Why red light is not darkness
Rodent facilities are full of red light, and the reasoning behind it is old and tidy: rats and mice have no long-wavelength cone, so red is invisible to them. You get to see what you are doing while the animal stays, functionally, in the dark.
We ran into this while building a visual–tactile discrimination task, where the dark condition had to actually be dark. It is not. Rats retain usable visual capacity under red illumination — enough to keep solving a task that the standard account says they should be blind to.
The photoreceptor story is not wrong so much as pitched at the wrong level. Absence of a matched cone class is not absence of absorption. Pigment spectra have long tails: something peaking near 500 nm still catches 620 nm photons, just inefficiently — and “inefficiently” stops mattering once the lamp is bright and the animal has all night to integrate. A behaving animal will use whatever signal it has, not the signal our shorthand says it should have.
Two things follow, and neither is subtle. An experiment that treats red light as the absence of a visual cue is not controlled for that cue. And housing that treats red light as night is not giving the animals night.
Neither is hard to fix. Both are easy to keep getting wrong, because the justification sounds mechanistic enough to stop the conversation.
The full argument is in eLife.