Towards a cognitive definition of colour vision

In recent years, colour vision abilities have been rather generously awarded to vari-ous invertebrates and even bacteria. This uncertainty of when to diagnose colour vi-sion stems in part from confusing what colour vision can do with what it is. What col-our vision can do is discriminate wavelength independent of intensity. However, if we take this as a definition of what colour vision is, then we might indeed be obliged to conclude that some plants and bacteria have colour vision. Moreover, there is a simi-lar confusion of what are necessary and what are sufficient mechanisms and behav-ioural abilities for colour vision. To humans, seeing in colour means seeing an image in which objects/lights have chromatic attributes - in contrast to the sensation that we have when viewing monochrome movies, or our experience in dim light when only rod vision is possible. The necessary basic equipment for this is to have at least two types of photoreceptors that differ in spectral sensitivity, and at least one type of spectrally opponent cell to compare the signals from the photoreceptors. Clearly, however, a necessary additional prerequisite for colour vision is to have vision, which entails the identification of shapes, sizes and locations of objects in the world. Thus if an animal has colour vision, it should see an image in which distinct objects/lights have colour attributes. This distinguishes colour vision from what has historically been called wavelength-specific behaviour: a type of behaviour triggered by fixed configurations of spectral receptor signals; however, we discuss difficulties in diagnosing wavelength specific behaviour as an indicator of the absence of colour vision.