Movement Enhances Tactile Sensitivity Through Prediction

Abstract

Tactile sensations decrease in felt intensity during movement, a phenomenon known as sensory gating. Competing theories attempt to explain why somatosensory perception is down-weighted during motion. We investigated how motor execution and prediction shape tactile sensations by precisely matching kinematics between active and passive movements. Perceived intensity was reduced in both conditions, but tactile sensitivity was enhanced rather than impaired during active movement compared with passive movement at matched kinematics. When passive movement velocity was predictable, sensitivity increased to the level of active movement, whereas temporal unpredictability impaired precision without affecting bias. These results show that predictions about imminent kinematics intrinsic to self-generated actions or learned during predictable passive motion sharpen tactile discrimination, while a separate movement-related process reduces perceived intensity. Our results reveal that movement does not globally suppress tactile processing but differentially affects bias and precision: perceived intensity decreases, while tactile discrimination improves when upcoming kinematics are predictable.

Introduction

It has long been known that tactile stimulation is perceived as less intense on effectors that are moving, a phenomenon commonly referred to as somatosensory gating (1). Somatosensory gating has been observed as a reduction in perceived tactile intensity(2, 3), changes in sensory discrimination(4, 5), and attenuated somatosensory-evoked responses(6–11). Here, we use the term tactile gating to denote the empirical phenomenon of movement-related modulation of tactile perception rather than to mean a single underlying neural mechanism. Multiple processes are thought to contribute to this phenomenon, including reafferent suppression(12), peripheral masking(13), and top-down modulation of sensory processing during movement(14, 15).

Somatosensory gating must be distinguished from sensory attenuation, which refers specifically to the reduction in perceived intensity of self-generated sensory consequences, such as self-touch or self-produced sounds(16–18). While sensory attenuation is typically attributed to predictive mechanisms linked to motor commands(19, 20), tactile gating encompasses movement-related changes in tactile perception that can also occur during passive motion(13, 21, 22).

For example, Kilteni and Ehrsson (23) proposed that predictive attenuation of self-generated touch primarily affects perceived intensity, while tactile gating during movement can modulate sensory precision. However, evidence across studies including earlier work from the same group and classic findings by Chapman(7) and colleagues demonstrates that both active and passive movements can influence the perceived intensity and sensory precision in a context-dependent manner. Moreover, tactile suppression can occur prior to movement onset and depends on expectations about forthcoming sensory input and task relevance, indicating anticipatory modulation of somatosensory processing (4, 11, 24-26).

Within predictive frameworks of sensorimotor control, the nervous system is thought to modulate not only the expected magnitude of sensory input but also its expected reliability(27). From this perspective, shifts in perceptual bias are consistent with a down-weighting of expected sensory input such as that arising from reafferent or predictive processes whereas changes in discrimination precision reflect modulation of the reliability or gain of sensory representations.

Despite extensive work on somatosensory gating, it remains unclear how movement-related suppression of tactile perception relates to predictive modulation of sensory precision. Many studies comparing active and passive movements report reduced tactile intensity in both conditions, suggesting that gating does not depend exclusively on motor commands (13, 21, 22). At the same time, recent work indicates that predictive processes can enhance perception of expected sensory outcomes rather than suppress them, particularly when predictions are precise and task-relevant (1, 28, 29).

In the present study, we addressed these issues by dissociating movement execution from prediction and by independently quantifying perceptual bias and sensory precision. In Experiment 1, we compared tactile perception during active and passive arm movements while closely matching movement trajectories and kinematics. In Experiment 2, we manipulated the predictability of passive movement dynamics and stimulus timing to test whether learned predictions about forthcoming kinematics are sufficient to sharpen tactile discrimination in the absence of voluntary action.

Based on this framework, we expected movement-related tactile modulation to differentially affect perceptual bias and sensory precision. Specifically, we hypothesized that perceived tactile intensity would be reduced during both active and passive movements relative to rest, reflecting movement-related suppression independent of agency. In contrast, we predicted that tactile discrimination precision would depend on the availability and reliability of predictive information, such that precision would be preserved during active movement and enhanced when passive movements were predictable, but impaired when predictions about upcoming kinematics or stimulus timing were unreliable.

Methods

Experiment 1

Participants

All experimental procedures were carried out in accordance with the ethical standards of the Declaration of Helsinki from 2024 and were approved by the local ethics committee of the Faculty of Mathematics and Natural Sciences of Heinrich Heine University, Duesseldorf.

Thirty-one right-handed participants (19–27 years, M = 21.97, SD = 2.20) were recruited. All participants provided written informed consent prior to participation. The study was approved by the local ethics committee of the Psychology Department of the Heinrich-Heine University Düsseldorf. Only right-handed individuals with no history of neurological or psychiatric disorders were included in the study.

Four participants were excluded due to unusable psychometric data (e.g. extremely noisy response patterns, indications of random guessing, or self-reported interference from the passive machine’s mechanical vibration). This resulted in a final sample of 27 participants (17 females, M = 22.04, SD = 2.31) for Experiment 1 analyses.

Software and Data Collection

The experiment was conducted on an Alienware Aurora R12 computer, running stimulus presentation and data collection software. Unity (C#) was used for real-time movement tracking, stimulus presentation, controlled vibration timing, and logging participant responses in the Active and Passive conditions. MATLAB (R2023a) was used for the stimulus presentation in the no-movement conditions (Self-Generated task and Control) and performed preliminary data processing. JASP (0.18.1.0) was used for statistical analysis, and the Arduino IDE was used to program the microcontrollers (Arduino Nano and Arduino Leonardo).

Apparatus and Materials

Participants were seated in front of an Asus ROG PG259QN monitor (1920 × 1080 resolution; 240 Hz refresh rate). Viewing distance varied depending on the condition: 90 cm for active movements and 110 cm for passive movements, due to the experimental setup. Soundcore Life Q30 noise-canceling headphones were used to eliminate auditory cues from the movement apparatus.

A vibrotactile actuator (diameter 1 cm) was attached to the median nerve area of the right forearm and controlled by an Arduino Nano, delivering vibrations each lasting 300 ms. In all tasks, participants were asked to discriminate between two vibrations. The first (reference) stimulus had a fixed frequency of 55 % of Duty Cycle, while the second (comparison) stimulus varied randomly across trials between 10 % and 90 % of Duty Cycle in 10 % of increments. Participants responded using a Pimoroni Keybow Mini (three-key keyboard) by pressing “1” if the second vibration felt weaker or “2” if it felt stronger. For ease of subsequent analysis, these responses were recorded so that the original “1” responses were labeled as 0 and the “2” responses as 1. See Figure 1A-C for task flow and apparatus. In all movement conditions, the first tactile stimulus was delivered during movement execution, whereas the second stimulus was delivered after movement completion.

Figure 1. Experimental set-up and task design.

(A) Schematic of the active task. Participants moved their right arm with a stylus from the first (left) to the second (right) sphere. A vibrotactile stimulus was delivered on the median nerve at one of four possible delays from movement onset (0, 100, 200, or 300 ms in the active condition, fixed at 200 ms in the passive condition). After reaching the second sphere, a second vibration was presented 200 ms later. Participants judged which vibration felt stronger by pressing a key (“1” = first stronger, “2” = second stronger) on the Pimoroni Keybow(5). (B) Active condition apparatus. The task display (1) guided the movement of a stylus tracked by the Touch X system (2). Vibrotactile stimulation was controlled by an Arduino Nano (7), with the actuator placed over the right median nerve (4). Responses were given via the Keybow. (C) Passive condition apparatus. The participant’s arm was moved by a rail system (3) driven by a stepper motor (Arduino Leonardo, 6). Passive movements varied in velocity (167, 200, or 250 mm/s). Vibrotactile stimuli were delivered 200 ms after movement onset and again 200 ms after reaching the target sphere. (D) Boxplots show movement latency, duration, and mean velocity for the active-movement condition across participants (N = 27). Dots represent individual participants; solid lines indicate kernel density distributions. (E) Top: example movement trajectory from a single participant showing one block of 300 repetitions. Bottom: group-average trajectory (black line) with shaded area representing ±1 SD across participants. Red circles mark the start and end positions of the stylus path.

Procedure Experiment 1

Participants completed three different experimental tasks in randomized order: Control Task, Active Movement Task, Passive Movement Task. The tasks were designed to investigate sensory bias and discrimination under varying conditions of movement initiation and predictability.

The Active and Passive conditions allowed comparison of self-initiated versus externally driven movement effects on sensory bias and discrimination precision. while the Control condition provided a baseline measurement of performance at rest.

In the Control task, participants remained still while holding the stylus. They received two sequential vibrotactile stimulations per trial, both delivered externally. The second vibration was presented 500 ms after the offset of the first, and participants judged which of the two vibrations felt stronger. This condition served as the baseline measurement of performance at rest and consisted of one block of 90 trials.

In the two movement tasks (Active and Passive), participants held a Touch X haptic stylus in their right hand while resting their elbow on an ergonomic mouse pad. The stylus’ movement was tracked in real time and displayed as a cursor on the screen. The task of the participant in both conditions was to reach a right sphere starting from a known position (Figure 1B-C).

In the Active Movement Task, participants actively moved the Touch X stylus between two target spheres displayed on the screen. Upon reaching the second sphere, they received two vibrotactile stimulations. The first vibration was delivered at 0, 100, 200, or 300 ms after movement onset (randomized), and the second vibration was delivered 200 ms after reaching the second sphere (Figure 1B). Participants completed four blocks of 100 trials each, with a 30-trial practice session preceding the first block.

In the Passive Movement Task, a custom-built rail system, driven by a stepper motor (controlled by an Arduino Leonardo), moved the participant’s arm passively along a 20 cm path, the movement path length was 20 cm in both movement conditions, at one of three predefined sustained speeds yielding velocities of 250 mm/s (0.8 s), 200 mm/s (1.0 s), and 167 mm/s (1.2 s) (Figure 1C). For each trial, the speed was randomly selected from these three options. The first vibration was delivered 200 ms after motion onset, and the second was administered 200 ms after movement completion. In total, participants completed 300 trials divided into 3 blocks. This fixed timing was chosen for the initial comparison in Experiment 1 to provide a consistent measurement point across all passive trials, in contrast to the Active condition where the predictable, self-generated nature of the movement allowed us to probe for temporal effects within the movement trajectory itself.

Experiment 2: Predictability of Movement Influences Sensory Discrimination

Experiment 2 was designed to directly test the hypothesis that the predictability of movement influences sensory discrimination precision. A separate sample of 20 participants was recruited and split into two groups for this purpose. The Fixed-Velocity Group (n = 10) experienced passive movements at a single, predictable velocity (medium speed, 200 mm/s) throughout the experiment. The Medium-Variability Group (n =10) with passive movements at two different velocities (slow 167 mm/s and medium 200 mm/s) randomized across trials. For comparison, the passive movement data from Experiment 1 (where three speeds were randomly interleaved: slow 167 mm/s, medium 200 mm/s, and fast 250 mm/s) served as the High-Variability group (n = 27), allowing for comparison across distinct levels of velocity predictability. The procedure otherwise replicated the passive movement task from Experiment 1 (first vibration 200 ms after onset, second 200 ms after offset), see figure 2A.

Figure 2. Predictability modulation during passive movements.

(A) Velocity Modulation. Schematic of the three levels of velocity predictability used to manipulate kinematic regularity. Low predictability (blue) included three possible velocities (167, 200, and 250 mm/s; 33% probability each), Medium predictability (green) included two velocities (167 and 200 mm/s; 50% probability each), and High predictability (orange) used a single constant velocity (200 mm/s; 100% probability). The right panel illustrates the corresponding movement trajectories, with vibrotactile stimuli (gray icons) delivered 200 ms after movement onset. (B) Temporal Modulation. Schematic of the manipulation of stimulus timing during passive movement. Low temporal predictability (blue) included three possible onset times (200, 400, or 600 ms after movement onset; 33% probability each), whereas High temporal predictability (orange) used a fixed onset time of 200 ms (100% probability). The right panel illustrates the timing of the tactile stimuli along the movement trajectory, highlighting temporal variability across conditions.

To disentangle the effects of kinematic predictability (i.e., movement dynamics) from temporal predictability (i.e., stimulus timing), we added a Temporal Unpredictability Group (n = 10) as a follow-up control. This group experienced passive movements at the same single, predictable velocity as the Fixed-Velocity group (200 mm/s), ensuring movement kinematics were perfectly predictable. However, for this group, the timing of the first vibrotactile stimulus was unpredictable, occurring randomly at one of three timestamps after movement onset: 200 ms, 400 ms, or 600 ms. The second vibration occurred, as in other conditions, 200 ms after movement completion (Figure 2B). This design isolates the effect of temporal uncertainty while holding kinematic predictability constant. The performance of this group was compared via an independent samples t-test to the Fixed-Velocity Group, which can be considered the fully predictable (kinematically and temporally) passive condition.

Trial Exclusion Criteria

In the Active condition, each trial included measurements of latency (the interval from trial onset or movement cue to actual movement initiation) and movement duration (the interval between movement onset and offset). The mean latency across participants was 517 ms (SD = 115 ms), the mean movement duration was 633 ms (SD = 139 ms), and mean velocity 419.94mm/s (SD = 115 mm/s) is shown in Figure 1D. We excluded all trajectories whose latency, duration, or spatial path fell outside ±2 standard deviations of each participant’s mean, ensuring that only consistent, straight movements were analyzed (see Figure 1E). After applying these criteria, 90.89 % of all trials were retained (929 of 10,200 trials excluded). Moreover, if participants initiated movement before the second sphere appeared, the trial was excluded. The remaining data were used for all subsequent psychometric and statistical analyses.

Data Analysis

Psychometric functions (cumulative Gaussians) were fitted to each participant’s response data for each condition to derive the Point of Subjective Equality (PSE) and the Just Noticeable Difference (JND). The PSE represents the comparison frequency judged equal in intensity to the 55% Duty Cycle reference and serves as a measure of sensory bias, while the JND, defined as the difference in comparison frequency between 50% and 75% “stronger” responses, provides a measure of discrimination precision. (Figure 3A–C). Each participant’s responses were plotted as a function of comparison stimulus intensity and fitted with a cumulative Gaussian distribution; the median of this function corresponds to the PSE, and its standard deviation corresponds to the JND. Repeated-measures ANOVAs examined the effects of Condition (Active, Passive, Control) in Experiment 1. Further ANOVAs tested the effects of Time (0, 100, 200, 300 ms) within the Active condition, Speed (Slow, Medium, Fast) within the Passive condition of Experiment 1, and Velocity Group (Fixed, Medium-Variability, High-Variability) in Experiment 2. Post hoc Bonferroni-corrected t-tests were used to explore significant main effects. Effect sizes (Cohen’s d for t-tests, partial eta squared η² for ANOVAs) were calculated. Because the contrasts of interest were defined a priori, we additionally verified that all reported effects remained unchanged when post-hoc tests were evaluated without Bonferroni correction Additional control analyses revealed that neither the timing of tactile stimulation within the movement nor movement velocity influenced PSEs or JNDs in the Active or Passive conditions (all ps >.26; see Supplementary Figures S1–S3).

Figure 3. Tactile sensitivity during active and passive movements:

(A–C) Psychometric functions from representative participants illustrating performance across baseline, passive, and active movement conditions. Curves plot the proportion of trials in which the second vibration was judged stronger as a function of comparison stimulus intensity (second stimulus, duty cycle %). (D–E) Group data for the Point of Subjective Equality (PSE; lower values indicate reduced perceived intensity, D) and Just-Noticeable Difference (JND; lower values indicate higher tactile precision, E). Each dot represents an individual participant; colored circles indicate group means ± SEM (red = active, green = passive). The shaded gray horizontal band represents the mean ± SEM of the baseline condition. Both active and passive movements produced reduced perceived intensity (lower PSE) relative to baseline, but discrimination precision (JND) was impaired only during passive movement. (F–G) Scatterplots of PSE (F) and JND (G) comparing active and passive conditions across participants. Each point represents one participant; the dashed unity line indicates equal values across conditions. PSE values clustered around the unity line, showing comparable intensity bias, whereas JNDs were consistently higher during passive movement, reflecting reduced tactile precision when motion was externally generated.

Results

Experiment 1

Participants actively moved a stylus whose position was continuously tracked by the Touch X device. In sessions with passive movements, participants held the stylus, which was then transported by a motor to the target location (Figure 1). At various times across active and passive movement execution, participants received a tactile vibration, whose intensity they had to judge against a reference vibration being presented after movement execution (for an example of psychometric functions, see Figure 3 A-C). A repeated-measures ANOVA revealed a significant main effect of Condition (F (2,52) =32.915, p<.001, η²=0.559; Figure 3D Post hoc tests revealed a significant underestimation of tactile intensity in both Active and Passive conditions compared to baseline, Active Movement (ΔMean=1.074, t(26)=8.391, p₍bonf₎<.001, d=1.744; Figure 3C) and Passive Movement (ΔMean=1.094, t(26)=7.478, p₍bonf₎<.001, d=1.777; Figure 3A) compared with Control at rest, indicating a bias towards reduced intensity during limb motion. No significant PSE difference was found between Active and Passive movement (ΔMean=0.020, t (26) =0.183, p₍bonf₎=1.000, d=0.033; see also Figure 3F).

To assess discrimination precision, we next examined just-noticeable differences across the three movement conditions, providing a measure of tactile sensitivity independent of perceptual bias, note that lower PSE values reflect reduced perceived intensity, whereas lower JND values reflect improved discrimination precision. A repeated-measures ANOVA on the JNDs revealed a significant main effect of Condition (F (2,52) =27.506, p<.001, η²=0.514; Figure 3E). Post hoc tests (Bonferroni-adjusted) showed impaired precision only in Passive Movement: Passive > Control (ΔMean=– 0.651, t (26) =6.978, p₍bonf₎<.001, d=–1.214) and Passive > Active (ΔMean=–0.620, t (26) =–5.413, p₍bonf₎<.001, d=–1.156). No difference was found between Control and Active (p₍bonf₎=1.000). Individual Active–Passive pairs are shown in Figure 3G.

To assess whether the precise moment of stimulation during active movement affected perception, we analyzed PSEs and JNDs across four onset times (0/100/200/300 ms Figure 1B, right). A repeated measures ANOVA revealed no effect of timestamp on PSEs (F (3,78) =0.495, p=.687) or JNDs (F (3,78) =1.341, p=.267).

To examine whether velocity influenced perception during passive movement, we compared PSEs and JNDs across three velocities (167/200/250 mm/s; Figure 1C, right). Repeated measures ANOVA found no main effect of velocity on PSEs (F (2,52) =2.466, p=.95) nor JNDs (F (2,52) =0.1.191, p=.319).

Experiment 2

In experiment 2 we asked whether predictability changes precision (Figure 2). Kinematic predictability was varied by how many passive speeds could occur: three, two or one constant speed (Figure 2A). To separate timing from kinematics, a Temporal-Unpredictability group kept speed fixed but randomized the first-stimulus onset (200/400/600 ms) (Figure 2B). Analysis of PSEs across the three groups with varying velocity predictability showed no significant difference (F (2,42) = 0.012, p =.988 Figure 4A, left). However, JND analysis revealed a significant effect of predictability (F (2,42) = 5.512, p =.007, η² = 0.208 Figure 4A, right). Post-hoc tests showed that JNDs were significantly smaller in the Fixed-Velocity condition (M = 1.203, SD = 0.480) than in the High-Variability condition (M = 1.979, SD = 0.600; t = –3.306, p₍bonf₎ =.006, d = –1.273), indicating improved tactile discrimination under predictable movement dynamics.

Figure 4. Predictability modulates tactile perception.

(A) Movement Predictability. Group mean Point of Subjective Equality (PSE; lower values indicate reduced perceived intensity, left) and Just-Noticeable Difference (JND; lower values indicate higher tactile precision, right) across three levels of velocity predictability (Low = blue, Medium = green, High = red). Error bars represent ± SEM. Dashed red lines indicate baseline mean values. PSEs did not vary with predictability, whereas JNDs decreased significantly as movement became more predictable (p < 0.01), indicating enhanced tactile discrimination under stable kinematics. (B) Spatiotemporal Predictability. Mean PSE (left) and JND (right) for passive movements with predictable (High, red) versus unpredictable (Low, blue) stimulus timing. JNDs were significantly lower under high temporal predictability (p < 0.05), showing that consistent stimulus timing improves tactile precision, while PSEs remained unaffected. Dots represent individual participants.

Asterisks indicate significance levels: p < 0.05 (*), p < 0.01 (**).

To determine if movement predictability could be explained by temporal predictability of stimulus occurrence, we compared the Fixed-Velocity group with the Temporal Unpredictability group (schematic: Figure 2B). In the latter, movement velocity was predictable, but stimulus timing was not. JNDs in the Temporal Unpredictability group were significantly higher than in the Fixed-Velocity group (time predictable) (t (17) = 1.96, p =.033, d = 0.90, Figure 4B, right). JNDs were larger in the unpredictable condition, indicating reduced tactile precision. PSE remained unaffected (Figure 4B, left).

These results demonstrate that the predictability of both movement dynamics and stimulus timing critically shape tactile discrimination. When participants could anticipate the upcoming velocity, their perceptual precision improved, indicating that learned predictions can substitute for internally generated motor signals in refining tactile sensitivity.

Discussion

Our findings refine classical views of somatosensory gating. While perceived tactile intensity decreased during both active and passive movements, tactile discrimination precision depended on the availability and reliability of predictive information. Specifically, precision was preserved during active movement and enhanced when passive movement dynamics were predictable. Because perceived intensity was comparable across active and passive conditions in the present study, movement-related reductions in tactile intensity did not appear to depend on voluntary motor commands. Previous studies have reported mixed findings in this regard, with some suggesting stronger suppression associated with motor commands (Chapman & Beauchamp, 2006), whereas others report comparable effects between active and passive movements (e.g., Yildiz et al., 2015; Juravle et al., 2017).Together, these results suggest that movement-related modulation of touch reflects the interaction of multiple mechanisms rather than a unitary suppressive process.

Previous accounts have suggested that reductions in tactile sensitivity are a necessary consequence of decreasing the precision of sensory evidence during movement (1) and that this suppression protects the fidelity of preparatory neural activity in motor circuits(2). Contrary to this idea, we observed that tactile discrimination precision was not reduced during active movement but was instead higher than during unpredictable passive movement. This dissociation indicates that somatosensory gating cannot be explained solely by a global down-weighting of somatosensory input. Rather, predictive information about forthcoming movement kinematics can selectively enhance sensory precision even when perceived intensity is concurrently reduced. We speculate that movement introduces uncertainty in the interpretation of tactile intensity for stimulation of the respective limbs. Purely mechanical factors likely constrain tactile receptivity during movements, for instance through backward masking. However, predictions of movement trajectories might counteract the effects of the mechanical factors on tactile sensations.

Experiment 2 identified the conditions under which high tactile sensitivity emerges: when the velocity of passive motion is predictable. These findings indicate that predictability of movement dynamics, rather than movement execution per se, governs tactile sensitivity (3). During active movements, such information likely arises from motor-based predictions; during passive movements, it can emerge through learning stable external dynamics. Anticipating the velocity profile of motion may facilitate more precise temporal and spatial allocation of sensory processing, sharpening tactile discrimination.

In contrast, the consistent reduction in perceived intensity across conditions replicates classical gating effects (4, 5). Peripheral and reafferent mechanisms probably contribute to this bias by reducing the responsiveness of somatosensory pathways during motion. Arikan et al. (30) similarly reported that suppression persists even when passive movements are predictable, supporting the idea that both central and peripheral processes shape gating. Crucially, the present findings show that enhanced sensory precision can arise independently of such reduction, indicating that predictive sharpening of tactile discrimination does not depend exclusively on efference-based motor predictions but can also derive from predictions about movement dynamics that are based on previous experience. This interpretation aligns with active-inference accounts, which propose that predicted sensory precision is dynamically modulated during movement.

These results help reconcile seemingly divergent findings in the literature. Classical studies of somatosensory gating demonstrated reduced tactile sensitivity during both active and passive movements (13, 21, 22), whereas more recent work has shown facilitation or sharpening of expected sensory outcomes during action (28, 29). Our data show that movement can simultaneously reduce the perceived intensity of tactile sensations while enhancing sensory precision when predictions about forthcoming kinematics are reliable. This dissociation helps explain why some studies emphasize suppression of tactile signals, whereas others report perceptual facilitation, depending on whether perceptual bias or discrimination precision is measured and on the predictability of sensory consequences.

The present findings also complement recent work showing that predicted tactile outcomes during active movement can enhance perceived intensity(31). Whereas those studies manipulated the predictability of the tactile stimulus itself, our manipulation targeted the predictability of the moving effector and its kinematics. This distinction may explain why predictability influenced perceptual bias in previous work but primarily modulated sensory precision in the present study. Together, these results suggest that different sources of prediction stimulus-based versus kinematic can exert dissociable effects on tactile bias and precision.

Crucially, predictive modulation of tactile precision need not depend exclusively on self-generated motor commands. Although efference-based predictions are available during active movement, similar predictive signals can be acquired through learning the regularities of externally generated motion. Passive movements with stable and predictable kinematics may therefore support accurate predictions of upcoming sensory input and enhance tactile discrimination, even in the absence of voluntary action (1, 28, 29)

From a neurophysiological perspective, a reduction of perceived intensity alongside enhanced discrimination under predictable kinematics may reflect distinct mechanisms within the somatosensory system. The reduction in perceived intensity likely arises from reafferent suppression at early sensory relays such as the dorsal horn or cuneate nucleus, where inhibitory interneurons gate cutaneous input during motion. In contrast, improved precision may depend on top-down modulation of cortical gain within primary somatosensory areas (S1). Preparatory activity in motor and premotor regions (2, 10)could convey predictive signals that fine-tune S1 representations, consistent with active-inference proposals of descending proprioceptive predictions (7). Moreover, the finding that learned predictability in passive movement elicits comparable sharpening suggests that cerebellar–thalamocortical circuits may acquire predictive control over tactile precision even in the absence of efference copy.

Although our design precisely matched kinematics and minimized attentional differences between active and passive movements, we cannot entirely exclude contributions from other movement-related factors such as proprioceptive feedback or stabilization effort. Nevertheless, the selective improvement of discrimination under predictable passive motion supports a central role for predictive processes whether motor-based or learned in refining tactile sensitivity.

In summary, movement-related tactile modulation reflects a dissociation between perceptual bias and sensory precision. A general reduction of the perceived intensity of tactile stimulation accompanies motion, likely reflecting peripheral and central reafferent suppression, while predictive information about forthcoming kinematics selectively enhances sensory precision. By dissociating these components and matching kinematics across active and passive movements, the present study demonstrates that predictive information plays a critical role in shaping tactile perception during movement, revealing that tactile gating involves not only the suppression of redundant input but also the sharpening of expected sensory information.

New & Noteworthy.

This study refines the classical view of somatosensory gating by dissociating bias and precision in tactile perception during movement. We show that while perceived intensity decreases in both active and passive motion, tactile discrimination improves when upcoming kinematics are predictable. These findings reveal that tactile gating involves not only reafferent suppression but also predictive sharpening of sensory precision.

Figure 5.

Data availability

The data is available in: https://doi.org/10.5281/zenodo.17414491

Supplementary analysis can be found: https://doi.org/10.5281/zenodo.18608119