16/5 DESIGN FOR A BRAIN 



either immediately affects the other, and that for some reason 

 all the other subsystems are inactive. 



The first point to notice is that, as the other subsystems are 

 inactive, their presence may be ignored ; for they become like 

 the 4 background ' of S. 6/1. Even if some are active, they can 

 still be ignored if the two observed subsystems are separated 

 from them by a wall of inactive subsystems (S. 14/8). 



The next point to notice is that the two subsystems, regarded 

 as a unit, form a whole which is ultrastable. This whole will 

 therefore proceed, through the usual series of events, to a terminal 

 field. Its behaviour will not be essentially different from that 

 recorded in Figure 8/8/5. If, however, we regard the same 

 series of events as occurring, not within one ultrastable whole, 

 but as interactions between two subsystems, then we shall observe 

 behaviours homologous with those observed when interaction 

 occurs between ' animal ' and l environment '. In other words, 

 within a multistable system, subsystem adapts to subsystem in exactly 

 the same way as animal adapts to environment. Trial and error 

 will appear to be used ; and, when the process is completed, the 

 activities of the two parts will show co-ordination to the common 

 end of maintaining the variables of the double system within the 

 region of its critical states. 



Exactly the same principle governs the interactions between 

 three subsystems. If the three are in continuous interaction, 

 they form a single ultrastable system which will have the usual 

 properties. 



As illustration we can take the interesting case in which two 

 of them, A and C say, while having no immediate connection 

 with each other, are joined to an intervening system B, inter- 

 mittently but not simultaneously. Suppose B interacts first with 

 A : by their ultrastability they will arrive at a terminal field. 

 Next let B and C interact. If B's step-functions, together with 

 those of C, give a stable field to the main variables of B and C, 

 then that set of J5's step-function values will persist indefinitely ; 

 for when B rejoins A the original stable field will be re-formed. 

 But if Z?'s set with C's does not give stability, then it will be 

 changed to another set. It follows that B's step-functions will 

 stop changing when, and only when, they have a set of values 

 which forms fields stable with both A and C. (The identity in 

 principle with the process described in S. 11/4 should be noted.) 



174 



