Non-invasive brain stimulation in cognitive sciences and Alzheimer's disease

This review addresses how non-invasive brain stimulation (NIBS) can interrogate and modulate human cognition and whether it holds therapeutic value in Alzheimer’s disease (AD). Context: After the advent of electroconvulsive therapy and the development of TMS and low-intensity tES, NIBS has become a versatile tool for basic and clinical neuroscience due to favorable safety and efficacy. Purpose: to summarize basic mechanisms of TMS, tES, and emerging ultrasound-based methods; review their applications in cognitive sciences; and synthesize evidence for NIBS as a diagnostic, prognostic, and rehabilitative approach in AD. Importance: Given the multifactorial and heterogeneous pathophysiology of dementia and limited disease-modifying therapies, NIBS offers a non-pharmacologic avenue to modulate network-level dysfunctions, inform biomarkers of synaptic plasticity and connectivity, and enhance outcomes when combined with cognitive training.

The paper provides a narrative synthesis across: (1) Mechanisms of NIBS: TMS (single-pulse, rTMS, theta-burst stimulation; Hebbian ccPAS), tES (tDCS, tACS, tRNS; high-density/multifocal net-tDCS), and ultrasound-based methods (low-intensity tFUS, transcranial pulse stimulation, TPS), plus emerging kHz-frequency electrical stimulation. Safety and practical guidelines are summarized with seizures as the rare major adverse event. (2) Cognitive neuroscience in healthy controls: Foundational TMS work demonstrated causal disruption of visual perception and motor output timing; DLPFC’s role in working memory and decision networks (rTMS interference during delay vs decision phases); lateralization and age-related changes in episodic memory (rTMS over DLPFC); precuneus involvement in source memory (cTBS improved retrieval by reducing source errors); somatosensory lateralization; language-motor interactions (action words activate motor cortex). tES studies showed left DLPFC anodal tDCS improves complex verbal problem-solving; anterior temporal lobe montage enhances insight performance; posterior parietal tDCS reduces false recognitions and improves source discrimination. (3) AD and MCI: TMS biomarkers reveal increased motor cortex excitability and medial/frontal motor map shifts in AD; impaired LTP-like but preserved LTD-like plasticity; reduced cholinergic SAI in AD (vs FTD); LTP-like plasticity predicts cognitive decline. Therapeutically, rTMS over DLPFC, parietal cortex, and precuneus shows domain-specific benefits, often enhanced by cognitive training. tDCS over temporoparietal/temporal or frontal regions improves memory, language, attention, and may alter plasma Aβ oligomerization tendency; tACS at 40 Hz enhances gamma power, increases perfusion, and shows preliminary effects on p-tau burden. TPS studies report improved neuropsychological scores and network upregulation (memory networks), mood benefits with SN connectivity normalization, and structure-function correlations. tFUS to the hippocampus shows short-term cognitive gains and metabolic improvements without necessarily opening the BBB. Combined NIBS-imaging/EEG work elucidates connectivity modulation and supports NIBS-derived biomarkers. (4) Future directions: Integration with AI/ML for diagnosis, patient stratification, protocol optimization; BCI coupling to drive activity-dependent plasticity.

Narrative review of the literature. The article synthesizes physiological principles, safety guidelines, and empirical findings from experimental and clinical studies (including sham-controlled trials and pilot studies) on TMS, tES, and TUS in cognition and Alzheimer’s disease. No explicit systematic search strategy, inclusion/exclusion criteria, or meta-analytic methods are reported.

- Mechanistic insights: NIBS modulates cortical excitability and network connectivity locally and distally; patterned protocols (e.g., TBS, ccPAS) leverage synaptic plasticity principles. Multichannel/net-tDCS enables network-targeted modulation; tFUS/TPS allow deeper, focal targeting with millimetric resolution (TPS clinically approved for AD in some regions). Safety is favorable with rare seizures primarily in high-risk contexts. - Cognitive neuroscience (healthy): TMS can transiently disrupt visual perception and voluntary movement timing; DLPFC engagement differs for memory encoding/retrieval with age-related changes; precuneus involvement in source memory (cTBS reduced source errors); parietal tDCS improves item/source discrimination; insight and complex problem-solving enhanced by targeted tDCS montages. - AD pathophysiology and biomarkers: AD shows increased motor cortex excitability and altered motor maps (Ferreri 2003). LTP-like plasticity is impaired while LTD-like is preserved (Koch 2012); LTP-like measures correlate with cognitive severity (Di Lorenzo 2016) and predict decline (Motta 2018). SAI reduction differentiates AD from FTD (Di Lazzaro 2006). Connectivity disruptions span large-scale networks early in disease. - Therapeutic effects in AD/MCI: • rTMS: With cognitive training, improved global cognition (Brem 2020; Vecchio 2022). In mild AD (ADAS-Cog ≤30), active rTMS+training showed benefits (Sabbagh 2020, n=131). Precuneus-targeted rTMS over 24 weeks slowed cognitive/functional decline and maintained cortical excitability with increased gamma activity (Koch 2022); MRI-based pilot showed macro/micro-structural preservation and increased precuneus connectivity (Mencarelli 2024, n=16). Left parietal rTMS improved associative memory in amnestic MCI (Cotelli 2012). Language benefits reported with DLPFC rTMS (Cotelli 2011). • tDCS/tACS: Temporoparietal/temporal anodal tDCS improved recognition memory (Ferrucci 2008; Boggio 2012). Home-based bi-frontal tDCS (2 mA, 30 min, 12 weeks) improved language, memory, attention, and reduced plasma Aβ oligomerization tendency (Kim & Yang 2023). Left DLPFC anodal tDCS improved episodic memory, naming, general cognition, and mood in MCI (Fileccia 2019) and, combined with cognitive training, improved working memory and attention with 6-month maintenance (Rodella 2022). 40 Hz tACS increased temporal lobe perfusion (Sprugnoli 2021), enhanced gamma power, and showed preliminary p-tau reduction (~2%) in mesial temporal regions (Dhaynaut 2022). • TUS/TPS: TPS improved neuropsychological scores and upregulated memory networks on fMRI (Beisteiner 2020); reduced cortical atrophy correlating with cognitive gains (Popescu 2021); improved depression with SN–vmPFC connectivity normalization (Matt 2022); reduced neuropsychiatric symptoms at 30–90 days (Shinzato 2024). tFUS to hippocampus yielded mild improvements in memory/executive/global functions (Jeong 2021) and increased hippocampal glucose metabolism with improved verbal learning without BBB opening (Jeong 2022). - Modulation of connectivity: TMS–EEG and TMS/fMRI elucidate causal network effects; tES/fMRI shows network-specific modulation; TPS alters functional connectivity in disease-relevant networks. EEG small-world metrics and alpha/delta features may serve as diagnostic/prognostic biomarkers. - AI/BCI: Machine learning using TMS parameters (SAI, SICI/ICF, LICI) differentiates dementias with high accuracy (Benussi 2020). EEG-based ML predicts rTMS response (Kayasandik 2022) and tDCS responsiveness with region-specific features (Andrade 2023). Virtual brain modeling optimizes tDCS montages (Luppi 2024).

The review synthesizes converging evidence that NIBS can causally probe and therapeutically modulate large-scale brain networks underlying cognition. In AD, characteristic synaptic and connectivity dysfunctions (e.g., impaired LTP-like plasticity, widespread network disconnection) align with NIBS’ mechanisms of action, supporting its role as both a biomarker (plasticity indices, EEG connectivity) and an intervention (targeted rTMS of precuneus/DLPFC/parietal cortex; temporo-frontal tDCS/tACS; ultrasound-based TPS/tFUS). Coupling NIBS with cognitive training appears to amplify effects, consistent with activity-dependent plasticity. Multimodal combinations (EEG, MRI, PET) reveal local and remote network impacts and may stratify responders. Preliminary biomarker shifts (e.g., Aβ oligomerization tendency, p-tau signals, perfusion, glucose metabolism) underscore potential disease-modifying pathways, though validation is pending. AI/ML and BCI approaches promise personalization of targets and parameters, improved diagnostics, and adaptive, closed-loop neuromodulation. Overall, the findings address the central question by demonstrating that NIBS can enhance cognition, track network dysfunction, and potentially slow decline in AD, especially with network-guided and combined interventions.

NIBS provides a multidisciplinary, translational framework to interrogate cognition and to deliver individualized neuromodulation in AD. Evidence supports TMS-, tES-, and ultrasound-based protocols improving specific cognitive domains and neuropsychiatric symptoms, with biomarkers of plasticity and connectivity offering diagnostic and prognostic value. Integrating NIBS with AI-driven modeling, EEG/MRI guidance, and BCI may optimize target selection, dosing, and closed-loop delivery. Future research should prioritize large, longitudinal, sham-controlled trials; standardized protocols; responder stratification; multimodal biomarker validation; and comparative effectiveness across NIBS modalities and combinations with cognitive rehabilitation and pharmacotherapy.

- Narrative (non-systematic) review without explicit search strategy or bias assessment. - Heterogeneity across studies in patient populations (MCI vs AD severity), stimulation modalities, targets, dosing schedules, and outcome measures limits generalizability. - Many trials are small, single-center, and short-term; durability of effects and optimal maintenance schedules are not fully defined. - Limited human data for some emerging modalities (kHz-FS; variability in tFUS responses); TPS evidence includes pilot/uncontrolled designs. - Biomarker findings (e.g., p-tau, Aβ oligomerization tendency, perfusion/metabolism) are preliminary and need replication. - Safety is generally favorable, but standardized risk stratification and monitoring for specific populations (e.g., seizure risk, BBB manipulation) require further study. - Lack of head-to-head comparisons across NIBS techniques and limited guidance on individualized targeting outside research settings.