Colour Vision: Primary Visual Cortex Shows Its Influence Dispatch

In a landmark statement, Crick and Koch [1] argued against primary visual cortex (V1) being the site for consciousness, citing the fact that V1 neurons did not display the properties necessary for colour constancy, a cornerstone of our conscious experience. Colour constancy is what makes the redness of a red apple stable under changing illumination. All proposed mechanisms for colour constancy require comparisons between lights reflected from different surfaces of the scene, to discount the overall effects of the light source spectrum. The general view until recently has been that neurons capable of making such comparisons do not occur until area V4 or beyond, where the receptive fields are sufficiently large [2]. This view is now challenged by new studies which have revealed that V1 neurons may be influenced by light outside their receptive fields [3,4] and that, within their receptive fields, they perform important steps in computing colours [5–7]. In his pioneering study twenty years ago, Zeki [8] specifically searched for cells whose responses mirrored our perception of colour: cells which, for example, would continue to signal ‘red’ as long as a surface appeared ‘red’, regardless of the illumination upon it. He found such cells only in V4, and termed them ‘colour-only’, as opposed to the ‘wavelength-only’ cells in V1, whose responses changed with changing wavelength, regardless of the perceived colour. In a new search, Wachtler et al. [3] instead looked for neurons whose responses were influenced by changes in the background colour, well outside of their classical receptive fields [9]. By definition, stimuli presented on their own outside the classical receptive field provoke no response from the cell, but, crucially, they may modulate the response to stimuli within the classical receptive field. This non-classical ‘contextual modulation’ has been demonstrated for a variety of visual attributes, for example, binocular disparity and texture orientation [10], and now luminance [4] and colour [3]. Colour contrast is, of course, a prime example of contextual modulation in perception: a bluish disk against a blue background appears less blue than when against a grey background (Figure 1, left). Colour contrast is also one of the most basic mechanisms by which colour constancy may be achieved, because factoring out the colour of the background is likely to factor out the illuminant colour. The latest search for colour-constant neurons has revealed a population of V1 neurons that report colour contrast, over a large scale [3]. These cells display several properties. When presented with a large patch of uniform colour entirely covering its receptive field against a grey background, each has a preferred colour but also responds well to nearby colours, some with tuning widths as broad as half the entire hue wheel. For most cells, a background of the preferred colour suppresses its response to the preferred colour, as well as, less strongly, to nearby colours. For some cells, a background opposite in colour to the preferred colour enhances its response. (To ensure that the background effects were not due to stray light falling in the receptive field itself, Wachtler et al. [3] adjusted each stimulus to be at least twice as large as the classical receptive field, and also verified that the background changes alone did not activate the neuron.) Thus, a cell that fired vigorously in response to a bluish patch against a grey background would respond less well when confronted with the same bluish patch against a blue background, as if it were signalling the reduction in chromatic contrast — or correcting for an overall bluish cast of the illumination, thereby reserving its best response for true-blue surfaces. Individual cells do not perfectly encode chromatic contrast: on average, the reduction in any one cell’s response is about two-thirds the reduction in contrast. Yet, over the whole population of 94 cells tested by Current Biology, Vol. 13, R270–R272, April 1, 2003, ©2003 Elsevier Science Ltd. All rights reserved. DOI 10.1016/S0960-9822(03)00198-2