Metraipolis › archive › dc_2e2c94f3e79d
Mechanistic Interpretability Reveals Shared Causal Subspaces in Brain-to-Speech Decoders — arXiv:2609.31992
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.31992
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T22:36:11.584Z
- SHA-256 of the archived text
1644e6d137fd33876204715846aeda5054852019bd225d928fe9dbc1fbdc1a49- Stable URL of this page
https://metraipolis.com/documento/dc_2e2c94f3e79d
How to cite
Mechanistic Interpretability Reveals Shared Causal Subspaces in Brain-to-Speech Decoders — arXiv:2609.31992 — arXiv (en), https://arxiv.org/abs/2609.31992 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_2e2c94f3e79d (Metraipolis archive dc_2e2c94f3e79d, sha256 1644e6d137fd33876204715846aeda5054852019bd225d928fe9dbc1fbdc1a49)
BibTeX
@misc{dc2e2c94f3e79d,
title = {Mechanistic Interpretability Reveals Shared Causal Subspaces in Brain-to-Speech Decoders — arXiv:2609.31992},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.31992},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_2e2c94f3e79d; sha256 1644e6d137fd33876204715846aeda5054852019bd225d928fe9dbc1fbdc1a49},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.31992. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Decoding covert speech, such as mimed or imagined, from brain activity is harder than decoding vocalized speech. Cross-modal transfer, where information from one speech form helps decode another, is a promising remedy; yet how a decoder internally represents and processes brain activity from different speech forms remains unclear. In this work, we ask: which internal neurons of a decoder carry cross-modal information, and are these neurons shared across different speech forms? To answer these questions, we leverage mechanistic interpretability, using recordings of the same sentences in vocalized, mimed, and imagined input pairs for activation patching. We insert the decoder's internal activity for a sentence in one condition into its processing of the same sentence in another and measure the change in decoding accuracy. We find that no single neuron drives this benefit; instead, it arises from small groups of neurons, with vocalized speech as the most useful source. These groups are largely condition-specific in the early stage of the decoder but overlap in the later stage. These findings point toward more data-efficient covert speech decoders through training objectives that encourage shared later-stage representations learned mainly from vocalized data. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.LG). Categorie arXiv: Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2609.31992 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-07T22:14:27.570Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).