THE NEURAL BASIS OF LEARNING 



'49 1 



'borrows' the thalamocortical circuit of the US in 

 the thalamus is not elucidated. The theory, despite 

 its emphasis on thalamic changes in learning, is said 

 by its author not to minimize the part played by the 

 cortex. We may consider it to be a somewhat elabo- 

 rate early attempt to account for the rather meager 

 information that the EEG studies have thus far pro- 

 vided. 



Other Neural Possibilities 



There is a group of theories that explain learning 

 in terms of brain events for which the all-or-none 

 law of nerve action is not particularly relevant. For 

 example, it was once proposed that an increase in the 

 "conductance' of nerve fibers, that is the amount of 

 message each fiber transmitted, might account for 

 learning. This notion was dismissed when nerves 

 were shown to convey all-or-none signals. Recent 

 neurophysiological research, however, establishes the 

 importance of graded events where neurons arc in 

 synaptic contact with each other, and so the notion 

 may again deserve further consideration. 



In this connection, Bishop (14) points out that the 

 all-or-none activities in the brain serve merely to con- 

 vey messages from one location to another; they are 

 set up by graded events in the first place and in turn 

 they produce their actions through graded-response 

 processes at the synapses upon which the) converge. 

 Grundfest (84) also emphasizes tlii^ idea. Graded 

 processes can be maintained at steadv siaics for long 

 times as compared to nerve impulses which vary 

 abruptly from all to nothing. What is needed to ex- 

 plain learning;, or at least retention, is long-continued 

 neural events. Since the graded responses in dendrites 

 have appropriate characteristics, it would therefore 

 be in, on or around them that one might look with 

 particular care for neural events peculiar to learning. 



Herrick (92, 93) sees in the neuropile an answer to 

 w r here the integrative processes of learning take place. 

 This neuropile is "a fabric of relatively unspecialized 

 nerve cells and very thin fibers'" in which, as an 

 anatomist sees it, the complex events of learning 

 might well occur. Unfortunately he cannot specify 

 what these might be, and so this idea, like so many 

 others, cannot be subjected to experimental test. 



By contrast, the cortical 'electrical field' theories 

 that have been advanced from time to time to ex- 

 plain retention in learning (as well as perception in 

 general) can be tested. Electrical conductors have 

 been implanted in and upon the cortex of some ani- 

 mals while insulators have been implanted in others; 



such devices must have distorted or destroyed very 

 effectively any existing cortical electrical field, yet 

 the complex learned behavior suffered minimally, if 

 at all, as a consequence (226). 



Mathematical Models 



There is one final class of theories to be considered. 

 Certain physicists and mathematicians oxer the vears 

 have been challenged by the complexities of the 

 learning process to develop explanatory formulations 

 for it. Besjinnin" perhaps with Rashevskv in 1938 

 (201), new contributions at a rate of at least one every 

 year have been made in this area. Recently, with the 

 advent of digital and analogue computers and theory, 

 the rate has been stepped up with the idea, perhaps, 

 that our advancing knowledge of complex switching 

 circuits in machines may have application to the 

 brain. The reader interested in these models of the 

 brain will want to consult the available contributions 

 (11, 56, 97, 100, 240, 24] I. 



Summary 



A large number of speculations have been ad- 

 vanced to explain the neural correlates lor learning. 

 Some of these are based upon a certain amount of 

 objective data about the brain. The most popular 

 schemes incorporate hypothetical changes at svn- 

 apses with hypothetical reverberating activity in 

 neuron chains. In the large collection of speculations 

 on record, the one (or ours! that will finally har- 

 monize with the facts may well be present but, if so, 

 there is no compelling experimental evidence for it 

 (or them) at this time. 



DISCUSSION AND SUMMARY 



This is the appropriate place in our exposition for 

 the authors to propose a comprehensive theory which, 

 without violating an) of the data, will explain what 

 happens in the brain during the process of learning. 

 In our opinion, however, this cannot be clone at the 

 present time. New^ information is currently being de- 

 veloped rapidly and, as this happens, the large gaps 

 in knowledge that still exist stand out more and more 

 clearly. Until some of these gaps are filled only the 

 most general of formulations seem warranted. In 

 this discussion, therefore, we shall merely point first 

 to some of the basic questions implied by the material 

 in the preceding sections and then consider one of 



