Barsalou's hypothesis that mental representations are constructed by selecting parts of percepts encounters the same difficulties as other empiricist theories: They cannot explain concepts for which instances do not share perceptible features (e.g., furniture) or for which there are no relevant percepts (e.g., the end of time). Barsalou's attempt to reduce falsity to failed pattern matching is an elementary error, and the generativity of his simulators cannot be attained without nonterminal symbols. There is not now, and there never was, any (...) reason to be interested in empiricist theories of knowledge. Abstraction is a fundamental aspect of human cognition. (shrink)

Localist networks are symbolic models, because their nodes refer to extra-mental objects and events. Hence, localist networks can be combined with symbolic computations to form hybrid models. Such models are already familiar and they are likely to represent the dominant type of cognitive model in the next few decades.

Learning is the acquisition of knowledge, not of input/output mappings. The distinction between statistical and relational learning, as Clark & Thornton define those terms, is not useful because all human learning is relational. However, prior knowledge does influence later learning and the sequence in which learning tasks are encountered is indeed crucial. Simulations of sequence effects would be interesting.