TRIAL AND ERROR REACTIONS IN MAMMALS 37 



a subject's first experience with the apparatus was merely that 

 of entrance into it, with all the doors propped open and with 

 food scattered about on the floor. After he had eaten he was 

 coaxed out of one of the entrance doors. This was repeated 

 until the subject was thoroughly familiar with the interior of 

 the apparatus as a feeding ground, and with the open exit doors 

 as the most convenient avenues of approach to a desirable 

 locality. In the case of the horse this differed in that the feeding 

 ground was, from the start, a place just beyond any of the 

 exit doors. 



The next step in training the subject for the experiment was 

 to close the entrance door as soon as he had entered the appa- 

 ratus. As soon as this had ceased to produce any evidence of 

 uneasiness the four exit doors were left partially closed, so 

 that the subject was compelled to exert some pressure against 

 any of the exit doors through w^hich he sought to escape. As 

 the training advanced the exit doors were left more and more 

 nearly closed until they were quite closed, although still un- 

 locked. The subject was considered " trained " as soon as he 

 had learned to seek the exit doors for escape from the apparatus, 

 and to push against them without hesitation. 



(2) Formal trials. The description of these trials, which 

 constitute the experiment proper, will be facilitated by speak- 

 ing of the various exit doors as if they were numbered " i," 

 "2," " 3 " and " 4," in order, from left to right. For the first 

 formal trial of any subject three of the exit doors were locked 

 by the experimenter, who pulled the appropriate button strings 

 without the subject's knowledge. The remaining exit door, 

 and only that one door was left unlocked. From within the 

 apparatus, therefore, all four exit doors looked alike, i. e., 

 closed. With the apparatus thus prepared, the subject was 

 placed within it for his first trial, and was allowed to choose 

 his own time for effecting his escape. As soon as he found the 

 one unlocked exit door and effected his escape he was rewarded 

 with food. 



For the second formal trial the one unlocked door of the first 

 trial was locked, and another of the exit doors was unlocked. 

 Thus, if door 4 were the unlocked door for the first trial, it would 

 be one of the three locked doors for the second trial. Each suc- 

 ceeding trial in a total series of 100 trials differed from the trial 



