Supplementary Data for "Bravura: an open immersive virtual reality paradigm for human reactive and proactive aggression "

Data for the Bravura analysis pipeline. Paths below are relative to the data root. Raw data raw/ Original data from AggressionProjectDataShare. MatlabEvents/ — Per-subject BioPac event files (.mat, one per subject across both cohorts) psap/ — PSAP task data (.mat, one per Cohort A subject) additional.xlsx — Additional participant metadata (PSAP button presses, presence questionnaire sum score) Markers legend.xlsx — BioPac event marker definitions shared/ Files shared across both cohorts. AggressionScript.xlsx — Preprogrammed opponent behaviour (shocks, wins per trial) trial_events.csv — Parsed per-trial data from BioPac events (duration, choice, outcome) Per-cohort data cohort_a/ Cohort A (N=126, 114 after exclusion, 112 for physiology). Task and questionnaire data aggroPerformance.xlsx — Per-trial shock decisions, latencies, choices (126 subjects x 30 trials) latPerformance.xlsx — Clean latency data (124 subjects) beliefs.xlsx — Belief ratings per opponent (0–10 scale) subject_ids.csv — Subject ID list outliers.mat — 12 excluded subject IDs outliers.txt — Outlier subject IDs (text format) shock_long.csv — Shock decisions in long format (consumed by brms_shocks, brms_shocks_overview) psap_ilr.csv — PSAP ILR-transformed compositions (consumed by brms_psap) shock_latency_long.csv — Shock latencies in long format (consumed by brms_latency) Physiology physio/physPerformance.xlsx — Physiological features (HR, HRV) across experimental blocks physio/baseline_hr.csv — Baseline heart rate per subject (consumed by brms_baseline_hr) physio/delta_hr_long.csv — Delta-HR in long format (consumed by brms_delta_hr) physio/physio_cardiac_long.csv — Cardiac features in long format (consumed by brms_physio_cardiac) physio/hrv_pca_loadings.csv — Varimax-rotated PCA loadings for HRV features Clustering clustering/clusters.csv — Cluster assignments per subject clustering/mc_k-means_3.npz — Consensus clustering label counts (1000 MC draws) clustering/mc_k-means_3_stats.txt — Consensus clustering summary statistics clustering/grid_k_solver.csv — k x solver silhouette/gap grid clustering/ablate_k.csv — Silhouette and gap statistics per k clustering/ablate_metric.csv — Silhouette when ablating individual features VBA model fitting Produced by pipeline/vba/ (MATLAB). See "VBA model outputs" below for details. vba/vba_input.xlsx — Transposed input matrix for MATLAB VBA vba/models/ — Per-grid-point model fits (256 .mat files) vba/bma/ — Bayesian model averaging (bma_results.mat, free_energy.mat) vba/vba_posteriors.mat — Posterior means (muPhi) and covariances (SigmaPhi) per subject VBA-derived summaries vba/coefficients.csv — BMA-averaged model coefficients [Kr1, Krc, Kp, Kwc] per subject vba/predictions.csv — Model-predicted P(shock) per trial vba/decisions.csv — Actual binary shock decisions (ground truth) vba/fit_metrics.csv — R2, accuracy, balanced accuracy, log evidence per subject vba/free_energy_matrix.csv — Free energy matrix (subjects x models) Simulation recovery vba/corr_preds.mat — Per-subject parameter correlation matrices and covariance predictions vba/cov_stats.mat — Condition numbers and determinants of parameter covariance matrices cohort_b/ Cohort B (N=44, 37 after exclusion). Same structure as cohort_a/ with these differences: Beliefs on 0–5 scale (rescaled to 0–10 in pipeline) No simulation recovery files