Reproduce “metasignal: A Python Package for Comprehensive Metacognitive Analysis and Decision-Making” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Saurabh Ranjan, Mukesh Makwana, Konstantina Sokratous, Brian Odegaard
METASIGNAL introduces an open-source Python package that unifies 17 signal detection theory (SDT) and metacognitive measures, previously fragmented across disparate implementations. The package was extensively validated against the Rahnev (2025) benchmark, achieving high numerical fidelity, especially for non-model-based metrics.
Key insights
- A unified Python implementation can achieve high numerical fidelity with existing benchmarked metacognitive measures, promoting standardization and reproducibility in the field.
- Strategic software design, emphasizing a pure-Python core and isolated complex dependencies, enhances accessibility and efficiency for diverse research needs without unnecessary overhead.
- While different optimizers can lead to minor numerical variations in model-based measures (e.g., `meta-d'`) for individual participants, core analyses and profiles maintain high correlation and statistical agreement across implementations.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.