# Are Stated Reasoning Steps Causally Load-Bearing?

arXiv: https://arxiv.org/abs/2609.27038 · HTML: https://mlsanity.com/paper/2609.27038/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Causal reasoning in LLMs, one of 17 directions in LLM reasoning.
- Builds on 8 of its 12 references that are in the map.
- First posted on arXiv 22 Sep 2026.

## Placed in

- LLM reasoning / [Causal reasoning in LLMs](https://mlsanity.com/lr/causal-reasoning-in-llms/)

## Builds on (most cited first, 8 of 8)

- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Measuring Faithfulness in Chain-of-Thought Reasoning](https://mlsanity.com/paper/2307.13702/) (2023)
- [Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safety](https://mlsanity.com/paper/2507.11473/) (2025)
- [How does Chain of Thought Think? Mechanistic Interpretability of Chain-of-Thought Reasoning with Sparse Autoencoding](https://mlsanity.com/paper/2507.22928/) (2025)
- [Breaking the Chain: A Causal Analysis of LLM Faithfulness to Intermediate Structures](https://mlsanity.com/paper/2603.16475/) (2026)
- [When Chain-of-Thought Fails, the Solution Hides in the Hidden States](https://mlsanity.com/paper/2604.23351/) (2026)
- [Bypassing the Rationale: Causal Auditing of Implicit Reasoning in Language Models](https://mlsanity.com/paper/2602.03994/) (2026)
- [When Reasoning Traces Become Performative: Step-Level Evidence that Chain-of-Thought Is an Imperfect Oversight Channel](https://mlsanity.com/paper/2605.11746/) (2026)

## Abstract (arXiv)

Chain-of-thought (CoT) monitoring assumes that the reasoning a model writes reflects the computation that directly produces its answer. Previous faithfulness metrics have been predominantly behavioral, as they simply edit the reasoning text and observe the resulting answer. However, our methodology aims to measure faithfulness causally at the activation level, specifically on self-generated reasoning. Unlike previous causal audits, which measure degradation, our interventions carry a known predicted target. In this way, each patch should switch the answer to a specific counterfactual entity derivable by construction. Specifically, we use synthetic multi-hop lookup tasks (2-6 hops). We patch the residual stream at the token span where the model states each intermediate step with the corresponding activations from a counterfactual run. For Qwen3-4B, 76.9% +/- 2.8% of stated steps are causally load-bearing (CLB) at the most responsive mid-network layer (random-position null: 11.3%; patching the underlying prompt fact: 83%, so stated steps carry approximately 96% of the achievable effect). Moreover, the standard behavioral test on the same items yields 88.2%, which overstates causal faithfulness by 11.4 percentage points (item-matched; 111:14 discordant pairs, p < 1e-15) and, for the easiest items, by up to 20 percentage points. This gap also has a clear capability dimension. Qwen3-1.7B is far less causally faithful overall (54.8%), with its faithfulness collapsing as reasoning depth increases (68% at 2 hops to 30% at 6), while Qwen3-4B remains relatively flat. Although stated reasoning can be causally meaningful, standard behavioral tests tend to overestimate its causal faithfulness, particularly on easier examples where model reasoning appears most fluent.