INTERACTION BETWEEN ADAPTATIONS 18/5 



system, coupled to a source of disturbance, tends to change its 

 step-functions to such values as will render it independent of the 

 source. Such a change must also occur if one part of a multis table 

 system is repeatedly disturbed by another part ; for the reacting 

 system possesses the essential properties, and the origin of the 

 disturbance is irrelevant. A subsystem is safe from such dis- 

 turbance when and only when its variables are independent of all 

 the other variables in the system. There is no necessity for me to 

 repeat the evidence here, for it is identical with that of Chapter 13. 

 It can therefore be predicted that as the various subsystems of a 

 multistable system act on one another, the tendency will be, as 

 time goes on, for the various subsystems to upset each other less 

 and less. 



If the nervous system is multistable it would show the same 

 tendency. It would thus show habituation twice : once in its 

 interactions with its environment and again between its various 

 component subsystems. Such ' intracerebral ' habituation will 

 tend to lessen the disturbing actions of part on part, and it will 

 therefore contribute to lessening the chaos described in S. 18/2. 

 But such a process will not always lead to complete adaptation ; 

 for its tendency, being always to remove interaction, is to divide 

 the whole into many independent parts. With some simple 

 environments such subdivision may be sufficient, as was noticed 

 in S. 17/3 ; but it contributes nothing towards the co-ordination 

 of reactions when a complex environment can be controlled only 

 by an intricate co-ordination in the nervous system. 



18/5. In the turbulence of many subsystems interacting, the 

 principle of ultrastability still holds and still acts persistently in 

 the direction of tending to improve the organism's adaptation 

 to its environment. It will still act selectively towards the 

 useful interactions. Suppose first that two subsystems interact 

 in such a way that, though individually adaptive, their com- 

 pounded reactions are non-adaptive/ A kitten, for instance, has 

 already learned that when it is cold it should go right up to the 

 warmth of its mother, and that when it is hungry it should go 

 right up to the redness of a piece of meat. If later, when it is 

 both cold and hungry, it sees a fire, it would probably tend, in the 

 absence of other factors, to go right up to it. But the very fact 

 that the interaction leads to non-adaptive behaviour provides the 



195 



