Fig. 3: Validation of the DSR framework using intracellular voltage recordings.

a Example recording of the membrane potential in a PV neuron on a single trial³². The trace contains spikes and subthreshold voltage fluctuations. b A relationship between the membrane potential and instantaneous firing rate for the example neuron in (a). For each of the 25 ms bins in the recording, we estimate the instantaneous firing rate by the ratio of the spike count to the bin size and plot it against the subthreshold membrane potential averaged over the bin after removing spikes (blue points). Averaging the data points within 1 mV voltage bins reveals a lawful relationship between the subthreshold membrane potential and the firing rate (black points, error bars are s.e.m. over the data points in each voltage bin), which can be approximated with a spline fit (black line). c To independently estimate the spiking irregularity of a neuron, we map its spikes to the operational time using the instantaneous firing rate computed from the subthreshold membrane potential at each time using the fitted voltage-to-firing-rate relationship in (b). The spiking irregularity is CV² of ISIs in the operational time (blue) and accounts for a small fraction of the total variability \({\,{\mbox{CV}}\,}_{{{\rm{raw}}}}^{2}\) of the ISIs in the real time (yellow), which also contains the firing rate variability. For all 8 neurons in the dataset, ϕ estimated with the DSR method from spike times alone (orange) closely corresponds with the spiking irregularity estimated independently from the subthreshold voltage, validating the assumptions of the DSR framework. The dashed line indicates the Poisson irregularity value of 1. d The fraction of the total spiking variability \({\,{\mbox{CV}}\,}_{{{\rm{raw}}}}^{2}\) attributed to the spiking irregularity was similar between the DSR method (y-axis) and the partitioning based on the subthreshold voltage (x-axis). Each data point represents one neuron. For each neuron, the recording was divided into 20 segments of 10 s duration each. Dots indicate the mean of the estimates across these segments, and error bars show the standard error of the mean (SEM). Source data are provided as a Source data file.