Rewrite Science - The Universal Semantic Calculus and Grammatical Cosmos
Symposium 10A - CASYS 07

The Logic of Nervous System Design
Brian D, Josephson FRS
Department of Physics, University of Cambridge J J Thomson Avenue, Cambridge CB3 OHE, United Kingdom
+44 1223 337200, bdil0@cam.ac.uk, http://www.tcm.phy.cam.ac.uk/~bdj10

Keywords: brain, skill development, language, representation. evolution

Abstract

This paper attempts to gain insight into the brain by perceiving it as a system that became gradually elaborated over time in response to a variety of needs, which elaborations led to it gradually advancing so as to populate new places in a notional 'skill space".
Achieving new skills, or higher levels of a given skill, demands performing certain additional computations, associated with modifications in the architecture, but these modifications can implement specific computation;, only to the extent that there are specific relationships between neural processes and ordinary computations (i..e. operations upon variables).

The processes supported by the nervous system, as required by the various specializations - must be structured in such a way that representational connections of this kind will come into existence and be maintained (cf. the symbol grounding issue). This involves for example connection schemes and activity patten-is evolving so that particular neural systems tend lo take upon particular roles.

In the task of organization of behaviour, a crucial role is played by systems whose activity corresponds to the specification of context. This is relevant in situations such as top-down or hierarchical organization, the ability to resume a task that has been interrupted, and the ability to address situations outside the current one, as for example with planning or many uses of language, Here there are connections with the .multi-level representational schemes of Karmiloff-Smith.

The practical application of these ideas is. just as with the practical application of an\ ideas relevant to the understanding of a particular kind of situation, is that a vast collection of data gets put into some kind of order; by adapting the concepts to specific situations it should be possible to understand why things work out the way that they do. as opposed simply to knowing that they do. The paper finishes with discussion of a number of specific cases.