This paper extends previous proposals concerning the design of the nervous
system given in papers at the ECHO and NESCI conferences (the ECHO paper
can be found at
http://www.tcm.phy.cam.ac.uk/~bdj10/papers/amiens/
and the NECSI paper at
http://www.tcm.phy.cam.ac.uk/~bdj10/papers/ICCS97/abstract.html).
The nervous system is idealised as a collection of capacities, generated by
corresponding mechanisms which drive the individual to attempt to
accomplish particular kinds of tasks, and provide support for such
attempts. The feasibility of such mechanisms is related to the
extendibility of the activities concerned, i.e. the ease with which
extended goals can be achieved on the basis of existing accomplishments.
If activities are extendible in principle in this sense, then the actual
nervous system design is a matter of arranging for the appropriate
exploration of the space of accomplishments to take place, with retention
of successes in the form of new modules which can be optimised, a process
leading to the emergence over time of the new capacity contingent upon the
design. This theory has implications for how we view capacities such as
language. We are programmed to attempt particular kinds of task in a
linguistic environment, and acquire thereby a system embracing the various
aspects of a given language. The operations performed by this system are
extendible to a wide range of other tasks: individuals explore these
possibilities and thereby create extensions to the language system. It is
the activities of these extensions that we see as language use. Similar
ideas apply to other fields of mental activity, such as logic.