Search

Detection of the smooth pursuit eye movement can be used to select an object moving on a computer screen (Vidal et al., 2013). However, moving objects may also become pursued without intention to select and become incorrectly selected. We hypothesized that intentional and spontaneous pursuits can be accompanied by different EEG patterns. We coregistered EEG and eye movement under two conditions: for intentional pursuit, participants were asked to select moving objects with gaze (pursuit caused target highlighting); spontaneous pursuit was provoked by an instruction to count dots on the moving objects. Both intentional and spontaneous pursuits were featured with an EEG lambda wave well-known for static gaze fixations. The intentional, but not spontaneous selection was accompanied, in all of the 20 participants, by an EEG wave, presumably related to the expectation of the feedback. This wave was similar, by time course and topography, to the EEG marker described for static object selection (Shishkin et al., 2016). The selection-related EEG marker did not differ between moving objects selection with smooth pursuit and static objects selection with dwells. Using a deep neural net (EEGNet, Lawhern et al., 2018), we successfully classified intentional vs. spontaneous pursuits. These results support the proposal of enhancing gaze interaction with an expectation-based passive brain-computer interface (Protzak et al., 2013), showing high robustness of the EEG marker related to the expectation of the interface feedback. The study was partially supported by the Russian Science Foundation, grant 18-19-00593 (development of techniques for data preprocessing and classification). [ABSTRACT FROM AUTHOR]

Unlike most human behaviors, eye movements with similar overt characteristics can be either automatic or intentional and consciously controlled. Contrasting these two types of eye movements may help to study mechanisms of intention and to search for the "markers" of intention in brain signals accompanying gaze- based interaction with a computer. We review our studies that employed this approach. Samples of gaze-based interaction and spontaneous gaze behavior were collected in participants who played "gaze-controlled" games. The data included EEG recorded under interaction using dwells (Shishkin et al., 2016) and smooth pursuit (Zhao et al., submitted), as well as MEG+EEG and (separately) fMRI recorded under interaction using dwells. Gaze dwells/pursuits were selected for analysis using a set of criteria maximizing contrast on the "intentionality" scale and minimizing influence of irrelevant factors. The EEG related to gaze interaction but not to spontaneous gaze behavior exhibited a component likely related to the feedback expectation, in line with Protzak et al. (2013). The MEG and the fMRI showed signs of brain activity that were unrelated to the feedback expectation and presumably more intention-specific. We are currently using contrasting interaction-related vs. spontaneous gaze behavior to develop an interface that on-the-fly classifies gaze events based on the neural markers, which we call the eye-brain-computer interface (EBCI). Such an interface may enable particularly fluent human-machine interaction. [ABSTRACT FROM AUTHOR]