Introduction

This book appeared in 2001, when artificial intelligence was methodologically plural. Bayesian inference, genetic algorithms, and symbolic systems each claimed territory. Neural networks, starved of compute, had receded from the field's center. The networks treated here are classical: fixed architectures whose explanatory power lives in connection weights between nodes. Yet the author frames them as connectionist models of mental representation — structures that mirror how brains encode and combine thought. That framing, read now, looks prescient.

In the two decades since, neural networks absorbed the field. They became, for most practical purposes, synonymous with AI. This report situates the book's arguments against that transformation — tracing where its claims held, where technology outran them, and what the convergence of deep learning and cognitive science now demands we reconsider.

Next: Outline of the Book

Prev: Introduction