362 JOSEPH PETERSON 



is to avoid the fallacy of assuming — and these authors, I believe, 

 do not assume this — that each response can be considered on 

 its own account, rather than in relation to other reactions con- 

 cerning vitally the welfare of the entire organism. It would 

 seem that with a different arrangement of the relations of the 

 two types of blind alleys in the circular maze, so that the objec- 

 tion here urged would be met, results in this maze would in gen- 

 eral agree with those obtained in the use of other mazes, 20 show- 

 ing that on the whole there is a progressive backward elimination 

 of errors in the maze. There are, of course, many circumstances, 

 making entrances to certain cut de sacs more probable than to 

 others, that tend against this general rule. No maze in exist- 

 ence has cut de sacs all presenting equal difficulty to the animal. 



The writer believes that in spite of the shortcomings of the 

 frequency factors as an explanation principle of maze learning, 

 the general considerations which he has discussed in the first 

 (the theoretical) part of this paper satisfactorily account for the 

 progressive backward elimination of errors in the maze, to the ex- 

 tent that it actually occurs, and also for the fact that the number 

 of entrances to blind alleys increases roughly with their distance 

 from the food box. There seems to be no ' retroactive associ- 

 ation " necessary, as Hubbert and Lashley rightly conclude. 



If frequency and recency factors play the unimportant part 

 in actual learning that our present data seem to indicate, .to 

 what neural and physical conditions, then, must we look for 

 the main factors that bring about the elimination of random 

 acts and the changes in behavior characteristic of learning? The 

 writer has attempted elsewhere to indicate in a general and 

 tentative way the answer to this question. In support of his 

 contention that our neural explanations have usually been so 

 simple as to throw us into a mechanical associationism which 

 finds difficulty in explaining the changes in behavior character- 

 istic of learning; that neural processes are inconceivably com- 

 plex so that the general consistency of the circumstances, organic 

 and extra-organic, forces short-circuiting of impulses in the cen- 

 tral nervous system,- — in support of this position the writer is 

 pleased to quote a few lines from Professor C. J. Herrick 21 which 

 have come to his attention since the former articles were written : 



20 Cf. Carr, H. A. Distribution and Elimination of Errors. (An abstract.) 

 Psychol. Bull, 1917, 14, p. 58. . 



21 Introduction to Neurology, 1916, p. 306. See also page 296 and Ch. XXI. 



