Tactile suppression during movement as optimal integration of somatosensory feedback across time

This study demonstrates that tactile suppression during movement is not a fixed gating mechanism but rather a consequence of optimal, dynamic state estimation where the nervous system continuously adjusts its reliance on internal predictions versus noisy sensory feedback based on the level of uncertainty.

The Gist

The Big Question: Why Don't We Feel Our Sleeves?

Imagine you are reaching across the room to grab a cup of coffee. As your arm moves, your sleeve slides against your skin. You probably don't even notice it, right? Yet, if someone else brushed that same spot on your arm, you'd feel it instantly.

Scientists have long known that our brains "turn down the volume" on touch sensations while we are moving. This is called tactile suppression. But how and why does the brain decide exactly when to turn the volume down and when to turn it back up?

For a long time, people thought the brain had a simple "gate" that just closed whenever we moved. But this new paper suggests it's much more sophisticated than a simple on/off switch.

The New Theory: The Brain as a Smart GPS

The authors propose that the brain isn't just ignoring touch; it's acting like a smart GPS navigation system that is constantly trying to figure out exactly where your hand is.

Here is the analogy:

• The Forward Model (The GPS Prediction): Your brain has an internal map. When you decide to move your hand, it predicts exactly where your hand should be a split second from now based on your muscle commands. It's like your GPS saying, "If you turn left here, you will be at the coffee shop in 10 seconds."
• The Sensory Feedback (The Real-Time Traffic): Your skin and muscles are constantly sending back noisy, messy data about what is actually happening. It's like the GPS getting a signal from a traffic camera that says, "Hey, there's a pothole!" or "The road is slippery."

The "Kalman Gain": The Volume Knob

The brain has to combine these two things: the prediction (where we think we are) and the reality (what our sensors feel).

The paper argues that the brain uses a mathematical rule (called Optimal Feedback Control) to decide how much to trust the prediction versus the reality. This decision-making process is like a volume knob for your sense of touch.

1. When the Prediction is Strong (Low Uncertainty):
   Imagine you are driving on a straight, empty highway you know perfectly well. Your GPS is 100% confident. You don't need to look at the road signs or listen to traffic reports constantly.

   • In the body: At the very start of a movement, your brain is very sure where your hand is going. So, it turns the "touch volume" way down. It ignores the sliding sleeve because it already knows exactly what to expect. Result: High Suppression.

2. When the Prediction is Weak (High Uncertainty):
   Now, imagine you are driving in a foggy, unfamiliar city. Your GPS is guessing. It's not sure if you are on the right street. Suddenly, you need to look at every sign and listen to every traffic report to correct your course.

   • In the body: As your hand moves, tiny errors (noise) build up. Your brain gets less sure about where your hand is. To fix this, it turns the "touch volume" back up. It needs to feel the sleeve sliding to correct its internal map. Result: Low Suppression (You feel more).

The Experiment: Testing the GPS

The researchers tested this by having people reach for a target on a screen while a tiny vibration was tapped on their forearm. They measured how well people could feel the vibration at different moments during the reach.

What they found:

• The Curve: Just as the theory predicted, people felt the vibration least right when the movement started (when the brain was most confident in its prediction). As the movement went on, the feeling of the vibration slowly came back (as the brain needed more sensory data to stay accurate).
• The Twist: They made the task harder by hiding the target until the very last second. This made the brain's initial "guess" very uncertain.
  • Result: When the brain was unsure at the start, it didn't turn the volume down as much. It kept the "touch volume" high immediately because it needed that sensory feedback to figure out where to go.

The Takeaway

This paper changes how we see the brain's "gating" mechanism. It's not a rigid wall that blocks all touch during movement. Instead, it's a dynamic, intelligent filter.

The brain is constantly asking itself: "How sure am I about where my hand is?"

• If I'm sure, I can ignore the noise (the sleeve sliding).
• If I'm unsure, I need to listen to the noise to correct my path.

In short: Your brain suppresses touch not because it's "busy" moving, but because it is doing a high-speed calculation to make sure you don't miss the coffee cup. It only ignores the feeling of your sleeve when it's confident it knows exactly where your arm is; the moment it gets confused, it turns the senses back on.

Technical Summary

1. Problem Statement

The central problem addressed is the mechanism behind tactile suppression (the reduction of sensitivity to touch) during voluntary movement. While it is well-established that the nervous system suppresses sensations from self-generated movement (e.g., not noticing a sleeve sliding over the arm while reaching), the computational principles governing the temporal dynamics of this suppression remain unclear.

Existing theories fall into two camps:

• Predictive Accounts: Suggest suppression arises from an internal forward model that predicts sensory consequences of motor commands (efference copy) and down-weights predicted inputs. However, these accounts often lack a quantitative explanation for why suppression varies dynamically based on task demands.
• Fixed Gating Accounts: Suggest suppression is a peripheral, fixed mechanism where strong movement signals mask weak tactile probes. This fails to explain why suppression is stronger in active vs. passive movements or why it modulates based on task uncertainty.

The authors propose that tactile suppression is not a fixed gate but a consequence of optimal state estimation, where the nervous system dynamically weights uncertain internal predictions against noisy somatosensory feedback to optimize movement control.

2. Methodology

Experimental Design

The study involved 31 human participants performing goal-directed reaching movements toward visual targets on a computer screen.

• Task: Participants reached from a fixed starting position to a target.
• Stimuli: Brief vibrotactile stimuli (250 Hz) were delivered to the dorsal forearm (targeting Pacinian corpuscles) at random time points ranging from ~250ms before movement onset to shortly after movement end.
• Measurement: Participants reported whether they detected the vibration. Sensitivity was quantified using signal detection theory ($d'$).
• Conditions:
  1. Certain Condition: Participants viewed the target location 1 second before the "Go" cue, allowing for precise initial state estimation.
  2. Uncertain Condition: The target appeared simultaneously with the "Go" cue, increasing the initial uncertainty regarding the hand's position relative to the target.
  3. Literature Comparison: The authors also re-analyzed data from a previous study (Colino et al.) where vision was removed during reaching to manipulate sensory feedback uncertainty.

Computational Modeling

The authors employed an Optimal Feedback Control (OFC) framework with signal-dependent noise to model the reaching movement.

• State Estimator: A Kalman filter was used to integrate a forward model prediction (based on motor commands/efference copy) with noisy sensory feedback.
• Key Variable: The Kalman Gain ($K_t$) represents the dynamic weight assigned to sensory feedback versus the internal prediction.
• Hypothesis: The model predicts that tactile suppression corresponds to the inverse of the Kalman gain. When the internal prediction is precise (low uncertainty), the system relies less on feedback (high suppression). When prediction uncertainty is high (e.g., due to signal-dependent noise or lack of visual info), the system relies more on feedback (low suppression).

3. Key Results

Temporal Dynamics of Suppression

• Behavioral Finding: Tactile sensitivity ($d'$) dropped significantly just before movement onset and remained suppressed during the early phase of the reach. Sensitivity began to recover after ~30% of the movement duration but did not fully return to pre-movement levels by the end.
• Model Alignment: The time course of the Kalman gain for relative hand position in the OFC model closely mirrored the behavioral sensitivity data. The model predicted high reliance on the forward model (low feedback gain/high suppression) at movement onset, transitioning to higher reliance on sensory feedback as movement uncertainty increased.

Manipulation of Initial Uncertainty

• Behavioral Finding: In the Uncertain Condition (where target position was unknown until movement onset), tactile suppression was significantly weaker during the initial phase of the movement compared to the Certain Condition.
• Interpretation: When the initial state estimate was uncertain, participants relied more heavily on somatosensory feedback to correct their trajectory, resulting in reduced suppression.
• Kinematics: Crucially, there were no significant differences in movement kinematics (reaction time, velocity, endpoint error) between the Certain and Uncertain conditions. This suggests the nervous system adjusted the weighting of sensory processing without altering the motor execution strategy.

Manipulation of Sensory Uncertainty (Vision)

• Re-analyzing data where vision was blocked during reaching, the authors found stronger suppression (lower sensitivity) compared to when vision was available.
• Model Explanation: Removing vision increases the uncertainty of the continuous sensory feedback. According to OFC, when sensory feedback is less reliable, the system should rely more on the forward model, leading to increased suppression. The model successfully predicted this shift.

4. Key Contributions

1. Normative Explanation of Suppression: The paper provides a quantitative, normative framework (OFC) explaining tactile suppression not as a fixed gate, but as a dynamic consequence of optimal state estimation.
2. Uncertainty-Dependent Weighting: It demonstrates that the nervous system continuously tracks and updates the uncertainty of its internal state estimates. Tactile suppression is modulated inversely to the reliability of the internal prediction:
   • High prediction certainty $\rightarrow$ Low feedback reliance $\rightarrow$ High suppression.
   • High prediction uncertainty $\rightarrow$ High feedback reliance $\rightarrow$ Low suppression.
3. Dissociation of Perception and Action: The study shows that the nervous system can adaptively modulate sensory processing (tactile sensitivity) based on task demands (uncertainty) without changing the kinematic output of the movement.
4. Unification of Literature: The OFC model successfully accounts for diverse findings, including the time course of suppression, the effects of active vs. passive movement, and the impact of visual occlusion on tactile sensitivity.

5. Significance

• Theoretical Impact: This work bridges the gap between predictive coding theories and optimal control theory. It moves beyond the qualitative idea of "prediction vs. sensation" to a quantitative model where the brain optimally balances noisy predictions and noisy feedback.
• Mechanism of Action: It suggests that the brain does not simply "turn off" sensory channels during movement. Instead, it dynamically tunes feedback gains to minimize error in state estimation. This implies that tactile suppression is a functional feature of motor control, allowing the system to ignore predictable noise while remaining sensitive to unexpected, task-relevant perturbations.
• Future Directions: The findings provide testable hypotheses for neurophysiological studies, suggesting that neural circuits (potentially involving the cerebellum and cortical areas) encode not just body state, but the uncertainty of that state, which drives the modulation of sensory gain.

In summary, the paper argues that tactile suppression is a sophisticated, adaptive mechanism driven by the brain's need to optimally estimate body state in the presence of uncertainty, rather than a simple, static filtering of sensory input.