Have you ever wondered why, when you look at a rainbow, you see distinct bands of colour? You may see red, orange, yellow, green, blue, indigo and violet (though more likely you will not be able to distinguish between indigo and violet). We know that the wavelength of light in the visible spectrum – http://colourware.wordpress.com/2009/06/29/colour-101/ – varies smoothly and continuously, so why don’t we see a smooth and continuous colour spectrum? Why do we see distinct colour bands?
In my opinion the reason we see bands is because of something called categorical perception. We tend to want to group things that we perceive together into one class or another. But this grouping is not just a matter of putting things into boxes; it has an impact on how we perceive those things. We see categorical perception everywhere – indeed, I have often wondered whether even the periodic table of chemical elements is a true and accurate representation of how the world is or whether it stems from our categorical perception.
A recent study by Skorinko at the University of Virginia and colleagues at Rice University (published in Psychology and Marketing, 2006) finds that consumers have a more positive reaction to products whose colours are given rather exotic and flashy names such as mocha compared with the same products that are given plainer and genric names such as brown.
And how is that linked to the early statements I hear you ask (even though you asked very quietly)? Well, the authors hypothesise that the reason for the improved consumer reaction to the fancy colours is …. categorical percetption. The fancy names stimulate a more positive category than their plainer alternatives. It is also suggested that more ambiguous descriptions (mocha as opposed to brown, for example) yield higher consumer acceptance and safisfaction. I cannot resist finishing this blog with the last line from the paper by Skorinko et al. (2006) who write:
Indeed, the judgement of “that we call a rose” seems to be influenced by its name (Shakespeare, 1595).
Really interestiing stuff to ponder; I would like to learn more.
Me too. Increasingly much of my research is about the use of colour as a marketing/advertising tool. However, I currently don’t have time to study the use of colour names in this respect. But I hope to come across more information on this topic in the future.
Your image of the colour spectrum does not show Cyan. I think this is a cultural blindspot, particularly in the West.
Your categorical perception may well be down to the way the cones operate in the eye. This is analogous to tuning between radio stations on AM The (tunable) response curve does not change how a transmitting station is perceived (beyond volume and a little distortion), so you get a ‘band’ of the same information as you tune through it.
On a cheap radio where the response curve is wide, , you may hear the difference between the main carrier of the adjacent stations which ‘beats’ together to produce an audible note at 10kHz (USA/Canada) 9kHz (rest of the world). This locally produced beat is analogous to magenta as we perceive it, a synthesized colour we make in our minds.
It’s not possible to show the full and proper spectrum on a display based on RGB and so any picture that we show is always going to be a representation rather than a perfect copy. However, cyan is not seen in the spectrum and there is no reason for it to be. From this post and others you seem to think that red, green, blue, yellow, cyan and magenta (the primary and secondary colours of additive colour mixing) should be the six colours in the spectrum. This is bizarre and the first time in 25 years I have ever heard anyone suggest this. There is little, if any link, between colour mixing and the colours seen in the rainbow or spectrum. The additive primaries are RGB but there is nothing particularly special or fundamental about these colours contrary to much popular opinion; no matter which colours you use as the primaries your colour gamut will be limited. In just turns out that using RGB the number of colours you can’t make by mixing them is fewest and hence they are an empirical and pragmatic choice as primaries. Certainly orange is one of the colours that can clearly be seen in the rainbow or spectrum. It seems to me that you would like orange not to be there and cyan to be there for ideological reasons; take a look at the rainbow next time you have a chance.
Your comment at tuning is really interesting. If you look at a plot of wavelength discrimination – there is a nice one here http://www.yorku.ca/eye/color_discrimination.htm – then you can see that towards the end of the spectrum we get quite poor at discriminating between nearby wavelengths; the reason for this is, I think, exactly as you suggest. Despite wavelength changing the ratio of cone responses hardly changes at all. I am not sure this explains the bands in the spectrum though.
Funny discussion about Cyan.
One interesting thing, if you try to take a digital picture of a rainbow, is that the “spectral violet” (the violet end of the visible EM) will usually turn very blue, sometimes totally blue – and the screen’s “Cyan” will replace the “azzuro” or what I call deep or intense “sky blue.” These are the worst distortions I have ever seen caused by a RGB picture encoding. RGB gamut’s very limited in the Orange-Yellow and the Blue-Green (“Cyan”) regions of the human gamut.
It’s a famous fact that there are some unprintable colours, another less famous one is the fact that some printable colors (like the phtalocyanine blue or “Cyan” of printers is not “displayable” on a computer screen or a TV).
***
About the rainbow. It is true that the limited gamut of a RGB screen does not allow a “true colour” representation of it.
However, in nature, most of the times (all the time except in an optics laboratory dark room), we see rainbows whose colours are “diluted” in white light: for example there’s a SKY behind, and this sky emits some whitish-bluish light that superimposes with the rainbow showing up there.
It is perfectly possible th use CIE colorimetry to determine the most saturated and “exact” rainbow that is possible to display on a RGB screen. The idea is to take the colorimetric coordinates of the spectral colours, to apply on them the spectral distribution of some source of light (daylight, any lamp’s light, etc.) and then to desaturate the results by adding “white light.”
I did that thing there with a D65 light distribution:
https://adrienlucca.wordpress.com/2012/08/26/application-of-the-additive-colorsystem-representing-spectral-colors/
drawing here:
https://adrienlucca.wordpress.com/2012/09/26/d65-n3-transparences-lampe-spectres-2012/
This guy did the same in RGB for a computer screen: http://casocouleurb3.files.wordpress.com/2012/10/srgb-combo.png?w=960
In both cases, the colours are “exact”!
Give these explanations, I would personally agree that “Cyan” is a colour of the rainbow. The condition for that is: this rainbow is not seen in a complete darkness.
Any Cyan is a (slightly) desaturated spectral colour (an average between a spectral colour and some white light)
***
This “categorical colour” part of your post is the most interesting one, nothing is more true than this fact: we never look at colour alone, there’s always something in the judgment that relies on a specific type of experience, linguistic or not. We never look at a colour, always at some paint, object, screen, skin, air-with-skylight-in-the-morning, etc. etc.
Always some kind of systematic intelligence is involved.