systematicity - A number of putative psychological properties or regularities go by the name of systematicity. These diverse regularities are meant to constitute explananda that are supposed to support the view that there exists a syntactically and semantically combinatorial language of thought. See productivity of thought, compositionality, symbolicism.
[Introduction][Systematicity argument][Systematicity of inference][References]
The locus classicus of the so-called systematicity argument(s) is Fodor and Pylyshyn, (1988). See systematicity argument of cognitive representations, compositionality of representations, productivity of thought, inferential coherence, systematicity of inference. Other developments of a systematicity argument may be found in Fodor (1987), Fodor & McLaughlin, (1990), and McLaughlin, (1993a), (1993b).
A number of somewhat different arguments have come to be clustered under the heading of "the systematicity argument". The locus classicus for the argument is the systematicity of cognitive representations argument found in Fodor & Pylyshyn, 1988. This discussion is the basis of the present account.
According to Fodor and Pylyshyn, there are intrinsic connections certain thoughts and certain other thoughts. Were a normal cognitive agent to lack some thoughts that cognitive agent would also lack certain other thoughts. The point is not, for example, that the lesioning of one thought would automatically bring about the lesioning of another thought. This would presumably be an implementational fact, rather than a psychological fact requiring a psychological explanation.
Fodor and Pylyshyn claim that the systematicity of cognitive representations is a psychological fact that is in need of an explanation. Why is there an intrinsic connection between some thoughts and other thoughts? A semantically and syntactically combinatorial language of thought is supposed to provide an explanation. Thoughts involve mental representations that are made up of meaningful parts. Thus, thinking that John loves Mary might involve as meaningful parts representations of John, loves, and Mary. If it is the case that the thought that John loves Mary and the thought John likes tomatoes both have a representation of John as proper parts, then one can see why it is that there is some intrinsic connection between the thought that John loves Mary and the thought that John likes tomatoes. Were one to lack a representation of John, then one would lack the ability to think both that John loves Mary and that John likes tomatoes.
Rival accounts of mental structure, such as the view that there exist mental representations that lack combinatorial structure, either do not or cannot explain the systematicity of cognitive representations. If a normal cognitive agent has only a single, syntactically atomic representation, say, Þ, that means that John loves Mary and another single, syntactically atomic representation, say, ß, that means that John likes tomatoes, then it is unclear why there should be any intrinsic connection between the thought that John loves Mary and the thought that John likes tomatoes. Why, on this view, should it be the case that normal cognitive agents that are unable to think that John loves Mary are also unable to think that John loves tomatoes?
Fodor & Pylyshyn suppose that this argument applies to connectionism, because, they suppose, connectionism is committed to mental representations that lack syntactic and semantic combinatorial structure.
Jerry Fodor and Zenon Pylyshyn suggest that normal cognitive reasoning is systematic. By this, they mean that if a normal cognitive agent that can perform one instance of a type of logical inference, then, ceteris paribus, that agent can perform other instances of that same type of inference. Thus, any normal cognitive agent that can perform one instance of conjunction elimination (modus ponens, modus tollens, disjunctive syllogism) can, ceteris paribus, perform another instance of conjunction elimination (modus ponens, modus tollens, disjunctive syllogism).
Fodor and Pylyshyn argue that this putative regularity in normal cognition provides reason to believe that there exists a syntactically and semantically combinatorial language of thought, because such a language of thought provides a good explanation of the systematicity of inference. The reason inference is systematic, according to Fodor and Pylyshyn, is that 1) there are syntactic and semantic atoms that are composed to form complex mental representations and 2) there are mechanisms that are sensitive to this semantic and syntactic form. The mechanisms use the similarities among inferences of a given type as the basis for performing all the inferences of a given type. Thus, all conjunctions may be supposed to have some common syntactic and semantic structure, e.g., the occurrence of a main conjunction symbol. The mechanism that performs once instance of a conjunction elimination, e.g. inferring P from P&Q, also ceteris paribus performs other instances of conjunction elimination. All instances of modus ponens may be supposed to have a common syntactic and semantic structure, so that processes that enable the inference from P and P -> Q to Q, ceteris paribus, also enable the inference from PvQ and (PvQ) -> R to R.
Rival accounts of mental structure and function, such as the view that there exist mental representations that lack syntactic and semantic combinatorial structure, do not or cannot explain why it is that normal cognitive agents that draw one instance of an inference of a given logical can, ceteris paribus, also draw other instances of the same logical type. Suppose that the mental process of inferring P from P&Q involves moving from the representation Þ, meaning P&Q, to a representation ß, meaning P. Why, then, should it be the case that there is also a mental process, say, inferring P from P&Q&R that involves moving from the representation µ, meaning P&Q&R, to a representation , meaning P?
Fodor & Pylyshyn suppose that this argument applies to connectionism, because, they suppose, connectionism is committed to mental representations that lack syntactic and semantic combinatorial structure.
Fodor, J. (1987). Psychosemantics. Cambridge, MA: MIT Press.
Fodor, J., and McLaughlin, B. (1990) 'Connectionism and the problem of systematicity: Why Smolensky's solution doesn't work.' Cognition, 35, 183-204.
Fodor, J., and Pylyshyn, Z. (1988). 'Connectionism and cognitive architecture: A critique.' Cognition, 28, 3-71.
McLaughlin, B. (1993a). 'The Classicism/Connectionism battle to win souls.' Philosophical Studies, 70, 45-72.
McLaughlin, B. (1993b). 'Systematicity, conceptual truth, and evolution.' In Hookway, C., and Peterson, D. (eds.). Philosophy and cognitive science. Cambridge. (pp. 217-234).
Thanks to our sponsors: Custom logo design by logobee