Hamilton, Unusual Reaction of a Dog. 337 



greater adequacy of reaction. But the relatively slight influence 

 of the expermienter's personal equation on tabulations ot results, 

 and its inevitably distorting influence on any report of the animal's 

 apparent intention, must, so far as one kind of certainty is con- 

 cerned, give precedence to mathematically derived inferences. 



First of all, then, w^e must take into account the fact that had 

 the animal been unmfluenced by any associations, mere chance 

 would have enabled him to make correct first choices in approx- 

 imately 25 per cent of the total number of trials. Or, had he 

 always struck the same pedal first, or the pedal bearing the same 

 color or odor, he would have made 25 per cent correct first choices. 

 Now, the 53 incorrect first choices unaccounted for by any of the 

 above-mentioned factors, and the 62 incorrect first choices ascribed 

 to factor "i" (the latter, smce they often came in series of more 

 than four each) can most conservatively be considered to be due 

 to nothing that impaired the animal's chances to make the average 

 chance 25 per cent correct first choices. 



The case is quite different with the loi incorrect first choices 

 ascribed to factor "2," where the animal chose the pedal that had 

 just released him; of necessity any choice due to the influence of 

 such an association would be an incorrect one, thus materially low- 

 ering his chance of attaining the average 25 per cent correct first 

 choices. But mere chance would have enabled him to strike first 

 the pedal that w^as last attached in no of the trials. This would 

 be a fatal objection to any attempt to demonstrate that the animal 

 was misled by such associations were it not for the fact that the 

 choose-first-the-last-successful-pedal reactions came, not at irregu- 

 lar intervals, but in definite and prolonged series. Thus, if p 3 

 had just released hmi, he would strike it on reentering the cage, 

 and p I proving to be the successful pedal, the next time he would 

 make an unsuccessful attempt to escape by striking p i, and so 

 on, until some new determinative of first choice became effective. 

 The same is true of the 102 incorrect first choices ascribed to factor 



"3-" 



It now becomes clear, I think, that to have made 122 correct 

 first choices the animal must have been influenced by the sign 

 boards; otherwise, the effects of the misleading experience-deter- 

 minants to first choice would have been reflected in a considerably 

 lower than the average chance per cent (25 per cent) of correct 

 first choices. 



