Voluntary Action, Evidence Accumulation, and Decision-Making
Some of our actions are responses to external events. Other actions don’t have immediate external triggers. We call these self-initiated or voluntary actions. In this project, working with Professor Patrick Haggard and others at the UCL Institute of Cognitive Neuroscience, I’m working on outlining how these self-initiated actions work.
The standard way to study self-initiated actions is to ask participants to press a button “whenever they feel like it”, and record what happens in the brain in the seconds before they do so. This has shown us that self-initiated actions depend on neurons in the medial prefrontal cortex, and in particular on the Supplementary Motor Area. Using EEG, we can see that activity in this area ramps up in the second or so prior to action, producing a signal we call the Readiness Potential (RP), pictured above.
We’re interested in what this pre-movement activity actually does. Originally people thought it was just motor preparation: the work the brain needs to do in between deciding to move and actually moving. However, if you ask people when they actually decided to move, they report a time just before they actually did so, well after the RP begins. Some people think this means that the decision to move is made unconsciously at the start of the RP, and we only become aware of it later.
The alternative is that the RP is actually part of the decision-making process itself. Our colleague Aaron Schurger suggested in 2012 that these kind of actions, where participants are told to press a button whenever they feel like it, might be triggered by random fluctuations in the brain reaching a threshold, and that the RP just reflects this random activity rising towards threshold. Others have argued that it reflects uncertainty or conflict in the decision to act, or temporal expectations and predictions about when an action will occur. In this project, we try to test these ideas.
To do this, we record EEG or fMRI while participants perform self-initiated actions as part of realistic tasks. This allows us to see how changes to the context in which a person decides to initiate an action affect the neural processes that make that action happen.