Finding RELIEF: Shaping Reasoning Behavior without Reasoning Supervision via Belief Engineering

TL;DR

This paper introduces RELIEF, a novel method to shape large reasoning models' behavior by aligning their internal reasoning beliefs with target traits, without relying on costly reasoning trace supervision, leading to improved efficiency and faithfulness.

Contribution

The paper reveals that LRMs have latent reasoning beliefs that can be manipulated through simple logit probing and proposes RELIEF, a belief engineering framework that improves reasoning behavior without supervision.

Findings

RELIEF matches or outperforms supervised baselines in efficiency and faithfulness.

Shifting reasoning beliefs effectively alters model behavior.

RELIEF requires lower training costs than traditional methods.

Abstract

Large reasoning models (LRMs) have achieved remarkable success in complex problem-solving, yet they often suffer from computational redundancy or reasoning unfaithfulness. Current methods for shaping LRM behavior typically rely on reinforcement learning or fine-tuning with gold-standard reasoning traces, a paradigm that is both computationally expensive and difficult to scale. In this paper, we reveal that LRMs possess latent \textit{reasoning beliefs} that internally track their own reasoning traits, which can be captured through simple logit probing. Building upon this insight, we propose Reasoning Belief Engineering (RELIEF), a simple yet effective framework that shapes LRM behavior by aligning the model's self-concept with a target belief blueprint. Crucially, RELIEF completely bypasses the need for reasoning-trace supervision. It internalizes desired traits by fine-tuning on synthesized, self-reflective question-answering pairs that affirm the target belief. Extensive experiments on efficiency and faithfulness tasks demonstrate that RELIEF matches or outperforms behavior-supervised and preference-based baselines while requiring lower training costs. Further analysis validates that shifting a model's reasoning belief effectively shapes its actual behavior.