Abstract

In this thesis, I show how decisions about the ontology of colour depend upon the empirical and conceptual relations among levels of explanation for vision. In Chapter 1, I show how the "received" Lockean view of colour is linked to Newton's theory of light and colour. In Chapter 2, I review extensively recent biological, psychophysical, and computational models of colour vision, and I discuss their relations. I also show how the ontological status of colour is linked to these levels of explanation for the specific case of colour constancy. This empirical discussion provides the foundation for Chapters 3 and 4 where I argue that both objectivism and subjectivism are untenable on extensive empirical grounds. ;In Chapter 3, I argue that the computational level of explanation is not sufficient to ground an objectivist view of colour. Colour cannot be identified with surface spectral reflectance, nor can our present conception of colour be replaced by a new concept of physical colour as surface reflectance. An adequate account of colour must, then, be experientialist. I conclude this chapter by showing how arguments against objectivism are not sufficient to establish subjectivism. ;In Chapter 4, I present a novel philosophical argument based on the comparative study of colour vision in different species. This body of research has been largely neglected both by philosophers and by those who work on computational colour vision. My "comparative argument" purports to show that an adequate account of colour must be both experientialist and ecological. Whereas computational objectivism runs afoul of the experiential nature of colour, neurophysiological subjectivism runs afoul of the ecological nature of colour experience. In contrast to both objectivism and subjectivism, I defend a new view of colour, which I call ecological experientialism. According to this view, colour corresponds to a type of property that is interactional, i.e., to a type of property that results from the biological and ecological interactions of perceiving and cognizing animals. Therefore, to account for colour we must state generalizations over the ecologically embodied visual experiences of perceiving animals.