Table of Contents

Rapid Motor Inhibition as a Mechanism to Prevent Outdated Movements

  1. Martin Rolfs1,3
  1. 1Department of Psychology, Humboldt-Universität zu Berlin, Berlin 10099, Germany
  2. 2Berlin School of Mind and Brain, Humboldt-Universität zu Berlin, Berlin 10099, Germany
  3. 3Bernstein Center for Computational Neuroscience Berlin, Berlin 10099, Germany

  1. Author contributions: C.K. and M.R. designed research; C.K. performed research; C.K. and M.R. contributed unpublished reagents/analytic tools; C.K. analyzed data; C.K. and M.R. wrote the paper.

 
Next Section

Abstract

Sudden environmental changes can render planned hand movements suboptimal or even counterproductive. To prevent the execution of outdated motor plans, the motor system may transiently inhibit actions following salient changes, allowing time to evaluate alternatives. While such a mechanism is well established for eye movements, its applicability to hand movements remains unclear. Here, we present findings from three online behavioral experiments and two lab-based replications designed to probe key features of this mechanism in manual responses: reflexive inhibition, temporal precedence, complete movement updating, and sensitivity to saliency. Participants of either sex performed rapid sequential tapping movements toward onscreen targets. At an unpredictable time, either a relevant change (a target displacement) or an irrelevant change (a brief luminance flash) occurred. We measured movement initiation rates following these changes and compared them with a no-change baseline. A significant transient inhibition of movement initiation followed both relevant and irrelevant changes. This inhibition preceded observable updates to the movement plan. At the time of inhibition release, the update to a movement plan was complete. Across experiments, we observed stronger inhibitory effects for more salient changes. The lab-based replication confirmed that the latency of this inhibitory response aligns with visuomotor reaction times. These results support the existence of a general-purpose, rapid inhibitory mechanism in hand movements analogous to inhibition in the oculomotor system. We propose that such inhibition provides a reflexive, domain-general safeguard against obsolete actions following unexpected changes.

Previous SectionNext Section

Significance Statement

As we act within dynamic environments, sudden changes can invalidate our planned movements. In such cases, rather than relying on continuous sensorimotor integration, the brain may employ a different mechanism: the rapid inhibition of potentially outdated actions. Our studies show that the human motor system exhibits abrupt, nonselective inhibition in response to unexpected changes. This response occurs before the selection of a new action, suggesting a central, preemptive control process. These findings highlight an automatic, stimulus-driven mechanism that interrupts ongoing motor activity. This work advances our understanding of how the brain prioritizes error prevention over movement execution in action planning, with implications for models of sensorimotor control.

Previous SectionNext Section

Introduction

Humans navigate dynamic environments, where actions depend on changing states of the world. Consider an ice hockey goalkeeper: as players and puck shift positions, they must continuously update predictions about where the puck might strike. Experts can anticipate these changes effectively, but novices are often caught off guard (Goettker et al., 2021). Reacting to a novel stimulus requires stimulus recognition (Donders, 1969; Franklin and Wolpert, 2008), and a new motor plan must be formed. This process takes 150–200 ms (Wolpert and Landy, 2012; Wong et al., 2015; Gallivan et al., 2018), during which outdated motor plans that were formed before the unexpected change may be executed unless actively interrupted. Executing obsolete actions wastes energy and can lead to unwanted outcomes—like conceding a goal. Thus, swiftly interrupting outdated movements to allow updated responses would be adaptive.

Saccadic inhibition: interrupting outdated actions

One candidate mechanism for rapid interruption exists in eye movement control. Humans make about three large eye movements (saccades) per second and frequent smaller shifts during fixation (microsaccades). When a sudden sensory change occurs—visual (Reingold and Stampe, 1999, 2002, 2004), auditory (Rolfs et al., 2005, 2008), or tactile (Ossandón et al., 2015, 2020, Badde et al., 2020)—these eye movements are briefly paused. This effect, known as “saccadic inhibition,” begins ∼60 ms after stimulus onset, peaks at ∼90 ms, and resolves by ∼130 ms, followed by a rebound in movement rate. Saccadic inhibition has been modeled as a pause in the evidence accumulation process that allows reevaluation of new information (Salinas and Stanford, 2018). This delay improves behavioral accuracy. For example, in double-step saccade tasks, participants make fewer errors when a distractor triggers saccadic inhibition (Buonocore et al., 2017). Contrary to this early interpretation of the pause as halting the evidence accumulation process, later modeling of neurophysiological activity in the frontal eye fields showed that pausing likely acts at the level of motor output rather than perception or decision-making (Shinn et al., 2022). When a mismatch between expected and actual sensory input is detected, the system suppresses the execution of an imminent eye movement to prepare a more accurate one. This approach offers an elegant solution: the integration of new information can continue while delaying the execution of (potentially obsolete) responses.

Saccadic inhibition exhibits four behavioral signatures:

  1. Reflexive inhibition: Any salient stimulus briefly suppresses saccade rates, regardless of task relevance (Reingold and Stampe, 1999, 2002, 2004).

  2. Temporal precedence: The interruption occurs before the stimulus is fully processed (Salinas et al., 2019).

  3. Complete movement update: The final response reflects the new target location, not a blend of old and new (Buonocore et al., 2017).

  4. Saliency dependence: Only stimuli above a perceptual threshold trigger inhibition (White and Rolfs, 2016); more salient stimuli elicit longer pauses (Bonneh et al., 2015, Scholes et al., 2015).

Does this mechanism extend to hand movements?

While eye movements primarily gather information, are brief, and energetically inexpensive, this is not true for other movements. We therefore asked: Does a similar rapid inhibition mechanism exist for skeletal muscles?

Initial evidence supports this idea. Corticomuscular resonance and action-stopping responses can briefly interrupt movements following sudden stimuli (Badry et al., 2009, Wessel and Aron, 2017, Novembre and Iannetti, 2021). The “pause-then-cancel” model describes a rapid inhibitory phase followed by cancelation of action (Diesburg and Wessel, 2021), resembling saccadic inhibition. However, no behavioral study has systematically examined parallels between saccadic inhibition and other motor systems.

We hypothesized that a mechanism akin to saccadic inhibition exists for hand movements (Fig. 1). To test this, we developed a task inspired by saccadic inhibition paradigms. We predicted four signatures: reflexive inhibition, temporal precedence, complete movement updating, and saliency dependence. Experiments were first conducted online, with participants completing tasks remotely on smartphones (“remote”). After confirming behavioral signatures, we replicated our findings in a lab-based, within-participant design to validate timing and confirm results in a controlled setting (“lab-based”).

Figure 1.
View larger version:
Figure 1.

Schematic depiction of the proposed mechanism. A visual signal (IN) is processed by the perceptual system through several steps, resulting in an action (OUT). Blue boxes indicate neurocognitive processing stages. Gray boxes indicate functional processing steps. Arrows indicate the flow of information. Pink arrows show information flow during the suggested inhibition process. Black arrows show information flow during information evaluation. Note that the inhibition path skips and does not influence the evaluation process (including late perceptual and motor control stages).

Previous SectionNext Section

Material and Methods

Experimental design

We designed our task—a sequential manual tapping paradigm—to imitate reading paradigms used to study saccadic inhibition (Reingold and Stampe, 1999, 2004). Participants tapped on a series of six dots (“movement targets”) from left to right with their index finger. The maximum time allowed to tap on all movement targets was 1,500 ms, introducing tight temporal constraints [average time per trial in Experiment 1 (remote): 1,102.07 ± 36 ms; mean ± SEM]. Movement targets were filled circles with one degree of visual angle (dva) diameter that disappeared as soon as they were touched. Each movement target was positioned around one of the six origin positions evenly spaced on a horizontal midline. For every trial, possible locations of the movement targets were jittered around the origin position by a random value between −1 and 1 dva along the x- and the y-axis.

At a random time point within 700 ms after the first tap, both behaviorally relevant (relevant) and behaviorally irrelevant (irrelevant) changes could occur. Both types of changes were present in 50% of all trials (independently chosen for relevant and irrelevant changes) and absent otherwise. Relevant and irrelevant changes were “absent” or “present” in an independently chosen 50% of all trials. We will refer to trials where both types of changes were absent as “baseline” conditions. Trials in which any change was present will be called “experimental” conditions. In the “relevant-only” condition, the relevant change is present, and the irrelevant change is absent. In the “irrelevant-only” condition, the relevant change is absent, and the irrelevant change is present. If both changes were present, the trials will be referred to as “combined” conditions.

Behaviorally relevant stimulus

The relevant stimulus was a jump—a change in the position of the movement targets. Each target jumped to a new location around the origin position (Fig. 2A, bottom panel). Targets that had previously been tapped on and disappeared remained invisible. Participants were instructed to tap on the target, even after it jumped, making this change relevant. We accepted responses within a radius of 1.5 dva around the target location, allowing the trial to continue even if participants accidentally tapped at the old location. The relevant change was the same across all experiments.

Figure 2.
View larger version:
Figure 2.

Paradigm, A, Illustration of the basic task and relevant changes used in all experiments, remote and lab-based. Participants were instructed to tap on the targets (black circles) from left to right. A relevant change appeared in 50% of all trials. In these trials, the movement targets changed their position to the illustrated dashed positions. On each trial, positions were jittered relative to six origin positions (white crosses; not shown in the experiment). B, Illustrations of the irrelevant changes in Experiments 1–3. In Experiment 1, the background changed to white for 33 ms. In Experiment 2, the targets changed to white. In Experiment 3, the background was black while the targets were gray, and the change was a white flash inside the targets. C, Distribution of tap onsets for each target (gray levels) relative to the first target being touched in Experiment 1 (remote, collapsed across all observers and conditions). The black bar indicates the time frame during which changes occurred. D, Touch onset times in Experiment 1 (remote) relative to the time of a change.

Behaviorally irrelevant stimuli

The irrelevant stimulus was a flash—a 33-ms-long change in luminance in the background (Experiment 1 remote and lab-based) or the movement target (Experiments 2, remote and lab-based, and Experiment 3, remote). Participants were instructed to ignore the flash and continue tapping on the movement targets. The type of flash differed between experiments and is described below. Because the studies were run in a browser, an exact value for the maximum luminance could not be measured. Instead, we prompted participants to set their screens to full brightness before starting the experiment and used the HEX color white (#FFFFFF) to implement a flash with maximum luminance.

Experiment 1: flashes in the background—medium contrast

Experiment 1 (remote and lab-based) showed movement targets with minimal luminance (HEX #000000, black) against an intermediately bright background (HEX #666666, gray). The background color changed from intermediate to maximum luminance during the irrelevant flash. Movement targets remained at minimal luminance throughout the flash (Fig. 2B, Experiment 1).

Experiment 2: flashes in the targets—high contrast

Experiment 2 (remote and lab-based) showed the targets and the background at the same luminance levels as Experiment 1. The movement targets—rather than the background—changed from minimal to maximal luminance during the flash (Fig. 2B, Experiment 2). Targets that had already been touched and had disappeared remained invisible.

Experiment 3: flashes in the targets—medium contrast

Experiment 3 (remote) showed movement targets at the intermediate and the background at minimal luminance, thus inverting the luminance scheme of Experiments 1 and 2 (Fig. 2B, Experiment 3). The irrelevant stimulus was a flash inside the movement targets, which turned from intermediate to maximal luminance. Note that this experiment combines the spatial location of the irrelevant change used in Experiment 2 (the movement target) with the contrast of the irrelevant change used in Experiment 1 (medium).

Remote studies, recruitment, and instruction workflow

Participants completed our experiment through the web browser on their smartphones and tablets. We adjusted our typical data collection approach to ensure data quality. This adjustment included splitting the experiment into four sessions, which were not longer than 40 min, including instructions and calibration. This change was made to ensure that participants would remain concentrated. All four sessions were identical in structure except for the first session, which included video instructions, quiz-style comprehension tests of the task, and giving informed consent. Participants completed 200 trials (50 per factor level combination) per session, for a total of 800 trials. Participants were given breaks during each session. Between sessions, submissions from each participant were checked for compliance with the instructions, and participants were excluded if their tapping behavior indicated that they did not follow the instructions.

Implementation

All three experiments were implemented in JavaScript with jsPsych version 6.3.1 (https://www.jspsych.org/6.3/; Leeuw, 2015; Leeuw et al., 2023). Individual trials were implemented with a custom-written jsPsych plugin, which was called with trial-specific parameters. The plugin used an html canvas element to draw experimental stimuli, providing better timing control than other html elements (Kuroki, 2021).

All experiments were hosted on an instance of a JATOS (https://www.jatos.org/) server (Lange et al., 2015) owned by Humboldt-Universität zu Berlin.

Calibration

In remote and lab-based studies alike, the size of the display and viewing distance were calibrated before each session. We modified the virtual chinrest plugin from jsPsych and Li et al. (2020) to touchscreen requirements. In the virtual chinrest procedure, participants first adjust the image of a credit card on their screen to the size of a physical credit card or driver's license. This method translates the screen size in pixels into millimeters or inches more precisely than relying on the CSS pixel size. In the original version of the virtual chinrest, the ratio of height and width of the credit card is fixed. In our version, participants were allowed to adjust them independently. We used this ratio to check the participant’s scaling accuracy. Credit card sizes were only accepted when the ratio between width and height was within 10% of the true ratio (1.588 <mml:math display="inline"><mml:mrow><mml:mi mathvariant="normal">card</mml:mi></mml:mrow><mml:mspace width=".1em"></mml:mspace><mml:mrow><mml:mi mathvariant="normal">width</mml:mi></mml:mrow><mml:mo>/</mml:mo><mml:mrow><mml:mi mathvariant="normal">card</mml:mi></mml:mrow><mml:mspace width=".1em"></mml:mspace><mml:mrow><mml:mi mathvariant="normal">height</mml:mi></mml:mrow></mml:math>cardwidth/cardheight ).

In the next step of the virtual chinrest, we estimated the blind spot position, allowing us to calculate the distance at which the participant was sitting from the screen. Participants were asked to focus with their left eye on a black square located at the outer right border of the screen while either closing or covering their right eye. A red circle moved from right to left, away from the black square. Participants had to tap on the screen when the red circle visually disappeared due to it entering the blind spot, prompting the procedure to start again. After five repetitions, the average position of the circle at perceived disappearance was recorded. This distance was assumed to be ∼13.5 dva—the average distance from the fovea at which the dot enters the blind spot (Li et al., 2020). Using the screen size in pixels and the translation into dva, the virtual chinrest plugin computes the participant's distance from the screen. We used this information to scale all movement targets to be uniform in dva across participants.

The calibrated distance to the screen was restrained to be between 20 and 60 cm, and participants confirmed the calibrated distance. If the calibration failed, participants were shown the instruction video (in remote studies) or were given additional help from the experimenter (in lab-based studies) and could retake the calibration. The mean viewing distance of participants in the remote sessions was 336 ± 3.96 mm (mean ± SEM). This translates to an average diameter of 6 ± 0.07 mm of the target. In lab-based sessions, the average viewing distance was 395 ± 5.49 mm, and the target size was 7 ± 0.10 mm.

Participants in remote studies

We recruited participants through Prolific (www.prolific.co), using the following filter: self-reported normal or corrected-to-normal vision, self-reported fluency in English, and an approval rate between 90 and 100% on prolific. In the description of each study, we informed participants that the complete experiment would consist of four sessions. For every session after the first, we invited only participants who had completed the previous session.

To create an incentive to complete every session, we used an incremental compensation structure, where participants received a higher compensation relative to the duration of each session as they progressed in the experiment. In all sessions, payment complied with Prolifics minimal pay regulations of 5£/h in the first study and 6£/h in the latter two experiments. One person participated in both Experiments 2 (remote) and 3 (remote).

Participants in Experiment 1 (remote) were on average 27.7 years old (range, 19–50). They were based in 11 different OECD countries. Thirteen participants self-reported their sex as female and seven as male. The average age in Experiment 2 (remote) was 28.0 years (range, 19–43). Participants resided in eight different OECD countries, and 9 indicated their sex as male and 11 as female. In Experiment 3 (remote), the average age was 29.7 years (range, 20–64), and participants resided in nine different OECD countries. Self-reported sex was 8 females and 12 males.

Pros and cons of our online methodology

Conducting these studies online on mobile devices allowed rapid, diverse data collection on complex motor behavior in naturalistic settings. However, it sacrifices precise control over stimulus timing, response measurement, and environmental factors. Touchscreen delays typically add 70–80 ms with up to 10 ms SD (Anwyl-Irvine et al., 2021), confounding absolute latency estimates but preserving relative timing across same-modality measures. Consequently, we cannot specify exact inhibition onsets. Still, our results demonstrate that inhibitory patterns are robust enough to emerge under low experimental control—a promising avenue for high-throughput motor–control research complementary to lab-based validation.

Lab-based replication, setup, and participants

Experiments 1 (lab-based) and Experiment 2 (lab-based) were replicated in a within-participant study. For this study, we invited participants to our lab at Humboldt-Universität zu Berlin. We maintained the online character of previous studies and asked participants to bring their phones. We presented the experiment in their browser, identical to the setup for online participants. In the lab, the experimenter ensured the screen's brightness was set to its full value, and notifications were turned off. Participants would sit in a chair in a room with constant artificial lighting. The smartphone was mounted on a holder, which itself was mounted on the table. This ensured a relatively constant viewing distance from the smartphone (Fig. 3).

Figure 3.
View larger version:
Figure 3.

Setup of the lab-based replication of Experiments 1 (lab-based) and 2 (lab-based). The smartphone was held inside the smartphone stand. A, Setup without the phone. B, Setup with the phone, with labels for the photodiode (1), the microphone (2), AprilTag markers (3), and the Raspberry Pi (4). C, Close-up of the phone screen, showing how the photodiode is mounted.

A photodiode (grove sensor, Seeed Technology; Fig. 3A) was mounted in the top-right corner of the smartphone. This light sensor recorded brightness changes in that part of the screen—the “photodiode rectangle.” This rectangle was set up in the online experiment to signal important time points, such as changes and block onsets. A microphone (grove loudness sensor, Seeed Technology; Fig. 3B) was mounted below the smartphone stand. This sensor registered the sound participants made when tapping on the screen. The photodiode and microphone readouts were recorded at 300 Hz with a Raspberry Pi 3+ (Raspberry Pi) with a grove head mount (Seeed Technology) running a custom Python script (Fig. 3D). At the beginning of each session, the device's light and sound recordings were checked to ensure the diode was placed correctly and the taps on the screen were audible. We recorded eye movements with a wearable PupilCore eye tracker (PupilLabs). These data served as a control to show that the relevant and irrelevant changes triggered saccadic inhibition. The data are included in the supplementary material (Text S1, Fig. S1).

At the beginning of each session, participants scanned a QR code for the scheduled session, which started the experiment on their phones. Each participant took part in four 1 h sessions. In two sessions, participants performed the task of Experiment 1 (lab-based), and in two separate sessions, they performed the task of Experiment 2 (lab-based). The order of the sessions was counterbalanced, with half of the participants starting with Experiment 1 (lab-based) and the other half starting with Experiment 2 (lab-based).

Participants were recruited through a local platform for Humboldt-Universität zu Berlin (Psychologischer Experimental-Server Adlershof) and through word of mouth. Twenty-one participants were recruited for the study, one of which was excluded due to technical problems. The remaining participants were, on average, 26.3 years old (range, 19–35). Six participants self-identified as male, 13 as female, and 1 as agender. Four participants were left-eye dominant, and one participant was left-handed. Participants were debriefed by the experimenter about the setup and the data types recorded by the photodiode, the microphone, and the eye tracker. They received the instructions about the task in a video format, like the online participants. Participants were compensated for their participation, either financially or with experimental participation credit.

Statistical analysis

Two types of data were collected in our studies. “Browser data” from the online study was collected in remote and lab-based studies. “Sensor data” from the photodiode and microphone were only collected in lab-based studies. Browser data from all three experiments were analyzed using the same pipeline. Sensor data were preprocessed to the same format as browser data and then analyzed with the same pipeline described below, where applicable.

Movement onset rate analysis

We aligned all touch offsets—equivalent to the onset of a movement—to the onset of a change (Fig. 2C,D). We computed each participant’s movement onset rate per second (from here, movement rate) from the movement onset times, following the procedure described in Rolfs at al. (Rolfs et al., 2008), with a kernel width of α = 1/50. This was done separately for each participant and factor combination.

We corrected the movement rates for the upper bound of movements that could have occurred by dividing by the number of trials that contributed to each time point (1 ms resolution). The final value of movement rates reflects the number of movement onsets/second in one standardized trial. These movement rates are shown in Figure S2.

As fewer responses were recorded toward the end of each trial (participants tended to slow down over the trial), we expressed movement rates as their difference from the mean baseline rate. The mean baseline rate was computed for each participant as the average movement rate in the null condition across all other participants. Normalized movement rates in experimental conditions were compared against normalized movement rates in the null condition with a cluster-based permutation test (see below).

Movement endpoint analysis

We analyzed whether movement endpoints discretely reflected the old (shown before) or the new (shown after the relevant change) movement target or whether there was an average between both. For this analysis, we collapsed across conditions with and without an irrelevant change. In trials without a relevant change, we simulated a movement target after the change by randomly permutating movement targets around the same origin, creating realistic alternative movement targets.

We computed the angle between the old and the new movement target. We rotated the touch positions by the inverse of this angle, aligning all touches in a 90° angle from the original target. In this alignment, the x-component of the touch position reflects the distance from the original target, and the y-component reflects orthogonal variation. We then normalized the touch position by dividing the x- and y-components by the distance between the old and the new target.

After these transformations, the touch position along the x-axis indicates the proportion of the distance traveled between the old and the new target. Values ∼0 are movement endpoints on the old, while values ∼1 are movement endpoints on the new target. We created time-by-position frequency heatmaps, where the time axis indicated the time a movement started (25 ms resolution), and the position axis indicated the normalized endpoint (0.1 proportional resolution). We compared the heatmaps from trials with a task-relevant change to heatmaps from trials without a relevant change with a cluster-based permutation test (Fig. 4D–F).

Cluster-based permutation tests

We used cluster-based permutation tests to compare timelines or heatmaps. In those tests, we computed the difference between each experimental condition and the baseline condition. For each point on a timeline or entry on a heatmap, we computed the t-value of the difference based on the mean and standard error between participants. Clusters above a critical t-value of 2.093 (corresponding to a two-sided student t-test with p = 0.05) were recorded. Next, we randomly permutated the condition labels 1,000 times and identified the largest clusters above the critical t-value arising in these random permutations. We took the 95% percentile of the largest clusters as the cutoff value for significance. The center of each cluster was defined as the average weighted by the t-value along each investigated axis (time for all two-dimensional comparisons, time and movement endpoint for the three-dimensional comparisons).

Signal detection for photodiode and sound sensors

Detecting changes in the photodiode data

An increase in the brightness of the photodiode rectangle accompanied all changes (relevant and irrelevant). This increase lasted 33 ms. The photodiode rectangle flashed regardless of whether changes were present or absent. Because the photodiode covered the flashing rectangle, the participants did not see these flashes. We detected time points in the light data stream when the brightness increased above a set threshold. Thresholds were set by hand to account for differences in screen brightness and positioning of the photodiode that changed the recorded signal. From the peaks detected above the threshold, we went backward to find the first time the brightness started increasing. This time point was recorded as the change onset.

Detecting movement onsets in the sound data

We applied a bandwidth filter to the recorded noise time series to extract the tap onsets from the sound data. The bandwidth filter was adjusted manually for each recording, with lower bounds between 30 and 100 Hz and upper bounds between 90 and 160 Hz. We detected peaks in the bandwidth-filtered data with a threshold set individually per participant. Going backward and forward in time, we detected the sound onset, indicating the contact between the fingertip and the phone. Note that we are recording the sound of participants tapping on the phone—which is the movement offset. In all analyses above, we used the movement onset to compute tapping rates. To calculate the movement onset from the detected offsets, we subtracted the relative timing between the movement offset and the movement onset recorded in the browser data.

Alignment of browser and sensor data

To align the data streams, we relied on a signal that all data streams could record. This signal was the photodiode rectangle changing between 1 s white and 1 s black five times. In the browser data, these time points were registered when the command to change the screen was sent. In the lab-based sensor data, the photodiode registered the light changes.

Code accessibility

Code and data are available through the Open Science Framework: Experiment 1 (remote), https://osf.io/fs985/; Experiment 2 (remote), https://osf.io/43dxc/; Experiment 3 (remote), https://osf.io/mkq84/; and Experiment 1 + 2 (lab-based), https://osf.io/tzmbr/.

Previous SectionNext Section

Results

Relevant and salient irrelevant changes are followed by movement inhibition

We first probed our data for the signature of inhibition: a drop in movement rates after salient changes. According to our hypothesis, we expected to see reduced movement rates, regardless of the behavioral relevance of the change. During the 100 ms before change onset, movement rates were on average 4.73 ± 0.07 s−1 (mean ± SEM) in the baseline condition of Experiment 1 (remote). In the same time window, movement rates in the experimental conditions did not differ from the baseline (combined, 4.68 ± 0.07 s−1; relevant-only, 4.67 ± 0.07 s−1; irrelevant-only, 4.65 ± 0.08 s−1). Experiment 2 (remote) (baseline, 4.78 ± 0.12 s−1; combined, 4.70 ± 0.11 s−1; relevant-only, 4.76 ± 0.10 s−1; irrelevant-only, 4.76 ± 0.10 s−1) and Experiment 3 (remote; baseline, 4.86 ± 0.11 s−1; combined, 4.89 ± 0.1 s−1; relevant-only, 4.9 ± 0.11 s−1; irrelevant-only, 4.84 ± 0.10 s−1) featured similar movement rates (Fig. S2).

As predicted, movement rates dropped in response to a stimulus change in all experimental conditions. In the combined condition of Experiment 1 (remote; Fig. 4A), movement rates were reduced to 62 ± 4% (mean ± SEM) of the baseline rate in a cluster between 266 and 646 ms (t-sum = −3,239.64). Experiment 2 (remote; 62 ± 4%; 278–643 ms; t-sum = −2,486.07; Fig. 4B) and Experiment 3 (remote; 66 ± 4%, 279–651 ms; t-sum = −2,923.45; Fig. 4C) showed similar reductions. These reductions in movement rates were followed by a rebound. In Experiment 1 (remote), movement rates increased to 113 ± 5% between 732 and 1066 ms (t-sum = 1,585.95). Similar increases were observed in Experiment 2 (remote; 117 ± 7%; 713–946 ms; t-sum = 890.54) and Experiment 3 (remote; 117 ± 7%; 741–1,053 ms; t-sum = 1,425.50).

Figure 4.
View larger version:
Figure 4.

Main results: movement rates and endpoints. A–C, Movement rates over time as a function of relevant (solid lines, present; dashed lines, absent) and irrelevant change (colored, present; black, absent). Error bands (shaded areas in panels A–I) indicate the 95% CI across participants. Horizontal lines at the bottom indicate significant clusters in which the movement rate (in a given condition) deviates from the baseline condition. D–F, Relative movement endpoint frequencies in trials with, compared with trials without, relevant changes. Heatmaps show the contrast relevant change present—relevant change absent of proportion of answers made toward a normalized location relative to the movement target shown before (blue line) and after (red line) the change. Clusters in which the difference is significant are highlighted with a black outline. Black symbols (+) indicate the center of the cluster. Dashed vertical lines illustrate the time window during which movements were inhibited after irrelevant changes.

Movement rates in trials with the relevant-only condition resembled movement rates in the combined condition. We observed an initial decrease in movement rates [Experiment 1 (remote), 65 ± 4%; 264–635 ms; t-sum = −3,055.55; Experiment 2 (remote), 65 ± 4%, 273–626 ms; t-sum = −2347.43; Experiment 3 (remote), 70 ± 4%; 297–659 ms; t-sum = −2,625.41] that was followed by a rebound [Experiment 1 (remote), 112 ± 4%; 759–1,066 ms; t-sum = 1,294.50; Experiment 2 (remote), 117 ± 7%; 791–944 ms; t-sum = 517.07; Experiment 3 (remote), 121 ± 7%; 738–1,054 ms; t-sum = 1,358.89].

We were particularly interested in movement rates in the irrelevant-only condition because we predicted that movement rates should be inhibited regardless of the behavioral relevance of the stimulus. We observed significant clusters of lower movement rates after irrelevant stimuli in Experiment 1 (remote; 93 ± 3%; 319–486 ms; t-sum = −535.27), where the stimulus was a medium-contrast flash in the background. We also observed a reduction in movement rates in Experiment 2 (remote; 91 ± 4%; 304–479 ms; t-sum = −661.18), where the stimulus was a high-contrast flash inside the movement targets. We did not find a significant reduction in movement rates in Experiment 3 (remote), where the stimulus was a medium-contrast flash inside the movement targets. A cluster around the time of an expected decrease in movement rates did not cross the threshold for significance (96 ± 3%; 422–446 ms; t-sum = −55.21). In Experiment 3 (remote), but not in the other experiments, we found a rebound-like pattern in the irrelevant-only condition (107 ± 5%; 809–1,000 ms; t-sum = 455.61).

Movement endpoints reflect two stages: a reduction of movements to the old target, followed by an increase of movements to the new target

In all three experiments, an analysis of movement endpoints showed an early reduction of movements toward the old target and a later increase in movements toward the new target. The reduction of responses toward the old target centered at a latency of 600 ms, and a relative position of −20% in Experiment 1 (remote; t-sum = −769.84; Fig. 4D). The other two experiments showed highly comparable patterns [Experiment 2 (remote), center at 600 ms; −20% relative position; t-sum = −792.74; Experiment 3 (remote), center at 600 ms latency, −10% relative position; t-sum = −563.50; Fig. 4E,F]. The increase in movement frequency toward the new target was centered 150 ms after the earlier cluster. In Experiment 1 (remote), this cluster center was at 750 ms and 110% relative location (t-sum = −1,016.55). Again, the other two experiments replicated the results from Experiment 1 (remote) [Experiment 2 (remote), center at 775 ms latency; 100% relative position; t-sum = 968.92; Experiment 3 (remote), center at 775 ms latency; 110% relative position; t-sum = 777.66].

Results from sensor data

The inhibition of hand movements starts 200 ms after any change

We replicated the movement rate analysis using the change onsets detected by a photodiode and the movement offsets detected in the sound recording of a microphone. These two measures are independent of processing times induced by the online device and provide an accurate estimate of the inhibition delay. The inhibition after a combination of the relevant and the irrelevant change lasted from 210 to 628 ms in Experiment 1 (lab-based; 63 ± 6%; t-sum = −2,721.99) and from 195 to 543 ms in Experiment 2 (lab-based; 54 ± 6%; t-sum = −2,481). When only the relevant change was displayed, hand movements were inhibited between 210 and 601 ms in Experiment 1 (lab-based; 65 ± 6%; t-sum = −2,176.37) and between 194 and 542 ms in Experiment 2 (lab-based; 60 ± 6%, t-sum = −2,404.38). After the irrelevant change, hand movements were inhibited between 204 and 374 ms for changes inside the movement target [Experiment 2 (lab-based), 88 ± 7%; t-sum = −557.83]. A comparable cluster was not found after irrelevant changes in the background.

The start point of corresponding inhibition clusters detected in the browser data of the lab-based study was systematically delayed by up to 100 ms and comparable in timing to the clusters found in the remote study. In the combined condition, the browser data showed inhibition between 291–592 ms [Experiment 1 (lab-based), t-sum = −2,073.19; +81 ms] and 282–604 ms [Experiment 2 (lab-based), t-sum = −1,902.33; +87 ms]. In the relevant-only condition, the inhibition was found between 309–606 ms [Experiment 1 (lab-based), t-sum = −1,743.78; +99 ms] and 289–592 ms [Experiment 2 (lab-based), t-sum = −2,004.01; +94 ms]. Finally, after irrelevant changes, movement rates were inhibited between 312–466 ms [Experiment 1 (lab-based), t-sum = −563.68; 102 ms] and 277–435 ms [Experiment 2 (lab-based), t-sum = −666.11; +83 ms]. The movement rates are shown in Figure 5.

Figure 5.
View larger version:
Figure 5.

Movement rates calculated from tapping sound and photodiode onset. Frequency of movement onsets detected in the sound data (y-axis) as a function of the onset of a change detected through the photodiode (y-axis). Panels (A) and (B) show normalized tapping rates relative to the baseline (dashed, black). Panels (C) and (D) show the significant clusters detected in the browser data from lab-based experiments to compare the delay visually. Conventions as in Figure 4.

Previous SectionNext Section

Discussion

Across three remote and two lab-based experiments, we showed that sudden, salient changes elicit a rapid inhibition of hand movements, followed by fully updated movements to the new target—evidence that the outdated motor plan was indeed discarded in favor of a motor plan that was based on new information. In analogy to saccadic inhibition in the oculomotor system, we hypothesized and confirmed three key behavioral signatures: (1) reflexive inhibition of movement rates after both relevant and irrelevant changes, (2) temporal precedence of inhibition before movement updates, and (3) complete movement updates reflecting new targets rather than blends. Weak evidence was present for the fourth signature, a saliency-dependent modulation of inhibition magnitude. These findings support a model in which motor output is briefly paused without disrupting perceptual decision stages, thereby facilitating accurate responses after unexpected environmental changes.

Reflexive inhibition: movement rate drops

The most pronounced signature of saccadic inhibition is a drop of movement rates after any type of unexpected sensory signal. Experiments 1 (remote and lab-based) and 2 (remote and lab-based) confirmed the same signature in hand movement rates: significantly fewer movements were made even after task-irrelevant brightness changes. This is evidence that the change triggers inhibition before its relevance is evaluated. An alternative explanation is that inhibition following irrelevant stimuli reflects temporal conditioning due to their consistent pairing with relevant changes. This explanation seems unlikely, however, as Experiment 3's (remote) medium-contrast flashes did not induce inhibition despite identical temporal structure.

The inhibition onset after irrelevant changes coincided temporally with inhibition following relevant changes, indicating that the early drop in movement rates in relevant conditions partly stems from this pre-evaluation pause. Following a relevant change, movement rates fell to ∼60% of baseline within 400–550 ms across experiments. In the browser data, these drops started ∼300 ms after the change. This timing was surprisingly late. We suspected that the inhibition timing was late due to technical limitations of the remote recording. In a lab-based replication, we explicitly checked this suspicion by creating alternative ways to record changes (through a photodiode) and response offsets (through a microphone). When measuring the onset of the inhibition between these data points, we found that inhibition started 100 ms earlier than previously observed—200 ms after the change appeared on the screen, drops in movement rates were detected through the microphone. This timing aligns with the usual time of stop-signal responses (Wessel, 2025) and the planning of novel responses (Biguer et al., 1982; Desmurget et al., 1998).

Delays in the movement responses following relevant changes could also be explained by a smooth correction of the hand movement midair, resulting in a longer trajectory. To confirm that the movement stops before the trajectory changes, high-resolution motion-tracking studies are required. In our present dataset, the pause in the tapping rate is longer after relevant changes. We argue that the timing of the short pause after irrelevant changes, which coincides with the onset of the long pause after relevant changes, is evidence for a general inhibitory mechanism that precedes any trajectory adjustments. The differences in pause duration observed after relevant and irrelevant changes can be explained through additional processes, such as trajectory adjustments.

Another alternative explanation for the drop in movement rates could be that saccadic inhibition indirectly mediates hand movement rates. In a supplementary analysis (Fig. S3), we split trials into those in which saccades followed a relevant or an irrelevant event (suggesting a lack of saccadic inhibition), and those in which no saccades occurred in the same time window. Movement rates in both subsets of trials were strikingly similar, supporting the conclusion that hand movement inhibition is not exclusively a by-product of saccadic inhibition.

Inhibition temporally precedes a complete movement update

We examined the movement endpoints during relevant changes by normalizing the response between old and new target locations. We observed that movement endpoints clustered cleanly around the old or the new target. There was no evidence of intermediate or “mixed” endpoints. Despite allowing responses within a generous spatial window (1.5 dva), participants fully updated their movements to the new target by the time they completed the action. Comparing the timing of inhibition after irrelevant changes with movement endpoint responses clearly shows that the automatic inhibition aligns with the reduction of responses to the old movement target and temporally precedes the cluster of movements to the new target by ∼150 ms. This comparison demonstrates that the inhibitory pause—visible in responses to irrelevant changes—is the signature of dropping outdated movements in favor of those whose planning relied on new information.

Saliency dependence

While we did not include an explicit manipulation of saliency, we can post hoc compare inhibition magnitudes across three brightness-change conditions (Fig. 6). The largest inhibition followed a full-screen flash from black to white (Experiment 1, remote and lab), the next largest followed a maximum-contrast flash inside the target area (Experiment 2, remote and lab), and no inhibition occurred for a medium-contrast flash inside the target (Experiment 3, remote). These results parallel saccadic inhibition findings: inhibition strength and duration scale with stimulus saliency (Bonneh et al., 2015; Scholes et al., 2015; White and Rolfs, 2016), not simply spatial coincidence with the movement target (Reingold and Stampe, 2004).

Figure 6.
View larger version:
Figure 6.

Mean and individual participant’s movement rate (normalized) at maximum inhibition after an irrelevant change. Maximum inhibition (the lowest recorded movement rate) occurred at 417 ms (Experiment 1), 418 ms (Experiment 2), and 434 ms (Experiment 3).

Because we did not systematically manipulate contrast and location within the same participants, future work should vary these parameters in a fully crossed design to confirm the threshold dynamics of manual inhibition.

Relation to existing motor inhibition models

Our findings align with rapid inhibitory signatures in skeletal-muscle control, such as corticomuscular resonance (Novembre et al., 2018), global action-stopping responses (Badry et al., 2009), and post-novelty slowing (Wessel and Aron, 2017), and with the “pause-then-cancel” model of action stopping (Diesburg and Wessel, 2021). In the pause-then-cancel framework, an initial fast pause halts motor output, and then a slower pathway cancels or alters the command. We propose that this initial pause is functionally equivalent to saccadic inhibition, serving a generic role for canceling outdated movements. Unlike traditional stop-signal tasks, where inhibition is task-explicit, our paradigm shows that inhibition can unfold automatically under time pressure to update actions in response to surprising stimuli.

A large and convincing body of research shows that movement updates can be fast, appearing just 100 ms after a stimulus onset (Brenner and Smeets, 1997, 2023; Wijdenes et al., 2011), and automatic—without signatures of overt stopping (Prablanc and Martin, 1992; Prablanc et al., 2003). The timing of inhibition in our studies is too slow to interrupt such fast, automatic online corrections, even after correcting for delays in the browser data. Studies of automatic movement updates often work with a limited set of optional targets (Elliott et al., 2017; Laquitaine and Gardner, 2018). In this case, multiple motor plans may be prepared in parallel, enabling rapid switching between them (Haith et al., 2015; Xivry et al., 2017; Enachescu et al., 2021; Meirhaeghe et al., 2023). In our study, we tried to prevent participants from being able to rely on pre-planned, alternative movement plans by randomizing the locations where the new targets could appear in each trial. Other studies that have shown remarkably fast reaction times have used small changes to the visual stimulus (Brenner and Smeets, 2023). In such cases, it is possible that these changes are indistinguishable from perceptual noise and integrated into a perception of a single, moving target, rather than a surprising, nonintegrable change.

Another notable difference is that many studies of online movement correction involve reaching movements covering larger distances than the hand movement in our paradigm. Our results could thus be specific to the small-distance movements in our paradigm and not generalize to larger movements.

To reconcile frameworks of smooth adjustments and automatic pauses, we suggest a prediction-error-like mechanism. For visual changes, such a mechanism is likely located in the visual cortex (V1). V1 is central in generating the phenomenon of saccadic inhibition (White and Rolfs, 2016; White et al., 2022; Malevich et al., 2024). It also connects to subcortical areas that control hand and other limb movements (Battaglia-Mayer et al., 2016; Cont and Zimmermann, 2021). Hence, V1 could generate a disruption signal, traveling to subcortical areas, blocking the execution of movements if the detected change is surprising enough to generate a prediction error. Regardless of the presence of a prediction error, integration of new visual information into a motor plan would continue. The difference would be that in the absence of a disruption of execution, the integrated answer would be translated into an overt response, whereas in the case of disruption, the integration would remain covert.

Previous SectionNext Section

Conclusion

Our sequential tapping paradigm reveals four behavioral signatures of rapid hand movement inhibition—reflexive inhibition, temporal precedence, full movement updating, and saliency dependence—closely mirroring saccadic inhibition phenomena. These results support a framework in which motor output is briefly paused upon detecting salient changes, preventing execution of outdated plans while leaving perceptual and decision processes uninterrupted. This pause ends as soon as a fully updated motor action is executed. This general motor inhibition mechanism likely operates across effectors to optimize adaptive behavior in dynamic environments, eliminating the need for upstream interruptions of sensory or cognitive processing. Future work with precise motion tracking and systematic stimulus manipulations will further elucidate the timing, neural pathways, and functional benefits of this ubiquitous inhibitory pause.

Previous SectionNext Section

Footnotes

    • Received July 7, 2025.
    • Revision received December 23, 2025.
    • Accepted January 10, 2026.

  • We thank Stefania Cionka and Laura Lyra, who helped in the early stages of developing the online task. Armin Israel helped collect data and recorded the instruction videos for Experiment 3. Casimir Ludwig gave feedback at different stages of the project. C.K. was funded by the Berlin School of Mind and Brain. M.R. was supported by the Heisenberg Program of the Deutsche Forschungsgemeinschaft, DFG (RO 3579/8-1, RO 3579/10-1, and RO 3579/12-1), and the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Grant Number 865715).

  • The authors declare no competing financial interests.

  • This paper contains supplemental material available at: https://doi.org/10.1523/JNEUROSCI.1319-25.2026

  • Correspondence should be addressed to Clara Kuper at clara.kuper{at}posteo.de or Martin Rolfs at martin.rolfs{at}hu-berlin.de.
Previous Section
 

References

  1. ↵
    1. Anwyl-Irvine A,
    2. Dalmaijer ES,
    3. Hodges N,
    4. Evershed JK
    (2021) Realistic precision and accuracy of online experiment platforms, web browsers, and devices. Behav Res Methods 53:1407–1425. https://doi.org/10.3758/s13428-020-01501-5
    Medline
  2. ↵
    1. Badde S,
    2. Myers CF,
    3. Yuval-Greenberg S,
    4. Carrasco M
    (2020) Oculomotor freezing reflects tactile temporal expectation and aids tactile perception. Nat Commun 11:3341. https://doi.org/10.1038/s41467-020-17160-1
    CrossRefMedline
  3. ↵
    1. Badry R,
    2. Mima T,
    3. Aso T,
    4. Nakatsuka M,
    5. Abe M,
    6. Fathi D,
    7. Foly N,
    8. Nagiub H,
    9. Nagamine T,
    10. Fukuyama H
    (2009) Suppression of human cortico-motoneuronal excitability during the stop-signal task. Clin Neurophysiol 120:1717–1723. https://doi.org/10.1016/j.clinph.2009.06.027
    CrossRefMedline
  4. ↵
    1. Battaglia-Mayer A,
    2. Babicola L,
    3. Satta E
    (2016) Parieto-frontal gradients and domains underlying eye and hand operations in the action space. Neuroscience 334:76–92. https://doi.org/10.1016/j.neuroscience.2016.07.009
    CrossRefMedline
  5. ↵
    1. Biguer B,
    2. Jeannerod M,
    3. Prablanc C
    (1982) The coordination of eye, head, and arm movements during reaching at a single visual target. Exp Brain Res 46:301–304. https://doi.org/10.1007/BF00237188
    CrossRefMedline
  6. ↵
    1. Bonneh YS,
    2. Adini Y,
    3. Polat U
    (2015) Contrast sensitivity revealed by microsaccades. J Vis 15:11. https://doi.org/10.1167/15.9.11
    Abstract/FREE Full Text
  7. ↵
    1. Brenner E,
    2. Smeets JBJ
    (1997) Fast responses of the human hand to changes in target position. J Mot Behav 29:297–310. https://doi.org/10.1080/00222899709600017
    CrossRefMedline
  8. ↵
    1. Brenner E,
    2. Smeets JBJ
    (2023) Continuous use of visual information about the position of the moving hand. Exp Brain Res 241:2001–2008. https://doi.org/10.1007/s00221-023-06658-x
    Medline
  9. ↵
    1. Buonocore A,
    2. Purokayastha S,
    3. McIntosh RD
    (2017) Saccade reorienting is facilitated by pausing the oculomotor program. J Cogn Neurosci 29:2068–2080. https://doi.org/10.1162/jocn_a_01179
    CrossRefMedline
  10. ↵
    1. Cont C,
    2. Zimmermann E
    (2021) The motor representation of sensory experience. Curr Biol 31:1029-1036.e2. https://doi.org/10.1016/j.cub.2020.11.032
    CrossRefMedline
  11. ↵
    1. Desmurget M,
    2. Pélisson D,
    3. Rossetti Y,
    4. Prablanc C
    (1998) From eye to hand: planning goal-directed movements. Neurosci Biobehav Rev 22:761–788. https://doi.org/10.1016/S0149-7634(98)00004-9
    CrossRefMedline
  12. ↵
    1. Diesburg DA,
    2. Wessel JR
    (2021) The pause-then-cancel model of human action-stopping: theoretical considerations and empirical evidence. Neurosci Biobehav Rev 129:17–34. https://doi.org/10.1016/j.neubiorev.2021.07.019
    CrossRefMedline
  13. ↵
    1. Donders FC
    (1969) On the speed of mental processes. Acta Psychol 30:412–431. https://doi.org/10.1016/0001-6918(69)90065-1
    CrossRefMedline
  14. ↵
    1. Elliott D,
    2. Lyons J,
    3. Hayes SJ,
    4. Burkitt JJ,
    5. Roberts JW,
    6. Grierson LEM,
    7. Hansen S,
    8. Bennett SJ
    (2017) The multiple process model of goal-directed reaching revisited. Neurosci Biobehav Rev 72:95–110. https://doi.org/10.1016/j.neubiorev.2016.11.016
    CrossRefMedline
  15. ↵
    1. Enachescu V,
    2. Schrater P,
    3. Schaal S,
    4. Christopoulos V
    (2021) Action planning and control under uncertainty emerge through a desirability-driven competition between parallel encoding motor plans. PLoS Comput Biol 17:e1009429. https://doi.org/10.1371/journal.pcbi.1009429
    CrossRefMedline
  16. ↵
    1. Franklin DW,
    2. Wolpert DM
    (2008) Specificity of reflex adaptation for task-relevant variability. J Neurosci 28:14165–14175. https://doi.org/10.1523/JNEUROSCI.4406-08.2008
    Abstract/FREE Full Text
  17. ↵
    1. Gallivan JP,
    2. Chapman CS,
    3. Wolpert DM,
    4. Flanagan JR
    (2018) Decision-making in sensorimotor control. Nat Rev Neurosci 19:519–534. https://doi.org/10.1038/s41583-018-0045-9
    CrossRefMedline
  18. ↵
    1. Goettker A,
    2. Pidaparthy H,
    3. Braun DI,
    4. Elder JH,
    5. Gegenfurtner KR
    (2021) Ice hockey spectators use contextual cues to guide predictive eye movements. Curr Biol 31:R991–R992. https://doi.org/10.1016/j.cub.2021.06.087
    Medline
  19. ↵
    1. Haith AM,
    2. Huberdeau DM,
    3. Krakauer JW
    (2015) Hedging your bets: intermediate movements as optimal behavior in the context of an incomplete decision. PLoS Comput Biol 11:e1004171. https://doi.org/10.1371/journal.pcbi.1004171
    CrossRefMedline
  20. ↵
    1. Kuroki D
    (2021) A new jsPsych plugin for psychophysics, providing accurate display duration and stimulus onset asynchrony. Behav Res Methods 53:301–310. https://doi.org/10.3758/s13428-020-01445-w
    Medline
  21. ↵
    1. Lange K,
    2. Kühn S,
    3. Filevich E
    (2015) “Just another tool for online studies” (JATOS): an easy solution for setup and management of web servers supporting online studies. PLoS One 10:e0130834. https://doi.org/10.1371/journal.pone.0130834
    CrossRefMedline
  22. ↵
    1. Laquitaine S,
    2. Gardner JL
    (2018) A switching observer for human perceptual estimation. Neuron 97:462-474.e6. https://doi.org/10.1016/j.neuron.2017.12.011
    CrossRefMedline
  23. ↵
    1. Leeuw JR
    (2015) Jspsych: a JavaScript library for creating behavioral experiments in a Web browser. Behav Res Methods 47:1–12. https://doi.org/10.3758/s13428-014-0458-y
    CrossRefMedline
  24. ↵
    1. Leeuw JR,
    2. Gilbert RA,
    3. Luchterhandt B
    (2023) Jspsych: enabling an open-source collaborative ecosystem of behavioral experiments. J Open Source Softw 8:5351. https://doi.org/10.21105/joss.05351
  25. ↵
    1. Li Q,
    2. Joo SJ,
    3. Yeatman JD,
    4. Reinecke K
    (2020) Controlling for participants’ viewing distance in large-scale, psychophysical online experiments using a virtual chinrest. Sci Rep 10:904. https://doi.org/10.1038/s41598-019-57204-1
    CrossRefMedline
  26. ↵
    1. Malevich T,
    2. Yu Y,
    3. Baumann MP,
    4. Zhang T,
    5. Hafed ZM
    (2024) Sound activates a dormant visual-motor pathway bypassing primary visual cortex. J Vis 10:532. https://doi.org/10.1167/jov.24.10.532
  27. ↵
    1. Meirhaeghe N,
    2. Riehle A,
    3. Brochier T
    (2023) Parallel movement planning is achieved via an optimal preparatory state in motor cortex. Cell Rep 42:112136. https://doi.org/10.1016/j.celrep.2023.112136
    CrossRefMedline
  28. ↵
    1. Novembre G,
    2. Pawar VM,
    3. Bufacchi RJ,
    4. Kilintari M,
    5. Srinivasan M,
    6. Rothwell JC,
    7. Haggard P,
    8. Iannetti GD
    (2018) Saliency detection as a reactive process: unexpected sensory events evoke corticomuscular coupling. J Neurosci 38:2385–2397. https://doi.org/10.1523/JNEUROSCI.2474-17.2017
    Abstract/FREE Full Text
  29. ↵
    1. Novembre G,
    2. Iannetti GD
    (2021) Towards a unified neural mechanism for reactive adaptive behaviour. Prog Neurobiol 204:102115. https://doi.org/10.1016/j.pneurobio.2021.102115
    CrossRefMedline
  30. ↵
    1. Ossandón JP,
    2. König P,
    3. Heed T
    (2015) Irrelevant tactile stimulation biases visual exploration in external coordinates. Sci Rep 5:10664. https://doi.org/10.1038/srep10664
    CrossRefMedline
  31. ↵
    1. Ossandón JP,
    2. König P,
    3. Heed T
    (2020) No evidence for a role of spatially modulated α-band activity in tactile remapping and short-latency, overt orienting behavior. J Neurosci 40:9088–9102. https://doi.org/10.1523/JNEUROSCI.0581-19.2020
    Abstract/FREE Full Text
  32. ↵
    1. Prablanc C,
    2. Desmurget M,
    3. Gréa H
    (2003) Neural control of on-line guidance of hand reaching movements. Prog Brain Res 142:155170. https://doi.org/10.1016/S0079-6123(03)42012-8
    CrossRefMedline
  33. ↵
    1. Prablanc C,
    2. Martin O
    (1992) Automatic control during hand reaching at undetected two-dimensional target displacements. J Neurophysiol 67:455–469. https://doi.org/10.1152/jn.1992.67.2.455
    CrossRefMedline
  34. ↵
    1. Becker W,
    2. Deubel H,
    3. Mergner T
    1. Reingold EM,
    2. Stampe DM
    (1999) Current oculomotor research. In: Physiological and psychological aspects (Becker W, Deubel H, Mergner T, eds), pp 249–255. New York: Kluwer Academic/Plenum.
    Google Scholar
  35. ↵
    1. Reingold EM,
    2. Stampe DM
    (2002) Saccadic inhibition in voluntary and reflexive saccades. J Cogn Neurosci 14:371–388. https://doi.org/10.1162/089892902317361903
    CrossRefMedline
  36. ↵
    1. Reingold EM,
    2. Stampe DM
    (2004) Saccadic inhibition in reading. J Exp Psychol Hum Percept Perform 30:194–211. https://doi.org/10.1037/0096-1523.30.1.194
    CrossRefMedline
  37. ↵
    1. Rolfs M,
    2. Engbert R,
    3. Kliegl R
    (2005) Crossmodal coupling of oculomotor control and spatial attention in vision and audition. Exp Brain Res 166:427–439. https://doi.org/10.1007/s00221-005-2382-y
    CrossRefMedline
  38. ↵
    1. Rolfs M,
    2. Kliegl R,
    3. Engbert R
    (2008) Toward a model of microsaccade generation: the case of microsaccadic inhibition. J Vis 8:5–5. https://doi.org/10.1167/8.11.5
    Abstract/FREE Full Text
  39. ↵
    1. Salinas E,
    2. Steinberg BR,
    3. Sussman LA,
    4. Fry SM,
    5. Hauser CK,
    6. Anderson DD,
    7. Stanford TR
    (2019) Voluntary and involuntary contributions to perceptually guided saccadic choices resolved with millisecond precision. Elife 8:e46359. https://doi.org/10.7554/eLife.46359
    CrossRefMedline
  40. ↵
    1. Salinas E,
    2. Stanford TR
    (2018) Saccadic inhibition interrupts ongoing oculomotor activity to enable the rapid deployment of alternate movement plans. Sci Rep 8:14163. https://doi.org/10.1038/s41598-018-32224-5
    CrossRefMedline
  41. ↵
    1. Scholes C,
    2. McGraw PV,
    3. Nyström M,
    4. Roach NW
    (2015) Fixational eye movements predict visual sensitivity. Proc Biol Sci 282:20151568. https://doi.org/10.1098/rspb.2015.1568
    CrossRefMedline
  42. ↵
    1. Shinn M,
    2. Lee D,
    3. Murray JD,
    4. Seo H
    (2022) Transient neuronal suppression for exploitation of new sensory evidence. Nat Commun 13:23. https://doi.org/10.1038/s41467-021-27697-4
    CrossRefMedline
  43. ↵
    1. Grafman J
    1. Wessel JR
    (2025) Action stopping. In: Encyclopedia of the human brain (Second Edition) (Grafman J, ed), pp 184–205. Oxford: Elsevier. https://doi.org/10.1016/B978-0-12-820480-1.00045-0
  44. ↵
    1. Wessel JR,
    2. Aron AR
    (2017) On the globality of motor suppression: unexpected events and their influence on behavior and cognition. Neuron 93:259–280. https://doi.org/10.1016/j.neuron.2016.12.013
    CrossRefMedline
  45. ↵
    1. White AL,
    2. Moreland JC,
    3. Rolfs M
    (2022) Oculomotor freezing indicates conscious detection free of decision bias. J Neurophysiol 127:571–585. https://doi.org/10.1152/jn.00465.2021
    CrossRefMedline
  46. ↵
    1. White AL,
    2. Rolfs M
    (2016) Oculomotor inhibition covaries with conscious detection. J Neurophysiol 116:1507–1521. https://doi.org/10.1152/jn.00268.2016
    CrossRefMedline
  47. ↵
    1. Wijdenes LO,
    2. Brenner E,
    3. Smeets JBJ
    (2011) Fast and fine-tuned corrections when the target of a hand movement is displaced. Exp Brain Res 214:453–462. https://doi.org/10.1007/s00221-011-2843-4
    CrossRefMedline
  48. ↵
    1. Wolpert DM,
    2. Landy MS
    (2012) Motor control is decision-making. Curr Opin Neurobiol 22:996–1003. https://doi.org/10.1016/j.conb.2012.05.003
    CrossRefMedline
  49. ↵
    1. Wong AL,
    2. Haith AM,
    3. Krakauer JW
    (2015) Motor planning. Neurosci 21:385–398. https://doi.org/10.1177/1073858414541484
  50. ↵
    1. Xivry J-JO,
    2. Legrain V,
    3. Lefèvre P
    (2017) Overlap of movement planning and movement execution reduces reaction time. J Neurophysiol 117:117–122. https://doi.org/10.1152/jn.00728.2016
    CrossRefMedline

Navigate This Article

  1. Published online before print 17 February 2026, doi: 10.1523/JNEUROSCI.1319-25.2026 The Journal of Neuroscience, 46(11): e1319252026
  1. AbstractFree
  2. » Full Text
  3. Full Text (PDF)
  4. Supplemental Material

ORCID

  1. Profile for Rolfs, M.https://orcid.org/0000-0002-8214-8556
Responses to This Article
Be the first to respond to this article
Submit a Response
View Similar Articles
Similar articles in this journal»
Advertisement