Hypothesis 4 was tested as follows:
For each of the 10 repetitions on each recorded neuron, the effect of light on AP frequency was calculated as the difference between the firing frequency in the LED condition and the firing frequency in the no LED condition. For each neuron, the median effect of light was calculated. The values calculated for all recorded neurons were used for descriptive statistics by calculating the average population effect ± 95% CI as well as the Cohen’s d ES (average of population effect /STD of population effect). Statistical tests were performed using unilateral one-sample Wilcoxon signed rank test (tested hypothesis, median effect <0). The σ risk level was set at 0.05. The experiment was conducted on 15 neurons. This sample size was determined based on a previously published experiment on MCs and MSNs, in which continuous light stimulation reduced firing frequency with an ES=-1 at -1.6, respectively (Ait Ouares et al. , 2019). Therefore, the statistical power to observe an effect as small as that produced by 1-s continuous light stimulation was between 97% and 99% (calculated with G*Power 3.1.9.2).
See supplementary table 1 for a summary of statistical analysis.
The results obtained from procedure 1 were used for exclusion criteria based on Rm (see later) as well as for exploratory analysis aimed at correlating the eventual effect of light stimulation with Rm.
Analyses were performed with custom scripts (Python 3.7).