PMC

A) Unified Modeling Language (UML) representation of the classes that implement the encoding and decoding analysis methods and store the relevant results. B) UML diagram of the remaining classes within nSTAT.

Time varying stimulus response. The within-trial stimulus is modulated by the across-trial stimulus gain in order to simulate neural plasticity across trials. . The simulated neural raster is generated via the time rescaling algorithm described in Section2.1.4 using the nSTAT CIF class. The true conditional intensity function is obtained from .

This particular example shows the typical workflow in the testing of candidate conditional intensity functions. For brevity, the workflow for decoding stimuli using the Point Process Adaptive Filter or the Hybrid Point Process Filter is not shown (see the Matlab help files for these examples for the corresponding code).

A) nSTAT results summary for the constant baseline firing rate model () for the mEPSC data in ). (top, left to right) original neural raster, autocorrelation function of the u_s’s, K-S Plot, and conditional intensity function estimate. B) (top, left to right) Raster of mEPSC data under varying magnesium concentration, autocorrelation function of the u_s’s, K-S Plot and CIF estimate. Note that the piecewise constant rate model yields a K-S plot that falls within the 95% confidence bands and produces rescaled event times whose autocorrelation function is closer to zero across all lags – suggesting independence of the rescaled times under λ_CONST-EPOCH. The estimated CIF (lower right panel) is able to capture the increased rate that is observed in ).