158 HARVEY CARR 



then invariably entered on all succeeding runs. The eighth 

 cul de sac was avoided six times, entered once, avoided once, 

 and then became fixed; in another case it was avoided for 28 

 trials, entered three times, avoided for five trials, and then 

 became fixed. The continual avoidance of an error for a number 

 of trials followed by invariable entrance can not be due to chance. 

 Chance may account for the first entrance but the sudden fixation 

 of the habit remains inexplicable. This type of behavior may 

 be readily explained by our hypothesis. This alley is so related 

 to the previous sections of the maze that the turning habits 

 necessitated by the latter dispose the animal to avoid it. But 

 these determining habits become altered with successive runs. 

 The avoiding disposition becomes weakened and finally sup- 

 planted by the opposite attitude. The alley is thus regularly 

 avoided for a while; there is indecision and alternation for a 

 short period; and this behavior is followed by sudden fixation 

 and invariable entrance. The fixation of this error is not due 

 to repeated entrances. In a sense this error was engrained in 

 the animal's organization before it was entered, because the 

 disposition is the outcome of the entire maze 4 habit developed 

 up to that time. 



CONCLUSIONS 



The temporal order in which the various cul de sacs are elim- 

 inated is roughly correlated sometimes with their spatial order 

 in reference to the food box and sometimes with their order of 

 proximity to the point of entrance. 



The temporal order of mastery of the cul de sacs is invariably 

 correlated with their order representing the increasing number 

 of errors made in each. The ease or rapidity of mastery of 

 any cul de sac is thus inversely related to its degree of attrac- 

 tiveness as measured in terms of number of entrances. The 

 problem as to the order of elimination of the cul de sacs must 

 be explained in large part in terms of the distribution of errors. 



The factors influencing the distribution of entrances among 

 the cul de sacs are the tendency to return, the character of the 

 motives actuating the animal, and peculiarities of the cul de sacs. 



* This hypothesis is closely related to Peterson's conception developed in his recent 

 article, Completeness of Response as an Explanation Principle in Learning, Psych. 

 Rev., 23, 153-162. 



