Fig. 2.

Distributions for each of the 4 types of similarity score for both monkeys. The y-axis of each plot is in units of probability density (the proportion of observations found in each bin, divided by the width of the bin). The x-axis indicates similarity score: correlation coefficient for X-corr, waveform, and autocorr; change in log mean rate for the bottom panel. The green line represents the distribution for comparisons between recordings from different channels, which are guaranteed to be different neurons. The black area represents similarity scores from the same channel on subsequent days that were classified by the algorithm as the same neuron. The red line represents the scores described in Synthetic data, where we ran the algorithm using only 3 of the scores so that we could validate the 4th score without creating circular logic. The blue line shows the Gaussian function representing the same-neuron distribution in the expectation-maximization procedure described in Tracking the same neurons.