Understanding the Brain p. 4 Josephson
5. The observer mechanism and the coordination of subsystems
Our scheme implies the presence of a large number of component subsystems, acting in accord with a range of
abstractions. In the present scheme the coordination between the parts is ascribed to the observer mechanism in
conjunction with the hierarchical organisation, or to problem-solving processes. Specifically, the observer
mechanism tests the performance of some combination of systems, and fixes the relevant information only if the
combination achieves some specified target. In a variant of this process, an abstract specification of a process
(as discussed later in connection with representational redescription) is tested instead, and in a further variant,
structures are created by rule and the rule tested (or it may be supplied by another person using language and
assumed to be valid).
In addition, aspects of the coordination may be innate and implicit in the hierarchical organisation, for example
in assigning priorities to particular activities in particular contexts. Or again, we can envisage the existence of a
level of integration higher than that of a specific plan, involving an abstraction that may be thought of as a life-
scheme.
6. Evolutionary aspects
There are many possible consistent constructions whose behaviour is consistent with survival, just as there are
many configurations for the body that are consistent with survival, with the different constructions relating to
different capacities. During the course of evolution new constructions, associated with particular abstractions,
develop, associated with new capacities. One may regard abstractions as a source of power in that they enable
particular things to be achieved. For example, naming things is an abstraction, associated with the existence of
a connection, which may be utilised in either direction, between a name and a thing named. Naming things
opens up new possibilities, whose realisation requires neural circuitry that can realise the processes concerned.
7. Application to language: preliminary concepts
The processes of language are more complex than mere naming but, according to the present proposals, arise in
the same general way, on the basis of a range of abstractions concerning language (of which the relationships
involved in naming form one), with corresponding processes. According to the observational emergence idea,
specific systems are developed in accord with specific tests. The proposals of Jackendoff (2002), who discusses
the various representations that appear to be involved in language, as well as also paths by which language
could have evolved from something very basic to modern language, are consistent with this general picture,
since we can consider each advance as involving a new abstraction. This will be discussed in some detail later
on, but first we discuss a key concept that arises in connection with any skill involving the use of abstractions,
namely that of representational redescription.
8. Subtler levels of behaviour: representational redescription
A concept relevant to understanding phenomena such as language is the representational redescription concept of
Karmiloff-Smith (1992), the basic idea of which is that having become competent at a given process we re-
represent our knowledge in a way that opens up more possibilities. An example is provided by the abstraction
cat , the use of which allows one to represent direectly knowledge relating to cats in general, rather than having
to treat each instance of a cat individually.