A Large-Scale Model of Spatio-Temporal Patterns of Excitation and Inhibition Evoked by the Horizontal Network in Layer 2/3 of the Visual Cortex

Abstract

Cortical processing of even the most elementary visual stimuli can result in the propagation of information over significant spatiotemporal scales. To fully understand the impact of such phenomena it is essential to consider the influence of both the neural circuitry beyond the immediate retinotopic location of the stimulus, including pre-cortical areas, and the temporal components of stimulus driven activity that may persist over significant periods. Two computational modelling studies have been performed to explore these phenomena and are reported in this thesis. I) The plexus of long and short range lateral connections is a prominent feature of the layer 2/3 microcircuit in primary visual cortex. Despite the scope for possible functionality, the interdependence of local and long range circuits is still unclear. Spatiotemporal patterns of activity appear to be shaped by the underlying connectivity architecture and strong inhibition. A modelling study has been conducted to capture population activity that has been observed in vitro using voltage sensitive dyes. The model demonstrates that the precise spatiotemporal spread of activity seen in the cortical slice results from long range connections that target specific orientation domains whilst distinct regions of suppressed activity are shown to arise from local isotropic axonal projections. Distal excitatory activity resulting from long range axons is shaped by local interneurons similarly targeted by such connections. It is shown that response latencies of distal excitation are strongly influenced by frequency dependent facilitation and low threshold characteristics of interneurons. Together, these results support hypotheses made following experimental observations in vitro and clearly illustrate the underlying mechanisms. However, predictions by the model suggest that in vivo conditions give rise to markedly different spatiotemporal activity. Furthermore, opposing data in the literature regarding inter-laminar connectivity give rise to profoundly different spatiotemporal patterns of activity in cortex. 2) The second computational modelling study considers simple moving stimuli. These stimuli are implicated in the 'motion streak' phenomenon whereby the movement of a visual feature can give rise to trajectory information that is not explicitly present. Published experimental data of an in vivo study in the cat has shown that a single small light square moving stimulus elicits activity in populations of neurons in primary visual cortex that are selective for orientations parallel to stimulus trajectory (Jancke 2000). In more recent, unpublished data, this work is extended to consider long term persistent cortical activity that is generated by similar stimuli. These data indicate that following initial cortical activation that appears to result directly from the stimulus, iso-orientation domains display persistent activity. Furthermore, initial activity is broadly tuned with respect to orientation whilst later activity is strongly selective for orientations that are parallel to the stimulus trajectory. Currently the generative processes involved have not been clearly defined. Hence the proposed thesis will contribute to a more complete understanding of the mechanisms responsible for such cortical representations of moving visual stimuli. More specifically this will be achieved by a large scale mean field model that will enable a thorough investigation of the anatomical and electrophysiological elements concerned with the observed spatiotemporal dynamic behaviour and will represent a significant region of cortex. In conjunction, an existing computational model of the retina will be integrated. In doing so this thesis will offer the notion that certain cortical representations are inextricably linked with earlier stages of the visual pathway. As such consideration of retinal processing is fundamental to the understanding cortical functions and failure to do so can only result in erroneous conclusions.