# Observable Patterns Are Not Explanations: A Causal-Geometric Analysis of Latent Reasoning Models

arXiv: https://arxiv.org/abs/2606.12689 · HTML: https://mlsanity.com/paper/2606.12689/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Prompting and chain-of-thought reasoning, one of 17 directions in LLM reasoning.
- Builds on 22 of its 58 references that are in the map.
- First posted on arXiv 10 Jun 2026.

## Placed in

- LLM reasoning / [Prompting and chain-of-thought reasoning](https://mlsanity.com/lr/prompting-and-chain-of-thought-reasoning/)

## Builds on (most cited first, 22 of 22)

- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Toy Models of Superposition](https://mlsanity.com/paper/2209.10652/) (2022)
- [The Linear Representation Hypothesis and the Geometry of Large Language Models](https://mlsanity.com/paper/2311.03658/) (2023)
- [Mechanistic Interpretability for AI Safety -- A Review](https://mlsanity.com/paper/2404.14082/) (2024)
- [Reasoning Models Don't Always Say What They Think](https://mlsanity.com/paper/2505.05410/) (2025)
- [Language Models Represent Space and Time](https://mlsanity.com/paper/2310.02207/) (2023)
- [Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safety](https://mlsanity.com/paper/2507.11473/) (2025)
- [Gradient Descent Happens in a Tiny Subspace](https://mlsanity.com/paper/1812.04754/) (2018)
- [How to use and interpret activation patching](https://mlsanity.com/paper/2404.15255/) (2024)
- [Implicit Chain of Thought Reasoning via Knowledge Distillation](https://mlsanity.com/paper/2311.01460/) (2023)
- [Chain-of-Thought Reasoning In The Wild Is Not Always Faithful](https://mlsanity.com/paper/2503.08679/) (2025)
- [The Hydra Effect: Emergent Self-repair in Language Model Computations](https://mlsanity.com/paper/2307.15771/) (2023)
- [Soft Thinking: Unlocking the Reasoning Potential of LLMs in Continuous Concept Space](https://mlsanity.com/paper/2505.15778/) (2025)
- [When Thinking Fails: The Pitfalls of Reasoning for Instruction-Following in LLMs](https://mlsanity.com/paper/2505.11423/) (2025)
- [Reasoning Theater: Disentangling Model Beliefs from Chain-of-Thought](https://mlsanity.com/paper/2603.05488/) (2026)
- [Do Latent Tokens Think? A Causal and Adversarial Analysis of Chain-of-Continuous-Thought](https://mlsanity.com/paper/2512.21711/) (2025)
- [Can Aha Moments Be Fake? Towards Quantifying Decorative and True Thinking in Chain-of-Thought](https://mlsanity.com/paper/2510.24941/) (2025)
- [The Dead Salmons of AI Interpretability](https://mlsanity.com/paper/2512.18792/) (2025)
- [Do Latent-CoT Models Think Step-by-Step? A Mechanistic Study on Sequential Reasoning Tasks](https://mlsanity.com/paper/2602.00449/) (2026)
- [The Illusion of Superposition? A Principled Analysis of Latent Thinking in Language Models](https://mlsanity.com/paper/2604.06374/) (2026)
- [Revisiting Anisotropy in Language Transformers: The Geometry of Learning Dynamics](https://mlsanity.com/paper/2604.08764/) (2026)
- [What Really Controls Temporal Reasoning in Large Language Models: Tokenisation or Representation of Time?](https://mlsanity.com/paper/2603.19017/) (2026)

## Built on by (most cited first, 4 of 4)

- [The Weight of Silence: A Causal Case for Weights Over the Scratchpad in Latent Chess Reasoning](https://mlsanity.com/paper/2607.20952/) (2026)
- [How Do Language Models Choose Between Context and Memory?](https://mlsanity.com/paper/2609.00753/) (2026)
- [Temporally-Resolved Token Attribution Reveals the Generation Dynamics of Diffusion Language Models](https://mlsanity.com/paper/2610.01177/) (2026)
- [What Matters for Latent Reasoning with Flow Matching](https://mlsanity.com/paper/2610.06666/) (2026)

## Abstract (arXiv)

Latent reasoning models (LRMs) replace explicit chain-of-thought with continuous thoughts. Recent work treats observable latent-state patterns, such as BFS-like frontiers and decodable arithmetic computation, as evidence for internal reasoning mechanisms. Evaluating two LRMs (Coconut and CODI) against controls lacking the proposed recurrence or curriculum, we find these patterns also appear in the controls and do not always causally affect behavior. Causal interventions reveal that latent-thought utilization is not binary but graded, scaling with a thought's causal effect on model behavior. Geometric analyses reveal this effect concentrates in low-rank directions whose step-to-step geometry grows more structured as their behavioral influence increases. Latent thoughts should therefore be treated as hidden computation, not hidden explanation: decodability, attention, or static structure alone cannot establish mechanism. LRM interpretability thus requires matched controls and causal tests.