Verbal descriptions, photographs and maps are reviewed for the kind of selective information they impart (§1). The need is established for a critical analysis of the assumption that we can not only map the invariant features of the physical world but also represent the optical world, the changing appearance of objects as it is conveyed to the camera and to the eye by reflected light. As long as this light was identified with the stimulus pattern causing visual sensations it could be assumed that appearances were uniquely determined and could thus be uniquely represented, but this view conflicts with the insight that there are many other variables influencing our visual experience. Visual sensations as such cannot be isolated by introspection but they can be aroused and manipulated by artifice based on the knowledge of the physiology and psychology of visual perception (painting, stereoscope, film, television). The surprise caused by such unexpected visual effects underlies the notion of illusion. Artistic experiments in registering and arousing visual sensations are discussed (§ 2). These experiments show the need to give subjectivism its due without falling into the trap of complete relativism. It is here that the consideration of maps and mapping styles is helpful. The keys and symbols adopted by map makers suggest that visual conventions can but need not rest on arbitrary choice and are rarely devoid of psychological effects. These effects, however, are independent of the truth or falsehood of the information compiled by surveying instruments of any kind. It is possible to predict what aspect of a physical array will be visible from any given point in space (§ 3). The theory underlying this prediction is that of central perspective based on the ‘visual cone’. The procedure is not reversible, the information imparted by a perspectival representation does not uniquely determine the object represented. This multivalence of monocular stationary vision (the tracing on the window pane) has given rise to many psychological puzzles concerning the determinants of appearances such as the constancy phenomena (§ 5). More recently J. J. Gibson has challenged the relevance of these puzzles and experiments and emphasized the resources afforded by the ambient light for the veridical perception by a moving organism of the invariant environment. This challenge has created a fresh problem for the theory of pictorial representation (including photographs) and made it more urgent to investigate the visual experience aroused by such representations in varying conditions (§5). A comparison between the information conveyed by pictorial representations and the information picked up by the eye inspecting a real scene may provide opportunities for testing Gibson’s account and help to clarify the limits of veridical perception. Probing reactions to distant prospects in reality, paintings or photographs may reveal that the experienced stability of their appearance can be upset. Even the clouds in the sky and the vault of heaven are subject to various perceptual interpretations which rarely impinge on our awareness. Far from justifying a representational relativism these variations confirm the need for an anchorage of representation in the objective procedures of perspective (§6). It is suggested that the indeterminacy of visual interpretation disposes of the time-honoured problem of the apparent curvature of the phenomenal world. Demands for an alternative system of perspectival representation which appeal to this experience rest on a confusion between the mirror and the map. We can map the physical world but not its variable and shifting appearance. This conclusion, however, is not intended to discourage artistic attempts to record a visual experience. On the contrary: all experiments on the hoardings, on the screen and in paintings probing our cognitive and emotional response to images should be of interest to the student of human reactions (§7).