114 ^Journal of Coynparative Neurology and Psychology. 



or negatively phototropic to a marked degree, and also showed a 

 preference for one color or another, the preferred color was taken 

 much darker than the other, if the animal was positively photo- 

 tropic, or lighter if negatively phototropic. The persistence of 

 the preference under these conditions showed it to be a true 

 color-preference. Graber's results for the two species of fish 

 were approximately the same and showed decided preference for 

 white over black, a lesser degree of preference for blue without 

 the ultra-violet rays over blue with the ultra-violet rays, for red 

 over green, and for green over blue (ultra-violet). The last- 

 mentioned could hardly be called a preference at all, and the 

 difference between red and green was so slight as to be reversed 

 when the green was made decidedly dark. In fact, the color- 

 preferences proper, as distinguished from the cases involving 

 ultra-violet rays, are scarcely marked enough to allow one to con- 

 clude from the experiments that the fish tested had the power of 

 discriminating colors. 



The subject of the following study was a female of the common 

 species, Semotilus atromaculatus, the creek chub or horned dace. 

 Our general plan upon beginning the investigation was to test 

 color-discrimination by establishing, if possible, an association 

 between a certain color and food. This method has a two-fold 

 advantage over that employed by Graber. First, it is a true 

 test of discrimination as distinguished from preference; and, 

 second, involving, as it does, "associative memory," the truly 

 psychic nature of the phenomena resulting will be admitted by 

 the most conservative biologists, whereas the "preference" 

 method, which involves reaction to present stimulation only, 

 establishes the existence merely of a tropism which may or may 

 not have a mental aspect. Of course, in the case of fish, where 

 intelligence has been shown to exist, by the experiments of 

 MoBius, Thorndike, Triplett and others,^ the presumption 

 would be for consciousness. 



The fish was kept throughout the experiments in a circular 

 glass tank 50 cm. in diameter and 45 cm. deep. The apparatus 

 used for feeding it consisted of two like pairs of dissecting forceps 

 w^hich were faced on the outer surfaces with four-cornered strips 



*Zeitschr. d. gesammt. Naturwiss., Bd 42, p. 89; Amer. Naturalist, Vol. 33, p. 923; Am. Jour, 

 of Psych., Vol. 12, p. 354. 



