# Reasoning or Fluency? Dissecting Probabilistic Confidence in Best-of-N Selection

arXiv: https://arxiv.org/abs/2601.13735 · HTML: https://mlsanity.com/paper/2601.13735/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 18 of its 27 references that are in the map.
- First posted on arXiv 20 Jan 2026.

## Builds on (most cited first, 18 of 18)

- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Large Language Models are Zero-Shot Reasoners](https://mlsanity.com/paper/2205.11916/) (2022)
- [Let's Verify Step by Step](https://mlsanity.com/paper/2305.20050/) (2023)
- [HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering](https://mlsanity.com/paper/1809.09600/) (2018)
- [GPQA: A Graduate-Level Google-Proof Q&A Benchmark](https://mlsanity.com/paper/2311.12022/) (2023)
- [Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters](https://mlsanity.com/paper/2408.03314/) (2024)
- [Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone](https://mlsanity.com/paper/2404.14219/) (2024)
- [CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledge](https://mlsanity.com/paper/1811.00937/) (2018)
- [CLEVR: A Diagnostic Dataset for Compositional Language and Elementary Visual Reasoning](https://mlsanity.com/paper/1612.06890/) (2016)
- [Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Prompting](https://mlsanity.com/paper/2305.04388/) (2023)
- [Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human Annotations](https://mlsanity.com/paper/2312.08935/) (2023)
- [Scalable Best-of-N Selection for Large Language Models via Self-Certainty](https://mlsanity.com/paper/2502.18581/) (2025)
- [Thought Anchors: Which LLM Reasoning Steps Matter?](https://mlsanity.com/paper/2506.19143/) (2025)
- [A Comprehensive Survey of Small Language Models in the Era of Large Language Models: Techniques, Enhancements, Applications, Collaboration with LLMs, and Trustworthiness](https://mlsanity.com/paper/2411.03350/) (2024)
- [AR-LSAT: Investigating Analytical Reasoning of Text](https://mlsanity.com/paper/2104.06598/) (2021)
- [PiCSAR: Probabilistic Confidence Selection And Ranking for Reasoning Chains](https://mlsanity.com/paper/2508.21787/) (2025)
- [Best-of-Majority: Minimax-Optimal Strategy for Pass@k Inference Scaling](https://mlsanity.com/paper/2510.03199/) (2025)
- [seqBench: A Tunable Benchmark to Quantify Sequential Reasoning Limits of LLMs](https://mlsanity.com/paper/2509.16866/) (2025)

## Built on by (most cited first, 2 of 2)

- [Reproducing and Stress-Testing Two Approaches to LLM Reasoning Reliability: Test-Time Probability Aggregation and Logic-Representation Editing](https://mlsanity.com/paper/2608.08514/) (2026)
- [Certainty Is Not Just Correctness: Rethinking Token-Level Certainty in LLM Reasoning](https://mlsanity.com/paper/2610.00296/) (2026)

## Abstract (arXiv)

Probabilistic confidence metrics are increasingly adopted as proxies for reasoning quality in Best-of-N selection, under the assumption that higher confidence reflects higher reasoning fidelity. In this work, we challenge this assumption by investigating whether these metrics truly capture inter-step causal dependencies necessary for valid reasoning. We introduce three classes of inter-step causality perturbations that systematically disrupt dependencies between reasoning steps while preserving local fluency. Surprisingly, across diverse model families and reasoning benchmarks, we find that selection accuracy degrades only marginally under these disruptions. Even severe interventions, such as applying hard attention masks that directly prevent the model from attending to prior reasoning steps, do not substantially reduce selection performance. These findings provide strong evidence that current probabilistic metrics are largely insensitive to logical structure, and primarily capture surface-level fluency or in-distribution priors instead. Motivated by this gap, we propose a contrastive causality metric that explicitly isolates inter-step causal dependencies, and demonstrate that it yields more faithful output selection than existing probability-based approaches.