Rogers T.T., McClelland J.L. Semantic Cognition. A Parallel Distributed Processing Approach

MIT Press, 2004, -440 p.Attempts to understand the basis of human semantic abilities are central to the effort to understand human cognition. Since antiquity, philosophers have considered how we make semantic judgments, and the investigation of semantic processing was a focal point for both experimental and computational investigations in the early phases of the cognitive revolution. Yet the mechanistic basis of semantic cognition remains very much open to question. While explicit computational theories were offered in the 1960s and into the 1970s, the mid-1970s saw the introduction of findings on the gradedness of category membership and on the privileged status of some categories that these theories did not encompass. Eleanor Rosch, who introduced most of these phenomena, eschewed any sort of explicit mechanistic theorizing.
Subsequently a new thrust of research has emerged within a framework that is often called ‘‘theory theory.’’ This framework has been very useful as a springboard for powerful experimental demonstrations of the subtlety and sophistication of the semantic judgments that adults and even children can make, but it has left the field without an explicit mechanistic theory of the representation and use of semantic knowledge, since the fundamental tenets of theory theory are general principles whose main use has been to guide the design of ingenious experiments rather than the explicit formulation of computational mechanisms.
The subtlety of the semantic capabilities of human adults and even of rather young children, and the ways these abilities evolve and become elaborated over the course of development, make the prospect of offering any sort of mechanistic framework for understanding semantic cognition daunting. Nevertheless, we find ourselves in the position of offering, if not a full characterization of the mechanistic basis of semantic knowledge, then at least some demonstrations of the properties of a type of mechanism that may suggest the general form such a characterization might take.
When we began our work we had a specific and rather narrow focus: to address the progressive differentiation of conceptual knowledge in development and the progressive deterioration of conceptual knowledge in dementia. As we explored a model that exhibited these properties, we found ourselves thinking of ways our approach might address other findings in the literature as well. While we would not disagree with those who may feel that we are still far from a complete model of human semantic abilities, we nevertheless have come to feel that our modeling work has addressed a sufficient range of basic findings for it to be useful to lay out what we have done in the form of an integrative theoretical statement. Our main goal in presenting our work at this point is to solicit the participation of others in its further development. We hope to gain the benefit of the reactions of others who have considered these matters. We also hope that our progress to date will seem promising enough that some researchers will join us in the further development of our approach.Categories, Hierarchies, and Theories
A PDP Theory of Semantic Cognition
Latent Hierarchies in Distributed Representations
Emergence of Category Structure in Infancy
Naming Things: Privileged Categories, Familiarity, Typicality, and Expertise
Category Coherence
Inductive Projection and Conceptual Reorganization
The Role of Causal Knowledge in Semantic Task Performance
Core Principles, General Issues, and Future Directions
A: Simulation Details
B: Training Patterns
C: Individuating Specific Items in the Input