From weng@cse.msu.edu Sat Mar 4 13:26:18 2000 From: Dr John J Weng Date: Sat, 4 Mar 2000 13:26:17 -0500 (EST) To: wdl@cse.msu.edu Subject: WDL Dis. #15 ======== WDL Dis. #15 (Bcc: all WDL participants) ====== George, > I conclude with the opinion that a completely developmental approach is not > optimal and is not followed by nature, human or otherwise. Cultures provide > animals with high-level programming, which increases both their > functionality and biases. Higher level organisms do not have enough > lifespan to learn everything they need from development. You made several great points, such as cultural impact and age impact to development. The idea of development includes all these points. Esther Thelen and other developmental psychologists will tell us the rich perspectives in developmental psychology: biological maturation, environmental learning, constructivist, cultural-context, among others. Again, we MUST NOT misunderstand development as ``starting from nothing,'' ``no programmed-in bias'' or anything like that. There exists no ``completely'' (if you mean exclusive) developmental approach. All the points you wrote should be considered in the design (for machine) or study (for organisms) of the following four levels of developmental mechanisms (see WDL Dis. #12): Inborn behavior level, Representation level, Architecture level, Timing level. The subject of human capability of understanding and using symbols is a very interesting subject. Based on the neuroscience literature that I am aware (just as a fan), I think that although brain can deal with symbols (only after extensive post-natal development), the brain DOES NOT use the (atomic) symbolic representation that our AI people are used to. Signals in the brain are largely numeric in nature (frequency of pulses, potentiation, chemical transmission, etc). Knowledge about each symbol is represented in the brain in a very distributed fashion, stored as sensorimotor traces across various cortices (e.g., visual, auditory, somatosensory, and limbic systems), evoked by sensory inputs, and observed externally as spatiotemporal behaviors (e.g., a sequence of motor control signals that generates the verbal word "table"). The neuroscientists on this discussion list are at a much better position to explain these than I. John