EFFICIENT CODING OF NATURAL SCENES IN THE LATERAL GENICULATE-NUCLEUS - EXPERIMENTAL TEST OF A COMPUTATIONAL THEORY

A recent computational theory suggests that visual processing in the r etina and the lateral geniculate nucleus (LGN) serves to recode inform ation into an efficient form (Atick and Redlich, 1990). Information th eoretic analysis showed that the representation of visual information at the level of the photoreceptors is inefficient, primarily attributa ble to a high degree of spatial and temporal correlation in natural sc enes. It was predicted, therefore, that the retina and the LGN should recode this signal into a decorrelated form or, equivalently, into a s ignal with a ''white'' spatial and temporal power spectrum. In the pre sent study, we tested directly the prediction that visual processing a t the level of the LGN temporally whitens the natural visual input. We recorded the responses of individual neurons in the LGN of the cat to natural, time-varying images (movies) and, as a control, to white-noi se stimuli. Although there is substantial temporal Between 3 and 15 Hz , the power of the responses had an average variation of only +/-10.3% . Thus, the signals that the LGN relays to visual cortex are temporari ly decorrelated. Furthermore, the responses of X-cells to natural inpu ts can be well predicted from their responses to white-noise inputs. W e therefore conclude that whitening of natural inputs can be explained largely by the linear filtering properties (Enroth-Cugell and Robson, 1966). Our results suggest that the early visual pathway is well adap ted for efficient coding of information in the natural visual environm ent, in agreement: with the prediction of the computational theory.