364 K. S. LASHLEY 



gradual fixation occurs, the problem of the neuro-chemistry of 

 learning is simplified by admission of the hypothesis that the 

 effects of the passage of the neural impulse upon later conduc- 

 tivity are direct and immediate. This hypothesis is more in 

 accord with such facts as are known concerning the alteration 

 of conductivity in regions of decrement (Lucas, '17), where the 

 learning process may, perhaps, be located ultimately, and with 

 the generally established facts of the deterioration of function 

 through disuse. 



The experimental evidence upon which the belief in a gradual 

 fixation of associations is based is far from convincing. It con- 

 sists primarily of the facts expressed in Jost's law, of occasional 

 records of improvement in complex functions during periods of 

 non-practice, and of the data upon the effects of distribution of 

 practice. All of this can be explained equally well by other 

 hypotheses and, in view of the extreme importance of the point 

 for physiological explanation, we should be careful not to accept 

 the assumption of a 'gradual setting' of new functional connec- 

 tions until some real evidence is advanced in support of it. 



In studies of the mechanism of learning the processes of adjust- 

 ment and of fixation must be distinguished as absolutely inde- 

 pendent variables. The former is, in lower animals and prob- 

 ably in primates also, solely the production of varied random 

 activity, through which the first adjustment to a new situation 

 is brought about; the latter is a process by which the recurrence 

 of certain of the random activities in future trials is rendered 

 more probable. Slow improvement in any function may result 

 either from difficulty in discovering efficient methods of per- 

 formance or from failure to fix as habits the methods which have 

 been hit upon by chance-. In an earlier paper (Lashley, '17a) 

 I have shown that a part of the superiority of short over long 

 practice periods is due to the fact that distributed practice per- 

 mits of greater variability of response and hence greater probabil- 

 ity of discovering new and successful modes of attacking the 

 problem, than does concentrated practice. The influence of the 

 distribution of practice is here exerted upon the process of adjust- 

 ment to the new situation and not upon that of fixation of the 

 random acts. 



