Rava A. da Silveira


After graduating from the University of Geneva, Rava A. da Silveira obtained his Ph. D. in physics from the Massachusetts Institute of Technology. Subsequently, he was a Junior Fellow of the Harvard Society of Fellows. Since 2004, he is a permanent faculty at the Ecole Normale Supérieure, in Paris, where he is affiliated to the Department of Physics and to the Department of Cognitive Studies.

While at MIT, Rava A. da Silveira worked on theoretical problems in statistical physics and soft matter physics. He studied disordered magnets, simple models of fracture and other breakdown processes, surface growth phenomena, and instabilities that occur in elastic and liquid films. At Harvard, he continued working on physics problems, developing theories for the behavior of frictional sand and for the behavior of elastic objects subjected to random forces. At the same time, his work extended to theoretical neuroscience and biophysics – his main interests today. In neuroscience, he studies the statistical properties of brain circuitry and simple aspects of its dynamics, as well as retinal processing of visual information. In biophysics, he investigates chemotaxis and other directed behaviors in simple organisms.


Sophisticated Visual Adaptation in the Retina

The brain is such a complex object that it allows a virtually unlimited choice of measurements, each of which may shed light on some aspects of its rich function. Similarly, the brain's biology comes with such a large number of parameters of potential relevance that theoretical approaches tend to remain underconstrained. Isolating simpler brain sub-systems or organisms with simpler brains may provide a route to the study of the evolved brain in its entirety. One such system, on which exquisitely detailed experiments may be carried out, is the retina – a literal piece of brain that sits in the eye. Many experiments illustrate the fact that the retina, simple as it is in structure, carries out sophisticated computations which turn visual input into a highly processed neural signal sent to the brain.

One of the feats of retinal processing is its ability to adapt to the statistics of visual input. Adaptation to light levels is a commonly experienced phenomenon: coming out of a dark room, one is momentarily blinded by sunlight, but soon the retina readjusts its dynamic range to afford comfortable sight again. Recent experiments demonstrate a far more elaborate ability: the retina adapts to geometrical features present in the visual input, a function that was, as yet, ascribed to the cortex. Spatial correlations that occur frequently in the visual field tend to be underemphasized by the retina, in the messages it sends to the brain, while the reverse holds for underrepresented correlations. We shall describe two competing theories for adaptation to spatial correlation. These rely on well-documented properties of cells and collective effects involving several cells in concert. We shall illustrate predictions of these theories and potential interplay of theory with experiment. Along the way, we shall argue that the biological mechanisms central to our theories may be more generally relevant for perception, beyond retinal processing, and we shall touch upon the role of theory in understanding biological phenomena.