In lieu of an abstract, here is a brief excerpt of the content:

1 Introduction The systematicity debate initially turned on the issue of the best explanation for the systematicity of cognition—then a property taken for granted, so that it barely required anything more than cursory exemplification. Connectionists challenged the idea that a “language of thought” of atomic constituents, plus formal combinatorial rules, was the only (best) approach to account for that claimed property of cognition. In these post-cognitivist times, we rather think that the proper reaction to the Fodor and Pylyshyn’s (1988) challenge is to deny that cognition is systematic in general. Systematicity rather seems a property intrinsically dependent upon language rather than cognition in general, since the typical examples of systematicity are in fact syntax-bound; in addition, when we examine nonverbal cognition, we don’t find the kind of systematicity required by the argument . Current post-cognitivist approaches to cognition, which emphasize embodiment and dynamic interaction, in its turn, also challenge the cognitivist assumption that the explanandum that a theory of cognition has to account for includes systematicity as a basic property of cognitive processes. While the general strategy to claim that cognition is systematic was based on structural parallels between language and thinking, inferring for thinking what was supposed to be obvious for language, it has also been proposed that the systematicity of cognition can be found in perception— on the understanding that perception is a cognitive process, and one that is clearly independent of language. In The Language of Thought, Fodor (1975) exemplified his general approach to cognition by appealing to three areas of research: language processing, decision making, and perception. However, at that time the systematicity as explanandum and the combinatorial structure of representational elements as explanans were not 15 From Systematicity to Interactive Regularities: Grounding Cognition at the Sensorimotor Level David Travieso, Antoni Gomila, and Lorena Lobo 372 David Travieso, Antoni Gomila, and Lorena Lobo clearly distinguished, and in fact, there was no explicit discussion of whether perception is systematic on the same model—the linguistic one— that was used to claim that cognition is systematic. But the basic idea is reasonably clear: perception is to count as systematic if it can be shown that perceptual capacities come in groups, if having one perceptual capacity involves having some others—just as understanding one sentence is connected to understanding many others. Marr’s (1982) theory, which tried to account for perception in terms of basic representations and formal rules to combine and transform them, displayed the same kind of explanatory strategy as that of a combinatorial syntax of representational units, and was taken by Fodor as support for his language of thought (LOT) approach. So, it could be said that one way to support the systematicity of perception is to show that it can be better explained by a version of this classical cognitive architecture. In this chapter, after briefly reviewing our previous arguments against the systematicity of cognition in general, and for a change in the order of dependence between thought and language—so that the systematicity of thought, when it is found, is parasitic on the systematicity of language—we will discuss the attempt to argue for the systematicity of perception and the contention that it is better explained by a combinatorial syntax of primitive representational units. We will first discuss the example offered to claim that perception is systematic, quite apart from a commitment to Marr’s approach to perception: the phenomenon of amodal completion, as presented by Aizawa (this vol.). We will argue that it falls short of proving that perception is systematic, because what is claimed to be a group of interconnected perceptual abilities is better viewed as just one; furthermore, the best explanation of amodal completion is not in terms of a compositional structure plus inferential processes, because it is a global, emergent pattern that arises from context-dependent interactions. The way it is discussed by Aizawa, moreover, can’t sustain the claim that perception is systematic in the required sense, because it is just a version of the standard illustration of cognitive systematicity by propositional interdependencies—those that we claim are language dependent. Next, we will discuss whether there is indeed a way to sustain the claim that perception is in fact systematic. Again, our response will be negative. Our strategy will consist in showing that, in general, it is wrong to try to explain perception in terms of basic representational, language-like units, plus formal rules of inference over those units. For the classicists, the very...

Share