Naturalistic viewing conditions can increase task engagement and aesthetic preference but have only minimal impact on EEG Quality

Abstract

Free gaze and the use of dynamically changing video stimuli are typically avoided in EEG exper- iments to avoid artefacts and confounds related to uncontrolled eye movements. Yet, often it is unclear whether these artificial secondary manipulations might have unwanted effects on the pri- mary measures of interest and for a growing number of research questions removing them would be beneficial: Among those is the investigation of visual aesthetic experiences, which typically involve open-ended exploration of highly variable stimuli. Here we aimed to quantify the effect of fixation task and using still vs. movie stimuli on EEG signal quality and several behavioral and physiological measures of interest during an aesthetic rating task. Participants observed scenes from landscapes and dance performances and rated each stimulus for both aesthetic appeal and their state of boredom while watching it. The scenes were presented either as dynamic video clips or static pictures, and participants observed them either with unconstrained gaze or under attempted fixation We recorded EEG, ECG and eyetracking from 43 participants. An auditory stream of 40Hz amplitude modulated pink noise was played during each trial and signal-to-noise-ratio (SNR) of the auditory steady-state response measured at the scalp was extracted as a proxy measure for overall EEG signal quality. The study including hypotheses and a priori power analysis was preregistered. We found that both behavioral ratings were influenced by the experimental conditions: boredom and aesthetic ratings were positively affected by dynamic video stimuli, indicating that these are experienced as more engaging; both these effects were stronger in dance. As already reported before, landscape stimuli were experienced as more appealing. Fixaton task, on the other hand, had no significant effect on the ratings which is encouraging given how canonically it is applied. Eye movements were significantly affected not only by viewing task, but by stimulus dynamics and content as well: we observed fewer eyeblinks, saccades and microsaccades in video stimuli, and fewer saccades but more microsaccades in dance than in landscape stimuli, with several significant interactions. EEG SNR, to our surprise, was barely affected by fixation task - despite only minimal preprocessing and no trial rejection. We nevertheless believe that the new metric is sensitive to capture noise: it was significantly correlated with the number of eye blinks, and after cleaning the dataset with an ICA based preprocessing pipeline the significant effect of fixation task and the correlation with blink rate vanished. We see these as promising results indicating that at least in the lab more liberal experimental conditions could be achieved without significant loss of signal quality. Specifically the use of dynamic video material bears a lot of potential for future investigations in human neurophysiological studies.