10/11 THE RECURRENT SITUATION 



outside our present view) provided only that if the presentation 

 of a particular problem P t got through to some set S t , then always 

 when P t is presented again the actions shall again go through to 

 S { . Such a case would occur if the connexions were, say, electrical 

 and made by plugging connexions at random into a plug-board. 

 Once made* they would ensure that recurrence of P t would give 

 the same pattern for the selection of S { ; and the change from P t 

 to some other problem, P i say, by involving some change in the 

 sensory input to R, would cause some change in the distribution 

 over the step-mechanisms. 



In the same way, if nerve-cells were to grow at random (i.e. 

 determined in their growth by local temporary details of oxygen 

 supply, mechanical forces, etc.) until their histological details were 

 established, and if the paths taken by impulses depended on the 

 concatentation of stimuli coming in, then the recurrence of P t 

 would always give access to S i3 and a change from P z to P„ by 

 changing the sensory stimuli, would change the distribution. 



An easy method by which J 7 may be provided is given in S. 16/13. 



These details need not detain us. They are mentioned only to 

 show that the basic requirements are easily met, and that the 

 mechanism meeting them may look far less tidy than Figure 

 10/9/1 might suggest. In this sense the Figure, though helpful 

 in some ways, is apt to be seriously misleading. In S. 16/12 we 

 return to the matter. 



10/11. In the previous sections, the various situations P l9 P 2 , 

 P 3 , . . . were arbitrary, and not assumed to have any particular 

 relation between them. A special case, common enough to be of 

 interest, occurs when the situations usually occur in a particular 

 sequence. Thus a young child, reaching across the table for a 

 biscuit, may have first to get his hand past the edge of the table 

 without striking it, then the hand past his cup without spilling 

 it, then past the jam without his sleeve wiping it, and so on: 

 a sequence of actions, each of Which calls for some adaptation. 

 Much of life consists of just such sequences. 



The system of Figure 10/9/1 can readily give such sequences 

 in which every part is adapted to its own little problem. The 

 situation of ' hand coming past the edge of the table and in danger 

 of striking it ' is P v say. Adaptation to this situation can occur 

 in the usual way, by the basic method of the ultrastable system. 



145 



