Sunday, May 30, 2010

What makes us different from animals?

The origin of so much bad thinking about both humans and animals is a recurring tendency to try to understand both in the same terms. Animal behaviour is incessantly anthropomorphized, while many try to explain human behaviour in solely biological-evolutionary terms.

So I thought it would be worth noting a recent report on a study into the difference in human and animal learning, and then comparing this to a more expansive analysis of the same issue. Curiously, two of the authors under discussion have almost, but not quite, the same surname.

First, the recent report on a study by Nielsen and Tomaselli: Copycat behavior in children is universal and may help promote human culture:
ScienceDaily (May 26, 2010) — Children learn a great deal by imitating adults. A new study of Australian preschoolers and Kalahari Bushman children finds that a particular kind of imitation -- overimitation, in which a child copies everything an adult shows them, not just the steps that lead to some outcome -- appears to be a universal human activity, rather than something the children of middle-class parents pick up. The work helps shed light on how humans develop and transmit culture.

Scientists "have been finding this odd effect where children will copy everything that they see an adult demonstrate to them, even if there are clear or obvious reasons why those actions would be irrelevant," says psychologist Mark Nielsen, of the University of Queensland in Australia. "It's something that we know that other primates don't do." If a chimpanzee is shown an irrelevant action, they won't copy it -- they'll skip right to the action that makes something happen.
[...]
For the experiments, the children were shown how to open a box -- but in a complicated way, with impractical actions thrown in. For example, the adult would drag a stick across a box, then use a stick to open the box by pulling on a knob -- which is a lot easier if you just use your fingers. Most of the children copied what the adults did, even if they'd been given the opportunity to play with the box first and figure out how it worked. This was just as true for Bushman children as for the Australian children.

But aren't the children just following the rules of what appears to be a game? "That kind of is the point," says Nielsen. "Perhaps not a game, but certainly, when I demonstrate the action, it's purposeful. So from the mind of a child, perhaps there's a reason why I'm doing this." This willingness to assume that an action has some unknown purpose, and to copy it, may be part of how humans develop and share culture, he says. "Really, we see these sorts of behaviors as being a core part of developing this human cultural mind, where we're so motivated to do things like those around us and be like those around us."

And now, for a deeper understanding of this, let's turn to Richard van Oort's analysis of the work of Michael Tomasello:

Tomasello calls this type of [chimpanzee] social learning emulative rather than imitative. The difference is that in emulative learning the disciple focuses not on the model’s particular behavior but on the objects with which the model is interacting. Chimpanzees, for instance, are very good at observing other chimpanzees interact with objects, but they do not then imitate the other’s behavior with respect to the objects involved. Rather, they are led to interact with the objects and discover for themselves the natural affordances of the particular objects attended to. For example, a young chimp may observe its mother crack a nut with a stone. It will then pick up a stone and discover that the stone makes a pretty good hammer. What the chimp has learned is not a particular behavior (the mother’s technique of nut cracking), but a fact about stones, and more precisely, a fact about the impact of stones on nuts. This is something that is learned by the chimp in its interaction with the stone and the nut, not by imitating its mother’s gesture toward the nut or stone. The chimp does not oscillate its attention between mother and the objects involved, in a conscious effort to reproduce her particular gesture. It therefore has not learned a new behavior by imitation. Hammering is something the chimp can do individually, given the natural affordances of the objects involved. Provided with a stone and a nut, and given the general primate capacity for grasping objects (Thank God for the opposable thumb!), the chimp discovers how to crack a nut. The behavior is emulative rather than imitative because, as Tomasello puts it, the chimp "focuses on the environmental events involved--the changes of state in the environment that the other produced--not on a conspecific’s behavior or behavioral strategy" (29).


In order to test his theory that chimpanzee learning is emulative rather than imitative, Tomasello devised a series of ingenious experiments testing both two-year-old children and chimpanzees. The experiments involved getting the subjects to imitate the behavior of a model. Tomasello describes how the children almost always insisted on imitating the behavior of the model, no matter how bizarre or inefficient it was. For example, if the model switched on a light by using her head, so would the children. Or if the model used a tool in an extremely inefficient fashion to reach an object, the children would use the tool in the same inefficient manner, despite the fact that the natural affordances of the objects presented a much easier way to do the same task. The chimps, on the other hand, simply experimented with the objects no matter which way had been demonstrated to them beforehand. Evidently, the children were imitating the model whereas the chimps were attempting to emulate the outcome of the experiment independently of the model’s particular behavior. Whereas the children were focused on the model’s behavior toward the goal, the chimps were focused on the outcome of the experiment. The difference is important because it explains why chimpanzees have such great difficulty learning to use symbols. The ability to separate behavior from outcome is necessary before the model’s gesture toward the object can be transformed into a genuine symbol that designates or "means" the object. What Tomasello’s experiments strongly suggest is that children, but not chimps, are predisposed to focus on the model’s behavioral stance toward the object. They are entering into the model’s particular intentional stance toward the object.

Joint Attention

Key to Tomasello’s ideas about language acquisition among children is his hypothesis concerning the joint attentional scene. Before nine months, children interact with the world much as other primates interact with the world. That is, they are aware of the objects around them and of other individuals interacting with those objects, but they never enter into the other’s intentionality toward those objects. If a pre-nine-month-old child is playing with an object and an adult walks into the room and says, "Look, let’s play with this!" while holding out a toy car, the child may reach out and grab the car and start sucking on it, or manipulating it, or whatever, but it pays no attention to the adult’s intentions toward the toy. It pays attention either to the toy or to the adult, but not to the relationship between adult and toy. In other words, the child does not think to itself, "Oh, mommy wants to show me this new toy," or even, "This person wants to show me this thing." On the contrary, if Tomasello is right, it does not even "think" at all, at least not in the way an adult thinks, which according to Tomasello is an internalized version of the kind of joint attentional scenes children first experience at nine months. Thus, in grabbing the toy the pre-nine-month-old child does not look from mommy to toy and back to mommy again. The toy is simply another object to interact with, but it receives no further significance beyond the child’s own interest in it. Tomasello’s argument is that chimps never really go beyond this "egocentric" stage of understanding objects, which is why primatologists never observe chimps pointing in the wild. That is, they do not distinguish between my intention toward objects and your intention toward objects, so there is no point in trying to get you to pay attention to my attention toward the object. As far as chimps and pre-nine-month-old children are concerned, there is only the point of view of the self, into which all other perspectives are innocently absorbed.

At around nine months of age, however, children begin to engage in what Tomasello calls joint attentional scenes. Initially, this begins with simple checking on the attention of an adult in relation to an outside object, but it quickly evolves into gaze following, when the child looks from adult to where the adult is looking, and to acts of pointing, when the child tries to direct the attention of an adult to some external object. What Tomasello is keen to stress in this ontogenetic "revolution" is the fact that the scene is fundamentally triadic in structure: "Joint attentional scenes are social interactions in which the child and the adult are jointly attending to that third thing, for some reasonably extended length of time" (97). The child’s attention shifts between the adult and the object to which both adult and child are attending. In this collective sharing of attention toward a central object, Tomasello sees the roots of symbolic culture, including language, symbolic play, and ritual.

Joint attentional scenes, however, are not examples of language, at least not language in the sense usually intended by philosophers or linguists.(4) They are rather the minimal condition of language. On the other hand, nor are they simply perceptual events of the kind that nonhuman primates and other animals engage in. All animals, including of course all humans, perceive the world around them and are able, on the basis of those perceptions, to form sensory-motor representations that allow them to anticipate events in the world, including the actions of other conspecifics. However, these anticipations are still perceptually based, in the sense that they are individually learned image schemas or sensory-motor representations. To borrow a usage from the evolutionary anthropologist Terrence Deacon, these perceptual and sensory-motor representations are indexical. They are based on the capacity of all animals to form categories of perceptual events, including categories of communicative events, such as the widely publicized example of vervet monkey distress calls.(5)

The scene of joint attention is quite different. Indeed, Tomasello claims that it bridges the gap between perceptual representation (which we share with all animals) and language (which only we possess). What differentiates the joint attentional scene from language in the narrow sense employed by linguists is the fact that language "abstracts" from the scene to include only its most portable aspect, which is the symbol or word itself. On the other hand, what differentiates the joint attentional scene from perceptual events is that it includes "only a subset of the child’s perceptual world" (97). That is, the joint attentional scene focuses the child’s attention on a central object, against which all other perceptual objects and events become background or "periphery."

Let me emphasize this difference between perceptual events and the joint attentional scene. Animals can of course focus their attention on discrete objects or events in the world, as when a cat tracks the movements of a mouse, or a chimp warily eyes the presence of a male rival. In the joint attentional scene, however, the child is not merely paying attention to the object, but to someone else’s attention toward the same object. In other words, the child grasps that the significance of the object is mediated by the attention of the other. It is this capacity to separate the other’s intention--his or her internally represented goal--from the perceptual reality of the object that distinguishes the joint attentional scene from otherwise superficially similar perceptual scenes. The child learns that the adult’s intention to the object is distinct from its own intention toward the same object. Moreover, in making this distinction between self and other, it lays the foundation for participating in an intentional relation that is truly collective or intersubjective, because in recognizing the difference between the other’s intentionality toward the object and the object itself, the child learns to take a perspective on the object distinct from its own. The child is now imitating a particular intentionality toward the object that is transposable to other scenes in which the object may appear. Tomasello calls this "role reversal imitation" (105) because it implies that the child is able to grasp that an adult’s intentional stance toward an object is something that can be adopted by the child itself. This is something we do all the time--indeed, whenever we use language. For example, suppose you tell me that the peculiar thing on your dining room table is a "grazza." Later, when my wife walks in the room and makes a face while staring at the peculiar object on the table, I turn to her and say, "Oh, that’s a grazza." I have not merely imitated the word, I have also reproduced your intentional stance toward it. That is, I have recreated the joint attentional scene by adopting your perspective, and this time I have reversed the roles because I am now instructing someone else, as you had instructed me before.

A skeptic might object that these joint scenes of attention are not so very different from the attentional scenes other animals engage in. Animals are not rigidly tied to the same perceptual construal of a particular object. A tree may represent a number of different things. Depending on the context, it may represent an escape route, a nesting site, or the location of food. But Tomasello’s point is not that perceptual scenes are inflexible but that they are always tied to the natural affordances of the objects and that these affordances are discoverable by the individual’s dyadic interaction with the object. At no point does the chimp seek confirmation from another conspecific in order to see that the tree is a nesting site or an escape route or source of food. It can discover these things for itself. Furthermore, it is impossible for the tree to represent all these things at the same time. The chimp does not choose between different representations of the tree that it can hold in its mind simultaneously. Rather, the representation of the tree remains a function of the chimp’s particular goal, which is either to eat, escape, or sleep. As Tomasello says, "the animal is attending to different affordances of the environment depending on its goal" (126).

In the case of symbolic attention, however, the goal is not defined by the practical affordances of the environment, but by the attentions of both individuals in the attentional scene. When a child points to a tree, the goal is to secure the adult’s attention to the same object. And this is, in the end, what language does. It secures the other’s attention toward some external object or, in the case of declarative sentences, some external idea or "signified." When it comes to specifically human cognitive functions, what is primary, as Emile Durkheim saw, is the social relation. It is the latter that mediates our more basic indexical perceptual and sensory-motor functions. The latter are basic functions that we share with all other animals. It is the mediation of these functions by the joint attentional scene that distinguishes human from animal cognition.

No comments: