38 ROBERT M. YERKES 



in the number of correct choices promptly resulted, and con- 

 tinued until on July 14 the ratio of choices was 1 to .54. It 

 appeared from these data that a relatively small number of 

 choices, say not more than ten a day, the rewards in connection 

 with which supplied the only food received by the animal, yielded 

 most favorable results. 



On July 16, the period of confinement in wrong boxes was 

 increased to sixty seconds, and it was so continued for a number 

 of days. But in the end, it became clear that the period of 

 thirty seconds, combined with a liberal reward in the shape of 

 desired food and a single series of ten trials per day, was most 

 satisfactory. The detailed data of table 2 indicate that at 

 this time Skirrl was making his choices by memory of the par- 

 ticular setting. 



Skirrl, on July 17 was evidently hungry and eager to locate 

 food, but seemingly unable to select the right box. In trial 5 

 (765th) of the series, he was punished by confinement in 

 box 8. When the doors were unlocked in order that the en- 

 trance door might be raised to release him, the lock-bar, sliding 

 under the floor, made a slight grating noise, and the instant the 

 entrance door was opened, he jumped out excitedly. He made 

 no outcry, hut as soon as he was out of the box, sat down, and taking 

 up his right hind foot, examined it for a few seconds. Having 

 apparently assured himself that nothing serious had happened, 

 he went on unconcernedly about his task. The presumption 

 is that the sound of the lock-bar, associated as it was with his 

 painful experience in box 1, revived the strongly affective ex- 

 perience of stepping on the nail. Psychologically described, the 

 sound induced an imaginal complex equivalent to the earlier 

 painful experience. The behavior seems to the writer a most 

 important bit of evidence of imagery in the monkey. 



Finally, on August 9, after ten hundred and seventy trials, 

 Skirrl succeeded in choosing correctly in the ten trials of a series, 

 and he was therefore considered to have solved the problem of 

 the second door from the right end of the group. 



On the following day, he was given a control series with the 

 settings which are presented on page 19 and also at the bottom 

 of table 2. In this series he chose correctly five times, — in other 

 words, as often correctly as incorrectly. An analysis of the 

 choices indicates, however, that two of the five correct choices 



