analiza aiqbench
This paper introduces RecToM, a framework that enhances Large Language Models' Theory of Mind capabilities by recursively constructing nested character perspectives to resolve higher-order belief questions. By reducing complex social reasoning tasks to actual-world queries within a specific constructed viewpoint, the method achieves state-of-the-art performance on benchmarks like Hi-ToM without relying solely on event filtering. This approach represents a significant step forward in enabling LLMs to handle asymmetric observations and infer agent beliefs more accurately.
Mind the Perspective: Let's Reason Recursively for Theory of Mind
arXiv:2606.11724v2 Announce Type: replace Abstract: Theory of Mind (ToM) reasoning requires inferring agents' beliefs from partial and asymmetric observations, which remains an open challenge for LLMs. Existing prompting-based approaches improve ToM reasoning through observable-event filtering or temporal belief chains, without explicitly modeling nested beliefs. We introduce RecToM, an inference-time framework for ToM reasoning that models nested beliefs via recursive perspective construction. RecToM constructs each character perspective from the preceding character perspective along the character chain specified by the question, reducing higher-order belief questions to actual-world questions within the final constructed perspective. We further provide a KD45 analysis showing that RecToM's perspective construction induces a well-formed belief modality beyond simple event filtering. Experiments on ToM benchmarks, including Hi-ToM, Big-ToM, and FanToM, across multiple LLM backbones show that RecToM consistently outperforms recent advanced approaches, achieving state-of-the-art performance. Notably, RecToM reaches 100\% accuracy on Hi-ToM with GPT-5.4 and Qwen3.5, a benchmark requiring higher-order ToM reasoning. Our code is available at https://github.com/you68681/rectom.
Abstract:Theory of Mind (ToM) reasoning requires inferring agents' beliefs from partial and asymmetric observations, which remains an open challenge for LLMs. Existing prompting-based approaches improve ToM reasoning through observable-event filtering or temporal belief chains, without explicitly modeling nested beliefs. We introduce RecToM, an inference-time framework for ToM reasoning that models nested beliefs via recursive perspective construction. RecToM constructs each character perspective from the preceding character perspective along the character chain specified by the question, reducing higher-order belief questions to actual-world questions within the final constructed perspective. We further provide a KD45 analysis showing that RecToM's perspective construction induces a well-formed belief modality beyond simple event filtering. Experiments on ToM benchmarks, including Hi-ToM, Big-ToM, and FanToM, across multiple LLM backbones show that RecToM consistently outperforms recent advanced approaches, achieving state-of-the-art performance. Notably, RecToM reaches 100\% accuracy on Hi-ToM with GPT-5.4 and Qwen3.5, a benchmark requiring higher-order ToM reasoning. Our code is available at this https URL.
Artificial Intelligence (cs.AI)
arXiv:2606.11724 [cs.AI]
(or arXiv:2606.11724v2 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2606.11724 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 247 cuvinte
Subiecte conexe
Mind the Perspective: Let's Reason Recursively for Theory of Mind
arXiv · 11 iun. 2026
The basic argument for making AI better at moral philosophy
LessWrong · 14 oct. 2026
The J-Space Debate, Agent Swarms, and Pacing Frontier AI - Digital Minds Newsletter #4
LessWrong · 18 sept. 2026
Illusionists should try to build hedonium
LessWrong · 17 iun. 2026