Interpretation of spiking variable S

When using the constrained_foopsi method for deconvolution the spiking variable S does not immediately correspond to number of spikes or the spiking probability in each timebin. It merely represents a measure of "deconvolved neural activity" that is loosely proportional to the firing rate of the neuron at each time. To convert into actual spike counts some reference point is needed to threshold and quantize this signal. For example, if you deconvolve DF/F traces and have knowledge of what change in DF/F units a spike is inducing, you can use this information to approximate the number of spikes (under certain linearity assumptions).

The more accurate but slower MCMC deconvolution method does produce actual spike counts for each sample it produces. If it's used, then the variable S corresponds to the expected number of spikes that are expected in each timebin.