Neural mechanisms of self-generated action sequences

Complex problems often allow multiple paths to a solution. Choosing and taking the best path is an important part of the executive cognition that underpins intelligent problem-solving behaviour. However, once a path is chosen, the motor system must be activated for executing it. This interface between problem-solving and self-generated action has rarely been studied. We recorded EEG movement-related potentials while 25 participants (7 males, 18 females) performed the “Tower of London” problem-solving task. In a control condition, participants merely followed instructed steps without planning for any goal, and thus without any sense that their movements solved a problem. Readiness potentials (RPs) preceding actions showed a more sustained preparatory negativity for self-generated than stimulus-driven movements. Critically, this effect was most pronounced at the first move of a sequence and diminished at later stages, indicating that preparatory activity is closely linked to the planning demands of sequence initiation. Consistent with this, contralateral motor β-band suppression was stronger for self-generated actions, particularly at sequence onset, but remained present across all moves, indicating that it is not selectively modulated by sequence position in the same way as the RP. Multivariate pattern analysis (MVPA) further showed that self-generated and stimulus-driven actions could be reliably distinguished throughout the entire preparatory period. Taken together, these results show a deep interaction between executive function and self-generated actions, and draw attention to the fact that, if a problem can be solved, then actually solving it generally requires executive cognition to trigger self-generated actions, based on a plan. Significance Statement Self-generated actions are often studied as isolated, arbitrary movements, whereas real behaviour typically involves solving problems through sequences of goal-directed acts. Using EEG during a modified Tower of London task, we show that neural markers of self-generated action are strongest when participants initiate a planned action sequence and diminish as the sequence unfolds. Readiness potentials, motor beta suppression and multivariate decoding all distinguished self-generated from stimulus-driven actions during preparation. These results show that the neural signatures of self-generated action are closely linked to planning demands, bridging the neuroscience of volition with the psychology of problem-solving.