Neural correlates of location–response compatibility in an immersive virtual-reality Attention Network Test: a multiverse electroencephalography analysis

Abstract

Immersive virtual reality (VR) can preserve the logic of laboratory attention tasks while altering the perceptual-action context in which attentional control is expressed. In this study, we examined the neural underpinnings of location-response compatibility in a VR adaptation of the Attention Network Test-Revised (ANT-VR), using a restricted preprocessing multiverse to account for uncertainty arising from defensible EEG analysis choices. Forty-four young adults contributed complete ANT-VR behavioural data. Target-locked EEG analyses were conducted across 192 preprocessing branches, with branch-level participant contributions varying after quality check and trial-count filtering. The contrast compared location-response incompatible with compatible trials and was balanced within participants across cue-target interval, cue condition, flanker congruency, and, for spatial-cue trials, conditions in which spatial cues were valid or invalid for subsequent targets. Across the multiverse, the N2pc-window posterior-lateralisation contrast could be defined in all branches and showed high directional stability: the median incompatible-minus-compatible effect was 0.30 µV [IQR: 0.22 to 0.41], with positive effects in 192/192 branches, nominal evidence in 79/192 branches, and Holm-corrected evidence in 42/192 branches. Comparison across ERP measures indicated that this pattern was more consistent for N2pc-window posterior lateralisation than for P1, posterior N1, frontocentral N2, P3, or response-referenced C3/C4 measures. Comparison across C3/C4 reference frames further constrained the interpretation: target-location-referenced C3/C4 showed the strongest effect, whereas response-referenced C3/C4 was weaker. Behavioural analyses showed no reliable compatibility differences. These findings suggest that location-response compatibility in immersive ANT-VR modulates target-locked lateralised neural activity associated with spatial selection and target-location coding, rather than producing broad sensory, conflict-related, P3-related, or specifically response-referenced modulation. Significance Statement Behavioural performance can conceal neural effects of spatial compatibility, particularly in immersive tasks where perception and action are closely coupled. In the Attention Network Test–VR, location-response incompatibility produced no reliable reaction-time or accuracy cost, yet target-locked EEG showed a directionally stable posterior-lateralisation shift in the N2pc window across 192 defensible preprocessing specifications. Target-location-referenced C3/C4 activity showed stronger and more consistent modulation than response-referenced activity, arguing against a purely response-preparation account. These findings indicate that task-irrelevant target location can alter spatial-selection or target-location coding even when behavioural performance is unchanged, while illustrating how multiverse analysis can test whether such neural conclusions depend on preprocessing choices.