A review of parallel distributed processing

Undoubtedly the most influential text in neural computing (or connectionism) has been that partly written by ang edited by David Rurnelhart and Jay McClelland on behalf of the ‘Parallel Distributing Processing’ Group in the USA. In contrast with the ring of engineering in the title, the sub-title ‘Explorations in the Microstructures or Cognition’, reveals the common interest of the 16 members of this group. I have set out to review in detail what is said in the 26 or so chapters of these two volumes. There are several reasons for this, but my intention is not to prevent anyone reading the volumes themselves. The main reason is diametrically opposite: I have had great pleasure in reading these books, particularly as they have made clear many issues that have occupied my mind over the last 20 years or so. So I hope that the short reviews will act as appetizers which direct the reader to the detailed ideas contained in the books. Furthermore, the authors of this present book make copious references to PDP chapters, and it is for completeness that it is necessary to make some of this material available. But the review is not a passive paraphrase. It is mostly my own understanding of the subject matter, which comes from a background of design and engineering of neural systems for application rather than brain modelling. I have also added personal comments not expressed in PDP books, in square brackets, where I felt it necessary.