Unified theories and theories that mimic each other's predictions.

sion of how SOAR might apply to the learning of language, its comprehension and production. Use is made here of the important notion of chunking that Newell rightly brings to the fore throughout his discussion of many different psychological processes. On the other hand, for those who have attempted to deal with the troublesome and intricate details of language learning, comprehension, or production, what is said is at the most a very general prolegomenon to some future genuine theory of language. It is perhaps my central criticism of the book that Newell can set before us what he claims is an exemplar of a unified theory of cognition and yet provide only a very sketchy treatment of language. Many psychologists and philosophers would argue that it is the capacity for the use of language that is the most cognitively discriminating feature of human cognition as opposed to that of other animals. Newell likes to mention various well-established experimental cognitive phenomena, and this Is something I commend. On the other hand, there must by now be more than 10,000 articles reporting substantial psycholinguistic data on language acquisition. There is no attempt on Novell's part to deal with any of this overwhelmingly large and complicated experimental literature. I do not mean to suggest that he should give us some kind of encyclopedic report on this literature, but I do mean to suggest that any candidate theory of cognition must come to terms with language in real detail in order to make a serious claim to being a prototheory of human cognition.