92 JOSEPH U. YARBROUGH 



from one period of delay to another. The general method used, 

 however, was to promote the animal as fast as possible, and only 

 demote when the records showed him unable to bridge the delay. 

 There were no special arrangements made for punishment in case 

 of error. It was easy to observe, however, that there was a 

 certain degree of punishment following each error. These pun- 

 ishments were: having to back out of a box, and having food 

 and freedom deferred for a longer period of time. Although the 

 cats were expected to go straight to the stimulus box, no wide 

 turn in their pathway is recorded wrong unless they approached 

 the entrance to one of the other boxes. The apparatus was 

 so constructed that the animal could not see the position of 

 an exit door, i.e., whether it was open or closed, without actually 

 approaching the particular box. 



IV 



EXPERIMENTAL RESULTS 

 1. Three Compartment Experiments 



A. Learning the association. — Although the primary purpose 

 of this investigation is a study of the delayed reaction proper, 

 it is well to make additional note of the learning process. Table 

 I gives the number of trials required by the cats of set A to learn 

 the association between the light and the getting of food. Each 

 cat was given 10 trials daily. Fay, the last reported in the table, 

 died at the end of 50 trials. Her results are reported, however, 

 because 75% of her last 20 trials were successful. 



TABLE I 



Cats Tested on Light 



Number Per cent 



Number Number Per cent correct of correct of 



Cat of trials correct correct last 50 last 50 



Bobby 130 96 73 47 94 



Jim 110 84 76 49 99 



Tom 170 112 65 45 90 



Fay 50 25 50 25 50 



The number of trials required by the cats on sound, set B, 

 are given in table II. The last cat reported in this table died 

 at the end of 40 trials. For this reason no record of her work 

 appears in the last two columns of the table. 



