Metraipolis › archive › dc_92fbd35c460b

Larry Caused the Car to Stop, But the Model Didn't Notice: Transformer Blindness to the M-Heuristic — arXiv:2609.37497

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.37497
License (as archived)
CC0-1.0
Archived
2026-10-07T16:56:57.990Z
SHA-256 of the archived text
662d6721b9f31fe5a2c1dd01de9681986f4372393534b63f489a83d38361247c
Stable URL of this page
https://metraipolis.com/documento/dc_92fbd35c460b

How to cite

Larry Caused the Car to Stop, But the Model Didn't Notice: Transformer Blindness to the M-Heuristic — arXiv:2609.37497 — arXiv (en), https://arxiv.org/abs/2609.37497 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_92fbd35c460b (Metraipolis archive dc_92fbd35c460b, sha256 662d6721b9f31fe5a2c1dd01de9681986f4372393534b63f489a83d38361247c)
BibTeX
@misc{dc92fbd35c460b,
  title = {Larry Caused the Car to Stop, But the Model Didn't Notice: Transformer Blindness to the M-Heuristic — arXiv:2609.37497},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.37497},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_92fbd35c460b; sha256 662d6721b9f31fe5a2c1dd01de9681986f4372393534b63f489a83d38361247c},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.37497. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Modern transformer models excel at capturing semantic relationships through sentence embeddings, yet their ability to perform pragmatic reasoning remains understudied. This paper investigates whether encoder-based transformers such as DeBERTa employ the M-Heuristic (the neo-Gricean principle that marked linguistic forms implicate marked meanings). We test this hypothesis by contrasting lexical causatives (e.g., ``Larry stopped the car'') with periphrastic causatives (e.g., ``Larry caused the car to stop'') using a Natural Language Inference framework. Our experiments across 188 conditions with 15 ambitransitive verbs reveal that DeBERTa, RoBERTa, and BART show no evidence of capturing the pragmatic distinction between these forms, with DeBERTa predicting ``Neutral'' for 100% of cases. Probing analysis initially suggested a representation-use dissociation, but control experiments reveal the probe was tracking syntactic complexity, not causative pragmatics. Semantic similarity over 30 triplets places periphrastic causatives closer to unmediated manner descriptions in 29/30 cases, opposite to M-Heuristic predictions in the embedding space. Under explicit metalinguistic framing, Gemini Flash-Lite reaches 100% with item-specific traces, so the principle is available under instruction yet unused in default NLI.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-30 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.37497 — 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 2026-10-07T16:27:03.298Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-30?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).