Summary: Our eyes make rapid, flicking movements called saccades several times per second. Despite these frequent jumps, the world appears stable. A recent study used afterimages—the faint ghost images that linger after looking at a bright flash—to reveal how the brain preserves visual stability.
By tracking afterimages in total darkness, the researchers showed the brain relies on an internal copy of its own motor commands to predict how the scene should shift when the eyes move. These predictions are remarkably precise, but they consistently fall short by about 6%, revealing a small, systematic bias in how the brain estimates eye movement consequences.
Key Facts
- Saccade stability: Although our eyes jump several times a second, perception remains steady because the brain predicts the visual outcome of those jumps before new visual input arrives.
- The 94% rule: On average the brain’s internal estimate of an eye movement corresponds to roughly 94% of the actual displacement. This consistent shortfall is known as hypometria.
- Efference copy: The brain uses a feedforward copy of the motor command to the eye muscles—an efference copy—to update its internal spatial map ahead of incoming visual feedback.
- Predictive remapping: Afterimages remain fixed on the retina, but the brain predicts where objects should land on the retina after a saccade; when that prediction conflicts with the retinal signal, the afterimage appears to move in space.
- Saccadic adaptation: When eye movements change over time (for example, becoming shorter due to fatigue), the brain’s predictive estimate adapts accordingly, but the small undershoot remains.
Source: TUB
Our visual experience is not a continuous scan of the environment. Instead, our eyes jump in quick saccades many times per second, shifting the retinal image each time. If perception relied solely on retinal input, the world would appear unstable—but the brain compensates using internal predictions and other mechanisms to maintain a steady view.
A study titled “High-fidelity but hypometric spatial localization of afterimages across saccades,” published in Science Advances, used afterimages to measure how well the brain predicts the sensory consequences of eye movements. Conducted by Richard Schweitzer, Thomas Seel, Jörg Raisch, and Martin Rolfs at the Cluster of Excellence Science of Intelligence in Berlin, the work shows that predictions are both high-fidelity and systematically undershot.
Using afterimages to isolate internal signals
Afterimages, known since antiquity, offer a unique window into the brain’s internal computations. They stay fixed on the retina even as the eyes move. In normal vision, rich visual input helps the brain correct and confirm its estimates, but in complete darkness those external cues are absent. That makes afterimages a powerful tool to isolate the brain’s feedforward signals.
In the experiment, participants sat in total darkness and first fixated a bright flash to generate an afterimage. They then made eye movements to a second briefly visible light. Once the afterimage was clear, brief probe lights were flashed at defined positions. Participants reported whether the afterimage appeared left of, right of, or aligned with each probe. Eye tracking recorded the actual gaze direction, allowing direct comparison between perceived afterimage position and real eye movement.
High accuracy, but a small systematic undershoot
Results showed afterimages appeared to shift with the eyes: larger saccades produced larger perceived displacements. However, perception did not match the eye movement perfectly. On average, perceived afterimage shifts reached about 94% of the actual saccade amplitude. This consistent undershoot—hypometria—was present across participants, directions, and saccade sizes, indicating a systematic bias rather than random noise. The discrepancy is subtle and typically imperceptible in everyday life, but it reveals how the brain updates spatial representations after each saccade.
Prediction precedes visual feedback
The researchers tested whether post-saccadic visual feedback determines perceived afterimage location by manipulating the visibility and position of the saccade target after the eyes landed. These manipulations did not alter where participants saw the afterimage. The evidence supports the idea that an efference copy—an internal duplicate of the motor command—drives predictions about how the visual scene should shift, allowing perception to anticipate the result of the eye movement rather than waiting for corrected visual input.
Perception adapts when eye movements change
Eye movements themselves can adapt over time: repeated misses or muscle fatigue lead to saccadic adaptation, shortening or lengthening saccades until accuracy is restored. When the experimenters induced adaptation, participants’ perceived afterimage shifts adjusted alongside their altered saccades. Yet the characteristic ~6% undershoot persisted, showing the bias exists both for natural and adapted eye movements.
Why a small error makes sense
This residual mismatch may be an adaptive feature rather than a flaw. In natural behavior, saccades frequently fall slightly short of targets. If the brain’s internal estimate mirrors that habitual tendency, perception remains reliably synchronized with the motor system’s actual performance. Prioritizing consistent alignment with the body’s dynamics over perfect mathematical accuracy could be more useful for everyday vision.
What afterimages teach us about visual stability
Afterimages stay retinotopically fixed, yet they appear to move because the brain predicts where objects should land on the retina after each saccade—a process known as predictive remapping. When that prediction matches incoming visual signals, objects appear stable. But because an afterimage does not update with the scene, the brain infers it must have moved in space. The magnitude of that perceived movement therefore reflects the brain’s predicted visual change.
“Afterimages are a valuable probe for studying how the brain keeps the visual world stable by forecasting the sensory consequences of its own movements,” says Richard Schweitzer. Insights from this work extend beyond basic vision science and could inform applications in robotics, virtual reality, and clinical studies of eye-movement disorders, where linking motor commands to sensory outcomes is essential.
At a glance
• Human eyes make rapid saccades several times per second, yet perception remains stable.
• Afterimages provide a way to isolate internal, predictive signals used to track eye movements.
• The brain predicts the sensory consequences of saccades with high fidelity.
• Predicted afterimage movement consistently undershoots actual eye movements by about 6% (approximately 94% gain).
• This systematic undershoot likely reflects the brain’s alignment with habitual motor behavior rather than a processing flaw.
• The findings clarify how predictive mechanisms support visual stability despite constant eye motion.
Key Questions Answered:
A: The brain subtracts the predicted retinal shift produced by an eye movement to keep the world stable. An afterimage is fixed to the retina, so when the brain predicts the retinal motion, it infers the afterimage must be moving through space to remain in the same retinal position.
A: This modest undershoot mirrors the common tendency of saccades to fall slightly short of targets. The brain’s internal estimate therefore reflects typical motor performance, keeping perception consistently aligned with actual eye behavior.
A: Yes. Motion sickness often arises from mismatches between visual input and internal movement predictions. Understanding the brain’s typical predictive gain could help VR systems minimize those mismatches and create more comfortable, natural experiences.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- The journal paper was reviewed in full for accuracy.
- Additional context was added by editorial staff.
About this visual neuroscience research news
Author: Maria Ott
Source: TUB
Contact: Maria Ott – TUB
Image: The image is credited to Neuroscience News
Original Research: Open access. “High-fidelity but hypometric spatial localization of afterimages across saccades” by Richard Schweitzer, Thomas Seel, Jörg Raisch, and Martin Rolfs. Science Advances. DOI: 10.1126/sciadv.aeb0557
Abstract
Humans generally experience a continuous, stable visual world despite frequent retinotopic shifts caused by saccades. Afterimages demonstrate a parallel phenomenon: although fixed on the retina, they appear to move in egocentric space whenever the eye moves. To probe this, observers localized afterimages relative to briefly flashed probes in complete darkness while their eye movements were tracked. A computational model accurately predicted afterimage displacement based on saccade size. The gain of afterimage movement was reliably hypometric, unaffected by postsaccadic visual feedback and saccadic adaptation, and inversely related to saccade gain. Within a head-centered localization framework, afterimage motion is driven by efference-based, feedforward predictions of saccadic consequences, highlighting afterimages as a useful tool for studying perceptual stability.