Materials and Methods

Participants

Thirty right‐handed or ambidextrous adults (17 female; mean age 26.1 years, SD = 9.6) volunteered to participate in this study. All participants had normal or corrected‐to‐normal vision, had no known sensory or motor dysfunctions, and were naïve to the hypotheses under investigation. No formal a priori sample size calculation was performed but was instead determined based on previous studies employing similar TMS paradigms (Leocani et al. 2000; Ibáñez et al. 2020). The single testing session lasted approximately 90 min. Before the start of a testing session, participants provided written informed consent and completed a safety questionnaire to screen for contraindications to the application of TMS (Rossi et al. 2011). This study and the protocols used were approved by the University of Ottawa Research Ethics Board (File number H‐05‐22‐8081) and were conducted in accordance with the seventh revision of the Declaration of Helsinki.

Task and Apparatus

Participants sat approximately 1.5 m in front of a 24″ computer monitor (Asus VG248; 144 Hz refresh) with their right forearm secured in a custom manipulandum, which restricted movement of the wrist to flexion and extension (Figure 1). Participants performed ballistic wrist extension and flexion movements within three separate RT paradigms: a simple RT task, a choice RT task, and a go/no‐go RT task. A fixation cross was shown in the middle of the computer screen, and each trial began with the removal of feedback from the previous trial. A blank screen was presented for 3–5 s, followed by the presentation of a visual warning signal. For the simple and go/no‐go tasks the visual warning signal consisted of the appearance of single box (7‐cm square, gray fill with a 3‐mm black outline) presented 10 cm to the right of the fixation cross. For the choice task, the visual warning signal was two identical boxes (as above) each presented 10 cm either side of the fixation cross. The warning signal was followed by a fixed foreperiod of 500 ms prior to the imperative stimulus. In the simple RT task, participants were instructed to perform a 20° right wrist extension upon presentation of the imperative go‐signal, which consisted of the on‐screen box filling in green. In the go/no‐go RT task, participants were instructed to execute a 20° right wrist extension if a green go‐signal was presented, but to withhold any motor response if the box turned red. In the choice RT task, participants were instructed to perform a 20° wrist extension if the box on the right of the fixation cross turned green or a 20° wrist flexion if the box on the left of the fixation cross turned green. Participants were instructed to execute their movements as soon as possible following the presentation of the go‐stimulus.

FIGURE 1: Schematic of experimental set up (top panel). Participants performed a right wrist extension (or flexion on choice trials) in response to a visual stimulus. TMS was applied over the left motor cortex and EMG data was collected from the right wrist extensors and flexors (blue). Schematic timeline, including the warning stimulus (−500 ms) and imperative stimulus (0 ms) for the various reaction time (RT) tasks, is shown at the bottom. TMS stimulation times within the time course of a trial are indicated with a TMS figure‐8 coil icon. Note the −2500‐ms TMS stimulation time was not included in the primary analysis (see Methods).

For each RT task, participants completed a single block of 70 experimental trials; block order was counterbalanced. Prior to each RT task block, participants completed 10 familiarization trials of that task. In the simple RT task, five additional catch trials were randomly included in which the imperative go‐stimulus did not occur. These were included in order to minimize anticipation of the go‐stimulus. In the go/no‐go and choice RT tasks, participants completed 35 randomized trials of each task alternative. Within each RT task block, participants were given the option to rest every 25 trials to minimize fatigue.

TMS

TMS was delivered during the reaction time tasks using a Magstim 2002 stimulator with a 70‐mm figure‐8 coil (Magstim Company Ltd., Whitland, UK) that was placed over the primary motor cortex (M1) representation of the right extensor carpi radialis (ECR) muscle. The target location was first approximated by finding the midpoint between a participant's nasion and inion (in the midsagittal plane) and left and right preauricular notches (in the frontal plane). From this point, a location 4 cm lateral (leftward) and 1 cm anterior was used as a starting point to determine the TMS location over M1 that generated the largest MEP in the right ECR (i.e., hotspot). During stimulation, the TMS coil was oriented such that the handle pointed backwards at approximately 45° relative to the midsagittal line of the head. Neuronavigation hardware and software (ANT Neuro Visor 2, Madison, WI) was used to save and replicate this location for all subsequent TMS applications. After the hotspot was located, the resting motor threshold (rMT) for the right ECR muscle was determined to the nearest 1% of stimulator output by finding the minimum intensity needed to elicit an MEP of 50 μV in five of 10 trials (Rossini et al. 2015). During the RT tasks, TMS was applied at 120% of each individual's rMT.

On each trial (except simple RT catch trials), TMS was applied over the motor hotspot for the ECR at one of six time points with respect to the onset of the imperative go‐stimulus: −500 ms (at warning stimulus onset), −300, −150, −100, −50, and 0 ms (coincident with the go‐signal). The TMS was only applied at one of these time points per trial, and each time point was repeated 10 times (random order). For the choice RT and go/no‐go tasks, there were two possible response alternatives; therefore, each TMS time occurred for each eventual response type on five separate trials. It was expected that response type (e.g., flexion vs. extension during choice RT task; “go” versus “no‐go” trials) should have no effect on MEP amplitude as the TMS application occurred prior to the imperative stimulus when the eventual required response was still unknown, and therefore, the preparatory/inhibitory processes would be expected to be similar on average at this time.

TMS was also applied on 10 trials for each task at −2500 ms (2000 ms prior to the warning stimulus) as it was assumed that MEP amplitude at this time would act as an index of “resting” baseline level activation. However, preliminary analysis indicated that there was no significant difference in MEP amplitude (p = 0.08) between the −2500 ms (between‐trial) application time and the −500‐ms application time (at the warning signal). In addition, the MEP amplitude was slightly larger at −2500 versus −500 ms, opposite to what would be expected for a baseline value. Given the brief 500‐ms foreperiod and relatively short inter‐trial interval (3–5 s), it appeared that participants in the current study likely remained in a task set, which resulted in elevated corticospinal excitability throughout the testing block. As such, trials where TMS was applied at −2500 ms were not able to be used as resting baseline and were discarded from analysis. To obtain a more accurate estimate of resting baseline corticospinal activation, 10 MEPs were collected in a subset of individuals (n = 12) immediately following determination of rMT with TMS output set to 120% of rMT for that individual. These MEPs were collected prior to providing any of the RT task instructions, and participants were instructed to sit quietly with their right arm resting comfortably in their lap. Inclusion of the −2500‐ms time point in the analyses did not meaningfully change the results or conclusions. Additional analyses including this time point are provided in the Supporting Information.

Recording Equipment

Surface electromyography (EMG) was collected from the ECR and flexor carpi radialis (FCR) muscles using bipolar preamplified double differential surface electrodes (Delsys, Bagnoli DE‐3.1; Delsys Inc., Natick, MA) connected to an external amplifier system (Delsys Bagnoli‐8) via shielded cabling. The location of EMG electrodes was cleaned using abrasive gel (Nuprep) and an alcohol swap to decrease electrical impedance, with electrodes placed parallel to the muscle fibers and attached to the skin via double‐sided adhesive tape. A reference electrode was placed on the medial epicondyle or olecranon of the right humerus. Raw band‐passed (20–450 Hz) EMG was digitally sampled at 4 kHz (PCIe‐6321, National Instruments) for 4.5 s starting 3 s prior to the imperative stimulus on each trial using a custom program written in LabVIEW software (National Instruments Inc.).

Data Reduction

EMG and MEP parameters were determined from the raw data using a custom LabVIEW analysis program. Integrated MEP amplitude (iMEP) was calculated by numerically integrating the rectified EMG values recorded from the ECR in a 30‐ms window beginning 15 ms following the time of TMS presentation. Integrated area under the curve has been argued to be a more valid amplitude measure for polyphasic MEPs, which are typically seen in nonhand effectors (Spampinato et al. 2023). Premotor reaction time (RT) was defined as the time from presentation of go‐stimulus to EMG onset of the agonist muscle and was only analyzed for the wrist extension task (go trials only) as it was the common movement task across all RT task blocks. EMG onset was determined as the first point at which full‐wave rectified and filtered EMG data (dual‐pass filtered using a 25‐Hz low‐pass second‐order elliptic filter) increased more than 2 standard deviations above the mean value recorded during the 500 ms prior to the warning signal and remained elevated for a minimum of 20 ms. EMG onsets were visually inspected and manually adjusted if necessary.

Practice trials were not included in the analyses nor were catch trials that occurred during the simple RT task block. Trials with the following errors were excluded from analysis: Anticipation (256 trials) was defined as trials where EMG onset of the agonist muscle occurred in less than 50 ms following a go‐stimulus. Slow RT (85 trials) was defined as EMG onset of the agonist muscle occurring more than 2.5 SD greater than the mean for each RT task. Movement errors (109 trials) consisted of incorrect responses during choice and go/no‐go RT tasks. Finally, trials were removed if peak‐to‐peak MEP amplitude was less than 50 μV (38 trials), or if there was excessive background EMG activity (RMS > 3× baseline and > 50 μV) prior to the TMS pulse (89 trials). Therefore, the final testing trial inclusion rate for analysis was 4943/5520 (89.5%).

Statistical Analysis

Premotor Reaction time was analyzed for the right wrist extension movement between RT tasks and TMS presentation times using a linear mixed effects model with interacting fixed factors of TMS time prior to the go‐signal (six levels: −500, −300, −150, −100, −50, and 0) and RT task (three levels: simple RT, choice RT, and go/no‐go RT), with random intercepts specified for participant, and random slopes specified for the effect of RT task condition (e.g., RT ~ TMStime × Task + (1 + Task|Participant)).

To determine if RT task type affected the degree to which corticospinal excitability was elevated above the resting state, integrated MEP amplitude was compared between resting baseline (measured in subset of 12 participants; see above) and when TMS was applied at the warning signal (i.e., “go” minus 500 ms) for each of the three RT conditions. These values were compared using a generalized linear mixed effects model with task condition (four levels: resting baseline, simple RT, choice RT, and go/no‐go RT) as a fixed factor. One benefit of linear mixed effects models is that they use maximum likelihood estimation, which accommodates unbalanced data by using all available observations without requiring participants to contribute data to every condition. Random intercepts were specified for participants, and the model used a gamma distribution for the residuals with a log link function (e.g., MEP amplitude ~ TaskCondition + (1|Participant), family = Gamma (link = “log”)). A gamma model was chosen because integrated MEP amplitudes exhibited strong positive skew (Ng and Cribbie 2017).

To investigate how corticospinal excitability evolved as the go‐signal approached, MEP amplitude in the wrist extensor was analyzed using a similar generalized linear mixed effects model with interacting fixed factors of TMS time prior to the go‐signal (6 levels: −500, −300, −150, −100, −50, and 0) and RT task (three levels: simple RT, choice RT, and go/no‐go RT). Random intercepts were specified for participants, and the model used a gamma distribution for the residuals with a log link function (e.g., MEP amplitude ~ TMStime × Task + (1|Participant), family = Gamma (link = “log”)).

Secondary Analyses

Previous research has shown a relationship between RT and the magnitude of MEP suppression whereby greater suppression is associated with shorter RT (Hannah et al. 2018). To investigate if this relationship was present in the current study, MEP suppression was calculated on a trial‐by‐trial basis. First, mean MEP amplitude was calculated for the trials in which TMS was applied at −500 ms (meanMEP‐500ms), then change in MEP amplitude from this mean was quantified for each trial at the 0 ms TMS time point (MEP0ms) using Equation (1) (with negative values indicating suppression relative to the 500‐ms time point). Change in MEP amplitude was then analyzed as a function of RT, across task conditions (three levels: simple RT, choice RT, and go/no‐go RT), with random intercepts specified for participant (e.g., MEP change ~ RT × Task + (1|Participant)).

(1) MEPchange=MEP0ms−meanMEP−500ms/meanMEP−500ms×100

Although trials with excessive background EMG were removed, MEP amplitude may nevertheless have been influenced by differences in background activation. To examine this possibility, for each trial a ratio was calculated between mean RMS EMG in the 100 ms prior to TMS and mean RMS EMG from a 100‐ms window just after the start of data collection (~3 s prior to the go‐signal). This ratio was analyzed using a linear mixed effects model with interacting fixed factors of TMS time and RT task, with random intercepts specified for participant (e.g., Ratio ~ TMStime × Task + (1|Participant)).

The R software package (Version 4.5.2; R Core Team 2025) was used for all analyses. The lme4 (Bates et al. 2026) and lmerTest (Kuznetsova et al. 2026) packages were used to fit the models and calculate p‐values. The anova() function was used to calculate p‐values for main effects and interactions using Satterthwaite's method. Type II Wald chi‐squared tests were used to calculate p‐values for main effects and interactions in gamma models. The emmeans (Lenth et al. 2026) package was used for pairwise contrasts with Tukey's HSD post hoc correction for multiple comparisons. The DHARMa (Hartig et al. 2024) and flexplot (Fife 2022) packages were used to assess assumptions of the models and to evaluate residuals. If the specified models exhibited convergence failures that could not be remedied via optimization algorithms, or if random slopes did not significantly improve model fit, the models were simplified by removing random slopes. All data are expressed as estimated marginal means with standard error or 95% confidence intervals. The sjPlot (Lüdecke et al. 2025) and flexplot (Fife 2022) packages were used to generate visuals.