MENTAL LIFE OF MONKEYS AND APES 47 



food once from each of the nine boxes, both doors of a given box 

 being open for the trial and all other doors closed. For this 

 feeding experiment, the doors were opened in irregular order, 

 and this order was changed from day to day. 



Systematic work with problem 2 began on May 3, with pun- 

 ishment of thirty seconds for mistakes and a liberal reward of 

 food for each success. Early in the series of trials it was dis- 

 covered that Sobke was likely to become discouraged and waste 

 a great deal of time unless certain aid were given by the experi- 

 menter. On this account, after the first two trials, the method 

 was adopted of punishing the animal by confinement for the 

 first ten mistakes in a trial., and of then, if need be, indicating 

 the right box by slightly and momentarily raising the exit door. 

 Every trial in w^hich aid was thus given by the experimenter is 

 indicated in table 5 by an asterisk following the last choice. In 

 the first series of trials for this problem, aid had to be given in 

 seven of the ten trials, and even so the series occupied seventy- 

 one minutes. It is possible that had no aid been given, the 

 work might have been continued successfully with a smaller 

 number of trials than ten per day. But under the circumstances 

 it seemed wiser to avoid the risk of discouraging and thus spoil- 

 ing the animal for use in the experiment. It should be stated, 

 also, that it proved impossible to adhere to the period of thirty 

 seconds as punishment in this series. For the majority of the 

 wrong choices confinement of not more than ten seconds was used. 



For the second series, given on May 4, the conditions were 

 unfavorable in that it was dark and rainy, and the noise of the 

 rain on the roof frightened Sobke. He refused to work after 

 the fourth trial, and the series had to be completed on the fol- 

 lowing day. The total time required for this series was seventy- 

 eight minutes. 



The work on May 6 was distinctly better, and the animal's 

 behavior indicated, in a number of trials, definite recognition 

 of the right door. He might, for example, make a number of 

 incorrect choices, then pause for a few seconds to look steadily 

 at the doors, and having apparently found some cue, run directly 

 to the right box. No aid from the experimenter was needed in 

 this series. 



On the following day improvement continued and the ani- 

 mal's method of choosing became definite and fairly precise. He 



