I have been thinking and publishing about the ontology of colour for a long time now, and I have slowly made some progress toward a defensible view. But now I have an idea that might bring it all together.
(Gerhard Richter: 1024 colours )
I am a realist about colour. Twenty-five years ago, my primary "reason" was the incredulous stare. "My car isn't blue? Really? What colour is it then? What?—It isn't any colour? Go away: how do I see it then?" Today, I would say something a little different.
There is, however, a disquieting consideration that plays to the anti-realist. "Human colours"—this is my term for red, green, brown, pink, and the rest—have a subject-dependent essence. Red, for example, cannot be physically defined: there is nothing intrinsically red about 700 nm light. It is red only because of its connection to a certain kind of colour-experience. As Newton remarked: "the Rays to speak properly are not coloured."
Moreover, the relations among human colours have no basis in perceiver-independent reality. Nothing can be both reddish and greenish, but this has nothing to do with things themselves for there is nothing in physical variables such as wavelength and reflectance that counts intrinsically as “reddish”. So if reddishness excludes greenishness, this has something to do with how the colours are processed and sorted into such categories, not with the real nature of colours. Stephen Palmer sums it up well: light is of different wavelengths independent of an observer, but color arises only within the observer. (I don't exactly agree with this, as you'll see.)
The issue of perceiver-relativity has come to the fore recently in puzzles about the “unique hues”. Every normal trichromat perceives some hue as “uniquely” green: i.e., as green with no admixture of either blue or yellow. There are similarly unique shades of blue, yellow, and red. Trichromats are consistent in their identification of these unique shades: they identify the same colour sample again and again as the one that is uniquely green. The trouble is that different hues are identified as unique by different perceivers. I may consistently identify 520nm as uniquely green; you may consistently identify 530 nm as so.
When there is this sort of disagreement about unique shades, who is right? There is no good reason to pick on one perceiver rather than another: you and I are equivalently placed with regard to our epistemic access to the colours. Realists such as Michael Tye and Alex Byrne insist that regardless of what you and I may perceive, some one shade is really unique green. But they allow that it is impossible to say which shade is, or even on what the question turns. Their position is tenable, they say: nobody can say whether the number of seconds that have elapsed now since the Big Bang is odd or even, yet it was one or the other and not both. Still realism with respect to the uniqueness of hues is precarious: nobody, including Tye and Byrne, know what it would mean to say that a 525 nm is uniquely green, though some other hue is perceived that way by me and by you. What is it exactly that you and I are getting wrong? They don't know. (Contrast the even/odd question.)
It has always seemed to me best to say that uniqueness turns on how colours are perceived, not on some intrinsic feature of them—a Newtonian line of thought. Each of us has a system that specifies colour by certain qualia. My system specifies the colour that fire engines have as a slightly bluish red. Maybe yours specifies the very same colour as a unique red, or even as slightly yellowish. However that may be, we can each identify and re-identify the colour; we can each base successful inductions on it.
This view is compatible with realism. We see real properties as colours; but our colour vision systems reveal these colours idiosyncratically in terms of red/green, blue/yellow, and light/dark. But this view has a lacuna. It doesn’t answer the question: What is red? Maybe it makes sense to say that I experience a light source or a surface redly, but I don't know what red is.
This is where my new idea comes in.
1. Red is a second-order property of surfaces, lights, and the like.
Surfaces and lights have colour; they are coloured. The colours that such things have are properties of these things: for instance, the colour of my table-top is a property of the table-top. Red, brown, etc.—the "human colours" as I have been calling them—are properties of not of table-tops and the like, but of colours. To say that my table-top is a reddish-brown is to say that the table-top's colour is reddish brown. Further:
2. These properties of colours—i.e., red, brown, and the other "human colours"—are relational. They are properties that colours have relative to colour vision systems.
Note that my view has some completely counter-intuitive consequences. It entails first of all that:
3. If C is a "human colour" (e.g., reddish-brown), then no surface is ever C. (No, my car isn't blue!)
We say that tables and the like are brown, but what we mean is that their colour is brown. My view entails, secondly, that
4. Red is not a colour.
Rather it is a property of certain colours.
And this is where I rest for now.
Recent Comments