DESIGN FOR A BRAIN 16/5 



The poorly joined environment 



16/5. We will finally consider the case in which the environment 

 consists of subsystems joined so that they affect one another only 

 weakly, or occasionally, or only through other subsystems. It 

 was suggested in S. 15/2 that this is the common case in almost 

 all natural terrestrial environments. 



If the degree of interaction between the subsystems varies, its 

 limits are: at the lower end, the iterated systems of S. 15/4 (as 

 the communication between subsystems falls to zero), and at the 

 upper end, the richly connected systems of S. 16/2 (as the com- 

 munication rises to its maximum). 



When the communication between subsystems falls much below 

 that within subsystems, the subsystems will show naturally and 

 prominently (S. 12/17). 



If such an environment acts within an ultrastable system, what 

 will happen ? Will adaptation occur ? As the discussion below 

 will show, the number of cases is so many, and the forms so 

 various, that no detailed and exhaustive account is possible. We 

 must therefore use the strategy of S. 2/17, getting certain type- 

 forms quite clear, and then covering the remainder by some appeal 

 to continuity: that so far as other forms resemble the type-forms 

 in their construction, to a somewhat similar degree will they 

 resemble the type-forms in their behaviour. 



16/6. To obtain a secure basis for the discussion of this most 

 important case, let us state explicitly what is now assumed: 



(1) The environment is assumed to be as described in S. 15/2, 

 so that it consists of large numbers of subsystems that have many 

 states of equilibrium. The environment is thus assumed to be 

 polystable. 



(2) Whether because the primary joins between the subsystems 

 are few, or because equilibria in the subsystems are common, the 

 interaction between subsystems is assumed to be weak. 



(3) The organism coupled to this environment will adapt by 

 the basic method of ultrastability, i.e. by providing second-order 

 feedbacks that veto all states of equilibrium except those that 

 leave each essential variable within its proper limits. 



(4) The organism's reacting part is itself divided into sub- 

 systems between which there is no direct connexion. Each sub- 



208 



