Bayesian Dynamical Modeling of Fixational Eye Movements

Area of Science:

• Neuroscience
• Computational Neuroscience
• Vision Science

Background:

• Human eye movements are constant, even during fixation, involving slow (drift, tremor) and fast (microsaccades) components.
• The statistically self-avoiding random walk (SAW) model has been proposed to describe the complex dynamics of physiological drift.

Purpose of the Study:

• To implement a data assimilation approach for the SAW model to analyze fixational eye movements and microsaccades.
• To investigate the relationship between the SAW model's activation and microsaccade occurrence using Bayesian parameter estimation.
• To explore individual differences in fixational eye movement behavior.

Main Methods:

• Utilized a data assimilation approach to implement the SAW model.
• Applied Bayesian parameter estimation to experimental high-resolution eye-tracking data.
• Analyzed the SAW model's likelihood function for individual human observers.

Main Results:

• Established a relationship between SAW model-predicted activation and microsaccade occurrence.
• Demonstrated that the model's latent activation correlates with microsaccade onsets and offsets.
• Provided evidence supporting a triggering mechanism for microsaccades.

Conclusions:

• The SAW model effectively captures individual variations in fixational eye movements.
• The SAW model serves as a valuable tool for studying the interplay between physiological drift and microsaccades.
• Findings enhance understanding of individual differences in microsaccade behavior and visual processing.