90 ■ CORA D. REEVES 



training experiments the fish show an initial avoidance of 

 red and that they rapidly become adapted to red and then 

 no longer avoid it. (Graphs 13 and 15 and discussion of 

 these in text.) It appears further that when first presented 

 with red and blue patches of matched brightness for the 

 human eye fish show without training, by manner and fre- 

 quency of approach to the patches, that the stimulus value 

 of the two is not the same for them. When in experiments 

 they have become adapted to red and blue patches, their 

 behavior is greatly altered when a gray patch of matched 

 brightness is substituted for the red. Many innate responses 

 indicate the different stimulus values for the fish that 

 lights of longer and shorter wave-lengths have. 



In spite of the fact that the fish discriminated red and 

 blue stimulus patches when they were matched in brightness 

 for the human eye and when the behavior of the fish indi- 

 cated that they were matched in brightness for them; it 

 is possible to postulate extraordinary powers of brightness 

 discrimination for the fish, powers far beyond those of 

 man, and to explain the results obtained with the use of 

 food association on that basis. Thus, postulating high 

 capacity to differentiate slight brightness differences, we 

 might explain the graph in figure 13 on the basis of such 

 discrimination. Failure to discriminate at slit width of 0.9 

 mm. would then be due to the fact that the stimulus patches 

 were nearly but not quite matched in brightness for the 

 fish. Recovery of discrimination at this slit-width could be 

 attributed to continued training at these slightly differing 

 brightnesses. To determine whether the fish possesses 

 extreme powers of brightness discrimination tests were 

 made to find by how much two white stimulus patches 

 must differ in intensity in order that the fish may learn 

 to go to the duller patch for food. It was found that in- 

 tensity differences of white light of 1 to 4 were necessary 

 for discrimination in the case of dace and of 1 to 3 in the 

 case of sunfish. The fish do not therefore under the con- 

 ditions of the experiments show powers of accurate bright- 

 ness discrimination sufficient to explain the graphs shown in 

 figures 13 and 15, nor sufficient to explain their innate dif- 



