BACKGROUND An important feature of motor-evoked potential (MEP) responses to transcranial magnetic stimulation (TMS) over the primary motor cortex (M1) is their substantial trial-to-trial variability. Ongoing brain activity at the time of TMS, such as the power and phase of the sensorimotor mu rhythm, influences MEP amplitude. It is unclear to what extent these endogenous brain states also shape the consistency of MEP amplitude across trials. OBJECTIVES We investigated trial-to-trial variability of MEP amplitude as a function of pre-stimulus mu dynamics. METHODS Twenty-nine healthy participants received 1200 single TMS pulses over the left M1 during simultaneous electroencephalography (EEG) recording; 10 participants completed two sessions to assess inter-session reproducibility. Trials were stratified based on pre-stimulus mu power, phase, and interhemispheric M1-M1 functional connectivity. Brain-state-resolved MEP variability was quantified using the coefficient of variation (CV) within subsets of trials defined by similar pre-stimulus mu dynamics. RESULTS Trial subsets characterized by high mu power were associated with reduced MEP variability, indicating more consistent corticospinal output. In contrast, mu phase did not significantly influence MEP variability at the group level. Mu power group-level effects were reproducible across sessions in an exploratory test-retest analysis and across experimental sites. CONCLUSIONS We show that inter-trial consistency of MEP amplitude reflects a structured, brain-state-dependent feature of the sensorimotor network. These findings contribute to our mechanistic understanding of trial-to-trial variability in TMS-evoked responses. This may provide a basis for future TMS approaches aimed at reducing response variability and improving the consistency of neuromodulatory effects.
Aim Repetitive Transcranial Magnetic Stimulation (rTMS) is a promising non-invasive neuromodulation technique for cognitive rehabilitation. We evaluated the cognitive effects of rTMS on neurological and psychiatric diseases to determine moderators of efficacy. Methods A meta-analysis of Randomized Controlled Trials (RCTs) comparing real versus sham rTMS to July 2025 was performed. Objective cognition reported as primary or secondary outcomes were extracted and Hedges’ g effect sizes computed. Univariate and multivariate three-level meta-analyses were performed to estimate pooled effect sizes and assess moderator influence. Results 748 measures from 137 studies (5972 participants) were extracted. Real rTMS produced a significant cognitive enhancement (g = 0.28, p < 0.0001, I2 = 37.3%). This was durable over 2-52 weeks follow-up. There was efficacy across all cognitive domains (g > 0.20, I2 = 35.8). Neurodegenerative (g = 0.35) and focal (g = 0.44) conditions demonstrated stronger effects than psychiatric conditions (g = 0.20). Although stimulation target was not a significant moderator overall, the Inferior Frontal Gyrus showed the largest effect (g = 0.45), with significant effects also at the left and right DLPFC and left parietal cortex (g = 0.29–0.45), all nodes of the Frontoparietal Network. 1-Hz stimulation produced a significantly larger effect (g = 0.39) than the 10-Hz reference (g = 0.19); 5-Hz, 20-Hz and iTBS were also significantly positive (g = 0.30–0.49). In multivariate models, adding 10 rTMS sessions improved cognitive outcomes (Δg = +0.077), whereas increasing the number of pulses per session trended toward reduced efficacy. All effects were confirmed in a left dorsolateral prefrontal cortex subgroup analysis. Subjective cognition was measured in only 6% of studies. A multivariate prediction tool is introduced: https://amsali.shinyapps.io/tms_cognition_prediction/ Conclusion This meta-analysis demonstrates the sustained efficacy of rTMS in enhancing cognitive function, with small-to-moderate effect sizes across diverse clinical populations and stimulation protocols. However, a significant gap remains regarding whether these objective neurocognitive gains translate into meaningful improvements in subjective cognition or functional outcomes.
Background: The Columbia University Autobiographical Memory Interview-Short Form (AMI-SF) is commonly used in electroconvulsive therapy (ECT) research, but reference data are lacking for determining whether decline observed after ECT exceeds expected retest change. Methods: Individual participant data from 99 depressed inpatients with no lifetime ECT exposure and 218 ECT recipients treated with right unilateral (RUL) or bitemporal (BT) ECT were analysed. Reliable decline was defined relative to depressed controls assessed over equivalent intervals using reliable change indices and standardised regression-based methods. Results: In depressed controls, the mean percentage of baseline autobiographical information reproduced was 76.9% at end-of-treatment, 75.1% at 3 months, 68.4% at 6 months, and 65.5% at 12 months. At end-of-treatment, approximately three in ten RUL patients and one in two BT patients showed reliable decline. At 12 months, corresponding proportions were one in ten after RUL and three in ten after BT ECT. On average, lower proportions at later follow-ups did not indicate recovery of memories not recalled at end-of-treatment; mean scores in ECT groups remained broadly unchanged during long-term follow-up while controls declined further. Conclusions: Autobiographical memory decline exceeding that expected in depressed controls was common acutely after ECT and remained substantially more frequent after BT than RUL ECT at one year follow-up. After RUL ECT, nine in ten patients did not meet reliable decline criteria at one-year follow-up. These psychometric thresholds describe statistically unusual change, not necessarily clinically meaningful retrograde amnesia.
