J.— PSYCHOLOGY. 187 



relationship. This he perfectly legitimately isolates as far as possible, 

 trying to state its characteristics in terms of the functions of the mechanism 

 immediately concerned. 



These are precisely the type of problem that the psychologist has most 

 often attempted to carry over into his own field. Let us consider one 

 point which grew out of the early experimental work on psychophysical 

 methods. Fechner believed that all the content of experience, from the 

 relatively simple sensation to the most complex reasoning content, was 

 definitely measurable. Mainly as a result of his own observations, but 

 aided also by the earlier work of Weber, he thought he could demonstrate 

 this experimentally by showing that stimulus intensities and sensation 

 intensities are related by a definite principle, the value of the unit sensation 

 varying with the modality of stimulation. In order to establish the zero 

 point for a sensation of given mode, and also the series of just noticeable 

 differences piled up from this point by increase of the stimulus, he carried 

 out a wonderful and patient series of experiments. He employed and 

 somewhat improved Weber's method of ' limits ' ; he developed the 

 method of ' mean error,' but he relied mainly upon the method, already 

 proposed by Vierordt, of ' right and wrong ' cases. Fundamentally, this 

 method consisted in the presentation to an observer of a series of stimulus 

 variables, each variable being presented a number of times over together 

 with some standard magnitude, the order of presentation of the variables 

 being haphazard. Thus for each stimulus variable a number of judgments 

 were obtained upon its equality with or difference from the standard. As 

 might have been expected, the observer was often doubtful and said so. 

 Now, argued Fechner, if these doubtful judgments were resolved by an 

 ideal observer, they would in the long run be equally divided on both 

 sides of the alternatives which confront him every time he judges under 

 these experimental conditions. Consequently, in determining the stimulus 

 value which gave a sensation difference threshold, he adopted the plan of 

 dividing the doubtful cases equally between ' right ' and ' wrong ' cases. 



This procedure was speedily and acutely criticised by G. E. Miiller. 

 Miiller pointed out, among other things, that the judgment in these cases 

 is not simply a function of the stimulus on the one hand and of the 

 sensitivity of the observer on the other hand. It is determined also by 

 the precision of the observer, and this will affect the actual significance of 

 the doubtful cases. But he did not go all the way with this notion ; he 

 thought that statistical manipulation could give him a measure, both of 

 precision and of sensitivity. 



Thus grew up a controversy which has proceeded voluminously ever 

 since, though at the moment it seems nearly to have worn itself out. 

 Broadly, four plans for dealing with these ' doubtful ' judgments have 

 been proposed and followed : 



(a) to divide them equally between ' right ' and ' wrong ' cases ; 



(6) to ignore them ; 



(c) definitely to instruct the observer to ' guess ' when he is 

 uncertain and then, if doubt still persists, to lump all cases together 

 with ' guesses of greater ' and ' guesses of less ' as constituting an 

 area within which stimulus variations have no corresponding sensation 

 differences ; 



