SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2603.05488 · ICML · 2026

Reasoning Theater: Disentangling Model Beliefs from Chain-of-Thought

Atticus Geiger, Jack Merullo, Eric Bigelow, Annabel Ma, Siddharth Boppana, Max Loeffler, Raphael Sarfati, Owen Lewis

arXiv · PDF · Open in the Atlas

Code that ran

We have not lifted any functions out of this paper's repositories yet, so there is nothing we have run. The repositories linked to it are listed below.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

We provide evidence of performative chain-ofthought (CoT) in reasoning models, where a model becomes strongly confident in its final answer, but continues generating tokens without revealing its internal belief. Our analysis compares activation probing, early forced answering, and a CoT monitor across two large models (DeepSeek-R1 671B & GPT-OSS 120B) and find task difficulty-specific differences: The model's final answer is decodable from activations far earlier in CoT than a monitor is able to say, especially for easy recall-based MMLU questions. We contrast this with genuine reasoning in difficult multihop GPQA-Diamond questions. Despite this, inflection points (e.g., backtracking, 'aha' moments) occur almost exclusively in responses where probes show large belief shifts, suggesting these behaviors track genuine uncertainty rather than learned "reasoning theater." Finally, probeguided early exit reduces tokens by up to 80% on MMLU and 30% on GPQA-Diamond with similar accuracy, positioning attention probing as an efficient tool for detecting performative reasoning and enabling adaptive computation. 12 * Equal contribution

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2603.05488")
get_code_for_paper("2603.05488")
have("2603.05488")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2603.05488.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2603.05488.svg)](https://syntology.ai/paper/2603.05488)

Connect an agent — have() is free.