Perception: 19ECVP abstract

Perception 1996: 19ECVP abstract

Dynamics in parvocellular and magnocellular pathways: Consequences for luminance and colour processing streams
W H A Beaudot (CSEM --- Centre Suisse d'Électronique et de Microtechnique SA,
Jaquet-Droz 1, CH 2007 Neuchâtel, Switzerland;
e-mail: ; WWW: http://wbeaudot.kybervision.net)

An achromatic neuromorphic model of the vertebrate retina has already accounted for X and Y pathways (Beaudot and Hérault, 1994 Perception 23 Supplement, 25) and has shown a temporal `coarse-to-fine' processing of spatial information (Beaudot et al, 1995 Perception 24 Supplement, 93). This model has been extended to colour vision. By taking into account the chromatic sensitivities of cones, functional properties of the parvocellular pathway are modelled. Approximating the responses of colour-opponent cells, the model provides a spatial multiplexing of luminance and chrominance information: sustained responses show spatial band-pass behaviour to luminance variations and low-pass behaviour to equiluminant colour changes. In addition the spatiotemporal inseparability for luminance in the parvocellular model leads to a temporal multiplexing of spatial luminance information: at higher temporal frequencies the spatial filtering is low-pass, conveying only luminance information.

Demultiplexing this mixed information suggests interactions between retinal channels. By locally combining additive and subtractive mechanisms between opposite parvocellular pathways (eg G⁺/R^-±R⁺/G^-), and an inhibition from the magnocellular pathway, the existence of at least three functional subchannels is predicted: (i) a transient, spatially low-pass channel, (ii) a sustained, spatially band-pass channel, dedicated to the analysis of luminance information in a spatiotemporally separable way (eg moving shadows and static textures), and (iii) a spatiotemporally low-pass, colour-opponent channel leading to colour induction, which is little affected by the presence of shadows and is more representative of objects. This hypothesis of spatiotemporal demultiplexing of luminance and chrominance information, which should presumably occur at an early cortical level, is in accordance with the multiple-processing-streams organisation of the primate visual system.