3 ms·

I thoroughly enjoyed the article, but I do tire of the “hundreds of olfactory receptors Vs JUST THREE colour receptors” trope. Perception of EMR is fundamentall

I thoroughly enjoyed the article, but I do tire of the “hundreds of olfactory receptors Vs JUST THREE colour receptors” trope. Perception of EMR is fundamentally simpler (just two receptors could do it, one at either end of the spectrum). Smell is fundamentally more complex, which is, I suppose, the point, but I feel that the naked repetition of the trope harms articles such as these, where acknowledging the physical complexity directly, without the trope, might better serve.

scaramanga 4y ago
What I take from that was the interesting point that you can take a principled approach about what "correct colour perception" means. But there's no such stable basis for deciding that with olfaction given that it seems like an ad-hoc jumble of molecules to sense and there's such a wide and, seemingly random, variety of gene variants in the gene pool.
thfuran 4y ago
(just two receptors could do it, one at either end of the spectrum). One for ULF, one for >TeV gamma, and damn everything in between? That doesn't sound great.
- adrian_b 4y ago
  I believe that it was meant that each of the 2 receptors is not tuned to a single frequency, but its spectral transfer function must have an approximately triangular shape, with the peak at the nominal frequency, and the slope of the spectral transfer function must be such that the transfer function does not become null before reaching the nominal frequency of the second receptor. In such a case, it is possible to determine both the amplitude and the frequency of any electromagnetic radiation, e.g. light, with a frequency intermediate between the 2 peak frequencies of the receptors, from the 2 values provided by the 2 receptors. However spectral transfer functions of such a shape can be approximated only in a narrow bandwidth. Therefore both in animal vision and in artificial vision, a larger number of receptors must be used to cover a large bandwidth, with approximately equal spacings between their peak frequencies, in order to be able to determine not only the intensity but also the color (i.e. the frequency) of incoming light.
  - thfuran 4y ago
    >In such a case, it is possible to determine both the amplitude and the frequency of any electromagnetic radiation, e.g. light, with a frequency intermediate between the 2 peak frequencies of the receptors, from the 2 values provided by the 2 receptors. Is it? How would you distinguish between monospectral illumination at a frequency between the two peaks and lower amplitude polyspectral illumination at each of the peaks? That ambiguity is essentially the basis of rgb displays.

scaramanga 4y ago

What I take from that was the interesting point that you can take a principled approach about what "correct colour perception" means. But there's no such stable basis for deciding that with olfaction given that it seems like an ad-hoc jumble of molecules to sense and there's such a wide and, seemingly random, variety of gene variants in the gene pool.

thfuran 4y ago

(just two receptors could do it, one at either end of the spectrum). One for ULF, one for >TeV gamma, and damn everything in between? That doesn't sound great.

adrian_b 4y ago
I believe that it was meant that each of the 2 receptors is not tuned to a single frequency, but its spectral transfer function must have an approximately triangular shape, with the peak at the nominal frequency, and the slope of the spectral transfer function must be such that the transfer function does not become null before reaching the nominal frequency of the second receptor. In such a case, it is possible to determine both the amplitude and the frequency of any electromagnetic radiation, e.g. light, with a frequency intermediate between the 2 peak frequencies of the receptors, from the 2 values provided by the 2 receptors. However spectral transfer functions of such a shape can be approximated only in a narrow bandwidth. Therefore both in animal vision and in artificial vision, a larger number of receptors must be used to cover a large bandwidth, with approximately equal spacings between their peak frequencies, in order to be able to determine not only the intensity but also the color (i.e. the frequency) of incoming light.
- thfuran 4y ago
  >In such a case, it is possible to determine both the amplitude and the frequency of any electromagnetic radiation, e.g. light, with a frequency intermediate between the 2 peak frequencies of the receptors, from the 2 values provided by the 2 receptors. Is it? How would you distinguish between monospectral illumination at a frequency between the two peaks and lower amplitude polyspectral illumination at each of the peaks? That ambiguity is essentially the basis of rgb displays.