In the early visual system of the cat, the feedforward pathway going from the lateral geniculate nucleus (LGN) to the primary visual cortex (V1) is well characterized both anatomically and functionally. On the other hand, despite a large amount of experimental work, there is poor agreement about functional roles of the feedback pathway from V1 to LGN. The common experimental approach compares system responses taken in the​ open-loop​ condition-probing thalamus in isolation from cortex-and in the​ closed-loop​ condition-probing the intact system. In the literature to date this approach has yielded inconsistent results. Open-and closed-loop results for some stimuli differ only by an additive factor, while for other stimuli there is a marked qualitative difference in the results.

In this computational study we investigated the thalamocortical loop with reference to an unprecedentedly broad set of experimental studies. We took published data for six types of experiments (with stimuli varying in luminance, contrast, spatial frequency, temporal frequency, size, and orientation) involving both open-and closed-loop conditions. We explored each condition with an integrative large-scale spiking model of the cat early visual system that includes LGN, peri-geniculate nucleus (PGN), and V1. The model is heavily constrained by the available literature at multiple levels: anatomical, functional, and statistical. The model is developed using the Mozaik workflow engine [1] with NEST [2] as backend simulator.