Brain Differently Changes its Algorithms in Parallel Processing of Visual Information

Abstract

Feedback from the visual cortex (V1) to the Lateral Geniculate Nucleus (LGN) in macaque monkey increase contrast gain of LGN neurons for black and white (B&W) and for color (C) stimuli. LGN parvocellular cells responses to B&W gratings are enhanced by feedback multiplicatively and in contrast independent manner. However, in magnocellular neurons corticofugal pathways enhance cells responses in a contrast-dependent non-linear manner. For C stimuli cortical feedback enhances parvocellular neurons responses in a very strong contrast-dependent manner. Based on these results [13] we propose a model which includes excitatory and inhibitory effects on cells activity (shunting equations) in retina and LGN while taking into account the anatomy of cortical feedback connections. The main mechanisms related to different algorithms of the data processing in the visual brain are differences in feedback properties from Vl to parvocellular (PC) and to magnocellular (MC) neurons. Descending pathways from Vl change differently receptive field (RF) structure of PC and MC cells. For B&W stimuli, in PC cells feedback changes gain similarly in the RF center and in the RF surround, leaving PC RF structure invariant. However, feedback influence MC cells in two ways: directly and through LGN interneurons, which together changes gain and sizes of their RF center differently than gain and size of the RF surround. For C stimuli PC cells operate like MC cells for B&W. The first mechanism extracts from the stimulus an important features in a independent way from other stimulus parameters, whereas the second channel changes its tuning properties as a function of other stimulus attributes like contrast and/or spatial extension. The model suggests novel idea about the possible functional role of PC and MC pathways.