# Sparse Autoencoders Map Brain-LLM Alignment onto Cortical Semantic Topography

arXiv: https://arxiv.org/abs/2605.23035 · HTML: https://mlsanity.com/paper/2605.23035/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Placed in LLM preference optimization methods, one of 10 directions in LLM alignment / RLHF / preference optimisation.
- Builds on 21 of its 68 references that are in the map.
- First posted on arXiv 21 May 2026.

## Placed in

- LLM alignment / RLHF / preference optimisation / [LLM preference optimization methods](https://mlsanity.com/alignment/llm-preference-optimization-methods/)
- Mechanistic interpretability / [SAE features and their applications](https://mlsanity.com/mechinterp/sae-features-and-their-applications/)

## Builds on (most cited first, 21 of 21)

- [Similarity of Neural Network Representations Revisited](https://mlsanity.com/paper/1905.00414/) (2019)
- [Sparse Autoencoders Find Highly Interpretable Features in Language Models](https://mlsanity.com/paper/2309.08600/) (2023)
- [Explaining by Removing: A Unified Framework for Model Explanation](https://mlsanity.com/paper/2011.14878/) (2020)
- [Finding Alignments Between Interpretable Causal Variables and Distributed Neural Representations](https://mlsanity.com/paper/2303.02536/) (2023)
- [A is for Absorption: Studying Feature Splitting and Absorption in Sparse Autoencoders](https://mlsanity.com/paper/2409.14507/) (2024)
- [Interpreting and improving natural-language processing (in machines) with natural language-processing (in the brain)](https://mlsanity.com/paper/1905.11833/) (2019)
- [Towards Principled Evaluations of Sparse Autoencoders for Interpretability and Control](https://mlsanity.com/paper/2405.08366/) (2024)
- [RAVEL: Evaluating Interpretability Methods on Disentangling Language Model Representations](https://mlsanity.com/paper/2402.17700/) (2024)
- [Testing the Predictions of Surprisal Theory in 11 Languages](https://mlsanity.com/paper/2307.03667/) (2023)
- [Language Models Learn Rare Phenomena from Less Rare Phenomena: The Case of the Missing AANNs](https://mlsanity.com/paper/2403.19827/) (2024)
- [Recurrence is required to capture the representational dynamics of the human visual system](https://mlsanity.com/paper/1903.05946/) (2019)
- [The Geometry of Concepts: Sparse Autoencoder Feature Structure](https://mlsanity.com/paper/2410.19750/) (2024)
- [Disentangling Syntax and Semantics in the Brain with Deep Networks](https://mlsanity.com/paper/2103.01620/) (2021)
- [Joint processing of linguistic properties in brains and language models](https://mlsanity.com/paper/2212.08094/) (2022)
- [Dissociating language and thought in large language models](https://mlsanity.com/paper/2301.06627/) (2023)
- [Neural Language Models are not Born Equal to Fit Brain Data, but Training Helps](https://mlsanity.com/paper/2207.03380/) (2022)
- [The Algonauts Project 2021 Challenge: How the Human Brain Makes Sense of a World in Motion](https://mlsanity.com/paper/2104.13714/) (2021)
- [Neural Brain: A Neuroscience-inspired Framework for Embodied Agents](https://mlsanity.com/paper/2505.07634/) (2025)
- [AudioSAE: Towards Understanding of Audio-Processing Models with Sparse AutoEncoders](https://mlsanity.com/paper/2602.05027/) (2026)
- [BrainExplore: Large-Scale Discovery of Interpretable Visual Representations in the Human Brain](https://mlsanity.com/paper/2512.08560/) (2025)
- [Brain-aligning of semantic vectors improves neural decoding of visual stimuli](https://mlsanity.com/paper/2403.15176/) (2024)

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

- [Interpreting Brain Responses to Language with Sparse Features from Language Models](https://mlsanity.com/paper/2606.06857/) (2026)
- [The Platonic brain bridge hypothesis: human brain networks as an architectural prior for multimodal large language models](https://mlsanity.com/paper/2609.10947/) (2026)

## Abstract (arXiv)

Intermediate layers of large language models (LLMs) best predict human brain responses to language, one of the most robust findings in computational neurolinguistics, yet why remains mechanistically unexplained. We address this gap by bridging sparse autoencoders (SAEs) from mechanistic interpretability with neural encoding models, decomposing GPT-2 XL and Llama-3.1-8B into 16K-32K interpretable features per layer. A human-validated taxonomy (kappa ≥ 0.74) reveals that semantic features alone recover 94% of peak encoding performance (r=0.285), substantially exceeding variance-matched baselines (p<0.001, d=1.31). Beyond this aggregate dominance, we test a novel cortical topography prediction: five semantic subcategories derived a priori from three independent neuroscience programs should map onto distinct brain regions. A formal convergence test confirms this alignment (Spearman ρ=0.72, p<0.001; hypergeometric p=0.007), demonstrating that SAE-discovered features recapitulate known cortical semantic organization at a granularity inaccessible to prior methods. SAE features further predict human reading times beyond lexical controls (DeltalogLik=38.4, p<0.001), and an exploratory prediction-error analysis provides preliminary evidence that the brain additionally encodes unexpected semantic content. Results generalize across English, Chinese, and French.