Exercise has long been celebrated as one of the most reliable non-pharmacological routes to better brain health, but a sweeping new review argues that the popular explanation—a single molecule released by working muscles that travels to the brain and upgrades cognition—is far too simple. Writing in Frontiers in Integrative Neuroscience, researchers led by Dongjing Yin and Cheng Luo of Chengdu College of University of Electronic Science and Technology of China propose that exercise-related brain remodeling emerges from a genuinely bidirectional dialogue between the brain and the body, in which the brain is not a passive recipient of peripheral signals but an active regulator that continuously shapes the very signals it later receives.

The review, published on 14 October 2026, confronts a fundamental problem in the exercise-neuroscience literature: the evidence spans molecular, cellular, organ, neural-circuit, imaging, and behavioral levels that differ dramatically in causal strength. Animal and cellular studies provide rich mechanistic detail, while human studies typically report circulating biomarkers or neuroimaging changes that are indirect and process-nonspecific. The authors argue that these levels cannot be stitched into a causal chain merely because their changes occur after exercise. Instead, they develop an integrative framework that explicitly separates what is well established at each biological level from what remains untested across levels.

A central conceptual contribution is the distinction between acute exercise responses and sustained brain remodeling. A single bout of exercise produces transient changes in lactate, cytokines, catecholamines, cerebral perfusion, arousal, and task performance over seconds to hours. These responses represent immediate regulatory states and candidate signals, not evidence of durable neural adaptation. Repeated exercise over weeks to months may instead produce persistent changes in synaptic regulation, neural circuitry, brain structure, and behavior. Acute responses may contribute to chronic adaptation when repeatedly engaged, but the authors insist this relationship cannot be assumed without temporally ordered longitudinal evidence.

The framework begins with the brain itself. Exercise is not initiated solely by peripheral metabolic disturbance; it is a centrally coordinated physiological state. Before and during movement, central motor commands activate skeletal muscle while simultaneously coordinating cardiovascular and respiratory adjustments to anticipated metabolic demands. The autonomic nervous system increases sympathetic outflow and modifies parasympathetic activity, influencing heart rate, cardiac output, vascular resistance, and blood pressure according to exercise intensity. Exercise also activates the hypothalamic-pituitary-adrenal axis, with glucocorticoid responses that depend on workload, duration, training status, circadian timing, and recovery conditions. Perception of effort—a centrally integrated experience linked to motor command—further regulates exercise intensity and task continuation, creating a behavioral route through which central processing shapes future brain-body interactions.

On the ascending side, contracting skeletal muscle generates several prominent candidate signals. Interleukin-6 rises with exercise and participates in substrate mobilization and post-exercise immune regulation in a source-, concentration-, and time-dependent manner. Lactate serves as both metabolic substrate and signaling molecule, entering the brain through monocarboxylate transport systems and linked experimentally to epigenetic processes. Irisin, cathepsin B, interleukin-15, and insulin-like growth factor 1 also respond variably to exercise mode and dose. Yet the review emphasizes that a peripheral concentration change alone cannot identify the dominant tissue source or prove delivery to a neural target. Muscle-derived brain-derived neurotrophic factor, for instance, appears well suited to local autocrine or paracrine functions at neuromuscular sites, whereas its contribution to circulating BDNF is limited and difficult to isolate.

The liver, adipose tissue, and bone add further endocrine-metabolic partners. When exercise is prolonged or nutritional availability is low, hepatic beta-hydroxybutyrate may rise, and beyond serving as an alternative fuel it can inhibit histone deacetylases and participate in histone beta-hydroxybutyrylation, mechanisms with potential epigenetic consequences for the brain. In aged mice, the liver-derived factor glycosylphosphatidylinositol-specific phospholipase D1 was identified as capable of transferring part of the neurogenic and cognitive effects of exercise. Adipose tissue contributes adiponectin, leptin, inflammatory mediators, and potentially extracellular-vesicle cargo, but training-related adipokine changes are often confounded by concurrent changes in fat mass and energy balance. Bone, an endocrine organ in its own right, may release osteocalcin in response to mechanical loading, with animal support for cognitive links but limited and inconsistent human evidence.

The gut microbiome, circulating immune-vascular milieu, and extracellular vesicles complete the peripheral signaling landscape. Some studies report exercise-related changes in short-chain-fatty-acid-producing taxa or in butyrate, propionate, and acetate, but human findings are heterogeneous and sensitive to diet, adiposity, medication, geography, and sampling strategy. Cytokines and endothelial mediators can act on cerebral vascular receptors and modify neurovascular or glial states without requiring a molecule to enter the brain parenchyma, though human blood markers are not brain-specific. Extracellular vesicles can transport proteins, microRNAs, lipids, and metabolites with increased cargo stability, but their tissue origin, biodistribution, interface crossing, target-cell uptake, and biologically effective dose remain major unresolved questions.

Communication between peripheral organs and the brain occurs through interacting hemodynamic, vascular, humoral, and neural pathways that operate simultaneously. Cerebral perfusion during exercise reflects the integrated influence of arterial carbon dioxide tension, cerebral metabolism, arterial pressure, neurogenic regulation, cardiac output, and cerebrovascular resistance. Hyperventilation during strenuous exercise can reduce arterial carbon dioxide and produce cerebral vasoconstriction even while cardiac output remains elevated, meaning that exercise-related changes in cerebral perfusion cannot be interpreted as direct measures of neuronal activation. The blood-brain barrier, formed by specialized endothelial cells with pericytes, basement membrane, and astrocytic endfeet, is better understood as a signaling platform than a passive wall: peripheral inflammatory stimuli can activate receptors and cyclooxygenase-related pathways in brain vessels, generating central-side mediators without bulk entry of the original molecule. The choroid plexus and circumventricular organs, with their fenestrated capillaries, provide additional sensing interfaces, though direct evidence that an exercise-elevated myokine is rapidly detected by a specific circumventricular organ and transmitted to a hippocampal target is lacking. Vagal afferents project to the nucleus tractus solitarius and distribute information to hypothalamic, limbic, and forebrain networks, but not every metabolic or inflammatory signal is directly encoded by vagal terminals.

Within the brain, candidate central responses include neurotrophic and synaptic signaling, glial-state regulation, adult hippocampal neurogenesis, and neurovascular adaptation. Locally produced BDNF binds tropomyosin receptor kinase B and activates MAPK/ERK-CREB, PI3K-Akt-mTOR, and PLC-gamma pathways regulating gene expression, survival, local translation, and synaptic plasticity. Exercise-induced adult hippocampal neurogenesis is among the most reproducible findings in rodents, yet the human situation is far more uncertain: studies have reported both very low or undetectable adult neurogenesis and persistent neurogenic cell populations in healthy adult and aging brains, and even if adult human neurogenesis occurs, its magnitude, functional contribution, and exercise responsiveness in living humans remain unestablished. Human neuroimaging provides complementary systems-level evidence—a landmark randomized trial in older adults reported approximately 2 percent mean hippocampal volume increase after one year of moderate-intensity aerobic training—but MRI volume and cortical thickness reflect mixtures of cellular, vascular, extracellular-space, and water-related processes that lack cell-type and process specificity.

The review ultimately argues that the weakest part of the evidence base is the bridge across levels. Few human studies demonstrate, in temporal order, that an organ-specific signal reaches or activates a defined interface, engages a particular neural cell population, produces a specific imaging change, and thereby causes cognitive benefit. Age, sex and hormonal state, genotype, disease stage, baseline function, exercise dose, recovery, and adherence further shape response heterogeneity, and the BDNF Val66Met polymorphism and APOE epsilon4 carriage illustrate genetic modifiers whose clinical utility remains unproven. The authors call for preregistered longitudinal interventions, source-resolved and time-resolved biomarker sampling, cross-species validation using physiologically anchored exercise intensity domains, multimodal target-engagement measures, and externally validated prediction models. Until such bridges are built, exercise-related brain remodeling is best understood not as a single molecule’s journey but as a dynamic, multi-organ conversation in which the brain and body continuously reshape one another with every step, stride, and repetition.

Subject of Research: Exercise-related brain remodeling through bidirectional brain-body crosstalk: an integrative narrative review

Article Title: Exercise-related brain remodeling through bidirectional brain-body crosstalk: an integrative narrative review

Article References: Yin, D., Luo, C., Hu, J., & Bi, S.-Z. (2026). Exercise-related brain remodeling through bidirectional brain-body crosstalk: an integrative narrative review. Frontiers in Integrative Neuroscience, 20, Article 1954740. https://doi.org/10.3389/fnint.2026.1954740

Image Credits: AI Generated

DOI: 10.3389/fnint.2026.1954740

Keywords: exercise, brain-body crosstalk, neuroplasticity, BDNF, exerkines, neuroimaging, cognition, blood-brain barrier, Exercise-related, brain, remodeling, bidirectional