DELAYED REACTION 29 



a,*^ with a medium gray. If the animal's Hmit of delay was 

 no better or was worse with this arrangement than before, the 

 animal was dropped from the experiment. If the limit were 

 better, the different backgrounds were removed and the sim- 

 ilar ones used again. The limit of delay with these was then 

 re-determined. If this third limit were greater than the second, 

 the effect of training could be evoked as an explanation of the 

 fact. But if it were markedly less than the second, the only 

 cause could be an association between the backgrounds and the 

 lights. The significance of this type of association will be dwelt 

 upon in detail later in this paper. 



One more point in general method remains to be considered. 

 This is the question of what percentage of correct choices shall 

 be taken as sufficient to justify further increase in the interval 

 of delay. With three discrimination boxes, pure chance would 

 lead the animal to make 33i%of correct reactions out of a long 

 series of presentations. But series used in experimentation are 

 very rarely long enough for chance to operate as theory de- 

 mands. Besides there are various other influences which enter 

 in to determine an animal's behavior above and beyond the 

 influence of the stimulus proper to the test. One such influence 

 is the position factor. Try as I might to eliminate this, most 

 of the animals acquired at various times during the experi- 

 ments more or less pronounced preferences for certain boxes. 

 And these preferences varied from animal to animal. Some- 

 times they were so strong that the regular tests had to be 

 stopped until the habit could be broken up. In the light of 

 this, although on the whole each box was presented to an 

 animal an equal number of times, in any one stage of the tests 

 such an equality might not be present. Thus an equality of 

 percentages among the various animals, in this case, would not 

 mean that they knew the problem equally well. Again hesi- 

 tancies and waverings must be noted in estimating how well 

 the animal is grounded in his appointed task. Further state- 

 ments concerning the value to be attached to the percentages 

 appear later in the paper (pp. 44, 46 and at intervals thereafter.) 

 Considerations such as these render it highly inadvisable to lay 

 down a rigid standard as to the number of trials and the per- 



^' a, b, and c will be used in this paper to designate the right, middle and left 

 light boxes respectively. 



