Reasoning Models Are Accurate but Unsound on Identification
On the map: LLM reasoning → Causal reasoning in LLMs
Abstract (arXiv)
A reasoning model asked whether a causal effect is recoverable from observational data can fail in two ways: it refuses an identifiable query or answers a nonidentifiable one. The latter is more consequential, as no observational data can validate the claimed formula. Measuring this failure requires queries that are provably non-identifiable, which prior evaluations lack, and grading that accepts correct formulas in any equivalent form, which string matching cannot provide. We build CERTID, a formal identification pipeline that addresses both limitations. CERTID uses the sound and complete causal identification algorithm ID to certify whether an effect is identifiable from a given graph and query, and verifies returned formulas against structural causal models whose interventional distributions are known exactly. CERTID further develops theoretical results to mitigate structural leakage, repair non-identifiable queries, and establish grading guarantees. We evaluate three frontier reasoning models (Gemini Flash, Gemini Pro, and GPT5.5) on 1,200 certified instances spanning 4 to 50 vertices. Accuracy proves a poor proxy for soundness: on identical instances, the false-claim rate on non-identifiable queries varies by seventeen-fold across models. We also find that models decide identifiability with 97-100% accuracy on graphs generated after the strongest model's training snapshot. Instances, the certification procedure, the verifier, and per-instance records are available at https://anonymous.4open.science/r/certid-D718.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Causal reasoning in LLMs, one of 17 directions in LLM reasoning. First posted on arXiv 2 Oct 2026.
Builds on
No reference list yet.
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2026CausalReasoningBenchmark: A Real-World Benchmark for Disentangled Evaluation of Causal Identification and Estimation
- 2026CausalArena: Benchmarking Causal Discovery in the Foundation Model Era
- 2026CausalDS: Benchmarking Causal Reasoning in Data-Science Agents
- 2026Symmetries and Causality: Causal Effect Identification Beyond IID Data
- 2026Leveraging Large Language Models for Causal Discovery: a Constraint-based, Argumentation-driven Approach
- 2025Causality can systematically address the monsters under the bench(marks)
- 2024Context-Aware Reasoning On Parametric Knowledge for Inferring Causal Variables
- 2026CausalT5k: Diagnosing Refusal and Failure Modes in Trustworthy Causal Reasoning Across Causal Rungs