DESIGN FOR A BRAIN 9/10 



any nature, if they were sited inside the neurons in which they 

 exerted their action. 



Evidently much more knowledge will be necessary before we 

 can identify accurately this part of the second-order feedback. 

 What is most important at the present time is that we should 

 avoid unwittingly taking for granted what has yet to be demon- 

 strated. 



Are step-mechanisms necessary ? 



9/10. Since S. 7/12 we have been considering adaptation when 

 the variables affecting the reacting part R of Figure 7/5/1 behave 

 as step-functions. The justification given then was that this case 

 is of central theoretic importance because of its peculiar simplicity 

 and clarity: even if the nervous system contained no step- 

 mechanisms, the student of the subject would still find considera- 

 tion of this form helpful to get a clear grasp of how a nervous 

 system can adapt. But is there no justification stronger than 

 this ? Does the evidence, perhaps, prove that the process of 

 adaptation implies the existence of step-mechanisms ? 



9/11. We have already seen (S. 7/16) that even so typical a 

 mechanism as a Post Office relay cannot be said to be (uncondi- 

 tionally) a step-mechanism, for some ways of observing it (e.g. 

 over microseconds or over years) do not show this form; and no 

 particular mode of observation can claim absolute priority. Thus 

 even if some object in a Black Box gave convincing evidence that 

 it was a Post Office relay, the observer still could not claim that 

 the real object was a step-mechanism unconditionally. 



9/12. On the other hand, contrasted with a full-function the 

 step-function has a remarkable simplicity of behaviour, and not 

 every real object can be made to show such a simplicity. To 

 say of the object in the Black Box that it can be made to show 

 behaviour of step-function type is thus to say something un- 

 conditionally true about it. 



Again, if a system of three variables is studied and found to 

 produce such a field as that of Figure 7/20/1 (in which three 

 possible dimensions are reduced to two two-dimensional planes), 

 then again the observer can claim that he has demonstrated some- 



128 



