GENERAL DISCUSSION 659 



formation and related structures) which tacihtatcs responses to 'important' inputs 

 and which favours and fixes 'good' outputs. 



In contrast to most models that have been offered, which are either empirically 

 descriptive of observed neural behaviour or are mathematically rigorous but little 

 related to real nervous systems, this model is based on neurophysiologically valid 

 assumptions and leads to mathematically rigorous consequences, some not easily 

 anticipated, which are congruent with many tacts of the complex behaviour of 

 mammals and man. 



Doty. All these schemes ignore the most ditiicult and probably the most impor- 

 tant element of the problem : the temporal order. Instead, their main reliance is 

 placed upon the convergence of excitations from multiple sources upon a common 

 network. This is a safe and reasonable assumption since such convergence is well 

 known for many regions of the nervous system (e.g. mesencephalic reticular 

 formation, Amassian and DeVito, 1954; amygciala, Machne and Segundo, 1956; 

 basal ganglia, Segundo and Machne, 1956; somatosensory cortex, Ricci, Doane 

 and Jasper, 1957). Dr Buser has just given us another excellent example for the 

 cortical level. Victor Wilson's demonstration (1955) that post-tetanic potentiation 

 in one multisynaptic spinal cord system is available through non-tetanized afferents 

 assures us these convergent networks exist at a functional level. Yet real as they are, 

 these effects give no adequate basis for explaining the formation of conditioned 

 reflexes. The temporal order of these effects is exactly the opposite of those in 

 conditioning. In all multi-syiiaptic convergence systems it is the antecedent excita- 

 tion which alters the response to a subsequent event. In forming conditioned 

 reflexes on the other hand, the antecedent excitation (the CS) has essentially no 

 effect upon the response to the subsequent stimulation (the US); rather it is the 

 effect produced by the antecedent excitation which becomes altered. One might 

 call this the 'temporal paradox' of conditioning. 



It is true that Dr Asratyan has just shown us that a linkage exists in a conditioned 

 reflex situation whereby the subsequent stimulation can elicit the effects of the 

 antecedent, a 'backward' linkage in terms usually employed for conditioned 

 reflexes. Perhaps the foregoing 'convergence theories' explain this phenomenon. 

 These effects, however, are feeble and normally overwhelmed completely by the 

 appearance of the conditioned reflex. Dr Asratyan showed equally well the prime 

 importance of the temporal factor in forming conditioned reflexes, indeed, a 

 previously formed CR was lost when the CS and US were given simtiltaneously. 

 Dr Giurgea and I have also added here to an extensive literature on the temporal 

 aspects of conditioning. Randomized or reversed temporal relations in the pairing 

 of cortical stimulations did not produce conditioned reflexes, whereas the same 

 stimuli present in the usual order did so. Appropriate temporal ordering is absolu- 

 tely essential to conditioning. Thus one can speculate about converging and 'grow- 

 ing' pathways indefinitely, and with a certain shadow of truth; but until the 

 reason for this 'paradoxical' temporal relation is clear, we will not understand the 

 processes of conditioning or learning. 



UU 



