1

Joint Attention, Communication and Mind

Naomi Eilan

Sometime around their first birthday most infants begin to engage in relatively sustained bouts of attending together with their caretakers to objects in their environment. By the age of 18 months, on most accounts, they are engaging in full-blown episodes of joint attention. As developmental psychologists (usually) use the term, for such joint attention to be in play, it is not sufficient that the infant and the adult are in fact attending to the same object, nor that the one’s attention cause the other’s. The latter can and does happen much earlier, whenever the adult follows the baby’s gaze and homes in on the same object as the baby is attending to; or, from the age of six months, when babies begin to follow the gaze of an adult. We have the relevant sense of joint attention in play only when the fact that both child and adult are attending to the same object is, to use Sperber and Wilson’s (1986) phrase, ‘mutually manifest’. Psychologists sometimes speak of such jointness as a case of attention being ‘shared’ by infant and adult, or of a ‘meeting of minds’ between infant and adult, all phrases intended to capture the idea that when joint attention occurs everything about the fact that both subjects are attending to the same object is out in the open, manifest to both participants.

The phenomenon of joint attention, under that description, was initially studied by developmental psychologists interested in the development of pre-verbal and early verbal communication during the second year of life, interested, that is, in the emergence of pointing and the transition from pointing to the accompaniment of such pointing with various pre-verbal ‘comments’ and finally, by the time children are two, to the production of basic sentences in the context of short conversations. And the importance of joint attention for the development of such early verbal communications is hard to overestimate. As Bruner (1977, p. 287) puts it, joint attention ‘sets the deictic limits that govern joint reference, determines the need for referential taxonomy, establishes the need for signaling intent, and eventually provides a context for the development of explicit predication’. But commenting on this work retrospectively some thirty years later Bruner (1995, p. 1) notes that there was a sense in which the phenomenon of joint attention itself was simply taken for granted. In particular, he says, ‘[e]pistemological questions [such as how infants and toddlers ‘come to know about other minds or how they come to realize that other minds know theirs’] never entered the discussion’. Nor, he says, did the questions of the nature of the mental concepts the babies and toddlers use when engaging in bouts of joint attention.

Turning now to philosophy, at least some of the epistemological issues raised by the phenomenon of joint attention, or analogues of them, have, of course, been intensely discussed, with respect to adults, under the heading of mutual or common knowledge. The concept question, that is the question of the kind of mental understanding implicated in joint attention, however, has not. Let me first say a few words about the former, and use them to lead on to the latter, which has, in recent years, become of interest to developmental psychologists.

The notion of mutual or common knowledge has been thought of as the key to explaining verbal communication, and more generally rational cooperative activity. Many analyses of it appeal to the possibility of infinitely iterated beliefs in order to explain what such openness consists in—and then try and find ways of either justifying stopping it at some arbitrary level of complexity or of providing subject’s with bases from which they can compute an infinite regress. Thus to take a much discussed example of Schiffer’s (1988), suppose you and I are sitting at a table with a candle between us. In such a situation, in normal conditions we will have mutual knowledge of the fact that we can both see the candle. A typical philosophical analysis of what must be true of me, say, if this is a case of mutual knowledge will ascribe to me at the very least the belief that you see the candle, the belief that you believe that I see the candle, the belief that you believe that I believe that you see the candle. The debate then turns on how one deals with the further, infinite iterations that many feel should follow to explain complete openness, given the finiteness of human minds.

Now, if something along these lines is the right thing to say about the kind of mutual awareness implicated in joint attention, it is hardly surprising that developmental psychologists had shown no interest in it. Not merely because it seems absurd to explain what is going on in the infant’s mind by appeal to the capacity for such infinitely iterated knowledge structures, but because one of the most robust finding in developmental psychology is that children before the age of four just don’t have the concepts of belief and knowledge needed for formulating even the first step. And one response, when presented with such an account, might be—that just shows that it is wrong to describe the toddler as engaging of bouts of joint attention. They may well be capable of bringing it about, causally, e.g. by pointing, that others attend to what they attend to, and they may respond to adults attempts to draw their attention to objects, but that is all there can be to it—in particular the idea of some kind of mutual awareness here must go by the board.

The alternative response is: the fact that babies engage in joint attention just shows, particularly vividly, what is wrong with the iterated belief ascriptions approach to explaining mutuality of awareness here, which is anyway unsatisfactory for the adult case as well. There is something utterly simple and basic about the transparency of our minds to each other in the case of joint attention which this whole kind of account misses. The very idea that we have to iterate beliefs ad infinitum in order to capture the phenomenon of mutual awareness only gets going because of an assumption of basic opacity as a starting point.

There is something very attractive and suggestive about this immediate reaction, but it turns out to be extremely difficult to articulate and make good. For one thing, if the phenomenon of joint attention is to take us forward on the epistemological front this will only happen if we can explain how we should account for the way in which each person represents what is going on in others' mind when it is achieved. Whatever in general we want to say about the kind of mental understanding we find in joint attention, it will presumably involve ascribing to the baby some kind of grip on the notion of attention. But what is attention? Despite the centrality of the concept of attention in our mental life, there has been very little philosophical work on what exactly our intuitive, everyday concept of attention actually means—about how it is connected with, on the one hand, grasp of physical concepts such as concepts of space and causation, and, on the other hand, grasp of mental concepts such as those of perception, action and various affective concepts. There a clear contrast here with other concepts possession of which has been the subject of intense research among developmental psychologists, such as the concepts of belief, desire and intentional action. So, solving for the children’s case with respect to the concept of attention is a matter of simultaneously discovering the role played in our adult common sense psychology by the concept of attention.

However, whatever we say about the child’s (and adult’s) concept of attention in general, this will only help with explaining the kind of transparency of minds we assume exists in cases of joint attention if we can also say something about the kind of social interaction we have in play in such cases, which seems, intuitively, to guarantee such transparency. After all, such transparency is not something we assume always exists, for example when we observe someone’s behavior and use that to form hypotheses about what they must be thinking. So there is something specifically about the joint attention situation, which it shares with other cases of social interaction, e.g. rational cooperation, verbal communication and so forth, in which we want to say that minds are transparent to each other. And this brings us to the third big issue raised by the phenomenon of joint attention.

Many philosophers and developmental psychologists share the intuition that treating the child’s developing mental capacities as emerging in the context of social interaction yields a picture of understanding that is fundamentally different from one on which mental understanding is a matter of theorizing about observed behavior. Here too, though there is something attractive and intuitively suggestive about the idea of a different picture here, it is not at all easy to formulate what this difference is exactly. What exactly is it that turns out to be different in the account we give of our everyday psychological terms, and of their referents, if we hold that mental life emerges in the context of social interaction. Indeed, what exactly do we mean by social interaction? And how does our understanding of development contribute to spelling out the answer?

So, there is the question of the kind of mutual awareness we find in joint attention, in virtue of which we think of minds as being transparent to each other (call this the ‘Epistemological Question’); and there is the question of the kind of understanding of attention which a child at the age of 1-2 might plausibly be thought of as possessing and bringing to bear on these situations (call this the ‘Concept Question’); and there is the question of how we should conceive of the kind of social interaction involved in joint attention, in virtue of which the concept of attention can be deployed in a way that yields mutual awareness (call this the ‘Social Question’). And finally there is the issue of how exactly these three questions are related to each other.

There are, in fact, many ways one might relate them. In what follows I propose one framework for doing so, which in turn yields one way of distinguishing between two different accounts of joint attention. These two types of theory give different answers to our three questions, and relate them in different ways. As I will be setting up issues, the difference turns on how each theory addresses two fundamental issues raised by the phenomenon of joint attention. (a) What, if any, is the connection between the capacity to engage in joint attention triangles and the capacity to grasp the idea of objective truth? (b) How do we explain the kind of openness or sharing of minds that occurs in joint attention? In developing these ideas I will be drawing heavily on the papers collected here. This is not at all to say, though, that the authors would necessarily agree with this way of setting up the issues. And it even less to say that the questions I will be raising begin to scratch the surface of the rich assortment of fascinating issues raised by each paper in its own right.

Part I: Locating the phenomenon of joint attention

To get going, it will help to have before us the following relatively uncontroversial breakdown of the looking patterns that lead up to the emergence of the phenomenon labeled joint attention, and of the attentional behaviours that develop in the course of it maturation. Up until the age of four of five months infants look mainly at their caregivers. Attentional focus switches to physical objects at about five months. Between the ages of six and nine months we find the beginning of gaze alternation between objects and adults, where this includes first bouts of gaze following, restricted by the visibility of the object to the infant. Pointing and more sophisticated forms of gaze- and point-following, coupled with the phenomenon of social referencing (in which infants appear to look to the adults to get emotional cues about how to react to new or unsettling stimuli) begin to take form between the ages of 10 and 12 months. Towards the end of this period children also start showing and giving objects to adults. First words emerge during the thirteenth month, on average, and from then on, until the end of the second year attentional behaviours become progressively sophisticated—for example we find progressively sensitive checks of where the adult is looking before, during and after pointing initiated by the infant, or showing of objects to adults, and the bouts of attending together to an object become longer and able to sustain the beginning of extended play with, and conversations about, the object(s) attended to.

On most accounts, joint attention in the sense we are interested in, begins to manifest itself at about twelve months, when we find the beginning of periods of sustained attending together to objects in the environment As I will understand the term ‘joint attention’, to say of an event that it is an event of two subjects (or more) jointly attending to the same object is to be committed, at least, to the truth of the following four claims about the event.

  1. There is an object that each subject is attending to, where this implies (a) a causal connection between the object and each subject, and (b) awareness of the object by each subject.
  2. There is a causal connection of some kind between the two subjects’ acts of attending to the object.
  3. The two subjects’ experiences exploit their understanding of the concept of attention.
  4. Each subject is aware, in some sense, of the object as an object that is present to both subjects. There is, in this respect, a ‘meeting of minds’ between both subjects, such that the fact that both are attending to the same object is open or mutually manifest.

Let us say that when these conditions are met we have in play a ‘joint attention triangle’. Very loosely, so-called ‘rich’ theories say that it is right to ascribe to 1-2 year olds participation in joint attentional triangles thus conceived. ‘Lean’ theories put pressure on condition c., and thereby on d. A radically lean claim would be that there is no understanding of any kind of the concept of attention at this age, and that, therefore between the ages of 1 and 2 conditions a. and b. only apply. A modestly rich account will say that there is some understanding of attention, but that it is not sufficient to generate mutuality, so conditions a-c only apply. (In the case of primate research, where all are agreed that mutuality is not achieved, all rich interpretations are modestly so: they say that some kind of understanding of attention is in play, whereas lean ones deny any).

There are, however, two critical ambiguities in this loose description of the rich/lean debate. Clarifying these will yield several additional stipulations about the way I will be using the notion of ‘joint attention’, and allow a more focused introduction of the kinds of questions I will be pursuing.

1.The term ‘rich’ is sometimes used to refer to what the causal coordinations of attention provide for between the ages of 1-2. A rich theory will say that these coordinations provide for mutual awareness; a lean theory will deny it. On another reading, a ‘rich’ theory is a theory that says sophisticated conceptual abilities are deployed by children engaged in causal coordinations of attention; a lean theory will deny it. The difference between these two readings come out when we reflect on the possibility of claiming that the causal coordinations of attention we find among 1-2 year olds are rich in the first sense, that is, yield mutual awareness, but that this requires ascribing to them only lean, unsophisticated conceptual abilities, which would be a lean claim in the second sense.

2. Condition c. refers to the mental understanding implicated in causal coordinations of attention among 1-2 year-olds. There are two distinct issues that fall under this heading. The first is the question of which particular mental concepts are implicated in joint attention conceived of as rich in the first sense. In particular this the question of what notion of attention we should be ascribing to children if we say that between the ages of 1-2 they participate in joint attention triangles. (This is what I earlier on called the ‘Concept Question’). The second is: what model of mental understanding should we use to explain the child’s use of these concepts. The central issue here is whether, and in what sense, the social interaction involved in joint attention suggests an alternative to the claim that children’s mental understanding is a matter of grasping a theory. (This is one aspect of what I earlier called the ‘Social Question’).

Now, as I will be using the notion of ‘rich’ throughout this chapter , unless I specify otherwise, it will refer to theories that hold that the kinds of causal coordinations of attention we find among 1-2 year olds provide for mutual awareness. In the first half of this chapter I will be concerned with building up to a progressively refined definition of such a theory. In the second half of the chapter we will be considering two current approaches to joint attention, both of which describe themselves as ‘rich’. As we will see, there is a question whether both can or would want to meet the requirements I will be using to define richness. At that point we will return to questions about the sophistication of abilities required of children participating in joint attention triangles. For the moment though, our concern will be to develop one substantive version of a rich theory in a way that will allow us to raise some of the fundemantal problems joint attention gives rise to.