Figure 8

Statistics of the normalized maximum speed labeling subjects on the Gamma plots for representative novice and expert. (A) Expert (a,b) MLE for each speed condition and training context (bag vs. no-bag) with 95% confidence intervals. (B) The corresponding Gamma probability density function (PDF) curves reveal in the expert a broad bandwidth of parameter values across training contexts. It also shows an unambiguous distinction between bag and no-bag conditions for each speed level. Speed levels are not confused by the expert’s kinesthetic data. (C-D) The novice however shows a narrow bandwidth of parameter values with no clear distinction between slow motions that are against the bag or towards a simulated opponent. The novice’s kinesthetic data does distinguish between the fast-bag condition and the other training contexts. Notice the degree of dispersion of the probability distribution measured through the Fano Factor (noise to signal ratio) $F=\frac{{\sigma }_w^2}{{\mu }_w}$ the ratio of variance to mean taken within the time window w (the time in ms to reach the peak velocity, on the order of 200 ms in this case) is indistinguishable in the novice for the slow case (6.47 × 10^-5 slow-bag vs. 6.30 × 10^-5 slow-no-bag) and for the fast case (1.76 × 10^-4 fast-bag vs. 2.47 × 10^-4 fast-no-bag). The novice can however differentiate between fast and slow (Wilcoxon ranksum test of equal medians p < 10^-3). Compare to the expert with Fano factors that distinguished speed within each training context (slow-bag 4.4 × 10^-4 vs. fast-bag 0.0015; slow-no-bag 9.7 × 10^-5 vs. fast-no-bag 4.08 × 10^-4).