Parameters

From the recordings of muscular activity, we determined the low or static level (10th percentile, RMS10), the median level (50th percentile, RMSMEDIAN), and the high level (90th percentile, RMS90) according to the amplitude probability distribution [22]. These parameters were determined for the overall task, screwing only (phase 1), and fastening only (phase 2). In addition, we calculated the within-day change of the MPF and RMSMEDIAN to assess whether there were signs of electromyographic muscular fatigue, determined as an increased RMSMEDIAN concomitant with a decreased MPF [23].

We calculated the relative cycle-to-cycle variability of RMS during screwing (phase 1) and fastening (phase 2) as a metric reflecting the size of motor variability. We defined one work cycle of screwing as screwing 1 screw (~40 s) and one work cycle of fastening as fastening 6 screws belonging to 1 vertical row (~28 s), resulting in 72 screwing and 12 fastening work cycles per day. The relative variability was assessed by calculating the RMS coefficient of variation (CV) per experimental day (RMSCV), i.e., the square root of the mean variance of RMS across screwing and fastening work cycles, divided by the RMSMEAN of the screwing and fastening work cycles.

**Figure 1.** Experimental task setup with an example of screwing using the t-handle screwdriver (left) and an example of fastening using the torque screwdriver (right).
