In the experimental data, we occasionally found that the distributions of neural or behavioral events had ‘outliers’ – samples that fell far from the distribution, often with a large gap between the outlier point and the nearest neighbor sample. To remove subjective choices from our analysis pipeline we developed a systematic automated algorithm for excluding outliers. Although this outlier exclusion is not subjective, it is still important to understand how it may impact the ability of our power-law fitting algorithm. Here, we benchmarked our outlier exclusion and fitting pipeline on synthetic samples drawn from known power-law distributions. We tested a family of truncated power-law distributions. In all cases, we set the minimum and maximum sample sizes to be 0.01 and 100, respectively. We tested a range of power-law exponents z=0.8 to z=2, where P(s) ~ s-z (like those observed in our experimental results) and sample sizes from N=100 to N=10,000. Finally, we tested outlier cutoff thresholds 0.01, 0.03, 0.06, and 0.1. (A) Our fitting algorithm successfully identified the correct exponent, independent of the outlier threshold (N was fixed at 5000 for all panel A results). (B) Best fit exponents were very slightly underestimated when the sample size was very small, less than about 200. Our typical sample size in the experiments was larger than this. (C) Since excluding outliers inevitably reduces the range of the data, power-law range was more sensitive to outlier threshold, but only for exponents above about 1.7. We note that the neural subsets that were strongly correlated with behavior had much lower exponents (closer to one), for which there is no substantial sensitivity to outlier threshold. Dashed line shows the upper bound on power-law range for N=5000. (D) Power-law range was also underestimated when sample sizes were small. Again, we emphasize for the outlier thresholds we used in our main results (0.03 and 0.06) and the typical sample sizes from the experiments (>400), the best-fit power-law range was not very sensitive to the outlier threshold.