# Unveiling the Reasoning Process of Large Language Models

arXiv: https://arxiv.org/abs/2603.29735 · HTML: https://mlsanity.com/paper/2603.29735/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning.
- Builds on 10 of its 37 references that are in the map.
- First posted on arXiv 31 Mar 2026.

## Placed in

- LLM reasoning / [Reasoning benchmarks and evaluation](https://mlsanity.com/lr/reasoning-benchmarks-and-evaluation/)

## Builds on (most cited first, 10 of 10)

- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [Gemma 3 Technical Report](https://mlsanity.com/paper/2503.19786/) (2025)
- [DeepSeek-V3 Technical Report](https://mlsanity.com/paper/2412.19437/) (2024)
- [The Remarkable Robustness of LLMs: Stages of Inference?](https://mlsanity.com/paper/2406.19384/) (2024)
- [Weight-sparse transformers have interpretable circuits](https://mlsanity.com/paper/2511.13653/) (2025)
- [Language Model Circuits Are Sparse in the Neuron Basis](https://mlsanity.com/paper/2601.22594/) (2026)
- [Rethinking Evaluation of Sparse Autoencoders through the Representation of Polysemous Words](https://mlsanity.com/paper/2501.06254/) (2025)
- [A Brain-like Synergistic Core in LLMs Drives Behaviour and Learning](https://mlsanity.com/paper/2601.06851/) (2026)

## Abstract (arXiv)

Large language models often reason beyond surface tokens, but the internal stage at which token-level information becomes abstract relational structure remains unclear. We investigate this question by analyzing how attention heads and layers transform information during autoregressive reasoning. Across mathematical and symbolic reasoning tasks, we observe a consistent layer-wise division of labor: outer layers mainly preserve and route input-related features, whereas middle layers reorganize them into more transferable rule-level representations. This interpretation is supported by representation geometry: middle-layer states occupy lower-dimensional manifolds and show stronger alignment across disjoint vocabularies that instantiate the same symbolic rules. It is further supported by causal interventions: removing middle-layer components identified by our interaction-based criterion produces substantially larger downstream changes and accuracy drops than removing components from other regions or at random. Together, these results suggest that abstract reasoning is not uniformly distributed across transformer layers, but is preferentially formed in a middle-layer computation stage that converts token-level information into reusable relational structure.