Abstract
When we match two surfaces' colours, are we primarily matching the light coming from those surfaces or are we primarily matching the colour contrast with the background? To find out, we asked eight subjects to set the colour and luminance of a 2 deg diameter test disk on a computer monitor to match a 2 deg diameter grey, reddish or greenish reference disk. Moving the computer mouse to the left or right decreased or increased the extent to which light from the test disk stimulated l-cones, and pushing it away or bringing it nearer increased or decreased m-cone stimulation. The disks were presented on various backgrounds. We assume that the variability in performance is proportional to the amplitude of the underlying signal. Thus, if subjects primarily match the colour contrast then they should be most accurate when the target and background are the same colour. This was indeed so (ignoring errors in luminance). If subjects had primarily matched the light from the targets themselves they would probably always have been most accurate for grey surfaces (smallest colour opponent signals) irrespective of the background colour. In fact, subjects were equally accurate for greenish or reddish disks on a grey background as for grey disks on a greenish or reddish background. However, similar matches with more complex backgrounds and with different colours near the two disks show that subjects do not just match the colour contrast at the borders between the disks and the background.