484 ADAPTATIONS TO PHOTIC QUALITY 



from green (A510-550m[x). One member of each pair was the positive 

 stimulus at first, and the fish was later retrained to the other stimulus — 

 a further attempt to eliminate preferences or any possible discrimination 

 on a brightness basis. 



White-Hineline's decidedly inconclusive work was far overshadowed 

 by the restricted but thorough and beautifully controlled investigations 

 of Reeves, a student of Reighard. Reeves was content to employ two 

 hues only, devoting her time and energy entirely to the elimination of 

 the brightness factor from this one discrimination. While other more 

 recent researches have yielded superficial information about much more 

 of the whole color-vision system of the fish, that of Reeves stands as a 

 model demonstration of an unquestionable discrimination of hue as hue. 



She used an adaptation of the Yerkes- Watson discrimination box 

 described in the 'ideal investigation' outlined above, and studied the 

 untrained and trained responses of several species, chiefly Semotilus 

 atromaculatus and Lepomis gibbosus, to white and (filtered) red and 

 blue lights. The blue filter passed the short-wave end of the spectrum 

 up to ^509m(l. Several different red filters were used, principally one 

 which transmitted only wavelengths longer than 589m[l. 



The dace (Semotilus) showed very poor brightness-discrimination in 

 preliminary experiments. At least they would not distinguish intensities 

 which differed in less than 1 :4 ratio. When two such intensities were 

 presented, apparently barely discriminable by the animal, the insertion of 

 the red filter in the path of the brighter light (without any other change 

 being made) produced a marked change in behavior, although the 

 intensities were presumably no longer discriminable, being too nearly 

 equal. The fish evinced a strong red-shyness, which however was tempor- 

 ary and in young individuals was absent. Dace were now trained positive 

 to a dim blue stimulus versus an intense red one. After training, the 

 intensity of the red was cut to 60%, 37%, then 20% of its original 

 value, the dace still going to the blue to seek food 90% of the time. 

 Repeated with more gradual reduction, the typical response was an 85%- 

 95% correct choice until a critical value of red intensity was reached, 

 at which the accuracy suddenly dropped to 60%! 



At this point, the two stimuli were obviously alike in some way for 

 the fish. A permanent performance of 50% or 60% accuracy would have 

 meant that they were wholly identical; but Reeves found that with no 

 further changes the accuracy eventually rose again to 83%, and upon 

 a still further dimming of the negative stimulus the performance went 



