The Aubert-Fleischl Paradox: Why Eye Tracking Makes the World Move Slow

Hold your thumb out at arm’s length, point your index finger on your other hand, and swipe your finger past your face from left to right. First, keep your eyes locked on your stationary thumb while watching your finger zip past in your peripheral vision. Now swipe it again at the exact same physical speed, but this time smoothly trace your index finger with your eyes as it moves.

To your visual cortex, the second pass looks noticeably slower.

The motion hasn't changed, but your brain's accounting system has. When your eyes remain locked on a fixed spot, a moving object sweeps directly across your retina, triggering a dense stream of raw visual signals that your brain interprets as rapid motion. But when you track the object with a smooth eye movement, its image stays nearly stationary on your retina. To figure out how fast the target is actually traveling, your visual brain has to estimate how fast your muscles are turning your eyeballs and add that extraretinal calculation back into the equation.

That math is systematically flawed.

Physiologist Hermann Aubert first published this discrepancy in 1886 alongside Ernst von Fleischl-Marxow, giving rise to what vision scientists call the Aubert-Fleischl phenomenon. When an observer actively tracks a target using smooth pursuit eye movements, they perceive its velocity as significantly slower compared to when they keep their gaze fixed on a stationary point and let the object cross their visual field.

Measured.

For over a century, neuroscientists worked to uncover why two pathways measuring the exact same physical event reach different conclusions. The human visual system relies on two distinct sources of speed information. The first is afferent motion—the direct optical flow of light sweeping across photoreceptors on the retina. The second is efferent motion, an internal calculation driven by motor commands sent to the extraocular muscles.

When your eyes are held still, the afferent signal does all the heavy lifting. Neurons in the primary visual cortex and middle temporal area fire in direct proportion to the speed at which contrast borders glide across the retinal grid. But when you pursue an object with your eyes, the target’s image rests almost motionless on the fovea—the central region of the retina. The retinal motion signal drops near zero.

To compensate, the visual cortex relies on extraretinal signals: an internal copy of the motor command sent to the eye muscles, known as a corollary discharge, alongside proprioceptive feedback from muscle spindles.

The glitch lies in how the brain weights these two inputs. Work on visual speed perception—such as Bayesian observer models developed by researchers like Stockerguer and colleagues—reveals that human perception treats sensory measurements as noisy estimates. When signals become uncertain, the brain pulls its final perceptual judgment toward a prior expectation—specifically, a baseline assumption that objects in the physical world are generally stationary or moving slowly.

Extraretinal motor signals carry significantly more statistical noise than direct, high-resolution retinal stimulation. Because the brain is far less certain about how fast the eyes are rotating than it is about light sweeping across retina cells, its Bayesian machinery leans more heavily on that internal prior for slowness. The brain essentially hedges its bets, discounting the speed of the motor tracking effort and registering a lower velocity overall.

The consequence is a persistent mismatch between sensory input and physical reality. When retinal motion drops and motor tracking takes over, the internal reconstruction underreports physical velocity.

Experiments examining this trade-off highlight just how fundamental the mechanism is across human senses. Researchers led by Alexander Schmidt and colleagues in 2019 demonstrated an exact tactile analogue of the Aubert-Fleischl phenomenon. When participants estimated the speed of a surface sliding under a stationary finger using touch alone, they perceived it as moving faster than when they actively tracked a moving surface with a guided hand movement—keeping the physical friction on the skin at zero. The same computational principle governs both visual and proprioceptive tracking: motor tracking signals underreport speed compared to passive surface displacement.

Every time you sweep your eyes across a room to follow a bird in flight, track a fastball headed over home plate, or trail a passing vehicle down the highway, your nervous system is actively subtracting extraretinal noise from your perception. You see the world slightly slower precisely because you chose to look directly at it.

Measured.

---

This article is AI-generated (synthetic) content, produced by an automated editorial system with human direction and review. Every claim is traced to published, peer-reviewed sources.