How can I make sense of the way (screen) colors are generated?

Question

OK here is some common knowledge I have from the net: a computer screen has 3 constituent colors of red, green, blue. To make all spectra of stuffs it just combines them.

We also know that red is about 650nm, blue 470 and violet 425. In real life, mixing yellow and blue creates green, a color in-between.

So how the hell can computers combine red & blue to give violet? In my understanding, (650+470)/2=425?? Using that absurd principle, then surely it's possible to create IR & UV light by mixing RGB on a computer screen?

Bonus question: to make things more perplexing, let's add biology in for a bit. Our eyes have 3 types of cones with maximum responsivity near computer's RGB values. When a natural light strikes our eyes - say from a pure 425nm violet flower - according to a Wikipedia picture I can say that the S-cones are highly activated while M & L are barely on. See, that's very different from the combination of red & blue excitation that screens give us. So... should we conclude that computer colors are "fake"?

Answer 1

The computer screen RGB color mapping protocol takes advantage of metamerism, which is the physiological optics effect wherein stimulation of the different color receptors in the human retina produces a full range of perceived colors in the brain- even though the retina is being presented with a combination of ONLY red, green, and blue wavelengths.

We can therefore conclude that the computer screen is exploiting metamerism to trick our brains into believing the screen generates a full-color spectrum when it actually does not.

Answer 2

As the other answer already tells you, computer colours are indeed fake, they rely on the way human beings perceive colour. I want to answer your question about generating UV and IR frequencies, but first, a detour. Let us say that the circle of minimum area that a human eye can discern has area A. For two such circles adjacently placed, the human eye can distinguish their colours. On the other hand, if we place smaller circles of two different colours $C_1$ and $C_2$ adjacent to each other, the eye discerns one blob of colour $(C_1+C_2)/2$, pretty much like your "absurd principle", except that we do not literally add wavelengths in our brain. Instead, the eye has three specific, independent colour receptors, and any other colour is indexed in the brain by what combination of these three generators is involved in it. The point is that the brain is not measuring wavelengths, it is measuring number of cone cells of each type excited by the incoming light, and reading off the colour from its inbuilt dictionary, say $$\bigg(\frac{2}{10} S+\frac{3}{10} M+\frac{5}{10} L\bigg)=\text{violet}$$ or some similar algorithm.

The reason we cannot detect UV or IR is now easier to address: these frequencies are not detected by any of the cone cells. So the combination is basically $$(S,M,L)= (0,0,0)$$ The number of distinct colours in the brain's vocabulary (considering perfectly functioning receptors and other mathematical idealizations necessary) is therefore the number of ways of partitioning the number of colour receptors we have into three subsets. If you place the subject in a perfectly dark room and illuminate the room with UV rays, the brain gets a $(0,0,0)$ signal, consults its dictionary, and reads it as 'black' (assuming the illuminating radiation isn't one that burns out the eyes, in which we could probably tell in retrospect that it wasn't black).