Neural Hardware for the Language of Thought: New Rules for an Old Game (2510.10251v1)

Abstract: The Language of Thought (LOT) hypothesis posits that at least some important cognitive processes involve language-like representations. These representations must be processed by appropriate hardware. Since the organ of biological cognition is the nervous system, whether biological cognition relies on a LOT depends on how neural hardware works. I distinguish between different versions of LOT, articulate their hardware requirements, and consider which versions of LOT are supported by empirical evidence. I argue that the Classical LOT hypothesis (Fodor 1975) is ruled out; the version of LOT that is best supported by empirical evidence is the Nonclassical LOT thesis that some neural representations mirror some of the structure of natural language and represent in a language-like way, yet they encode information nondigitally and are processed by ordinary (nondigital, and hence Nonclassical) neural computations that rely not only on syntactic structure but many other features.