Many otherwise puzzling aspects of the way we see brightness, colour, orientation and motion can be understood in wholly empirical terms. The evidence reviewed here leads to the conclusion that visual percepts are based on patterns of reflex neural activity shaped entirely by the past success (or failure) of visually guided behaviour in response to the same or a similar retinal stimulus. As a result, the images we see accord with what the sources of the stimuli have typically turned out to be, rather than with the physical properties of the relevant objects. If vision does indeed depend upon this operational strategy to generate optimally useful perceptions of inevitably ambiguous stimuli, then the underlying neurobiological processes will eventually need to be understood within this conceptual framework.