Spatial coding and response redundancy in parallel visual pathways of the marmoset Callithrix jacchus

• Published in: Visual neuroscience Vol. 22; no. 4; p. 479
• Main Authors: Forte, Jason D, Hashemi-Nezhad, Maziar, Dobbie, William J, Dreher, Bogdan, Martin, Paul R
• Format: Journal Article
• Language: English
• Published: England 01-07-2005
• Subjects: Action Potentials - physiology; Animals; Callithrix; Geniculate Bodies - cytology; Models, Neurological; Neurons - physiology; Orientation; Photic Stimulation - methods; Visual Cortex - cytology; Visual Pathways - physiology; Visual Perception - physiology
• ISSN: 0952-5238
• DOI: 10.1017/S0952523805224094

Many neurons in the primary visual cortex (area V1) show pronounced selectivity for the orientation and spatial frequency of visual stimuli, whereas most neurons in subcortical afferent streams show little selectivity for these stimulus attributes. It has been suggested that this transformation is a functional sign of increased coding efficiency, whereby the redundancy (or overlap in response properties) is reduced at consecutive levels of visual processing. Here we compared experimentally the response redundancy in area V1 with that in the three main dorsal thalamic afferent streams, the parvocellular (PC), koniocellular (KC), and magnocellular (MC) divisions of the dorsal lateral geniculate nucleus (LGN) in marmosets. The spatial frequency and orientation tuning of single cells in the LGN and area V1 were measured, using luminance contrast sine-wave gratings. A joint spatial frequency-orientation response selectivity profile was calculated for each cell. Response redundancy for each population was estimated by cross-multiplication of the joint selectivity profiles for pairs of cells. We show that when estimated in this way, redundancy in LGN neurons is approximately double that of neurons in cortical area V1. However, there are differences between LGN subdivisions, such that the KC pathway has a spatial representation that lies between the redundant code of the PC and MC pathways and the more efficient sparse spatial code of area V1.