Andrea Perna

(paintings by Henri Matisse)

Our brain is able to extract meaningful information about a visual scene from simple sketchs made of only a few traits. In particular, the image positions at luminance edges and lines are those that carry most information.
To some extent, either edges or lines can be used indifferently without affecting the recognizability of the scene.

The visual system processes in parallel the different attributes of visual images. There are visual areas specialized respectively in the analysis of colour, of movement, of brightness. In the striate cortex, neurons are selective for a narrow range of spatial frequencies and a particular image location. Each area will take a different amount of time to process information and it will give an estimation of scene attributes possibly incongruent with that of other areas. Yet we usually are not aware of this multitude of information pathways. We have the conscious experience of a single, stable and unitary visual scene and we respond to it with coherent and in general appropriate motor reactions.

To what extent and how the information processed in parallel by different neurons and different brain areas is integrated into a common estimation of the scene we are seeing? Does our perceptual experience of an image attribute result from averaging of all estimations from different cells or brain areas? Is the averaging linear or is it performed in some different metrics? Or simply one single mechanism determines the final perception?

Both edges and lines are important for the recognition of visual objects, but edges (and only them) play an additional important role in determining the appearance of the surfaces they enclose. This is shown by the existence of visual illusions such as the "Cornsweet Craik O'Brien illusion" represented here. The circle is exactly the same gray level as the rest of the image, but we perceive it as brighter. In some way the brain infers the characteristics of the circle and of the external surface from the high spatial frequency edges.

The different brightness of the circle is not in the image nor on the retina, but it becomes explicit somewhere in the brain. What is the nature of the signals that encodes surface information? Are there neurons in retinotopic visual areas firing in response to the illusory brightness? Or the surface attributes -and the surface itself- is simply ignored by the brain?