UC San Diego
Surround processing in the awake rodent dLGN
- Author(s): Sriram, Balaji
- et al.
The early visual system is responsible for encoding a complex spatiotemporal pattern of light. It transform this pattern into spikes which can then be read by downstream neurons. What strategies should neurons employ while transforming the input pattern? In the best of all possible worlds, neurons encode only information that other nearby neurons haven't already encoded leading to a highly sparse highly non-redundant coding scheme. What receptive field features lead to such a sparse, non- redundant coding scheme. For linear receptive fields, the presence of a classical surround correlates well with the response sparseness and de-correlatation. I measure the strength of this surround in the rodent dLGN and find that the surround strength in the un-anesthetized rodent is optimized for maximal sparseness and decorrelation for a majority of cells. Apart from the linear receptive field, neurons in the dLGN of many species are known to have powerful non-linear processing. Further, multiple response features of dLGN neurons have been attributed to these non -linear effects. I show that a simple linear model is capable of explaining many of these features. However, I identify multiple other response features that linear models are inherently incapable of explaining. I show that rodent dLGN neurons show atleast some of these features