The New Science of the Mind
Page 9
Clark and Chalmers are often interpreted as claiming that the sentence "The Museum of Modern Art is on 53rd Street," located in Otto's notebook, is identical with Otto's belief that the Museum of Modern Art is on 53rd Street. This interpretation is, presumably, too simplistic. More accurately, the idea is that when the sentence in the notebook is being deployed by Otto in the right sort of way, then, and only then, can it count as among Otto's beliefs. The guiding principle here is, as Clark and Chalmers put it:
If, as we confront some task, a part of the world functions as a process which, were it done in the head, we would have no hesitation in accepting as part of the cognitive process, then that part of the world is (so we claim) part of the cognitive process (1998, 8)
This claim, however, is unacceptably vague. If the "part of the world" in question is the sentence in Otto's notebook, then we must ask: how can the sentence function as a process? This apparently muddled talk seems to mask two possible interpretations:
1. The sentence in Otto's notebook, when appropriately deployed by Otto, and so situated in a context composed of the right sorts of surrounding psychological states and processes (Otto's perception of the sentence, his desire to see the exhibition, etc.) is one of Otto's beliefs. This interpretation identifies a token cognitive state-a belief-with an external structurea sentence.
Box 3.2
The Curious Case of Otto
Here is a classic-if it is correct to describe something barely a decade old as classic-thought experiment devised by Clark and Chalmers (1998). Otto is in the early stages of Alzheimer's disease. Otto has a friend, Inga, who is not similarly afflicted. Inga sees advertised in the newspaper an exhibition that she would like to see. The exhibition is taking place at the Museum of Modern Art. Inga thinks for a moment, recalls that MoMA is on 53rd Street, and sets off. Inga, it seems, has the belief that MoMA is on 53rd Street. And she had this belief even before she consulted her memory. Prior to the act of memory, of course, the belief was not occurrent. But most beliefs are like this: most of the beliefs we have we are not aware of having at any given moment. I believe that MoMA is on 53rd Street, but this is a belief I consciously entertain (usually) only when I am thinking about this thought experiment. Most of the time most of our beliefs exist in dispositional form: they exist in our tendencies or dispositions to do and say things in certain circumstances. Prior to her consciously accessing it, Inga's belief existed in dispositional form. And the basis of this disposition was that the belief was somewhere in Inga's memory, waiting to be accessed.
Otto, like many patients with Alzheimer's disease, relies on information contained in the environment to facilitate his day-to-day life. Otto carries a notebook around with him wherever he goes and, upon learning new information that he decides is sufficiently useful, he writes it down in his book. So whenever he needs previously acquired information, he looks it up. Otto's notebook, in effect, plays for him the sort of role usually played by biological (i.e., brain-based) memory. Otto reads about the exhibition at MoMA and, wanting to see it, looks in his notebook. This tells him that the museum is on 53rd Street, which he is where he duly goes.
According to Clark and Chalmers (C&C), "it seems reasonable to suppose that Otto believed the museum was on 53rd Street even before consulting the notebook" (1998, 12). This parallels the claim that Inga possessed the belief prior to her act of memory, for, according to C&C:
In relevant respects the cases are entirely analogous: the notebook plays for Otto the same role that memory plays for Inga. The information in the notebook functions just like the information constituting an ordinary non-occurrent belief; it just happens that this information lies beyond the skin. (1998, 12)
Inga sees the exhibition advertised in the newspaper, decides that she wants to see it, recalls that MoMA is on 53rd Street, and heads off in that direction. Similarly, Otto sees the exhibition advertised in the newspaper, decides he wants to see it, looks in his notebook and sees that MoMA is on 53rd Street, and heads off in that direction. The entry in the notebook that reads "The Museum of Modern Art is on 53rd Street" seems to function the same way in Otto's psychology as the corresponding belief functions in Inga's psychology. That is, the entry in Otto's notebook interacts with his desires (for example, to see the exhibition) in the same way that Inga's belief interacts with her desires. And it issues in the same actions: both Otto and Inga set off in the direction of 53rd Street. So, C&C ask: why shouldn't we regard the entry in Otto's notebook as one of his beliefs? They argue, notoriously, that there is no reason-no justifiable reason-to deny that the entry in Otto's notebook is one of his beliefs. Otto's beliefs exist not just in his head. They also exist in his notebook. But this is true not just of Otto: it is true of the rest of us as well.
2. The process of manipulating and/or exploiting the sentence is a properly cognitive part of an overall cognitive process. The overall process in question would be that of remembering or believing. The manipulation involved would be that of opening the book to the relevant page and orienting the page so that the sentence is open to detection by Otto. The manipulation, therefore, transforms the information contained in the sentence from the merely present to the available-available to Otto and/or to his subsequent processing operations. In virtue of playing this role, the manipulation of the sentence forms a properly cognitive part of the overall process of remembering or believing.12
The second interpretation claims that manipulation of an external structure is a properly cognitive part of a larger cognitive process. But it stops short of identifying the structure thus manipulated with a cognitive state. The version of the extended mind defended in this book is that implicated in the second interpretation. And there are very good reasons, I think, for rejecting the first interpretation. For those of you not antecedently convinced, you'll find the reasons in box 3.3.
My version of the extended mind-as a form of process rather than structure externalism-has little use for cognitive states. It is framed purely in terms of cognitive processes. This, I think, is the best way to understand the extended mind: as a process-oriented account. Therefore, on my version of the extended mind, the entry "The Museum of Modern Art is on 53rd Street" is not identical with any belief of Otto, no matter how he uses this entry, and no matter what the psychological milieu in which it is situated.
Box 3.3
How Not to Understand the Extended Mind
In "The Extended Mind," Clark and Chalmers (1998) are commonly regarded as defending this claim:
(0) The sentence "The museum of Modern Art is on 53rd Street" is, in appropriate circumstances, identical with one of Otto's beliefs.
Roughly, the circumstances are appropriate when the sentence is being deployed by Otto in an appropriate way (where such deployment requires Otto to also be engaged in suitable internal processing operations). I am not convinced that Clark and Chalmers are committed to (0), or even that they intend it, but I shall put aside this issue of interpretation. This interpretation of their view is widespread. For example, in what is sure to be an influential recent review of Clark's (2008) book Supersizing the Mind, Jerry Fodor (2009) writes (in connection with the original argument of C&C): "Since the content of Otto's notebook is derived (i.e., it's derived from Otto's thoughts and intentions with a 't'), the intensionality of its entries does not argue for its being part of Otto's mind." In other words, the entries in Otto's notebook are not beliefs because, unlike beliefs, their intentionality is merely derived. This intended criticism of Clark and Chalmers makes sense only if we assume that they are claiming that the entries in Otto's notebook are identical with (some of) his beliefs. Perhaps they do indeed intend this claim. I hope not; because the claim is untenable. This does not, however, entail rejection of the extended mind thesis; there is another, far more plausible, way of understanding this thesis.
The thesis of the extended mind is a thesis about mental tokens, not types. EMT is, among other things, a thesis about the location of mental items, and it is not clear
where, if anywhere, mental types are located. Therefore, (0) should be understood as asserting an identity between a sentence token-a physical inscription in Otto's notebook-and a token belief. The problem with this identification, I shall argue, is that the token sentence is the wrong sort of token to be identical with a token belief. (0) is, ultimately, incoherent.
In developing the problem, let us first recall two claims that are (partially) constitutive of the concept of a token. First, tokens are dated, unrepeatable entities. The eruption of Vesuvius in 79 AD is not token-identical with any subsequent eruption of that volcano. A token comes, goes, and never recurs. Second, tokens obey the transitivity of identity: if token x = token y, and token y = token z, then x = z. The problem with (0) is that it makes it impossible to satisfy both conditions.
Suppose, now, Oscar is deploying his notebook in a way deemed acceptable by C&C. Then, according to (0), the following claim is true:
(1) The sentence-token "The Museum of Modern Art is on 53rd Street" is identical with Otto's token belief that the Museum of Modern Art is on 53rd Street.
However, suppose that someone else is also deploying the notebook in a way deemed acceptable by C&C. For example, suppose Inga has also succumbed to the ravages of Alzheimer's and has decided to make use of Otto's book rather than going to the trouble of making one of her own. With suitable tinkering, I think we can make the example satisfy the sorts of conditions that C&C stipulate for Otto. Thus, we suppose that every time Otto made an entry in his book, Inga witnessed this and consciously endorsed the information Otto entered, and so on. We can suppose that the notebook is as easily and reliably accessible to Inga as it is to Otto, and so on. I won't push this point, however, for reasons to follow shortly: even if we deny that Inga can deploy the book in the right way, the same problems arises for Otto's repeated use of the book. In such a circumstance, it seems, (0) would commit us to the following:
(2) The sentence-token "The Museum of Modern Art is on 53rd Street" is identical with Inga's token belief that the Museum of Modern Art is on 53rd Street.
However, this means, by the transitivity of token-identity, that Otto's belieftoken is identical to Inga's belief-token. This is not, it must be emphasized, a case of Otto and Inga sharing two token beliefs of the same type. It is a case of two distinct people having the same token belief. And this is, of course, impossible if tokens are dated, unrepeatable entities.
Perhaps C&C might try to preclude this possibility by denying that anyone but Otto can deploy the book in the right way. I am not sure this strategy would work (see above), but I shall not press this point here. Even if this strategy were effective, it does not rule out the following problem. Suppose Otto, at a given time t, deploys his notebook in a way that, according to (0), qualifies the sentence-token as identical with a belief. Thus:
(3) At time t, the sentence-token "The Museum of Modern Art is on 53rd Street" is identical with Otto's token belief that the Museum of Modern Art is on 53rd Street.
However, if Otto can do this at time t, then there is nothing preventing him from deploying the notebook in the relevant way at a distinct time, t*. Thus:
(4) At time t*, the sentence-token "The Museum of Modern Art is on 53rd Street" is identical with Otto's token belief that the Museum of Modern Art is on 53rd Street.
It again follows from the transitivity of identity that Otto's token belief at t is identical with his token belief at t*. Once again, this is not a case of Otto instantiating two distinct belief-tokens of the same belief-type. Rather, it is a case of him possessing one and the same belief-token on two distinct occasions. And this is, again, impossible: tokens are dated, unrepeatable entities. In other words, (0) has the unfortunate consequence that one and the same belief-token can be possessed by the same subject at different times (and arguably by more than one subject at the same or different times). This is incoherent, and (0) must, therefore, be rejected.
What has gone wrong? The sentence-token is a physical inscription in Otto's notebook. As such, it can be deployed by Otto on more than one occasion. But the sentence-token is token-identical with itself on all of these occasions of use. Therefore, by the transitivity of identity, Otto's belief-tokens must be token-identical with themselves on all occasions in which they occur. And this is incoherent. What is needed is fairly clear. We need a way of individuating tokens such that when Otto deploys the sentence-token at time t it qualifies as a distinct token from the one deployed at distinct time t*. But what this means, in effect, is that we must reject the identification of the sentence- or inscription-token with a token belief possessed by Otto. That is, we must replace (0) with:
(0*) Otto's deployment, at time t, of the sentence in his notebook is (when it meets appropriate conditions) part of the process of his believing that the Museum of Modern Art is on 53rd Street.
The process in question is, of course, a process-token. Thus, when Otto deploys the sentence in his notebook in the same way at a distinct time t*, this automatically qualifies as (part of) a distinct process-token. The same would be true if Inga were to deploy the sentence in a way that met appropriate conditions (or ones deemed acceptable by C&C). On each occasion we have a distinct process-token, and the problem of incoherence is thereby avoided.
If the above arguments are correct, (0) should be rejected. However, the rejection of (0) is not equivalent to the rejection of the thesis of the extended mind. Although (0) provides a common interpretation of the claims of the extended mind, it is not, in fact, fruitful to understand this thesis as asserting the identity of mental state-tokens with external structure-tokens. A far more fruitful approach, one implicated in (0*) and untouched by the rejection of (0), is to understand the thesis of the extended mind as a claim about the partial composition of token cognitive processes: when Otto deploys his notebook in the right sorts of way, then this is part-a genuinely cognitive part-of the overall process of believing or remembering the location of the museum. We might call this version of the extended mind thesis "process externalism," and contrast it with the sort of "structure externalism" implicated in (0)-which identifies mental state-tokens with environmental structure-tokens.
The thesis of the extended mind should not be understood as claiming that cognitive states can be identical with environmental structures. Properly understood, the thesis makes no claim about cognitive states at all. It is a thesis that concerns cognitive processes and it claims some of these processes are, in part, composed of processes of manipulating, exploiting, or transforming environmental structures. It is the things we do with external structures-our manipulation, exploitation, and transformation of them-that constitute properly cognitive parts of overall processes of cognition. This is compatible, of course, with the idea that environmental structures can form parts of cognitive processes. They do so, but not because they are identical with cognitive states. It is the manipulation of environmental structures that forms part of the cognitive process; and so these structures form part of the process only insofar as they are part of the process of their being manipulated.
4 The Embedded Mind
The thesis of the extended mind is, first and foremost, a thesis of the composition or constitution of cognitive processes: some cognitive processes are partly composed of environmental processes. As such, the extended mind parallels the third interpretation of the thesis of the embodied mind, according to which some cognitive processes are partly composed of wider bodily structures and processes. In both cases, this claim of composition is an ontic claim that should be distinguished from epistemic claims about how we understand cognitive processes. However, as we saw in connection with the thesis of the embodied mind (henceforth simply "the embodied mind"), there is another ontic way of interpreting this thesis: as a claim of dependence rather than composition or constitution. Interpreted in this way, the embodied mind claims that (some) cognitive processes are dependent for their successful functioning on wider bodily structures and processes.
Therefore, we might expe
ct that there is also a corresponding way of interpreting the thesis of the extended mind. This expectation is, in effect, correct-but misleading. When interpreted in this way, what we end up with is not another version of the extended mind, but a quite distinct thesis: the thesis of the embedded mind (henceforth simply "the embedded mind"). Although often confused with the extended mind, the embedded mind is a very different thesis-far weaker and less interesting.
According to the embedded mind, cognitive processes are often (and on some versions essentially) embedded in the environment. The usual way of thinking about cognitive processes is in terms of what they do; in terms of their function. Take visual perception, for example. As we saw in the earlier discussion of Marr's theory of vision, a common way of understanding perception is as an input-output function from stimulus to visual representation. The stimulus is the retinal image-a pattern of light intensity distributed across the retina. The function of visual perception is to transform this into a visual representation of the world, a function that it completes in a series of stages. We should be careful to distinguish two things. On the one hand, there is the general idea that perception and other cognitive processes are best understood in terms of functions. This idea is known as functionalism-at least on one of the meanings of that widely used term. We shall look at functionalism in more detail in a later chapter. On the other hand, there is the identification of a particular role with which a cognitive process such as perception is to be defined. These claims are quite different. To say that perception is defined by its functional role-by what it does-is one thing. To say that perception is defined by this specific functional role is quite another. For example, you might accept that perception is defined by its functional role, but deny that the role is one of transforming retinal images into visual representations.