Amplified Does Not Mean Predictive: Reasoning Behaviors in Thinking Models

2026-08-13

AI Summary

This study addresses the misalignment between behavioral amplification and accuracy improvement in reasoning training by proposing the Behavioral Lift metric and establishing a cross-modal reasoning behavior taxonomy. Leveraging over 10,000 annotated trajectories and multi-model empirical analysis, this work quantitatively reveals, for the first time, the discrepancy between training preferences and actual effectiveness. Findings indicate that high-value behaviors such as confidence calibration remain under-reinforced, whereas self-correction is disproportionately amplified. By elucidating the necessity of process-level optimization, this research introduces novel evaluation dimensions and optimization directions for enhancing the reasoning quality of large language models.

Abstract

Which reasoning behaviors are associated with correct answers in reasoning models, and does reasoning-oriented training amplify those behaviors? This distinction is important because reasoning-oriented training can make traces look more deliberative without amplifying the behaviors most tied to model correctness. We quantify this mismatch with Behavioral Lift, a metric that measures how much correctness changes when a behavior is present versus absent in a model's reasoning trace. Across 15 models and 6 benchmarks spanning text-only and vision-language reasoning, we annotate 15,282 traces with a taxonomy whose core behaviors are defined for both LLM and VLM traces. We find evidence for an Amplification-Lift Gap, in which thinking models strongly amplify self-correction, hypothesis testing, and uncertainty acknowledgment, while the highest-lift behaviors are confidence calibration, knowledge alignment, and self-awareness. Confidence calibration is among the strongest positive signals of correctness in both modalities, yet is barely amplified; uncertainty acknowledgment is amplified by 3--7$\times$, yet is weakly or negatively associated with correctness. We find that reasoning-oriented training does not preferentially amplify the highest-Lift behaviors, motivating process-level objectives that reward calibrated and grounded reasoning rather than surface form alone.

Problem

• Reasoning Behaviors
• Amplification-Lift Gap
• Thinking Models
• Behavioral Lift
• Correctness

Innovation

• Behavioral Lift
• Amplification-Lift Gap
• Confidence Calibration
• Process-level Objectives
• Reasoning Behaviors