CHAPTER 11 



The Fully-joined System 



11/1. The Homeostat is, of course, grossly different from the 

 brain in many respects, one of the most obvious being that while 

 the brain has a very great number of component parts the Homeo- 

 stat has, effectively, only four. This difference does not make 

 the theory of the ultrastable system totally inapplicable, for much 

 of the theory is true regardless of the number of parts, which is 

 simply irrelevant. Nevertheless, we are in danger, after spending 

 so much time getting to know a system of four parts, of developing 

 a set of images, a set of working mental concepts, that is seriously 

 out of proportion if considered as a set of working concepts with 

 which to think about the real brain. Let us therefore consider 

 specifically the properties of the ultrastable system that has a very 

 great number of parts. We shall find that one difficulty becomes 

 outstanding, and we shall spend the remainder of the book dealing 

 with it, for it is the main problem in the adaptation of large 

 systems. 



Adaptation-time 



11/2. Suppose the Homeostat were of a thousand units. Such 

 a size goes only a small fraction of the way to brain-size, but it 

 will serve our purpose. Such a system will have a thousand relays 

 (F, Figure 8/2/3); let us suppose that all but a hundred are 

 shorted out, so that the system is left with one hundred essential 

 variables. This number may be of the same order of size as the 

 number of essential variables in a living organism. 



Since these variables are essential, adaptation implies that all 

 are to be kept within their proper limits. Let us try a rough 

 preliminary calculation. Simplifying the situation further, let us 

 suppose that the step-mechanisms, as they change, give to each 

 essential variable a 50-50 chance of going within, or outside of, 

 its limits, and that the chances are independent. (The case of 

 independence should be considered, by the strategy of S. 2/17, 



148 



