Is the Brain a Calculating Machine? 145 



A great deal of effort has been devoted to attempts to build up 

 models, using these principles, which will possess at least some of the 

 abilities of brains. In the first place, can we find a basis for memory? 

 A good many years ago Dr Lorente de No pointed out that a num- 

 ber of neurones arranged in a circle would act as a permanently re- 

 verberating circuit. A pattern of impulses fed into the circuit at any 

 point will continue to circulate indefinitely. It is only necessary that 

 the neurones should be able to provide sufficient energy and that the 

 ring should be sufficiently long so that the first neurone is in a recep- 

 tive state when the message arrives back. Patterns of impulses could 

 thus be stored in the brain in a permanent or semi-permanent form. 

 Drs McCulloch and Pitts have investigated mathematically the pro- 

 perties of such networks of neurones and suggested that they provide 

 a possible basis of memory since the patterns of impulses will con- 

 tinue to circulate long after the immediate occasion which produced 

 them has passed. 



Although there is no doubt that innumerable circuits exist in the 

 brain, it is doubtful if this suggestion offers a satisfactory theory of 

 memory. Lashley has extended the idea somewhat. If the brain con- 

 tains innumerable circuits, how does a circulating impulse know how 

 to stay on its ov/n circuit and to ignore many others? Lashley has sug- 

 gested that a pattern of excitation spreads out from the neurones 

 originally stimulated. The activity which starts at one point in the 

 cortex may spread out and give rise to radiating waves of activity. 

 When many points are stimulated simultaneously, the different waves 

 of activity will interact with each other like the waves formed when a 

 number of stones are thrown into a pond, giving rise to interference 

 patterns and to a pattern of standing waves. This would perhaps give 

 a basis for understanding how general characteristics (called *univer- 

 sals') are recognized, irrespective of their size and position. The inter- 

 ference patterns finally produced would be similar, and would not 

 depend on such details. 



An alternative proposal was made by Drs McCulloch and Pitts. 

 They suggested that the sensory impulses are received on several 

 layers of neurones, which are, as it were, tuned to different scales of 

 magnitude. Each of these is brought, in turn, into a receptive condi- 

 tion by a 'scanning' mechanism. The signals sent on to the rever- 

 berating circuits are thus the sum or average of those produced by the 

 original pattern, when measured on the different scales, so that what 

 matters is the essential character of the pattern rather than its size 

 or position. 



Hebb has gone still further and suggested that the conducting cir- 

 cuits between neurones are themselves modified by the passage of 



K 



