62 ROBERT M. YERKES 



laboratory to take charge, I found that Sobke was no longer 

 trying to solve the problem as when I had gone away. His 

 attitude had changed in that he had become indifferent, care- 

 less, and obviously discouraged with his task. 



I immediately set about reinstating the former attitude by 

 lessening the number of trials and the punishment, and by in- 

 creasing the value of the reward, but my best efforts, continuing 

 up to August 28, failed markedly to improve the condition. The 

 number of correct choices did somewhat increase, but at no 

 time did the animal attain the degree of success which he had 

 achieved as early as July 31 in the eleventh series of trials. 



During the last two weeks of experimentation, all possible 

 efforts were put forth to discover the best combination of re- 

 wards and punishments. Punishment was varied from to 

 confinement of sixty seconds, and man}^ kinds of food in different 

 amounts were tried as rewards, but in spite of everything 

 Sobke failed to improve markedly. From time to time, notably 

 on August 12 and 21, he exhibited peculiarly strong resentment 

 toward me and repeatedly attempted to attack me. 



The outcome of my experiments with problem 4 is peculiarly 

 interesting in that it indicates the importance of a favorable 

 attitude toward the work and the extreme risk from disturbing 

 or discouraging conditions. It seems not improbable that had 

 the work progressed without change in experimenter, or method 

 of procedure, and above all without the disturbance of the paint- 

 ing, Sobke might have solved problem 4 within a few days. 

 This is by no means certain, however, for in problems 2 and 3 

 the ratio of right to wrong choices instead of increasing steadily 

 increased very irregularly. 



The detailed results for this problem are given in table 7. 

 Reactive tendencies which appear are : (a) persistent choice of the 

 end boxes followed, subsequently, by (b) the tendency to locate the 

 middle box directly. This proved fairly easy when the number 

 of boxes involved was only three as in settings 1, 4, 7, and 10. 

 Setting 4 was most difficult of all, because box 9 was avoided 

 or ignored. When the number of open boxes was as great as 

 five, as in settings 2 and 8, the task was obviously- more difficult, 

 but whereas success in setting 2 appeared early, in setting 8 it 

 failed to appear during the course of experimentation. For 

 the settings 3, 6, and 9, involving either seven or nine open boxes, 



