# Perceptual punctuation: fixational eye movements reveal segmentation of auditory streams

## TL;DR

Fixational eye movements can reveal how people perceive and organize complex sounds over time, even when the sounds are ambiguous.

## Contribution

This study shows that eye movement patterns reflect perceptual organization of auditory streams, not just physical properties.

## Key findings

- Eye-velocity spectra showed percept-dependent power shifts between 2 and 4 Hz.
- Microsaccade reaction times varied based on the reported auditory percept.
- Fixation dynamics reflected bistable fluctuations in auditory stream perception.

## Abstract

Perception operates as rhythmically structured sampling in which temporal predictions determine when incoming signals are weighted. Fixational eye movements carry opposing consequences, enhancing acuity yet inducing brief peri-saccadic suppression, suggesting that their timing is paced by expected, salient rhythms. Auditory scenes can be parsed into competing streams that unfold over time. If fixation dynamics are shaped by temporal expectation, and auditory streaming imposes a percept-dependent temporal structure on otherwise identical acoustics, then fixational eye movements might provide a window into how listeners parse sound over time. We asked whether fixational eye movements reflect the perceived rather than the physical temporal organization of an ambiguous ABA– pattern.

While listeners fixated and either attended High, Low, or All tones (Experiment 1, n = 15) or freely reported their percept (Experiment 2, n = 15), we recorded binocular eye position (500 Hz) and quantified microsaccade (MS) dynamics and eye-velocity spectra.

Across both experiments, eye-velocity spectra showed a percept-dependent redistribution between 2 and 4 Hz, with relative power shifting with the instructed/reported stream. A normalized 4–2 Hz index (ΔPSD) separated Low-tone from High-tone percepts across procedures. Time-resolved analyses further revealed within-trial waxing-and-waning of 2 vs. 4 Hz dominance, consistent with bistable fluctuations in maintaining a stream. Moreover, microsaccade reaction time (msRT), aligned to the onset of the sound sequence, differed significantly depending on the percept.

These findings extend oculomotor inhibition beyond discrete events, positioning fixation dynamics as a sensitive, report-free marker of auditory scene organization. We discuss mechanistic links to temporal attention and active sensing, and implications for a multisensory timing framework.