DESIGN FOR A BRAIN 8/12 



the front two magnets by a light glass fibre so that they had to 

 move together. Figure 8/11/1 shows a typical record of the 

 changes. Three units were joined together and were at first 

 stable, as shown by the response when the operator displaced 

 magnet 1 at D v At J, the magnets of 1 and 2 were joined so 

 that they could move only together. The result of the constraint 

 in this case was to make the system unstable. But the instability 

 evoked step-mechanism changes, and a new terminal field was 

 found. This was, of course, stable, as was shown by its response 

 to the displacement, made by the operator, at D 2 . But it should 

 be noticed that the new set of step-mechanism values was adjusted 

 to, or ' took notice of ', the constraint and, in fact, used it in the 

 maintenance of stability; for when, at R, the operator gently 

 lifted the fibre away the system became unstable. 



There are other unusual problems, of course, for which the 

 Homeostat's repertoire contains no solution; putting too powerful 

 a magnet at one side to draw the magnets over would set such a 

 1 problem ' ; so would a shorting of the relay F (Figure 8/2/3). 

 In such a situation the Homeostat, or any ultrastable system, 

 would have no state of equilibrium and would thus fail to adapt. 

 So, too, would a living organism, if set a problem for which its 

 total repertoire contained no solution. 



Some apparent faults 



8/12. It will be apparent that the principle of ultrastability, as 

 demonstrated by the Homeostat, does not seem to represent 

 adequately the great richness of adaptations developed by the 

 higher animals ; with some of the inadequacies we shall deal later 

 in the book. There are, however, some ' faults ' of the ultrastable 

 system that are found on closer scrutiny actually to support the 

 thesis that the living brain adapts by ultrastability. We will 

 examine them in the next few sections. 



8/13. If the relation of S. 7/5 does not hold between the essential 

 variables and the step-mechanisms, that is, if an ultrastable 

 system's critical surfaces are not disposed in proper relation to 

 the limits of the essential variables, the system may seek an 

 inappropriate goal or may fail to take corrective action when the 

 essential variables are dangerously near their limits. 



118 



