Examining Post-Error Behavior in a Complex Multitasking Environment

Human operators are more prone to errors under high-workload conditions. However, error-commission research in cognitive science has been limited to studying behavior in single-choice reaction time tasks, which do not represent the complex multitasking scenarios faced in the real-world. In the current paper, prior evidence for a cognitive error-monitoring mechanism was applied toward predictions for how humans would react after making errors in a more ecologically valid multitasking paradigm. Previous work on neural and behavioral indices of error-monitoring strongly supports the idea that errors are distracting and can deplete attentional resources. Therefore, it was predicted that after committing an error, if a subject is subsequently presented with two simultaneously initiated task alerts (a conflict trial), they would be more likely to miss their response opportunity for one task and stay tunneled on the other task that originally caused the error. Additionally, it was predicted that this effect would dissipate after several seconds (under different lag conditions), making the error cascade less likely when subsequent tasks are delayed before presentation. A Multi-Attribute Task Battery was used to present the paradigm and collect post-error and post-correct performance measures. The results supported both predictions: Post-error accuracy was significantly lower as compared to post-correct accuracy (as measured through post-trial error rates). Post-trial error rates were also higher at shorter lags and dissipated over time, and the effects of pre-conflict performance on post-trial error rates was especially noticeable at shorter lags (although the interaction was not statistically significant). A follow-up analysis also demonstrated that following errors (as opposed to following correct trials), participants clicked significantly more on the task that originally caused the error (regardless of lag). This continued task-engagement further supports the idea that errors lead to a cognitive tunneling effect. The study provides evidence that in a multitasking scenario, the human cognitive error-monitoring mechanism can be maladaptive, where errors beget more errors. Additionally, the experimental paradigm provides a bridge between concepts originating in highly controlled methods of cognitive science research and more applied scenarios in the field of human factors.