Review: Advances in Cognitive Science, Volume 2. Theory and Applications, Perspectives on Dyslexia 2

This book consists of ten articles divided into a theoretical section, covering symbolic and subsymbolic processing in learning and problem solving, and an applications section, concerned with computer vision, object and face recognition, arithmetical and semantic concepts, and industrial control. Many of the articles also contain useful reviews of relevant publications in both French and English. In his editorial introduction, Guy Tiberghien, of Grenoble University, claims that cognitive science is a research program with its own methodology, experimental paradigms, theories, and models. His definition of 'cognitive' is information processing in the CNS, including subsym-bolic and automatic levels as low as synaptic parameters of connectionist paradigms. He argues that mental phenomena of a symbolic nature are produced by other units situated at an inferior subsymbolic level. However, he admits to some difficulties in understanding the role of higher-level processes in cognition (such as affect), and finds the process of belief generation "totally opaque". Although these demarcation issues may be forced upon us by the academic system, they are unlikely to benefit cognitive science in the long run. Where they reduce communication between cognitive scientists and higher and lower levels of analysis they weaken the constraints which knowledge of strategic goals and implementational details can impose on irrelevance and error at the level of algorithmic production. There is also the question of scientific coherence and economy; clearly there are complex processes of recognition, communication, and adaptive control taking place both at microbiological and social levels, and powerful developmental constraints operate between levels. Ideally, cognitive science should provide a unified developmental account for data processing at all levels. Particular difficulties can be caused by considering information processing in the CNS out of its cultural and social context, as if each brain develops monadically in isolation, without benefit of educational processes or cultural evolution. Sociobiology has highlighted the individual's need to develop the cognitive capacity to model the strategies of rivals and allies in competitions for scarce resources essential to fitness, and social psychology has revealed group pressure on individuals to conform to the beliefs values and programs of successful cultures. In this context, it was interesting to find that none of the more theoretical papers take explicit account of the social dimension. Murphy and Winieski (chapter 1) elegantly demonstrated that learning iV-feature concepts does not involve laboriously recording the N(N-l) correlations between individual features and that only the N links from features to …