the case. Building on the abilities to imitate and make believe. where researchers such as Henry Wellman and Joseph Perner have shown that children develop the ability to represent other’s mental states in stages. and in monkeys’ superiority over prosimians. One of the outcomes of this evolutionary trend is our ability to be (folk) psychologists. but in their superiority over monkeys. Humphrey calls this introspective ability. where an individual A tries to make B believe something to be true which A knows to be false.
for reciprocal altruism to maintain itself. However. it requires a mechanism to detect these cheats and deny them reciprocity.My proposal for the evolution of human intelligence ends up in the same place as Humphrey’s. This leads to selective pressures for deceptive behavior. First of all. and should be bold in suggesting hypotheses as to the universe which science is not yet in a position to confirm or confute.7 Thus. Specifically. I am following Bertrand Russell’s vision for philosophy: that it should be “comprehensive. predicting others’ behaviors according to their desires for rewards—and perhaps thwarting others’ desires when they conflict with one’s own. Wilfrid Sellars (1956) offers a myth about the development of our abilities to understand others’ and our own mental states in terms of behavioral theories. to explain why animals might be selected to be aware of their own and other’s mental states—to have a theory of mind.”8 As a hypothesis. animals in a complex social system which involves reciprocal altruism are poised between cooperation and competition. despite the genetic and anatomical similarities between modern humans and our nearest relatives (the common and bonobo chimps). in the way we are. humans have diverged significantly in certain
There is an important debate between theory theory of mind and simulation theory of mind (see Carruthers and Smith. my proposal for the evolution and development of free will requires a number of assumptions. my goal is to explain more fully how we got there—that is. that this is a proposal. especially comparative ethology and developmental psychology. including our own. usually subtle cheating aimed at gaining more than one gives in a cooperative relationship. Indeed. My account might be seen as offering the selective pressures that would lead our ancestors (like Sellars’ character Jones) to develop a “folk psychology.” 5
. my proposal attempts to bridge together several areas of research to come up with a possible explanation for why we are able to be represent mental states. Assumptions Proposals for the evolution of cognitive abilities face several daunting problems. belief theory of mind allows an animal. however. this purposeful misrepresentation is what we generally mean by deception. Finally. The next level. one based on substantial but probative research from a wide range of fields. I should stress. Thus. though he presents it more as a cultural invention. to predict specifically how another animal. which I will discuss in section 5 below. These pressures to deceive and to detect deception create a feedback loop. His story might perhaps be seen as an evolutionary account of theory of mind. along with the emotional motivations to withhold reciprocity from and perhaps punish cheats. may attempt to satisfy its desires. given its beliefs (its perceptions and perhaps memories). From Bertrand Russell’s “Logical Atomism” (879). and this detection of deception is tied to our understanding of other’s intentions. 1996). false-belief theory of mind allows an animal to manipulate another’s beliefs so that they misrepresent reality—and perhaps to notice when one’s own beliefs are being manipulated. However. leading to increasingly complex levels of theory of mind. detection of deception would be selected. But these should always be presented as hypotheses. acting more like a cognitive psychologist.
2. The most basic level is desire theory of mind which allows an animal to be a sort of behavioral psychologist.
Furthermore.” beyond those needed to control bodily functions. Furthermore.11 difficulty of childbearing. given the significant costs of big brains (in terms of energy use. encephalization quotient (EQ) is the ratio of an animal’s actual brain size to its “expected” brain size based on its body size as compared to other species.9 The traits that remain relatively unchanged in the living species (e. we can thus assume that higher EQ is a derived trait. is which cognitive abilities were selected for and what features of the environment selected for them. makes up two percent of our body mass yet uses 18 percent of our energy. perhaps by natural selection. and the traits that changed (e. there must have been a corresponding selective advantage to having relatively larger brains. bipedalism in humans) are the derived traits that evolved. quadripedal locomotion in monkeys) represent the ancestral trait. These brain-based behaviors are “cognitive abilities. theories for increases in primate brain size and cognition fall into two camps. such as language. with a nearly threefold increase from chimpanzees through the extinct homo ancestors to modern humans. The human brain. Often. But we do have some resources.cognitive abilities. primates’ increased brain size must allow for behaviors which are advantageous in the particular selective environment in which they evolved.g. and since cognitive abilities don’t fossilize and since selective environments have changed. This involves looking at the traits of groups of related living species and “building” common ancestors based on their shared traits. energy costs translate to fitness costs. for instance. as well as brain complexity. one arguing that the increases were due to selective pressures from a complex ecological environment and the other arguing that they were due to pressures from a complex social environment. Jerison (1973) presents the most detailed defense of the use of EQ as a measure of intelligence.g. different pressures may have been involved during speciation events than during more gradual changes during the evolution of a species. 227-234). Presumably. He also argues that EQ corresponds to “extra neurons. 6
. Only the liver and intestines are more costly. thought to be the part of the brain most related to intelligence. because the increases occurred at different times in a variety of environments. EQ may also be measured more specifically to indicate the relative size of the animal’s neocortex.” The question. Drawing a sharp distinction between these two types of selective
See Byrne (1995. especially ch. One resource is the comparative method.12 Because big heads are not adaptive in themselves. we have limited resources to speculate about how our cognitive abilities evolved and why our brains got so big. which is a measure of brain size relative to body size). the brain. and incapacity of infants). One particularly relevant evolutionary trend in primates is an increase in brain size (specifically encephalization quotient [EQ]. and in the anatomical feature that underlies cognition. however. then assuming any divergent traits evolved since the living species diverged from this common ancestor. 2). Numerous extinct ancestors in the Austrolopithicus and Homo lineages separate us from the apes. increase from prosimians to monkeys to apes. Fossil evidence may help to confirm this assumption about the traits of the ancestral common ancestors. And the comparative method suggests that the relations of ancestral to derived traits is generally represented by the above ordering of families (prosimians to monkeys to apes to humans). then.10 EQ.
12 11 10 9
See Byrne (1995. It is highly unlikely that any single selective pressure drove increased EQ. This method provides a case for evolutionary trends and selection for particular traits. Technically.
thus leading to larger. ecological and social environments cannot always be clearly distinguished. and which traits are spandrels. and for correlations between EQ and group size.
It would also help the proposal if these cognitive abilities evolved in distinct clades facing similar selective pressures. such as dolphins. larger) social groups. they are spandrels. such as increased brain size and complexity.g. I will present arguments for why a particular type of environment—a complex social setting including reciprocal altruism—may have selected for particular cognitive abilities that presumably required larger (and more complex. they are made possible by a large. and that develop relatively early in humans and without significant variation across cultures. That is. but these abilities are not adaptations. there is some evidence that marine mammals. except if environmental variation itself can be described as a selective pressure (these issues may apply to explanations for the evolution of sexual reproduction as well). my proposal generally sides with the social complexity theorists and focuses on describing the aspects of the social environment that would select for particular cognitive abilities. “Foraging Behaviour and the Evolution of Primate Intelligence. literacy is currently predictive of decreased reproductive rates). My proposal attempts to avoid the problem posed by spandrels by looking for relatively low-level cognitive abilities that are found. Dolphins also perform other behaviors discussed in Figure 3. at least in rudimentary form. like reading and writing. more complex brains. for instance. in turn. such as mirror-self recognition. but especially for those involving phenotypic plasticity. selection for frugivorous foraging may have led to an initial increase in brain size (in this case. Indeed. Indeed. simply allowed by larger. this claim raises another problem for any explanation of the evolution of increased brain size and cognitive abilities. Rather. However. for mapping and remembering the locations of seasonal fruits). non-adaptive traits that arise as effects of adaptations?14 Humans use their big brains for all kinds of useful things.16 My proposal relies on another important assumption: that the evolution of these cognitive abilities parallels the development of the abilities in human children. These questions are problematic for any evolutionary explanation. see Dunbar (1988). 7
. which in turn allowed for even larger brains (because of increased available energy in the form of fructose). and which abilities were. for instance. there has been some cross-cultural research that shows similar trends in the order and timing of the onset of cognitive abilities associated with theory of mind.15 Indeed. and language use. See Gould and Lewontin (1978). they do not exist because they increased reproductive success (in fact. interconnected) brains. pretend play. It may be that the selective pressures from different aspects of the environment were mutually reinforcing. Astington (1993).13 Nonetheless.” in Byrne and Whiten (1988). have evolved theory of mind abilities and exhibit reciprocal altruism. composing music and computing quadradic equations. in primates. For instance. then pinpointing the environmental factor becomes a moot point. I will discuss some of these
See Milton. if variable environments are what selects for phenotypic plasticity. 16 See. because. suggesting convergent evolution. this social complexity may then have selected for further increases in brain size. more complex brains? Which traits evolved because they increased relative reproductive success (and hence are adaptations).pressures surely oversimplifies the story. flexible brain that was selected for other purposes. and thus perhaps for increasingly complex (e. which cognitive abilities were actually selected for.
but the cost to A is “regained” by a benefit received from B at some later time. the development of folds. Let us turn. Restak (1986). pseudoreciprocity. . (following Piaget’s reasoning) it may be that a certain order in the development of some cognitive abilities is logical and natural. not closely related. then.
Trivers (1971. “mindreading abilities are built up from simpler precursors. In child development this organization arises as the brain grows and increases in complexity (e. Reciprocal Altruism Trivers defines altruism as “behavior that benefits another organism. suffers a cost (in terms of reproductive success) by behaving in a way that benefits another animal. one point at which the development of human cognitive abilities diverges from those of apes is on tests for false-belief theory of mind.18 In the evolution of primates. whereas monkeys do not. . 277. 35). These abilities follow a similar progression in both phylogeny and ontogeny (see Figure 3 below).g. and mirror self-recognition. It may be that this improvement in theory of mind. so that evolution may have followed a similar course. to the theories and evidence supporting this account for the evolution of free will. Trivers sets up the model for reciprocal altruism: one animal. 6). or gyri and sulci. and mutualism (e. it appears that apes. See also p.g. B. which show more evidence of reciprocal altruism and deception than do other primates.”19 To explain how such altruism between unrelated animals is possible. many types of cooperative hunting).
See. Other abilities precede and are perhaps necessary for theory of mind. As Andrew Whiten remarks. Trivers’ definition has since been refined to differentiate it from byproduct beneficence. . while being apparently detrimental to the organism performing the behavior. The argument.parallels in theory of mind and deception in the following sections. pretend play. which is especially significant in deception and detection. and in development human brains do grow and become more complex (actually losing neurons while increasing synaptic connections). in the cortex). however. adult EQ is the measure used to compare phylogenetic differences. Specifically. A. increases in EQ represent not only increases in overall brain size but also in neocortex size and in complexity. such as imitation. An apparent problem for my proposal (raised by Steve Geisz) is that EQ actually decreases in infants as they develop. 4. Instead. is not that ontogeny recapitulates phylogeny (in the sense suggested by Haeckel). Furthermore. 8
. as measured by synaptic interconnectivity and folds in the cortex. for example. This is because their body grows faster than their brains. which start off disproportionately big. there may simply be logical reasons why step B precedes step C. Unlike these types of cooperative behavior. However.”17 The basic idea is that the brain must be at a certain level of organization before it is able to demonstrate certain abilities. evolved in human ancestors and represents the key to our representational abilities. also possess some levels of theory of mind. reciprocal altruism requires that the altruistic act is actually costly at the time of the act and that it is separated from the reciprocal benefit by a significant
Whiten (1991. The development and evolution of the brain may thus underlie the particular abilities leading up to theory of mind.
de Waal and Luttrell (1988). The best evidence so far is for vampire bats. and food sharing. Are there any such social structures? It is difficult to ascertain. a culturally developed system would not require irrationally strong emotions. See. which are sometimes disproportionate to the costs and benefits involved. and moralistic aggression. An analogy: byproduct beneficence is to reciprocal altruism as safe driving is to polite driving. high mutual dependence. many of which are between unrelated individuals. since the reciprocity involves regurgitating blood. exchange of various
See Taylor and McGuire (1988) and Mesterton-Gibbons and Dugatkin (1992). Bird alarm calls may not carry a cost to the caller.period of time.20 “Significant delay” is of course vague. these emotions suggest that evolution played a role in human reciprocity. such as long-lasting relationships. Trivers’ third example of humans is of course much more effective.21 The question is whether he is right that our reciprocal relations and their underlying emotions are the result of natural selection. The larger the social group (e. and especially of the corresponding emotions. ethologists must determine what counts as an altruistic act. but may make it difficult to cheat. Furthermore. the difficulty of their type of foraging may allow for group selection of reciprocity. Marine mammals may offer an example of convergent evolution following the same progression of cognitive abilities as I am describing in primates (see note 15). in order to maintain itself. The type of social structure that will most likely lead to deception and detection of deception involves long-term reciprocal relations involving an exchange of various kinds of benefits. and even if they do. humans have emotions that seem appropriate for maintaining reciprocity. city). especially baboons and apes. With vampire bats. because.g. some marine mammals. The similarities between the systems of reciprocity. He notes that all human societies have the features his model predicts. To test for reciprocal altruism. Bats certainly do not seem to demonstrate theory of mind or deception. for example. Cleaning symbiosis is better viewed as byproduct beneficence between two species which have co-evolved (like hosts and parasites). Trivers’ own examples of cleaning symbiosis and bird alarm calls do not seem to fit the refined model of reciprocal altruism. such as long-term relationships between unrelated members. and with what values. but my proposal does not claim that reciprocal altruism will necessarily select for these cognitive abilities. gratitude. This claim would be strengthened if our primate relatives demonstrated reciprocal altruism. which would suggest it evolved in our common ancestors before the explosion of human culture. I lose (a little) time perhaps hoping the favor will be returned by someone in the future. across every human society supports the claim that reciprocity has its roots in evolution. such as trust. I drive safely for my own benefit and other drivers gain the benefit of safety from my act. This also requires knowing the relatedness of the participants to differentiate it from kin altruism. but without this temporal element and the “absorption of cost” by the altruist. 9
22 21 20
. the less likely drivers will be polite? As Robert Brandon points out. not simply a cultural invention. only that it is likely. they may be accounted for by kin altruism. for instance.22 Many primate social groups show signs of reciprocal relationships. but if I let someone enter traffic at a busy intersection. Only a few examples of reciprocal altruism in animals have been well documented. and Rothstein (1988). reciprocal altruism would lose its significance for my proposal since it would not set up the intense selective pressures for cheating. Taylor and McGuire (1988). assign to the act cost and benefit values measured in terms of fitness. and some primates. and then attempt to observe when such acts are occurring. between whom.
any population which evolves altruistic behavior among non-kin could be easily invaded by cheats who receive benefits while never returning them. anthropologists often discuss the complicated reciprocal systems of human societies (e. and perhaps culminating with theory of mind. It thereby provides selective pressure for a host of cognitive abilities. Thus. reciprocal altruism offers an evolutionary paradox: while it allows for extensive cooperation beyond spatially limited kin altruism and beyond temporally limited mutualism. but it is likely our nearest relatives demonstrate it. Thus. alarm calls. whether such deception appears in primates. reciprocity would most likely become an evolutionary stable strategy only in populations which could avoid invasion by cheats. and grooming. beginning with the abilities to associate certain movements with future behavior (an understanding of agency). and perhaps punishment for cheating. The complicated fighting coalitions of baboons and the food sharing of chimpanzees offer numerous and subtle opportunities for withholding reciprocity (fully or. This arms race could lead to the evolution of the ability to recognize intentions. we should look for evidence of deception to see. more subtly. So long as the system of reciprocal exchange remains in place. An animal that can avoid getting a reputation for cheating but can still get more than it gives in a reciprocal exchange will reproduce more than its conspecifics. Thus. Mesterton-Gibbons and Dugatkin  and Rosthein ). reciprocity allows the evolution not only of a communal marketplace but also of embezzlement and castigation.benefits. beginning with the ability to recognize individuals as altruists and cheats. reciprocity also selects for cheating. as is the case with full-blooded cheats. Theories include group selection models and slow expansion of kin altruism to the larger group (see respectively. To maintain reciprocity. deny them benefits. and to calculate costs and benefits given and received. and to inhibit acting on immediate motivations. to recognize that the direction of eye gaze indicates another’s perceptual information.23 Because of the time delay between altruistic acts. Subtle cheating would not necessarily evolve in a society of reciprocal altruists (nothing necessarily evolves). subtle cheats could take over a population of “naive” altruists and altruism would dissipate. and perhaps even punish them.g. presumably by detecting them and withholding altruistic acts. first. however. fighting coalitions. detection of cheating. But even if such detection evolved. In anthropomorphic terms. at great cost to the sucker altruists. In addition to selecting for the cognitive abilities mentioned above. Thus. especially in their reciprocal exchanges. and then whether some forms of deception may involve theory of mind. only in part). there will be a selective advantage for increasing subtlety. to remember which is which. a population needs to be able to detect subtle cheats. reciprocal altruism may be rare. and the ability to recognize individuals and their relationships to each other in familial and hierarchical groups. reciprocal altruism may set up a feedback loop involving selective pressures for increasingly complicated deception and then for increasingly effective detection of deception.
Significant theoretical problems are also involved in explaining how reciprocal altruism could be introduced into a population. 10
. but I suggest turning the equation around to see whether reciprocal altruism selected for certain kinds of intelligence. economies) in terms of the intelligence they require. benefits remain for subtle cheats who return favors but find ways to receive more than they give. Thus. but it would offer a significant selective advantage if it did evolve. The rarity of reciprocal altruism results from the obstacles it must overcome to become stable in a population. such as food. Indeed.
So. Furthermore. powerfully selected for in the arms race between predator and prey. camouflage) or learned solely through reinforcement (by transferring a normal behavior to a new. as Dawkins and Krebs note. so. for instance. even a “higher” primate—the primatologist—may miss it. creative deception involves deliberate misrepresentation. Creative deception can be defined as “acts by an agent deployed so that the target is likely to misrepresent what the acts signify. Most significantly. and within species. Andrew Whiten and Richard Byrne (1988) attempted to overcome these problems by gathering anecdotal and experimental evidence of deception from a wide variety of researchers. and examples in the wild will appear anecdotal. As Samuel Butler wrote. if the deception is creative. Deception at these levels occurs often both between species. it would seem that selection would weed out “those who allow themselves to be deceived” too often. the researcher must try to determine that the deceptive tactic is not the result of prior reinforcement. Nahmias. a behavior learned by trial and error (even if only one trial). MIT Press). See also Guzeldere. The latter. but reciprocal altruism offers more numerous and more subtle opportunities to deceive.”25 Thus. while the goals sought may be due to behavioristic reinforcement. its goal is to be difficult to detect. call it “creative deception. when examining deception in evolutionary terms. is the hardest evidence to ascertain (especially without language) and also the most significant aspect for my proposal. But the pressure to counter this deception is equally pervasive.26 Creative deception is difficult to test for in experimental set-ups. Rather than conflicting. Deception.” 11
. the pressure to deceive and detect is pervasive among all creatures that perceive and react to each other. Indeed. it also encourages inferring others’ behavior from your own. of course. and Deaner (forthcoming in The Cognitive Animal. They then tried to determine whether the reports of deception involved a target acting on a false belief and also whether the agent intentionally caused that false belief. if defined simply as misrepresentation. in animal communication “the actor is selected to manipulate the behavior of the reactor” (1978: 383). since targets will try to detect and avoid it. Testing for such creative deception is notoriously difficult. simulation theory of mind may be a precursor for theory theory of mind. during mating rituals. —Machiavelli. But differentiating behavioristic deception from the next level. The Prince Though Machiavelli’s political commentary seems true enough. Deception One who deceives will always find those who allow themselves to be deceived. the deceptive means used to attain these goals are the result not simply of past reinforcement but of prediction about others’ behavior. determining whether an animal can differentiate
Since reciprocity occurs between conspecifics (creatures that behave in similar ways). perhaps even simulating how you would behave given a particular situation. to the advantage of the agent. 25 This definition is only slightly modified from Whiten and Byrne’s definition of “tactical deception” (1988). may be completely hardwired (for example. misleading context after being rewarded for performing the behavior in that context). “The best liar is he who makes the smallest amount of lying go the longest way.24 The difficulty in determining how much of deceptive behavior is genetic and how much is learned simply marks one example of a problem posed by most complex animal behavior.5. the selective advantage of deception seems pervasive.” is even more difficult.
Creative deception should also be relatively rare.
distinguishing between overt behavior and the deceptive intentions that sometimes underlie it” (1991: 276). (2) covering up sexually induced outbursts to avoid angering a dominant. only to follow Belle to the right location. Whiten and Byrne received the most submissions for creative deception in chimpanzees and other apes and few submissions for prosimians and non-primates. unlike objects. their motivation to move towards food) because they recognize how a conspecific (or sometimes a human trainer) will behave based on their current behavior. Furthermore. The situation escalates to the point where Belle leads the group to the wrong site or the site of a smaller food cache. 12
. they control their normal behavior (e.
This experiment was run by Charles Menzel (see Byrne. involving the agent’s knowing how it is manipulating the target’s beliefs. Belle inhibits her normal behavior of moving immediately to the food source.between knowing that a deceptive act works and knowing how it works is the key to determining if the animal is representing mental states in a way that suggests theory of mind. Rock inhibits his behavior of attending to another’s movement towards food. which. A similar experiment was run on ring-tailed lemurs (prosimians) by Rob Deaner at Duke University. most of the examples still fall to Lloyd Morgan’s canon and can be explained with a behaviorist interpretation. but that certain behavior (often the inhibition of normal behavior) may prevent the target from acting. albeit a complicated one. Whiten and Byrne’s findings have been questioned both because they often rely on anecdotal evidence (as one critic said. (3) hiding fear grins in conflicts. They classify most monkey deception as less creative than ape’s. Whiten suggests that chimpanzees “offer the strongest indications that an agent entertains some concept of deception itself. to move towards food). indicate that apes are able to inhibit their normal behavior (e. as in the cases where the target pretends be preoccupied and also in cases where apes become angry when they have
See Commentary on Whiten and Byrne (1988). This behavior often includes the agent’s concealing (as in 2 and 3 above). 247. or creating a false image (as in 4). act based on their perceptions and desires. The quotation is from Irwin Bernstein.g. Belle. 1995: 132). While there has been no conclusive evidence for such high-level intentionality. these examples indicate that the agent recognizes that the target has a desire and will act on it. who has been shown the location of some hidden fruit which the dominant chimp. Indeed. would takes from her if he had access to it. He found no evidence of intentional deception. When released with Rock. “the plural of anecdote is not data”) and because. and (4) pretending to see predators to escape pursuit. Belle either avoids the hidden fruit until Rock is preoccupied or leads the group away from the food. Rock. body orientation and eye gaze) indicate future behavior.27 Nonetheless. A commonly cited example of such deception involves a subordinate chimp.
See Whiten and Byrne (1988) for more examples of deception. except for baboons. as well as more recent experiments. consistent with this proposal. apes offer the best evidence for detection of deception. also have high EQ and whose fighting alliances seem to involve reciprocal altruism. some critics claim. distracting (as in 1 and Belle’s false leads).29 If specific past reinforcement can be ruled out. They recognize that certain animals are agents in the world which.g.g. their survey. but Rock then counters by pretending to be preoccupied. and in light of that understanding.28 Other examples of deception cited by Whiten and Byrne include apes and baboons (1) pretending to be injured to avoid confrontations. And both apes seem to recognize how their own and the other’s behavior (e.
to see which animals have various levels of theory of mind. Sarah.30 This evidence suggests a correlation between primate species that engage in reciprocal altruism and those that have demonstrated creative deception (notably baboons and the great apes. such as intentions. some argue that theory of mind is impossible without language). Nonetheless.31 Definitive results are elusive. Individuals are said to have a theory of mind when they make inferences based on unobservable mental states that provide a coherent. as well as when children develop these levels. See Perner (1991). whose vision is obscured. While Sarah surely cannot “think as others think” to the extent of a mob boss.been tricked (sometimes by their human trainers). because she can understand how others’ desires motivate their future behavior. Frans de Waal proposes that chimpanzees demonstrate reciprocity in their grooming alliances. and also what he calls moralistic aggression in their punishment of nonreciprocators and enemies. could predict the appropriate behavior for a human subject trying to achieve a goal. Theory of mind One thing I learned from pop was to try to think as other people around you think. especially when working with apes and infants who lack language and hence cannot be asked about what they think or what they think others think (indeed. especially the common and bonobo chimps). the best evidence for such punishment has also been found in chimpanzees. (2) choosing a human helper who can see where food is located as opposed to one who cannot see the food. as I will discuss below. The case for creative deception would be strengthened if there were further tests for whether an animal understands the relationship between mental states. explanatory framework (like scientific theories). Mario Puzo’s The Godfather The term “theory of mind” was introduced by Premack and Woodruff (1978) who studied whether a chimpanzee. such as using a key to escape from a locked cage or plugging in a record player to make it work. These tests include (1) showing another animal. she was able to predict what a human would do based on his desires—that is. This indicated to Premack and Woodruff that Sarah has a desire theory of mind. as is suggested by reciprocal altruism. Indeed. and (3) choosing on later trials a human who earlier accidentally spilled
See de Waal (1996. the correct lever to pull to obtain food. and food sharing. while most apes pass at least the desire tasks as well as some tests for taking on the perspective of another animal and perhaps even understanding another’s intentions. my proposal calls for trying to determine whether deception is sometimes accompanied by detection of deception as well as punishment of deceivers. 5. the research so far indicates that our nearest primate relatives can pass some theory of mind tasks—tasks that human children begin passing in their third year. 150-162). how he would solve the problem in way specifically related to his apparent goal. —Don Corleone. and behavior. A wealth of research has since occurred. This is where theory of mind research is helpful. fighting coalitions. So far. 13
. Monkeys fail both desire tasks and belief tasks. both on primates and human children.
so they say Joe will look where he believes the bananas are.g. Sarah). the children will also answer “coins” (and they are not lying to cover up for their mistake). where will Joe look for bananas?).juice as opposed to one who intentionally spilled it.” she is shown that the pack actually contains coins. for example. Another test involves asking a child what she thinks will be in a pack of M&M’s.g. some tests ask a child where a character will look for an object. most three year olds will answer “coins. False-belief theory of mind.32 Apes.
Perner (1991. For example.
Surprisingly. her limbs) and the rest of the world. But four year olds can represent—as distinct from the actual situation and their own beliefs—Joe’s false belief about the item’s location. When she is asked what another child (new to the scene) will say when asked what the pack of M&M’s contains. not where they actually are.35 Indeed. 189-202). 14
.34 Josef Perner also offers evidence that children at age four begin to perform what he calls “genuine deception. Four year olds pass the test. however. Joe knows there are bananas in the refrigerator. however. Children pass desire theory of mind tasks beginning around age two. Wellman (1990) and Perner (1991) for descriptions of these experiments. the child is told there are bananas in the cupboard and in the refrigerator. is not acquired until around age four. when asked what they had earlier believed was in the pack of M&M’s.33 They recognize that they have a point of view that is distinct from others’ view of the world. have not shown evidence of false-belief theory of mind. Perner and some other developmental psychologists believe that such deception
These tests were devised by Daniel Povinelli (reported in Byrne 1995). or whether a character in a story will be happy if he finds a desired object. but then sees Joe’s mom move them to the cupboard while Joe is gone. the well-worn intuitions of mothers and teachers confirms it: most of them report that children begin telling deliberate lies around the age of four. For example. “M&M’s. It indicates an ability to represent a situation which does not correlate with reality and to represent another’s beliefs as distinct from one’s own.” manipulating others in order to induce in them false beliefs about reality. an understanding that perhaps begins with the infant’s distinction between the part of the world she controls (e. However. children do not pass belief tasks until age three.” demonstrating their inability to distinguish their own knowledge about the world from others’ knowledge of the world. See. After she responds. given what the child and the character know about a situation (e. In addition to experiments that test for such deception. because they assume other’s beliefs are the same as their own. where will the child say Joe will look for the bananas when he returns? Most three years olds will say Joe will look in the cupboard. developing an understanding of subjectivity. These tasks often involve asking the child what a character will do next given his prior behavior towards a goal (like the tests on the chimp. Piaget recognized the significance of the limits of the infants’ will in the development of their understanding of subjectivity and objectivity. if the child sees Joe find bananas in the refrigerator.
g. Harris’ chapter in Carruthers and Smith (1996). “Theory theory” suggests that our mind-reading abilities rest on our ability to recognize patterns of behavior and organize them under theoretical mental constructs (like beliefs and desires). In fact.e. we feel sad when we see bad things happen to movie characters—but we likely use theory of mind to predict others’ complex behavior—for instance. to try to keep him ignorant. she may inhibit her own desire to take them out and eat them while he is around.37 I am not convinced the difference between these two views is as significant as the debate implies. for instance. which we then apply to our own mental states (notice the similarities between this view and the view of social psychologists discussed in Chapter 3). there seems to be distinctions between desire. So. some researchers believe it is impossible to have belief theory of mind without some understanding of false belief. For instance. In any case. as I do not believe their outcome substantially affects my proposal. A might give a false alarm call to make B leave so she can eat her bananas in peace. Such behavior is similar to the concealment involved in some primate deception. delineating sharply between the three levels of theory of mind may be painting too simplistic a picture. belief. But belief theory of mind would allow A to predict B’s behavior based on what he has seen (an important aspect of understanding belief is connecting seeing with knowing). and only if they could talk to us about mental states would we believe
See. and even adults seem to use both inferential methods depending on the type and complexity of the situation (we may simulate to understand another person’s emotional state—for instance. simulating another’s point of view. they learn that saying they did not eat the cookies may not work if they have crumbs on their face). However. “Simulation theory” suggests that we understand other’s mental states by imagining how we would feel and how we would behave in a situation like the one they are in—i. Children become better liars as they recognize the many factors that may influence others’ beliefs (e. Thus. 15
. if A knows B wants her bananas and will try to take them. and false-belief theory of mind which underlie varying competencies in deception. for instance.requires the representational abilities involved in false-belief theory of mind. I will not try to adjudicate these particular debates. Increasingly complex theory of mind certainly seems to allow for more complex types of deception. I believe simulation likely provides a building block for theory of mind. we try to “get in the head” of the Godfather to figure out what he will do next).36 A more substantial debate involves delineating between theory of mind and simulation. For instance. Some researchers have also questioned whether theory of mind requires language. It allows A to recognize when B is ignorant. desire theory of mind allows an individual A to predict B’s behavior based on his desires or goals. For instance. Carruthers and Smith (1996) contains several chapters dealing with this controversy. A could turn her back from B while she eats her bananas. Smith writes. As P. So. and to try to change his beliefs to her advantage. But false-belief theory of mind seems to exponentially increase the possibilities for creative deception.K. for instance. “Only if chimpanzees could talk to each other about mental states would they have evolved mind reading. Children with language and falsebelief theory of mind can tell lies in order to make their targets believe what is not the case.
More generally. and it certainly evolved first” (1995: 233). Indeed. there is a long tradition in philosophy. theory of mind seems to underlie selfconscious knowledge. I argue that in order to have a belief. This controversy will require much more research. In introspecting on our conflicting motivational states. the ability to distinguish between belief and reality allows us to imagine various possible futures and consider how our motivations may influence which of them occurs. 16
. as evolutionary psychologists would suggest. I am not concerned about the effect of this debate on my proposal. the main evidence for the former coming from autistics who seem to have a specific deficit in theory of mind which might be caused by damage to a specific part of the brain. Furthermore. Some researchers (such as Donald) have further suggested that mimetic culture. it is necessary to have the concept of belief. So.”38 And Donald Davidson offers a priori arguments for a related point in “Rational Animals” (1982): “First. perhaps even consciousness. but my proposal suggests that theory of mind is possible without language (e. in apes) and may even be a step in the evolution of linguistic abilities. teaching.g. if theory of mind represents an advancement of already-present general processing abilities. it more likely evolved by natural selection. we must be able to represent them as motivational states and understand their relation to actions and the world. writes that “prior to the evolution of a system as revolutionary as human language. that views language as uniquely human and as necessary for rational thinking. I am inclined to agree with those researchers who turn the equation around to suggest that representational abilities like those involved in theory of mind underlie the evolution and development of language. It seems that the ability to represent states of affairs that do not exist is crucial to the symbolic aspect of language—distinguishing symbols from the objects they represent—and that the ability to represent other’s mental states is crucial to the conversational aspect of language— understanding.39 Indeed. The idea behind these suggestions is that theory of mind involves representing one’s own and others’ mental states. for instance. Second. Richard Byrne claims with even more vigor: “The ability to imagine other mental viewpoints is a necessary precursor to language.them. However. While language certainly makes it easier to test for theory of mind and surely increases the breadth of our mental states. it may still have evolved by natural selection. and this stage involved mimetic and representational abilities found in apes and prelinguistic human ancestors. I argue that in order to have the concept of belief one must have language” (478). 354). for instance. It may seem that if theory of mind is a module. such knowledge is the basis of free will.40 This includes. Merlin Donald. Autistics lack theory of mind abilities but they possess language and the concept of belief. that you do not know what I know when I inform you of what I know. and even some human technology could not develop without these representational abilities. going back to Descartes.
Davidson’s argument faces a possible counterexample from autistics. the cognitive stage had to be set” (1991: 164). it is difficult to imagine how the activity of deliberation would be possible without representing at least two possible alternatives for the future. This ability may have been the adaptation which in turn underlies many exaptations and spandrels that require a similar sort of representational ability.
There is a significant debate about whether theory of mind is a modular (domain-specific) ability or a general (central) processing ability. the cognitive abilities I have identified with free will.
In Carruthers and Smith (1996. and as I have suggested. I believe.
6. To many.42 But some chimps (and other apes) have been raised within a human family and trained to use sign language (or some other symbolic language). Children. We accept that children have varying degrees of free will and responsibility according to the degree to which they can know what they are doing and control themselves accordingly. However. See also Byrne (1995: 224). especially those that include reciprocal altruism. They also pass mirror self-recognition tests and can be taught symbolic language and self-reference. In addition. in deception and teaching). they have demonstrated the abilities to follow other’s eye gaze. have shown evidence of reciprocity and intentional deception. Conclusion: Do Chimps Have Free Will? Though the precise interpretation of theory of mind still needs refinement. imitate others. but this question need not be answered with a resounding “no. and take account of other’s ignorance (e. and by the resulting selective pressures to deceive and detect deception. But consider how we would answer the question of whether a 4-year-old child has free will—or perhaps an 8-year-old or a 13-yearold. their human trainers anthropomorphize them. The human “parents” see the
See Figure 3. surely increases their ability to conceptualize their desires and contrastively compare them. it may sound counterintuitive to suggest that any animal has any degree of free will—we generally consider free will unique to humans. they also grow up learning the moral values of their culture. as I mentioned above. especially chimps. Not surprisingly.g. Despite most culture’s coming-of-age ceremonies. children learn language. suggesting a particular ordering of cognitive skills in both evolution and development. nor do we treat them that way.41 In sum. These differences correlate not only with a developmental sequence in children but also with the phylogenetic sequence from monkeys to apes to humans. and influence their motivations accordingly? Do chimps have free will? Probably not. as they grow up. apes appear to have at least the rudiments of an understanding of their own and other’s mental states. and false-belief theory of mind based on age of onset in children. we do not believe children go from lacking free will to having it overnight. gain responsibility as well as free will. the evidence so far supports differences between desire. The great apes. the degree to which they possess free will increases at the same time as they learn to exercise those abilities specifically in terms of their culture’s specific moral obligations. an important difference between chimps and children is that children not only develop their self-knowledge and self-control as they mature. See Taylor (1977).” Since an agent possesses free will to the degree that she possesses the requisite cognitive abilities to know herself. That is. belief. Does this mean that they can introspect on their motivational states. I have suggested that this progression was driven by the social complexity involved in some primate societies. So. 17
. Furthermore. as they acquire the ability to act in accord with their knowledge of themselves and the world. which. we can say that chimps may possess some degree of free will—they may at least be able to inhibit immediate (first-order) motivations in order to achieve goal they recognize as more important. predict other’s behavior based on their desires. identify which of them they prefer to move them. they acquire the relevant knowledge of the values in terms of which their actions will be judged by others in their culture. and this includes treating them as responsible for some of their behavior.
and they blame the chimps for breaking rules when they “should have known better. We can learn to know ourselves better. Washoe: “[They] left their ‘chimp myths’ behind and built respectful relationships with the individual members of Washoe’s family. We gradually grow into free and responsible creatures. the development of free will is tied more closely to our evolutionary and genetic heritage. 18
. But. we can learn techniques of self-control and strength of will so that we can more effectively act on our selfknowledge. They were rewarded with one thing only: chimpanzee friendship” (1997: 291). and being objects of. There are limits to how much a chimp can develop its free will. Free will involves cognitive processes that we develop given any normal upbringing.
Accounts from these human trainers are filled with the language of relationships and responsibility. Moral education and experience increases moral responsibility. and we are likely making a mistake if we really endow them with free will and responsibility. For instance. Nonetheless. like any other cognitive ability. free will may be developed and improved. (1962: 75). the full range of human and moral attitudes. those that apply to all of us because of our human nature and those that apply to each of us because of our individual history. and our reflective knowledge and self-control works within boundaries.chimps as able to manipulate and deceive. But when it comes time to punishing them. I would suggest. The more we know. Roger Fouts writes of new volunteers to care for his chimp. Practical education and experience increases free will. but are not yet truly capable of either. moral responsibility involves specific content relative to our particular upbringing. the more we can move within those boundaries and perhaps even push them back.”43 These “parents” take rudimentary reactive attidudes towards their ape charges in the same way other parents do. Much of who we are is given. And most significantly. we usually recognize that they “couldn’t help it” precisely because we believe they either didn’t know any better or didn’t know how to control themselves. There are also limits to how much we can develop our free will. we can learn how to improve our ability to consider our motivations rather than acting on them without thinking. Knowledge is power. Compare Peter Strawson’s point about how we move from the objective attitude to the reactive attitudes in our relationships with children: Thus parents and others concerned with the care and upbringing of children cannot have to their charges either kind of attitude [objective or reactive] in a pure or unqualified form. But within any boundaries there are degrees of freedom. In addition to learning the content of our culture and hence increasing our ability to act responsibly. while the development of moral responsibility is tied more closely to our maturation within a particular culture with its particular set of moral obligations. They are dealing with creatures who are potentially and increasingly capable both of holding. many of us take similar attitudes towards our pets. Of course.