TRANSFER FACTORS IN MAZE LEARNING 339 



The transference of the adjustment to maze-situation-in-general 

 affects principally the initial trials in the new maze. Here we 

 seem to have a case of preliminary adjustment to a problem- 

 situation of a general type analogous to the preliminary adjust- 

 ment found to play a large part in human learning. Apparently, 

 a rat in learning a maze under the conditions of the usual type of 

 experiment has for a first problem the getting used to the presence 

 of the culs-de-sac, and the very steep initial descent of his learning 

 curve is due in part to this fact. 12 



An incidental part of the experimentation above reported gave 

 evidence of another factor of transfer. As described, the sub- 

 jects of group IV were in their training series run with each single 

 daily trial in a new maze. The series of mazes used is shown in 

 figure 4. It was stated above that (a) the outside dimensions of 

 all mazes remained constant, (b) the entrance door was always on 

 the same side, and (c) the exit door and food box was always on 

 the same side, and opposite the entrance. As to results, it was 

 indicated above that the animals of this group seemed gradually 

 to learn not to repeat an error during the one run in each maze 

 (table 2). Now, furthermore, after the first few trials the ex- 

 perimenter observed, or thought he observed, that the animals 

 were developing a sense of direction for the food box; i.e., that 

 whenever confronted with a choice of paths they tended to turn 

 toward it rather than away from it. By this is meant a turning 

 toward the food box side of the maze, which was constant, and 

 not necessarily toward the door of the food box as it occurred 

 in that particular maze. (That they could not have learned the 

 location of the food box in terms of true path or paths to be taken 

 is obvious from the conditions of the problem ever-changing 

 culs-de-sac.) To get an exact proof or disproof of this rough 

 observation the writer constructed a table as follows (table 3). 

 Each maze pattern was inspected and the blind alleys classified 

 under three heads: (a) those which when entered involved a 

 turn by the subject toward the food box side of the maze; (b) those 

 which involved a turn away from that side and toward the 



12 Wiltbank (op. cit., pp. 36-37) speaks similarly of "error elimination activity" 

 as transferable. 



