DESIGN FOR A BRAIN 17/9 



levels: 2£'s, F's, and so on. Would that it were so ! When it 

 occurs, the whole act of adaptation (really life-long as we saw in 

 S. 10/2) can be divided into portions; then the practical scientist 

 can study the system portion by portion, level by level, and can 

 thus greatly simplify its study. The Homeostat was designed 

 partly so as to enable two levels to be obviously distinguishable: 

 (1) the four continuous variables at the magnets and (2) the 

 discontinuous variables on the uniselectors. When a system has 

 this natural internal division, the observer can take advantage 

 of the fact to describe the somewhat complex whole in three 

 stages, each considerably simpler: the continuous system and its 

 properties, the discontinuous system and its properties, and the 

 interaction between them. But when the whole system is not so 

 divisible it remains merely a fearfully complex whole, not capable 

 of reduction, and therefore as intractable to the scientist as the 

 examples in S. 16/3. 



This book inevitably concerns itself with the case in which the 

 essential variables are divisible clearly into levels: the primary 

 levels (of E v E 2 , E 3 , E A ) in Chapters 7 to 10, and then a sharply 

 differentiated F in Chapters 11 to the present. In this it was 

 again following the strategy of S. 2/17, getting a clear grasp of 

 the manageable cases so that they could serve as a basis for at 

 least a distant survey of the unmanageable. The reader will now 

 appreciate that the simplicity of the earlier chapters was essentially 

 a didactic device, not resembling the actual complexity of actual 

 organisms. In fact, their real complexity is greater, by many 

 orders of size, than that considered here. Thus, the reacting part 

 R of Figure 7/5/1, which looks so simple, may not only contain 

 the complexities of the multistable system (Figure 16/6/1) but 

 also, in the higher organisms, many subsystems of the form of 

 Figure 7/5/1 itself, each with its own little sub-essential variables 

 and sub-adaptations; for much adaptation to long- term goals is 

 achieved by finding suitable sets of sub-goals, perhaps in complex 

 sequences of timing and conditionality. Thus once we have used 

 the carefully simplified forms of Figures 7/5/1 and 16/6/1 to 

 establish our understanding, we must be prepared to admit that 

 in the real brain the same principles work in a complexity that is 

 of an altogether higher order, one that may well prove to be for ever 

 beyond the detailed comprehension of the human scientist, who has 

 an I.Q. limited, for all practical purposes, to something below 200. 



226 



