Fig. 7: Merging arising from synchronous drives to multiple synergies.
a In all detected instances of W synergy merging between Sedent0/Elite subject pairs, the C(t)’s of the to-be-merged (Sedent0) and merged (Elite) muscle synergies were resampled across time to produce cycle averages. The Sedent0 C(t)’s were then fit to a non-negative linear combination model to identify the combination coefficients that best explained the C(t) of the merged, Elite synergy. Here, the C(t) merging of S0-5, 6, and 8 from 2 Sedent0 subjects into S0-5 + 6 + 8 in eight Elite subjects (n = 15 instances of merging) is shown as an example. Merging coefficients for S0-5, 6, and 8 were 0.09 ± 0.11 (mean ± SD), 0.84 ± 0.11, and 0.0017 ± 0.0033, respectively. The coefficient of S0-6 obviously dominates this combination. b Shown here is the combination coefficients for the C(t)’s (per model in a) for the 10 most prevalent merging combinations in Elite (averaged across merging instances). In 7 of 10 merging combinations, the C(t) of 1 to-be-merged synergy dominated the combination (largest combination coefficient = 78–99% of total coefficient values). In the remaining three (marked by x), the combination coefficients were not statistically significant across the to-be-merged synergies (S0-6 + 7, p = 0.34; 5 + 8, p = 0.16; 7 + 11, p = 0.56; 1-way ANOVA). Numbers of merging instances, from top to bottom, were, n = 56, 15, 7, 8, 22, 8, 6, 8, 11, 6. c A model that accounts for muscle-synergy merging by reassigning multiple synergy-encoding interneurons (W1 and W2) to be driven by the same oscillator for one of the original synergies (C*(t) = C1(t)), and with the original C2(t) ceasing to be active (blue dotted line). This model is consistent with results in b. Source data for a, b are available as a Source Data file.
