Hostname: page-component-76d6cb85b7-pn7tm Total loading time: 0 Render date: 2026-07-11T12:51:22.980Z Has data issue: false hasContentIssue false
  • English
  • Français

Keeping it real: Language models implement algorithms to solve linguistic tasks

Published online by Cambridge University Press:  01 July 2026

Raphaël Millière Open the ORCID record for Raphaël Millière [Opens in a new window]
Raphaël Millière*
Affiliation:
Faculty of Philosophy, University of Oxford, UK raphael.milliere@philosophy.ox.ac.uk
*
*Corresponding author.
Get access Rights & Permissions [Opens in a new window]

Abstract

Futrell & Mahowald argue that language models (LMs) discover linguistic structure as “real patterns.” I contend this framing underplays what mechanistic interpretability uncovers: LMs implement specific algorithms to solve linguistic tasks. Under the framework of causal abstraction, we can rigorously test whether LMs converge on algorithms posited by linguistic theory, which further supports the authors’ conciliatory proposal.

Information

Type
Open Peer Commentary
Information
Behavioral and Brain Sciences , Volume 49 , 2026 , e216
Check for updates
Copyright
© The Author(s), 2026. Published by Cambridge University Press

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

Article purchase

Temporarily unavailable

References

Geiger, A., Harding, J., & Icard, T. (2025). How Causal Abstraction Underpins Computational Explanation (No. arXiv:2508.11214). arXiv. https://doi.org/10.48550/arXiv.2508.11214 CrossRefGoogle Scholar
Geiger, A., Ibeling, D., Zur, A., Chaudhary, M., Chauhan, S., Huang, J., Arora, A., Wu, Z., Goodman, N., Potts, C., & Icard, T. (2025). Causal abstraction: A theoretical foundation for mechanistic interpretability. Journal of Machine Learning Research, 26(83), 1–64.Google Scholar
Millière, R., & Rathkopf, C. (2026). Anthropocentric bias in language model evaluation. Computational Linguistics, 52(1), 379–388. https://doi.org/10.1162/COLI.a.582 CrossRefGoogle Scholar
Piccinini, G., & Craver, C. (2011). Integrating psychology and neuroscience: Functional analyses as mechanism sketches. Synthese, 183(3), 283–311. https://doi.org/10.1007/s11229-011-9898-4 CrossRefGoogle Scholar
Williams, D. J. (2025). Two senses of medium independence. Mind & Language, 40(4), 437–447. https://doi.org/10.1111/mila.12542 CrossRefGoogle Scholar