Humans combine probabilistic and pragmatic cues when judging presuppositions in conditionals, while LLMs show variable human-like alignment and often lack coherent pragmatic reasoning even when their ratings match.
For large proprietary models (e.g., GPT-5 and Gemini-2.5-Flash), a single run on 360 samples costs approximately $14 CAD for explanation gen- eration
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Presupposition and Reasoning in Conditionals: A Theory-Based Study of Humans and LLMs
Humans combine probabilistic and pragmatic cues when judging presuppositions in conditionals, while LLMs show variable human-like alignment and often lack coherent pragmatic reasoning even when their ratings match.