Colours, Categories and the Pursuit of Objectivity

Also available as video or audio.

Are colours real? Are they things that objectively exist in the universe? Or do they exist only in the mind of the beholder?

The answer to all of these questions is, I think, not really. Colours are a human response to complex — but objectively real — features of the physical world. I want to look at what lessons colour can teach us about other kinds of categories that we might hope to use to describe reality.

What is Colour?

The sensation of colour is uncontroversially real: nearly all sighted humans describe having experiences of colour. But where do those experiences come from? It is worth digging in to the science of colour perception, to get to its philosophical implications.

Most human eyes contain three kind of cone cells, which are the kind of nerve cells that respond to colour. These are known as L, M and S cones, for the ‘long, medium and short’ wavelengths that they respond to. It would be tempting to refer to them as red, green and blue cones, these being the three colours of light that can be combined to produce all known colours, but this might be confusing. As the graph below shows, M cells respond most strongly to green light, but also to yellow and orange light; the peak of L cells’ response is actually a wavelength that we would usually call yellow, and the ranges of the two types of cones overlap heavily. Genuinely red light only starts at considerably higher wavelengths than L’s peak response — when the waves get too long to trigger the M cells. S cells respond preferentially to the shortest light waves, known as ‘blue’ or ‘violet’, but they also respond to a range of wavelengths shared with both of the other two cones.

Image from the colour vision entry on Wikipedia

Most light includes waves of many different wavelengths; this is why you can split white light up into different colours using a prism. There is no one wavelength that corresponds to purple light, which is why you’ll occasionally hear people claim that purple is not a colour. It absolutely is, it’s just not a spectral colour, a colour of the rainbow. It’s the colour (or group of colours) that you get when short-wavelength and long-wavelength light combines.

Brown, on the other hand, is arguably not a colour at all — or anyway, not a colour that light or objects can ever really be. That’s because the colour that we perceive arises from a complex interaction between the cone cells the light is focused on and the ones around it, as well as our expectations of the scene and what other light we have been exposed to recently. “Brown is always a dark color next to a lighter background” according to the physicist Richard Feynman in his lecture on colour perception, and it is easy to show that objects we see as a muddy orange against a dark background are understood as brown when given light-coloured surroundings.

The Dress, one of the internet’s most famous photographs.

Anyone who was on the internet in 2015 will probably remember The Dress, an extreme example of the power of expectations in determining perceived colour. A photo was circulated of a dress which had either blue and black stripes, or white and gold stripes. The question proved fascinatingly divisive, with many people firmly in one camp or the other, while some found that they could switch the way they read the photo depending on things like the brightness of their monitor and what else was around. Out in the physical world, the stripes were unmistakably blue and black, but if you assumed that it was poorly lit under a blueish light in the photo, the white-and-gold interpretation made sense.

xkcd explains; further explanation

Apparent colour can vary depending on lighting in surprising ways. ‘Green’ plants often look red under ultraviolet light, for example. I once wore a grey hat to the London Planetarium, only to find that (unlike other grey objects) it was clearly green under their lights. I used to have a green tent and a green jumper, but when I sat in one wearing the other, it looked orange! It’s useful to be able to talk about grey hats and green jumpers, but colour is not a property of objects in a straightforward way.

It is not just context and expectations that determine perceived colour: it turns out that some people’s cone cells are just tuned a bit differently from others. People have long wondered whether their experience of red is the same as other people’s, and it’s an open philosophical question. Either way, the fact is that whether a given colour is perceived as yellow or green depends on which version of the M and L cone cells somebody has inherited. We call some people ‘red-green colourblind’ because their M and L cones respond to wavelengths that are closer together than they are for other people, but even among people considered to have ‘normal’ colour vision, there is significant variation in the responses of the photopigments in their cone cells. It’s not clear that any one version of any cone is objectively superior to the others, but there are certainly some configurations that are more convenient for the society we live in.

Some people lack one type of cone entirely, rendering the whole concept of red (for example) meaningless to them, except as something other people describe. Some people (tetrachromats) seem to have four kinds of cone cells, allowing them to distinguish colours that the rest of the world views as identical — a superpower, perhaps, but to what advantage in a society that doesn’t recognise these distinctions?

Here is a flag planted in front of the blinding light of the sun. Clouds around the sun are quite often colourful like this, but you usually can’t see it because they’re too bright. Wait… does even still count as colourful if it’s so bright it looks like blinding white to the naked eye?

Constructing categories critically

Colour occupies, at bare minimum, a three-dimensional space (we have some flexibility as to which three dimensions: they could be red, green and blue; cyan, magenta and yellow; hue, saturation and lightness; yellow-blue, red-green and brightness). Efforts to name parts of that space can be useful, but inevitably they are somewhat arbitrary, even before we account for the variability of cone cells — where does blue end, and green begin? How many other colours should we name in between? Is turquoise a kind of green? Is pink a kind of red?

It seems to be part of human nature to categorise — at the very least, it is a necessary part of verbal communication, and the thinking that underlies it. The real world doesn’t always work that way, though! By naming things we are often imposing divisions on a multidimensional landscape. We pass them through a distorting, simplifying filter to make them easier to talk about. In biology, arguments rage on about where one species ends and another begins, or how many levels there should be in the hierarchy of taxonomy. People observing the same objective, measurable facts about the external world devise different models to describe these facts, and reach different conclusions.

Similar debates crop up a lot in discussions of identity, where (for example) terms like ‘gay’ or ‘white’ may or may not be considered specific enough, depending on the person and the context. They may also mean different things to different people. Things like race, gender and sexuality are socially constructed, which doesn’t exactly mean that they aren’t real, and certainly doesn’t imply they’re not meaningful; money is socially constructed too, but its effects on the world are enormous.

What social construction means is that the way we understand and categorise things is to some extent based on agreements between people, which may be more or less arbitrary. Race, gender and sexuality all relate to objectively observable features of people, but as with colours, there is no one correct way to carve up and label the collections of features underlying these supposed categories. The concept of race is an unscientific mess, with no racial categories holding up to close inspection, but racism is still very much real. Sex is widely assumed to be binary, but it, too, is largely socially constructed: we can definitively test whether someone has Y chromosomes, or produces eggs, or has high testosterone, but none of these map straightforwardly onto how people understand the categories ‘male’ and ‘female’. None of them are even considerations when medical professionals tick one box or the other after birth.

In psychiatry, a taxonomy of human variation is codified in diagnostic manuals. The discipline is considered part of medicine and has a veneer of objectivity, but its categories are generally based on human judgements of behaviour considered to be a problem. Few of these have well-understood neurological bases; those that do not are purely socially constructed, by a small group of qualified professionals. It is extraordinarily difficult to make sense of the relationships between these categories: Similar behaviours might arise from different neurological, cognitive and social sources; superficially different behaviours might share the same cause, on one or more of those levels.

If what we are interested in is how people experience the world, or what gives rise to these experiences, psychiatric categories are not necessarily the best guide. Psychiatrists, after all, are doctors in the business of identifying and treating psychopathology; if something can’t be understood as a syndrome or condition, then — profesionally speaking — they are just not interested. This approach has butted up against the fact of neurodiversity in recent decades: human minds just do vary a great deal, and it is only by judging them against some invented standard of ‘normality’ that variations can be judged to be pathological.

The accepted symbol for neurodiversity is an infinity symbol in rainbow colours. Image adapted from one by Mrmw.

Where does this leave the validity of psychiatric categories, which include mental illnesses as well as neurodevelopmental differences (or neurovergences) like autism, ADHD and dyslexia? It certainly doesn’t mean that people sorted into these categories are not really different: just because there is no clear line between green and blue, that doesn’t mean that green and blue aren’t different! Even speakers of languages with no word for turquoise are capable of distinguishing it from ultramarine.

What we do need to bear in mind is that categorising human differences is far more complex and difficult than categorising colours, and we haven’t been at it for nearly as long. When even something as seemingly basic as colour hides so many complications, it is no surprise that the human mind is tough to categorise decisively!

Labels like ‘autism’ and ‘ADHD’ have been useful in identifying people who need particular kinds of support, and more useful still in helping people to find community; but there is no reason to think that the current definitions of these things are final. The ‘Autism Spectrum Disorder’ category in current diagnostic manuals combines several previous categories, for example. It has been constructed from the outside, without the benefit of internal insight or neurological observations. It may be sub-divided again in future, or it may instead swallow more categories still. Some, especially whose experiences straddle those associated with multiple diagnoses, prefer to self-identify as neurodivergent, or monotropic — rejecting the formal diagnostic categories.

The world is complex, and humans are some of the most complex things in it. Language has a way of putting things in boxes, however much richness might be lost that way — still, at least it allows us to talk about things! Do we really need the term ‘ecru’? I have no idea, but, well… what else would you call it?

Wikipedia on ecru




Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store