TMS and Cognitive Decline: What the Research Shows and What Remains Uncertain

When families describe memory loss, they often use one word for changes that may involve several different functions. A person may have difficulty learning new information, retrieving a name, maintaining attention, organizing a task, finding words, judging space, or carrying out familiar daily activities. Those functions depend on overlapping but distinct brain networks.

That complexity is important when reading research about transcranial magnetic stimulation. TMS does not stimulate "memory" as though memory were a single switch. A study selects a cortical target, delivers a particular pulse pattern, and measures certain outcomes in a defined group of participants. The result applies most directly to that protocol and that population.

Some studies in Alzheimer's disease and related cognitive disorders have reported encouraging changes in selected cognitive or functional measures. Other well-designed studies have not found active stimulation superior to sham. The honest conclusion is not that TMS has been proven to restore memory, and it is not that the research has no value. The evidence contains a signal worth studying, together with substantial uncertainty about who may respond, which protocol matters, and whether any effect is durable or disease modifying.

Alzheimer's disease and other causes of cognitive decline disrupt communication within distributed brain networks. The changes are not limited to one isolated "memory center." Regions involved in memory, attention, executive function, language, orientation, and daily function interact continuously.

TMS gives researchers a way to influence activity in a targeted cortical region without surgery. Repeated stimulation may alter cortical excitability and activity within connected networks. That provides a plausible scientific reason to study whether selected protocols can change performance on cognitive or functional measures.

Plausibility is not proof. A mechanism that makes sense biologically can still fail in a clinical trial. The meaningful question is whether carefully controlled studies show a reproducible benefit that is large enough to matter, persists long enough to be useful, and applies outside one research center.

Cognitive TMS studies differ substantially. Many earlier trials targeted the left or bilateral dorsolateral prefrontal cortex because of its role in attention, working memory, and executive networks. Other studies have targeted the precuneus, a region within the default mode network that is closely connected with memory systems and is affected in Alzheimer's disease. Some protocols stimulate one site. Others stimulate several cortical sites.

Studies also differ in frequency, intensity, coil design, number of pulses, length of the intensive phase, use of maintenance sessions, and targeting method. Some use standard scalp coordinates. Others use MRI-guided neuronavigation or combine TMS with EEG information to individualize the target.

Some trials deliver TMS alone. Others pair stimulation with cognitive exercises. A positive result from a precuneus protocol cannot automatically be used to validate bilateral prefrontal stimulation, multisite stimulation, or TMS combined with cognitive training. They are related technologies, but they are not the same intervention.

Research studies commonly use instruments such as the Mini-Mental State Examination, the Alzheimer's Disease Assessment Scale Cognitive Subscale, the Clinical Dementia Rating Scale Sum of Boxes, and tests of specific memory or executive tasks. Some trials also measure activities of daily living, behavior, caregiver observations, EEG connectivity, or imaging.

Each measure answers a different question. A score on a global cognitive scale may detect an average group difference, but it does not automatically tell us whether a person became more independent, communicated more easily, managed medication more safely, or retained the change after the study ended.

A statistically significant difference is also not always a clinically meaningful difference. Statistical significance asks whether an observed result is unlikely to be due to chance under the study's assumptions. Clinical significance asks whether the magnitude of change matters in real life.

For a progressive neurodegenerative condition, "less worsening than sham" and "improvement from baseline" are different outcomes. Both can be important, but they should not be described as though they mean the same thing.

A 2022 randomized, double-blind, sham-controlled study evaluated personalized repetitive TMS directed at the precuneus over 24 weeks in 50 patients with mild-to-moderate Alzheimer's disease. The active group showed differences from sham on selected cognitive and functional outcomes. The trial was notable because it used a network-based target and followed participants longer than many earlier studies.

A related 2025 randomized study extended the precuneus approach to 52 weeks. Forty-eight participants were randomized, and 32 completed the study. The active group had less worsening than the sham group on the prespecified Clinical Dementia Rating Scale Sum of Boxes outcome and differences on selected secondary cognitive, daily-function, and behavioral measures.

These findings are scientifically important. They suggest that target selection, individualized network mapping, and maintenance stimulation deserve further study. They do not establish that TMS reverses Alzheimer's disease. The longer study was conducted at a single center, had a modest sample, experienced meaningful attrition, and included investigator financial interests that were disclosed in the publication. The authors themselves called for multicenter confirmation.

A 2024 multisite, randomized, double-blind, sham-controlled trial evaluated bilateral dorsolateral prefrontal rTMS in more than 150 participants with mild-to-moderate Alzheimer's disease. Participants received different active-treatment durations or sham stimulation and were followed for short- and longer-term outcomes.

The trial did not find active stimulation superior to sham on the primary cognitive outcome. This negative result is not a reason to ignore the positive studies. It is a reason to avoid treating "TMS" as one uniform therapy.

The neutral trial used a different target and protocol from the precuneus studies. That leaves several possibilities. The target may matter. The dose or treatment duration may matter. The sham response may matter. Patient selection may matter. It is also possible that some encouraging findings will not reproduce reliably in larger independent studies.

Good evidence review requires holding those possibilities at the same time rather than selecting only the explanation we prefer.

Some research combines TMS with structured cognitive exercises. The theory is that stimulation may influence the neural network while training engages the same or a related cognitive function.

A 2020 randomized, double-blind, sham-controlled study of 50 participants paired left prefrontal rTMS with face-name associative-memory training. The active-stimulation group improved more on the trained associative-memory task and showed a difference in visuospatial reasoning. Response varied with disease severity and education.

That is an encouraging domain-specific finding. It should not be converted into the broader claim that TMS restores memory or changes the course of Alzheimer's disease. Improvement on a trained task can reflect learning that is specific to the exercise. The critical questions are whether benefits generalize to untrained abilities, daily function, and long-term outcomes.

Systematic reviews of TMS or other noninvasive brain stimulation combined with cognitive training have reported favorable pooled effects in some analyses. They also describe small trials, mixed stimulation methods, variable targets, differing cognitive programs, incomplete follow-up, and inconsistent results across scales. Pooling studies can increase statistical power, but it cannot remove weaknesses or biological differences in the underlying trials.

This distinction is central. A study can report a difference on a cognitive test without showing that the underlying neurodegenerative process has been stopped or reversed.

To establish disease modification, researchers would want converging evidence that an intervention changes the long-term clinical course, preserves meaningful function, and ideally influences validated biological markers in a way that corresponds with patient benefit. That requires sufficiently large studies, credible sham control, prespecified outcomes, appropriate follow-up, and independent replication.

Current TMS research has not established that stimulation removes amyloid, prevents tau-related degeneration, regenerates lost neurons, prevents dementia, or permanently restores previously lost memory. Changes in EEG connectivity or network activity can be scientifically useful, but they are not by themselves proof of clinical disease reversal.

The FDA's 2019 advisory-panel review of the neuroAD system provides important context. That system paired TMS with cognitive training. Its pivotal United States study was prospective, multicenter, double-blind, and sham controlled, but it did not meet its prespecified primary effectiveness endpoint. FDA reviewers identified substantial uncertainty about clinically meaningful benefit.

That historical review applies to one integrated device and protocol, not to every form of TMS now being studied. It still demonstrates why favorable secondary analyses or subgroup findings cannot replace a successful prespecified primary endpoint.

Regulatory review and scientific publication answer related but different questions. A promising paper can justify more research. FDA clearance for a specific indication requires evidence sufficient for the applicable regulatory standard for that device and use.

I do not think the most useful question is simply, "Does TMS work for Alzheimer's disease?" That question combines too many different interventions and outcomes.

I would ask:

The literature becomes more understandable when those questions are asked consistently.

The studies below were selected for methodological relevance and for their contribution to a balanced understanding of the field. Inclusion does not mean that a paper proves effectiveness or that its findings apply to every device, protocol, or patient.

Current research does not establish that TMS reverses Alzheimer's disease, prevents progression from mild cognitive impairment to dementia, restores every form of lost memory, stops neurodegeneration, replaces established neurologic care, or produces a predictable benefit for an individual patient.

The cognitive TMS literature is neither empty nor settled. It includes randomized studies with encouraging results, a larger neutral multisite trial, mixed findings from combined cognitive-training protocols, and a regulatory history that reinforces the importance of prespecified outcomes and clinically meaningful benefit.

The strongest conclusion is that target, protocol, patient selection, outcome measure, and follow-up matter. The research justifies careful continued investigation. It does not justify advertising TMS as a proven method to reverse Alzheimer's disease, restore memory, or stop dementia.