140 VINNIE C. HICKS 



elimination of the cul de sacs constitutes apparently less than 

 two-thirds of the maze problem. Again, time is the only prac- 

 ticable criterion for most of the various problem boxes, inas- 

 much as they offer no conditions by which a unit of error can 

 be established. On the other hand, error is the more practicable 

 criterion for most discrimination tests, since the time of the 

 reaction may become too small to be used advantageously, while 

 errors are easily defined and enumerated. 



Several valid theoretical reasons can be adduced in support of 

 Watson's rejection of the error criterion. There is the difiQculty 

 of defining an error. Shall errors be confined to entrances into 

 cul de sacs, or shall they also include total and partial returns 

 along the true pathway, turning around 360 degrees, stops to 

 scratch or rest, etc., etc.? There is the further difficulty of de- 

 fining the unit of error. This embraces the general question 

 whether all errors are to be regarded as of equal value. What 

 relative value shall be assigned to a total return in comparison 

 to a partial return? Shall entrances into a complex cul de sac 

 be given the same value as into a simple one? In case an animal 

 runs back and forth a number of times inside a blind alley, shall 

 this be given the same value as the case of a simple entrance? 

 In the progressive elimination of a blind alley, we find all degrees 

 of entrance from a thorough exploration on the one hand to 

 a mere hesitation at the opening at the other. Shall all of these 

 entrances be regarded of equal value? How far is it necessary 

 for an animal to enter a blind alley in order to constitute an 

 error? It is not our purpose to answer these questions at this 

 point, but to observe that a psychology that pretends to be 

 comparative is in duty bound to state its results in such a fashion 

 that data obtained at different times and by different observers 

 shall be strictly comparable. It is evident that error curves will 

 differ markedly unless the unit of error is defined accurately 

 and used consistently. From my own experience and that of 

 others, it is a difficult matter to use an error criterion consist- 

 ently. It would be a difficult matter for any group of experi- 

 menters to agree upon a standard unit of error, and they probably 

 would encounter difficulty in using this standard in a consistent 

 and comparable manner. 



Time also presents certain difficulties. Rats almost invariably 

 " fool " around the entrance box of a maze for a variable length 



