Marius Comper

Nature Communications · 30 August 2026 · Article in Press

Same image, a different eye jump

For curious readers and science communicators, choose a jump and a probe to see which kind of signal is favoured. A 30 August 2026 Nature Communications paper separates the measured human result from an explanatory diagram, not a clinical self-test.

The short answer

For the low probe, 6° versus 1° meant 37% higher sensitivity. For the high probe, the same comparison showed no difference.

Read the primary paper CC BY 4.0
The same image passes through three eye jumps An original diagram with one shared stimulus band and short, medium, and long arcs for 0.4, 1, and 6 degree jumps. The two lower bands show coarse and fine signal as conceptual weights. ONE IMAGE / THREE JUMPS original diagram, not an experimental figure shared stimulus 0.4° 1° 6° coarse · 0.1 cycles/degree fine · 2.5 cycles/degree
What you are seeing One stimulus band is drawn across three paths. Path length is the variable; coarse and fine are conceptual weights, not participant data.

01 · The result

The jump also selects scale

A saccade does more than move the fovea to a new place. It changes how a static image becomes a luminance signal over time. That is the idea tested here.

c/° means cycles per visual degree. The 0.5 c/° reference stays fixed; the probe and saccade amplitude change.

02 · The operation

Switch the probe, keep the image

Conceptual model

Change only the variable you want to follow. The diagram calculates a schematic split between coarse and fine signal. The published result stays separate, directly below it.

1. Choose the eye jump
2. Choose the probe

Fixed reference: 0.5 c/°. It does not change when you switch the probe.

Weight in the schematic1° · coarse probe
0–100, conceptual index
Coarse signal68
Fine signal32

For the coarse probe, the schematic shifts weight toward coarse signal as the jump gets longer.

This is a parameterized conceptual model. It is not participant data, a screen measurement, or a calibrated reproduction.

What the published comparison says

For the coarse probe, the 6° versus 1° comparison found 37% higher sensitivity for 6°. That result comes from human measurement, not the schematic index above.

03 · The people in the study

What the participants did

The psychophysical experiments use 9 unique participants. The main 1°–6° comparison has N = 9 for Low-Probe, while the three-amplitude experiment has N = 6, including three new participants. Datasets used separately for natural eye-movement analysis are not added to form one experimental N.

9unique participants in psychophysical experiments
N = 9main 1°–6° comparison
N = 6three-amplitude experiment (3 new participants)
Low-Probe
0.1 c/° 6° vs 1°
N = 9 37% higher sensitivity for 6°
t(8) = 9.097 · p < 0.001 · d = 0.98
High-Probe
2.5 c/° 6° vs 1°
N = 8 No measured difference
t(7) = −0.066 · p = 0.949
High-Probe
2.5 c/° 0.4° vs 1° and 6°
group result · N = 6 0.4° better than both
p = 0.016
Low-Probe
0.1 c/° 0.4°, 1° and 6°
N = 6 6° > 1° and 0.4°
p = 0.001 and p = 0.010

Those p-values do not mean the same thing. The p < 0.002 result for the same direction across all six people is an individual bootstrap analysis, not the group result at p = 0.016.

04 · The mechanism

The signal changes over time

A static grating becomes a sequence of luminance changes when the eye moves across it. Longer, faster saccades shift energy through time differently from short saccades. The page compresses that idea into a bar without turning a simulation into a clinical observation.

Schematic temporal flow. It is not an eye trace and does not use a participant’s signal.

In the separate passive simulated-saccade analysis, the researchers used four observers. That component shows that luminance transients can preserve the direction of the effect, but it is separate from the instructed-saccade participant results.

05 · Edges

The result has precise boundaries

On average, 52% of trials were retained, so about 48% were discarded. The archived source reports this average, not a per-participant range. Those filters are part of the measurement and matter when reading differences between conditions.

01

The analysis uses a pre- and post-saccadic spectral window. It cannot perfectly separate the contributions of the two periods.

02

A browser cannot guarantee visual angle, viewing distance, 200 Hz timing, or calibrated contrast.

03

This page does not reproduce the gaze experiment and cannot tell you what you see. It is not a self-test and gives no medical advice.

Source status: the article is marked “Article in Press”. It is accepted and citable, but it is not yet the Version of Record. If the paper or supplements change, this page needs another check.

06 · Check

The sources stay visible

The primary paper supports the conclusion, status, and licence. Its supplements support the sample sizes, comparisons, analysis windows, and discarded-trial rate. The preprint remains earlier context, not a replacement for the published paper.

  1. 01
    Li, Y. H., Cox, M. A., Victor, J. D. et al. “What one sees depends on how far the eye has moved.” Nature Communications, 2026.Paper and Article in Press status ↗
  2. 02
    Supplementary Information, Reporting Summary, and Transparent Peer Review file.Supplementary file ↗ Reporting Summary ↗ Peer Review ↗
  3. 03
    bioRxiv preprint, 19 September 2025, DOI 10.1101/2025.09.17.676280.Preprint metadata ↗
  4. 04
    Licence for the published material: Creative Commons Attribution 4.0 International.CC BY 4.0 ↗

Reader-facing method: the comparisons and sample sizes are taken from the published edition and its supplements. The interactive index uses the formulas declared in the project data and is explanatory only.

↑