Metraipolis › archive › dc_fd64800f74b2
Disentangling Statistical Preemption from Entrenchment in Language Models' Avoidance of Overgeneralization — arXiv:2609.01794
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.01794
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T04:16:30.449Z
- SHA-256 of the archived text
9b945c718eeb436f911623d69bb4ee964506ac39fb24d5dd1428aa945cc4bdb9- Stable URL of this page
https://metraipolis.com/documento/dc_fd64800f74b2
How to cite
Disentangling Statistical Preemption from Entrenchment in Language Models' Avoidance of Overgeneralization — arXiv:2609.01794 — arXiv (en), https://arxiv.org/abs/2609.01794 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fd64800f74b2 (Metraipolis archive dc_fd64800f74b2, sha256 9b945c718eeb436f911623d69bb4ee964506ac39fb24d5dd1428aa945cc4bdb9)
BibTeX
@misc{dcfd64800f74b2,
title = {Disentangling Statistical Preemption from Entrenchment in Language Models' Avoidance of Overgeneralization — arXiv:2609.01794},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.01794},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fd64800f74b2; sha256 9b945c718eeb436f911623d69bb4ee964506ac39fb24d5dd1428aa945cc4bdb9},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.01794. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
How do learners avoid overgeneralizations such as Tom laughed me without explicit negative evidence? Constructionists have posited two proposals that describe indirect negative evidence against overgeneralizations: preemption (which privileges exposure to near-synonymous construction---e.g., she made him laugh) vs. entrenchment (all exposures to a verb's grammatical usages, including cases like He laughed). We disentangle these hypotheses by running controlled rearing experiments on LMs trained on child-caregiver conversations, where we systematically remove preemptive vs. non-preemptive evidence. We find that while LMs avoid overgeneralizations, they do not show preemption at a verb-specific level, instead showing weak but non-zero evidence of abstract preemption. Combined with results from analyzing the LMs' training dynamics, we find that LMs treat competing structures as indirect positive---as opposed to negative---evidence in the verb-specific condition. Insofar as preemption is the more plausible route to avoiding overgeneralizations in humans, our results point the need for there to be sensitivities to indirect negative evidence in neural network learners, and suggest new human experiments to test abstract preemption. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-03 in cs.CL). Categorie arXiv: Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2609.01794 — 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 2026-10-17T03:26:49.759Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-03?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).