Pause and Reflect: Conformal Aggregation for Chain-of-Thought Reasoning

2026-05-13

AI Summary

This work addresses the problem of high-confidence erroneous answers in chain-of-thought reasoning by proposing the first conformal aggregation method with finite-sample guarantees. The approach aggregates multiple reasoning chains through weighted scoring at inference time and dynamically decides whether to abstain from answering to control the confidence error rate. Requiring no model retraining, it integrates self-consistency sampling with inference-time calibration, revealing that score separability is a key condition for improving selection accuracy. Notably, the method can predict accuracy gains using only calibration data. On GSM8K, it achieves a 90.1% selection accuracy with an abstention rate below 5%, substantially outperforming majority voting baselines (82%) and consistently meeting target error rates across diverse models and benchmarks.

Abstract

Chain-of-thought (CoT) reasoning with self-consistency improves performance by aggregating multiple sampled reasoning paths. In this setting, correctness is no longer tied to a single reasoning trace but to the aggregation rule over a pool of candidate paths, making aggregation uncertainty the central challenge. This issue is critical where confidently incorrect answers are far more costly than abstentions. We introduce a conformal procedure for CoT reasoning that directly addresses aggregation uncertainty. Our approach replaces majority voting with weighted score aggregation over reasoning paths and calibrates an abstention rule using conformal risk control. This approach leads to finite-sample guarantees on the confident-error rate--the probability that the system answers and is wrong. We further identify score separability as the key condition under which abstention provably improves selective accuracy, and derive closed-form expressions that predict accuracy gains from calibration data alone. The method is fully inference-time, and requires no retraining. Across four benchmarks, four open-source models, and three score classes, realized confident-error rates are consistent with the prescribed targets up to calibration-split and test-set variability. Our method achieves $90.1\%$ selective accuracy on GSM8K by abstaining on less than $5\%$ of problems, compared with $82\%$ accuracy under majority-voting baseline.

Problem

• Chain-of-Thought reasoning
• aggregation uncertainty
• conformal prediction
• abstention
• confident-error rate

Innovation

• Conformal Aggregation
• Chain-of-Thought Reasoning
• Abstention Mechanism
• Confident-Error Rate
• Score Separability