Re: How can one measure what colors the eyes of an animal perceive?

Dear Barbara,

It is a very good attitude to ask « how do we know that » when you hear 
something. In this case, you probably know that the first steps of color 
analysis take place in the retina. Receptors called « cones » are 
sensitive to specific ranges of wavelenght and are essential to color 
perception. In humans and some animals, there are 3 types of cones, 
respectively sensitive to short, medium and long wavelenghts. In 
daltonism, a particular type of cone is missing and the range of percieved 
colors is diminuished. Some experiments in normal and « daltonian » human 
retina fragments have shown that short, medium and long wavelenghts are 
absorbed by cones pigments in the normal and that in daltonian, a 
particular range of wavelenght is not absorbed at all. That is, it is 
possible to correlate the absence of a specific type of cone with the 
range of color that is percieved by a person. In animals, it is possible 
to determine how many types of cones are presents in the retina. Since we 
know that 3 types of cones are essential to percieved the full color range 
that normal people experienced and that daltonian people see a particular 
range of colors that is determined by the types of cones that remain, we 
can inferred about the range of colors that are probably percieved by a 
given animal on the basis of how many types of cones are present in its 
retina.

In order to precise the actual range of colors that an animal percieve, it 
is possible to make behavioral studies with this animal. In this type of 
experiment, the animal is placed in front of two colored panels and must 
approach one of those (previously chosen by the experimentator) in order 
to get a food reward. Suppose that an animal stands before a red and a 
green panel and must reach the green panel to get food. The experimentator 
must design the experiment in such a way that he/she can be sure that the 
animal will learn the task on the only basis of color discrimination.  As 
example, the panels must be randomly placed in the right and left 
positions between each trial (and a lot of trials must be done !) to avoid 
the animal learning that the reward is systematically obtained when it 
reach the left (or right) panel. In the case of color discrimination 
testing in animals, one determinant parameter concern the calibration of 
the presented stimuli. Two panels (red and green) may actually be 
dinstinguished even in the case where the animal actually does not 
percieve colors. Indeed, it is possible that the two panels differ, not 
only by their colors but also by their brightness. Therefore, it is 
necessary to measure the amount of luminance that is reflected by the two 
panels in the testing environment and to set them at same values. In 
that way, the differences between the two panels will only concern their 
respective colors. As you can see, behavioral studies about color 
perception in animals require very carefull precautions and can only be 
conducted with well calibrated stumuli. 

I hope this helps,
Eric