Research map / Probabilistic and scientific modeling / Causal discovery

Improving Causal Explanations

NeurIPS 2026 · Probabilistic and scientific modeling › Causal discovery · approach: Counterfactual identification and bounds

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes a causal framework for explaining model predictions to address the disagreement problem among attribution methods, drawing on cognitive science, law, and philosophy (abstract only; details unavailable).

Problem

Explanation methods for model predictions produce conflicting attributions, and there is no principled causal criterion to determine which attribution is correct (the disagreement problem), which matters in high-stakes settings.

Approach

Grounds explanation in a principled causal framework drawn from cognitive science, law, and philosophy, as an alternative to ad hoc attribution methods; the abstract is truncated, so the specific formalism is not stated.

Limitations

• Only the abstract was available; full method and experiments unknown

Closest papers by meaning

• A New Approach to Backtracking Counterfactual Explanations: A Unified Causal Framework for Efficient Model Interpretability (ICML 2025 · similarity 0.89)
• Everything, Everywhere, All at Once: Is Mechanistic Interpretability Identifiable? (ICLR 2025 · similarity 0.87)
• Reasoning About Actual Causes in Nondeterministic Domains (AAAI 2025 · similarity 0.87)
• Dissenting Explanations: Leveraging Disagreement to Reduce Model Overreliance (AAAI 2024 · similarity 0.87)
• Generating Likely Counterfactuals Using Sum-Product Networks (ICLR 2025 · similarity 0.87)
• From Probability to Counterfactuals: the Increasing Complexity of Satisfiability in Pearl's Causal Hierarchy (ICLR 2025 · similarity 0.87)