EEG microstates are brief metastable states of EEG activity defined by non-overlapping potential maps. The association of specific microstates with certain tasks is well established (Tarailis et al., 2023). Microstates B and E are typically considered the visual and interoceptive microstates, respectively. The present study aimed to investigate how microstate activity interacted with performance in a visual task (N2 posterocollateral potential, N2pc), as a heuristic of the trade-off between external and internal attention. In this task, participants identify the position of a lateralised object among distractors.

Microstates were fitted to data from 40 participants in the ERP Core dataset (Luck, 2020) using the modified K-means algorithm (Tarailis et al., 2023). Microstates sequences were predicted for each trial, and microstate coverage—the proportion of samples assigned to a given microstate—was analysed alongside task reaction time (RT).

As would be expected, microstate coverages were all negatively correlated – as they exist within a finite timeline; the strongest negative correlation was between microstates B and E. RT was positively correlated (Pearson) with the coverage of the microstate B prior to stimulus presentation (r=0.036, p<.001). A stronger positive correlation was noted in terms of the rate of the recruitment of microstate B (r=0.113, p<.001), as well as a negative correlation for microstate E (R=-0.072, p<.001). The coverage of microstates B and E during stimulus presentation positively and negatively correlated (respectively) with the RT (r=0.076, p<.001; R=-0.028, p<.001).

The positive correlation between RT and the visual microstate appears counter-intuitive, but may reflect the paradoxical interaction of spontaneous and task-evoked activity (He, 2013), and the effect of effort-related recruitment of a microstate for "harder" (and slower) trials. The inverse relationships between "visual" and "interoceptive" microstates suggest a potential "attentional budget" that may oscillate between internal and external processing, to be further informed through computational modelling.