364 JOSEPH PETERSON 



random acts will survive, i.e., that it brings about the learning, 

 we have found by flipping coins that probability does afford an 

 explanation of how the animal finally reaches the food box in 

 the maze, but that it fails to explain alone or in connection with 

 recency how the useless acts are eliminated. Recency and 

 frequency factors do not seem to explain how the short-cuts in 

 behavior characteristic of learning come about. Probability and 

 the effects of recency and frequency factors supplemented by 

 certain visceral directive factors, do, however, seem to account 

 in a satisfactory manner both for the elimination of cul de sacs 

 in a progressive backward order, roughly speaking, and also 

 for the greater number of entrances to cul de sacs in the first 

 part of the trail and for the general correlation between the 

 number of such entrances and the distances of the respective 

 cul de sacs from the food box. 



Tabulations of the reactions of seventeen rats in their first 

 three trials in three different mazes — six in one, four in another, 

 and seven in the third — show that, contrary to certain current 

 views, over 50% of the rat's early critical choices at bifurca- 

 tions in the maze are the opposite of what would be expected 

 on the basis of recency and frequency factors. Responses favor- 

 ing expectations on recency and frequency increase and finally 

 reach 100% when the learning is complete. This, however, is 

 not evidence that these factors bring about the learning. The 

 converse is true: the modification called learning increases fre- 

 quency and recency responses. It is suggested that this may 

 also be true of other types of learning. 



There seems to be clear evidence of the operation in learning 

 of visceral factors controlling, for the general demands of the 

 organism, the associations which are formed. Choices at bifur- 

 cations in the maze are not predictable on the basis of frequency 

 and recency alone as applied to individual responses; each re- 

 sponse must be considered in the light of the whole situation 

 to which the animal as a unitary organism is reacting. The 

 elimination of random acts, of entrances to cul de sacs, seem to 

 be comprehensible only on this basis. This seems to indicate 

 that the laws of association are not the dominantly controlling 

 factors that they have credit for being in current psychology. 



An analytic method of studying learning in the maze is devel- 

 oped, one which may be applied to other simple types of learn- 

 ing when the necessary controls are available. 



