Reproducibility package: Method-conditioned residual structure after deterministic forcing removal in a single-observer perceptual-dynamical task

This repository contains all raw data and analysis code required to reproduce the quantitative findings, figures, and tables reported in the manuscript: "Method-conditioned residual structure after deterministic forcing removal in a single-observer perceptual-dynamical task" Yosuke Yoshimura (2026) Contents: - data/: Six canonical JSON observation logs (one per experimental condition: A–E plus 30-minute stationarity session), and condition_metadata.csv (session–condition mapping, frame counts, durations). - analysis/: Primary analysis pipeline (exp18_analysis.py) and six figure generation scripts (make_figure1.py through make_figure6.py). All scripts expect to be run from the repro_package/ directory. - figures/: Submission-quality PNG figures (Figures 1–6) and source/ subdirectory containing per-figure source CSV files (Figures 1–6). - docs/: Manuscript draft (paper2-draft-v0.1.md), claim audit (CLAIM_AUDIT_PAPER2.md), and reproducibility guide (REPRODUCIBILITY_README.md). All 32 quantitative claims in the manuscript are mapped to their data source, script function, and recomputed value in CLAIM_AUDIT_PAPER2.md. Two numerical mismatches were identified and corrected in the manuscript (ΔAIC minimum, z_shuffle range). CRITICAL: Always use xy_beat_phase (not beat_phase) for Q-removal and beat-phase binning. See README and figure_generation_audit.md for details. Software: Python ≥ 3.11, numpy ≥ 2.4, scipy ≥ 1.17, matplotlib ≥ 3.10. No proprietary software required.