CHAPTER 11 



Fully Connected Systems 



11/1. In the preceding chapters all major criticisms were post- 

 poned : the time has now come to admit that the simple ultra- 

 stable system, as represented by, say, the homeostat, is by no 

 means infallible in its attempts at adaptation. 



But before we conclude that its failures condemn it, we must 

 be clear about our aim. The designer of a new giant calculating 

 machine and we in this book might both be described as trying 

 to design a ' mechanical brain '. But the aims of the two 

 designers are very different. The designer of the calculator wants 

 something that will carry out a task of specified type, and he 

 usually wants it to do the work better than the living brain can 

 do it. Whether the machine uses methods anything like those 

 used by the living brain is to him a side-issue. My aim, on the 

 other hand, is simply to copy the living brain. In particular, 

 if the living brain fails in certain characteristic ways, then I want 

 my artificial brain to fail too ; for such failure would be valid 

 evidence that the model was a true copy. With this in mind, 

 it will be found that some of the ultras table system's failures 

 in adaptation occur in situations that are well known to be just 

 those in which living organisms also are apt to fail. 



(1) If an ultrastable system's critical surfaces are not disposed 

 in proper relation to the limits of the essential variables (S. 9/1), 

 the system may seek an inappropriate goal or may fail to take 

 corrective action when the essential variables are dangerously 

 near their limits. 



In animals, though we cannot yet say much about their critical 

 states, we can observe failures of adaptation that may well be due 

 to a defect of this type. Thus, though animals usually react 

 defensively to poisons like strychnine — for it has an intensely 

 bitter taste, stimulates the taste buds strongly, and is spat out 

 — they are characteristically defenceless against a tasteless or 

 odourless poison : precisely because it stimulates no nerve-fibre 



130 



