Book review of Jeffrey Elman

Jeffrey Elman, Elizabeth A. Bates, Mark H. Johnson, Annette Karmiloff-Smith, Domenico Parisi, Kim Plunkett:
RETHINKING INNATENESS (MIT Press, 1996)

Home | The whole bibliography | My book on Consciousness

An international group of distinguished psychologists (Jeffrey Elman, Elizabeth Bates, Mark Johnson, Annette Karmiloff-Smith, Domenico Parisi, Kim Plunkett) got together to discuss where does knowledge come from. Their perspective is unabashedly connectionist and interactionist/constructivist. Knowledge comes from the interaction of nature and nurture, and is acquired through a connectionist process.

The first part of the book deals with basic biological facts, and demistifying the "digital gene" metaphor, and emphasizing that the relationship between genotype and phenotype is nonlinear (a small difference in genotype can cause a huge difference in phenotype, as in the cases of chimps and humans).

Then the book shifts gear and delves into connectionist models that simulate child development Basically, the idea seems to be to find a mathematical function that adequately describes child learning and then find a neural network that reproduces that mathematical function. Thus we are guided to what amounts to a differential calculus for studying children (replete with derivatives and dynamic equations).

Then we shift gear again and we are guided through a tour of the brain and how it develops. The idea, again, is to show that the dynamics is the one that we would expect from neural networks.

The mission of the book is fundamentally methodological: it shows that connectionism (neural networks) constitute a valid framework to study child development.

The thesis enunciated at the beginning is only implicitly explored. Basically, the authors show that "innate" knowledge does not need to be hardwired. "Innate" knowledge can be determined by constraints in space and in time, that limit what we can "know". Thus it is not really innate.