COLOR-VISION RESEARCH PROCEDURE 469 



numbers of times in the long run, but the alternation must not be at all 

 regular or the animal will probably catch on to that. Some investigators, 

 using choice-boxes for any sort of comparative psychological work, flip a 

 coin before each trial to decide which side the positive stimulus is to have. 

 In the course of a long experiment — and they are always long — there 

 will be equal numbers of heads and tails. 



When the animal has done his daily stint in trials — maybe ten, maybe 

 a hundred, depending upon his capacity for work and the speed with 

 which he gets filled up on the amounts of food which are big enough to 

 interest him as rewards, you let him finish his dinner in his cage in the 

 next room, and start with the next animal. The animals will get no more 

 food until the next day's experimental period, ensuring that they will 

 then be hungry and willing to work. Perhaps the food drive will not 

 prove sufficient, and you may have to wire the floor of the choice-box so 

 that you can give the woodchuck a light shock if he turns to the wrong 

 side. The stimulus with which you shock him will now be the positive 

 stimulus, and must not be varied; for he will now be making the associ- 

 ation 'not blue = pain'. If you assume that he associated 'blue = safety', 

 and change the other stimulus, he may seem to be unable to discriminate 

 between the two, simply because he no longer knows what to avoid. 

 Obviously, you must not both reward him for going to one stimulus and 

 punish him for going to the other, or you cannot alter either and there- 

 fore cannot find out whether they can be made to look alike to him with- 

 out looking alike to you. 



The stimuli themselves may be squares of colored paper or they may 

 be patches of light cast on ground-glass screens from behind. If they are 

 of paper, you can have the advantage of working with fully light-adapted 

 animals, but it will be harder to make certain that the discrimination is 

 not on a basis of brightness alone. If they are lights, the room will need 

 to be darkened. They may be colored by being passed through gelatine, 

 glass, or liquid filters; or they may be more nearly monchromatic lights 

 selected by slits from a broad, bright spectrum. In any case no wave- 

 lengths present in one should also be present in the other. It must be 

 convenient to alter their intensity over a great range without changing 

 their hue, though of course only one will be so altered in any given 

 course of training. This will require changing the distance of the lights 

 from the screens, or interposing various thicknesses of ground glass or 

 smoked wedges which do not change the size or shape of the stimulus- 

 patches. 



