254 F. M. GREGG AND C. A. McPHEETERS 



finally made immediately after the first presentation. With 

 the positive series, the animal became impatient to react after 

 the first presentation and in fact many such premature re- 

 sponses did occur. In all probabilhy immediate reaction to the 

 first member would also have developed if this tendency had 

 not been punished by the withholding of food. (2), The final 

 member was never effective in the discrimination. The con- 

 clusion is supported by the facts already cited as to the ten- 

 dency to correct premature responses to the positive group, 

 and the usual reaction after the first presentation of the nega- 

 tive group. This third element was common to the two groups 

 in the original experiment, and logically it is obvious that a 

 common element can not serve as a basis of distinction. This 

 training on the original series would naturally tend to make 

 the animal neglect this final member in those cases where it 

 did offer a possible basis for discrimination. As a matter of 

 fact those groups which did differ as to the third component 

 gave a discriminative accuracy no higher than the normal. 

 (3), While the third member is useless so far as discrimination 

 is concerned, yet it did possess a function in the positive group. 

 As Cole asserts, this component "was not a neglected element 

 of the situation." 4 It functioned, however, merely to release 

 the discriminative response which had already been aroused by 

 the preceding members of the group. (4), As to the efficacy 

 of the second member, our results are far from conclusive. 

 Exigencies of time at the close of the college year prevented 

 an adequate completion of the tests. The facts as to premature 

 responses indicate that this element was sometimes effective 

 in the positive series, but that it constituted no part of the 

 negative stimulus. In test no. 5, the groups 1-1-1 and 1-2-3 

 were both regarded as positive stimuli, but the percentage of 

 responses was slightly greater in the latter case. This fact 

 indicates a slight effectiveness of the second element. Some 

 results of test no. 6 corroborate the assumption. A comparison 

 of the last two groups shows that the accuracy of discrimina- 

 tion is slightly greater when the two series are different as to 

 the second member than when they are identical in this respect. 

 The conclusions thus far advanced are consonant with all 

 of the facts except those of the first control test. With the 



4 Op. cit., p. 258. 



