Context shapes action understanding in social interactions

Abstract

This repository contains processed data from dyadic interaction tasks (rock-paper-scissors), including classified hand poses and corresponding wrist movement trajectories for all valid games in each session. Each session folder contains: A) The MAT file wrist_trajectories.mat contains the extracted wrist movement data. Data are organized as a struct array (game_rep), where each element corresponds to one game. For each game, wrist trajectories are provided separately for both players and segmented into three movement phases: • phase_AB • phase_BC • phase_CD (final phase until the end of the movement) Each phase contains the preprocessed wrist trajectory time series for the respective player. B) The CSV file classified_poses_final.csv contains the behavioral labels for all retained games within each session. Each row corresponds to one game and includes: • game index • classified hand pose for player one • classified hand pose for player two • outcome variables for both players (e.g., win/loss/draw) • Only games that passed all preprocessing, cleaning, and alignment steps are included in these files. The analysis scripts and raw data (e.g., C3D files) are part of an ongoing project and will be made publicly available upon its completion; in the meantime, they can be obtained from the corresponding author upon reasonable request. Found a bug? Please contact Johannes.keck@uni-jena.de