Temporal dynamics of dual-task interference in the brain in a simulated driving environment

Dual-task interference occurs when the brain's limited cognitive capacity leads to performance impairments during overlapping tasks. We aimed to investigate the time profile of this phenomenon using EEG, multivariate pattern analysis (MVPA), and drift-diffusion modeling (DDM). Participants performed a tone discrimination task, followed by a lane-change task with short or long stimulus onset asynchrony (SOA) in a simulated driving environment. Dual-task interference increased lane-change reaction time, attributable to changes in decision and post-decision times, as indicated by DDM. MVPA findings revealed decreased decoding accuracy for the lane-change task in short SOA compared to long SOA and single-task conditions, highlighting interference. Using MVPA temporal generalization, we investigated how interference affects neural pattern stability over time, revealing disruptions as early as ∼250 ms after the lane change stimulus onset in short SOA trials, suggesting partial parallel processing in early stages. To assess stability across task conditions, we also applied MVPA conditional generalization. This analysis showed a delayed above-chance decoding accuracy (starting at ∼450 ms) in short SOA trials compared to long SOA and single-task conditions, suggesting a bottleneck in later processing stages. MVPA searchlight analysis further revealed a reduction in task-specific information, progressing from occipital and parietal regions (resonsible for perceptual and central processing) to frontal regions (responsible for decision-to-action mapping) in short versus long SOA trials. Overall, our findings suggest that tasks are processed partially in parallel during the first hundred milliseconds, particularly in perceptual and decision stages. Beyond ∼450 ms, competition exists in routing of information to motor areas, causing serial processing and delays for the second task.