Retinal model comparison

We are starting to analyze complete retinal model circuitry. In a preliminary step, we illustrate the consequences of selecting different OS, Bipolar and RGC modeling assumptions.

The script v_rgcModel runs tests of the eight combinations of os (biophys vs. linear), bipolar (rectified vs. linear) and RGC (coupled and uncoupled).

That script produces various visualizations to the response of a RGC mosaic to a moving bar stimulus.

os 0 corresponds to the linear outer segment, and os 1 corresponds to the biophysical outer segment. bipolar 0 corresponds to the linear bipolar mosaic, and bipolar 1 corresponds to the rectified bipolar mosaic.

We will keep exploring. Preliminary observations.

• There are large consequences to imposing rectification on the bipolar compute
• The RGC coupling does not have a big impact on the mean responses to the moving bar; but it may have an impact on the correlations, which have to yet to explore. So, analyzing the correlations is on our list.
• The Biophysical and Linear os models differ significantly, particularly at the onset transients. Some of this is our issue because we aren't sure we got the system perfectly into the equilibrium uniform gray background state. We will add more uniform gray frames at the start of the stimulus to equilibrate the system prior to introducing the moving bar.

JRG and I think we might want to have the option of running the mosaic response to the uniform field and stabilizing the cmosaic. And then for subsequent tests introduce the stimulus starting with the system in the state at the end of the uniform field. This might reduce the need to run the uniform field cases all the time. Not sure that this is right, but we are considering.