Computational and Neural Mechanisms of Planning

Abstract

In daily life, we often plan future actions without immediate feedback, relying on a sequence of interdependent decisions. Most choices involve multiple steps rather than isolated decisions, making planning an essential cognitive process. However, how humans plan in novel, real-world scenarios is still not fully understood. To explore this, we developed a task where participants planned in realistic situations by selecting one option from three categories (e.g., cakes, locations, decorations for a birthday party). Participants' responses and gaze patterns were tracked. The results showed that participants prioritized decisions based on subjective importance and difficulty, typically first selecting what they found most important and easiest. Their behavior reflected complex internal planning, navigating a decision tree whose complexity increased with each additional option. In a second experiment, we modified the task and added EEG recording to investigate the neural dynamics underlying planning behavior. This allowed us to capture brain activity associated with decision prioritization and internal evaluation processes. Together, these experiments provide new insights into how people engage in multi-step planning by evaluating multiple alternatives through imagination, internal sampling, and prioritization, both behaviorally and neutrally.