Error Awareness

While humans typically detect the majority of their errors (aware errors), some errors go unnoticed (unaware errors). In real life, these unaware errors might have the most negative consequences. For example, accident investigations have shown that unaware errors have led to serious train, ship, and airplane accidents. In the lab, we study error awareness using behavioural as well as neuroimaging methods. We aim to better understand what characterises unaware errors, what factors influence the frequency of unaware errors, and how unaware errors could be avoided.

Post-error Adaptations

Individuals sometimes commit errors, but they usually intend to reduce the number of errors as much as possible. Accordingly, humans show behavioural and neural adjustments after errors (Danielmeier & Ullsperger, 2011), which presumably have the goal to prevent further errors. For example, people often slow down their reaction times after errors (post-error slowing), show improved performance (post-error improvement in accuracy) or increase selective attention to focus on task-relevant information (Danielmeier et al., 2011, 2015).

However, these post-error adjustments can be more or less pronounced, and individuals do not always manage to reduce their error rate after suboptimal performance. Therefore, it has been discussed for some time if post-error adjustments mainly reflect functionally relevant adaptive behaviour or if they mainly reflect mal-adaptive processes, such as slowing down due to surprise, but not necessarily to improve. Increasingly, evidence suggests that post-error adjustments have both adaptive and maladaptive elements which might act on different time-scales (Ullsperger & Danielmeier, 2016).

In the lab, we aim to identify the factors that determine the extent and adaptiveness of post-error adjustments. In addition to healthy young participants, we investigate post-error adaptations in patients with mental health disorders.

Learning from Errors

We also investigate how people learn from errors. There is an ongoing debate in the literature whether people learn more from positive or from negative feedback. The answer to this question might depend on the context during encoding of the to-be-learned items. We have used different decision-making tasks to elicit positive and negative feedback and associate it with specific objects presented during these tasks. We then use computational models to estimate the size and valence of prediction errors during encoding. Additionally, we use functional neuroimaging to investigate modulations in performance monitoring and learning-related networks during the encoding phase.

Neurometabolite levels associated with performance monitoring and post-error adjustments

In a recently developed line of research, we aim to associate performance monitoring processes and post-error adjustments with neurometabolite levels (glutamate, glutamine, GABA) in the medial frontal cortex of healthy participants and individuals with psychosis.