Corey J. Keller
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Corey J. Keller has a total of 444 co-authors. Here are the top co-authors by number of shared publications:
- Christopher C. Cline
Department of Psychiatry and Behavioral Sciences, Stanford University
17 shared publications - Jessica Marie Ross
Stanford University
14 shared publications - Ashesh D. Mehta
The Feinstein Institutes for Medical Research, Northwell Health, Manhasset, NY, USA
10 shared publications Christopher C. Cline
Wu Tsai Neurosciences Institute, Stanford University, Stanford, USA
10 shared publications- Juha Gogulski
Department of Neuroscience and Biomedical Engineering, Aalto University School of Science, Espoo, Finland
10 shared publications - Sara Parmigiani
Department of Psychiatry & Behavioral Sciences, Stanford University School of Medicine, CA, USA
10 shared publications
Latest Publications
- Open Access
SafeLens: A Statistically Controlled Nonlinear Machine Learning Framework for Characterizing Hidden Dynamics in Event-Locked Neural Activity
Machine learning has advanced rapidly over the past decade, yet event-locked neural activity is still analyzed predominantly with traditional statistics and linear models. While traditional linear methods provide computationally tractable statistical control for pre-selected features or trial-averaged waveforms, they miss subtle nonlinear and trial-varying neural dynamics. Nonlinear models can capture these complex patterns, but their feature reliance is opaque and notoriously difficult to threshold statistically. We introduce SafeLens (Searchlight Analysis of Fluctuation Events, Leveraging Extensive Null Sampling) to combine the power of nonlinear machine learning with the rigorous false-positive control of traditional statistical inference. At each position of a sliding window, a random forest learns to discriminate event-locked activity from interleaved resting-state neural activity using multiscale temporal derivatives. Randomized feature ablation inverts slope polarity in held-out trials to map the random forest models’ reliance on specific rising versus falling excursions across the time-frequency space. SafeLens then establishes an empirical null by running the same pipeline on separate resting-state recordings, training forests to discriminate rest from rest. This empirical null translates each channel’s complex nonlinear resting dynamics into a model-matched noise floor on the map’s linear reliance scale, allowing a single threshold to control the family-wise error rate against the channel’s own background structure rather than a summary statistic of its baseline. When we injected simulated cortico-cortical evoked potentials (CCEPs) into resting intracranial EEG recordings, SafeLens recovered ~2.5 times as many polarity-resolved deflections as trial-average thresholding and the closest data-driven method, maintaining its advantage under trial-to-trial variability while yielding ~5–11 times fewer false components, ~2–5 times lower mean boundary error, and superior single-trial latency tracking as a timing prior. In eight patients, CCEPs remained detectable above chance until 1,040 ms after stimulation, compared with 570 and 455 ms for those benchmarks. Together, these findings establish SafeLens as a novel solution for enabling nonlinear models to uncover subtle, trial-varying neural dynamics under strict statistical control. Author Summary: Much of what we know about brain function comes from measuring how neural activity responds to specific events, from sensory stimuli and cognitive demands to targeted pulses of brain stimulation. Tracking these responses is fundamental for mapping how brain networks communicate in health and disease and for guiding interventions that modulate neural circuits. Because each response is buried in ongoing background activity and changes across repetitions, isolating it is difficult. Existing methods average repetitions inside short, hand-picked time windows, restricting researchers to seeing only the earliest, most stereotyped part of a response while leaving later, fluctuating activity invisible. Here, we introduce a new approach that combines the pattern-finding sensitivity of modern machine learning with the strict false-positive control of traditional statistics. Our method searches the period after each event without a preset window, using recordings of the brain at rest both to learn how a response differs from normal background activity and to measure what the model detects by chance alone. Using this tool, we show that human brain responses to electrical stimulation last over twice as long as previously recognized, revealing that brief perturbations trigger a longer cascade of neural activity than previously thought.
Open AccessSafeLens: A Statistically Controlled Nonlinear Machine Learning Framework for Characterizing Hidden Dynamics in Event-Locked Neural Activity
Machine learning has advanced rapidly over the past decade, yet event-locked neural activity is still analyzed predominantly with traditional statistics and linear models. While traditional linear methods provide computationally tractable statistical control for pre-selected features or trial-averaged waveforms, they miss subtle nonlinear and trial-varying neural dynamics. Nonlinear models can capture these complex patterns, but their feature reliance is opaque and notoriously difficult to threshold statistically. We introduce SafeLens (Searchlight Analysis of Fluctuation Events, Leveraging Extensive Null Sampling) to combine the power of nonlinear machine learning with the rigorous false-positive control of traditional statistical inference. At each position of a sliding window, a random forest learns to discriminate event-locked activity from interleaved resting-state neural activity using multiscale temporal derivatives. Randomized feature ablation inverts slope polarity in held-out trials to map the random forest models’ reliance on specific rising versus falling excursions across the time-frequency space. SafeLens then establishes an empirical null by running the same pipeline on separate resting-state recordings, training forests to discriminate rest from rest. This empirical null translates each channel’s complex nonlinear resting dynamics into a model-matched noise floor on the map’s linear reliance scale, allowing a single threshold to control the family-wise error rate against the channel’s own background structure rather than a summary statistic of its baseline. When we injected simulated cortico-cortical evoked potentials (CCEPs) into resting intracranial EEG recordings, SafeLens recovered ~2.5 times as many polarity-resolved deflections as trial-average thresholding and the closest data-driven method, maintaining its advantage under trial-to-trial variability while yielding ~5–11 times fewer false components, ~2–5 times lower mean boundary error, and superior single-trial latency tracking as a timing prior. In eight patients, CCEPs remained detectable above chance until 1,040 ms after stimulation, compared with 570 and 455 ms for those benchmarks. Together, these findings establish SafeLens as a novel solution for enabling nonlinear models to uncover subtle, trial-varying neural dynamics under strict statistical control. Author Summary: Much of what we know about brain function comes from measuring how neural activity responds to specific events, from sensory stimuli and cognitive demands to targeted pulses of brain stimulation. Tracking these responses is fundamental for mapping how brain networks communicate in health and disease and for guiding interventions that modulate neural circuits. Because each response is buried in ongoing background activity and changes across repetitions, isolating it is difficult. Existing methods average repetitions inside short, hand-picked time windows, restricting researchers to seeing only the earliest, most stereotyped part of a response while leaving later, fluctuating activity invisible. Here, we introduce a new approach that combines the pattern-finding sensitivity of modern machine learning with the strict false-positive control of traditional statistics. Our method searches the period after each event without a preset window, using recordings of the brain at rest both to learn how a response differs from normal background activity and to measure what the model detects by chance alone. Using this tool, we show that human brain responses to electrical stimulation last over twice as long as previously recognized, revealing that brief perturbations trigger a longer cascade of neural activity than previously thought.
- Open Access
- 2 November 2026
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 123 (45)
Transcranial magnetic stimulation to the dorsolateral prefrontal cortex modulates single-neuron activity in humans
Transcranial magnetic stimulation (TMS) to the dorsolateral prefrontal cortex (dlPFC) is an established treatment for depression, yet the fundamental question of how cortical stimulation influences single neurons in deep brain circuits critical for mood regulation remains unanswered. Using a rare opportunity to record from intracranial microelectrodes while TMS was delivered in four neurosurgical patients, we evaluated the responses of 185 single neurons. Artifact cleaning enabled single-neuron spikes to be resolved as early as 8 ms after stimulation. We found that single-pulse TMS to the left dlPFC elicited time-locked firing responses in 46% of neurons (26% facilitated only, 19% suppressed only, 1% both). These responses occurred within deep bilateral cortical and subcortical structures, including the putamen and thalamus. Single-pulse TMS tended to facilitate spiking in striato-thalamic putative inhibitory neurons from ~8 ms, peaking at ~80 to 100 ms, and decaying until ~600 ms. In contrast, putative excitatory neurons and limbic areas were suppressed with a slower time course lasting to ~1,000 ms. TMS-evoked potentials recorded from intracranial depth macroelectrodes to quantify network activity revealed that the modulation of responsive neurons was positively correlated with cortico-striato-thalamic network activity and anticorrelated with limbic network activity in one participant with sufficient network and unit sampling. These results suggest that dlPFC TMS recruits a high percentage of neurons spanning cortical and subcortical brain regions bilaterally. The recruitment pattern selectively involves increasing the firing rate of inhibitory neurons in executive control networks while suppressing limbic excitatory drive, providing a candidate cellular mechanism for how cortical stimulation modulates distributed brain networks.
- 2 November 2026
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 123 (45)
Open AccessTranscranial magnetic stimulation to the dorsolateral prefrontal cortex modulates single-neuron activity in humans
Transcranial magnetic stimulation (TMS) to the dorsolateral prefrontal cortex (dlPFC) is an established treatment for depression, yet the fundamental question of how cortical stimulation influences single neurons in deep brain circuits critical for mood regulation remains unanswered. Using a rare opportunity to record from intracranial microelectrodes while TMS was delivered in four neurosurgical patients, we evaluated the responses of 185 single neurons. Artifact cleaning enabled single-neuron spikes to be resolved as early as 8 ms after stimulation. We found that single-pulse TMS to the left dlPFC elicited time-locked firing responses in 46% of neurons (26% facilitated only, 19% suppressed only, 1% both). These responses occurred within deep bilateral cortical and subcortical structures, including the putamen and thalamus. Single-pulse TMS tended to facilitate spiking in striato-thalamic putative inhibitory neurons from ~8 ms, peaking at ~80 to 100 ms, and decaying until ~600 ms. In contrast, putative excitatory neurons and limbic areas were suppressed with a slower time course lasting to ~1,000 ms. TMS-evoked potentials recorded from intracranial depth macroelectrodes to quantify network activity revealed that the modulation of responsive neurons was positively correlated with cortico-striato-thalamic network activity and anticorrelated with limbic network activity in one participant with sufficient network and unit sampling. These results suggest that dlPFC TMS recruits a high percentage of neurons spanning cortical and subcortical brain regions bilaterally. The recruitment pattern selectively involves increasing the firing rate of inhibitory neurons in executive control networks while suppressing limbic excitatory drive, providing a candidate cellular mechanism for how cortical stimulation modulates distributed brain networks.
- Open Access
- 29 October 2026
- Published by Elsevier in Transcranial Magnetic Stimulation
A Pragmatic Framework for Comparing TMS Treatment Schedules in Real-World Clinical Practice
- 29 October 2026
- Published by Elsevier in Transcranial Magnetic Stimulation
Open AccessA Pragmatic Framework for Comparing TMS Treatment Schedules in Real-World Clinical Practice
- Open Access
- 19 October 2026
- Published by Springer Nature in Molecular Psychiatry
- p. 1-2
Correction: Neurophysiological signatures of Stanford Neuromodulation Therapy in treatment resistant depression
- 19 October 2026
- Published by Springer Nature in Molecular Psychiatry
- p. 1-2
Open AccessCorrection: Neurophysiological signatures of Stanford Neuromodulation Therapy in treatment resistant depression
- 18 September 2026
- Published by Springer Nature in Nature Mental Health
- Vol. 4 (11) , 1642-1659
Early brain functional connectivity changes induced by antidepressants and placebo
- 18 September 2026
- Published by Springer Nature in Nature Mental Health
- Vol. 4 (11) , 1642-1659
Early brain functional connectivity changes induced by antidepressants and placebo
- Open Access
iTBS induces a suppression-rebound pattern in human cortical excitability
Intermittent theta burst stimulation (iTBS) is a widely used form of brain stimulation, developed on the premise that it promotes long-term potentiation (LTP) and neural excitation. However, this is not validated in higher-order neocortex, and evidence suggests its excitatory properties may have been overstated. To address this, we recruited 20 neurosurgical patients to undergo sessions of intracranial iTBS to varied targets. We probed changes in cortical reactivity with single pulse electrical stimulation, finding that iTBS resulted in approximately 1 minute of decreased cortical reactivity in the theta and high-frequency ranges in regions strongly connected to the stimulation site. However, we observed a subsequent rise in reactivity leading up to the next iTBS session with parallel changes in resting state excitation/inhibition balance. Finally, we generalized these findings to 3 neurosurgical patients who underwent iTBS delivered by transcranial magnetic stimulation, demonstrating a similar suppression-rebound pattern. Our results suggest iTBS promotes a dynamic process of cortical suppression followed by rebound increases in excitability.
Open AccessiTBS induces a suppression-rebound pattern in human cortical excitability
Intermittent theta burst stimulation (iTBS) is a widely used form of brain stimulation, developed on the premise that it promotes long-term potentiation (LTP) and neural excitation. However, this is not validated in higher-order neocortex, and evidence suggests its excitatory properties may have been overstated. To address this, we recruited 20 neurosurgical patients to undergo sessions of intracranial iTBS to varied targets. We probed changes in cortical reactivity with single pulse electrical stimulation, finding that iTBS resulted in approximately 1 minute of decreased cortical reactivity in the theta and high-frequency ranges in regions strongly connected to the stimulation site. However, we observed a subsequent rise in reactivity leading up to the next iTBS session with parallel changes in resting state excitation/inhibition balance. Finally, we generalized these findings to 3 neurosurgical patients who underwent iTBS delivered by transcranial magnetic stimulation, demonstrating a similar suppression-rebound pattern. Our results suggest iTBS promotes a dynamic process of cortical suppression followed by rebound increases in excitability.
- Open Access
- 2 September 2026
- Published by Springer Nature in Molecular Psychiatry
- p. 1-13
Neurophysiological signatures of Stanford Neuromodulation Therapy in treatment resistant depression
Treatment-resistant depression (TRD) affects approximately 30% of patients with major depressive disorder. Stanford Neuromodulation Therapy (SNT), a high-dose intermittent theta-burst transcranial magnetic stimulation protocol, produces rapid antidepressant effects, but its neurophysiological mechanisms remain unclear. Here, we used longitudinal TMS-EEG to characterize the progressive neurophysiological changes induced by SNT, assess their site-specificity, and explore whether baseline neural markers are associated with clinical response. We conducted a double-blind, randomized, sham-controlled trial at Stanford University (2017–2018; analysis August 2024–October 2025) in 24 TMS-naïve participants with TRD (Montgomery-Åsberg Depression Rating Scale ≥20; ≥1 failed antidepressant trial). Participants were randomized to active (n = 12) or sham (n = 12) SNT, consisting of 10 sessions per day over 5 consecutive days targeting the left dorsolateral prefrontal cortex (90,000 pulses). TMS-EEG was acquired at two baseline sessions, before and after each treatment session, and at 1-month follow-up (14 TMS-EEG sessions in total). Active SNT progressively reduced cortical excitability at the treatment site, with significant decreases by day 3 in the early window component (−27.9%; P < 0.01), while no changes were observed at the vertex control site. Site-specific comparisons confirmed early window reductions only at the left dorsolateral prefrontal cortex (t₂₂ = −3.82; P < 0.001). SNT also selectively decreased estimated medial prefrontal source activity consistent with the subgenual anterior cingulate cortex (sgACC) across sessions (F₁₃,₂₂₂ = 4.93; P < 0.001), with effects persisting at 1-month follow-up. In an exploratory analysis in the active group (n = 12), higher baseline estimated sgACC source activity was associated with greater clinical improvement (r = −0.67; P = 0.023); although promising, the latter preliminary finding requires replication in larger, adequately powered samples before predictive utility can be established. These findings indicate that SNT induces progressive, site-specific cortical modulation and selective downstream effects on estimated sgACC source activity. Early cortical excitability changes represent candidate neurophysiological markers of SNT response, while the observed association between baseline sgACC activity and clinical outcome, while preliminary, motivates prospective investigation of subcortical source activity as a potential predictor of treatment response in larger trials. ClinicalTrials.gov Identifier: NCT03068715.
- 2 September 2026
- Published by Springer Nature in Molecular Psychiatry
- p. 1-13
Open AccessNeurophysiological signatures of Stanford Neuromodulation Therapy in treatment resistant depression
Treatment-resistant depression (TRD) affects approximately 30% of patients with major depressive disorder. Stanford Neuromodulation Therapy (SNT), a high-dose intermittent theta-burst transcranial magnetic stimulation protocol, produces rapid antidepressant effects, but its neurophysiological mechanisms remain unclear. Here, we used longitudinal TMS-EEG to characterize the progressive neurophysiological changes induced by SNT, assess their site-specificity, and explore whether baseline neural markers are associated with clinical response. We conducted a double-blind, randomized, sham-controlled trial at Stanford University (2017–2018; analysis August 2024–October 2025) in 24 TMS-naïve participants with TRD (Montgomery-Åsberg Depression Rating Scale ≥20; ≥1 failed antidepressant trial). Participants were randomized to active (n = 12) or sham (n = 12) SNT, consisting of 10 sessions per day over 5 consecutive days targeting the left dorsolateral prefrontal cortex (90,000 pulses). TMS-EEG was acquired at two baseline sessions, before and after each treatment session, and at 1-month follow-up (14 TMS-EEG sessions in total). Active SNT progressively reduced cortical excitability at the treatment site, with significant decreases by day 3 in the early window component (−27.9%; P < 0.01), while no changes were observed at the vertex control site. Site-specific comparisons confirmed early window reductions only at the left dorsolateral prefrontal cortex (t₂₂ = −3.82; P < 0.001). SNT also selectively decreased estimated medial prefrontal source activity consistent with the subgenual anterior cingulate cortex (sgACC) across sessions (F₁₃,₂₂₂ = 4.93; P < 0.001), with effects persisting at 1-month follow-up. In an exploratory analysis in the active group (n = 12), higher baseline estimated sgACC source activity was associated with greater clinical improvement (r = −0.67; P = 0.023); although promising, the latter preliminary finding requires replication in larger, adequately powered samples before predictive utility can be established. These findings indicate that SNT induces progressive, site-specific cortical modulation and selective downstream effects on estimated sgACC source activity. Early cortical excitability changes represent candidate neurophysiological markers of SNT response, while the observed association between baseline sgACC activity and clinical outcome, while preliminary, motivates prospective investigation of subcortical source activity as a potential predictor of treatment response in larger trials. ClinicalTrials.gov Identifier: NCT03068715.
- Open Access
- 13 June 2026
- Published by Springer Nature in Research Square
Sensory Entrained TMS (seTMS) enhances motor cortex plasticity
Neural excitability fluctuates with sensory events, creating windows of opportunity to enhance brain stimulation. Repetitive transcranial magnetic stimulation (TMS), including intermittent theta burst stimulation (iTBS), is a promising treatment for neurological and psychiatric disorders, but does not account for fluctuations in neural excitability, likely contributing to variable outcomes. Sensory Entrained TMS (seTMS) leverages sensorimotor oscillations to enhance corticospinal responses, but the sustained effects as a repetitive protocol are unknown. We extended seTMS to iTBS measuring motor-evoked potentials (MEPs) as a physiological readout in a randomized crossover study comparing standard iTBS with sensory entrained iTBS (se-iTBS, n = 20). se-iTBS more than doubled the MEP effect (55% vs. 26% MEP enhancement) and persisted for at least 30 minutes. Notably, more than 80% of participants showed larger responses with se-iTBS at all time points. se-iTBS may provide a robust and practical framework for optimizing TMS that bridges electrophysiological mechanisms and clinical applications.
- 13 June 2026
- Published by Springer Nature in Research Square
Open AccessSensory Entrained TMS (seTMS) enhances motor cortex plasticity
Neural excitability fluctuates with sensory events, creating windows of opportunity to enhance brain stimulation. Repetitive transcranial magnetic stimulation (TMS), including intermittent theta burst stimulation (iTBS), is a promising treatment for neurological and psychiatric disorders, but does not account for fluctuations in neural excitability, likely contributing to variable outcomes. Sensory Entrained TMS (seTMS) leverages sensorimotor oscillations to enhance corticospinal responses, but the sustained effects as a repetitive protocol are unknown. We extended seTMS to iTBS measuring motor-evoked potentials (MEPs) as a physiological readout in a randomized crossover study comparing standard iTBS with sensory entrained iTBS (se-iTBS, n = 20). se-iTBS more than doubled the MEP effect (55% vs. 26% MEP enhancement) and persisted for at least 30 minutes. Notably, more than 80% of participants showed larger responses with se-iTBS at all time points. se-iTBS may provide a robust and practical framework for optimizing TMS that bridges electrophysiological mechanisms and clinical applications.
- Open Access
Characterising and Minimising Step and Filtering Artifacts in TMS-EEG Recordings
Transcranial magnetic stimulation combined with electroencephalography (TMS-EEG) enables direct measurement of cortical reactivity via TMS-evoked potentials (TEPs). Interpretation of early TEP components however, is highly sensitive to stimulation and hardware-related artifacts. We identified and characterised a persistent, non-neural ‘step-drift’ artifact unexpectedly present in recent TMS-EEG recordings from our group. We show that the artifact is distinct from previously described TMS pulse and discharge/decay artifacts and likely reflects a hardware interaction phenomenon. We demonstrated that amplifier settings, but not TMS pulse shape, substantially influenced artifact expression, with DC-coupled recordings with no online high-pass filter reducing step amplitude compared with AC-coupled recordings with a high-pass filter. Simulations additionally revealed that filtering over the step-drift artifact introduced pronounced ringing and edge artifacts, highlighting the need to address this artifact prior to data processing. We propose a processing pipeline incorporating robust polynomial detrending and a modified Butterworth filter with autoregressive extrapolation that minimised TEP distortion in both simulated and real data containing the step-drift artifact. Together, these findings provide practical recommendations for both preventing and correcting step-drift artifacts and underscore the need for formal definition and routine recognition of this artifact to improve reproducibility and data quality in TMS-EEG research.
Open AccessCharacterising and Minimising Step and Filtering Artifacts in TMS-EEG Recordings
Transcranial magnetic stimulation combined with electroencephalography (TMS-EEG) enables direct measurement of cortical reactivity via TMS-evoked potentials (TEPs). Interpretation of early TEP components however, is highly sensitive to stimulation and hardware-related artifacts. We identified and characterised a persistent, non-neural ‘step-drift’ artifact unexpectedly present in recent TMS-EEG recordings from our group. We show that the artifact is distinct from previously described TMS pulse and discharge/decay artifacts and likely reflects a hardware interaction phenomenon. We demonstrated that amplifier settings, but not TMS pulse shape, substantially influenced artifact expression, with DC-coupled recordings with no online high-pass filter reducing step amplitude compared with AC-coupled recordings with a high-pass filter. Simulations additionally revealed that filtering over the step-drift artifact introduced pronounced ringing and edge artifacts, highlighting the need to address this artifact prior to data processing. We propose a processing pipeline incorporating robust polynomial detrending and a modified Butterworth filter with autoregressive extrapolation that minimised TEP distortion in both simulated and real data containing the step-drift artifact. Together, these findings provide practical recommendations for both preventing and correcting step-drift artifacts and underscore the need for formal definition and routine recognition of this artifact to improve reproducibility and data quality in TMS-EEG research.
- Open Access
Cognitive load modulates cortical excitability in the human prefrontal cortex
The prefrontal cortex dynamically adjusts its excitability to meet changing cognitive demands, yet our understanding of how these rapid state changes unfold in real time remains limited. Existing evidence relies predominantly on correlational neuroimaging and invasive electrophysiology in clinical populations, neither of which affords causal measurement in healthy humans. Here, using concurrent single-pulse TMS-EEG during a parametric cognitive multi-source interference task (MSIT) in 27 healthy participants, we causally probed left prefrontal excitability across conditions spanning a range of cognitive interference levels. Prefrontal excitability scaled systematically with cognitive load, peaking at intermediate incongruency levels where trial-type uncertainty was greatest, and differentiating high-from low-conflict trials at the single-subject level. Critically, prefrontal excitability and midfrontal theta power showed parallel state dependency patterns, linking causal perturbation of prefrontal circuit responsiveness to an established oscillatory marker of cognitive control. The observed excitability modulation was selective to the left prefrontal cortex: stimulation of right dlPFC, pre-SMA, and vertex in a separate control cohort (N=5) produced no load-dependent effects. These results position prefrontal excitability as a real-time, regionally selective physiological readout of how the human prefrontal cortex adapts to cognitive demand. Highlights: We evaluated the links between cognitive interference, prefrontal excitability, and intrinsic theta activity during a multi-source interference task Cognitive load, estimated based on behavioral performance, was greatest at at intermediate interference levels (33% congruent), where trial-type uncertainty was maximal Prefrontal excitability, measured via single-pulse TMS-evoked potentials, scaled with cognitive load Effects were observed at the single subject level and were specific to prefrontal stimulation Prefrontal excitability changes and midfrontal theta power showed parallel scaling with cognitive load, linking causal probing to established oscillatory markers This noninvasive technique provides causal measurement of prefrontal engagement during cognitive processing The methodology demonstrates potential for tracking brain state dynamics in both healthy cognition and neuropsychiatric disorders
Open AccessCognitive load modulates cortical excitability in the human prefrontal cortex
The prefrontal cortex dynamically adjusts its excitability to meet changing cognitive demands, yet our understanding of how these rapid state changes unfold in real time remains limited. Existing evidence relies predominantly on correlational neuroimaging and invasive electrophysiology in clinical populations, neither of which affords causal measurement in healthy humans. Here, using concurrent single-pulse TMS-EEG during a parametric cognitive multi-source interference task (MSIT) in 27 healthy participants, we causally probed left prefrontal excitability across conditions spanning a range of cognitive interference levels. Prefrontal excitability scaled systematically with cognitive load, peaking at intermediate incongruency levels where trial-type uncertainty was greatest, and differentiating high-from low-conflict trials at the single-subject level. Critically, prefrontal excitability and midfrontal theta power showed parallel state dependency patterns, linking causal perturbation of prefrontal circuit responsiveness to an established oscillatory marker of cognitive control. The observed excitability modulation was selective to the left prefrontal cortex: stimulation of right dlPFC, pre-SMA, and vertex in a separate control cohort (N=5) produced no load-dependent effects. These results position prefrontal excitability as a real-time, regionally selective physiological readout of how the human prefrontal cortex adapts to cognitive demand. Highlights: We evaluated the links between cognitive interference, prefrontal excitability, and intrinsic theta activity during a multi-source interference task Cognitive load, estimated based on behavioral performance, was greatest at at intermediate interference levels (33% congruent), where trial-type uncertainty was maximal Prefrontal excitability, measured via single-pulse TMS-evoked potentials, scaled with cognitive load Effects were observed at the single subject level and were specific to prefrontal stimulation Prefrontal excitability changes and midfrontal theta power showed parallel scaling with cognitive load, linking causal probing to established oscillatory markers This noninvasive technique provides causal measurement of prefrontal engagement during cognitive processing The methodology demonstrates potential for tracking brain state dynamics in both healthy cognition and neuropsychiatric disorders