ABSTRACT
The systematic class of classical, vector-based types of neuroarchitectures comprises those connectionist models that attempt to solve the "variable binding" problem by integrative synchronization mechanisms. For this purpose, vector and tensor constructions are used in the context of continuous mathematics. The criteria of systematicity and compositionality for complex mental representations, which are guaranteed in the symbol structures of the Classical Symbol Theory in the context of discrete mathematics, are also preserved in the transformations into relevant vector-based constructions and mechanisms of the connectionist neuroarchitectures. Based on the basic assumption in modern cognitive science that neurocognition is computation, the fundamental question for Paul Smolensky is: What kind of computation takes place in the human brain or mind that is which model of a cognitive neuroarchitecture is to be preferred. Smolensky's subsymbolic theory of (neuro-)cognition stands to classical symbol theory, in the relationship between micro and macrotheory, for example in analogy to the relationship between classical and statistical thermodynamics in physical chemistry.
