September 2019
Volume 19, Issue 10
Open Access
Vision Sciences Society Annual Meeting Abstract  |   September 2019
Rapid and robust online saccade detection
Author Affiliations & Notes
  • Richard Schweitzer
    Department of Psychology, Humboldt-Universität zu Berlin, Germany
    Bernstein Center for Computational Neuroscience Berlin, Germany
    Berlin School of Mind and Brain, Humboldt-Universität zu Berlin, Germany
  • Martin Rolfs
    Department of Psychology, Humboldt-Universität zu Berlin, Germany
    Bernstein Center for Computational Neuroscience Berlin, Germany
Journal of Vision September 2019, Vol.19, 83. doi:https://doi.org/10.1167/19.10.83
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Richard Schweitzer, Martin Rolfs; Rapid and robust online saccade detection. Journal of Vision 2019;19(10):83. https://doi.org/10.1167/19.10.83.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Research on active vision heavily relies on gaze-contingent stimulus manipulations triggered by the onset of a saccade. Rapid online saccade detection is particularly essential when investigating trans-saccadic perception. Most paradigms, however, use either spatial-boundary techniques or absolute velocity thresholds. The former is reliable but slow; the latter is faster but unreliable, especially at higher sampling rates or noise levels. Here we present a velocity-based algorithm for online saccade detection (inspired by Engbert and Kliegl’s widely used algorithm for microsaccade detection), that outperforms traditional techniques in speed and accuracy while allowing for the flexible adjustment of detection criteria. Saccades are detected using two-dimensional velocity thresholds estimated based on the variance of preceding fixation data, compensating for noisy or dropped samples by smoothing and interpolation. An optional direction criterion limits detection to predefined saccade directions, increasing robustness against false alarms from fixational eye movements and blinks. We validated the algorithm using datasets of >34,000 recorded saccades. Saccades were detected online as early as 3 ms after their onset (ground truth detected offline) with false alarm rates of < 1%. Even when corrupting the data by adding Gaussian noise (SD up to 0.1 degrees of visual angle) and randomly removing up to 30% of all samples, high accuracy was preserved while detection latency increased by 10 ms at most. Finally, we developed a perceptual test to verify the algorithm’s usefulness for gaze-contingent, intra-saccadic presentations. Using a projection system operating at 1440 Hz, we displayed a Gabor patch drifting at saccadic peak velocities for 14 ms. While this stimulus was invisible under fixation, it was reliably detected when presented upon online saccade detection. Using a photodiode, we confirmed that stimuli were physically shown 20 ms after saccade onset (average, including all system latencies). Thus, our algorithm provides a useful tool for investigating trans-saccadic perception.

Acknowledgement: Studienstiftung des deutschen Volkes and Berlin School of Mind and Brain (RS) and Deutsche Forschungsgemeinschaft DFG (grants RO3579/2-1, RO3579/8-1, RO3579/10-1) (MR) 
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×