On colour: Theory is bunk

<!--[if gte mso 9]> <![endif]--><!--[if gte mso 9]> Normal 0 false false false EN-GB X-NONE X-NONE <![endif]--><!--[if gte mso 9]> <![endif]--><!--[if gte mso 10]> <![endif]--><!--StartFragment-->
Graphic: Bart Rose. Duck-egg blue
Graphic: Bart Rose

The way things seem to be

The best questions about colour are all philosophical but the artworld prefers slogans. It took to Theory like a duck to water. Like the American constitution, Theory is built on dodgy foundations presented as self-evident. We are born equal. We have subjective sense experiences about which we cannot possibly be mistaken. The real world is a speculative domain about which we can never be quite sure.

Ruminations of this sort circulated persuasively in lucid eighteenth-century English before they acquired a life of their own in the European languages that derive their active verbs grammatically from abstract nouns. By the last third of the twentieth century translations back into English had become so exquisitely unintelligible that only art students could be persuaded that a career path lay in committing passages to memory.

The bemused idea that the way things really are is at best a matter of speculation, whereas the way things seem to be is known with certainty, came before postmodernism. Painters educated in the craft tradition that preceded Theory believed that although nobody can say for sure what colour the sky really is, each one of us knows exactly what colour it seems to be. They would paint the upper halves of their canvases with a colour that the sky seemed to them to be, defying all dissenters. “You may see it differently,” they would concede, “but so what? Are you calling me a liar?”

These seemings-to-be that allegedly stand outside any shadow of doubt acquired a variety of technical names such as “phenomenal experiences,” “sense-data,” “sensibilia” and “qualia.” Sense impressions” is perhaps the nearest to plain English. What our sense impressions of colour share with sense impressions acquired through other sensory modes, such as the smooth feel of glass, the acrid smell of ammonia and so on, is a notional location. They are usually assigned a place somewhere inside the perceiver’s head, with their ownership contested between the traditional contenders for proprietorship of this space: the brain and the mind.

Graphic. Bart Rose. Sky Blue/Shades of doubt
Graphic: Bart Rose

Visual sense impressions have been located, more precisely, on the retina at the back of the eyeball in a skein of cells called “colour receptors.” More recently, Neuroscience has nominated a less concentrated “cortical projection area” further along the optic nerve where it has merged with other issues of the brain. But there has always been anxiety about such talk. The trouble is that giving sense impressions a precise physical location suggests that they are accessible in principle to public scrutiny, and this puts their subjectivity at risk. If neuroscientists know precisely where to look for the blue sense impression that a perceiver is experiencing, why can’t they insert a probe that will measure its shade exactly?

According to the story, I can describe the proportions of the rectangular patch of blue that occupies my visual field when I am looking through my window at the sky, but I cannot say just how far in front of me this rectangle is standing. Moreover, precision cannot be achieved by hauling it back inside my head. If I say “It doesn’t stand out in front of my real eye of course, but in front of my mind’s eye,” then I shall need to say where the mind’s eye in my mind’s head is figuratively situated. If I put it figuratively at the front of my mind’s head, beside my real eye, its visual field will be squashed flat.

Stories of a phenomenalistic sort in which sense impressions are subjective and indubitable are told in various ways, all of them have a disastrous implication. If I cannot doubt the colour of my own sense impressions, and if I have no access at all to yours, then a sense impression that you report as blue might not be the one that I would report as blue if I were having it myself. For all I can ever know (because I can’t get inside your head to check it out) a sense impression that you are calling blue might be the very one that I would call yellow if I were having it myself.

Graphic. Bart Rose. Speculation
Graphic: Bart Rose

Theory has a triumphant response to this familiar conundrum. We owe our salvation from the chaos of incommensurable subjectivities to the sovereign power of language. If we have been taught within our shared cultural milieu to apply the name blue to the colour of the sense of impression that we get when we look at the sky, then harmony will prevail. It will be unnecessary, unrewarding – and perhaps even nonsensical – to speculate about the real colours of our sense impressions. Indeed, it will be unnecessary, unrewarding – and perhaps even nonsensical – to speculate about the real colours of those notoriously public things that might be held notionally responsible for causing us to have sense impressions.

But here’s the rub. To say that language rules and that questions of fact are spurious is to say that the actual colour of a sense impression of my own can be no more reliably known to me than the actual colour of one of yours can be known to me – or indeed, to you! The contrast between subjective certainty and objective doubt on which Theory seemed to stand has collapsed in ruins.

We may well ask: how did creatures without and mediating language ever respond to different coloured things in regular and consistent ways? Why do they not all thrash randomly about in a chaos of incommensurable perceptual subjectivities, unable even to learn that the offer of a banana can regularly win friends.

We must think again; this time without the crippling handicap of Theory.

Graphic Bart Rose. Objects bluer than they appear
Graphic: Bart Rose

The way things really are

Imagine that we are naturalistic painters of the old school, standing side by side facing a whitewashed wall about thirty metres away under a bright blue sky. We set up our easels and start work, first by making the upper parts of our canvases look as much like the sky as we can, from where we stand. If perfect simulation were possible the tops and the outer edges of our canvases would disappear against the backdrop of sky. (More will be said about practical imperfection in a moment).

Next, we put our paintings temporarily aside and move on to Experiment 2, in which we shall together set about turning the white wall itself into a picture of the sky. It will be a bigger picture than we usually paint. It will be unusually long and low in its proportions and our station point will be further away than we normally stand when painting pictures. Never mind. We shall rush to and fro with big rollers and trays of the re-mixed colours that will be needed to modify the colour of the wall so that, from our original standpoint, we can barely see the line where the wall ends and sky begins.

Such perfect camouflage will be hard to achieve in practice because of such incidental properties as the texture of the bricks, their mortar lines, cast shadows and shifts of light intensity. In spite of this our goal is clear. We shall push our capacity to discriminate between wall and sky, purely in respect of colour, as far as it will go. As we approach this limit we shall become more inclined to say that, from where we stand, the wall and the sky seem to be the same colour. Or we may say that they now look the same colour, or that they look as if they are the same colour.

These locutions differ slightly – perhaps importantly – in sense, but none of them requires our commitment to Theory’s indubitable sense impressions. There are different ways of admitting that, whether the wall and the sky are really alike or really different in their colours, we find ourselves unable or almost unable to discriminate sensorily between them in this respect. It is a sort of failure that does not turn ultimately on the colour names that we have been taught. It is a sort of failure that must occasionally beset the sensory systems of creatures that have been taught no colour names at all.

Sensory discrimination failures occur for a pair of intimately related reasons. If two things are in fact alike in some respect, then sensory discrimination between them in this respect is logically impossible under any circumstances, not only for a given species but for every species of perceiver. Quite differently: when two things are unalike in some respect the members of a given species may be unable to discriminate between them in this respect, although the members of another species find it easy. Or, differently again: the members of a given species (such as ourselves) may in certain circumstances be unable to discriminate in some respect between things that actually differ, although in more favourable circumstances they can do so.

Patterns of sensory discrimination failure that are regularly shared by the members of a given species are a natural product of evolution, and they must be exploitable by creatures with even the most rudimentary capacity to act purposefully. Among ourselves, the practices of picture-painters and picture-users are a dramatic illustration. A patch of canvas so coloured that, under standardly repeatable viewing conditions, is barely discriminate from the sky is wonderfully apt for use as a representation of the sky. In a similar way, a flatly presented disc is apt for use as representation of the moon.

In the latter case, of course, the discrimination failure is partly attributable to the fact that the representation and its subject are alike in the relevant respect of circumferential shape, but in the former case it is not so. The failure is attributable entirely to the viewing conditions. So-called “resemblance theories” of pictorial representation are often promoted, and often derided, without either party to the dispute attending properly to the distinction.

Graphic by Bart Rose. Landscape painting
Graphic: Bart Rose

Let us move on to Experiment 3, in which we return to our original station point and set up again the pictures that we painted in Experiment 1. Now we shall take into account the way in which we have modified the colour of the wall by adding a representation of it to our pictures, just as we originally painted in our representation of the sky.

The word “representation” has not yet been defined. It must be understood primarily as the purposeful substitution of one thing for another thing, and not primarily as the linguistic signification of one thing by another thing. This latter understanding is one of the heresies of Theory. The ability of active perceivers to achieve useful goals by substituting one thing for another thing and therefore exploiting common patterns of discrimination failure, must have long preceded language. Indeed, without the stepping stone of this fundamental representational capacity the evolution of language itself would be inconceivable.

Long before language emerged, cephalopods such as the octopus acquired the useful trick of representing their backgrounds by modifying their own skin colours, thus exploiting a locally prevalent range of discrimination failure among competing animals, with obvious survival advantages. The angler fish uses its lure to represent items of prey sought by other predators, exploiting the fact that there is a respect – it may or may not be the respect of colour – in which their victims regularly fail to discriminate between a rewarding and a highly unrewarding item of attention.

In evolutionary terms, representational practices must have improved exponentially as their users discovered a capacity to exploit discrimination failures that are not simply attributable to properties shared between a representation and its subject, but also to circumstantially contrived scenarios in which a representation and its subject differ. This is the prime scenario of pictorial representation. Of course, the emergence of language with its powers of signification quickly made it possible to overlay pictures that are dependent upon discrimination failure with layers of implicit text. In spite of this the distinctive feature of picturing, in contrast with linguistically mediated forms of representation, remains what it has always been. What is distinctive about pictorial representation is its exploitation of the effective substitutability of one thing for another that is enabled by our naturally shared patterns of sensory discrimination failure.

Another important point. In the illustration printed here, our knowledge that the various areas representing, respectively, the wall, the sky and of the two pictures of the wall and the sky are all the same colour is not attributable in a logically circular way to our inability to discriminate between them. On the contrary, our inability to discriminate between them is attributable to their being the same colour. It is a colour to which a more precise name can be given than the “pale blue” of common speech. Printers can generate a sample of this colour in one place, and another sample in another place, in such a way that when the two samples are brought together they will prove to be sensorily indiscriminable.

This claim is empirical. Theorists should try it out.

Donald Brook wall diagram
Donald Brook, Experiment 3, diagram

In summary

We live as perceivers in a world in which our evolutionary history – much of it shared with other species – has shaped our ability to exploit natural patterns of discrimination failure for representational purposes. Some of these failures relate to our encounters with things that are alike, while others relate in a more subtly contextual way to things that are as a matter of fact different. The effective substitutability of one thing for another just because our powers of discrimination fail in regular ways is the basic principle of picturing. It relies in no way upon the putative experience of subjective sense impressions; and a fortiori upon their putative indubitability.

A developed language most certainly enables competent users to embellish their pictures with layers of signification, but signification is neither a necessary nor a sufficient condition of picturing. The point is not easy to argue, but it can be stated in just three words: Theory is bunk.


Donald Brook is Emeritus Professor, Flinders University.