Quiet eye can be measured automatically from the electrical signal of the eyes
1 study · 10 archers measured
The summary today
Only one group has tried, at the University of Mersin. Ten club archers, six arrows each, recorded with electro-oculography, the electrical potential measured around the eye; an expert marked the start and end of the quiet eye by hand in the sixty stretches of signal, and five neural networks learned to reproduce those annotations. The two with temporal memory reach an accuracy of about 93 and 92 per cent, against 51 to 68 for a traditional classifier, and in the best conditions they miss the onset by 7 to 9 milliseconds and the end by 4 to 6. The result depends on cleaning the signal, though: on the raw trace the best network drops to 84 per cent. Three serious limits remain: ten archers from a single club, whose discipline and competitive level the paper does not give, a validation that may have mixed arrows from the same archer between training and testing, and a reference that is one expert's annotation rather than the video measure the quiet eye literature usually relies on.
Also phrased as: "measuring the quiet eye"; "electro-oculography in archery"; "quiet eye and artificial intelligence"; "sensors for the archer's gaze"
In practice
The quiet eye is usually measured with an eye tracker and a video annotated frame by frame, a slow job that stays in the laboratory; on the field the idea is going round that light sensors will soon be enough to measure it during training.
Evidence
-
2025
For Archers Published later
10 club archers, 6 arrows each, 60 stretches of signal annotated by an expert: the networks with temporal memory recognise the quiet eye with an accuracy of about 0.93 (CNN-LSTM, F1 0.954) and 0.92 (CNN-GRU, F1 0.928), against 0.51 to 0.68 for a support vector machine depending on signal quality. The abstract gives 95 and 93 per cent, the table 93 and 92, in the published version too: the table is followed here.
Automated detection of quiet eye durations in archery using electrooculography and comparative deep learning models · Fatma Söğüt, Hüseyin Yanık, Evren Değirmenci, İnci Kesilmiş, Ülkü Çömelekoğlu
-
2025
Context Archers Published later
The measurement only holds if the signal is cleaned: the best network goes from about 0.84 on the raw signal to 0.91 with a Butterworth filter and 0.93 with wavelets added, and the mean error on quiet eye onset falls from 20 to 9 to 7 ms. The traditional classifier stays at 0.51 on the raw signal.
Automated detection of quiet eye durations in archery using electrooculography and comparative deep learning models · Fatma Söğüt, Hüseyin Yanık, Evren Değirmenci, İnci Kesilmiş, Ülkü Çömelekoğlu
-
2025
Context Archers Published later
With the cleaned signal the end of the quiet eye is located better than its onset by every method: 4 against 7 ms for CNN-LSTM, 6 against 9 for CNN-GRU, 21 against 30 for the traditional classifier. Between the two best networks the difference is not always significant.
Automated detection of quiet eye durations in archery using electrooculography and comparative deep learning models · Fatma Söğüt, Hüseyin Yanık, Evren Değirmenci, İnci Kesilmiş, Ülkü Çömelekoğlu
Status history
- 2025 Reported The first attempt in the bibliography: ten club archers recorded with electro-oculography, sixty stretches of signal annotated by hand by an expert, and two neural networks with temporal memory that recognise the quiet eye with an accuracy of around 92 to 93 per cent, against 51 to 68 for a traditional classifier. On the raw signal the best network drops to 84 per cent. It is a methods study on ten people, in a single laboratory and with no comparison against the video measure: reported.
Related
Related articles
- Quiet Eye and Visual Training: the Science of Stable AimingQuiet eye in archery: what studies on archers have measured, why there is no right duration to chase, and how to work on the gaze before the release.
- The future of archery: technology, research, evolutionWhat has already been measured about technology in archery: virtual reality, sensors, video, the quiet eye, camera scoring and shooting machines.
Research lines
Cite
Campagna, M. (2026). Quiet eye can be measured automatically from the electrical signal of the eyes (Version 2). In Evidence Registry. Retrieved 16 October 2026, from https://matteocampagna.com/en/registry/mind/automatic-quiet-eye-measurement-eog
@misc{campagna2026_quieteyemisuraautomatica,
author = {Campagna, Matteo},
title = {Quiet eye can be measured automatically from the electrical signal of the eyes},
year = {2026},
howpublished = {Evidence Registry, matteocampagna.com},
url = {https://matteocampagna.com/en/registry/mind/automatic-quiet-eye-measurement-eog},
note = {Version 2; accessed 2026-10-16}
}
Follow this claim
RSS is an address you paste into a feed reader: each time this claim changes status, adds a year to its history or is revised, a new item appears there.
Registry updates by email
At most one email a month, sent only when something has changed: claims that change status, new claims, Lab predictions registered or resolved, new studies and new books.
Sign-ups are handled by Systeme.io, the service that runs the newsletter. Every email has a link to unsubscribe with one click.
Other claims in this domain
- ReportedA larger working memory does not make an archer more accurate3 studies · 79 archers measuredStatus since 2012
- ReportedA longer quiet eye improves shooting accuracy3 studies · 58 archers measuredStatus since 2017
- ReportedA mental profile tells archers apart from athletes in other sports only in part5 studies · 550 archers measuredStatus since 2018
- ReportedA mental training programme improves the elite archer's mood1 study · 16 archers measuredStatus since 2023
- ReportedA mental training programme reduces burnout in elite archers1 study · 16 archers measuredStatus since 2023