1. Disentangling cognition and eye movements in EEG using generalized additive mixed models.
- Author
-
Meghanathan, Radha Nila, Nikolaev, Andrey R., and van Leeuwen, Cees
- Subjects
- *
EYE movements , *ELECTROENCEPHALOGRAPHY , *MOVEMENT sequences , *COGNITION - Abstract
When EEG is co-registered with eye movements in free-viewing tasks, EEG response to cognitive processes of interest maybe confounded with that to sequential eye movements (Dimigen et al 2011; Nikolaev et al 2016). These two can be disentangled using generalized additive mixed models (GAMM) with eye movement parameters and experimental conditions as predictor variables. Since GAMM can account for non-linearity in data, it is particularly suited to modeling EEG. GAMM was used to investigate the role of visual saliency in saccade guidance in an EEG eye movement co-registration study. Participants were asked to search, during an 8s interval, for a contour formed by 7 collinear Gabor elements embedded in a field of similar but randomly oriented distractors. Participants were then asked to indicate whether the contour was present, which was the case in half of the trials. Since the interval preceding saccade onset (presaccadic interval) involves saccade planning to the next fixation location, EEG epochs in this interval were modeled using GAMM with visual saliency of the subsequent fixation location and perisaccadic eye movement parameters as predictors. Besides the effect of fixation duration and saccade size on pressaccadic EEG, low presaccadic EEG amplitude was associated with high saliency at the next fixation location in both contour-present and contour-absent conditions (van Humbeeck et al 2018). This reveals the role of bottom-up saliency in saccade guidance. Using GAMM, the effects of visual saliency on saccade guidance were successfully isolated from the effect of eye movements on EEG. [ABSTRACT FROM AUTHOR]
- Published
- 2019