Photo by Vincent J Musi/National Geographic


The pointing ape

How a chimpanzee named Clint trained a psychologist to question human exceptionalism and reconsider the intelligence of apes

by David Leavens + BIO

Photo by Vincent J Musi/National Geographic

Human language, as far as we know, is unique in the animal kingdom. There has been long-standing academic debate over whether linguistic communication should be viewed as a biological adaptation or a cultural invention. On the one hand, the range of human articulation far surpasses that of our nearest living relatives, the great apes (chimpanzees, bonobos, gorillas and orangutans), suggesting that we are evolutionarily adapted for speech. On the other hand, there are thousands of human languages, most of them mutually unintelligible, consistent with the idea of cultural selection. Both views, however, are rooted in the idea that humans are neuroanatomically and cognitively distinct from other animals.

We last shared a common ancestor with the other great apes around 6 or 7 million years ago. Yet our brains were, essentially, ape-sized until around 2 million years ago, and did not reach contemporary proportions until sometime within the past 200,000 years. Material evidence suggests that humans used ochre to make drawings on rocks only within the past 75,000 years; and representational objects began to appear in the archaeological record about 40,000 years ago. Unfortunately, speech and gesture do not fossilise, but if we consider that no ape can speak (beyond a tiny handful of laboriously produced utterances) and if we assume that language requires a brain larger than an ape’s, we can define a window of time between, roughly, 2 million years ago and 40,000 years ago in which spoken language emerged. Based on a range of data from archaeological to neurophysiological, the computational neuroscientist Michael Arbib estimates this window at around 1.5 million to 100,000 years ago. Somewhere in this time span, we became qualitatively distinct from all other animals by any reasonable measure.

But to what degree are humans truly unique? Psychology tells us that humans, alone among animals, have the capacity to theorise about the contents of other minds. If you’re reading this essay and wondering where I’m going with this, you’re displaying this capacity. We humans also regulate our own behaviour based on the outcomes of such computations. This is a core tenet in many branches of the cognitive sciences today: the idea that our mentations cause our behaviour. In terms of our language behaviour, human children are magnificent test subjects because every child who masters a language (and this describes the overwhelming majority of humanity) transitions from being a creature without any apparent capacity for symbolic communication, akin to other animals, to being a creature who can skilfully produce and comprehend complex utterances that are, apparently, unique in the world. If we can understand the changing competencies of human children, then, the argument goes, we can discern those infant and toddler capabilities that facilitate this language learning. We can glean the ‘psychological toolkit’ that human babies apply to their social environments to produce their native languages.

Because it is so easy to study children, the literature on this issue is immense. A sub-area in this active research domain involves identifying the competencies present in preverbal children while absent from our living relatives, the great apes. An ability displayed by preverbal children but not adult great apes would be seen as an adaptation unique to us. For decades, the sine qua non of human preverbal communicative exceptionalism was the pointing gesture. A language-competent individual can name an entity or event to which she would like to draw the attention of her social partner; a preverbal child armed with a pointing finger can accomplish much the same.

In the early 1990s, it was a nearly universal axiom in psychology that pointing was a human adaption for creating a ‘referential triangle’ between two people. At that time, I had no particular reason to doubt this story, but quite by happenstance I met someone who gave me grounds to reconsider pointing as a human adaptation in the human toolkit for language. That ‘someone’ was Clint, an adolescent chimpanzee, and this is the story of how he trained me to question the mainstream scientific perspective on pointing as an evolved cognitive adaptation for the acquisition of symbols.

I should confess right up front that I fell in love with Clint. When I met him, in 1994 at Georgia State University, he was a cheerful, rambunctious 14-year-old chimpanzee. His home cage was right near the entrance to the great ape wing and, as soon as he’d see me come in, he would greet me in his boisterous, hail-fellow-well-met fashion, standing on all fours, head bobbing up and down. I would wheel a computer cart up to his cage, start the software program, and then usually we chased each other along the side of the cage playing grab hand, or taking turns tickling each other. Then I’d wander off to set up a second computer for another chimpanzee in a different part of the facility, leaving Clint to play his computer games, returning from time to time to refill his automated reward-delivery system with peanuts and grapes. At the time, I had a bachelor’s and a master’s degree in anthropology, but was in the process of switching to psychology. It seemed that my anthropology degrees qualified me to make coffee for experimental psychologists and do the bidding of chimpanzees.

Handsome Clint; Rest in peace. Photo courtesy Bill Hopkins

One day, I came back to Clint’s cage and saw him point with his index finger at a grape that had fallen on the floor, due to a technical problem with the automated reward-delivery system. The grape was out of his reach, and he pointed to it, making loud raspberry sounds (like a Bronx cheer), looking back-and-forth between me and the fruit. Now, you don’t need a PhD in experimental psychology to be able to interpret this signalling behaviour, right? However, without significant indoctrination into late 20th-century intellectual fashions you might not realise that Clint’s pointing was, at the time, theoretically impossible. Almost everybody knew, at that time, that human pointing – this ability to capture and redirect the attention of another being to a specific entity – was part of our unique adaptation for language. In language, we refer to things with words. We say, for example: ‘Look at the dog!’ or ‘I think that blue car is following us.’ Symbolic reference seems easy enough as we speak in daily interactions, but hop on a plane and travel somewhere where people don’t speak your language, and the same words just don’t work for us the way they do back home. This is because languages are, for the most part, mutually unintelligible. The whole system breaks down if we don’t have a shared symbolic code. But pointing will often work to establish joint reference.

Clint points repeatedly at malfunctioning computer. In Panel A, Clint quietly awaits the return of the experimenter, holding on to the cage mesh with his left hand, without gesturing. In Panel B, as the experimenter enters (offscreen), Clint rotates his hand and extends his fingers in a point to the computer monitor. In Panel C, as the experimenter leans in to peer at the monitor screen (arrow), Clint relaxes his fingers. In Panel D, Clint has retracted his hand and points again. Images originally recorded on VHS videotapes. From Developments in Primate Gesture Research (2012) by David Leavens, p190. Courtesy John Benjamins Publishing Company

This highlights different kinds of reference. In speech, there is a mostly arbitrary relationship between a symbol and the thing it refers to. The word ‘big’ is not bigger than the word ‘little’, for example. The word ‘dog’ does not sound like a dog, and so on. In contrast, the relationship between a pointing gesture and its referent is not arbitrary – the pointing hand acts like a geometric ray, so that while a point might not usually resemble the referent, it nevertheless has a spatial relationship with it. Pointing is an interactive skill in human infancy. Children begin to follow pointing gestures to targets in their fields of view by about nine months of age; by approximately 12 months of age, they can follow points to more distant objects or locales. Children also begin to produce pointing gestures for others, at roughly the same age that they begin to speak, around the end of their first year of life.

The consensus was that pointing was part of our species-unique neurocognitive adaptations for speech

In the mid-1980s, a series of studies by the education scholar Peter Mundy in Miami and others had established that aspects of children’s pointing behaviour correlated with their later vocabularies. Children can use another’s pointing gesture to focus on an entity while their interlocuter states the name of the object, forming visual-auditory associations. Children can also elicit verbal labels for objects when they, themselves, point and their caregivers respond (‘Yes, that’s a doggie!’). Children’s production and comprehension of pointing, in Western cultural contexts, significantly predicts their later language skills.

The British psychologist George Butterworth and others, meanwhile, had shown that children’s pointing was biased towards the right hand, and thus under the control of their left cerebral hemispheres (because the left side of the brain controls movement on the right side of the body, and the right side of the brain controls the left side of the body). Because speech is also, in most people, significantly represented in our left cerebral hemispheres, some researchers began to view pointing and speech as functionally linked in the brain, part of the same alleged evolutionary adaptation for establishing joint reference. Thus, in the 1990s, there was an emerging consensus that pointing was part of our species-unique neurocognitive adaptations for speech.

Finally, it was widely asserted by numerous researchers that humans’ nearest living relatives, the great apes, do not point. In his book The Lopsided Ape (1991) outlining a theory of language origins, the psychologist Michael Corballis in New Zealand noted that ‘pointing with the outstretched arm and index finger at visual objects in visual space seems to be unique to humans’. That same year, the Canadian psychologist Merlin Donald argued that only humans displayed intentionality in their gestural signalling, a conclusion justified, allegedly, by an ‘absence of spontaneous pointing behaviour’ in home-reared chimpanzees.

The 1990s saw an explosion of theories about human cognitive evolution, most of it based, in part, on the assumption that we, uniquely, could coordinate the attention of two or more individuals by pointing. These ideas had, at that time, a certain prima facie validity: pointing creates a sort of referential triangle between the pointer and the observer; only humans were known to routinely communicate with referential gestures about specific events and entities in the world; pointing seems to facilitate the acquisition of speech; pointing tended to reflect the same left cerebral hemisphere dominance as speech in most humans; and nobody had reported pointing as part of the gestural repertoires of great apes. Pointing was seen as a bridge to speech and an adaptation unique to humans, one that evolved around 6 million years ago; that is, after our split from the other great apes.

Yet here was Clint, a chimpanzee, casually refuting a suite of sophisticated theoretical models. I was excited by the apparent pointing behaviour, and I ran down to the laboratory office to report my observations to the lab director, Bill Hopkins. As it happened, he was relatively unmoved by my report, and the reason for that is a key part of this story: he had worked for years at the Language Research Center at Georgia State, which housed the renowned language-trained apes Sherman, Austin and Lana (all chimpanzees) along with Kanzi, the language-trained bonobo. (Bonobos used to be known as pygmy chimpanzees, Pan paniscus, and are a distinct species from common chimpanzees, Pan troglodytes.)

It turns out that apes who have been cross-fostered by humans very frequently point during interactions with people. So, initially, Hopkins didn’t find the observation particularly exciting. A few days later, though, Kim Bard, the director of an adjacent laboratory, was visiting the office, and Hopkins turned to me and said: ‘Hey, Leavens, why don’t you tell Dr Bard about your [pause for comic effect] observation.’ So, I did and Bard – a developmental psychologist – recommended that we study the behaviour, given its prominence in theories of human cognitive and communicative development. We set out to determine whether Clint’s apparent pointing behaviour was (a) communicative and (b) intentionally communicative.

At that time, intentional communication was almost universally believed to be (no surprise, here) a cognitive specialisation unique to humans. Why that was the case, and how people had come to believe that they were measuring intentional communication in our own children is a long and fascinating intellectual story that has not yet been told in full. Here, I will cover the high points.

There are two ‘flavours’ of intentionality in contemporary scientific discourse: one is philosophical, and the other – folk-psychological. When philosophers write about intentionality in communication, the gist of it is that people are construed as having minds that exist in relation to things in the world, real or imagined; that is, there is a quality of ‘aboutness’ between states of mind and the entities with which minds are involved.

Almost all contemporary cognitive science and psycholinguistics, then, assumes that entities that we perceive and about which we have feelings, beliefs, attitudes and so forth are represented in a kind of inner conceptual archive, where representations of these entities are stored. Whether this archive is made up of concepts that we acquire through our social environments or whether instead we have concepts, first, and language gives us the ability to make them explicit is one of the central debates of the previous century. Linguistic reference is – broadly speaking – the act of making my invisible, mental representation of an entity explicit through speech (or sign) so that your invisible, latent mental representation of that entity is somehow brought before your ‘mind’s eye’. The debate around this subject fills library shelves worldwide, but the heart of the matter concerns the assumption that a multitude of inner mental elements is mapped on to our public, shared world. Whether people argue for a ‘language-first’ or ‘concept-first’ sequence, there seems to be broad agreement that intentionality involves minds having relationships with the world outside the body. Let’s call this Intentional Communication Type 1.

Psychologists had charted the developmental pathway in children from middle-class, Western families

A second relevant aspect of intentionality comes from Western folk psychology. In this worldview, people are seen as acting for reasons, and intentions are – painting with a very broad brush, here – both motivations for and causes of action. Thus, at the dinner table, a person might point to a salt shaker. A folk-psychological interpretation of this quotidian vignette is that the signaller desires salt, and uses the gesture to manipulate a dinner partner to deliver the salt shaker. The idea is that the signal is used as a social tool. The child pointing to a distant milk bottle, the ape pointing to an unreachable grape – these are examples of the use of gesture in goal-directed sequences. This, we’ll call Intentional Communication Type 2.

How does all of this relate to Clint’s pointing? At that time, 25 years ago, developmental psychologists had charted the typical developmental pathway towards what they called intentional communication, primarily in children from middle-class, Western families. The onset of pointing by these children, typically around the end of the first year of life, was taken as evidence for a kind of nonverbal reference: evidence that even human infants displayed a mind-world relationship, held to be (you guessed it) unique in the animal kingdom. Pointing was seen as a developmental bridge to symbolic reference, a way to make entities manifest for mutual contemplation.

In parallel to this interest in the development of Type 1 intentional communication, other researchers were interested in how pointing indexed communicative motivations, or Type 2 intentional communication. At that time, some linguists were characterising utterances in terms of their objectives. For example, directives were utterances used to request actions; expressives to broadcast one’s psychological state; and representatives to assert something about the state of the world, among other motivations.

In 1975, the American psychologist Elizabeth Bates and her colleagues distinguished two central motivations for communicative pointing in young children: protoimperative (issuing a directive) and protodeclarative (expressing a feeling or thought). These terms were described as ‘proto’ because they were seen as continuous, spanning from preverbal communication through the later epoch of true speech. In this classic and immensely influential paper (along with numerous follow-up studies), the authors characterised pointing by human children as reflecting these communicative goals, thus linking linguistic utterances with nonverbal communication; in their view, children evinced these communicative motivations in their preverbal communicative behaviour before they developed linguistic means to express these same motivations. To put it another way, the kinds of communicative goals revealed by analysis of adult speech were evident in human infants, suggesting psychological continuity across pre-speech and post-speech epochs.

This was a significant advance, primarily because Bates and colleagues seemed to link early communication to well-established findings in human cognitive development. Late in the first year of life, human children begin to display relatively sophisticated goal-directed activities known as ‘coordinated secondary circular reactions’. For example, if reaching for a toy, they might brush aside a bottle that is situated between them and the toy. This kind of coordination, putting together the moving of the bottle and then displaying a reach-and-grab for the toy is an advanced skill in human infancy.

Bates and her colleagues had demonstrated a growing ability to relate particular means to achieve particular ends. Protoimperative communication (pointing to request an unreachable toy) was the infant’s use of a social agent (the means) to obtain a desirable object, whereas protodeclarative communication (pointing to express delight at a dog chasing a ball) was the infant’s use of a signal (the means) to obtain a caregiver’s response.

This kind of behaviour has been demonstrated in animals, too. One of the first to apply this framework to animal communication was the Dutch psychologist Frans Plooij, who studied the development of communication between infant and mother chimpanzees at Gombe National Park in Tanzania. In 1978, Plooij reported that chimpanzee mothers will raise their infants’ arms above their heads, to facilitate grooming along their flanks and under their arms, but older juvenile chimpanzees will approach their mothers and adopt this posture in apparent attempts to request grooming from their mothers. Later, Bard (who, recall, encouraged us to study Clint’s pointing behaviour) studied the development of how young, Bornean orangutans requested food from their mothers.

Plooij also noted two examples of protodeclarative communication, where chimpanzee infants communicate to elicit attention to themselves: leaf-grooming and running away with an object. As Plooij described it, leaf-grooming is when:

an individual will suddenly, as if out of the blue, take a leaf and begin grooming it intensively. In the majority of cases, this results in another individual coming over to him or her and observing closely what is going on. Nothing special was ever observed to be on the leaves groomed in this way: it seems more likely that what is special is nothing intrinsically to do with the leaf itself, but more to do with the role leaf-grooming plays in social interaction. The leaf-grooming individual obtains the attention of others.

Of course, anybody who has watched children will have seen them take an object and run away with it, while looking back at their social partners. Plooij described this quintessentially protodeclarative behaviour in infant and juvenile chimpanzees.

This gazing back at the social partner is what signals to an observer that the behaviour is social in its function, that communication is directed to some specific individual. Human children, from Western cultures where the development of pointing is most studied, tend to develop this concomitant visual checking behaviour somewhere around the beginning of their second year of life. This gaze alternation between a referent and a social partner is a key behavioural marker of intentional communication in our own species. Plooij showed that wild chimpanzees also displayed this visual checking.

Yet by the summer of 1994, when I began my work with Clint, nobody had ever set up a video camera and manipulated the presence and absence of an observer to see whether apes, or any other animal, used manual gestures as explicitly communicative signals – and this seemed to me to be an astonishing gap in the scientific literature. If we were going to convincingly establish whether Clint’s finger extensions towards food were explicitly communicative gestures, then we would need to know what he was doing when unreachable food was visible, but there was nobody around to retrieve it for him. We reasoned that if Clint extended his fingers towards unreachable food in the absence of somebody situated to retrieve it, then this would falsify the interpretation that Clint was pointing, and imply that we had been richly overinterpreting his behaviour.

This question exemplifies the heart of scientific enquiry – develop a hypothesis (Clint is pointing), then set up a situation that would make it possible, in principle, to emphatically reject that interpretation.

Clint was signalling, in frustration, that he’d tear the apparatus to pieces – if only he could get his hands on it

Here, then, we had a series of research questions that we could ask about Clint’s pointing behaviour: (a) we could determine whether Clint ‘pointed’ when there was nobody around, establishing clearly whether the apparent signals were communicative; (b) we could measure whether Clint alternated his gaze between the apparent objects of his points and the human observers, thus determining whether his visual orienting behaviour formed a pivot point in a nonverbal referential triangle, like human children do; (c) we could also measure whether Clint accompanied his apparent points with calling behaviour, as humans often do too – we will call out or bang on something with our hands, apparently to elicit somebody else’s attention; and finally, (d) we could ask whether, like human children, Clint tended to persist in signalling when his initial points were not followed by delivery of the requested food.

So, in late July, while Clint was continuing to play computer games for an hour or two every day, we set up a video camera outside his home cage to capture his behaviour, both when an observer was present and when absent. Meanwhile, the automated reward-delivery apparatus continued to occasionally and randomly misdeliver peanuts or grapes to the floor, so we had large blocks of time captured on film where Clint was in the presence of unreachable food, but nobody was around for him to ask to deliver it to him. We coded whether or not Clint stuck his fingers through the cage mesh, whether he displayed gaze alternation when doing this in the presence of an observer, and whether he displayed calling or other auditory behaviour while extending his fingers through the mesh.

The outcome of the study was unambiguous and, in some ways, surprising. For example, the primary experimenter (yours truly) was in the habit of carrying around peanuts in his lab coat pocket, and both of Clint’s cagemates, Anna and Flora, frequently pointed to that pocket. Therefore, we also counted their apparent points, in addition to Clint’s.

With respect to the question of whether the chimpanzees extended their fingers towards food, the results were clear. The three chimpanzees displayed 256 finger extensions directed outside their home cage: only two (less than 1 per cent) were displayed in the absence of a human observer. Those two actions are worthy of comment: after moving Clint from a relatively easy discrimination task on the computer to a more difficult discrimination, his error rate increased, dramatically. After a run of incorrect responses, accompanied by a buzzer and no reward delivery, Clint stood upright and jabbed the fingers of both hands towards the reward dispenser, twice in quick succession – it was as if he was signalling, in frustration, that he would tear the apparatus to pieces, if only he could get his hands on it. Anyone who has ever struggled with a faulty tool or device can sympathise with him, I think.

So, we can say with a high degree of certainty, these finger extensions were communicative signals. We coded gaze alternation only from Clint (because the other two chimpanzees were frequently signalling from a raised platform, and their heads were not in view), and we found that 76 per cent of Clint’s 167 pointing gestures were accompanied by gaze alternation between the human observer and the referents of his points; this is comparable to pointing by human children, after about one year of age. In contrast, Clint vocalised during only 24 per cent of his points, which is substantially less than human children, when they point – we think that, perhaps, Clint was aware that he had our attention, so didn’t discern a need to display attention-getting behaviour. Finally, we determined that Clint was significantly (in a statistical sense) more likely to display a second pointing gesture if the first wasn’t followed by delivery of the requested item. Thus, by these measures, Clint’s finger extensions were pointing gestures, and they were intentionally communicative by the same criteria that human pointing gestures are considered to be intentional: they were used only in the presence of a human, they were accompanied by gaze alternation between the human and the referents, and Clint persisted in signalling more often when initial communicative bids failed.

I was over the moon with these findings. I knew that we had the first experimental study of pointing in great apes, the first to establish baseline behaviour in the absence of an observer, and among the first to provide strong evidence of intentional communication in great apes (indeed, any animal species) under controlled, experimental conditions. Bursting with enthusiasm and intellectual pride, a few months later I was putting the finishing touches on our manuscript describing these unprecedented findings, when Hopkins happened to ask me if I’d seen a paper by Josep Call and Mike Tomasello in the Journal of Comparative Psychology on orangutans, an Asian great ape species. I had not seen it, so I raced over to the library and dug out the December 1994, issue, and found their report, entitled: ‘Production and Comprehension of Referential Pointing by Orangutans (Pongo pygmaeus).’

My heart sank all the way down to my ankles, as it became clear that we had been scooped, and scooped very competently. Unlike our study, Call and Tomasello had measured whether their two orangutans understood pointing, which was a question that we had not considered. Unlike our study, they had experimentally manipulated the location of the referents (treats in boxes), whereas we had just left that to the vagaries of chance. Also, unlike in our study, they presented a condition in which the apes had to indicate the location of a tool that a human observer needed to access the containers that had been baited with treats, which was an unprecedented scientific finding. Like our study, they filmed the animals in the absence of a human observer, so we were even scooped there.

In essence, they reported that a sign-language-trained orangutan was much more competent at using and understanding pointing gestures than another orangutan, who had been born and raised in a biomedical research centre. Overall, their study was innovative, important and very competently conducted, and for weeks I was pretty much inconsolable. (I felt like the time when, as a teenager bursting with pride over my first motorcycle – an old, beat-up Honda 350cc – I drove 100 miles south to San Jose in California, to see a girl I fancied. I showed up, covered in road grime and oil, only to see her boyfriend drive up in a shiny, new, yellow Chevrolet Camaro, and I knew that I had no chance with this young woman.) But, eventually, I regained perspective and saw some of the unique strengths of our study, which was published in December 1996, also in the Journal of Comparative Psychology.

Since then, with my collaborators, I’ve studied manual gestures, mostly pointing, in hundreds of chimpanzees, about 100 human infants, and more than 1,000 human adults. It turns out that the patterns that Clint displayed are commonplace in captive chimpanzees. In a 1998 study, Hopkins and I showed that 53 out of 115 chimpanzees who were presented with unreachable food pointed to it. In 2004, we filmed 101 chimpanzees to demonstrate that, in three different experimental conditions, they didn’t gesture when nobody was around to see them do it. In 2005, in a study with 29 chimpanzees, we showed that if we gave them the banana they requested, 100 per cent of them ceased gesturing entirely, whereas they persisted and elaborated in their signalling – for example, by communicating in multiple modalities – if we gave them less desirable biscuits of primate chow. More recently, in 2015, we demonstrated that chimpanzees do often point to desirable food from a distance, an ability claimed in 2014 by Tomasello, Marloes van der Goot and Ulf Liszkowski to be unique to humans and an index of the cognitive ability to recognise psychological common ground with our social partners. We don’t think this conclusion follows from the evidence at all; rather, we think this kind of pointing simply indicates a learned appreciation of the possibilities afforded by the environment.

The significance of these findings, and related research by other research groups, is profound. Contrary to centuries of argumentation, the ability to highlight an entity for mutual contemplation, to raise a topic, is not uniquely human, nor is a shared symbol system necessary for this activity. In the cognitive sciences, a considerable body of contemporary work is being conducted under the assumption that humans display unique cognitive adaptations for collaborative activity, as indexed by our pointing behaviour. For theorists more interested in questions of what make humans so distinct among primates – and there is no dispute that we are exceptional primates – the utility of pointing as evidence for human cognitive uniqueness has been somewhat diminished.

There is now a near-consensus that apes in captivity do point to request objects (so-called ‘imperative pointing’). Theoretical interest in the human realm has shifted towards what is known as declarative pointing, drawing attention to something for the benefit of a social partner. One point of view contends that declarative pointing developmentally prefigures later commenting (‘It’s raining’) or signs of uniquely human altruism, as when children point to dropped keys for which an adult is looking. This view of declarative pointing in humans, called the ‘rich interpretation’, contrasts with the original ‘lean’ interpretations by Bates, who portrayed pointing by preverbal children as an instrumental act to elicit some kind of response from adults.

Many animals point: dogs point (there are even breeds called pointers), dolphins point, Australian magpies point

There’s evidence to support the ‘rich interpretation’, but it isn’t absolute. For instance, Hélène Cochet and Jacques Vauclair in Scotland have studied the handedness of pointing in young children and adults. They find that declarative pointing is significantly more right-handed than children’s manipulative activities; but imperative pointing is about as right-handed as these manipulative acts, which is consistent with the idea that the motivations differ between declarative and imperative points.

Do apes also point declaratively, not just to obtain something but to strengthen the social bond? There are reports of declarative pointing by great apes dating back to the early 20th century, with W N Kellogg and L A Kellogg even publishing a photograph of declarative pointing by an infant chimpanzee in their book The Ape and the Child (1933). Some insist that apes do not naturally point among themselves, even in a transactional way. The pointing we documented in Clint, they argue, is merely an artefact of captivity. The truth is nuanced: wild apes have, rarely, been shown to point in both declarative and informative ways. Yet it is also true that informative pointing is commonplace only among precisely those great apes who have had the most intensive exposure to Western, middle-class patterns of nonverbal communication: language-trained apes.

What we think Clint and the other pointing apes have to tell us is that pointing emerges in situations in which it is useful, and that utility is defined outside the brain of the pointing individual, be it ape or human. Our babies spend a lot of time in conditions of physical restraint, strapped into car seats, feeding chairs, backpacks, slings and so on, so they encounter interesting entities on a daily basis that they cannot directly grasp. Pointing emerges in this referential problem space, an intrinsic feature of the lives of babies, at least in Western, postindustrial societies.

Wild apes, in contrast, can freely locomote almost anywhere in their environments, starting as young as about six months of age. When we take great apes and put them into enclosures, they see objects they desire, and they develop tactics to manipulate people outside the cages to deliver those items. In this referential problem space, pointing emerges. Pointing, thus, implicates features of the signaller’s directly experienced social and physical environments during development; to the extent that pointing reflects cognitive activity, key elements of that cognitive process – such as barriers to action – exist outside the brain and body of the pointing individual. Pointing is a behaviour shaped by desire and circumstance.

The old idea of pointing as evolutionary adaptation for referential signalling is significantly challenged by the finding that, for chimpanzees, pointing is very common in research centres and zoos while extremely rare in the wild. Moreover, it turns out that many animals point, often using the long axis of their bodies: dogs point (there are even dog breeds called pointers), dolphins point, Australian magpies point, and pointing has been reported in at least 12 different nonhuman primate species.

What this means for understanding the evolution of language is that our ancestors didn’t have to evolve a specialised manual gesture to foster language acquisition in that window of time when language or protolanguage originally appeared, 1.5 million to 100,000 years ago. Our ancestors who invented language were already pre-adapted for pointing, and as our infants became ever more helpless for ever longer periods of early development, pointing became an increasingly useful tool for social manipulation in our species.

Clint died young – he was only 24 years old when he passed, still in his prime; chimpanzees in captivity routinely live into their 50s and 60s. His contributions to science were enormous: he was the first chimpanzee to have had his genome sequenced, and he participated in scores of studies, including some of the first applications of personal computers to animal learning and many of our studies of chimpanzee communication. As we’ve seen, he instigated a significant line of research into how chimpanzees use gestures to manipulate people outside their cages, by training me to retrieve food for him. He was also my friend; his impact on me as both a person and a scientist was profound. He lived in a cage, but it was my privilege to serve as Man Friday to a chimpanzee.

Dedicated to the memory of Clint, a chimpanzee (1980-2004).