DESIGN FOR A BRAIN 13/2 



importance only as a typical, generic, or ' random ' element of the 

 set it belongs to. Though the argument will often appear verbally 

 to be focused on an individual system, it is directed really at the 

 properties of the set, the individual system being introduced only 

 as a means to an end. 



We shall have much to do, in what follows, with systems con- 

 structed in some 4 random ' way. The word will always mean that 

 we are discussing some generic system so as to find its typical 

 properties, and thus to arrive at some precise deduction about the 

 defined set of systems. 



13/2. A set of systems of special importance for the later 

 chapters is the set of those systems that are made of parts that 

 have a high proportion of their states equilibrial, and are made 

 by the parts being joined at random. 



More precisely, assume that we have before us a very great 

 number of parts, assumed to be fairly homogeneous, so that there 

 is a defined ' universe ', or distribution, of them. Each is assumed 

 to be state-determined, and thus to have in it no randomness 

 whatever. As a little machine with input, if it is at a certain state 

 and in certain conditions it will do a certain thing; and it will do 

 this thing whenever the state and conditions recur. 



We now take a sample of these parts by some clearly defined 

 sampling process and thus arrive at some particular set of parts. 

 (It is not assumed that all parts have an equal probability of being 

 taken.) Again we take a sample from the possible ways of 

 joining them, taking it by a clearly defined sampling process, and 

 thus arrive at some one way of joining them. 



The particular set of parts, joined in the particular way, now 

 gives the final system. 



This particular final system, be it noticed, is state-determined. 

 It is not stochastic in the sense of being able, from a given state 

 and in given conditions, to undergo various transitions with 

 various probabilities. Thus the particular system is not random 

 at all. The randomness enters with the observer or experi- 

 menter; he is little interested in the particular system taken by 

 the sampling, but is much interested in the population from 

 which the particular system has come, as the neurophysiologist is 

 interested in the set of mammalian brains. The ' randomness ' 

 comes in because the observer faces a system that interests him 



172 



