Static masks and saccadic velocity profile jointly reduce perceived motion: Evidence from simulated saccades

This study demonstrates that visual mechanisms alone, specifically static masking and the naturalistic temporal structure of saccadic velocity profiles, significantly reduce perceived motion during simulated eye movements, thereby supporting perceptual continuity across saccades.

The Gist

The Big Mystery: Why Don't We See the World Blur When We Blink?

Imagine you are driving a car at 100 mph. If you suddenly swerve your head to the left, the view out your window should look like a chaotic, blurry smear, right? Yet, when you move your eyes quickly (a movement called a saccade) to look at something new, the world doesn't blur. It feels like a seamless, high-definition movie where the camera just "cuts" to the next scene.

Scientists have long wondered: How does our brain hide this blur?

For a long time, we thought the brain had a "mute button" (called saccadic suppression) that turned off our vision the moment our eyes moved. But this new study suggests the answer is more like a magic trick involving two specific visual effects: Static Masks and the Shape of the Movement.

---

The Experiment: Simulating a "Fake" Eye Movement

The researchers didn't ask people to actually move their eyes. Instead, they kept the participants' eyes perfectly still and projected a moving picture onto a screen.

• The Scene: Imagine a wall covered in static TV snow (pink noise).
• The Action: Suddenly, the entire wall of snow slides to the left or right very quickly, then stops.
• The Trick: Sometimes, they flashed a "freeze-frame" of the snow before and after the slide. This is called a mask.

They wanted to see if these freeze-frames (masks) could trick the brain into thinking the movement was smaller or slower than it actually was, even without the eyes moving.

The Two "Magic Tricks" Discovered

The study found that the brain uses two distinct tricks to hide motion:

1. The "Flashbulb" Effect (Static Masks)

Think of this like a photographer taking a picture. If you take a photo of a moving car, but you also flash a bright light right before and right after the car moves, the movement gets lost in the glare.

• What they found: When the researchers flashed the "before" and "after" images (the masks), the participants' brains drastically underestimated how far the image moved.
• The Speed: This happens incredibly fast—within about 15 milliseconds. That's faster than a camera shutter. It's like the brain gets "dazzled" by the static images, so the motion signal gets drowned out before it can be fully processed.

2. The "S-Curve" Effect (Velocity Profile)

This is the most surprising part. The researchers compared two types of movement:

• Type A (Constant Velocity): The image moves at a steady, robotic speed (like a train on a track).
• Type B (Natural Saccade): The image speeds up quickly, hits a top speed, and then slows down quickly (like a real eye movement).

The Result: Even without any flashbulbs or masks, the "Natural S-Curve" movement looked smaller and slower to the participants than the "Robot Train" movement, even though they covered the exact same distance in the exact same time.

The Analogy: Imagine two runners running a 100-meter dash.

• Runner A starts at full speed instantly and stops instantly.
• Runner B accelerates gently, sprints, and then brakes gently.
• Even if they finish at the same time, our brain perceives Runner B as having run a shorter distance. The "start and stop" nature of the movement confuses the brain's motion detectors, making the movement feel less significant.

The "Blind Spot" in Our Confidence

The researchers also asked the participants: "How sure are you about what you saw?"

You might expect that if the image was blurry or masked, people would say, "I'm not sure." But that's not what happened.

• The Analogy: Imagine you are trying to guess the number of jellybeans in a jar that is covered by a foggy glass. You guess 500. The real number is 100. You are wildly wrong, but you feel 100% confident in your guess.
• The Finding: People were confident in their wrong answers. Their brains didn't realize the motion signal had been "deleted" by the masks. The brain just filled in the gap with a guess, and the person felt certain about it. This suggests the "hiding" happens so early in the visual system that the brain never even knows it missed the signal.

Why Does This Matter?

This study changes how we understand our vision. It tells us that:

1. We don't need a "mute button": We don't necessarily need the brain to turn off our vision to stop the blur. The visual system is smart enough to use the "before" and "after" pictures to mask the motion automatically.
2. Movement shape matters: The way an object moves (accelerating and decelerating) is a built-in feature that helps hide motion. Our eyes are designed to move in a specific "S-curve" pattern, and our brain is tuned to ignore that specific pattern to keep our world feeling stable.
3. We are easily fooled: Our brain constructs a stable reality by stitching together fragments of information, often ignoring the messy, blurry parts in between.

The Bottom Line

The reason the world doesn't look like a blurry mess when you look around is a combination of visual static (the images before and after the move) and the specific rhythm of your eye movement. Your brain is a master editor, cutting out the "blurry scenes" so seamlessly that you never even notice the cut happened.

Technical Summary

1. Problem Statement

Saccadic eye movements displace the retinal image at high speeds (hundreds of degrees per second), yet observers rarely perceive this motion or the resulting scene displacement, a phenomenon known as saccadic omission. While saccadic suppression (extra-retinal signals reducing visual sensitivity) is a well-known mechanism, it is insufficient to fully explain omission, as intra-saccadic motion can be perceived if the post-saccadic image is absent or if the stimulus is stabilized.

Previous research suggested visual masking by pre- and post-saccadic scenes plays a critical role. However, two key questions remained unresolved:

1. Amplitude Range: Previous studies on isolated visual masking were limited to a single saccade amplitude (6 dva), failing to address how masking operates across the full range of natural saccade sizes (6–18 dva).
2. Kinematic Contribution: It was unclear whether the specific velocity profile of a saccade (characteristic acceleration and deceleration) contributes to motion reduction independently of the static flanking masks, or if the saccade is treated merely as a uniform motion event.

2. Methodology

The authors employed simulated saccades presented to fixating observers to isolate visual factors from extra-retinal motor signals.

• Stimuli: Full-field, repetitive 1/f pink noise patterns (RMS contrast 0.1664) spanning the screen. The patterns were horizontally periodic to eliminate salient spatial endpoints, ensuring motion perception was driven by global displacement rather than feature tracking.
• Experimental Design: Two experiments manipulated:
  • Amplitude: 6, 12, and 18 degrees of visual angle (dva).
  • Duration: Varied to match natural saccade main sequences (Experiment 1) or crossed orthogonally (Experiment 2).
  • Velocity Profile: Two types were compared:
    1. Saccadic Profile: Derived from averaged real saccade recordings (rapid acceleration/deceleration).
    2. Constant Profile: Uniform velocity matched in amplitude and duration.
  • Masking: Forward and backward static masks of equal duration (0, 20, 40, 80, 320 ms) were applied.
• Participants:
  • Experiment 1: 10 participants (Amplitude and Velocity reports collected in separate sessions).
  • Experiment 2: 11 participants (Amplitude reports only; independent manipulation of amplitude and duration).
• Procedure: Participants reported perceived motion direction and magnitude (amplitude via arrow adjustment; velocity via moving reference stimulus) and confidence ratings (Exp 1 only). Eye movements were monitored to exclude trials with actual saccades.
• Analysis: Bayesian multilevel nonlinear regression was used to model perceived amplitude/velocity as an exponential decay function of masking duration. Key parameters included:
  • $N_0$: Perceived magnitude without masking.
  • $\lambda$ (Halftime): Rate of decay (time to reach 50% suppression).
  • Asymptote: Perceptual floor under full masking.

3. Key Contributions

• Quantification of Visual Masking: Demonstrated that purely visual transients (static masks) are sufficient to reduce perceived motion amplitude and velocity by ~50% within ~15 ms across a wide range of saccade sizes, without any eye movement.
• Isolation of Velocity Profile Effects: First to show that the saccadic velocity profile itself acts as a "self-masking" signal, reducing perceived motion even in the absence of external static masks.
• Dissociation of Amplitude and Duration: Showed that while amplitude and duration covary in natural saccades, they have opposing effects on masking dynamics, which cancel out along the natural "main sequence."
• Metacognitive Analysis: Investigated whether observers are aware of masking-induced degradation via confidence ratings, finding no reliable correlation between confidence and accuracy.

4. Key Results

A. Visual Masking Dynamics

• Rapid Suppression: Perceived motion decayed exponentially with mask duration. The halftime (time to 50% suppression) was remarkably consistent across conditions, averaging ~15 ms (14.3–16.9 ms).
• Invariance: This rapid masking rate was largely invariant across saccade amplitudes (6–18 dva) and response modalities (amplitude vs. velocity reports), suggesting a shared early motion representation is suppressed before spatial and temporal estimates diverge.
• Residual Perception: Even with long masks (320 ms), a small residual motion percept remained (asymptote), which increased slightly with stimulus duration but was largely independent of amplitude.

B. The Saccadic Velocity Profile Effect

• Self-Masking: Motion following a naturalistic saccadic velocity profile was perceived as smaller and slower than constant-velocity motion of the same amplitude and duration, even with 0 ms external masking.
• Scaling: This additional attenuation increased with saccade amplitude. For velocity reports, the saccadic profile reduced perceived speed by ~10 dva/s at 6 dva, ~34 dva/s at 12 dva, and ~39 dva/s at 18 dva compared to constant motion.
• Independence: The profile effect persisted even under full masking (reduced the perceptual floor), suggesting that kinematic structure and external masking operate via partially separable mechanisms.

C. Metacognition

• Confidence vs. Accuracy: Confidence ratings did not track masking-induced motion reduction.
  • For amplitude reports, confidence slightly increased with masking duration.
  • For velocity reports, confidence slightly decreased but remained high despite large errors.
  • Correlations between confidence and response error were near zero or weakly negative.
• Conclusion: Observers lack metacognitive access to the degradation of motion signals caused by masking, implying the suppression occurs at a pre-perceptual or early sensory stage.

5. Significance and Implications

• Mechanism of Saccadic Omission: The study provides strong evidence that visual masking alone can account for a substantial portion of saccadic omission. The visual system does not require extra-retinal motor commands to suppress motion; the temporal structure of the retinal image shift (acceleration/deceleration) combined with pre/post-saccadic transients is sufficient.
• Role of Kinematics: The saccadic velocity profile is not just a motor output but a visual stimulus property that actively contributes to motion attenuation. The gradual acceleration/deceleration may reduce the "suddenness" of the motion signal or generate competing motion signals (acceleration/deceleration phases) that interfere with coherent motion perception.
• Temporal Dynamics: The ~15 ms halftime suggests that motion masking operates on the order of tens of milliseconds, consistent with early sensory processing (e.g., motion-energy models) rather than higher-level cognitive processes.
• Perceptual Continuity: These findings suggest that the visual system constructs perceptual continuity across eye movements through a combination of external masking (pre/post-saccadic scenes) and internal "self-masking" driven by the specific kinematic signature of the eye movement itself.

In summary, the paper establishes that the visual system's inability to perceive saccadic motion is largely a result of rapid, early visual masking mechanisms that are sensitive to both the static context and the specific kinematic profile of the motion itself.