The (r)evolution of primate cognition does the social intelligence hypothesis lead us around in anthropocentric circles

In particular, the social intelligence hypothesis or, as it is now more commonly known, the social brain hypothesis has enjoyed great success as an explanation for why the anthropoid mon

To appear in: The Routledge Handbook of the Social Mind (ed Julian Kiverstein)

Please do not cite without permission

The (R)evolution of Primate Cognition: Does the Social Intelligence Hypothesis Lead us Around in Anthropocentric Circles?

Louise Barrett

1 Introduction: anthropomorphic reflexivity

The British comedian, Eddie Izzard, has a joke about squirrels, which centres on the way they often pause suddenly while eating As Izzard suggests, its as if they’ve just remembered

something Izzard mimics a squirrel eating a nut before pausing dramatically and asking himself:

“Did I leave the gas on?” There is a pause for laughter, before Izzard resumes his imaginary nut-eating, saying dismissively: “Nah…of course I didn’t!! I’m a f**king Squirrel!” As well as getting an even bigger laugh, the notion of a squirrel anthropomorphically commenting on the folly of attributing an anthropomorphic thought to a squirrel captures the reflexivity of human thoughts and action, adding a further layer to the joke For anyone interested in recent

developments in comparative cognition, yet another layer is added by the recognition that this same reflexive quality structures much of the debate over the use of anthropomorphism as a scientific strategy, with worries raised over whether ascribing particular traits to other species, or our refusal to do so, is a reflexive response to the way in which we wish to see ourselves

What makes this all even more mind-bending is that human reflexivity is itself argued to have arisen evolutionarily (and ontogenetically) as a result of the intense sociality that

characterizes our species: Dewey, Mead and Vygotsky for example, all suggested that human selves, and our ability to see our selves as selves, arose from the internalization of those around

us to produce a “generalized” or “social other” (Dewey 1958; Mead 1934; Vygotsky 1997) In this view, sociality is fundamental to our individuality, representing both its source and cause Similar ideas about the evolutionary pressures generated by social life have also been put

forward to explain the evolution of cognitive abilities across the primate order as a whole

(Humphrey 1976; Dunbar 1998) In particular, the social intelligence hypothesis or, as it is now more commonly known, the social brain hypothesis has enjoyed great success as an explanation for why the anthropoid monkeys and apes, including humans, should possess the largest brains for their body size across the animal kingdom

In what follows, I briefly trace the history of the social intelligence hypothesis,

highlighting its strengths and weaknesses as a general theory of cognitive evolution I then go onto consider how the social intelligence hypotheses links to current debates on

anthropomorphism in ways that preserve the essential anthropocentrism that lies at the heart of both As a parting shot, I suggest that this outcome is inevitable if we accept the social

intelligence hypothesis as our best explanation for primate cognitive evolution, which perhaps explains why we continue to find ourselves in the same mind-bending zone as Izzard’s forgetful squirrel

2 A Brief History of the Social Intelligence Hypothesis

Following Alison Jolly’s early suggestion that primate social life “preceded the growth of primate intelligence, made it possible, and determined its nature” (Jolly 1966, p 506), Humphrey

(1976) subsequently (and independently) made a similar claim based on his own observations of captive monkeys and wild apes As he pointed out, the mountain gorilla, which lives, effectively,

in a giant bowl of salad, faces a set of ecological problems that are simple and finite, and no more demanding of intelligence than those that apparently face other, less brainy, creatures Add

in the demands of social life, however, and you have a game-changer: obligate sociality places a premium on the ability to predict how other animals are likely to act, a capacity that makes it possible to anticipate, ameliorate or avoid altogether the competition that arises when animals are forced to live together and share resources In other words, animate beings, with agendas and goals of their own, generated the selection pressure to acquire a specific kind of psychological insight into others, and a more ‘creative’ form of intelligence; one that is fundamentally different from the practical knowledge needed to forage effectively and efficiently

More specifically, Humphrey (1976) made the case for primates as ‘social gamesmen’ (p.309), rather like human chess-players: they needed to both preserve the overall structure of the group while “exploiting and out-manoeuvring others” (p.309) The idea of primates as

“calculating beings” capable of a “special form of forward planning” (p.309) gained even greater prominence with the publication of the “Machiavellian Intelligence” (Byrne & Whiten 1989) As the editors were at pains to note, the word “Machiavellian” referred only to the need for a

specifically psychological understanding of others, rather than the need for manipulativeness or deception, although these clearly formed part of the argument, and were especially prominent in Humphrey’s work

Empirical tests of the theory then followed, presenting evidence to suggest that social factors had indeed been more important than ecological factors in promoting brain size

evolution: Dunbar (1992, 1995) found a positive correlation between neocortex ratio and group size across the anthropoid primates, while simultaneously demonstrating that ecological factors did not show any significant relationship This was interpreted as showing that primates used specifically social strategies to help cope with the ecological problems they faced (e.g.,

competition for food resources); problems that, in turn, had been generated by selection

pressures favouring group living (most specifically, avoiding predation: van Schaik 1983)1 Further studies reinforced the association between social life and brain size, by correlating

neocortex measures with complex social behaviours like promiscuous mating (Pawlowski et al 1998), grooming clique size (Kudo & Dunbar 2001) and rates of deception (Byrne & Corp 2004) This led to the further claim that, much as Humphrey (1976) had argued previously, it was the ability to represent and manipulate specific kinds of information about social

relationships (as opposed to, say, improved memory capacities), that both selected for increased brain size, and then placed limits on the size of group that a given species was able to sustain More specifically, the form of social intelligence indexed by enlarged brain size “principally focuses on the ability to use knowledge about other individuals' behavior—and perhaps mind-states—to predict and manipulate those individuals' behavior” (Dunbar 2003, p.167) What was noticeable in this shift to empirical investigation, then, were the assumptions that large neocortex size supported a highly specific kind of cognitive ability, and that group size could be used an appropriate indicator of social complexity, without all that much in the way of independent empirical support for either of these positions

In recent years, this has led some researchers to question whether evolutionary size

increases in particular parts of the brain can, in fact, be traced causally to sociality in the manner implied (Healy & Rowe 2007; Healy & Rowe 2013; Rowe & Healy 2014) The rather nebulous concept of social complexity used in these analyses may also explain why, when empirical work

was extended to other species,2 the nature of the relationship proved to be distinctively different

to that of the primates, with pair-bonding (i.e., the formation of an enduring relationship between adult mates) emerging as the key correlate of enlarged brain size (Dunbar & Shultz 2007; Shultz

& Dunbar 2007) This naturally required some reconfiguring of the original hypotheses, with a more precise delineation of the quantitative versus qualitative demands of sociality Among the primates, the use of both a quantitative measure of group size and relative neocortex size rested

on the assumption that tracking a large number of dyadic relationships generated a significant cognitive burden As described above, this quantitative demand was accompanied by the

(implicit) assumption that animals also possessed a conceptual understanding and recognition of relationship quality, and attempted to manipulate others using this information accordingly (Dunbar 1998) The discovery that pair-bonded species have the largest relative brain sizes in ungulates and birds has therefore been taken to indicate that it is primarily the quality of

relationships, and not their quantity, that is evolutionarily important, thus reversing the original interpretation (Dunbar & Shultz 2007; Shultz & Dunbar 2007; see also Dunbar & Shultz 2010) The aspect of social life now considered to have selected for the enhanced brain size of the anthropoid primates compared to other species is the manner in which the pair bond has been generalised to include all group members (Dunbar 2009, p.564) characterizes this as “an

important phase transition in the form of the social brain effect” that took place at an early stage

of primate evolution The question that arises from all this is: why are pair-bonded relationships

so cognitively demanding? Dunbar (2009) suggests this question has no ready answer because

we also lack a ready answer to the prior question of how animals go about the day to day

business of maintaining a pair bond; so far, we have only considered their value in an

evolutionary-functional sense and not from a proximate-psychological/physiological perspective (i.e., we consider only the way in which they contribute to fitness, and not how they are

maintained within the context of an individual lifetime)

This may well be true, but it is equally true to say that this applies to the original social brain hypothesis The cognitive demands of social life were based on a series of assumptions, put forward by Humphrey and elaborated on by Dunbar and colleagues, that social life was complex because it required the manipulation of social information about various third-party relationships, and a form of prospective knowledge about one’s own relationships and those of others This, however, stemmed from the existing knowledge that the anthropoid primates had larger brain sizes than expected for their body sizes, and that they were all intensely social The reasoning therefore ran from the observation of large brains to the kinds of complex processing that large brains would afford in a social context (which, one could argue, were based largely on a folk psychological projection of our own abilities in this domain) No convincing evidence was offered to suggest that the social life of non-human primates actively required the use of flexible, highly cognitive, prospective strategies in real-time Evidence in support of the evolutionary argument, such as the correlation between group size and brain size across the primate order (Dunbar 1995), was then taken as implicit support for the postulated proximate behavioural and cognitive mechanisms by which individual animals increased their survival and reproductive success

Part of the reason for this seems to lie in the anthropocentric focus of the original theory, where explaining the evolutionary origin of our own extraordinarily large brains was integral to the whole project Questions about other primates' social cognition were posed in ways that privileged the evolutionary origins of cognitive abilities like language, analogical reasoning and theory of mind skills, because these were seen as essential and fundamental to understanding

how people are able to predict what other people will do, despite evidence to the suggest that can achieve without the need to model the mental states of others (e.g., Hutto 2004; Gallagher 2001; Andrews 2007a) Research efforts were (and still are) geared explicitly to detecting these

abilities or, more commonly, their precursors, in monkeys and apes, either to reinforce our own uniqueness or to identify how our own skills in these domains have been derived from

evolutionarily simpler mechanisms (e.g Dunbar 2003; Bergman et al 2003; Call & Tomasello 2008; Zuberbühler 2000; Cheney & Seyfarth 2005; Seyfarth & Cheney 2013)

This criticism doesn’t apply to all theories of human cognitive evolution, where the focus

is more squarely placed on identifying the unique path taken following the evolution of our own species, rather than tracing our capacities through our common ancestors as such Sterelny (2003,

2007, 2012) for example argues that we are “creatures of feedback”, where our uniqueness lies

in the nature of the feedback mechanisms that connect, among other things, the cultural

environments we construct and inhabit, human social learning, individual expertise and human life history processes As he notes: “[a]s hominids made their own worlds, they indirectly made themselves” (Sterelny 2003, p 17) Heyes (2012) makes a somewhat similar argument when she suggests that both the “grist and mills” of our lives—the knowledge and know-how needed to deal with the world, and the social learning processes that allow us to acquire this know-how respectively—are both culturally inherited, in contrast to views which suggest these social

learning “mills” are biologically inherited: according to Heyes (2012), our social learning

mechanisms are cultural adaptations, not evolved functional specializations Equally,

Tomasello’s “Vygotskian intelligence hypothesis” (Moll & Tomasello 2007; Tomasello & Moll 2010) posits unique motivations and cognitive skills that allow humans to understand others as cooperative agents with whom one can share collaborative actions (“shared intentionality”), an ability apparently not found in other apes In his work on ape intelligence, however, Tomasello (Tomasello & Call 2006; Call & Tomasello 2008; Schmelz et al 2011) does seem to offer a largely anthropocentric “top down” view, where apes are compared and contrasted with those of humans, with respect to capacities like the attribution of certain kinds of mental states

This same ‘top down’ view, where our own capacities colour theories of what we should look for in other species, can be seen in the hypotheses put forward to explain why pair-bonding should generate high cognitive demand As Dunbar (2009) sees it, there are two options: (i) intense pressure for finely-tuned mate choice competences and (ii) the need to coordinate and synchronize behaviour These are not mutually exclusive, but a “critical tests” analysis pitting the predictions of the hypotheses against one another was argued to come down in favour of behavioural coordination One reason for this, Dunbar (2009) suggests, is the need for

individuals to anticipate their mate’s needs so as to ensure that both members of the pair are able

to meet their nutritional and other requirements He then goes on to say:

“Being attentive to the mate’s needs so as to ensure that he/she can achieve

his/her daily nutrient intake has many of the hallmarks that would be

recognized as theory of mind in humans In effect, pairbonded species have to

be able to engage in perspective-taking, a phenomenon that is widely accepted

as being a prerequisite for mentalizing (or theory of mind: Hare et al 2001,

2006) Hence, pair bonded monogamy can perhaps be seen as laying the

foundations for the kinds of advanced social cognition found (albeit in limited

form) in primates and (perhaps uniquely in full-blown form) in humans”

(Dunbar 2009, p.568)

This has, in turn, led to the search for such perspective-taking skills on the part of pair-bonded species, with at least one study on Eurasian jays reporting positive evidence (Ostojić et al 2013) (although this conclusion is contingent on the precise way in which the data were analyzed)3 There is also evidence to suggest that behavioural coordination is important to breeding success: zebra finches (which are not, it should be noted, particularly large brained) that were allowed to choose their own mates displayed higher fitness and better behavioural coordination than

‘forced’ pairs placed together by the experimenters (Ihle et al 2015) In contrast, a study on New Caledonian crows, demonstrated that tasks involving behavioural coordination can be learned readily and rapidly without any form of cooperative cognition (i.e., understanding the importance

of timing and the role of the partner) (Jelbert et al 2015) There is also an experimental study of both wild and captive vervet monkeys that revealed they can learn to coordinate their behaviour

in complex ways on the basis of individual reinforcement learning alone, with no evidence for either perspective-taking or social learning (Fruteau et al 2013).4

This equivocal pattern of results is not too surprising given that the reasoning presented for perspective-taking as a necessary condition for behavioural coordination follows the same logic

as the original social intelligence hypothesis: assume the end-point of the process to be human social intelligence and then project this backward onto other species to arrive at the kinds of traits that can make sense of the behavioural data Indeed, there seems to be an even greater explanatory need for other species to possess such cognitive traits (or their precursors) so that the social intelligence hypothesis continues to cohere as a general theory of brain size evolution The anthropocentric origins of the social intelligence/social brain hypothesis thus continue to push the study of social cognition along a trajectory that ultimately may hamper rather than enhance our understanding of other species (see also Barrett & Würsig 2014)

3 Not Everything is About You…

Looking explicitly for precursors of human socio-cognitive attributes, for example, judges other species’ capacities against a human standard, and by necessity implies they will only meet some fraction of this standard, when it would be more productive to consider these as adaptations

in their own right (Tyler 2003; Barrett 2011; Barrett 2015a,b) More insidiously, the success of the social brain hypothesis has led other social species to be judged against a specifically primate standard — the crow family have been characterised as “feathered apes”, while studies of the cetaceans emphasis their convergence with the cognitive abilities of chimpanzees As the apes themselves are tethered to a human standard, this means that, despite an overt evolutionary emphasis, humans continue to lie at the centre of the comparative cognition project, even if they

no longer occupy the top spot on the ‘great chain of being’

This more subtle form of anthropocentrism continues to pervade both theoretical and empirical treatments of the social intelligence hypothesis because this is not a formal theory, but one that makes extensive use of a folk psychological intentional stance to construct and test its hypotheses Penn (2011) describes the methodological principles of this “comparative folk psychology” as the process by which a behaviour that appears clever to human eyes is

systematically investigated, such that instinct, stimulus-bound associative learning or simply random guessing are all ruled out as possible explanations of a behaviour Alternative

explanations based on what humans would be thinking if they were behaving like the subjects in that same context are then offered and the claim is made that it is appropriate (and more

parsimonious) to conclude that the non-human subjects are thinking what human subjects would

be thinking or, at least, thinking functionally equivalent thoughts or their ‘precursors’

This notion of functional equivalence is even more problematic than Penn (2011) suggests because it has licensed a form of ‘as if’ evolutionary reasoning that actively eliminates any need for the mechanism to be specified in detail (Barrett et al 2007) Animals are argued to act ‘as if’ they possess particular psychological mechanisms because their behaviour produces the

appropriate functional consequences This, in turn, may stem from the fact that Dunbar’s original analyses were situated purely at the functional level, designed to identify the factors that would potentially enhance the fitness of their possessors, and did not explicitly set out to link particular proximate mechanisms to their ultimate function Thus, all that is required to test hypotheses relating to social intelligence is that animals act in ways consistent with the proposed

evolutionary function of the behaviour, while the nature of the underlying mechanism remains utterly opaque The use of folk psychological ‘as if’ reasoning thus obscures the fact that studies using this tactic generally provide no solid evidence of the ultimate fitness benefits of a

behaviour, nor do they demonstrate the operation of a specific proximate mechanism (Barrett et

al 2007; Barrett 2011) Instead, they sit uncomfortably in between, offering a “proximatey-ultimatish” level of explanation

There is also the related problem that, in most studies following such reasoning, there is the implicit equation of apparent behavioural complexity with cognitive complexity, when there is,

in fact, no necessary connection between the two For example, one interpretation of the food-cache protection strategies of scrub jays is that they are able to attribute mental states, take the perspective of another bird, and so understand when their caches are at risk of pilfering by

others This behavioural outcome has also been modelled, however, by assuming that re-caching

is motivated simply by a general desire to cache more under conditions of stress, which itself is determined by the presence of onlookers, and by a bird’s own unsuccessful recovery attempts (van der Vaart et al 2011, 2012; for further examples see Barrett 2011) The overt patterning of behaviour in both cases is the same, and the outcomes are equivalent This suggests that

mechanisms based on the notion of mental state attribution are just ‘simpler for us’ to understand intuitively, rather than detecting something in behaviour that necessitates a more complex

cognitive mechanism as its cause

Having said this, a recent test of van der Vaart’s (2012) model on real scrub-jays suggested that their stress hypothesis could not adequately account for patterns of caching (Thom &

Clayton 2013) The degree to which it tested the stress hypothesis is open to question, however: the study recorded only the number of items cached, whereas van der Vaart et al.’s (2012)

hypothesis predicts an increased frequency of caching and re-caching (i.e., instances where the same item could be cached several times over) What remains true, however, in both the original work and in this more recent test is that the more complex cognitive mechanism, which is

expressed in folk psychological terms, provides no mechanistic explanation at all van der

Vaart’s (2012) simulated jays, by contrast, posit clearly specified mechanisms that provide a promising explanation of caching behaviour Thus, for all we know, “mental state attribution” might simply boil down to contextually-influenced tendencies to behave in particular ways Indeed, this is the burden of the critique presented by Penn (2011) He argues that folk psychological explanations prosper because comparative psychologists have, to all intents and purposes, abandoned the fundamental tenets of the cognitive revolution Instead of embracing the idea that cognitive processes are computational, rule-governed, algorithmic processes, Penn (2011) argues that comparative psychologists aim to dismiss this possibility in favour of

demonstrating that animals have an “understanding of” or “insight into” a particular folk

psychological concept The word “cognitive”, he argues, has thus become synonymous with

“mentalistic”, “conscious” or “insightful” or, as Tomasello & Call (2006, p 371) put it, with the

notion that “apes really do know what others do and do not see” (emphasis added) This, Penn

(2011) argues is actually a strongly anti-cognitivist position: the mentalistic mechanisms are presented as alternatives to computational or algorithmic processes, when they are really just

short-hand folk psychological description of such processes In many ways, then, Penn (2011)

echoes the arguments of the radical behaviourists, when he suggests that contemporary

comparative psychologists seem content to trade in folk psychological idiom (what Skinner would call “mental fictions”) as explanations of behaviour (Skinner 1977)

For Penn (2011), the solution to this problem is obvious: open up the “black box”, cash out folk psychological descriptions in terms of computational, algorithmic or neural levels of

explanation, and fulfill the promise of the cognitive revolution Abandoning folk psychological explanations would, he suggests, allow us “to imagine a rosier future”, one where “we study

non-human social cognition from a cognitive perspective” (Penn 2011, p 262, emphasis in the

original) Such a move would allow us to “build a scrub jay that thinks like a scrub jay” (p.262) But would it?

One reason why this might not work as well as intended is because the computational approach that Penn advocates is not as free of human intentionality as he supposes Although we treat computational cognitivist theories as essentially species-neutral, this is not the case (e.g., (Brooks 1991; Barrett 2015a) The problem domains (e.g., symbolic algebra, geometrical

problems, natural language understanding and vision) of the original artificial intelligence project were all “bench-marked against the sorts of tasks humans do within those areas” (Brooks 1991,

p 140) These problem domains were also defined and refined by researchers in ways that

simplified the task at hand, abstracting away most of the details Thus, as Brooks (1991) points out, humans had employed most of the intelligence needed to solve a particular task (i.e., the part

of the process involving abstraction) well before the computational models or artificial systems were let loose on it In other words, given that the relevant details of such models were decided

by humans in advance, the notion of intelligence captured by models of this nature was

fundamentally tied to human capacities The possibility therefore exists that there is a deep anthropocentrism built right into the core of computational theories of mind (see Barrett 2015 for

a more detailed discussion of this) Hence, any theory articulated in computational terms,

particularly those that make use of a representational framework, may continue to build a scrub jay that continues to think rather like a human Indeed we can already see this in existing work: the programming language, Act-R, used by van der Vaart et al (2012) is explicitly designed to mimic the human brain.5

4 Moving Beyond the Brain

This argument links to a broader criticism of computational theories in general, and the social intelligence/social brain hypothesis in particular, as both neglect the perception-action mechanisms by which animals engage with the world Indeed, by defining cognition as a brain-bound process of computation, such theories actively encourage such neglect, treating sensory and motor abilities as “peripherals” that convey input to and output from the brain, and are not relevant to understanding cognitive processes as such (an aspect that reinforces the idea that computational theories are species neutral) This neglect of bodily engagement with the

environment is problematic because selection has acted for much longer on these perceptual and

motor systems than on the kinds of “high-level” processes identified as cognitive, and it seems likely that perceptual and motor capacities influence the form that any such cognitive processes take, as well as raising the possibility that, in some instances, intervening cognitive processes may not be needed at all (Brooks 1999; see also Barrett 2011 and 2015 for more detailed reviews

of this alternative stance)

With respect to the social intelligence/social brain hypothesis specifically, this failure to consider perceptual and motor abilities means that differences between primates and other

species inevitably are downplayed, even though these may make a very real difference to the kinds of sociocognitive strategies open to a species This seems to make it even less likely that

we will build a ‘scrub jay that thinks like a scrub jay’ What is also lost in a solely brain-based view of social intelligence, then, is any notion that the selection pressures acting on non-primate species may compete with those social pressures argued to increase brain size Holekamp et al (2013) in her work on spotted hyenas, for example, has argued that selection on the ability of the jaws to resist high stresses and exert strong bite forces may have limited the extent of brain expansion in the genus (and carnivores more generally) Brain size expansion reduces the area available for muscles within the zygomatic arches, and smaller muscles would mean weaker bite forces In the hyena case, the feeding apparatus seems to have been under stronger selection than the brain, and Holekamp suggests these morphological constraints explain why hyenas show less behavioural flexibility than primates, despite striking similarities in social organization (see also Barrett & Würsig 2014 for a similar argument with respect to cetacean cognition)

This brings us back to the issue of anthropomorphism Recently, there have been a number

of positive arguments in favour of an anthropomorphism as a valid scientific strategy that are couched in slightly different terms to more standard arguments What I want to argue here is that these similarly contain a deep anthropocentrism, despite ostensively arguing to the contrary

5 Anthropomorphism as “ethnocentrism”

Debates about anthropomorphism and its appropriateness have a long history, coming and going in cycles, reflecting the dominant schools of thought at a given place and time This story

is well rehearsed, from Descartes, to Darwin, and then onto behaviourism, the cognitive

revolution and the rise of cognitive ethology As in all areas of life then, there are trends in science, and currently we are in a phase where similarities between species (or rather similarities between humans and other species) are emphasized and differences are downplayed In recent years, there has been a steady rise in the number of empirical claims for human-like cognitive traits in other species, bolstered by the application of an evolutionary framework In line with this, there have been a number of cogent arguments in favour of an anthropomorphic strategy, or

at least one that does not display a bias against attributing “high-level” or “human-like” abilities

to other species

Andrews (2007a,b, 2011) for example, suggests that the issue at stake is not whether

humans use folk psychology to understand other animals, but whether animals use their own form of folk psychology to understand each other This in turn raises questions over what folk psychology actually is Andrews notes that the traditional definition is some form of

mind-reading” i.e., the attribution of mental states—beliefs and desires—to other people in order to predict their behaviour The problem with this definition is that, as she demonstrates, we do not always attribute beliefs and desires to predict what our fellow humans will do: I can predict that

my office-mate will head straight for the kettle when she arrives at work in the morning because this is what she always does, not because I believe she desires a cup of tea, and believes that the

kettle will help her achieve this by boiling the water for her It is equally clear that human

children engage in folk psychological behaviours before they gain any understanding of belief This means that the traditional question to ask in a "critter psychology”—do animals attribute beliefs and desires to others in order to predict behaviour?—simply won't do, because it asks more of animals than it does of humans (Andrews 2007, 2011) Buckner (2013) makes a very similar argument, in which he coins the phrase ‘anthropofabulation' to describe our tendency to inflate human psychological capacities, and then insist that only those non-human animals that can perform at this (over-inflated) level can be said to possess the psychological trait or

mechanism in question

Andrews (2007) suggests the way out of this quandary is to change the premise that folk psychology requires the attribution of mental states to others, to the notion that folk psychology

is the understanding that others are minded, intentional creatures She argues that this involves two elements: practices of social interaction and the possession of abstract mentalistic concepts (i.e., animals possess beliefs of some kind, and these are instrumental in enabling effective behaviour, even if they cannot attribute such beliefs to each other) On this basis, Andrews considers there to be sufficient evidence to conclude that chimpanzees’ possess some kind of folk psychology by which they understand conspecifics (e.g., Hare et al.'s (2000, 2001)

experiments that are argued to show that chimpanzees "understand what others see”)

Chimpanzee folk psychology "need not be as robust or as complex as our own, but the use of mentalistic concepts to engage in social interaction counts" (Andrews 2007, p 204) From a positive perspective, this argument leads to the conclusion that folk psychology is not a

monolith: there may be many ways for other social creatures to predict behaviour and engage socially with each other

On the negative side, there may still be a problem with Andrews’ assessment here, one that arises empirically, rather than logically The data she uses to support her position involved chimpanzees being tested within an experimental paradigm that adopts a Western, scientific conception of human folk psychology Within this paradigm, the only mentalistic concepts that chimpanzees can display are those of the experimenters whose European-American folk

psychology informed the design Despite Andrews’ arguments, and despite the greater ecological validity that is said to characterize Hare’s form of testing (based on competition rather an

cooperation), the conceptual framework of the studies she uses to support her argument remains strongly anthropocentric and, arguably, is “ethnocentric” to boot

On the one hand, perhaps this doesn’t matter if all we wish to argue for is functional

continuity across species, and simply make the case that other species possess some kind of folk psychology of their own On the other hand, broadening the concept of folk psychology and making it more ‘pluralistic’ does not eradicate this deeper problem of anthropocentrism because, empirically speaking, human folk psychology remains at its centre This problem is thrown into sharper relief when we consider the animals’ failure to pass such tests: is this because they truly

do not possess the capacities being tested for? Or do they fail because the test is not “culturally appropriate” for the species in question, reflecting only one particular conception of what

understanding others entails? How would this influence our view of their folk psychology? It seems that a concern about the “colonization” of the animal mind by a particular Western

scientific view of what folk psychology remains, even if we decide that criticism of

anthropomorphism per se is baseless

The same perhaps is true of Andrews (2007b, 2011) arguments in favour of using “folk expert opinion”, and the functionality of attributions to assess which traits can be attributed to

other animals besides ourselves Here, her argument is that we can elicit folk psychological assessments of other species from people who spend time caring for other species, such as zoo-keepers, or those who spend long periods of time observing other species, like fieldworkers The veracity of such assessments can then be adjudicated by how useful they are in predicting and controlling behaviour, which is something that can be put to rigorous scientific test Given that

we use this approach to attribute psychological properties to other humans who are unable to do

so for themselves (such as infants and very young children, or people who are profoundly

disabled in some way), Andrews argues that we should not be shy of using a similar approach with other species

From a pragmatic point of view, this approach has much to recommend it, especially in a care-giving context where attributing mental states may very well enhance overall welfare and well-being of other species However, this strategy will work regardless of whether animals actually do possess psychological states, or whether we merely project them Scientifically, this approach may also work for the reasons Andrews gives, namely, that folk expertise is not the end point, but the starting point for further scientific investigation (although this would seem to fall prey to Penn’s criticism that folk psychological explanation is not a good basis for scientific investigation)6 Moreover, if we are happy with folk expertise as applied to human infants, then

we should similarly be happy with it application to other species, especially the primates where its application is licensed by the functional continuity between human minds and those of other species

There is, however, a further scientific worry here, which Andrews doesn’t fully discuss, which is that the personality inventories and other scales used to elicit this folk expertise have been created in the context of a particular (Western, individualistic) scientific way of thinking about human psychology, which may not even apply to all humans (Greenfield 1997; Greenfield

et al 2003) This being so, it becomes much less clear whether these will permit access to a non-human folk psychology that captures how animals might think about each other

What is interesting here is that Andrews & Huss (2014) make this same argument, but from the opposite perspective They point out that “[t]he problematic properties are those that require

a degree of interpretation to identify, those that are still more opaque than transparent Those human properties that currently defy a robust scientific account are also those that are most often cited as problematically anthropomorphic.” (p.716) But this is precisely the problem as Penn (2011) sees it: a folk psychological account is more opaque and requires more interpretation in virtue of the fact that it offers no real hypotheses nor explanation of the specific cognitive

mechanisms involved In their account, however, Andrews and Huss (2014), suggest that

“selective skeptics”, like Penn (2011), regard folk psychological explanations as problematic simply because they offer potentially false accounts of human behaviour, and thus we compound this error when we apply such explanations to other species Andrews and Huss (2014) therefore suggest that this anti-anthropomorphic stance ultimately cannot work in the way the skeptics would like because “the distinction between folk psychological concepts and scientific

psychological concepts will not map onto the distinction between anthropomorphic human properties and shared properties”, but this is not the argument being made The skeptics are not arguing that folk psychological attributions are anthropomorphic and therefore impermissible Rather, they are arguing that folk psychological attributions are impermissible because they fail

to identify any kind of cognitive mechanism that could be put to the test, whether these are found

in human or non-human animals The ‘mental representations’ that researchers like Povinelli and