ICCNS 2008

Spiking dynamics of perceptual grouping in the laminar circuits of visual cortex

Jasmin Leveille, Stephen Grossberg, Ennio Mingolla and Massimiliano Versace

Department of Cognitive and Neural Systems, and Center of Excellence for Learning in Education, Science, and Technology, Boston University, Boston, MA 02215

Visual perceptual grouping of collinear image or scenic inducers is thought to rely largely on long-range horizontal connections in layer 2/3 of primary visual cortex (Gilbert et al., 1996) and extrastriate cortex (Von der Heydt et al., 1984). Perceptual grouping data have been simulated by the Boundary Contour System neural models of perceptual grouping (Cohen and Grossberg, 1984; Grossberg, 1984; Grossberg and Mingolla, 1985), which predicted the existence of grouping cells that obey a bipole property whereby grouping can occur inwardly between pairs or greater numbers of similarly oriented and co-axial inducers, but not outwardly from individual inducers. These grouping interactions have been further clarified in LAMINART models that incorporate detailed laminar cortical interactions among identified cells in cortical layers 1, 2/3, 4, and 6 (Grossberg, Mingolla, and Ross, 1997; Grossberg and Raizada, 2000). These models and their refinements have explained and predicted a wealth of psychophysical, anatomical, and neurophysiological evidence. They have not, however, incorporated spiking dynamics to address data that depend on how individual spikes and bursts of spikes are influenced, in particular synchronized, by perceptual grouping circuits. The Synchronrous Matching Adaptive Resonance Theory, or SMART, model of Grossberg and Versace (2007) does incorporate spiking dynamics to clarify how multiple thalamocortical and corticocortical regions can synchronize their dynamics in response to matching bottom-up and top-down signals, but does not incorporate perceptual grouping circuits. The current model begins to unify these two streams of model development. It includes identified V1 cells in Layers 4 and 2/3, connected via AMPA and GABA synapses with axonal propagation delays, that together realize bipole grouping properties in a spiking neural network. Model cells are implemented as two compartments, where one compartment is a dendrite and the other one is a soma governed by Hodgkin-Huxley equations. The model quantitatively simulates data from neurophysiological experiments in monkeys, cats, and ferrets that record from cortical area V1 (18) in response to grouping stimuli, and demonstrates how grouping facilitates fast synchronization of spiking in linked layer 2/3 cells. Supported in part by the NSF (SBE-0354378).