170 The Animal Mind 



even when their brightnesses were considerably varied. 

 Yet the probability appeared that these discriminations 

 were based merely on brightness differences, for after a 

 mouse had learned to choose green rather than red, when it 

 was offered a choice between light and darkness, it uni- 

 formly preferred the former, although untrained mice 

 showed no such preference. Apparently, then, the green 

 had been previously discriminated simply as the lighter of 

 the two impressions, and to the eye of the mouse, as to that 

 of the color-blind human being, red looks an extremely 

 dark gray. 



In some experiments of the writer's (756) on the rabbit, 

 the method was used of presenting a color with various 

 grays, in successive experimental series, and finding whether 

 or not there existed a gray with which the color was confused. 

 This is the only adequate way of dealing with the bright- 

 ness error. We found that while the rabbit could be taught, 

 by rewarding it with food for right choices, to distinguish 

 a standard red paper from a number of different gray 

 papers, it invariably failed when a very dark gray, almost 

 black, was presented with the red. Two objections which 

 have been urged against the use of colored papers were 

 met in these experiments. In the first place, it is argued 

 that papers of different colors may differ in surface texture : 

 the possibility that our rabbits reacted to this clew we 

 eliminated by occasionally substituting red and gray vel- 

 vets for our red and gray papers, a change that did not at 

 all affect the rabbits. Secondly, it has been urged that 

 when colored papers are pasted on cards, they are apt to 

 show wrinkles that might identify them: this we obvi- 

 ated by pirming on our papers afresh in successive experi- 

 ments. 



Obviously, however, since colored papers do not give 



