Fixational eye movements - keeping the eyes on target

This thesis presents three different strands of research into fixational eye movements: the detection of microsaccades, the interaction between fixational eye movements and the visual stimulus, and the interplay of microsaccades and drift. Fixational eye movements actively sample the visual stimulus, generating spatial and temporal information about its structure while also preventing fading of the percept. Two new microsaccade detection methods are presented: the binocular correlation method most suited to data from video-based eye trackers and the monocular microsaccade detection method suited to microsaccade detection from high resolution adaptive optics eye tracking. These methods are systematically compared to the established standard method in the field by (Engbert & Kliegl, 2003) using the F1 score of the precision-recall metric. Manually labelled eye movement data will be made available online to be used as training and testing data for future microsaccade detection methods. The active responses of fixational eye movements to visual stimuli of different eccentricities, grating orientations, and levels of blur were also investigated. It was found that fixational eye movements exhibit systematic changes in response to stimulus eccentricity and orientation, but not to the different levels of blur tested. To explore possible underlying mechanisms data are presented on correlations of unreferenced motion thresholds and centre estimations with measures of fixational stability. Both were good predictors of fixational stability, suggesting that these or mutual underlying processes contribute to fixational eye movement control. Microsaccades and drifts are compared for their relative contribution to fixational stability. Traditionally microsaccades were considered to be correcting for the random error introduced by drift, but the data examined in this thesis suggest a complimentary role between microsaccades and drift in stabilising gaze.