338 'Jourual of Comparative Neurology and Psychology. 



Passing now to a consideration of the character of the adequate 

 reactions, it should be said that so far as observation of the indi- 

 vidual trials "s^^as concerned, the majority of correct first choices 

 might have been due to mere chance. But I was able to record 

 23 incidents like the following. The animal inspected the sign 

 board, pedal cards, and pedals, turned his head to look at me, 

 wagging his tail (my position was at a point 20 feet directly in front 

 of the door), and then, confining his attention exclusively to the 

 sign board, devoted some time to sniffing its surface. Following 

 this he went up and down the row of pedals, sniffing their cards; 

 then pausing at an unattached pedal, he raised his paw as if to 

 strike, but desisted, withdrawing the uplifted paw slowly, with- 

 out having struck the pedal. Again he returned to the sign board 

 and sniffed its surface, following which he passed down the line 

 of pedals, sniffing their cards as he went, until he came to the at- 

 tached pedal (/. e., the one affording him the same stimuli as the 

 sign board); this he struck and was released. There was not a 

 single failure when he behaved in this manner. 



An examination of the correct first choice curve will show that 

 there was but little, if any, improvement at the end of the 440 

 experiments with sign boards. This agrees with the fact that 

 the kind of behavior just described never became a fixed mode of 

 reaction. Usually, after he had once reacted so adequately, he 

 quickly chose the previously successful color or odor, thus falling 

 into error. 



It is interesting that the odor reinforcement was of so little value 

 to him. That particular series of experiments would have been 

 continued much longer had it not been for the fact that the animal 

 finally learned a very effective and easy manner of finding the 

 attached pedal. Starting at the third or fourth pedal to the right, 

 he would pass to the left, striking the pedals in succession until 

 the shock of the discharged trigger sent him scurrying to the door. 

 If he started at p 3, and p 4 was attached, he would go down the 

 line to p I in the usual manner, then walk directly to p 4 and strike 

 it. Reference to the curves shows that in the third from the last 

 set of 20 experiments he made 29 errors, and struck the attached 

 pedal first in 5 of the 20 trials. That he made only 29 instead of 

 the "average chance" 30 errors, is due to the fact that in switch- 

 ing from the method of starting at p 4 (a method which he pur- 

 sued for 13 successive trials), to the method of starting at p 3, he 



