DISTRIBUTION AND ELIMINATION OF ERRORS IN MAZE 157 



increases for the first fifteen runs, and then slowly decreases 

 for the remaining thirty-five trials. , More errors were made 

 at the end of the experiment than at the beginning. In con- 

 sidering the individual records, the general features of the group 

 curve are also characteristic of that of fifteen of the animals. 

 In maze III, the difficult group of cul de sacs is composed of 

 alleys 5, 7, and 8. In maze IV, alley 3 was relatively easy 

 while 5, 6, and 9 were the difficult ones. 



This difficulty of any cul de sac must be explained mainly 

 in terms of the animal's organization in reference to it, for all 

 of the blind alleys of mazes III and IV were highly uniform in 

 character. Each consisted of a single straight runway sixteen 

 or twenty inches long. This reduction of the difficulty of a 

 cul de sac to the animal's disposition toward it allows of a 

 common explanation for both the initial and final distribution 

 of errors. The disposition of an animal to enter or avoid an 

 alley is a result in part of the habits of turning already devel- 

 oped. As the maze is mastered, these habits become profoundly 

 modified. Returns and repeated explorations are inhibited and 

 several cul de sacs may be eliminated. Since the attractiveness 

 of a cul de sac is a function of the maze habit and this habit 

 is altered in the course of mastery, it is evident that the initial 

 and the final attractiveness of an alley must be to some extent 

 independent variables. With this conception the disposition to 

 enter an alley may actually increase as well as decrease, and 

 individual exceptions to the group attitude are possible. 



Such an explanation is feasible and the conception possesses 

 some degree of a priori rationality, but any convincing proof of 

 the hypothesis is more difficult. Chance can not account for 

 the error curve of alley 5 in fig. 1, for the records of fifteen of 

 the sixteen animals conform to its general features. There has 

 been no objective change in the alley itself. Evidently it is 

 the attitude of the animals towards this situation that has been 

 altered. Nor will chance account for the distribution of errors 

 of any individual animal. One rat did not enter alley 5 during 

 the first five runs but made one entrance per trial for the re- 

 maining 45 runs. Another animal avoided this alley for eleven 

 runs, alternated between success and error for the next six 

 trials, and the error then became fixed. In another case this 

 cul de sac was avoided twice, entered once, avoided twice, and 



