Online research in philosophy

For reasons internal to the concepts of thought and causality, a series of representations must be semantics-driven if that series is to add up to a single, unified thought. Where semantics is not operative, there is at most a series of disjoint representations that add up to nothing true or false, and therefore do not constitute a thought at all. There is necessarily a gulf between simulating thought, on the one hand, and actually thinking, on the other. It doesn't matter how perfect the simulation is; nor does it matter how reliable the causal mechanism involved is. Where semantics is inert, there is no thought. In connection with this, this paper also argues that a popular doctrine - the so-called 'computational theory of mind' (CTM) - is based on a confusion. CTM is the view that thought-processes consist in 'computations', where a computation is defined as a 'form-driven' operation on symbols. The expression 'form-driven operation' is ambiguous, and may refer either to syntax-driven operations or to morphology-driven operations. Syntax-driven operations presuppose the existence of operations that are driven by semantic and extra-semantic knowledge. So CTM is false if the terms 'computation' and 'form-driven operation' are taken to refer to syntax-driven operations. So if CTM is to work, those expressions must be taken to refer to morphology-driven operations. But, as previously stated, an operation must be semantics-driven if it is to qualify as a thought. Thus CTM fails on every disambiguation of the expressions 'formal operation' and 'computation'.

This paper explores the difference between Connectionist proposals for cognitive a r c h i t e c t u r e a n d t h e s o r t s o f m o d e l s t hat have traditionally been assum e d i n c o g n i t i v e s c i e n c e . W e c l a i m t h a t t h e m a j o r d i s t i n c t i o n i s t h a t , w h i l e b o t h Connectionist and Classical architectures postulate representational mental states, the latter but not the former are committed to a symbol-level of representation, or to a ‘language of thought’: i.e., to representational states that have combinatorial syntactic and semantic structure. Several arguments for combinatorial structure in mental representations are then reviewed. These include arguments based on the ‘systematicity’ of mental representation: i.e., on the fact that cognitive capacities always exhibit certain symmetries, so that the ability to entertain a given thought implies the ability to entertain thoughts with semantically related contents. We claim that such arguments make a powerful case that mind/brain architecture is not Connectionist at the cognitive level. We then consider the possibility that Connectionism may provide an account of the neural (or ‘abstract neurological’) structures in which Classical cognitive architecture is implemented. We survey a n u m b e r o f t h e s t a n d a r d a r g u m e n t s t h a t h a v e b e e n o f f e r e d i n f a v o r o f Connectionism, and conclude that they are coherent only on this interpretation.

Roger Schank's research in AI takes seriously the ideas that understanding natural language involves mapping its expressions into an internal representation scheme and that these internal representations have a syntax appropriate for computational operations. It therefore falls within the computational approach to the study of mind. This paper discusses certain aspects of Schank's approach in order to assess its potential adequacy as a (partial) model of cognition. This version of the Language of Thought hypothesis encounters some of the same difficulties that arise for Fodor's account.

Classical computational modellers of mind urge that the mind is something like a von Neumann computer operating over a system of symbols constituting a language of thought. Such an architecture, they argue, presents us with the best explanation of the compositionality, systematicity and productivity of thought. The language of thought hypothesis is supported by additional independent arguments made popular by Jerry Fodor. Paul Smolensky has developed a connectionist architecture he claims adequately explains compositionality, systematicity and productivity without positing any language of thought, and without positing any operations over a set of symbols. This architecture encodes the information represented in linguistic trees without explicitly representing those trees or their constituents, and indeed without employing any representational vehicles with constituent structure. In a recent article, Fodor (1997; Connectionism and systematicity, Cognition , 62, 109-119) argues that Smolensky's proposal does not work. I defend Smolensky against Fodor's attack, and use this interchange as a vehicle for exploring and criticising the “Language of Thought” hypothesis more generally and the arguments Fodor adduces on its behalf.

The arguments that Fodor (1987: 150-52) gives in support of a Language of Thought are apparently straightforward. (1) Linguistic capacities are "systematic", in the sense that if one understands the words 'John loves Mary' one also understands the form of words 'Mary loves John'. In other words, sentences have a combinatorial semantics, because they have constituent structure. (2) If cognitive capacities are systematic in the same way, they must have constituent structure also. Thus there is a Language of Thought. The essential connection between language and thought that the argument requires is: Since the function of language is to express thought, to understand a sentence is to grasp the thought that its utterance standardly conveys. So from the systematicity of sentences it follows that anyone who can grasp the thought that John loves Mary can grasp the thought that Mary loves John. Thought must be as systematic as language, for the best empirical explanation of the psychological fact that one who grasps the thought that John loves Mary can grasp the thought that Mary loves John is that grasping a thought is standing in some thinking relation to a complex entity whose constituents are MARY, X LOVES Y, and JOHN and semantic relations among MARY, X LOVES Y, and JOHN. So sayeth Fodor (1987).

I critically explore various forms of the language of thought (LOT) hypothesis. Many considerations, including the complexity of representational content and the systematicity of language understanding, support the view that some, but not all, of our mental representations occur in a language. I examine several arguments concerning sententialism and the propositional attitudes, Fodor's arguments concerning infant and animal thought, and Fodor's argument for radical concept nativism and show that none of these considerations require us to postulate a LOT that is innate or otherwise distinct from spoken languages. Instead, I suggest that we maintain the more conservative hypothesis, supported by introspection, that some of our thoughts occur in the languages that we speak.

The Language of Thought Hypothesis is often taken to have the fatal flaw that it generates an explanatory regress. The language of thought is invoked to explain certain features of natural language (e.g., that it is learned, understood, and is meaningful), but, according to the regress argument, the language of thought itself has these same features and hence no explanatory progress has been made. We argue that such arguments rely on the tacit assumption that the entire motivation for the language of thought consists in explaining the explanandum that allegedly generates the regress. But this tacit assumption is simply false. The Language of Thought Hypothesis is a cogent view and one with considerable explanatory advantages.

Fodor and Pylyshyn's critique of connectionism has posed a challenge to connectionists: Adequately explain such nomological regularities as systematicity and productivity without postulating a "language of thought" (LOT). Some connectionists like Smolensky took the challenge very seriously, and attempted to meet it by developing models that were supposed to be non-classical. At the core of these attempts lies the claim that connectionist models can provide a representational system with a combinatorial syntax and processes sensitive to syntactic structure. They are not implementation models because, it is claimed, the way they obtain syntax and structure sensitivity is not "concatenative," hence "radically different" from the way classicists handle them. In this paper, I offer an analysis of what it is to physically satisfy/realize a formal system. In this context, I examine the minimal truth-conditions of LOT Hypothesis. From my analysis it will follow that concatenative realization of formal systems is irrelevant to LOTH since the very notion of LOT is indifferent to such an implementation level issue as concatenation. I will conclude that to the extent to which they can explain the law-like cognitive regularities, a certain class of connectionist models proposed as radical alternatives to the classical LOT paradigm will in fact turn out to be LOT models, even though new and potentially very exciting ones.

1 *Common Sense Conception of Beliefs and Other Propositional Attitudes 2 What is the Language of Thought Hypothesis? 3 Status of LOTH 4 Scope of LOTH 5 *Natural Language as Mentalese? 6 *Nativism and LOTH 7 Naturalism and LOTH.

According to the Computational/Representational Theory of Thought (CRTT ? Language of Thought Hypothesis, or LOTH), propositional attitudes, such as belief, desire, and the like, are triadic relations among subjects, propositions, and internal mental representations. These representations form a representational _system_ physically realized in the brain of sufficiently sophisticated cognitive organisms. Further, this system of representations has a combinatorial syntax and semantics, but the processes that operate on the representations are causally sensitive only to their syntax, not to their semantics. On this approach, a first pass account of propositional attitudes is the following (cf. Field 1978: 37 and Fodor 1987: 17).