Neural Competition and Structured Interference in Muscle Synergies: Mapping the Limits of the Linear Summation Hypothesis
Muscle synergies are modeled as additive building blocks that linearly combine to generate motor outputs, an assumption central to decomposition approaches like non-negative matrix factorization (NMF). This model is relevant for wearable electromyography, where synergies can provide a compact representation of muscle coordination for human monitoring. However, linear frameworks neglect potential interactions arising from shared neural and muscular resources. We propose that the non-orthogonality of synergies may create overlapping muscle recruitment, leading to resource-dependent competition and sub-additive interactions. To investigate the boundaries of the linear assumption, we introduced a sub-additive coupling term into a biomimetic toy model, systematically modulating temporal co-activation (f) and interference magnitude (α). Our simulations show that structured reconstruction errors emerge under high-interference conditions, while NMF exhibits topology-driven resilience across a wide range of parameters within the tested biomimetic model. The algorithm exploits the geometric distinctiveness of spatial weights to resist nonlinear distortions until it reaches a critical co-activation threshold. Rather than invalidating the synergy construct, our findings map the boundaries of NMF robustness. Under extreme overlapping demands, nonlinear interference may be misattributed to altered synergy weights or distorted activation patterns. The proposed framework provides a controlled computational basis for assessing the reliability and interpretability of synergy-based signal processing in wearable human-monitoring systems.
Authors
- Alessandro Scano (ORCID: https://orcid.org/0000-0002-1896-9325)
- Cristina Brambilla (ORCID: https://orcid.org/0000-0002-3201-8315)
- Valentina Lanzani (ORCID: https://orcid.org/0009-0005-6944-4931)
- Nicol Moscatelli (ORCID: https://orcid.org/0009-0000-2500-6781)
Institutions
- National Research Council (IT)
- Institute of Intelligent Industrial Systems and Technologies for Advanced Manufacturing (IT)
Publication Details
- Journal
- Sensors
- Published
- 2026-11-12
- DOI
- https://doi.org/10.3390/s26227143
- Primary Topic
- Muscle activation and electromyography studies
- Type
- article
- Field-Weighted Citation Impact
- 0.00