Knowledge, intelligence and expertise

This course examines why constraints linked to cerebral capacity tend to inflect the induction of new concepts. The idea that limitation of capacity offers benefits for efficient learning in humans is developed to show that expertise is a product of both complexity and simplicity. Occam's razor is a key idea to link human and artificial intelligence. The course also describes the differential contribution to the development of intelligence in humans of the three following layers: phylogeny, sociogeny and psychogeny. We then review the Flynn effect (the debate on whether the rise in IQ scores truly corresponds to higher general intelligence in the recent evolution), the relationship between memory capacity and intelligence, memory training, deliberate practice, individual differences, and cultural effects. Comparison is made with animal learning and IQ tests are scrutinized in that respect. Computational models (e.g. CHREST) are studied and bridges are made to artificial intelligence (AI). The AI section of the course first considers symbolic AI approaches to knowledge representation, extraction and reasoning (formalisms, ontologies, and semantic Web), and distributed AI architectures focusing on multi-agent approaches. A second part covers machine learning. If most of the course focuses on current perspectives in psychology and AI, other aspects in neuroscience are also integrated to cover brain reorganization.