Physiologically based simulations and models of feedback in the visual system

The significant progress of the past few decades in mammalian visual system research has focused on the “feedforward” pathway: retina to lateral geniculate nucleus (LGN) and then through the hierarchy of cortical areas. Abundant “backward” or feedback connections also exist but have been relatively neglected. Recently an important experiment by Sillito et al. (1994) [Nature 369:479–821 showed that cortical feedback can synchronize LGN cells, demonstrating a previously unknown and potentially powerful cortical influence on the LGN. How the cortex causes the synchronization was not clear, however, and this thesis reports on analysis, modeling, and simulation studies to suggest possible physiological mechanisms. Analysis of the Sillito et al. (1994) data revealed bursts which strongly contributed to the correlation. A physiologically based model, using low threshold (LT) calcium channels and a hypothetical excitatory/inhibitory (e/i) structure of cortical feedback, replicated the experimental results in detail. Simulations without either the LT channels or e/i failed. The model passed an unexpected test when correlations of anomalously long duration that were found in the experimental data (Brody, 1997) [Ph.D. thesis, Caltech] were also found in the model. An investigation showed that LT channels and a spindle-like oscillation were responsible for the model\u27s long duration correlations. Another feedback model, using stimulus specificity and a modified e/i structure, was used to simulate experiments by Motter (1994) [J. Neuroscience 14:2178–89; 2190–9] on attention modulation in cortical area V4. The model reproduced the experimental results, although some discrepancies indicated that the model\u27s single-layered cortical networks were too simple. The importance of the e/i feedback structure in these models is complemented by increasing experimental evidence for its existence, and further experimental and theoretical work on its nature and function would be worthwhile to pursue