240 BRAIN MECHANISMS AND LEARNING 



sees that the animal seems set to learn one particular thing and notliing else. The 

 instinctive mechanism is ready to take it just so tar along the line towards perfection 

 of the particular performance but no further and from there on learning takes over, 

 hi such a situation the suggestion that the fundamental mechanisms are of the same 

 kind seems valuable. Also, it has been suggested that when animals are not in 

 perfect health the innate behaviour may fail to come out automatically in its full 

 perfection but can nevertheless be perfected by learning. But there is one point 

 where I think the differences are great: in innate motor patterns the action is often 

 highly elaborate and the innate adjustment may be very fine; but the innate 

 receptive mechanism by which responses arc released is seldom or never capable b}- 

 itself of coping with complex stimuli, hi other words releasers, if they have to be 

 recognized innately, without any learning coming into the process, must be very 

 simple. So in complex releasing situations learning has pertorce to take a corre- 

 spondingly more important part, since the innate component cannot take the 

 animal very far. hi this respect of complexity there does seem to be a significant 

 difference between the mechanisms for organizing motor patterns and those 

 controlling perceptual recognition. 



EccLES. Dr Galambos refers to overlapping neuronal fields for various functions 

 of the medulla and suggests to us that some new stimulus or new input will 

 galvanize this group of neurones into an entirelv new response. There need be no 

 mystery about this if you consider the mode of operation of neuronal networks. 

 I don't think we have considered these properties sufficiently. We must think of the 

 nervous system as composed of immense assemblages of neurones with patterns of 

 connections at each synaptic relay. If an impulse is going to be propagated, there 

 must be convergence — and therefore there must be a lot of collusion (if you like) 

 in neuronal operations, hi any knid of possible connections in neuronal networks it 

 can be demonstrated thai if you alter the input, you alter the output. The particular 

 situation that Dr Galambos mentions provides an example. In the whole of this 

 medullary assemblage of neurones there are all kinds of pathways and the particular 

 input you get by pharyngeal tickling operates in one particular way to give you the 

 vomiting output. We have to think of neuronal patterns and how varying 

 inputs give varying outputs and how modifications can be produced in this way by 

 habituation and learnnig. 



Galambos. There are of course several models one could use as one attempts to 

 visualize the principle of organization at work in neuronal networks and nuclei. 

 Dr Eccles prefers the input-output model in which the essential phenomena in- 

 volved are exclusively synaptic transmissions and excitation and inhibition. Surely 

 there is no law, however, against looking for some other organizing principle. No 

 less an authority than that grand old man of American comparative anatomy, C. 

 Judson Herrick, has said something is going on in the neuropile that we do not 

 understand. 



Olds. From Dr Gaiambos's slides, it seems that a common message pervades the 

 system at a certain stage of learning before the response is fully developed. We used 

 to think that the message was coded by the place to which it was projected, or b)- 

 the system of units to which it was projected, but tlois pervasiveness suggests that 

 the message may somehow be characterized by its pattern. This brings me to my 

 question — is there any characteristic difference observed between this complex 

 event, depending on the nature of the input signal or the nature of the reinforce- 



