{"slug": "larry-caused-the-car-to-stop-but-the-model-didn-t-notice-transformer-blindness-m", "title": "Larry Caused the Car to Stop, But the Model Didn't Notice: Transformer Blindness to the M-Heuristic", "summary": "Encoder-based transformers DeBERTa, RoBERTa, and BART show no evidence of capturing the M-Heuristic pragmatic distinction between lexical causatives such as \"Larry stopped the car\" and periphrastic causatives such as \"Larry caused the car to stop,\" according to an arXiv paper (2609.37497v1) testing 188 conditions across 15 ambitransitive verbs. DeBERTa predicted \"Neutral\" in 100% of cases, and semantic similarity over 30 triplets placed periphrastic causatives closer to unmediated manner descriptions in 29 of 30 cases, opposite to M-Heuristic predictions. Under explicit metalinguistic framing, Gemini Flash-Lite reached 100% accuracy with item-specific traces, indicating the principle is available under instruction but unused in default natural language inference.", "body_md": "arXiv:2609.37497v1 Announce Type: new \nAbstract: 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.", "url": "https://wpnews.pro/news/larry-caused-the-car-to-stop-but-the-model-didn-t-notice-transformer-blindness-m", "canonical_source": "https://www.machinebrief.com/news/larry-caused-the-car-to-stop-but-the-model-didnt-notice-tran-j1bg", "published_at": "2026-09-30 04:00:00+00:00", "updated_at": "2026-09-30 04:47:21.857925+00:00", "lang": "en", "topics": ["natural-language-processing", "large-language-models", "machine-learning", "ai-research"], "entities": ["DeBERTa", "RoBERTa", "BART", "Gemini Flash-Lite", "arXiv"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/larry-caused-the-car-to-stop-but-the-model-didn-t-notice-transformer-blindness-m", "markdown": "https://wpnews.pro/news/larry-caused-the-car-to-stop-but-the-model-didn-t-notice-transformer-blindness-m.md", "text": "https://wpnews.pro/news/larry-caused-the-car-to-stop-but-the-model-didn-t-notice-transformer-blindness-m.txt", "jsonld": "https://wpnews.pro/news/larry-caused-the-car-to-stop-but-the-model-didn-t-notice-transformer-blindness-m.jsonld"}}