DESIGN FOR A BRAIN 10/5 



will appear on two successive nights at the same place, followed, 

 on the third night, by a lethal mixture of the same food and poison 

 (the method of ' pre-baiting '). Environments such as this are 

 intrinsically complex. Complete adaptation here (under the 

 assumptions made) demands the reaction-pattern: eat, eat, 

 abstain. This reaction-pattern is more complex than the simple 

 reaction-pattern of eating, or of abstaining: for the three parts 

 must be related, and the triple organised holistically. 



In this chapter we shall consider the other case, of frequent 

 occurrence, in which what is appropriate to the disturbance is 

 conditional on which disturbance it is, but not on when it occurs 

 in the sequence of disturbances. 



So far, the ultrastable system (represented, say, by the Homeo- 

 stat) has been presented (e.g. in the Figures throughout Chapter 8) 

 with changes of parameter-value such that the later value is 

 merely different from the earlier; now we consider the case in 

 which the parameter takes a sequence of values, e.g. 



in which repetitions occur at irregular intervals, and in which a 

 response to P 2 , say, if adaptive on P 2 s first occurrence, is also 

 adaptive to P 2 on its later occurrences. 



When this is the case, the opportunity exists for advantage to 

 be taken of the fact that P 2 can be responded to at once on its 

 later occurrences, without the necessity of a second exploratory 

 series of trials and errors. 



This case is particularly important because (S. 8/10) it includes 

 the case in which the changes of P-value correspond to changes 

 from one environment to another. Suppose, for instance, that a 

 wild rat learns first to adapt to conditions in a stable {P 2 ), then 

 to conditions in a nearby barn (P 3 ), and so on. Haying adapted 

 first to the stable and then to the barn, its survival value would 

 obviously be enhanced if it could return to the stable and at once 

 resume the adaptations that it had previously developed there. 

 An organism with such a power can accumulate adaptations. 



10/5. To see what is necessary, let us see what happens in the 

 Homeostat. A little reflection, or an actual test, soon shows that 

 the present model is totally devoid of such power of accumulation. 

 Thus in Figure 8/4/1 the reversal at R 2 restores the external 



140 



