Using N2pc variability to probe functionality: Linear mixed modelling of trial EEG and behaviour

Abstract

This paper has two concurrent goals. On one hand, we hope it will serve as a simple primer in the use of linear mixed modelling (LMM) for inferential statistical analysis of multimodal data. We describe how LMM can be easily adopted for the identification of trial-wise relationships between disparate measures and provide a brief cookbook for assessing the suitability of LMM in your analyses. On the other hand, this paper is an empirical report, probing how trial-wise variance in the N2pc, and specifically its sub-component the N_T, can be predicted by manual reaction time (RT) and stimuli parameters. Extant work has identified a link between N2pc and RT that has been interpreted as evidence of a direct and causative relationship. However, results have left open the less-interesting possibility that the measures covary as a function of motivation or arousal. Using LMM, we demonstrate that the relationship only emerges when the N_T is elicited by targets, not distractors, suggesting a discrete and functional relationship. In other analyses, we find that the target-elicited N_T is sensitive to variance in distractor identity even when the distractor cannot itself elicit consistently lateralized brain activity. The N_T thus appears closely linked to attentional target processing, supporting the propagation of target-related information to response preparation and execution. At the same time, we find that this component is sensitive to distractor interference, which leaves open the possibility that N_T reflects brain activity responsible for the suppression of irrelevant distractor information.

Keywords: Attention; Linear mixed modelling; Multimodal data; N2pc.