Eli5 : the principle of univariance
In: Biology
We have three types of cones in our retina, L (long), M (medium) and S (short) denoting the relative wavelength that maximally excites each cone. However.. If you want to see a big range of the visible spectrum, you can’t have sensors that detect one wavelength each, otherwise you’d see only three wavelengths and would be blind to everything else. The cones are actually responsive to a spectrum of wavelengths with a peak at a given wavelength. So they could detect say 500 to 600 nm, with max detection at 450 nm. Now the problem is, these spectra are a bit wide, so they overlap from different cones. Roughly, for S it’s 400-530, M 430-650, and L 430-700. You can see a huge overlap, especially between M and L. If you shine monochromatic light say at 500 nm, all cones are activated, so how can you tell what is the actual color? Well that’s easy, because 500 nm will make each cone elicit a signal of different intensity (neuron changes frequency of signal to the brain), and you can then tell which receptor has maximum activation and know the color. But that doesn’t solve another giant problem. You see, light varies with two variables, one is wavelength sure, but the other is intensity. But the output of cone cells is just one dimensional, intensity of their signal. This means a highly intense off-peak light would give the exact same signal intensity from a given cone as a low intensity on-peak light. And now you can’t tell which color it is. It gets worse because the light we see is rarely ever monochromatic anyway which means multiple wavelengths are contained in it. This means you have mathematically an infinite number of possibilities for a given light when you get signal from a cone. How is this solved? Well it becomes necessary to utilize signal from all the cones and process that to backcalculate which color you’re seeing, this part is pretty hard to explain in a comment. You can probably figure it out yourself, just pick some light color, say a combination between two wavelengths, one in red and one in green. And then look up the spectra of the cones on Google. You can draw the different combinations of intensities and wavelengths that will give you the same output signal from a given cone. You’ll see the possibilities are infinite. Then try to do that when looking at the signature signal (the combination of all cone outputs) and you’ll see you can actually represent a given light by a given signature and resolve this issue.
Edit: English
Latest Answers