Agency has often been assumed to be metacognitive. We are trying to unpack this assumption: What do we mean, exactly, when we say that agency is metacognitive? Are our assumptions valid?
We collect behavioural data and test computational models to compare agency and motor metacognition.

In our first paper, we show that the computations underlying agency judgments differ conceptually from those underlying confidence judgments. We think that this is a good reason to be careful with names, and use one term for computationally unambiguous processes. See also a twitter thread for a quick summary.

We collaborated with Roy Salomon and his team for this project.

In another line of research, we generally asked questions about agency using metacognitive methods. We reasoned that measuring the precision of metacognition of agency in different conditions would allow us to compare them in a way that explicit agency ratings does not, because they are subject to response biases.
In three studies using this approach, Angeliki Charalampaki found, first, that we more confidently attribute actions to ourselves when they had the outcome that we intended them to have. She also suggests that the mechanisms that the brain uses to attribute agency over action outcomes may not be the same as those we use to attribute agency over the actions themselves. And, finally, she found a way to manipulate not just the the visual sensory consequences of an action (as most studies do) but also the _tactile_ consequences of an action, and measured participants' agency judgments and metacognition, compared to more traditional manipulations.

Along the way, we found that measuring metacognition of agency can, indeed, help in comparing conditions directly. But also that it comes with all sorts of problems!