A key human ability is to learn from outcomes by identifying their cause (“assigning credit”), and to figure out whether one’s actions have any influence on our environment. I will present a study in which we examined how participants assign credit to themselves versus others and estimate their control over the environment. For this, we developed a novel behavioural paradigm, in which participants had to infer from feedback how well they and another player perform, and how much control each of them exerts over the environment. We employed this paradigm in a 3T fMRI study (n=34) and an online study (n=36). We fit Bayesian learner models to participants’ behaviour to estimate how an optimal learner would assign credit.

Under high uncertainty, participants engaged in specific patterns of exploratory behaviour (‘active disambiguation’, AD) to better disambiguate the feedback. Crucially, during AD, people changed their performance to test the effect of their behaviour on the feedback, thereby inferring their control. We found that this is consistent with that of an optimal Bayesian learning model. The supramarginal gyrus (SMG) in the inferior parietal lobe signalled both AD and high uncertainty during action generation. In the outcome phase, we found that the SMG not only tracks the inferred control, but also the components necessary for inferring control. These results illuminate how people assign credit and learn about control by engaging in different action types, and also which neural mechanisms underly these behaviours.