Schedule of Events | Search Abstracts | Invited Symposia | Symposia | Poster Sessions | Data Blitz

EEG Decoding of Goal Representations During Multistep Actions

Poster Session C - Sunday, March 8, 2026, 5:00 – 7:00 pm PDT, Fairview/Kitsilano Ballroom

Bettina Bustos¹ (), Jiefeng Jiang, Eliot Hazeltine; ¹University of Iowa, Psychological and Brain Sciences

Action selection is guided not only by the immediate outcomes of actions but also by their contribution toward overarching goals. Traditional one-step tasks conflate goals with expected outcomes, making it difficult to dissociate these processes. Here, we used a novel multi-step construction task in which participants sequentially selected between line segments to build target shapes. This design allowed us to examine how people maintain goal representations while tracking intermediate outcomes. Behaviorally, participants (N = 27) selected actions that advanced the construction toward the intended goal, showing sensitivity to both goal progress and the current state of the shape options. Choice behavior was also biased towards the selection of lines that contributed towards goal shapes that were closer to completion, indicating dynamic tracking of goal status across the sequence. These findings demonstrate that participants maintain an internal model of multiple goals and dynamically update it after each action. These results provide a behavioral foundation for EEG decoding analyses that dissociate neural representations of goals and intermediate action effects during multi-step behavior. Preliminary EEG work has focused on training a line decoder in a separate oddball task, which can reliably distinguish between individual line segments at above-chance accuracy. This decoder will be used to test whether representations of specific lines—and by extension, evolving goal states—can be tracked during multi-step construction behavior.

Topic Area: EXECUTIVE PROCESSES: Goal maintenance & switching