Disorders of consciousness, or DoC, arise when insults—traumatic brain injury, anoxic damage, hemorrhage, infection—disrupt the distributed neural networks that generate awareness and arousal. Clinicians classify patients along a spectrum from coma through vegetative state or unresponsive wakefulness syndrome to the minimally conscious state, which is itself subdivided by whether language responses persist. Diagnosis is notoriously error-prone: the review cites a 37 percent misdiagnosis rate, and while the global incidence remains unknown, estimates hover around 201 cases per 100,000 annually. Because the mechanisms underlying consciousness are still poorly understood, and because spontaneous recovery can mimic therapeutic effects, the field has struggled to establish whether neurostimulation genuinely works. The Coma Recovery Scale-Revised remains the reference behavioral measure, tracking the two core components of consciousness—awareness and arousal—through standardized clinical assessment.

The review’s methodology was deliberately strict. Following the PRISMA extension for scoping reviews, the researchers searched PubMed, SCOPUS and Web of Science, with a final search performed in March 2026. Crucially, they only included studies with more than one participant, a quantitative measure of consciousness recovery, and a control or comparator group—a restriction designed to separate stimulation effects from the natural recovery that confounds uncontrolled reports. The initial search generated 1,318 records; after duplicate removal, dual screening and full-text review, nineteen studies met the criteria. An updated search in March 2026 added three more, bringing the final tally to twenty-two. Eighteen were randomized controlled trials, most employing sham controls, with healthy controls used in four studies. All subjects were adults with mixed etiologies—traumatic, anoxic, infective, hemorrhagic—with a mean age of 48.1 years, 43.7 percent female, and an average of 16.9 months since the inciting event.

The anatomical map that emerged is strikingly lopsided. Thirteen of the twenty-two studies targeted the frontoparietal network, the circuit spanning the dorsolateral prefrontal cortex and posterior parietal regions that supports executive function, attention and working memory—often called the central executive network. The dorsolateral prefrontal cortex alone was the primary target in fourteen studies, suggesting a pronounced research bias toward this single region. Three studies engaged the default mode network, integral to internal awareness, via the posterior parietal cortex and precuneus; one combined frontoparietal and default mode stimulation; one targeted the salience network through transcutaneous auricular vagal nerve stimulation of the bilateral auricles; two stimulated the sensorimotor network via the primary motor cortex; and two used invasive deep brain stimulation—one to the thalamocortical pathway through the bilateral centromedian-parafascicular nuclei, one to the medial pallidum. Transcranial direct current stimulation and repetitive transcranial magnetic stimulation tied as the most frequent modalities, each used in eight studies, with stimulation durations typically ranging from five to fourteen days—though the pallidal deep brain stimulation protocol extended up to eleven months.

Technically, the two dominant modalities work in fundamentally different ways. Transcranial direct current stimulation delivers weak, subthreshold currents through scalp electrodes, subtly biasing cortical neuronal membrane potentials toward excitation or inhibition depending on polarity. Repetitive transcranial magnetic stimulation uses rapidly alternating magnetic fields to induce electric currents that depolarize cortical neurons directly, with stimulation frequency determining whether the net effect is excitatory or inhibitory. Both are non-invasive and, across the review, maintained good safety profiles. The review also catalogued newer variants—transcranial pulsed current, alternating current, random noise, and intermittent theta burst stimulation—reflecting a field experimenting with how best to entrain residual network activity.

Here is where the story gets genuinely interesting. Among the twelve studies that also used electroencephalography and the two that used functional imaging, improvements in functional connectivity—particularly increased alpha-band coherence—were observed after stimulation, and these electrophysiological shifts often correlated with individual-level gains in clinical consciousness. In other words, when the stimulation worked at the level of the person, the brain’s networks visibly reorganized in parallel. Yet when the reviewers aggregated results across cohorts, the picture fractured: only six of the twenty-two studies reported significant group-level improvement in consciousness-related behavioral outcomes on the Coma Recovery Scale-Revised. Those six responders spanned the frontoparietal network, the default mode network, a combined frontoparietal-default mode cross-network approach, and the salience network—with no common stimulation modality, no concentration at the most frequently studied network, and median time from injury ranging from roughly 1.5 to 43 months. Four used repetitive transcranial magnetic stimulation, one used direct current stimulation, and one used auricular vagal nerve stimulation.

Why the disconnect between individual and group results? The review’s authors argue that heterogeneity is the culprit. Cohort sizes ranged from 3 to 50 participants. Etiologies varied from traumatic brain injury, which can follow prolonged recovery trajectories, to anoxic injury, which follows more predictable patterns. Clinical subtypes varied from minimally conscious state-only cohorts to mixed populations including unresponsive wakefulness syndrome and vegetative state. Stimulation protocols were applied uniformly despite these differences, neglecting critical individual variations in structural cerebral integrity and baseline functional connectivity that can be identified through functional MRI or electrophysiology. One study even used EEG electrode landmarks to determine stimulation targets—a practical but error-prone approach given individual anatomical variability, particularly in patients whose brains may have structurally altered. The reviewers also note that brief stimulation durations, limited prospective follow-up, and absent dose-response data further confound interpretation, since cerebral recovery is optimized by sustained, repetitive stimulation that drives lasting synaptic reorganization.

Safety, at least, is encouraging. Neurostimulation modalities generally demonstrated favorable profiles, though adverse event reporting was inconsistent, with eleven studies not formally reporting any. Only one study—an invasive deep brain stimulation to the medial pallidum—reported an adverse event attributable to the intervention: one participant developed a bronchopulmonary infection and died fifteen months after electrode implantation, with no acute neurological signs beforehand. Participants receiving auricular vagal nerve stimulation reported seizures, pneumonitis and deep vein thrombosis in both treatment and sham groups, with no significant difference between them and no attribution to stimulation effects.

The deeper conceptual problem the review surfaces is that consciousness is not the product of any single network. Neural networks share overlapping anatomical regions, and it is their complex interplay that gives rise to awareness. Studies classed as engaging the default mode network via the parietal hub likely co-modulated the frontoparietal network through shared parietal nodes. The positive outcomes observed across studies may therefore reflect wider coordinated restoration of neuronal activity across multiple networks rather than the effect of any isolated target. The absence of a network-specific response pattern is not evidence that chosen targets were irrelevant, the authors caution—rather, the small number of responders and the heterogeneity of modality, etiology and protocol limit the dataset’s power to detect network-specific effects. The network classifications should be read as an organizing framework, not proof of selective engagement.

Where does this leave the field? Network-targeted neurostimulation is promising but remains investigational. The review’s authors propose a clear roadmap: future trials should employ individualized structural and functional mapping guided by residual network integrity rather than anatomical landmarks, stratify participants by baseline connectivity and clinical subtype, apply standardized outcome definitions, and explore dual-network or dual-modality stimulation—approaches that have not yet been investigated. A shift toward tailored, patient-specific neuromodulation strategies, they argue, may finally move the field toward precision neurorehabilitation, optimizing therapeutic outcomes for consciousness recovery in ways that uniform protocols never could.

Subject of Research: Neuroscience

Article Title: What are the neural networks that have been modulated by anatomically targeted therapies to promote recovery of consciousness? A scoping review

Article References: Thakker, R. D., Karabacak, M., Aydin, S., Budihal, B. R., Cummins, D. D., Panov, F., & Margetis, K. (2026). What are the neural networks that have been modulated by anatomically targeted therapies to promote recovery of consciousness? A scoping review. Frontiers in Neurology, 17, Article 1882395. https://doi.org/10.3389/fneur.2026.1882395

Image Credits: AI Generated

DOI: 10.3389/fneur.2026.1882395

Keywords: neurostimulation, disorders of consciousness, neuromodulation, tDCS, rTMS, deep brain stimulation, neurorehabilitation, neural, networks, been, modulated, anatomically