OBSERVATIONAL EVIDENCE FOR COLOR VISION 467 



fly means something to a trout; but experimental evidence is more satis- 

 fying to the souls of scientists and — when it is sympathetically inter- 

 preted to them — of sportsmen as well. These worthies have suffered 

 abundantly with the psychologist's disinclination to try to study trout in 

 small experimental aquaria, and with his warnings to the angler against 

 assuming that if a mud-minnow has demonstrable color vision, a trout 

 must have it also. 



In our consideration of experimental procedures and findings, it may 

 be said at the outset that much of the scientific literature itself is largely 

 or wholly worthless. Almost always this is owing to incaution regarding 

 the big danger mentioned above. In the work of one prolific investigator 

 — Carl von Hess — it is also the result of certain assumptions which no 

 one before or since has deemed it at all wise to make. Even where the 

 researcher has made only correct assumptions and has been fully awake 

 to all possible errors of interpretation, he has not gotten far — no one 

 could — if his method has made use only of unlearned and untrained 

 responses by the animal. These can be valuable, but only methods in- 

 volving the training of the animal can be very fruitful, or make possible 

 anything like a complete analysis of a color-vision system. 



If the animal is to be made 'positive' to one of two or more stimuli— 

 in other words, trained to evidence a discrimination if he can make one 

 — he must be capable of trial-and-error learning. The angler may be im- 

 patient with the scientist for spending his time with the wrong species; 

 but the scientist is as often annoyed by the rigors of his code, which for- 

 bids him to use any methods but those he considers the best, for this 

 often prevents him from working upon the very species about which he is 

 most curious. For if an animal is very stupid, like an opossum or a guinea- 

 pig, he might have flamboyant colors in his brain, yet we might not be 

 sure that he had any. If, like a snake, he eats infrequently and responds 

 to mild punishment for his errors by getting angry, sulky, or flighty, he 

 may be both highly intelligent and richly color-perceptive; but we may 

 not be able to help him to prove it by giving him a proper incentive to 

 work to make discriminations. The animal must get the idea of what is 

 expected of him, must be willing to work for frequent small rewards, and 

 must be able to do the task involved in demonstrating a choice between 

 stimuli to the experimenter. 



A Sample Ideal Procedure for Investigation — The reader will un- 

 derstand better the difficulties involved in really good color-vision 

 research, and will be able to see for himself the loopholes in some of the 



