Causal Graph Recovery in Neuroimaging through Answer Set Programming

ICLR 2026 Conference Submission20673 Authors

Keywords: answer set programming, causal learning, neuroscience, graph theory

TL;DR: We propose Real-world noisy RASL (RnR), an ASP-based method that models undersampling in fMRI to recover more accurate and robust causal brain networks, improving F1-scores by 12% over existing approaches.

Abstract: Learning directed causal graphs from time-series data poses significant challenges, especially in fMRI where slow sampling rate obscures fast neural interactions. This temporal mismatch leads to undersampling, which can make multiple graphs equally plausible. We address this problem by explicitly modeling undersampling effects when recovering causal graphs. Our approach employs answer set programming (ASP) to enforce domain-specific constraints and optimize soft observational constraints, thereby identifying a Markov equivalence class for the resulting graph solutions. By customizing an ASP solver to collect multiple near-optimal solutions, we obtain not only the single best-fitting graph but an equivalence class of high-scoring graphs for expert consideration. This method, called Real-world noisy RASL (RnR), can also act as a meta-solver: it refines the output of other causal discovery algorithms by accounting for undersampling biases. In simulations and empirical brain network data, RnR produces more accurate causal graphs than state-of-the-art methods, improving F1-scores by an average of 12\% by reducing false connections. We demonstrate that RnR is robust to varying undersampling rates – maintaining high precision and recall even as sampling becomes more sparse – whereas baseline methods degrade significantly. Our results suggest that incorporating undersampling-aware constraints via ASP yields more reliable and interpretable brain connectivity estimates from fMRI time series, closing the gap between neural dynamics and observational data.

Primary Area: applications to neuroscience & cognitive science

Submission Number: 20673