38 G. V. HAMILTON 



immediately preceding it according to the same principle: the 

 unlocked exit door of the immediately preceding trial was always 

 one of the three locked exit doors during the present trial. 



A second condition of the formal trials was this: during each 

 subject's loo trials, each exit door was left unlocked for twenty- 

 five trials and locked for seventy-five trials, and care was taken 

 by the experimenter to avoid a discoverable sequence in select- 

 ing an ever- varying exit door to be left unlocked. Even the 

 most sophisticated human subject was thus left in ignorance as 

 to which of three inferentially possible doors of exit would open 

 when pressed against until he had actually tried one or more 

 of these doors. Ten consecutive trials were given each subject 

 daily for ten successive days. 



The older human subjects were given no preliminary training, 

 but were frankly told that no tricks would be played, and that 

 they were expected to leave the apparatus by way of an exit 

 door. The human infant was given essentially the same train- 

 ing that the animals received, except that in his case toys were 

 substituted for food when my commendation proved insufficient 

 as a motive for reaction. 



Before we enter into a discussion of results, two obvious 

 defects of the above method must be taken into account: (i) 

 The situations may have had different sensory values for the 

 different subjects. For example, the dogs of my experiment 

 were unmistakably guided by local odor signs in their discrimina- 

 tion of one exit door from another, whilst these signs were of no 

 value to the human subjects, and probably of but little value 

 to the cats and monkeys. On the other hand, the monkeys 

 may have detected fine differences of visual appearance of the 

 inner surfaces of the exit doors, whereas these doors certainly 

 presented no such differences for the human subjects, and may 

 have been visually indistinguishable to the cats, dogs, and horse. 



A fairly satisfactory solution of this difficulty was effected 

 by the preliminary training, which taught the subjects to dis- 

 criminate among the various exit doors by their differences of 

 spatial position. 



(2) There was no adequate measurement of the reactive value 

 of the motives supplied to the various subjects for escaping from 

 the apparatus with the greatest possible speed and accuracy. 

 Yerkes' and Dodson's (i) discoveries concerning the relation 



