Friday 9 November 2007

Why colours look the way they do

I went to a fascinating paper by Nick Unwin on ‘Why colours look the way they do’ yesterday evening but - or should that be ‘and’? - have been left baffled.

Colours have been taken by philosophers to be the best examples of qualia and hence often described as ‘nomological danglers’ because, so the thought goes, they dangle free of the physical laws. But in fact there’s been much work to explain colour vision and thus, as Nick put it, to take some small steps towards bridging the explanatory gap between our account of the physical world and the phenomena of consciousness. (He recommended Hardin’s Colour for Philosophers as a good start on this.)

One expression of the idea that colours are nomological danglers is that they might be inverted. For all we know, other people might see red where we see green and so on. If the inversion were systematic, it would not be detectable in the judgements they made about the colours of objects. But as Hardin and others have argued, there are in fact constraints on this because the relations between colours (including non-spectral colours like brown) are complex. Not just any inversion would work. In his paper, Nick argued that the ‘warmth’ of some colours would make inversions of red and blue detectable but that it might still be possible to mirror colours across the ‘warm-cold’ axis. He then discussed, in a fairly open ended way, whether this was conceivable, by looking, eg, at way shaded green grass would have to be translated into some forms of purpley pink (I think) and creating pictures of this.

What confused me was this. In the project of eliminating such sceptical possibilities as inverted colours what is needed are internal relations within the colour system which make no other systematic mapping possible. Indeed Nick called for the development of just such a pure phenomenal language. (Connections which are merely associative - between the colours and, say, thermally warm things - do not have anti-sceptical effects because a colour inverted subject might simply have different associations. Warm fires might look green to her.) But such internal features of colours do not seem to help much in explaining how colours are part of the natural world: closing the explanatory gap. Articulating such relations seems to be more describing how colours, as a whole, look than why they look as they do.

To put it another way, the project of eliminating scepticism here seems to require that there are sufficient internal, non-associative features of colours. But it seems odd that the project of explaining colour appearance (a project which includes discussion of the rods and cones of the eyes and of opponent processing of those outputs) should be held hostage to the fortunes of a particular anti-sceptical argument. In other projects of naturalising phenomena, scepticism doesn’t get a look in, let alone pride of place.

But as Nick pointed out, an explanation should be contrastive. It should explain why things are one way and not another. So, unusually perhaps, he argued that explaining colours does require this detour. That seems to me to be such an odd result that I’m left baffled.