Fixational saccades during grating detection and discrimination

• Fixational saccades did not consistently adapt to task-relevant stimulus properties.
• In detection, they were similar in grating-with-noise and noise-only displays.
• This suggests that they did not mainly subserve optimization of information uptake.
• The results also showed high interindividual variability.
• In detection, saccades were fewer and smaller in trials with highly visible stimuli.

We investigated the patterns of fixational saccades in human observers performing two classical perceptual tasks: grating detection and discrimination. First, participants were asked to detect a vertical or tilted grating with one of three spatial frequencies and one of four luminance contrast levels. In the second experiment, participants had to discriminate the spatial frequency of two supra-threshold gratings. The gratings were always embedded in additive, high- or low-contrast pink noise. We observed that the patterns of fixational saccades were highly idiosyncratic among participants. Moreover, during the grating detection task, the amplitude and the number of saccades were inversely correlated with stimulus visibility. We did not find a systematic relationship between saccade parameters and grating frequency, apart from a slight decrease of saccade amplitude during grating discrimination with higher spatial frequencies. No consistent changes in the number and amplitude of fixational saccades with performance accuracy were reported. Surprisingly, during grating detection, saccade number and amplitude were similar in grating-with-noise and noise-only displays. Grating orientation did not affect substantially saccade direction in either task. The results challenge the idea that, when analyzing low-level spatial properties of visual stimuli, fixational saccades can be adapted in order to extract task-relevant information optimally. Rather, saccadic patterns seem to be overall modulated by task context, stimulus visibility and individual variability.