The authors call the first approach restorative normalization, or RN. It is the philosophy underlying most rehabilitation efforts after stroke, psychiatric illness or neurodegeneration: if neural activity has drifted from a healthy pattern, stimulation should nudge it back. The second approach they name compensatory amplification, or CA. Rather than normalizing activity, CA deliberately enhances repurposed brain processes—alternative networks, rhythms or strategies that the nervous system spontaneously deploys when its usual routes are compromised. The distinction may sound subtle, but the researchers argue it has profound consequences for how stimulation targets are chosen, how patients are selected for trials, and how success is defined.

The framework rests on four pillars drawn from cognitive neurophysiology. The first is multiple realizability: the idea, long established in cognitive science, that the same behavior or cognitive function can be supported by more than one neural configuration. Degeneracy and redundancy in brain networks mean that a lesion to one circuit does not necessarily abolish the function it served, because parallel circuits can take over. The second pillar is multiscale neuroplasticity. The brain adapts at many levels simultaneously—from molecular and synaptic changes to large-scale network reorganization—and these nested layers of plasticity provide the raw material that compensatory amplification can exploit.

The third enabling factor is precision readiness. Modern neuromodulation now possesses an unusually rich toolkit: transcranial direct, alternating and random noise stimulation, rhythmic transcranial magnetic stimulation, focused ultrasound, and invasive deep brain stimulation, increasingly guided by individualized connectome maps and closed-loop control. Electric field modeling, lesion network mapping and personalized targeting have matured to the point where clinicians can, in principle, deliver stimulation with circuit-level specificity. The fourth pillar is activity selectivity: the recognition that stimulation interacts with the brain’s ongoing state, and that the same protocol can have opposite effects depending on which neural populations are active when the current arrives. Together, these four factors make it feasible not merely to perturb the brain, but to selectively strengthen the compensatory processes that matter.

Stroke rehabilitation offers the clearest clinical arena in which the two strategies interplay. Traditional restorative approaches have often sought to dampen the unaffected, contralesional hemisphere, on the theory that it exerts excessive inhibition over the damaged side. But a substantial body of imaging work shows that many well-recovered patients rely heavily on exactly those contralesional motor areas and ipsilateral pathways. For such patients, suppressing the workaround would be counterproductive; amplifying it could be the better treatment. The authors argue that patient stratification—identifying who depends on compensatory circuits and who retains the capacity for true restoration—should become a central design principle in stimulation trials rather than an afterthought.

The framework’s reach extends well beyond stroke. In neurodegenerative disease, compensatory signatures appear remarkably early. Studies have documented prefrontal recruitment in older adults carrying amyloid-beta pathology, altered hemispheric asymmetry in mild cognitive impairment, and cortical compensation in cognitively unimpaired Parkinson’s disease patients. Indeed, one recent analysis found that clinical severity in Parkinson’s disease tracks the decline of cortical compensation itself. If those compensatory mechanisms can be detected—and the authors point to high-density electrical stimulation protocols that restored working memory and long-term memory function in older adults by resynchronizing rhythmic brain circuits—then amplification strategies might delay decline or expand residual processing capacity even as underlying pathology progresses.

Psychiatry presents a different but equally compelling case. Compensatory network activity has been documented in schizophrenia, major depression, obsessive-compulsive disorder, anxiety and attention deficit hyperactivity disorder, sometimes marking resilience and sometimes maladaptation. Deep brain stimulation studies in depression and obsessive-compulsive disorder have revealed that therapeutic effects correlate with measurable changes in frontostriatal and cingulate dynamics, and that stimulation responses depend on the patient’s moment-to-moment brain state. A compensatory amplification lens suggests that some of these interventions may succeed not by normalizing pathological activity but by reinforcing adaptive workarounds—and that distinguishing adaptive from maladaptive compensation could sharpen target selection in precision psychiatry.

Healthy aging, too, falls within scope. Classic findings such as the HAROLD model of reduced hemispheric asymmetry and the posterior-to-anterior shift in aging neural recruitment describe how older brains reorganize to preserve performance. A 2023 meta-analysis by the same group concluded that transcranial alternating current stimulation improves cognition across healthy, aging and psychiatric populations. Framing such gains as compensatory amplification rather than restoration, the authors contend, yields testable predictions: stimulation should be most effective when it is timed and tuned to amplify signatures of successful compensation, and those signatures—oscillatory synchrony patterns, network recruitment profiles, behavioral strategy shifts—can be measured before treatment ever begins.

The perspective is deliberately conceptual rather than empirical; it reports no new data. But its authors argue that positioning compensatory amplification alongside restorative normalization as a core design principle can do three concrete things: sharpen target selection by asking which neural process a protocol is meant to strengthen, guide stratified treatments by matching patients to the strategy their brains can actually use, and translate the vast literature on compensatory signatures into specific, falsifiable stimulation protocols. In an era when neuromodulation is moving from crude blunt instruments toward circuit-precise, state-aware interventions, the question is no longer only how to repair the brain, but when to amplify what the brain is already trying to do for itself.

Subject of Research: Neuromodulation strategies for restoring and amplifying brain function through restorative normalization and compensatory amplification

Article Title: Neuromodulation for restoring and amplifying brain function

Article References: Grover, S., Wen, W., & Reinhart, R. M. G. (2026). Neuromodulation for restoring and amplifying brain function. Nature Neuroscience. https://doi.org/10.1038/s41593-026-02434-6

Image Credits: AI Generated

DOI: 10.1038/s41593-026-02434-6

Keywords: neuromodulation, brain stimulation, compensatory amplification, restorative normalization, neuroplasticity, stroke rehabilitation, neurodegeneration, psychiatry, cognitive aging, deep brain stimulation, tACS, working memory

Transcranial alternating current stimulation, commonly known as tACS, delivers weak electrical currents through electrodes placed on the scalp. Unlike direct-current stimulation, which maintains a relatively constant polarity, tACS oscillates between positive and negative phases at a selected frequency. Researchers use this rhythmic stimulation to influence patterns of neural activity that may be associated with attention, memory, perception, mood or other brain functions. The underlying idea is that external electrical rhythms may interact with the brain’s own oscillations, a phenomenon sometimes described as neural entrainment. Yet the effects of tACS can vary substantially from one person to another because brain anatomy, electrode placement, tissue conductivity and intrinsic neural dynamics differ across individuals.

The new closed-loop strategy attempts to address that variability. In a conventional stimulation experiment, investigators typically choose parameters such as current strength, frequency, electrode arrangement and stimulation duration before the session begins. Those settings may be based on previous research or an anatomical model, but they do not necessarily reflect the participant’s real-time brain state. A closed-loop system, by contrast, measures a biological signal, evaluates whether the desired response is emerging and then uses that information to adjust or optimize the intervention. In this trial, real-time fMRI serves as the measurement component, while tACS provides the controlled input to the brain.

Functional MRI does not record electrical impulses directly. Instead, it detects changes in blood oxygenation through the blood-oxygen-level-dependent, or BOLD, signal. When populations of neurons become more active, local changes in blood flow and oxygen use can alter the magnetic properties of surrounding tissue. These changes allow researchers to map activity across the brain with high spatial resolution. Although the BOLD signal is slower than the underlying neural events, it can reveal where stimulation-related changes are occurring. Using the imaging data during the experiment creates the possibility of identifying whether the brain is responding in the intended region or network and whether the selected stimulation settings are producing a measurable effect.

Combining fMRI and tACS, however, is technically demanding. Electrical stimulation inside an MRI scanner can generate artifacts in the imaging data, while the magnetic environment imposes strict safety and equipment requirements. The stimulation hardware must be designed to operate within the scanner without interfering with image acquisition or creating unacceptable risks. Researchers must also separate genuine physiological changes from signals caused by the stimulation equipment, electrode leads, movement or scanner noise. Real-time analysis adds another layer of complexity because images must be acquired, processed and interpreted quickly enough to inform the next stage of stimulation rather than merely being analyzed after the session is over.

The trial’s randomized design is important because it provides a structured way to compare conditions and assess whether changes are associated with the adaptive stimulation procedure rather than with expectation, repeated scanning or ordinary fluctuations in brain activity. Randomization can help reduce systematic differences between experimental conditions, while a proof-of-concept framework allows investigators to determine whether the full technical pipeline can function in practice. That pipeline includes participant preparation, electrode placement, MRI acquisition, artifact management, real-time signal processing, decision-making and stimulation control. Establishing that these components can work together is a necessary step before larger studies can evaluate clinical effectiveness.

The most significant promise of this approach is personalization. The brain is not a fixed electrical circuit with identical wiring from one person to the next. Even when two participants receive the same stimulation protocol, their brains may respond differently because of variations in skull thickness, cortical folding, network connectivity and baseline oscillatory activity. Real-time fMRI-guided optimization could eventually allow researchers to identify which stimulation parameters are most effective for a particular individual. Instead of assuming that a single frequency or electrode montage will work equally well for everyone, future systems might adjust the intervention according to each participant’s measured neural response.

Such technology could have implications for research into psychiatric and neurological conditions, although the present work should not be interpreted as proof that the system is ready to treat patients. Noninvasive stimulation is being investigated for conditions including depression, anxiety, chronic pain, addiction and cognitive disorders, but results across studies have often been mixed. One reason may be that stimulation protocols are not sufficiently sensitive to individual biology or to changes in brain state during an intervention. A responsive system could help researchers test whether adapting stimulation in real time improves consistency. It could also offer a way to study causal relationships between brain rhythms, distributed neural networks and behavior.

The approach may also change how scientists think about experimental control. Rather than treating the brain as a passive object that receives a predetermined dose of stimulation, closed-loop neuromodulation treats it as a dynamic system that continuously provides feedback. That perspective is already influential in other areas of neuroscience, including deep-brain stimulation and brain-computer interfaces. Applying it to tACS with real-time fMRI is especially ambitious because it combines a relatively accessible form of stimulation with one of the most information-rich tools for measuring human brain function. If refined, the method could help bridge the gap between broad population-level protocols and truly individualized interventions.

For now, the study is best understood as an engineering and methodological milestone rather than a finished therapy. The researchers’ central contribution is to demonstrate the feasibility of linking real-time functional imaging with adaptive transcranial electrical stimulation in a randomized experimental framework. Future investigations will need to determine how reliably the system identifies meaningful neural responses, how long those responses last, whether they translate into changes in behavior or symptoms and whether the approach can be scaled beyond specialized research scanners. The work nevertheless points toward a striking possibility: brain stimulation that does not merely send commands into the nervous system, but listens to the brain and adjusts its strategy in response.

Subject of Research: Closed-loop transcranial alternating current stimulation guided by real-time functional magnetic resonance imaging.

Article Title: Closed-loop transcranial electrical brain stimulation with fMRI: A proof-of-concept randomized trial of real-time fMRI-guided tACS optimization.

Article References: Soleimani, G., Kuplicki, R., Mulyana, B. et al. “Closed-loop transcranial electrical brain stimulation with fMRI: A proof-of-concept randomized trial of real-time fMRI-guided tACS optimization.” Translational Psychiatry (2026). https://doi.org/10.1038/s41398-026-04319-5

Image Credits: AI Generated

DOI: https://doi.org/10.1038/s41398-026-04319-5

Keywords: tACS, transcranial electrical stimulation, functional MRI, fMRI, closed-loop neuromodulation, brain stimulation, neurotechnology, personalized medicine, neuroscience, neural oscillations