With half private input noise, PV inhibition lowered detection of a fixed local signal at one cell in two claustrum model motifs

We tested when a released claustrum PV circuit improves detection of a small input change. We used two deposited local motifs, each with 500 projection neurons and 100 PV cells. We made 64 random realizations per motif. The fixed signal had spatial width parameter alpha=2. Half the input noise variance was private; half came from a spatial field. Field lengths had 31 values from 0.001 to 1 mm. Correlation between the projection neuron and PV fields ranged from zero to one. Relative to removal of PV input, the central cell detection score fell in every case. Ratios were 0.57327–0.96583 for the cortical projection motif and 0.85761–0.98700 for the subcortical projection motif. An exact variance equation certified the complete correlation interval at each tested length. Pooling 25 nearby output cells changed the result. At a 0.2 mm field length with the same field driving both classes, median ratios rose to 1.147 and 1.033. An added threshold calculation at 1 mm put the required shared variance fraction near 0.61–0.63 for one cell and 0.065 for the pooled output. Narrower signals also changed the sign. Direct response solves, covariance propagation, nonlinear source dynamics and a clean repeat checked the calculations. The contribution is a conditional model test and threshold tool. It does not measure animal noise or repeat the original animal filtering tests. The package preserves new code, source versions and hashes, parameter facts, seeds, response vectors, full results, figures and repeat instructions.

claustrum · PV inhibition · structure and function · noise correlation · signal detection · population readout · rate model · reproducible research