GENERAL DISCUSSION 655 



the same name may yet differ in details to the extent that divergent results are 

 obtained with animals having presumably identical brain lesions. Dr Rosvold has 

 already given us an interesting example of such a situation in which the amount of 

 energy an animal must exert in responding may determine whether or not a 

 dehcit is seen. Other important variables are the degree oi dehnition ot tigure from 

 ground; whether stnnuli are presented simultaneously or successively; the nature 

 and amount of punishment given for incorrect response and the like, hi a word 

 investigators working with learning in relation to brain lesions must be acutely 

 aware of the behavioural variables in their experiments and must exercise them- 

 selves excessively to maintain consistency in these details throughout an experi- 

 mental series, and, finally must give good enough account ot these variables that 

 others may determine precisely what had been done. 



FOURTH topic: NATURE OF THE CHANGE 



EccLES. As an investigator of the properties ot nerve cells and neural pathways in 

 the central nervous system, I have been particularly interested in the various 

 theoretical accounts of the neural changes that are responsible for conditioned 

 reflexes. There have been many attempts to explain learning and conditioning by 

 the growth and development ot synapses, as tor example by Tanzi in 1893; by 

 Cajal in 1911, and by Hebb in 1949; but no model has been presented that con- 

 forms with known neurophysiological properties of neurones, and at the same 

 time accounts tor the simplest phenomena of learning, i.e. conditioned reflexes ot 

 the Pavlovian type. Some years ago in m\- book The Ncuropliysiolo{;ical Basis of 

 A4iiid, I discussed critically two types oi these postulated changes and presented in 

 outline a new hypothesis that at least had the merit ot being based on properties ot 

 the nervous system that had been experimentally demonstrated. So far as I know 

 this hypothesis has aroused ver)- little interest either from psychologists or from 

 investigators of conditioned reflexes, so I welcome this opportunity to present it to 

 you in the hope that there will be some critical discussion. 



Fig. I A shows diagrammatically an explanation of conditioning that has been 

 developed by Konorski and Gastaut. The conditioned and unconditioned stimuli 

 (CS and US) activate respectively the emitting centre (EC) and the receiving centre 

 (RC). They postulated that the potential connections between these two centres 

 (dotted lines) are transformed into 'actual excitatory connections' when excitation 

 in the tirst centre coincides in time with a rise in excitation in the second centre. 

 This change in connections between the centres was envisaged as being due to a 

 growth and multiplication ot synaptic connections, but unfortunately there is no 

 experimental evidence that activation of a nerve centre (RC) could cause the 

 development of connections onto it from an adjacent activated centre (EC) or even 

 an increased eflicacy of connections already in existence. An alternative postulate is 

 illustrated in Fig. iB, which shows simply that the CS and US lines are assumed to 

 converge on the same neurones, activation bv the US in close temporal contiguity 



