GLM as a tool for modeling cellular integration of input. (A) Frozen noise injected into an EP neuron over multiple trials (top). Example of 5 intra-cellular potentials (different colors) of the neuron during repeated injections of the frozen noise (middle). Raster display of the neuronal spiking over the same trials (bottom). (B) Schematic structure of the GLM: Input passing through the stimulus filter is summed with the output spikes serving as the input to the post spike filter and a constant bias term and passing through a non-linear function (exponent) to generate an input to a Poissonian spike generator. (C) Left - The GLM spiking responses to a novel fluctuating current stimulus. From top to bottom: input stimulus, raster plots of spiking activity for repeated stimulus presentations over 50 trials of the experimental neuron (black) and the GLM reconstruction (red), PSTHs of the responses of the real neuron (black), and the GLM response (red). The prediction accuracy is assessed using the PCC between the experimental and model PSTHs. Right - The derived stimulus filter and post spike filter of the fitted GLM. The post spike filters are shown in their exponentiated form. (D) Histogram of the Pearson Correlation Coefficients (ρ) between the PSTH of the neuronal responses to the frozen noise and the PSTH of the model response to the same noise in the GP (red), EP (blue) and SNr (green).