This is a very counterintuitive conclusion, but it is well supported by the evidence – in contrast to the embedded assumption about visual perception in our language – ‘naive realism’. A statement like ‘the grass is green;’ implying a correspondence between intrinsic properties of grass and our perception; is scientifically misleading. But if I insist on expressing my belief with the phrase ‘my perception of the grass is green,’ our conversation may not go well. So, to facilitate conversation, we resort to  expedient shortcuts at the expense of accuracy. Philosophers delight in pointing out where this habit leads us astray.

Now to return to the topic of colour: we might ask, ‘what functional advantage determined the evolutionary success of trichromatic colour vision in humans and a very few primates?’ The simplest explanation is that achromatic object perception; or shape recognition based entirely on luminance clues; is less capable of generating unified perceptions. For example, imagine a theoretical potential evolutionary ancestor, without colour vision. She is looking through a thicket of tree branches toward a forrest, within which a hungry leopard hides. Her achromatic (grey scale) analysis of the scene did not enable her to discern ‘likeness’ across discontinuous parts  of the scene, so she could not generate a perception of a leopard. Hence extinction. Compare that with a theoretical ‘successful’ ancestor with genes that enabled trichromatic perception, providing her with the capacity to discern a colour similarity between the disconnected areas of ‘leopard’ that were visible through the trees. She produced a unified perception of a leopard hiding and had a higher chance of passing on her genes – an instance of ‘natural selection.’

A sophisticated interpretation of the environment began before the capacity for language had developed. It is likely that our evolutionary psychology, which led to our capacity to assign categorically unifying values to arrays of perceptual stimuli, made it useful to assign colour categories to visual perceptions since unifying perceptions according to colour helps us to form reliable beliefs about the world. Once language and social networks developed we learnt to communicate our beliefs more efficiently than other animals, and to pay more attention to differences between colour categories. It is a historical anomaly that 
Newton’s rather illogical ‘colour octave’ has become normative, since it fails to accurately express our  categorical colour perceptions – as evidenced by different menus of colour names across cultures. The ‘idea’ of colour supervened on spectral contrast detection as a way of thinking and talking about the perception. The aesthetic appreciation of colour is a cultural rather than a physiological skill. 

The problems we have in discussing our colour perceptions in light of our scientific understanding of the world are a specific and especially clear example of a wider problem relating to how we talk about consciousness. ‘Colour conversation’ clearly illustrates how naive realism and empirical facts about the world come apart when analysed scientifically. Whenever this happens the smart money is on the scientific view of reality. Unfortunately changing the way we talk about consciousness is a social privilege that must be earned. Even if linguistic pedantry was an acceptable mode of discourse, truth making in public is frequently denied to the cognoscenti. So grass is not about to lose ‘its’ greenness.