2. CAPITULO II MARCO TEÓRICO
2.1. ANTECEDENTES INVESTIGATIVOS
what they (or the processes that generate them) have been selected for; rather, it is because of the content that they have that they (or the processes that generate them) are selected for.
mental representations. Even if she set herself the task of uncovering the syntactic workings of, say, the visual module (rather than that of explaining an intentionally-characterised cognitive capacity) she could not afford to ignore the semantic properties of the symbols manipulated by that module. This is because one needs to have a pretty good idea of what information the visual module is processing before one has a realistic chance of working out any of the syntactic details. In general, in order to uncover the syntactic workings of a cognitive module psychologists proceed by first addressing such questions as: what information would it be useful for the module to generate for it to support the cognitive capacity that it underlies? What information could it possibly generate given the nature of the external environment? And so on. It is only when such questions have been answered (when some light has been shed on the semantic activity of the module in question) that the syntactic details can be uncovered. In the absence of a semantic story, the psychologist will not know what she is looking for and will have no clues as to how to formulate and test plausible hypotheses. Of course, it doesn't follow from this that semantic properties have a role to play in the description of the syntactic workings of a cognitive module or in an explanation of such a m odule's syntactic capacities. For such purposes the semantic properties hitherto identified should be, as it were, rubbed out. But they cannot be rubbed out from an explanation of an intentionally-characterised cognitive capacity for the reasons that I have outlined above.
In short, then, the semantic properties of the symbols manipulated by subpersonal systems cannot be ignored by the psychologist given her explanatory ambitions. But why must any appeal to syntax and syntactic operations be made? What would be wrong with an account w hich merely analyses a subpersonal system 's inform ation- generating activity into a series of information gathering substeps along these lines: the system starts off with such and such information from which it extracts some other information, from which it extracts some more information, and so on? The answer is that without the syntactic details it is a mystery as to how the system works out the information that it works out. As it generates information by means of computation, to give a complete account of how a system generates the information that it generates, it is
necessary to tell the syntactic story. In the case of the subpersonal systems that support and underlie cognition there is always the worry that if the syntactic story is missing one will have attributed to the brain miraculous powers, that is, powers to generate information from information that just couldn't be had by an unintelligent, mechanical system.
It is worthwhile emphasising the beauty of computation from the perspective of the scientific psychologist. In connection with the mind, what is most exciting about computers is not the possibility of building intelligent computers, but rather the actuality that computers are completely stupid. Computers generate meaningful symbols from meaningful symbols and in so doing work things out, solve problems, discover facts, and the like, by entirely mechanical means that involve no exercise of intelligence. This fact about computers raises the possibility that subsystems of the brain engage in computational activity and thus generate information from information by entirely mechanical and unintelligent means. To attribute to such subsystems powers to generate information from information by means of computation in the course of accounting for cognitive capacities is thus not to posit intelligent agents in the brain in any way that threatens circularity or infinite regress.^^ It is this fact about computers and computation that leads scientific psychologists to think of the subpersonal systems that underlie and support cognition as engaging in computational activity. And once they see subpersonal systems as engaging in computational activity, in perform ing their information-generating tasks, psychologists commit themselves to the task of telling the syntactic story in the course of accounting for cognitive capacities.
In the course of accounting for my capacity to work out my bank balance we saw that it was important to appeal to facts about the world. These facts are such that if they didn't hold certain symbol- m anipulating moves would be incapable of generating relevant information. The same point holds of the psychological case. For example, the visual module would not be able to detect such features of surfaces as changes in texture and colour were it not for the fact
that there is a systematic relationship between the texture and
colour of a surface and the intensity and wavelength of the light that
it reflects. Hence, the psychologist must specify these facts, for a failure to do so would leave it a mystery as to the relevance and success of the computational operations described by her explanation. Indeed, such facts about the world will, in a certain sense, lead the way in the construction of psychological explanations. For facts about the world will provide the psychologist with clues as to how subpersonal systems generate relevant information.
What we have arrived at now is a more or less complete account of the nature of scientific psychological explanation. In summary, here is how that account goes. Scientific psychology aims to account for intentionally-characterised cognitive capacities; in other words to describe how we exercise such capacities. A fundamental assumption is that underlying and supporting such capacities are neurophysiologically-realised subpersonal systems that are capable of generating symbols from symbols by means of computation. These subpersonal systems facilitate cognition because the symbols they m anipulate have semantic properties, so that in exercising their symbol-manipulating capacities they generate information from information. In other words, given information as input they generate further information as output. The information generated in this way is precisely the kind that enables the cognitive system as a whole to form the personal-level intentional states that manifest the cognitive capacity in question. Consequently, central to explaining a cognitive capacity is the construction of an account of how the subpersonal system underlying the cognitive capacity in question performs its information-generating task; that is, how it generates the information that it produces as output from the information that it takes as input. As it performs this task by generating information- bearing symbols from information-bearing symbols by means of computation, such an account m ust include reference to both semantic and syntactic properties. It must describe the formal language employed, the syntactic rules that are applied to the symbols of that language, and the semantic properties of these symbols. A failure to tell the syntactic story will constitute a failure to specify fully how the subpersonal system generates the information that it generates. And a failure to tell the semantic story will leave it a mystery as to the point and relevance of the system's symbol- manipulating activity, for that activity only has a point and relevance
if the symbols manipulated have appropriate semantic properties. Furthermore, those facts about the external world that enable the system to work out the information that it needs from information that it already has by means of the computational operations that it performs must also be specified. Thus the semantic and the syntactic are, as it were, intertwined in scientific psychological explanations. That they are so intertw ined is a product of the explanatory ambitions of scientific psychology and the fundamental assumptions of that discipline.
This account of the nature of scientific psychology would appear to imply that scientific psychological explanations are what Cummins calls "property theories" (see fn. 16). In explaining cognitive capacities by appeal to simpler underlying capacities, psychological explanations are not causal explanations of particular events that fit the deductive- nomological model of explanation. However, they can be seen as being causal explanations that effectively appeal to causal laws (or at least to counterfactual-supporting causal generalisations) and for this reason I am hesitant to call them property theories. As we have seen, cognitive capacities (along with the information processing and computational capacities that underlie them) correspond to causal generalisations. Consequently, a system's having such a capacity is just a matter of a certain causal generalisations being true of its input- output behaviour. Moreover, whenever an individual exercises a cognitive capacity, a causal process takes place that is subsumed by the causal generalisation that corresponds to that capacity. And that causal process will be made up of constituent causal processes each of which are exercises of underlying capacities and are thus subsumed by the causal generalisations corresponding to those capacities. (That is why references to underlying information processing and computational capacities can be freely interchanged with references to what goes on causally whenever such capacities are exercised). Effectively then, psychological explanations explain causal generalisations by appealing to the underlying causal generalisations in virtue of which they hold. And they can also be seen as describing what goes on causally whenever the cognitive capacity in question is exercised. For these reasons, scientific psychological explanation is a form of causal explanation.
2.7 Fodor's account of psychological explanation
How does Fodor's understanding of the nature of scientific psychology compare with my account? Does my account place any pressure on any of Fodor's views? One difference that we have seen concerns the nature of the RTM that scientific psychology is commited to. According to Fodor, scientific psychology is committed to RTM as a theory of such personal level PAs as beliefs, an assertion that I have rejected. In actual fact, the RTM that underlies scientific psychological theorising is a theory about subpersonal representational states, and not beliefs, desires, and the like.
A consideration of some of Fodor's most explicit pronouncements on the explanatory ambitions of scientific psychology and the nature of the explanations that it endeavours to produce would appear to generate a second point of conflict. Fodor has a tendency to represent scientific psychology as being a sharpened up version of folk psychology which, like the latter, seeks to produce singular causal explanations of PA tokenings and behavioural events that appeal to PAs, and which fit the deductive-nomological model of explanation. The primary difference between the two psychologies is that the scientific version is more thorough, rigorous and research oriented, and so constantly seeks to add to its stock of causal generalisations so
as to expand its explanatory p o w e r s . 3 7 Such a view of the nature of
scientific psychology and the explanations that it produces would appear to conflict with the account that I have developed in this chapter.
However, the above points notwithstanding, I am hesitant to accuse Fodor of holding a mistaken account of the nature of scientific psychology for there is much that he writes that sits happily with my account. First, he has represented scientific psychology as seeking to explain cognitive capacities by appealing to the simpler underlying capacities and operations of cognitive subsystems.38