Evaluation of Deontic Conditional Reasoning in Large Language Models: The Case of Wason's Selection Task
This work addresses the underexplored domain specificity of LLM reasoning, providing insights into how their performance varies across rule types and parallels human cognitive biases, which is incremental as it builds on prior comparisons of LLM and human reasoning.
The study tackled the problem of evaluating conditional reasoning in large language models (LLMs) by introducing a new Wason Selection Task dataset with deontic modality, finding that LLMs reason better with deontic rules and exhibit matching-bias-like errors similar to humans.
As large language models (LLMs) advance in linguistic competence, their reasoning abilities are gaining increasing attention. In humans, reasoning often performs well in domain specific settings, particularly in normative rather than purely formal contexts. Although prior studies have compared LLM and human reasoning, the domain specificity of LLM reasoning remains underexplored. In this study, we introduce a new Wason Selection Task dataset that explicitly encodes deontic modality to systematically distinguish deontic from descriptive conditionals, and use it to examine LLMs' conditional reasoning under deontic rules. We further analyze whether observed error patterns are better explained by confirmation bias (a tendency to seek rule-supporting evidence) or by matching bias (a tendency to ignore negation and select items that lexically match elements of the rule). Results show that, like humans, LLMs reason better with deontic rules and display matching-bias-like errors. Together, these findings suggest that the performance of LLMs varies systematically across rule types and that their error patterns can parallel well-known human biases in this paradigm.