Non-invasive brain stimulation offers a promising route for modulating large-scale neural dynamics in conditions such as frontotemporal dementia, Alzheimer’s disease, mild cognitive impairment and obsessive–compulsive disorder. Yet fundamental stimulation parameters—where, when, for how long, and at which frequency—remain poorly defined and highly subject-specific. We view the brain as a multistate, metastable dynamical system that continually transitions between quasistable attractors. From this perspective, neurorehabilitation becomes a problem of reshaping the system’s dynamical landscape: guiding activity toward favourable attractor states and re-establishing the flexible repertoire that underlies healthy function.
To formalise and quantify this principle, we develop a control-theoretic framework for regulating the emergent frequency of a whole-brain Kuramoto network constructed from empirical structural connectivity. Using linear–quadratic control applied to the linearised system, together with phase–frequency analysis, we derive optimal control signals capable of steering the network toward a desired collective frequency. We show that the dominant attractor depends jointly on intrinsic regional frequencies and on global network parameters such as coupling strength and network size. Our results reveal a clear force–duration tradeoff, demonstrate that stimulation aligned with intrinsic network resonances minimises control energy, and identify topological constraints that determine which dynamical regimes are attainable.
Together, these findings provide a principled basis for designing personalised neuromodulation strategies. By linking stimulation parameters to the underlying physics of whole-brain dynamics, our framework moves beyond restoring a single “normal” pattern and instead supports the recovery of a healthy, adaptable repertoire of brain states.