I recently wrote a little post about a colour recognition plug-in I wrote for OpenCFU. To better tell the difference between colours, I transformed the RGB values in the image into their corresponding L*a*b* so that I could use a reasonable measure of colour difference. With that metric, this image shows the colours in the right-most panel looking the most similar to me (d-values refer to the size of the colour perception difference, bracketed values are the RGB colour points for each square):
I then asked someone else after publishing the post which colours were the most similar to her, and she told me that it was definitely not the set at the right. So I’ve been thinking for a while now: “what’s the deal with colour perception”.
Here’s roughly what I already knew about colour perception, at least at a mechanistic level:
- Most people detect three different colours in the ‘cone’ cells of the eye (this is typically diminished for the varying forms of colour-blindness).
- A separate set of ‘rod’ cells are better adapted to seeing changes in brightness.
- Some people (almost always women because the relevant genes are carried on the X chromosome) have a fourth pigment type in their eyes, possibly permitting them to see more colours.
Could this fourth pigment be the source of the disagreement between my friend and I? She is significantly more likely than me to possess four colour pigments than me. Unfortunately, this is a particularly difficult hypothesis to test. In the study cited above, genotypes were analysed to identify people with a fourth pigment type. It would be much simpler if there were an effective test similar to the Ishihara colour blindness test to determine this (below – people with normal colour vision will see the number 74).
As far as I could find, there is no simple test for this. Newcastle University’s Tetrachromaticity Project, which searches for people with four colour pigments, claim that that most RGB colour gamuts cannot test for how many cones people can see, and while some tests are floating around on the internet, I haven’t found any with decent scientific support.
But what gives us our different colour perception, varying from person to person. At least at a biological level, the three pigments in normal colour-vision people have the following response to different wavelengths of light (data is from the Colour & Vision Research Laboratory, plotted using Gnuplot).
The short (S) wavelength cones are more sensitive to blue light, and the gene encoding for it is located on chromosome 7, whilst the medium (M) and long (L) wavelength cones are located on the X chromosome, hence the prevalence of red-green colour-blindness amongst males. So when colours are perceived, these three wavelength sensitivity relationships allow a single wavelength to be converted to a colour in our brains. What these curves don’t show though is the variations in sensitivity between individuals. There are many processes that are involved in the translation of the wavelength of light to a colour, and at every step individual characteristics can slightly change the interpretation of that wavelength between individuals.
As for the fourth colour cone? It definitely exists, and seems to have a response similar to the medium and long wavelength cones, coming from small variations to these pigments (if I’ve understood the relevant article correctly). But it’s significance in effecting colour perception isn’t always very strong, so it might not be why my friend and I perceived those two colours differently. Maybe one day personal genomics will be able to help answer that question more concisely, but for now there is certainly enough scope within individual variation to allow two people to disagree about colours, without searching too deeply.