# Mechanistic Interpretability of Brain-to-Speech Models Across Speech Modes

arXiv: https://arxiv.org/abs/2602.01247 · HTML: https://mlsanity.com/paper/2602.01247/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in Knowledge and representations inside LLMs, one of 13 directions in Mechanistic interpretability.
- Builds on 19 of its 31 references that are in the map.
- First posted on arXiv 1 Feb 2026.

## Placed in

- Mechanistic interpretability / [Knowledge and representations inside LLMs](https://mlsanity.com/mechinterp/knowledge-and-representations-inside-llms/)

## Builds on (most cited first, 19 of 19)

- [Sparse Autoencoders Find Highly Interpretable Features in Language Models](https://mlsanity.com/paper/2309.08600/) (2023)
- [HiFi-GAN: Generative Adversarial Networks for Efficient and High Fidelity Speech Synthesis](https://mlsanity.com/paper/2010.05646/) (2020)
- [Towards Automated Circuit Discovery for Mechanistic Interpretability](https://mlsanity.com/paper/2304.14997/) (2023)
- [Mechanistic Interpretability for AI Safety -- A Review](https://mlsanity.com/paper/2404.14082/) (2024)
- [Towards Best Practices of Activation Patching in Language Models: Metrics and Methods](https://mlsanity.com/paper/2309.16042/) (2023)
- [How to use and interpret activation patching](https://mlsanity.com/paper/2404.15255/) (2024)
- [Localizing Model Behavior with Path Patching](https://mlsanity.com/paper/2304.05969/) (2023)
- [Open Problems in Mechanistic Interpretability](https://mlsanity.com/paper/2501.16496/) (2025)
- [DeWave: Discrete EEG Waves Encoding for Brain Dynamics to Text Translation](https://mlsanity.com/paper/2309.14030/) (2023)
- [EEG-GPT: Exploring Capabilities of Large Language Models for EEG Classification and Interpretation](https://mlsanity.com/paper/2401.18006/) (2024)
- [The Cognitive Revolution in Interpretability: From Explaining Behavior to Interpreting Representations and Algorithms](https://mlsanity.com/paper/2408.05859/) (2024)
- [Beyond Transcription: Mechanistic Interpretability in ASR](https://mlsanity.com/paper/2508.15882/) (2025)
- [NeuroAI for AI Safety](https://mlsanity.com/paper/2411.18526/) (2024)
- [Mechanistic Interpretability Needs Philosophy](https://mlsanity.com/paper/2506.18852/) (2025)
- [Unboxing the Black Box: Mechanistic Interpretability for Algorithmic Understanding of Neural Networks](https://mlsanity.com/paper/2511.19265/) (2025)
- [A Convolutional Framework for Mapping Imagined Auditory MEG into Listened Brain Responses](https://mlsanity.com/paper/2512.03458/) (2025)
- [Interpretable Embeddings of Speech Enhance and Explain Brain Encoding Performance of Audio Models](https://mlsanity.com/paper/2507.16080/) (2025)
- [BrainStratify: Coarse-to-Fine Disentanglement of Intracranial Neural Dynamics](https://mlsanity.com/paper/2505.20480/) (2025)
- [From Black Box to Biomarker: Sparse Autoencoders for Interpreting Speech Models of Parkinson's Disease](https://mlsanity.com/paper/2507.16836/) (2025)

## Built on by (most cited first, 1 of 1)

- [Interpretable MEG Decoding of Perceived Speech: Cortical Sources and the Stimulus Features That Drive Retrieval](https://mlsanity.com/paper/2608.01481/) (2026)

## Abstract (arXiv)

Brain-to-speech decoding models demonstrate robust performance in vocalized, mimed, and imagined speech; yet, the fundamental mechanisms via which these models capture and transmit information across different speech modalities are less explored. In this work, we use mechanistic interpretability to causally investigate the internal representations of a neural speech decoder. We perform cross-mode activation patching of internal activations across speech modes, and use tri-modal interpolation to examine whether speech representations vary discretely or continuously. We use coarse-to-fine causal tracing and causal scrubbing to find localized causal structure, allowing us to find internal subspaces that are sufficient for cross-mode transfer. In order to determine how finely distributed these effects are within layers, we perform neuron-level activation patching. We discover that small but not distributed subsets of neurons, rather than isolated units, affect the cross-mode transfer. Our results show that speech modes lie on a shared continuous causal manifold, and cross-mode transfer is mediated by compact, layer-specific subspaces rather than diffuse activity. Together, our findings give a causal explanation for how speech modality information is organized and used in brain-to-speech decoding models, revealing hierarchical and direction-dependent representational structure across speech modes.