Ken Aizawa

This paper has a two-fold aim. First, it reinforces a version of the "syntactic argument" given in Aizawa (1994). This argument shows that connectionist networks do not provide a means of implementing representations without rules. Horgan and Tlenson have responded to the syntactic argument in their book and in another paper (Horgan & Tlenson, 1993), but their responses do not meet the challenge posed by my formulation of the syntactic argument. My second aim is to describe a kind of cognitive architecture. This architecture might be called a computational architecture, but it is not a rules and representations architecture nor the representations without rules architecture that Horgan and Tlenson wish to endorse.

Many years after the publication of “A Logical Calculus of the Ideas Immanent in Nervous Activity,” Warren McCulloch gave Walter Pitts credit for contributing his knowledge of modular mathematics to their joint project. In 1941 I presented my notions on the flow of information through ranks of neurons to Rashevsky’s seminar in the Committee on Mathematical Biology of the University of Chicago and met Walter Pitts, who then was about seventeen years old. He was working on a mathematical theory of learning and I was much impressed. He was interested in problems of circularity, how to handle regenerative nervous activity in closed loops....For two years Walter and I worked on these problems whose solution depended upon modular mathematics of which I knew nothing, but Walter did.. In this paper, we will fill in some of the details regarding Pitts’s interest in problems of circularity, regenerative activity in closed loops of neurons, and modular mathematics, and the way in which they relate to “A Logical Calculus.”

Abstract: There has recently been controversy over the existence of 'multiple realization' in addition to some confusion between different conceptions of its nature. To resolve these problems, we focus on concrete examples from the sciences to provide precise accounts of the scientific concepts of 'realization' and 'multiple realization' that have played key roles in recent debates in the philosophy of science and philosophy of psychology. We illustrate the advantages of our view over a prominent rival account ( Shapiro, 2000 and 2004 ) and use our work to rebut recent objections to the long-standing claim that psychological properties are multiply realized. For we use scientific evidence, in combination with our more precise theoretical framework, to show that we have strong reason to believe that psychological properties are indeed multiply realized both at the biochemical and neuronal levels.

  Terry Horgan and John Tienson have suggested that connectionism might provide a framework within which to articulate a theory of cognition according to which there are mental representations without rules (RWR) (Horgan and Tienson 1988, 1989, 1991, 1992). In essence, RWR states that cognition involves representations in a language of thought, but that these representations are not manipulated by the sort of rules that have traditionally been posited. In the development of RWR, Horgan and Tienson attempt to forestall a particular line of criticism, theSyntactic Argument, which would show RWR to be inconsistent with connectionism. In essence, the argument claims that the node-level rules of connectionist networks, along with the semantic interpretations assigned to patterns of activation, serve to determine a set of representation-level rules incompatible with the RWR conception of cognition. The present paper argues that the Syntactic Argument can be made to show that RWR is inconsistent with connectionism

This is a plausible reading of what Clark and Chalmers had in mind at the time, but it is not the radical claim at stake in the extended cognition debate.[1] It is a familiar functionalist view of cognition and the mind that it can be realized in a wide range of distinct material bases. Thus, for many species of functionalism about cognition and the mind, it follows that they can be realized in extracranial substrates.[2] And, in truth, even some non-functionalist views of cognition apparently allow for the possibility that cognition extends into the external world.[3] So, the (logical, conceptual, or nomological) possibility of extended cognition seems to us not the kind of radical view the advocates of this view have often implied. This is not, of course, to assess or pass judgment on the truth of these possibilities; it is only to note that they are not what most agitates people about the hypothesis of extended cognition. Framing the radical extended cognition hypothesis is a more delicate matter than framing the modal extended cognition hypothesis, but something like the following is in the ballpark. The radical extended cognition hypothesis maintains that, in many mundane cases of tool use, human cognitive processes extend into the tools. The principal reason this hypothesis is so delicate is that there remains much room for dispute about what constitutes a.

This chapter discusses the flaws of Clark’s extended mind hypothesis. Clark’s hypothesis assumes that the nature of the processes internal to an object has nothing to do with whether that object carries out cognitive processing. The only condition required is that the object is coupled with a cognitive agent and interacts with it in a certain way. In making this tenuous connection, Clark commits the most common mistake extended mind theorists make; alleging that an object becomes cognitive once it is connected to a cognitive agent is a “coupling-constitution fallacy.” From this fallacy, many hastily proceed to the conclusion that the object or process constitutes part of the agent’s cognitive apparatus or cognitive processing.

An important question in the debate over embodied, enactive, and extended cognition has been what has been meant by “cognition”. What is this cognition that is supposed to be embodied, enactive, or extended? Rather than undertake a frontal assault on this question, however, this paper will take a different approach. In particular, we may ask how cognition is supposed to be related to behavior. First, we could ask whether cognition is supposed to be behavior. Second, we could ask whether we should attempt to understand cognitive processes in terms of antecedently understood cognitive behaviors. This paper will survey some of the answers that have been given in the embodied, enactive, and extended cognition literature, then suggest reasons to believe that we should answer both questions in the negative

Philosophical interest in situated cognition has been focused most intensely on the claim that human cognitive processes extend from the brain into the tools humans use. As we see it, this radical hypothesis is sustained by two kinds of mistakes, the confusion of coupling relations with constitutive relations and an inattention to the mark of the cognitive. Here we wish to draw attention to these mistakes and show just how pervasive they are. That is, for all that the radical philosophers have said, the mind is still in the head.

In his recent critical notice of The Bounds of Cognition in this journal, Justin Fisher advances a set of concerns that favor the hypothesis that, under certain circumstances, cognitive processes span the brain, body, and world. One is that it is too much to require that representations in cognitive process must have non-derived content. A second is that it is possible that extended objects bear non-derived content. A third is that extended cognition might advocate the extension of certain general categories of cognition. A fourth is that Bounds misapplies Andy Clark and David Chalmers’ so-called “parity principle.” The purpose of this rejoinder is to show how Fisher’s concerns can be, or have already been, addressed

One way that scientifically recognized properties are multiply realized is by “compensatory differences” among realizing properties. If a property G is jointly realized by two properties F1 and F2, then G can be multiply realized by having changes in the property F1 offset changes in the property F2. In some cases, there are scientific laws that articulate how distinct combinations of physical quantities can determine one and the same value of some other physical quantity. One moral to draw is that in such cases we have the multiple realization of a single determinate, “fine grained” property instance that is exactly similar to another instance. As simple as this moral is, it has ramifications for a number of recent discussions of multiple realization in science. Taken collectively, these ramifications indicate that multiple realization by compensatory adjustments merits greater attention in the philosophy of science literature than it has hitherto received

Despite the considerable attention that the systematicity argument has enjoyed, it is worthwhile examining the argument within the context of similar explanatory arguments from the history of science. This kind of analysis helps show that Connectionism, qua Connectionism, really does not have an explanation of systematicity. Second, and more surprisingly, one finds that the systematicity argument sets such a high explanatory standard that not even Classicism can explain the systematicity of thought.

Despite the prominence of the systematicity argument in the debate between Classicists and Connectionists, there is extremely widespread misunderstanding of the nature of the argument. For example, Matthews (1994), has argued that the systematicity argument is a kind of trick, where Niklasson and van Gelder (1994), have claimed that it is obscure. More surprisingly, once one examines the argument carefully, one finds that Fodor, Pylyshyn, and McLaughlin, themselves have not fully understood it. 1 In part as a result of this, many Connectionists who have tried to meet the challenge of explaining the systematicity of thought have been misled about what this challenge involves (e.g. Pollack 1990; Smolensky 1990; Niklasson and van Gelder 1994). I have five principal objectives in this paper. First, I want to respond to those who believe that the systematicity argument is mere obscurity by providing a clear presentation of it. Second, and at the same time, I want to respond to those who believe the argument is mere trickery by showing it to be an instance of a rather familiar form of legitimate scientific reasoning. Third, having seen what sort of reasoning is involved in the argument, I wish to indicate how certain attempts to meet the challenge of systematicity are inadequate. Fourth, having seen the explanatory standard set by the systematicity argument, I want to indicate how even Fodor, Pylyshyn, and McLaughlin, fail to see how Classicism fails to meet the standard. Fifth, and finally, I wish to indicate what consequences the foregoing considerations have for the prospects of Classicism and Connectionism.

1. Do models formulated in programming languages use explicit rules where connectionist models do not? 2. Are rules as found in programming languages hard, precise, and exceptionless, where connectionist rules are not? 3. Do connectionist models use rules operating on distributed representations where models formulated in programming languages do not? 4. Do connectionist models fail to use structure sensitive rules of the sort found in "classical" computer architectures? In this chapter we argue that the answer to each of these questions is negative.

Obviously perception is embodied. After all, if creatures were entirely disembodied, how could physical processes in the environment, such as the propagation of light or sound, be transduced into a neurobiological currency capable of generating experience? Is there, however, any deeper, more subtle sense in which perception is embodied? Perhaps. Alva Noë’s theory of en- active perception provides one proposal. Noë suggests a radical constitutive hypothesis according to which (COH) Perceptual experiences are constituted, in part, by the exercise of sensorimotor skills. On Noé's view, bodily processes form a constitutive element in per- ceptual experiences. By contrast, it is more commonly supposed that bodily processes have at most a causal role to play in the genesis of perceptual experiences. Roughly stated, (CAI-l} Perceptual experiences are caused, in part, by the exercise of sensorimotor skills. Clearly these two hypotheses offer distinct conceptions of the embodiment of perception. One view maintains that there is a con- stitutive dependency between bodily processes and perceptual expe- riences, where the other maintains that there is a causal dependency. This paper will make the case that Noe fails to offer any evidence that. supports (COH) over (CAH) and that there is experimental evidence that favours (CAH) over (COH).E

InPsychosemantics Jerry Fodor offered a list of sufficient conditions for a symbol “X” to mean something X. The conditions are designed to reduce meaning to purely non-intentional natural relations. They are also designed to solve what Fodor has dubbed the “disjunction problem”. More recently, inA Theory of Content and Other Essays, Fodor has modified his list of sufficient conditions for naturalized meaning in light of objections to his earlier list. We look at his new set of conditions and give his motivation for them-tracing them to problems in the literature. Then we argue that Fodor's conditions still do not work. They are open to objections of two different varieties: they are too strong and too weak. We develop these objections and indicate why Fodor's new, improved list of conditions still do not work to naturalize meaning.

Among the many ideas that go by the name of “enactivism” there is the idea that by “cognition” we should understand what is more commonly taken to be behavior. For clarity, label such forms of enactivism “enactivismb.” This terminology requires some care in evaluating enactivistb claims. There is a genuine risk of enactivist and non-enactivist cognitive scientists talking past one another. So, for example, when enactivistsb write that “cognition does not require representations” they are not necessarily denying what cognitivists claim when they write that “cognition requires representations.” This paper will draw attention to instances of some of these unnecessary confusions.