An Electroencephalographic Investigation of the Impact of Eye Movements in a Memory Probe Task

Lateral saccades represent a major source of noise and confounds, particularly for event-related potentials (ERPs) that rely on hemispheric imbalances in neural activity elicited by lateralized stimuli during central fixation. These include lateralized ERPs such as the contralateral delay activity (CDA), which indexes visual working memory (VWM) load. Due to its relatively small amplitude and strict fixation requirement, the CDA is particularly vulnerable to contamination from eye movements, which usually cause the contaminated trial to be discarded. In this context, independent component analysis (ICA) offers an alternative to the traditional epoch rejection method, as it removes ocular artifacts without discarding entire trials. However, ICA's effectiveness may be limited if saccade-related activity is not fully removed, or if trials in which participants directed their foveae toward a lateral target stimulus are retained, leading to bilateral representation. In the present study, we compared the efficacy of ICA and epoch rejection in preserving CDA features when participants were allowed to saccade. Participants were asked to memorize an array composed of a variable number of laterally displayed colored squares. In half of the experiment, participants had to keep their gaze at fixation, whereas they had to saccade toward the memoranda in the other half. The memory array was displayed for either 100 ms or 500 ms to examine how the post-saccade physical availability of the memoranda influenced CDA amplitude and latency. The results showed that ICA correction preserved the quality and defining features of the CDA component as well as, or in some respects better than, epoch rejection. Notably, the post-saccade physical availability of the memoranda affected the latency of the CDA, with shorter offset latency observed when the memoranda were exposed for 500 ms compared to 100 ms, likely reflecting post-saccade retinotopic recoding of the memoranda.