Fixational Eye Movements as Active Sensation for High Visual Acuity

Academic Background

Human visual perception is a complex process, especially when we attempt to maintain stable gaze. Even then, the eyes still produce small involuntary movements known as fixational eye movements (FEM). These FEM typically include drifts and microsaccades. Previous studies have shown that despite causing retinal image jitter, the human visual system can perceive details finer than the amplitude of FEM. This phenomenon has sparked widespread interest in the scientific community: why do FEM not only fail to impair visual acuity but may also positively influence it?
To address this question, researchers combined theory and experiment to investigate how FEM affect visual coding and acuity under different conditions. By studying the dynamics of FEM and their impact on retinal neural activity, the study aimed to explain how FEM play an active role in high-acuity tasks and uncover the underlying neural coding mechanisms.

Paper Source

This research was conducted by Trang-Anh E. Nghiem, Jenny L. Witten, Oscar Dufour, Wolf M. Harmening, and Rava Azeredo da Silveira. The research team hailed from several prestigious institutions, including École Normale Supérieure in Paris, the Institute of Molecular and Clinical Ophthalmology Basel, the Department of Ophthalmology at Rheinische Friedrich-Wilhelms-Universität Bonn, and the Department of Economics at the University of Zurich. The paper was published in PNAS (Proceedings of the National Academy of Sciences) on February 4, 2025, titled “Fixational eye movements as active sensation for high visual acuity.”

Research Process and Results

1. Experimental Design and Participants

The study recruited 17 healthy adult participants (8 males, 9 females, aged 10 to 42 years) and used an adaptive optics scanning laser ophthalmoscope (AOSLO) to record FEM trajectories during a visual discrimination task. Participants had to identify the orientation of a Snellen E letter (up, down, left, or right), with the size of the letter adjusted via a Bayesian adaptive staircase procedure ranging from 0.6 to 1.6 arcmin. Each experiment included 20 trials, repeated five times.

2. Recording and Analysis of FEM Trajectories

Using AOSLO, the research team recorded retinal motion and stimulus position at a high temporal resolution (approximately 960 Hz). FEM trajectories were modeled as a two-dimensional diffusion process, and the diffusion coefficient (D) was calculated to quantify the magnitude and dynamics of FEM. Results showed that FEM trajectories followed a random diffusion process, with significant variability in the diffusion coefficient across participants.

3. Retinal Response Model

To simulate the effect of FEM on retinal neural activity, the research team developed a retinal ganglion cell (RGC) response model. This model assumed Gaussian-distributed receptive fields with spatiotemporal kernels. RGC firing rates were simulated using a Poisson process, and a Bayesian classifier was employed to classify stimulus orientation. The classifier accumulated evidence from retinal neural activity, updating the posterior probability distribution to infer stimulus direction and position.

4. Comparison of Model and Experimental Results

The study found that intermediate-amplitude FEM significantly improved discrimination of fine stimuli, while too small or too large FEM impaired visual acuity. These results aligned closely with experimental data, validating the model’s effectiveness. Additionally, participants dynamically adjusted FEM amplitude based on stimulus size, maintaining it within a near-optimal range to maximize visual acuity.

Conclusion

By combining experiments and theoretical models, the study revealed the dual role of FEM in high-acuity tasks: they refresh retinal image content, making RGC activity more sustained and enhancing encoding of fine details; and they introduce temporal variations, providing additional stimulus information. The study also found that FEM amplitude remains within a near-optimal range, further supporting its positive role in visual perception.

Highlights

1. Significance: This study is the first to reveal the positive role of FEM in high-acuity tasks and propose the underlying neural coding mechanisms, filling a gap in the field.
2. Methodological Innovation: The study combined high-precision eye-tracking technology (AOSLO) with Bayesian classifiers, offering a new methodological framework for visual perception research.
3. Application Value: The findings provide a theoretical basis for diagnosing and treating visual perception disorders and offer inspiration for designing visual algorithms in artificial intelligence.

Additional Valuable Information

The research team also explored the applicability of FEM in natural viewing conditions and proposed directions for future research, including extending to more complex visual tasks and broader stimulus sets. Moreover, the study data and analysis codes are publicly available on the Mendeley Data platform, providing valuable resources for subsequent research.

Through this study, scientists have not only unraveled the mystery of FEM in visual perception but also emphasized the importance of active sensing in the visual system, paving the way for future visual science research.