Deep Hidden Cognition Facilitates Reliable Chain-of-Thought Reasoning

TL;DR

This paper presents a novel method to improve the reliability of Chain of Thought reasoning in large language models by using intrinsic model activations to evaluate and select the most truthful reasoning paths, significantly enhancing accuracy.

Contribution

It introduces a confidence predictor based on attention head activations to calibrate reasoning accuracy and dynamically select the best reasoning paths, advancing CoT reliability.

Findings

Outperforms state-of-the-art CoT methods in accuracy and reliability.

Effective across mathematical, symbolic, and commonsense reasoning tasks.

Applicable to both unimodal and multimodal large models.

Abstract

Chain of Thought (CoT) reasoning has demonstrated remarkable deep reasoning capabilities in both large language models (LLMs) and multimodal large language models (MLLMs). However, its reliability is often undermined by the accumulation of errors in intermediate steps. This paper introduces an novel approach to calibrate the CoT reasoning accuracy by leveraging the model's intrinsic veracity encoding. We discover that specific attention head activations reliably reflect the truthfulness of reasoning steps in CoT. Based on this insight, we train a confidence predictor to evaluate the correctness of each reasoning step using these truthfulness-sensitive activations, dynamically selecting the most plausible reasoning path via beam search. Experimental results demonstrate that our method significantly outperforms the state-of-the-art baselines (e.g., Few-Shot CoT, Self-Consistency, and Self-Evaluation Guided Beam Search) across the mathematical, symbolic, and commonsense reasoning tasks, exhibiting superior accuracy and reliability in both unimodal and multimodal settings. We further validate the approach on large reasoning models, confirming its applicability to specialized reasoning models. Additionally, we explore the role of the model's self-correction ability in CoT reasoning. This work provides a novel reliability improvement path for CoT reasoning with broad application potential.