Abstract

We have recently shown that the efficiency in stopping a response, measured using the stop signal task, is related to GABAA-mediated short-interval intracortical inhibition (SICI) in the primary motor cortex. In this study, we conducted two experiments on humans to determine whether training participants in the stop signal task within one session (Experiment 1) and across multiple sessions (Experiment 2) would increase SICI strength. For each experiment, we obtained premeasures and postmeasures of stopping efficiency and resting-state SICI, that is, during relaxed muscle activity (Experiment 1, n = 45, 15 male participants) and SICI during the stop signal task (Experiment 2, n = 44, 21 male participants). In the middle blocks of Experiment 1 and the middle sessions of Experiment 2, participants in the experimental group completed stop signal task training, whereas control participants completed a similar task without the requirement to stop a response. After training, the experimental group showed increased resting-state SICI strength (Experiment 1) and increased SICI strength during the stop signal task (Experiment 2). Although there were no overall behavioral improvements in stopping efficiency, improvements at an individual level were correlated with increases in SICI strength at rest (Experiment 1) and during successful stopping (Experiment 2). These results provide evidence of neuroplasticity in resting-state and task-related GABAA-mediated SICI in the primary motor cortex after response inhibition training. These results also suggest that SICI and stopping efficiency are temporally linked, such that a change in SICI between time points is correlated with a change in stopping efficiency between time points.

INTRODUCTION

The ability to abort an inappropriate response is a crucial aspect of daily life. The time it takes for individuals to complete this process is estimated using the stop signal task through the stop signal RT (SSRT) measure (Logan & Cowan, 1984). Research has revealed impaired SSRT in conditions such as attention-deficit hyperactivity disorder (Oosterlaan, Logan, & Sergeant, 1998), obsessive–compulsive disorder (Menzies et al., 2007), and gambling disorder (Chowdhury, Livesey, Blaszczynski, & Harris, 2017). During stopping, areas such as the right inferior frontal gyrus and the pre-SMA have been hypothesized to engage the hyperdirect pathway of the BG. This causes inhibitory feedback over the primary motor cortex (M1), which then cancels the initial go command (Zhang et al., 2015; Chao, Luo, Chang, & Li, 2009; Stinear, Coxon, & Byblow, 2009; Li, Yan, Sinha, & Lee, 2008; Li, Huang, Constable, & Sinha, 2006; Aron, 2007). This inhibitory feedback is evident in the suppression of corticospinal excitability observed 100–150 msec after the onset of a stop signal, as measured using single-pulse TMS (spTMS; van den Wildenberg et al., 2009; Coxon, Stinear, & Byblow, 2006). Within M1, there is also evidence that GABAergic interneurons are activated during stopping, which is probed using the paired-pulse TMS (ppTMS) short-interval intracortical inhibition (SICI) protocol (Hermans et al., 2019; Lindberg et al., 2016; MacDonald, Coxon, Stinear, & Byblow, 2014; Coxon et al., 2006). SICI is observed when the amplitude of a motor evoked potential (MEP) elicited by a suprathreshold TMS pulse is reduced when a subthreshold pulse precedes it by an ISI of 1–6 msec, an effect thought to be mediated by GABAA neurotransmission (Ziemann, Lönnecker, Steinhoff, & Paulus, 1996).

We have recently extended the evidence for the involvement of SICI in stopping, showing that shorter SSRTs are correlated with stronger SICI (Chowdhury, Livesey, Blaszczynski, & Harris, 2020; Chowdhury, Livesey, & Harris, 2019a, 2019b; Chowdhury, Livesey, Blaszczynski, & Harris, 2018). In this study, we determined whether training participants in the stop signal task would lead to increased SICI strength. There is mixed evidence regarding the effects of stop signal task training on inhibitory control. Some studies have found no improvement in SSRT after training (Enge et al., 2014; Cohen & Poldrack, 2008), whereas others have shown improvements (Guerrieri, Nederkoorn, & Jansen, 2008; Logan & Burkell, 1986) as well as training-related neuroplasticity in prefrontal brain regions involved in response inhibition (Aasvik et al., 2017; Kühn et al., 2017; Berkman, Kahn, & Merchant, 2014; Manuel, Bernasconi, & Spierer, 2013). Whereas previous research has focused on changes in activity in prefrontal regions involved in stopping, our study determined whether stop signal training would lead to increased M1 GABAA-mediated SICI strength. Moreover, rather than only focusing on overall changes, we adopted an individual differences approach, determining whether changes in SICI are correlated with changes in SSRT.

To address our research question, we ran two experiments. In the first experiment, participants completed one session of stop signal task training, and we obtained premeasures and postmeasures of resting-state SICI (i.e., during relaxed muscle activity). This experiment would inform us as to whether training had the potential to influence SICI. The second experiment used two spaced interim sessions of training, which arguably should result in stronger improvements in inhibitory control compared with when sessions are massed together (Berkman et al., 2014; Enge et al., 2014). Moreover, we obtained premeasures and postmeasures of resting-state SICI and SICI measured during the stop signal task, using a similar methodology to our previous study (see Chowdhury et al., 2019a).

METHODS

Participants

For Experiment 1, there were 45 participants (15 male participants; 44 self-reported as right handed, and one self-reported as left-handed); 22 were in the experimental condition, and 23 were in the control condition. For Experiment 2, there was a different set of 44 participants (21 male participants; 42 self-reported as right-handed, and two self-reported as left-handed); 23 were in the experimental condition, whereas 21 were in the control condition. Three participants (two experimental, one control) had relatively high resting motor thresholds (RMTs; ∼70% of maximum stimulator output). For these participants, if SICI had been measured during the stop signal task, the high number of TMS pulses at a high intensity would cause the TMS coil to rapidly overheat. For this practical reason, these participants only completed the resting-state SICI component of the protocol.

Participants were undergraduate students from the University of Sydney School of Psychology. Informed consent was obtained, and the risk of an adverse response to TMS was assessed for all participants using a safety screening questionnaire (based on Rossi, Hallett, Rossini, Pascual-Leone, & The Safety of TMS Consensus Group, 2009). All procedures were approved by the University of Sydney Human Research Ethics Committee (protocol number 2016/718).

Training Protocol

Figure 1 shows the protocol for the experimental and control groups of each experiment. For both groups of Experiment 1, a resting-state SICI measure was followed by an SSRT measure. After this, the experimental group completed four additional blocks of the stop signal task, which took ∼15 min in total. The control group instead completed a choice RT task consisting entirely of go trials. After this, an SSRT measure was taken for both groups, followed by a resting-state SICI measure. For Experiment 2, there were four sessions, each session separated by 1–7 days for each participant. Sessions 1 and 4 were identical for both groups, involving an estimate of SSRT and a measurement of SICI at rest and during the stop signal task. Resting-state SICI, in this case, was measured after the initial SST measurement so that all SICI measures were collected at the same point in the experimental sequence. The groups differed in their experience of Sessions 2 and 3. The experimental group completed two sessions (each ∼20 min) of the stop signal task, whereas the control group received a task identical to the control group of Experiment 1. The mean number of days was 1.68 (range = 1–5) between Sessions 1 and 2, 2.23 (range = 1–5) between Sessions 2 and 3, and 2.86 (range = 1–7) between Sessions 3 and 4. Independent samples t tests showed that there were no significant differences between the experimental and control groups in these intersession intervals (ps > .05).

Figure 1. 

Training protocols for the experimental (Exp) and control (Cont) groups of Experiment 1 and Experiment 2.

Figure 1. 

Training protocols for the experimental (Exp) and control (Cont) groups of Experiment 1 and Experiment 2.

Stop Signal Task

Participants completed a visual stop signal task, based on the “STOP IT” program (Verbruggen, Logan, & Stevens, 2008), run in MATLAB (The MathWorks). The stimuli were presented against a mid-gray background. The two go signals were a black arrow pointing left or a black arrow pointing right. Participants responded by pressing one of two keys. Arrows had a height of 1.7° and a width of 1.5° of visual angle viewed from approximately 75 cm from the screen. On some trials, a stop signal (a blue square: 0.7° × 0.7° of visual angle) would occur after the arrow cue, indicating that participants were required to stop their initiated response. They were informed that, on some of the stop trials, the blue square would appear early and it would be easier to stop, but on other trials, the stop signal would occur late and it would be difficult to stop. Participants were also instructed to respond as quickly and accurately as possible to the arrow and not to delay their responses in anticipation of the stop signal. Such response slowing was monitored by observing RTs during each block. Data were excluded from a block if there was evidence of increasing go RTs as the task progressed (an increase of more than 1 msec per go trial; see Verbruggen, Chambers, & Logan, 2013), as response slowing to avoid inhibition failures leads to an inaccurate estimate of SSRT. We also instructed participants to keep go RTs as similar as possible throughout the entirety of the experiment. This was so that participants did not equate improvement in the task with becoming quicker at the task, given progressively faster go RTs across training have been argued to hinder SSRT improvement (Enge et al., 2014). On each trial, a black fixation dot (0.2° × 0.2° of visual angle) was presented for 500 msec, followed by a go signal for 1500 msec, which represented the maximal RT. On stop trials, the initial SSD was set at 250 msec. On trials where inhibition was successful, the SSD was increased by 25 msec. On trials where inhibition was unsuccessful, the SSD was decreased by 50 msec. This staircase ensured an overall stopping rate close to 66%. This staircase was chosen over the traditional 50% staircase to maximize the number of successful stop trials in the subsequent TMS phase of Experiment 2, so as to provide reliable measurement of MEPs and SICI related to stopping (see Chowdhury et al., 2019a; van den Wildenberg et al., 2009). As such, this staircase was also implemented in Experiment 1 to determine whether SSRT improvements could occur using this staircase. The integration method (Verbruggen et al., 2013) was used to calculate SSRT. This involved subtracting the mean SSD from the nth go RT, where n represents a point on the go RT distribution where the integral of the RT curve is equivalent to p(respond|signal).

Experiment 1

For Experiment 1, the choice task involved participants pressing the left arrow key with their right index finger when they saw a black arrow pointing left or pressing the right arrow key with their right middle finger when they saw a black arrow pointing right. As will be described in the TMS Protocol section, SICI was measured from the right index finger and not the right middle finger, which may have justified the use of a simple RT task involving the right index finger only. However, studies (e.g., Leunissen, Zandbelt, Potocanac, Swinnen, & Coxon, 2017) have shown that choice RT stop signal tasks (i.e., two or more go cues) lead to less biased estimates of stopping performance compared to simple RT stop signal tasks (i.e., one go cue), which is why the middle finger response was included in the task.

For both groups, the task began with 32 practice trials (where corrective feedback was presented). This was followed by a block of 96 experimental trials, with 25% of trials being stop trials. After this, the experimental group received four blocks of 96 experimental trials, whereas the control group received four blocks of the same task without stop trials. This choice RT task consisted of 83 go trials per block, with the remaining 13 trials being replaced by a blank intertrial interval of equivalent duration to each go trial. We chose 83 trials to match the number of go responses made in the experimental group given the ∼33% failed stopping rate. After these four blocks of the stop task or the choice RT task, all participants completed a final block of the stop task (with no further practice trials). SSRT was calculated for each block separately. Because of strategic slowing, stop signal task data were excluded for three control participants (one block each) and two experimental participants (two blocks for one, one block for the other). These participants were excluded from the overall pre–post SSRT analysis.

Experiment 2

For Experiment 2, the stop signal task was identical to that in Experiment 1, except for the following changes. For Session 1, for both groups, the task began with 24 practice trials (where corrective feedback was presented). For Session 4, there were no practice trials for either group. There were six blocks of 30 experimental trials in Sessions 1 and 4. The stop signal probability was increased to 33% rather than 25% to increase successful stop trial sampling during the TMS phase. The other difference from Experiment 1 was the use of a bimanual task rather than a unimanual task; participants were required to use the right and left index fingers to press the right enter key or the left control key depending on arrow direction. This was because the TMS phase aimed to observe MEP modulation as a function of which hand was selected for the response. The SSRT estimates from Sessions 1 and 4 were used to determine the critical SSD and TMS time point for the TMS stage (described in the next section). As such, this stage was called the “calibration stage.” For the two interim training sessions (Sessions 2 and 3), the experimental group received five blocks of 90 trials (900 trials in total over both training sessions). The starting SSD of Session 2 was determined based on the last SSD from Session 1. This SSD was continually adjusted using the staircase, and these adjustments continued on to Session 4. For the control group, participants completed a choice RT task, which involved responding to right and left arrows in the absence of stop trials. They received five blocks of 80 trials (to match the number of go responses to the experimental group), with the remaining 10 trials replaced by a blank intertrial interval of equivalent duration to each go trial. The starting SSD used for Session 4 was taken from the last SSD from Session 1. One experimental participant was responding with close to 0% stop accuracy during Session 2. Two control participants were responding with the wrong fingers on the wrong buttons during Sessions 2 and 3 (thus data were not recorded from these participants). Data from these participants were excluded from the analysis on interim SSRTs and go RTs.

TMS Protocol

EMG Preparation

Surface EMG recordings were obtained from the right first dorsal interosseous (FDI) muscle. The recording areas were cleaned using an abrasive pad and an alcohol swab. Three conductive adhesive hydrogel electrodes were then placed on the right and left hands: The positive electrode was placed on the left lateral side of the index knuckle, the negative electrode was placed on the belly of the FDI muscle, and the ground electrode was placed on the outer wrist bone. The EMG raw signal was amplified 1000× and band-pass filtered (1 Hz to 2 kHz, Powerlab 26T), digitized at a sampling rate of 4 kHz. We enabled a filter that detected mains noise (50 Hz) because of electrical equipment every second. The filter removed this noise from the signal whenever it was present. The signal was stored on a computer for visual display online and analysis offline using customized software.

TMS Preparation

Participants wore a cotton cap, used to mark the location of M1, and were seated with their head on a chin rest 75 cm from the computer monitor. Monophasic pulses of TMS were delivered over left M1 using a 70-mm figure-of-eight coil (D702) connected to a Bistim2 stimulator (Magstim). The coil was positioned at a tangent to the scalp, with the handle pointed backward, rotated 45° down from horizontal to deliver pulses with a posterior-to-anterior current direction. The optimal scalp position for coil placement over the left M1 was defined as the point where stimulation evoked the largest MEP on a single trial from the contralateral FDI muscle. Once a suitable location for stimulation was identified, the coil was locked in position using an articulated support (Manfrotto variable friction magic arm). Each participant's RMT was then determined by systematically changing stimulator output in increments/decrements of 1%, to find the minimum intensity that elicited an MEP of 50–100 μV in at least 5 of 10 trials (Rothwell et al., 1999). We obtained premeasures and postmeasures of RMT for both experiments. The value of the RMT for each participant was used to determine the intensities of the conditioning and test pulses (S1 and S2). In line with studies investigating optimal S1 and S2 intensities for SICI and for consistency with our previous studies (Chowdhury et al., 2018, 2019a, 2019b, 2020), we used 80% RMT for S1 (Ni, Gunraj, & Chen, 2007; Kujirai et al., 1993) and 110% RMT for S2 (Garry & Thomson, 2009).

Resting SICI Measurements

For Experiment 1, resting-state SICI was measured immediately before and after participants completed the six blocks of the behavioral task. To measure SICI, either spTMS or ppTMS was delivered in six blocks of 10 trials. Three of these blocks (30 trials in total) were spTMS trials, where S2 was delivered alone. Three blocks (30 trials in total) were ppTMS trials, where S1 and S2 were delivered with an ISI of 3 msec consistent with other SICI/response inhibition studies (Chowdhury et al., 2019a, 2019b; Hermans et al., 2019; Cirillo, Cowie, MacDonald, & Byblow, 2018; Chowdhury et al., 2018; Kratz et al., 2009; Waldvogel et al., 2000). The blocks of each type (spTMS or ppTMS) were randomly intermixed. Similar to our previous studies (Chowdhury et al., 2018, 2019a, 2020), the TMS pulses were delivered while participants performed a simple task that required them to keep track of the number of vowels among a stream of letters presented on the computer. This task was used to keep participants occupied and alert while resting-state SICI measurements were taken, as any loss of alertness or fatigue may influence MEP amplitudes. The use of a similar counting task did not influence resting-state SICI measurements (Thomson, Garry, & Summers, 2008). There was no relationship between when TMS pulses were delivered and the presentation of any specific type of letter. There was a random interval of 5–6 sec between pulses. For Experiment 2, resting-state SICI was measured after the pre- or post-SSRT measure was taken. SICI was probed in the same manner as in Experiment 1, except with 20 spTMS trials and 20 ppTMS trials (i.e., four blocks in total).

SICI Measurement during the Stop Task

In Experiment 2, after the resting-state SICI protocol, participants completed the stop signal task while receiving TMS pulses to M1. This task was programmed in Psychopy (Peirce, 2007). The stop signal task consisted of the same stimuli as in the calibration stage (black right or left arrows as the go signals and a blue square as the stop signal). However, instead of using a staircase, the SSD was fixed to the critical delay calculated from the calibration phase, which represented a 66% chance of stopping for each participant. Because the refresh rate of the monitor was 60 Hz, the SSD was rounded to the nearest 16.67 msec so that TMS probes could be precisely timed from the onset of the stop signal. In total, there were 360 trials presented in 10 blocks of 36 trials: Five of these blocks involved delivery of spTMS, whereas the other five involved the delivery of ppTMS, with a 3-msec ISI. spTMS and ppTMS blocks occurred in a randomly determined order. TMS was delivered on go trials (20 per block, 200 in total), where the go cue signaled either a right-hand response (100 in total) or a left-hand response (100 in total). Thus, for go trials, there were 50 spTMS trials and 50 ppTMS trials each for right- and left-hand conditions. TMS was also delivered on stop trials (10 per block, 100 in total), where half of the trials were right-hand stop trials (50 in total) whereas the other were left-hand stop trials (50 in total). Thus, for stop trials, there were 25 spTMS trials and 25 ppTMS trials each for right- and left-hand conditions. MEPs (30 single pulse, 30 paired pulse) were also obtained during fixation on a sample of additional go trials (four per block, 40 in total) and stop trials (two per block, 20 in total) wherein TMS was not delivered during the trial. On these trials, TMS was delivered after the fixation dot but before the go cue. When TMS was delivered during the trial, a single or paired pulse was delivered 200 msec after the onset of the stop signal on stop trials. We chose the 200-msec time point because our previous study (Chowdhury et al., 2019a) showed that the strongest SICI differences as a function of stopping performance were observed around this time point. TMS was fixed to the corresponding time point on go trials (i.e., equivalent to the critical SSD + 200 msec relative to go signal onset). On fixation trials, the TMS pulse was delivered 200 msec after the onset of the fixation dot to control for the impact of onset of a visual stimulus on MEPs. The interval between the offset of the go cue and the start of the next trial was 4–5 sec. Figure 2 shows a schematic of go and stop trials as well as the time points where TMS was delivered.

Figure 2. 

Schematic of go and stop trials where spTMS or ppTMS was delivered in Experiment 2. On trials where TMS was delivered during the trial, pulses were delivered 200 msec after the stop signal on stop trials and were fixed at an equivalent time for go trials. On fixation trials, TMS was delivered 200 msec after the onset of the fixation dot. The stop signal delay was set as the delay that represented a 66% chance of inhibition as estimated from the initial calibration phase.

Figure 2. 

Schematic of go and stop trials where spTMS or ppTMS was delivered in Experiment 2. On trials where TMS was delivered during the trial, pulses were delivered 200 msec after the stop signal on stop trials and were fixed at an equivalent time for go trials. On fixation trials, TMS was delivered 200 msec after the onset of the fixation dot. The stop signal delay was set as the delay that represented a 66% chance of inhibition as estimated from the initial calibration phase.

MEP Analysis

Custom software (available at github.com/nicolasmcnair/MEPAnalysis) was developed in Psychopy to calculate the peak-to-peak amplitude of each MEP. SICI was calculated as the ratio of the paired-pulse MEP (ppMEP), over the single-pulse MEP (spMEP). This ratio was then log transformed to correct for the positive skew that is inherent in the distribution of ratios of two positive numbers (see Chowdhury et al., 2018, 2019a, 2020; Tran, Harris, Harris, & Livesey, 2019; Poole, Mather, Livesey, Harris, & Harris, 2018). MEPs were excluded if there was head movement (i.e., if the center of the TMS coil visibly moved away from the marked hotspot). When this occurred, the TMS coil was readjusted over the hotspot with roughly the same coil orientation and angle. For Experiment 1, visible head movement occurred for one block for one participant, and this block was excluded. For Experiment 2, this occurred for four participants, such that one block was excluded for each of these participants. Given Experiment 2 involved measuring MEPs during a behavioral task, an exclusion criterion was required to eliminate trials with confounding EMG activity (no criteria were required for Experiment 1 given muscles were relaxed during SICI measurements). For all trial types, MEPs were also excluded if there was any EMG activity greater than 100 μV peak to peak (Chowdhury et al., 2019a; Seet, Livesey, & Harris, 2019; Chiu, Aron, & Verbruggen, 2012) in the 50-msec time window (Chowdhury et al., 2019a; Coxon et al., 2006) preceding the TMS pulse. Note that, for all analyses, only successful stop trials were analyzed. Given the stopping success rate and MEP exclusion criteria, across Sessions 1 and 4, there was a mean of 13.34 (SD = 5.8) spMEPs and 13.33 (SD = 5.37) ppMEPs remaining for successful right-hand stop trials and 15.67 (SD = 4.60) spMEPs and 15.65 (SD = 4.36) ppMEPs remaining for successful left-hand stop trials. Moreover, there was a mean of 26.15 (SD = 11.08) spMEPs and 25.67 (SD = 10.72) ppMEPs remaining for right-hand go trials, and 42.80 (SD = 10.09) spMEPs and 42.71 (SD = 9.76) ppMEPs remaining for left-hand go trials. Two paired-samples t tests showed no differences in stop success rate between paired-pulse blocks and single-pulse blocks for selected and nonselected stop trials, respectively (ps > .05), suggesting pulse type did not significantly influence stopping accuracy.

Experimental Design and Statistical Analysis

Behavioral Measures

To confirm that initial SSRTs did not differ between the experimental and control groups, Bayesian independent samples t tests were conducted, with a Bayes factor (BF01) greater than 1 indicating more support for the null hypothesis relative to the alternative hypothesis (Rouder, Speckman, Sun, Morey, & Iverson, 2009). To observe whether post-SSRTs improved in the experimental versus control group, a 2 (Time: pre vs. post) × 2 (Training: experimental vs. control) mixed-model ANOVA was conducted. For Experiment 2, we conducted a 2 (Time: pre vs. post) × 2 (Training: experimental vs. control) × 2 (Stage: calibration vs. TMS stage) mixed-model ANOVA on go RT and stop accuracy (α corrected to .025 to correct for multiple comparisons), with a specific focus on the Stage factor (a) to ensure that the TMS phase was still leading to an overall 66% success rate and (b) to determine whether RTs were significantly slower in the TMS stage compared to the calibration stage. For the middle four blocks of Experiment 1 and the middle 10 blocks of Experiment 2, we ran a 2 (Training: experimental vs. control) × 4 or 10 (Time: four blocks or 10 blocks) ANOVA on go RTs and SSRT (Training factor was not present for SSRT). These analyses tested whether SSRT changed in Sessions 2 and 3 and whether there was evidence of proactive inhibition in the experimental group (i.e., slowing of go RTs because of presence of stop trials). Alpha (α) was adjusted to .025 to correct for multiple comparisons.

RMT and spMEP Amplitudes

For Experiment 1, a 2 (Time: pre vs. post) × 2 (Training: experimental vs. control) mixed-model ANOVA was conducted to confirm that resting spMEP amplitude and RMT did not differ between groups (α corrected to .025 to correct for multiple comparisons). This same ANOVA was run on RMT, resting spMEP amplitude, and fixation spMEP amplitude for Experiment 2 (α corrected to .0167 to correct for multiple comparisons). In addition, for Experiment 2, we ran a 2 (Time: pre vs. post) × 2 (Training: experimental vs. control) × 2 (Trial: go vs. stop) × 2 (Response hand: left vs. right) mixed-model ANOVA to (a) confirm that spMEP amplitudes and changes in spMEP amplitudes did not differ between groups and (b) determine spMEP amplitude modulation as a function of trial type and response hand.

Resting-state SICI

To confirm that resting-state SICI from Session 1 did not differ between the experimental and control groups of each experiment, a Bayesian independent samples t test was conducted. For both experiments, a 2 (Time: pre vs. post) × 2 (Training: experimental vs. control) mixed-model ANOVA was conducted to determine whether resting-state SICI increased in strength in the experimental versus control group.

SICI Modulation during the Stop Signal Task

For Experiment 2, a 2 (Time: pre vs. post) × 2 (Training: experimental vs. control) mixed-model ANOVA was conducted on SICI during fixation to determine whether group differences existed in SICI measured during the intertrial interval and whether this interacted with time. To confirm that SICI measured during go or stop trials did not differ between the experimental and control groups from Session 1, Bayesian independent samples t tests were conducted on each condition (right-hand stop, left-hand stop, right-hand go, left-hand go). A 2 (Time: pre vs. post) × 2 (Training: experimental vs. control) × 2 (Trial: go vs. stop) × 2 (Response hand: left vs. right) mixed-model ANOVA was conducted on SICI. This ANOVA was conducted to (a) understand how SICI was modulated as a function of trial type and response hand across all participants, particularly whether SICI was stronger on stop trials relative to go trials and whether this effect was greater when participants prepared to respond with the right hand (where MEPs were measured) versus the left hand, and (b) determine whether SICI increased after training and whether this effect interacted with trial type and response hand. Note that, for all ANOVAs, epsilon corrections were made when sphericity assumptions were violated.

Pearson Correlation Analyses

For both experiments, we determined whether changes in SICI between pre and post were correlated with changes in SSRT. We also examined the correlations between the following variables: premeasures and postmeasures of SSRT, resting-state SICI, and mean SICI during all successful stop trials (Experiment 2). These correlation analyses would provide information as to (a) whether SICI changes were related to the key response inhibition measure, (b) whether SICI and SSRT were correlated at each time point, and (c) the test–retest reliability of the SICI and SSRT measures. On the basis of a recent study investigating the reliability of TMS parameters (Davila-Pérez, Jannati, Fried, Cudeiro Mazaira, & Pascual-Leone, 2018), correlations greater than .75 were considered to reflect high test–retest reliability, correlations between .5 and .75 were considered to reflect moderate test–retest reliability, and correlations less than .5 were considered to reflect low test–retest reliability. Data were analyzed using SPSS (Version 26) and JASP (Version 0.11.1).

RESULTS

Experiment 1 Results

Behavioral Data

Figure 3 shows pre- and post-SSRT for the experimental and control groups, for Experiment 1. Table 1 shows further behavioral data (go RT, stop accuracy, SSD) across all participants at pre and post, for Experiment 1. An independent samples t test revealed no significant difference in initial SSRT between the experimental and control groups, t(39) = 0.614, p = .543, BF01 = 2.812. Results of the 2 (Time: pre vs. post) × 2 (Training: experimental vs. control) mixed-model ANOVA revealed a significant main effect of Time, F(1, 38) = 4.29, p = .045, such that post-SSRTs were shorter than pre-SSRTs. There was no main effect of Training, F(1, 38) = 0.011, p = .917, and no significant interaction between Time and Training, suggesting that training did not influence the magnitude of reduction in SSRT between pre and post, F(1, 38) = 0.781, p = .382. For the middle four blocks, a one-way ANOVA on SSRT revealed no significant effect of Time, F(3, 60) = 1.033, p = .385, such that the experimental group did not show a significant change in SSRT across the four blocks of training. A 2 (Training: experimental vs. control) × 4 (Time: 4 blocks) mixed-model ANOVA on go RT revealed no significant change in go RTs across the four blocks, F(2.282, 95.864) = 1.57, p = .20, no significant difference in go RTs between the experimental and control groups, F(1, 42) = 0.001, p = .976, and no significant interaction between Training and Time, F(2.282, 95.864) = 1.056, p = .37. Thus, the results of the middle blocks suggest SSRT did not change across the four blocks, nor was there evidence of any proactive inhibition, because go RTs were not significantly faster in the absence of stop trials.

Figure 3. 

SSRT (± SEM) at pre and post for the experimental and control groups, for Experiment 1.

Figure 3. 

SSRT (± SEM) at pre and post for the experimental and control groups, for Experiment 1.

Table 1. 
Means and Standard Deviations for Behavioral Measures, RMT, and spMEP Amplitude for Experiment 1 at Pre and Post, for Each Group
 PrePost
ExperimentalControlExperimentalControl
Go RT (msec) 410.91 (73.72) 403.78 (80.48) 397.39 (64.72) 386.64 (78.60) 
SSD (msec) 127.45 (58.04) 126.47 (70.96) 131.16 (63.08) 115.03 (74.14) 
Stop accuracy % 65.42 (5.59) 65.39 (5.88) 65.34 (3.94) 65.13 (4.44) 
RMT (%MSO) 52.32 (9.62) 50.96 (7.89) 52.27 (8.70) 51.13 (7.84) 
Resting spMEP amplitude (μV) 655.81 (487.39) 814.19 (698.44) 780.86 (638.90) 795.42 (684.38) 
 PrePost
ExperimentalControlExperimentalControl
Go RT (msec) 410.91 (73.72) 403.78 (80.48) 397.39 (64.72) 386.64 (78.60) 
SSD (msec) 127.45 (58.04) 126.47 (70.96) 131.16 (63.08) 115.03 (74.14) 
Stop accuracy % 65.42 (5.59) 65.39 (5.88) 65.34 (3.94) 65.13 (4.44) 
RMT (%MSO) 52.32 (9.62) 50.96 (7.89) 52.27 (8.70) 51.13 (7.84) 
Resting spMEP amplitude (μV) 655.81 (487.39) 814.19 (698.44) 780.86 (638.90) 795.42 (684.38) 

RMT and spMEP Amplitudes

Table 1 shows the mean RMT and spMEP amplitudes at pre and post, for Experiment 1. A 2 (Time: pre vs. post) × 2 (Training: experimental vs. control) mixed-model ANOVA on spMEP amplitudes and RMT revealed no significant main effects of Time and Training, nor were there any significant interactions between these variables (all ps > .025). This provides evidence that spMEP amplitude sizes and RMT were matched across groups.

Resting-state SICI

Figure 4 shows pre and post resting-state SICI for the experimental and control groups for Experiment 1. An independent samples t test revealed no significant difference in initial resting-state SICI between the experimental and control groups, t(43) = 0.445, p = .659, BF01 = 3.13. Results of the 2 (Time: pre vs. post) × 2 (Training: experimental vs. control) mixed-model ANOVA on resting-state SICI revealed no significant main effect of Time, F(1, 43) = 1.061, p = .309, and no significant main effect of Training, F(1, 43) = 0.093, p = .761. There was a significant interaction between Time and Training, such that there was a significant increase in resting-state SICI from pre to post in the experimental group relative to the control group, F(1, 43) = 7.701, p = .008. Post hoc t tests on this interaction showed there was stronger SICI at post versus pre for the experimental group, t(21) = 2.565, p = .018, but no difference in SICI between post and pre for the control group, t(22) = −1.296, p = .208, suggesting the interim choice RT task did not lead to a change in SICI.

Figure 4. 

Resting-state SICI (± SEM) at pre and post for the experimental and control groups, for Experiment 1. The ordinate axis represents the log-transformation of the ratio of the ppMEP-to-spMEP amplitude. A more negative number represents stronger SICI.

Figure 4. 

Resting-state SICI (± SEM) at pre and post for the experimental and control groups, for Experiment 1. The ordinate axis represents the log-transformation of the ratio of the ppMEP-to-spMEP amplitude. A more negative number represents stronger SICI.

Correlation Analyses

Figure 5 shows the difference in pre- and post-resting SICI plotted against the difference in pre- and post-SSRT, for Experiment 1. The correlation between these two was r(38) = .619, p < .001 (experimental group: r(18) = .72, p < .001; control group: r(18) = .50, p = .025), such that an increase in SSRT correlated with a strengthening of SICI. These correlations were still significant when excluding the data points from Figure 5 that showed considerably larger SICI increases relative to other data points (i.e., a SICI change value of larger than −0.2). Table 2 shows the pattern of correlations between resting SICI, spMEP, and SSRT at pre and post, for Experiment 1.

Figure 5. 

Difference in resting-state SICI between pre and post graphed against difference in SSRT between pre and post, for Experiment 1.

Figure 5. 

Difference in resting-state SICI between pre and post graphed against difference in SSRT between pre and post, for Experiment 1.

Table 2. 
Pattern of Correlations between Pre and Post SICI, spMEP, and SSRT Measures for Experiment 1 across All Participants
 Pre Resting SICIPost Resting SICIPre-SSRTPost-SSRTPre-spMEP
Post resting SICI .73**         
Pre-SSRT .38* .28       
Post-SSRT .05 .38* .57**     
Pre-spMEP −.05 .09 −.17 .009   
Post-spMEP −.18 −.08 −.21 −.07 .73** 
 Pre Resting SICIPost Resting SICIPre-SSRTPost-SSRTPre-spMEP
Post resting SICI .73**         
Pre-SSRT .38* .28       
Post-SSRT .05 .38* .57**     
Pre-spMEP −.05 .09 −.17 .009   
Post-spMEP −.18 −.08 −.21 −.07 .73** 
*

Significant at the .05 level.

**

Significant at the .01 level.

Overall, correlation analyses provide evidence of a link between changes in SICI and changes in SSRT overtime—an increase in SICI between pre and post correlates with a shortening of SSRT between pre and post, with this correlation being numerically stronger for those who receive stop signal task training. Moreover, both SICI and SSRT showed moderate test–retest reliability.

Experiment 2 Results

Behavioral Measures

Figure 6 shows pre- and post-SSRT for the experimental and control groups for Experiment 2. Table 3 shows further pre and post behavioral data (SSD, go RT at each stage, stop accuracy at each stage) across all participants, for Experiment 2. An independent samples t test revealed no significant difference in initial SSRT between the experimental and control groups, t(43) = 0.018, p = .861, BF01 = 3.317. The 2 (Time: pre vs. post) × 2 (Training: experimental vs. control) mixed-model ANOVA on SSRT revealed that the main effect of Time was not significant, F(1, 42) = 3.895, p = .055. The interaction between Time and Training also fell short of statistical significance, such that the difference in pre- and post-SSRT was not significantly larger for the experimental group relative to the control group, F(1, 42) = 2.86, p = .098. However, when running a paired-samples t test comparing pre- and post-SSRT for each group, there was a significant improvement in SSRT for the training group, t(22) = 2.96, p = .007, but not the control group, t(20) = 0.178, p = .861. When running the 2 (Time: pre vs. post) × 2 (Training: experimental vs. control) × 2 (Stage: calibration vs. TMS phase) mixed-model ANOVA on RT and stop accuracy, there was no significant main effect of Time, Stage, and Training, nor were there any significant interactions between these variables (all ps > .025). One exception was an interaction between Time, Stage, and Training on accuracy, F(1, 39) = 5.688, p = .022, such that accuracy for the calibration and TMS stage became more similar for the control group at post versus pre, when compared to the experimental group. Overall, this analysis suggests that (a) RTs and accuracy did not, on average, significantly change between the calibration phase and the TMS phase (i.e., use of the fixed SSD in the TMS phase did not change behavioral performance) and (b) there were no significant overall group differences in RT or accuracy in either stage.

Figure 6. 

SSRT (± SEM) at pre and post for the experimental and control groups, for Experiment 2.

Figure 6. 

SSRT (± SEM) at pre and post for the experimental and control groups, for Experiment 2.

Table 3. 
Means and Standard Deviations for Behavioral Measures (During the Calibration and TMS Stage), and RMT and spMEP Amplitudes for Experiment 2 at Pre and Post, for Each Group
 PrePost
ExperimentalControlExperimentalControl
Go RT at calibration stage (msec) 421.76 (101.82) 415.16 (102.00) 413.27 (128.76) 381.60 (74.26) 
Go RT at TMS stage (msec) 452.09 (128.43) 424.98 (108.38) 428.05 (128.41) 400.20 (62.16) 
SSD at calibration stage (msec) 152.28 (99.66) 147.08 (103.02) 170.62 (130.19) 126.43 (78.85) 
Stop accuracy % at calibration stage 65.00 (0.04) 65.63 (0.02) 65.80 (0.03) 64.80 (0.05) 
Stop accuracy % at TMS stage 67.06 (0.20) 65.04 (0.15) 58.68 (0.18) 67.43 (0.14) 
RMT (%MSO) 49.46 (9.81) 48.62 (8.92) 49.89 (10.79) 48.29 (9.29) 
Resting spMEP amplitude (μV) 499.76 (575.80) 804.70 (833.46) 615.91 (717.06) 744.54 (697.67) 
 PrePost
ExperimentalControlExperimentalControl
Go RT at calibration stage (msec) 421.76 (101.82) 415.16 (102.00) 413.27 (128.76) 381.60 (74.26) 
Go RT at TMS stage (msec) 452.09 (128.43) 424.98 (108.38) 428.05 (128.41) 400.20 (62.16) 
SSD at calibration stage (msec) 152.28 (99.66) 147.08 (103.02) 170.62 (130.19) 126.43 (78.85) 
Stop accuracy % at calibration stage 65.00 (0.04) 65.63 (0.02) 65.80 (0.03) 64.80 (0.05) 
Stop accuracy % at TMS stage 67.06 (0.20) 65.04 (0.15) 58.68 (0.18) 67.43 (0.14) 
RMT (%MSO) 49.46 (9.81) 48.62 (8.92) 49.89 (10.79) 48.29 (9.29) 
Resting spMEP amplitude (μV) 499.76 (575.80) 804.70 (833.46) 615.91 (717.06) 744.54 (697.67) 

For Sessions 2 and 3, a one-way ANOVA across the 10 blocks of stop signal task revealed no significant effect of Time, F(9, 188) = 1.82, p = .063, such that the experimental group did not show a significant change in SSRT across the 10 blocks of training. A 2 (Training: experimental vs. control) × 10 (Time: 10 blocks) mixed-model ANOVA on go RT revealed no significant change in go RTs across the 10 blocks, F(5.378, 209.734) = 1.321, p = .224, and no significant difference in go RTs between the experimental and control groups at the .025 level, F(1, 39) = 4.834, p = .031, nor was there any interaction between Training and Time, F(5.378, 209.734) = 1.11, p = .354. Overall results from Sessions 2 and 3 suggest SSRT did not change throughout the 10 blocks. Although there was a trend toward go RTs being faster in the control group compared to the experimental group, this was not statistically significant, suggesting evidence of proactive inhibition in the experimental group because of the presence of stop trials was not conclusive.

RMT and spMEP Amplitudes

Table 3 shows the mean premeasures and postmeasures of RMT and spMEP amplitudes at rest. A 2 (Time: pre vs. post) × 2 (Training: experimental vs. control) mixed-model ANOVA on fixation spMEP amplitude, resting spMEP amplitudes, and RMT revealed no significant main effects of time and group, nor was there any significant interactions between these variables (all ps > .0167). This provides evidence that resting and fixation spMEP amplitude sizes and RMT were matched across groups.

Figure 8 shows mean spMEP amplitudes for each of the trial types (fixation, go, and stop trials as a function of response hand). Results of the 2 (Time: pre vs. post) × 2 (Training: experimental vs. control) 2 (Trial: go vs. stop) × 2 (Response hand: left vs. right) mixed-model ANOVA on trial spMEP amplitudes revealed a significant main effect of Time, such that spMEPs were larger during Session 4 than Session 1, F(1, 39) = 6.264, p = .017. There was a main effect of Trial, such that MEP amplitudes were larger for go trials than for stop trials, F(1, 39) = 92.146, p < .001. There was a main effect of Response hand such that MEPs recorded during right-hand responses were larger than MEPs recorded during left-hand responses, F(1, 39) = 92.156, p < .001. This difference was larger during Session 4 than Session 1, as indicated by the significant interaction between Time and Response hand, F(1, 39) = 4.66, p = .037. There was an interaction between Trial and Response hand, F(1, 39) = 102.776, p < .001, such that the MEP difference between right- and left-hand responses was larger for go trials than stop trials. The key finding was that there was no significant interaction between Time and Training, F(1, 39) = 1.44, p = .237, such that the change in spMEP amplitudes across Sessions 1 and 4 did not differ between the experimental and control groups, confirming that training on the stop signal task did not influence spMEP amplitudes. All other main effects and interaction effects were nonsignificant (ps > .05). Overall, this provides evidence that the pattern of spMEP amplitudes during the trials did not differ between the experimental and control groups, suggesting that the groups were matched in terms of the overall spMEP amplitudes as well as changes in spMEP amplitudes as a function of time, trial type, and response hand.

Paired-samples t tests showed that mean spMEPs (across Sessions 1 and 4) were facilitated relative to fixation for right-hand go trials, t(40) = 8.118, p < .001, and suppressed relative to fixation for left-hand go trials, t(40) = −6.75, p < .001. Such suppression of MEPs in the nonselected hand of a choice RT task has been reported previously (Duque, Lew, Mazzocchio, Olivier, & Ivry, 2010). Mean spMEPs (across Sessions 1 and 4) were also suppressed relative to fixation for right-hand stop trials, t(40) = 6.063, p < .001, and left-hand stop trials, t(40) = 7.496, p < .001. Furthermore, there was stronger suppression during left-hand stop trials relative to left-hand go trials, as confirmed by a paired-samples t test, t(40) = 3.984, p < .001. This suggests that, during left-hand stop trials, a stopping mechanism acted in addition to the inhibitory mechanism for left-hand go trials to further suppress spMEPs.

Resting-state SICI

Figure 7 shows pre- and post-resting-state SICI for the experimental and control groups for Experiment 2. An independent samples t test revealed no significant difference in resting-state SICI for Session 1 between the training and control groups, t(42) = 0.451, p = .654, BF01 = 3.093. Results of the 2 (Time: pre vs. post) × 2 (Training: experimental vs. control) mixed-model ANOVA on resting-state SICI revealed a significant main effect of Time, F(1, 42) = 5.89, p = .02, such that post-SICI was, on average, stronger than pre-SICI. Time did not interact with Training, F(1, 40) = 1.162, p = .287, such that the difference in SICI between pre and post was not significantly larger for the training group relative to the control group.

Figure 7. 

Resting-state SICI (± SEM) at pre and post for the experimental and control groups, for Experiment 2. The ordinate axis represents the log-transformation of the ratio of the ppMEP-to-spMEP amplitude. A more negative number represents stronger SICI.

Figure 7. 

Resting-state SICI (± SEM) at pre and post for the experimental and control groups, for Experiment 2. The ordinate axis represents the log-transformation of the ratio of the ppMEP-to-spMEP amplitude. A more negative number represents stronger SICI.

Figure 8. 

Mean spMEP amplitudes (± SEM) averaged across presessions and postsessions for fixation, go, and stop trials as a function of response hand (right or left). Note that MEPs were recorded from the right FDI.

Figure 8. 

Mean spMEP amplitudes (± SEM) averaged across presessions and postsessions for fixation, go, and stop trials as a function of response hand (right or left). Note that MEPs were recorded from the right FDI.

SICI Modulation during the Task

Figure 9 shows pre- and post-SICI during fixation, during go and stop trials, for both right- and left-hand responding. Figure 9A shows these data for the experimental group, whereas Figure 9B shows these data for the control group. Both groups show similar SICI for each trial type initially and similar SICI during fixation at both time points. The most striking difference between groups is that the experimental group shows an increase in SICI for stop trials from pre to post (the red lines in Figure 9A), with similar effects on both right-hand stop trials and left-hand stop trials, suggesting that the change in SICI is global. No pre-to-post changes appear to occur for the control group.

Figure 9. 

A shows the pre- and post-SICI (± SEM) for each condition for the experimental group. B shows the pre- and post-SICI (± SEM) for each condition for the control group. A more negative value indicates more SICI, whereas a value of 0 indicates no mean inhibition of the paired pulse relative to the single pulse. Both groups show similar fixation SICI for both time points, similar initial SICI for each trial type, and weaker SICI during the trial than during fixation. However, the experimental group shows an increase in overall SICI and SICI modulation between go and stop trials from pre to post relative to the control group. The improvement in SICI modulation between go and stop trials appears to be global, applying to both right- and left-hand trials. Data are offset for illustrative purposes.

Figure 9. 

A shows the pre- and post-SICI (± SEM) for each condition for the experimental group. B shows the pre- and post-SICI (± SEM) for each condition for the control group. A more negative value indicates more SICI, whereas a value of 0 indicates no mean inhibition of the paired pulse relative to the single pulse. Both groups show similar fixation SICI for both time points, similar initial SICI for each trial type, and weaker SICI during the trial than during fixation. However, the experimental group shows an increase in overall SICI and SICI modulation between go and stop trials from pre to post relative to the control group. The improvement in SICI modulation between go and stop trials appears to be global, applying to both right- and left-hand trials. Data are offset for illustrative purposes.

A 2 (Time: pre vs. post) × 2 (Training: experimental vs. group) mixed-model ANOVA on SICI during fixation revealed no main effects and interactions (all ps > .05). For Session 1, an independent samples t test showed no difference in SICI between the experimental and control groups for right-hand go trials, t(39) = −1.507, p = .14, BF01 = 1.337, left-hand go trials, t(39) = 0.099, p = .922, BF01 = 3.256, right-hand stop trials, t(39) = 0.535, p= .596, BF01 = 2.916, and left-hand stop trials, t(39) = −0.248, p = .806, BF01 = 3.189. Results of the 2 (Time: pre vs. post) × 2 (Training: experimental vs. control) × 2 (Trial: go vs. stop) × 2 (Response hand: left vs. right) mixed-model ANOVA on SICI revealed a trend for SICI being stronger on stop trials relative to go trials, F(1, 39) = 3.152, p = .084. There was a significant interaction between Time and Trial, such that the difference in SICI between stop and go trials was larger for Session 4 than for Session 1, F(1, 39) = 7.093, p = .011. The SICI difference between go and stop trials was numerically larger for right-hand responding than left-hand responding, but this fell short of significance, F(1, 39) = 3.714, p = .061. There was a significant interaction between Time and Training, such that the overall increase in SICI strength after training was larger in the experimental group than the control group, F(1, 39) = 10.897, p = .002. There was some evidence to suggest that this training-induced increase in SICI was also larger on stop trials than go trials, as suggested by a near-significant three-way interaction between Time, Training, and Trial, F(1, 39) = 3.976, p = .053. All other main effects and interaction effects were nonsignificant (all ps > .05).

Overall, there were two main patterns of results for SICI modulation during the task. First, in terms of overall SICI modulation during the trial, there was a trend for SICI being stronger during stop trials than go trials. There was also a trend toward this difference being stronger during right-hand responses than left-hand responses, which is indicative of SICI modulation on stop trials specifically when the right hand was selected for the response. This selective modulation can be contrasted with global modulation, evident when modulation occurs regardless of whether the relevant hand has been selected for action. We remain cautious of this result because of the lack of a significant effect. The most relevant finding in this article was that the experimental group showed an overall increase in SICI in Session 4 relative to the control group. There was also a trend toward improvement in go versus stop SICI modulation, such that those who received training became better at differentially modulating SICI during response execution versus inhibition. Again, we remain cautious of this latter effect as it failed to reach significance.

Correlation Analysis

Figure 10 shows the difference between pre- and post-SICI during successful stops plotted against the difference between pre- and post-SSRT, for Experiment 2. The Pearson correlations between SICI change and SSRT change was r(42) = .22, p = .149, for resting SICI (experimental group: r(21) = .37, p = .09; control group: r(19) = .05, p = .83) and r(39) = .67, p < .001, for SICI during successful stops (experimental group: r(19) = .73, p < .001; control group: r(18) = .59, p = .005). Table 4 shows the pattern of correlations between pre- and post-resting-state SICI, successful stop SICI, and SSRT, for Experiment 2.

Figure 10. 

Difference in SICI measured during all successful stop trials between pre and post graphed against difference in SSRT between pre and post, for Experiment 2.

Figure 10. 

Difference in SICI measured during all successful stop trials between pre and post graphed against difference in SSRT between pre and post, for Experiment 2.

Table 4. 
Pattern of Correlations between Pre- and Post-SSRT and Mean SICI (Both at Rest and during Successful Stop Trials), for Experiment 2, across All Participants
 Pre Resting SICIPost Resting SICIPre-SSRTPost-SSRTPre Stop SICI
Post resting SICI .54**         
Pre-SSRT .49** .27       
Post-SSRT .41** .36* .63**     
Pre stop SICI .62** .52** .74** .44**   
Post stop SICI .43** .41* .33* .61** .50** 
 Pre Resting SICIPost Resting SICIPre-SSRTPost-SSRTPre Stop SICI
Post resting SICI .54**         
Pre-SSRT .49** .27       
Post-SSRT .41** .36* .63**     
Pre stop SICI .62** .52** .74** .44**   
Post stop SICI .43** .41* .33* .61** .50** 
*

Significant at the .05 level.

**

Significant at the .01 level.

Given overall spMEP amplitudes were higher in Session 4 than in Session 1, we determined whether changes in SSRTs were explained by changes in spMEP amplitudes rather than SICI. This is important because stronger SICI has been shown to relate to higher spMEP amplitudes (Opie & Semmler, 2014). To do so, we conducted a multiple linear regression with stop trial SICI change (averaged across the left and right hands) and stop trial spMEP change (averaged across the left and right hands) as predictor variables and SSRT change as the outcome variable. These regressions were conducted on the experimental and control groups separately. For the experimental group, we found that stop trial SICI change was still a significant predictor of SSRT change when controlling stop trial spMEP change (b= 57.68, t = 4.318, p < .001, sr = 0.703), whereas stop trial spMEP change was not a significant predictor (b = 0.001, t = 0.25, p = .805, sr = 0.041). For the control group, we also found that stop trial SICI change was still a significant predictor of SSRT change when controlling for stop trial spMEP change (b = 44.335, t = 3.384, p = .004, sr = 0.623), whereas stop trial spMEP change was not a significant predictor (b = −0.014, t = −1.09, p = .291, sr = −0.207). This analysis suggests that SSRT change was specifically explained by stop trial SICI change rather than stop trial spMEP change.

Overall, the correlation analyses show a strong link between SICI and SSRT, although this time, the relationship was evident for SICI measured on successful stop trials rather than resting-state SICI. An increase in SICI strength during successful stops between pre and post correlates with a shortening of SSRT between pre and post (i.e., more efficient stopping). Furthermore, the SICI and SSRT correlations were numerically higher for SICI during successful stops than for resting SICI. Finally, resting-state SICI, SICI during successful stops, and SSRT all showed moderate test–retest reliability.

The data supporting the findings of this study are available online.1

DISCUSSION

In this study, we conducted two experiments to determine whether training participants in the stop signal task within one session (Experiment 1) and across two sessions (Experiment 2) would lead to increased SICI strength. For each experiment, at pre and post, we measured stopping efficiency and resting-state SICI (Experiment 1) and SICI during the stop signal task (Experiment 2). In the middle blocks of Experiment 1 and the middle sessions of Experiment 2, participants in the experimental group completed stop signal task training, whereas control participants completed a similar task without the requirement to stop a response. For Experiment 1, the experimental group showed increased resting-state SICI strength. The most notable result of Experiment 2 was that stop signal task training led to increased SICI strength during the trial in the experimental versus control group, with a trend toward a stronger effect on stop trials than on go trials (i.e., more pronounced go vs. stop SICI modulation). Although there were no overall behavioral improvements in stopping efficiency for both experiments, any improvements at an individual level were correlated with increases in SICI strength at rest (Experiment 1) and during successful stopping (Experiment 2). Moreover, SICI correlated with SSRT at each time point, and SICI and SSRT measures themselves showed moderate test–retest reliability. These results were observed in the experimental and control groups matched for RMT, overall spMEP amplitudes, and spMEP changes during the task as well as initial SSRT and SICI at rest and during the task.

Response Inhibition Training Strengthens M1 GABA-mediated Neurotransmission

This is the first study to show that response inhibition training can strengthen GABAA-mediated SICI within M1. SICI was measured during two states: at rest (Experiments 1 and 2) and during the stop signal task (Experiment 2). Experiment 1 consisted of training within one session, with SICI measured shortly after the final training block, whereas for Experiment 2, training occurred across two sessions, with SICI measured, on average, 3 days after the final training block. We observed that resting-state SICI increased in strength for Experiment 1 but not for Experiment 2. However, task-related SICI yielded improvements days after the final training block. This pattern of results might suggest that training produced a transient effect on SICI that was not task related (measured at rest) and also produced longer lasting task-related changes in SICI (measured during response stopping). However, caution is required when comparing data between the experiments given aspects of the stop signal tasks differed (bimanual vs. unimanual task, 33% vs. 25% stop signal probability).

An interesting observation evident in Figure 9 is that, at Session 4, individuals trained in the stop signal task show levels of SICI on successful stop trials that are comparable to baseline levels during fixation. In contrast, the control group seemed to show less SICI compared to baseline. This is consistent with findings from our previous study (Chowdhury et al., 2019a), which showed that participants who were better at stopping were able to recover baseline levels of SICI whereas slower stoppers exhibited less SICI relative to baseline. This suggests the ability to recover SICI back to baseline levels is a key feature of efficient stopping.

Although SSRT in Experiment 2 significantly improved in the experimental group, this was not statistically different from the control group. Thus, our results are in line with previous studies that have failed to find significant behavioral effects of response inhibition training (Enge et al., 2014; Cohen & Poldrack, 2008). The null findings for SSRT improvement might relate to the higher variability in behavioral measures relative to neural measures, suggesting larger sample sizes may be required to find such behavioral effects. The lack of behavioral training effects in our study might also be attributed to the training session in Experiment 1 being relatively short and the training blocks in Experiment 2 being inadequately spaced across days (for instance, delivering 10 blocks across 10 days instead of across 2 days). Indeed, some studies that have found behavioral effects of response inhibition training have used short multiple sessions spaced across more days of training (e.g., Berkman et al., 2014). An alternative explanation for the absence of a behavioral effect for Experiment 2 is that the control group completed too much of the stop signal task, especially when TMS was delivered, reducing the opportunity to observe a significant interaction between time and training.

Despite the absence of overall training effects on SSRT, this study focused on individual differences in improvement and how this related to neural changes. A shortening of SSRT was correlated with increased resting-state SICI strength (Experiment 1) and increased SICI strength on successful stop trials (Experiment 2). This suggests that the increase in SICI were related to the key response inhibition measure. Another explanation for the neural changes would be that, during the middle blocks of Experiment 1 and middle sessions of Experiment 2, the experimental group exhibited anticipatory proactive inhibition (slowing of go RTs) because of the presence of stop trials, and it is proactive inhibition that drove the SICI changes. However, in both experiments, there was no conclusive evidence of a proactive inhibition effect. Overall, our pattern of results suggests SICI and SSRT are closely linked, such that a change in SICI between time points is correlated with a change in stopping efficiency between time points.

The Role of SICI in Response Inhibition

Previous work has shown that SICI is recruited during stopping of actions (Hermans et al., 2019; Lindberg et al., 2016; MacDonald et al., 2014; Coxon et al., 2006), such that SICI is stronger on stop trials than on go trials. Our results are in line with these studies, but in addition, we have shown that a change in stopping performance across time is correlated with a change in SICI. Furthermore, we showed a trend toward the SICI difference between stop trials and go trials being larger when the contralateral hand was selected for a response compared with when it was not selected. The finding that there is potentially selective SICI modulation on stop versus go trials is in line with another showing that SICI is modulated in a selective manner during proactive response inhibition (Cirillo et al., 2018) and in line with our own study showing that there is a hemisphere-specific influence of SICI over stopping performance (Chowdhury et al., 2019b). If the go versus stop SICI modulation during right-hand (selected) versus left-hand (nonselected) responding was equivalent, it would suggest SICI is modulated globally, similar to spMEP modulation during standard stop signal tasks (Wessel, Reynoso, & Aron, 2013; Cai, Oldenkamp, & Aron, 2012; Majid, Cai, George, Verbruggen, & Aron, 2012; Badry et al., 2009). This might suggest that the stopping mechanisms acting on spMEPs are dissociable from the stopping mechanisms involving SICI. We remain cautious of this effect given that the relevant result fell short of significance, and we interpret this effect as an avenue for future research to explore. Caution is also required when comparing SICI between go and stop trials and between selected and nonselected trials given that spMEP amplitudes and trial availability differ between these conditions.

Reliability of the SICI–SSRT Relationship across Time within and across Sessions

In our previous work (Chowdhury et al., 2018, 2019a, 2019b), we showed that individual differences in SSRT are correlated with individual differences in SICI at one time point, but we did not assess the reliability of this relationship across time. This study shows that there is still a significant correlation between SICI and SSRT assessed at a second time point within the same session and at a session approximately 1 week after the first session. The resting SICI–SSRT relationship was ∼.38 on both occasions for Experiment 1, and for Experiment 2, the correlations were .49 at pre and .36 at post. The relationship between SSRT and SICI during successful stop trials was .74 at pre and .62 at post. Any reductions in these correlations across time might be because of training-related effects that reduce variance in SSRT and/or SICI and the opportunity to see a correlation. This pattern of relationships, along with our previous work, also seems to suggest that the relationship between SICI and SSRT is stronger during successful stops than when SICI is measured at rest (our resting-state SICI–SSRT correlations have varied between .36 and .63, whereas our correlations between SICI and SSRT during stopping have varied between .58 and .81).

Limitations and Outstanding Questions

SICI showed moderate test–retest reliability (both at rest and during successful stop trials) and consistently correlated with SSRT, giving us reason to be confident that, as a neurophysiological measure, it is both reliable and indexes a behaviorally relevant property of inhibitory function. Nevertheless, it should be noted that our SICI measure is not optimized for each individual because we used a paired-pulse protocol with fixed ISI (3 msec) and fixed conditioning pulse intensity (80% RMT). This means that we may be underestimating the strength of SICI for some individuals, and activity other than GABAA receptor activity may have been probed (Peurala, Müller-Dahlhaus, Arai, & Ziemann, 2008). Thus, an interesting future question would be to replicate this study using individualized TMS parameters that produce optimal SICI and therefore reduce measurement noise for each participant.

Given TMS was delivered at a fixed 200 msec after stop signal onset, it is possible that faster stoppers were simply closer to completing the stopping process than slower stoppers at the point where TMS was triggered. If so, SICI was probed closer to the end of SSRT for a participant who showed improved SSRT, suggesting any increase in SICI could be related to TMS timing rather than an improvement in inhibitory control. This may have justified tailoring the timing of SICI relative to each individual's SSRT. However, our previous study that did just this (Chowdhury et al., 2019a) showed that the relationship between SSRT and SICI was the same regardless of when TMS was delivered. On stop trials, those with slower SSRTs were still unable to recover SICI to baseline levels close to the end of their SSRT. Thus, we would expect to see results similar to those of this study even if TMS were delivered close to the end of each participant's SSRT, such that initially slower stoppers would be unable to recover SICI during stopping, and an improvement in SSRT would be accompanied by an increase in SICI.

Although our study adds to our understanding of the link between SICI and SSRT (i.e., change in SICI correlating with change in SSRT), future research is required to understand causal connections linking changes in SSRT and change in SICI. Indeed, it is possible that the SICI changes were mediated by prefrontal neuroplasticity, which has been observed after response inhibition training (Berkman et al., 2014; Manuel et al., 2013). Causal study designs can be developed using theta-burst stimulation (Chung, Hill, Rogasch, Hoy, & Fitzgerald, 2016) or transcranial direct current stimulation (Biabani et al., 2018). These techniques have been shown to modulate SICI, and thus, the effect of this on stopping performance can be established.

Conclusion

This study is the first to show that stop signal task training leads to changes in M1 GABAA-mediated SICI, measured at rest as well as during the task. The results add to our understanding of neuroplasticity within M1, and although the behavioral effects of training were nonsignificant, the results raise the possibility of using response inhibition training to strengthen inhibitory circuits in M1, which has major clinical applications for disorders associated with impaired response inhibition.

Acknowledgments

This research was supported by Discovery Project grant DP190100410, from the Australian Research Council.

Reprint requests should be sent to Nahian S. Chowdhury, School of Psychology, University of Sydney, Camperdown, NSW 2006, Australia, or via e-mail: ncho1135@uni.sydney.edu.au.

Note

1. 

The data can be retrieved from the University of SydneyLabArchives eNotebook with the identifier DOIs: http://dx.doi.org/10.25833/4cvn-nj63 for Experiment 1 and http://dx.doi.org/10.25833/aeyx-tm92 for Experiment 2.

REFERENCES

REFERENCES
Aasvik
,
J. K.
,
Woodhouse
,
A.
,
Stiles
,
T. C.
,
Jacobsen
,
H. B.
,
Landmark
,
T.
,
Glette
,
M.
, et al
(
2017
).
Effectiveness of working memory training among subjects currently on sick leave due to complex symptoms
.
Frontiers in Psychology
,
7
,
2003
.
Aron
,
A. R.
(
2007
).
The neural basis of inhibition in cognitive control
.
Neuroscientist
,
13
,
214
228
.
Badry
,
R.
,
Mima
,
T.
,
Aso
,
T.
,
Nakatsuka
,
M.
,
Abe
,
M.
,
Fathi
,
D.
, et al
(
2009
).
Suppression of human cortico-motoneuronal excitability during the stop-signal task
.
Clinical Neurophysiology
,
120
,
1717
1723
.
Berkman
,
E. T.
,
Kahn
,
L. E.
, &
Merchant
,
J. S.
(
2014
).
Training-induced changes in inhibitory control network activity
.
Journal of Neuroscience
,
34
,
149
157
.
Biabani
,
M.
,
Aminitehrani
,
M.
,
Zoghi
,
M.
,
Farrell
,
M.
,
Egan
,
G.
, &
Jaberzadeh
,
S.
(
2018
).
The effects of transcranial direct current stimulation on short-interval intracortical inhibition and intracortical facilitation: A systematic review and meta-analysis
.
Reviews in the Neurosciences
,
29
,
99
114
.
Cai
,
W.
,
Oldenkamp
,
C. L.
, &
Aron
,
A. R.
(
2012
).
Stopping speech suppresses the task-irrelevant hand
.
Brain and Language
,
120
,
412
415
.
Chao
,
H. H. A.
,
Luo
,
X.
,
Chang
,
J. L. K.
, &
Li
,
C.-S. R.
(
2009
).
Activation of the pre-supplementary motor area but not inferior prefrontal cortex in association with short stop signal reaction time—An intra-subject analysis
.
BMC Neuroscience
,
10
,
75
.
Chiu
,
Y.-C.
,
Aron
,
A. R.
, &
Verbruggen
,
F.
(
2012
).
Response suppression by automatic retrieval of stimulus–stop association: Evidence from transcranial magnetic stimulation
.
Journal of Cognitive Neuroscience
,
24
,
1908
1918
.
Chowdhury
,
N. S.
,
Livesey
,
E. J.
,
Blaszczynski
,
A.
, &
Harris
,
J. A.
(
2017
).
Pathological gambling and motor impulsivity: A systematic review with meta-analysis
.
Journal of Gambling Studies
,
33
,
1213
1239
.
Chowdhury
,
N. S.
,
Livesey
,
E. J.
,
Blaszczynski
,
A.
, &
Harris
,
J. A.
(
2018
).
Variations in response control within at-risk gamblers and non-gambling controls explained by GABAergic inhibition in the motor cortex
.
Cortex
,
103
,
153
163
.
Chowdhury
,
N. S.
,
Livesey
,
E. J.
,
Blaszczynski
,
A.
, &
Harris
,
J. A.
(
2020
).
Motor cortex dysfunction in problem gamblers
.
Addiction Biology
,
e12871
.
Chowdhury
,
N. S.
,
Livesey
,
E. J.
, &
Harris
,
J. A.
(
2019a
).
Individual differences in intracortical inhibition during behavioural inhibition
.
Neuropsychologia
,
124
,
55
65
.
Chowdhury
,
N. S.
,
Livesey
,
E. J.
, &
Harris
,
J. A.
(
2019b
).
Contralateral and ipsilateral relationships between intracortical inhibition and stopping efficiency
.
Neuroscience
,
415
,
10
17
.
Chung
,
S. W.
,
Hill
,
A. T.
,
Rogasch
,
N. C.
,
Hoy
,
K. E.
, &
Fitzgerald
,
P. B.
(
2016
).
Use of theta burst stimulation in changing excitability of motor cortex: A systematic review and meta-analysis
.
Neuroscience & Biobehavioral Reviews
,
63
,
43
64
.
Cirillo
,
J.
,
Cowie
,
M. J.
,
MacDonald
,
H. J.
, &
Byblow
,
W. D.
(
2018
).
Response inhibition activates distinct motor cortical inhibitory processes
.
Journal of Neurophysiology
,
119
,
877
886
.
Cohen
,
J. R.
, &
Poldrack
,
R. A.
(
2008
).
Automaticity in motor sequence learning does not impair response inhibition
.
Psychonomic Bulletin & Review
,
15
,
108
115
.
Coxon
,
J. P.
,
Stinear
,
C. M.
, &
Byblow
,
W. D.
(
2006
).
Intracortical inhibition during volitional inhibition of prepared action
.
Journal of Neurophysiology
,
95
,
3371
3383
.
Davila-Pérez
,
P.
,
Jannati
,
A.
,
Fried
,
P. J.
,
Cudeiro Mazaira
,
J.
, &
Pascual-Leone
,
A.
(
2018
).
The effects of waveform and current direction on the efficacy and test–retest reliability of transcranial magnetic stimulation
.
Neuroscience
,
393
,
97
109
.
Duque
,
J.
,
Lew
,
D.
,
Mazzocchio
,
R.
,
Olivier
,
E.
, &
Ivry
,
R. B.
(
2010
).
Evidence for two concurrent inhibitory mechanisms during response preparation
.
Journal of Neuroscience
,
30
,
3793
3802
.
Enge
,
S.
,
Behnke
,
A.
,
Fleischhauer
,
M.
,
Küttler
,
L.
,
Kliegel
,
M.
, &
Strobel
,
A.
(
2014
).
No evidence for true training and transfer effects after inhibitory control training in young healthy adults
.
Journal of Experimental Psychology: Learning, Memory, and Cognition
,
40
,
987
1001
.
Garry
,
M. I.
, &
Thomson
,
R. H. S.
(
2009
).
The effect of test TMS intensity on short-interval intracortical inhibition in different excitability states
.
Experimental Brain Research
,
193
,
267
.
Guerrieri
,
R.
,
Nederkoorn
,
C.
, &
Jansen
,
A.
(
2008
).
The interaction between impulsivity and a varied food environment: Its influence on food intake and overweight
.
International Journal of Obesity
,
32
,
708
714
.
Hermans
,
L.
,
Maes
,
C.
,
Pauwels
,
L.
,
Cuypers
,
K.
,
Heise
,
K.-F.
,
Swinnen
,
S. P.
, et al
(
2019
).
Age-related alterations in the modulation of intracortical inhibition during stopping of actions
.
Aging
,
11
,
371
385
.
Kratz
,
O.
,
Diruf
,
M. S.
,
Studer
,
P.
,
Gierow
,
W.
,
Buchmann
,
J.
,
Moll
,
G. H.
, et al
(
2009
).
Effects of methylphenidate on motor system excitability in a response inhibition task
.
Behavioral and Brain Functions
,
5
,
12
.
Kühn
,
S.
,
Lorenz
,
R. C.
,
Weichenberger
,
M.
,
Becker
,
M.
,
Haesner
,
M.
,
O'Sullivan
,
J.
, et al
(
2017
).
Taking control! Structural and behavioural plasticity in response to game-based inhibition training in older adults
.
Neuroimage
,
156
,
199
206
.
Kujirai
,
T.
,
Caramia
,
M. D.
,
Rothwell
,
J. C.
,
Day
,
B. L.
,
Thompson
,
P. D.
,
Ferbert
,
A.
, et al
(
1993
).
Corticocortical inhibition in human motor cortex
.
Journal of Physiology
,
471
,
501
519
.
Leunissen
,
I.
,
Zandbelt
,
B. B.
,
Potocanac
,
Z.
,
Swinnen
,
S. P.
, &
Coxon
,
J. P.
(
2017
).
Reliable estimation of inhibitory efficiency: To anticipate, choose or simply react?
European Journal of Neuroscience
,
45
,
1512
1523
.
Li
,
C.-S. R.
,
Huang
,
C.
,
Constable
,
R. T.
, &
Sinha
,
R.
(
2006
).
Imaging response inhibition in a stop-signal task: Neural correlates independent of signal monitoring and post-response processing
.
Journal of Neuroscience
,
26
,
186
192
.
Li
,
C.-S. R.
,
Yan
,
P.
,
Sinha
,
R.
, &
Lee
,
T.-W.
(
2008
).
Subcortical processes of motor response inhibition during a stop signal task
.
Neuroimage
,
41
,
1352
1363
.
Lindberg
,
P. G.
,
Térémetz
,
M.
,
Charron
,
S.
,
Kebir
,
O.
,
Saby
,
A.
,
Bendjemaa
,
N.
, et al
(
2016
).
Altered cortical processing of motor inhibition in schizophrenia
.
Cortex
,
85
,
1
12
.
Logan
,
G. D.
, &
Burkell
,
J.
(
1986
).
Dependence and independence in responding to double stimulation: A comparison of stop, change, and dual-task paradigms
.
Journal of Experimental Psychology: Human Perception and Performance
,
12
,
549
563
.
Logan
,
G. D.
, &
Cowan
,
W. B.
(
1984
).
On the ability to inhibit thought and action: A theory of an act of control
.
Psychological Review
,
91
,
295
327
.
MacDonald
,
H. J.
,
Coxon
,
J. P.
,
Stinear
,
C. M.
, &
Byblow
,
W. D.
(
2014
).
The fall and rise of corticomotor excitability with cancellation and reinitiation of prepared action
.
Journal of Neurophysiology
,
112
,
2707
2717
.
Majid
,
D. S. A.
,
Cai
,
W.
,
George
,
J. S.
,
Verbruggen
,
F.
, &
Aron
,
A. R.
(
2012
).
Transcranial magnetic stimulation reveals dissociable mechanisms for global versus selective corticomotor suppression underlying the stopping of action
.
Cerebral Cortex
,
22
,
363
371
.
Manuel
,
A. L.
,
Bernasconi
,
F.
, &
Spierer
,
L.
(
2013
).
Plastic modifications within inhibitory control networks induced by practicing a stop-signal task: An electrical neuroimaging study
.
Cortex
,
49
,
1141
1147
.
Menzies
,
L.
,
Achard
,
S.
,
Chamberlain
,
S. R.
,
Fineberg
,
N.
,
Chen
,
C.-H.
,
del Campo
,
N.
, et al
(
2007
).
Neurocognitive endophenotypes of obsessive–compulsive disorder
.
Brain
,
130
,
3223
3236
.
Ni
,
Z.
,
Gunraj
,
C.
, &
Chen
,
R.
(
2007
).
Short interval intracortical inhibition and facilitation during the silent period in human
.
Journal of Physiology
,
583
,
971
982
.
Oosterlaan
,
J.
,
Logan
,
G. D.
, &
Sergeant
,
J. A.
(
1998
).
Response inhibition in AD/HD, CD, comorbid AD/HD+CD, anxious, and control children: A meta-analysis of studies with the stop task
.
Journal of Child Psychology and Psychiatry
,
39
,
411
425
.
Opie
,
G. M.
, &
Semmler
,
J. G.
(
2014
).
Modulation of short- and long-interval intracortical inhibition with increasing motor evoked potential amplitude in a human hand muscle
.
Clinical Neurophysiology
,
125
,
1440
1450
.
Peirce
,
J. W.
(
2007
).
PsychoPy—Psychophysics software in Python
.
Journal of Neuroscience Methods
,
162
,
8
13
.
Peurala
,
S. H.
,
Müller-Dahlhaus
,
J. F. M.
,
Arai
,
N.
, &
Ziemann
,
U.
(
2008
).
Interference of short-interval intracortical inhibition (SICI) and short-interval intracortical facilitation (SICF)
.
Clinical Neurophysiology
,
119
,
2291
2297
.
Poole
,
B. J.
,
Mather
,
M.
,
Livesey
,
E. J.
,
Harris
,
I. M.
, &
Harris
,
J. A.
(
2018
).
Motor-evoked potentials reveal functional differences between dominant and non-dominant motor cortices during response preparation
.
Cortex
,
103
,
1
12
.
Rossi
,
S.
,
Hallett
,
M.
,
Rossini
,
P. M.
,
Pascual-Leone
,
A.
, &
The Safety of TMS Consensus Group
. (
2009
).
Safety, ethical considerations, and application guidelines for the use of transcranial magnetic stimulation in clinical practice and research
.
Clinical Neurophysiology
,
120
,
2008
2039
.
Rothwell
,
J. C.
,
Hallett
,
M.
,
Berardelli
,
A.
,
Eisen
,
A.
,
Rossini
,
P.
, &
Paulus
,
W.
(
1999
).
Magnetic stimulation: Motor evoked potentials. The International Federation of Clinical Neurophysiology
.
Electroencephalography and Clinical Neurophysiology
,
52(Suppl)
,
97
103
.
Rouder
,
J. N.
,
Speckman
,
P. L.
,
Sun
,
D.
,
Morey
,
R. D.
, &
Iverson
,
G.
(
2009
).
Bayesian t tests for accepting and rejecting the null hypothesis
.
Psychonomic Bulletin & Review
,
16
,
225
237
.
Seet
,
M. S.
,
Livesey
,
E. J.
, &
Harris
,
J. A.
(
2019
).
Associatively-mediated suppression of corticospinal excitability: A transcranial magnetic stimulation (TMS) study
.
Neuroscience
,
416
,
1
8
.
Stinear
,
C. M.
,
Coxon
,
J. P.
, &
Byblow
,
W. D.
(
2009
).
Primary motor cortex and movement prevention: Where stop meets go
.
Neuroscience & Biobehavioral Reviews
,
33
,
662
673
.
Thomson
,
R. H. S.
,
Garry
,
M. I.
, &
Summers
,
J. J.
(
2008
).
Attentional influences on short-interval intracortical inhibition
.
Clinical Neurophysiology
,
119
,
52
62
.
Tran
,
D. M. D.
,
Harris
,
J. A.
,
Harris
,
I. M.
, &
Livesey
,
E. J.
(
2019
).
Motor memory: Revealing conditioned action tendencies using transcranial magnetic stimulation
.
Journal of Cognitive Neuroscience
,
31
,
1343
1353
.
van den Wildenberg
,
W. P. M.
,
Burle
,
B.
,
Vidal
,
F.
,
van der Molen
,
M. W.
,
Ridderinkhof
,
K. R.
, &
Hasbroucq
,
T.
(
2009
).
Mechanisms and dynamics of cortical motor inhibition in the stop-signal paradigm: A TMS study
.
Journal of Cognitive Neuroscience
,
22
,
225
239
.
Verbruggen
,
F.
,
Chambers
,
C. D.
, &
Logan
,
G. D.
(
2013
).
Fictitious inhibitory differences: How skewness and slowing distort the estimation of stopping latencies
.
Psychological Science
,
24
,
352
362
.
Verbruggen
,
F.
,
Logan
,
G. D.
, &
Stevens
,
M. A.
(
2008
).
STOP-IT: Windows executable software for the stop-signal paradigm
.
Behavior Research Methods
,
40
,
479
483
.
Waldvogel
,
D.
,
van Gelderen
,
P.
,
Muellbacher
,
W.
,
Ziemann
,
U.
,
Immisch
,
I.
, &
Hallet
,
M.
(
2000
).
The relative metabolic demand of inhibition and excitation
.
Nature
,
406
,
995
998
.
Wessel
,
J. R.
,
Reynoso
,
H. S.
, &
Aron
,
A. R.
(
2013
).
Saccade suppression exerts global effects on the motor system
.
Journal of Neurophysiology
,
110
,
883
890
.
Zhang
,
S.
,
Tsai
,
S.-J.
,
Hu
,
S.
,
Xu
,
J.
,
Chao
,
H. H.
,
Calhoun
,
V. D.
, et al
(
2015
).
Independent component analysis of functional networks for response inhibition: Inter-subject variation in stop signal reaction time
.
Human Brain Mapping
,
36
,
3289
3302
.
Ziemann
,
U.
,
Lönnecker
,
S.
,
Steinhoff
,
B. J.
, &
Paulus
,
W.
(
1996
).
The effect of lorazepam on the motor cortical excitability in man
.
Experimental Brain Research
,
109
,
127
135
.