HUMAN REACTIONS IN A MAZE 117 



indicative of the absence of higher mental processes. The 

 functional presence of rational processes would be evident in 

 an abrupt descent. Hobhouse, while not explicitly accepting 

 the assumption, called attention to the abrupt initial descent of 

 Thorndike's curves and maintained that, according to the assump- 

 tion, rational processes were functionally present. Watson ques- 

 tions the assumption and argues that this abrupt initial drop 

 of learning curves can be legitimately explained in other terms. 



The slope characteristic of any curve is determined by three 

 factors which are to some extent independent variables. These 

 are (a) the number of trials, (b) the surplus values, and (c) 

 relative rate of elimination for different trials. Obviously, the 

 slope of a curve increases as the number of trials decreases, 

 when the other two factors remain constant. The abruptness 

 of the slope is also directly proportionate to the size of the 

 gross values, when the other factors remain constant. Likewise, 

 two curves may be alike as to the gross values and the number 

 of trials, and yet their slope may be radically different; one 

 may exhibit a straight line descent, while the abrupt initial 

 drop may characterize the other. Obviously, Hobhouse and 

 Watson are discussing this " abrupt initial descent " as compared 

 with the remaining portion of the curve, and this is the third 

 factor mentioned, viz., the relative rate of elimination. Hob- 

 house assumes that Thorndike also referred to this feature of 

 the curve. Thorndike's term " gradual slope " is to some extent 

 ambiguous and the writer has often harbored the suspicion that 

 the emphasis in Thorndike's mind was placed upon the " number 

 of trials " rather than upon the " relative rate of elimination." 

 Obviously a learning process which extends over twenty trials 

 may be termed " more gradual " than one which terminates in 

 four trials. 



Our results indicate that any inference from such a general 

 characteristic of a curve is not feasible, because we are dealing 

 with a complex phenomenon due to several independently 

 variable factors. Our results indicate that the rational status 

 of a group of animals cannot be inferred from the slope of a 

 curve in so far as this slope is dependent upon the number of 

 trials or the relative rate of elimination. They indicate, more- 

 over, that inferences as to intelligent status are legitimate in 

 so far as the slope is determined by the factor of total values 



