Thanks to Eric Lundquis for typing this up and putting it on the internet. It’s an experiment done by the army and cited by Rubin, M. L., and Walls, G. L. (1969). Fundamentals of visual science. Springfield, Ill.: Thomas, p. 546, which is in turn cited Sekuler, R., and Blake, R. (1994). Perception (3rd ed.). Springfield, Ill.: Thomas, pp. 62-63:
The following story dramatizes how photopigments determine what one can see. During World War II, the United States Navy wanted its sailors to be able to see infrared signal lights that would be invisible to the enemy. Normally, it is impossible to see infrared radiation because, as pointed out earlier, the wavelengths are too long for human photopigments. In order for humans to see infrared, the spectral sensitivity of some human photopigment would have to be changed. Vision scientists knew that retinal, the derivative of vitamin A, was part of every photopigment molecule and that various forms of vitamin A existed. If the retina could be encouraged to use some alternative form of vitamin A in its manufacture of photopigments, the spectral sensitivity of those photopigments would be abnormal, perhaps extending into infrared radiation. Human volunteers were fed diets rich in an alternative form of vitamin A but deficient in the usual form. Over several months, the volunteers’ vision changed, giving them greater sensitivity to light of longer wavelengths. Though the experiment seemed to be working, it was aborted. The development of the “snooperscope,” an electronic device for seeing infrared radiation, made continuation of the experiment unnecessary (Rubin and Walls, 1969). Still, the experiment demonstrates that photopigments select what one can see; changing those photopigments would change one’s vision.