DESIGN FOR A BRAIN 16/9 



join. Always we can predict that their step -mechanisms will stop 

 changing when, and only when, the combinations are all stable. 

 Ultrastable systems, whether isolated or joined in multistable 

 systems, act always selectively towards those step-mechanism 

 values which provide stability. 



16/9. At the beginning of the preceding section it was assumed, 

 for simplicity, that the process of dispersion was suspended, for 

 we assumed that the two subsystems interacting remained the 

 same two (e.g. A and B of Figure 16/6/1) during the whole pro- 

 cess. What modifications must be made when we allow for the 

 fact that in a multistable system the number and distribution of 

 subsystems active at each moment may fluctuate by dispersion ? 



The progression to equilibrium of the whole, to a terminal field, 

 and thus to adaptation of the whole, will occur whether dispersion 

 occurs or not. The effect of dispersion is to destroy the indi- 

 viduality of the subsystems considered in the previous section. 

 There two subsystems were pictured as going through the complex 

 processes of ultrastability, their main variables being repeatedly 

 active while those of the surrounding subsystems remained 

 inactive. This permanence of individuality can hardly occur 

 when dispersion occurs. Thus, suppose that a multistable 

 system's field of all its main variables is stable, and that its repre- 

 sentative point is at a state of equilibrium R. If the representa- 

 tive point is displaced to a point P, the lines from this point will 

 lead it back to R. As the point travels back from P to R, sub- 

 systems will come into action, perhaps singly, perhaps in com- 

 bination, becoming active and inactive in kaleidoscopic variety 

 and apparent confusion. Travel along another line to R will 

 also activate various combinations of subsystems; and the set 

 made active in the second line may be very different from that 

 made active by the first. 



In such conditions it is no longer profitable to observe par- 

 ticular sybsystems when a multistable system adapts. What 

 will happen is that so long as some essential variables are outside 

 their limits, so long will change at step-mechanisms cause com- 

 bination after combination of subsystems to become active. But 

 when a stable field arises not causing step-mechanisms to change, 

 it will, as usual, be retained. If now the multistable system's 

 adaptation be tested by displacements of its representative point, 



212 



