Abstract accepted at ARVO Meeting 1997


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; fax: +41 38 205 763;
e-mail: ; WWW: http://wbeaudot.kybervision.net)

Purpose. This work models the temporal dynamics of luminance and colour perception in the primate visual system, with the aim of revealing their underlying neural basis.

Methods. Temporal dynamics along the retino-cortical pathway are predicted by a neuromorphic modelling of primate retina: spatial, temporal and chromatic aspects of retinal processing are taken into account based on electrical coupling among neurons, leaky integrator for membrane properties, and spectral sensitivities of cones responses. Synaptic interactions in OPL and IPL provide a model of signal processing performed by P and M retinal pathways through spatiotemporal and chromatic transfer functions.

Results. Approximating the responses of colour-opponent cells, the model shows a spatial and temporal multiplexing of luminance and chrominance information. However, behavioral separability of pattern and color perception suggests an early demultiplexing of this information. By locally combining additive and subtractive mechanisms between opposite P pathways (e.g., G+/R-±R+/G-), and an inhibition from the M pathway, the existence of at least three functional subchannels is predicted by the model: (i) a transient, spatially low-pass channel, (ii) a sustained, spatially band-pass channel, dedicated to luminance analysis in a spatiotemporally separable way, and (iii) a spatiotemporally low-pass, colour-opponent channel.

Conclusions. The early cortical demultiplexing of P signal predicts (i) a ``Coarse-to-Fine'' spatial processing for luminance information due to the intrinsic spatiotemporal inseparability in the retinal model, (ii) a ``Fine-to-Coarse'' spatial processing for colour information, and consequently (iii) a ``Luminance-to-Colour'' processing for low-pass spatial information. Further experimental investigation is required to test the plausibility of these predictions.


© 1996 CSEM S.A.
E-mail: