Healthy brain function requires a balance between stable integration across brain areas for coordinated activity and brief periods of desynchronization that allow subsystems to reconfigure and express specialized functions. Metastability, a concept from statistical physics and dynamical systems theory, has been proposed as a key signature of this balance. Neuroscience research has used markers of metastability to study cognitive performance, healthy aging, meditation, sleep, responses to drugs, and to characterize psychiatric conditions and disorders of consciousness. However, the term is often used heuristically or inaccurately, making the literature difficult to navigate. This paper provides a comprehensive review of metastability in neuroscience, covering its scientific and historical foundations, practical estimators, and a critical analysis of recent theoretical developments to clarify misconceptions.
In cognitively normal older adults, β-amyloid (Aβ) buildup in the default mode network impairs learning independently of tau, but higher dopamine synthesis capacity in the dorsolateral striatum can recover that learning performance. Aβ-positive individuals show reduced default mode network activity in response to error feedback, which relates to poorer learning. Computational modeling indicates that Aβ disinhibits the default mode network during error processing, and dopamine synthesis capacity in the dorsolateral striatum rebalances effective connectivity between the default mode network and frontostriatal network, counteracting Aβ-related disruption. These findings suggest that dopaminergic function can partially compensate for Aβ-related learning deficits through network rebalancing, offering a candidate mechanism for cognitive resilience in preclinical Alzheimer's disease.
A whole-brain computational model of the corticothalamic system, built from empirical data on targeted and diffusely projecting thalamocortical nuclei, reproduces key features of propofol anesthesia: reduced network integration, lower state diversity, impaired susceptibility to perturbation, and decreased corticocortical coherence. These signatures indicate suppressed information transfer across the cerebral cortex. Selectively stimulating the matrix thalamus in the model restores signatures of conscious arousal, matching empirical results in macaques, and produces wake-like information processing states. The findings suggest that matrix thalamocortical projections modulate large-scale cortical attractor dynamics to enable the complex communication states that support conscious awareness.