Really what you need to do is match the wavelenghts that our rods and cones in our eyes are most sensitive to. That should be the optimal solution in terms of energy usage.

Perhaps. But it might not be pleasant to the eye.

The 'red' cones alone can't tell the difference between orange and brick red. The idea is to stimulate all 3 types of cones and the rods in the same proportion as daylight does. RGB LEDs should have the cones covered; I wonder if the problem are the rods? Do we need 4-component LEDs?

I don't know if all the cells in your eye are perfectly tuned to three peeks. I expect biological systems to show some variance.

The mechanism is chemical/physical, though, and depends on the absorption spectrum of a certain pigment molecule for each type of receptor. I strongly suspect the variance in those is likely to be tiny.

The fact that you can see a full spectrum -- as in a rainbow, thin-film refraction on a puddle on a paved street, etc. -- should be an indication that the band-pass for neural firing is relatively wide. If each optical receptor had a narrow acceptance window (the way that individual atoms absorb), there would be gaps in the rainbow. Remember that ALL of the colours you see in a rainbow are pure, composed entirely of photons of a particular wavelength proper to each of those colours. It's not RGB; it's not CMYK.

We can, to an extent, simulate a full spectrum through additive and subtractive colour mixing of "primaries", but it is only a simulation. (And Edwin Land demonstrated that we don't actually need all three of R, G and B in order to create the illusion of a full-colour picture with only a comparatively tiny gamut loss. It is a fragile thing, though, and requires the integrative function and cognitive mapping of our brains.

Similarly, we can easily tell the difference between full-spectrum lighting and discontinuous spectrum lighting. Discontinuous spectrum lighting makes us uncomfortable. None of the colours looks quite right. Food is unappetizing, pictures and people less attractive. Did the light-makers just miss the right mix of frequencies by a hair? No -- any scheme that relies on a small number of discreet wwavelengths will cause a similar discomfort. We may not know why we feels the way we feel at the time, but we sho' 'nuff knows we'se feelin' it.

You make a good point - I forgot about indirect light and banded spectra interacting with surfaces.