The New Science of the Mind Read online

  The second imperative facing the amalgamated mind is the provision of an adequate and suitably motivated mark or criterion of the cognitive. This is required for two reasons. First, it is needed to defend the thesis of the extended mind against the charge that the arguments used to support it in fact only support the thesis of the embedded mind (as well as subsidiary concerns pertaining to cognitive bloat, causal-constitutive confusions, etc.). Second, it is needed to defend the thesis of the embodied mind from the (corresponding) charge that the arguments used to support it only support the claim that cognitive processes depend on wider bodily structures and processes rather than being constituted by, or composed of, such structures and processes. That is, it is needed to show that the arguments for the compositional interpretations of the theses of the extended and embodied mind establish no more than the dependence interpretations.

  These two requirements are connected in a way that is not obvious but is deeply significant. In the next chapter, I shall provide a mark or criterion of the cognitive of the required sort. This criterion is made up of four conditions, the satisfaction of which, I shall argue, are sufficient for a process to count as cognitive. Three of the conditions are relatively unproblematic. The fourth, however, is distinctly tricky. The fourth condition is an ownership condition on cognitive processes: any process that is to qualify as cognitive must be owned by some cognitive subject or organism. There are no cognitive processes without a subject. Developing an adequate account of the conditions under which a subject can own its cognitive processes is a deceptively difficult task, one to which much of the second half of this book will be devoted. The project, here, will to be to develop an account of in what the ownership of a cognitive process consists. In answering this question, however, I shall, in effect, develop a novel argument for the amalgamated mind. This is an argument that (a) does not depend on functionalism, and (b) unifies the embodied mind and the extended mind under a single guiding principle. If successful, it provides the strongest imaginable support for the amalgamated mind.

  Thus, the criterion of the cognitive I am going to develop in the next chapter provides the link between both problems facing the amalgamated mind. The criterion allows us to reject the arguments directed at the thesis of the embodied mind and the thesis of the extended mind taken individually. And satisfying the fourth condition of the criterion-the ownership condition-requires the development of an argument for the amalgamated mind that subsumes both embodied and extended arms of this view-subsumes them in equal measure and for the same reasons. This argument will make it clear why both the embodied mind and the extended mind are essentially versions of the same idea. Thus, if all goes well, the criterion of the cognitive will kill two birds with one dialectical stone.

  1 Criteria: What Are They Good For?

  In what sense, and to what extent, does a science need to understand its subject matter? The answer to this question may seem obvious. Of course a science needs to understand what it is supposed to be investigating. How else could it investigate it? However, matters are not as obvious as they seem. Does physics, for example, require a precise-or even adequateunderstanding of what makes something physical in order to proceed? Does biology require a precise definition of what makes something biological? The answer to both seems to be "no." Physics has survived various significant shifts in our understanding of what it is to be physical. And the general idea that biology is the science of living things has learned to deal with the fact that there is no consensus on what makes something living. At most, we might think that an understanding of what it is to be physical or what it is to be biological is something that slowly emerges from the progress of the relevant sciences. If so, it is not something that is required in order for each science to conduct its business.

  Among cognitive science practitioners one can, I think, discern two types of attitude. On the one hand, there are pragmatists who think that cognitive science is simply in the business of modeling cognitive processes, at varying levels of abstraction. Deeper questions of selfinterpretation-what exactly is it that I am doing?-are left to philosophers, and are strictly tangential to the practice of cognitive science itself. The other attitude, more plausibly I think, regards a deepening understanding of the concept of cognition as something built into the cognitive scientific enterprise itself-as something that slowly emerges from the enterprise rather than guides it from the outset. In this, they echo a tradition that sees science not just as an activity concerned with interpreting the world, but also as one that has the possibility of self-interpretation built into it as one of its basic ingredients (Heidegger 1927/1962). But the idea that cognitive science requires a proper understanding of what cognition is before it can set about its legitimate business is one that finds few, if any, defenders.

  This chapter will be concerned to develop a criterion of cognition. Very roughly, this is a criterion that tells us when a process counts as a cognitive one. However, this is-emphatically-not because I think cognitive science in general needs one in order to be able to conduct its business. Rather, although cognitive science in general does not need a criterion of cognition, the amalgamated mind most certainly does. As we saw in the previous chapter, if the amalgamated mind is to have any genuine substance, then it cannot be allowed to collapse into claims about environmental and bodily embedding or dependence; for these are claims that can be easily assimilated into the old science. But if it is to avoid collapsing into claims of environmental or bodily dependence, the amalgamated mind must be able to make the case that the bodily and environmental exploitation and manipulation that it invokes are genuinely cognitive parts of overall cognitive processes that contain an irreducible neural element. The goal, therefore, is to show that manipulation and exploitation of bodily and environmental structures can satisfy a plausible criterion of the cognitive. If they can do this, then processes of manipulating or exploiting bodily and environmental structures will have every right to be regarded as genuinely cognitive components of an overall cognitive process, rather than merely causal contributors to an overall cognitive process.

  2 Criteria of the Criterion

  A mark of the cognitive is a criterion that specifies when a process qualifies as cognitive. A criterion that can do this provides what is known as a sufficient condition for a process to count as cognitive. One might also want the criterion to specify when a process does not count as cognitive. If it can do this, then it provides what is known as a necessary condition for a process to count a cognitive. I am not going to pursue a criterion of this latter sort. The mark of the cognitive I am going to develop and defend in this chapter provides only a sufficient condition for a process to count as cognitive-at least I shall claim no more than this. For reasons that will become clear, this is all the amalgamated mind needs in order to respond to the objections we have identified in the previous chapter. As we shall also see shortly, there is another reason, stemming from the way in which a mark of the cognitive is best motivated and defended.

  Given that the mark of the cognitive will provide a sufficient condition for a process to qualify as cognitive, the next question is: how does it do this? It does so by listing a set of conditions such that, if a process were to satisfy them, that would be enough for the process to count as cognitive. The conditions collectively make up a criterion of cognition. If the arguments of the previous chapter are correct, it is clear why the thesis of the extended mind requires a criterion of the cognitive: all the recognized objections to the thesis seem to depend, directly or indirectly, on its perceived failure to provide such a criterion. It is also clear why the thesis of the embodied mind requires such a criterion: without it, the thesis is vulnerable to the charge that the arguments for it establish only the claim that cognitive processes are dependent on wider bodily structures and processes and not the claim that they are, in part, constituted by such structures and processes. What is not yet clear, however, and even ignoring for the moment questions of its specific content, is precisely wh
at sort of thing the criterion should be. In particular, questions can be raised about both the scope and character of the desired criterion.

  Consider, first, the issue of scope. The term "cognition" can, in effect, be spelled with a big or small "c." With a big "C," the sense of the term is fairly narrow. In this sense, cognition is routinely opposed to perception (and, of course, sensation). That is, in this sense, cognition is restricted to postperceptual processing. However, there is a broader sense of the term, "cognition" with a little "c," where it is understood to include perceptual processing. Reference to cognition in this chapter, and indeed the rest of the book, should be understood to be reference to cognition in the more general sense: "cognition" with a little "c." And the proposed criterion of "the cognitive" is one that attempts to demarcate items that are cognitive in this broader sense from items that are not.

  Consider, now, the issue of the character of the criterion. What sort of thing would the mark of the cognitive have to be? Here, two quite distinct projects need to be distinguished. On the one hand, there is the philosophical project of naturalizing the mind. If we were engaged in this project, we might take ourselves-intelligibly but, I think, mistakenly-to require a reductive definition of the expression "the cognitive" in terms that are entirely noncognitive (or, initially at least, less than fully cognitive). On the other hand, there is the cognitive-scientific project of understanding how cognitive processes work. This project consists in functional decomposition of these processes into progressively simpler constituents. To engage in this project, we do not require a naturalistic reduction of cognition-in the form of a definition of "cognition" in noncognitive terms. Rather, we simply require a means of demarcating, with a reasonable but not necessarily indefeasible level of precision, those things in which we are interested from those things in which we are not.

  I shall propose and defend a criterion of this latter sort. This sort of criterion, I shall argue, is sufficient to defuse the standard objections to the theses of the extended mind and the embodied mind-the theses that together make up the amalgamated mind. The idea underpinning the criterion is that if we want to understand what cognitive processes are, we had better pay close attention to the sorts of things cognitive scientists say are cognitive and then try to identify the general principles-the general conception of cognition-that explains why they say what they say. This is not to say that we must restrict ourselves to the pronouncements of cognitive scientists. Cognitive scientists, like the rest of us, can be mistaken, and they can be confused. It is possible that in certain cases, the general conception of cognition we find in cognitive-scientific practice may be incompatible with specific determinations of cognitive status or lack thereof. In part, the purpose of a criterion of the cognitive is to allow us to render consistent the general conception of cognition that is embodied in cognitive-scientific practice with specific judgments about which processes are cognitive. That is, the goal of this process is to render consistent general principles and specific judgments insofar as this is possible. In doing so, as John Rawls puts it in a different but structurally similar context, we work from both ends with the goal of achieving reflective equilibrium between our specific judgments of cognitive status and the general conception of cognition that underlies those judgments.

  Our starting point, however, must be cognitive-scientific practice itself, and the specific judgments, made by cognitive scientists, concerning what processes count as cognitive. Thus, the criterion of the cognitive I propose here will be motivated and defended, in the first instance, by way of a careful examination of cognitive-scientific practice.

  3 The Criterion

  Before we get to the motivation and defense of the criterion, however, let us first look at what the criterion is. I shall argue that when we examine cognitive-scientific practice, what we find is an implicit mark or criterion of the cognitive that looks like this:

  A process P is a cognitive process if:

  1. P involves information processing-the manipulation and transformation of information-bearing structures.

  2. This information processing has the proper function of making available either to the subject or to subsequent processing operations information that was, prior to this processing, unavailable.

  3. This information is made available by way of the production, in the subject of P, of a representational state.

  4. P is a process that belongs to the subject of that representational state.'

  Before motivating and defending this criterion, it is necessary to make a few introductory remarks, both in general and about each of its conditions.

  General Remarks

  The criterion is, as I emphasized earlier, presented only as providing a sufficient condition for a process to count as cognitive. If a process satisfies these four conditions, it is a cognitive process. It does not follow that if a process does not satisfy these conditions, it is not a cognitive process. There may be other ways in which a process can count as cognitive. However, these potential other ways are not embodied in or recognized by current cognitive-scientific practice, and this, as I have also stressed, is to be the starting point for motivation and defense of the criterion.

  Condition (1)

  The idea that cognition involves information processing is now commonplace. Indeed, it is a central plank of the conception of cognition embodied in the cognitive science tradition. In its classical form, cognitive science was understood to involve the postulation of internal configurations of an organism or system: configurations that carry information about extrinsic states of affairs. The concept of information employed is, in essence, that elaborated by Claude Shannon (1948), or a close variant thereof. According to Shannon, information consists in relations of conditional probability. On the version championed by Shannon, a receptor r carries information about a source s only if the probability of s given r is 1 (cf. Dretske 1981). Other, less sanguine versions associate the carrying of information with an increase in conditional probability, although not necessarily to the value of 1. That is, r carries information about s only if the probability of s given r is greater than the probability of s given not-r (Lloyd 1989).

  Whichever explication of the concept of information is employed, the underlying vision of cognition is unaffected. Cognitive processes are understood as a series of transformations performed on information-bearing structures. These transformations will be effected according to certain rules or principles-principles that effectively define the character of the type of cognitive process in question. Shortly, we shall examine one influential example of this general vision.

  Condition (2)

  The second condition relies heavily on the concept of proper function. I shall understand this in what is known as an etiological sense championed by Millikan, among others (Millikan 1984, 1993). The proper function of something is what it is supposed to do, or what it has been designed to do. The proper function of the heart is to pump blood; the proper function of the kidneys is to process waste, and so on. Invoking the idea of proper function in the criterion of cognition is to remind us that the concept of cognition is, in part, a normative one. Cognitive processes can function well or badly, properly or improperly; they are defined in terms of what they are supposed to do, not what they actually do. This distinction between what a cognitive process does and what it is supposed to do can manifest itself in at least four ways:

  i. Although the proper function of a cognitive process, P, is to make available information that was previously unavailable, it might not fulfill this function in particular cases. Indeed, it might never fulfill this function, owing either to faults in the mechanisms that realize it, or to faults in the mechanisms that are supposed to receive the information it supplies. Similarly, a defective heart that is never able to pump blood is still a heart with the proper function of doing so. And if a heart is connected to arteries that are blocked, then it will not fulfill its proper function.

  ii. In addition to performing its proper functi
on, P might also do all sorts of things that are not its proper function-because these things do not explain why the mechanism has proliferated and is extant today. Similarly, the heart, in addition to pumping blood, also makes noise and produces wiggly lines on an electrocardiogram. These other causal effects of the heart, however, are not part of its proper function. The reason, very roughly and there is devil in the details, is that hearts that made noise but did not pump blood would not have been selected for; but, conversely, hearts that pumped blood without making noise would have been selected for.

  iii. P might have the proper function of making information available to a subject or subsequent operations only in the presence of environmental contingency C, and C may sometimes fail to obtain; in principle, it may never obtain. Similarly, the function of the sperm cell is to fertilize the female egg, but the vast majority of sperm cells fail to fulfill this function, largely owing to inopportune circumstances.

  iv. P may have the proper function of making information available only in combination with a number-perhaps a vast number-of other processes. In such circumstances, P's fulfilling its proper function is dependent on these other processes fulfilling their proper functions. Similarly, kidneys cannot process waste without being connected up, in the right way, to a circulatory system that produces this waste.

  Whether something qualifies as a cognitive process is a matter of what it is supposed to do, and not simply a matter of what it does. Defective cognitive processes are still cognitive processes. This is why the second condition of our criterion invokes the idea of function. However, when the idea of function is applied to cognitive processes, this involves a further distinction that has no real echo in the case of hearts, kidneys, and sperm cells. There are, broadly speaking, two sorts of general function that a cognitive process might have. Both functions involve the making available of information. What distinguishes them is to what they make this information available. Some cognitive processes make information available to the cognitive organisms or cognizing subject. Remembering is a cognitive process that, if successful, makes information available to me. So too do perception, reasoning, and so on. However, some cognitive process make information available not to a cognitive organism such as myself, but only to further nonconscious processing operations. For example, on Marr's (1982) theory of vision (discussed earlier, and soon to be discussed again), the transformation of the retinal image into the raw primal sketch makes information available not to the subject of vision (for example, you or me) but only to the subsequent nonconscious processes whose function is to transform the raw primal sketch into the full primal sketch.