STIMULI DURING DELAYED REACTION IN DOGS 273 



"CD" on two compartments. Resumption of the "A" and 

 "B" types with the three compartments had been so successful 

 that a ten second delay was used at once on beginning the "C" 

 trials. Here the food was thrown to the dog after completion 

 of the correct reaction. The records obtained here show per- 

 haps better than any others the learning curve, beginning very 

 low and rising gradually to a practical mastery of the problem. 

 The first twenty-five trials were very poor, only seven being 

 correct. Of the errors, eight were because the dog refused to 

 leave the box at the release. The dog gave no attention at 

 all to the stimulus and when he left the box on release it was 

 more a result of training than a reaction to a formed associ- 

 ation. The inattention was due to the lack of hunger. 



On January 24th, 1914, the dog was more alert, and tried to 

 solve the problem but was not very successful. Out of sixty 

 trials only twenty-seven or 45% of them were correct. This 

 low result was due to the dog's inability to retain the associa- 

 tions for the discrimination of the three compartments over a 

 delay period of ten seconds, and was not due to inattention 

 to the stimulus. 



On January 27th, 1914, twenty-five trials were given and 

 the benefit of the previous training began to be evident, for 

 seventeen or 68% of the trials were successful. The last nine 

 trials were all successful. 



The record of the twenty-one trials given on January 28th, 

 1914, were not so successful as those of the preceding day, 

 having only eleven or 52%, of the trials as correct reactions. 

 The falling off was due to the indifference of the dog to the 

 stimulus in the last ten trials. The average for the first eleven 

 trials was 73% and for the last ten only 30%. 



February 16th, 1914 was the next day on which the "ten 

 second delay" trials were given, and of eighteen trials sixteen 

 or 88% of them were correct. The last twelve trials showed 

 perfect reactions. The records made on this date were the 

 best obtained on the ten second delay problem, for fifteen trials 

 on February 10th, 1914 showed eleven, or S3% of them correct. 



The learning curve obtained in these trials with the ten second 

 delay shows the growth of the power of carrying the associ- 

 ations over the delay period the most perfectly of any of the 

 delay series. All show the same general results, but none show 



