Views Are My Own, But Also Yours: Benchmarking Theory Of Mind Using Common Ground

Soubki Adil, Murzaku John, Jordehi Arash Yousefi, Zeng Peter, Markowska Magdalena, Mirroshandel Seyed Abolghasem, Rambow Owen. ACL 2024

[Paper]    
Attention Mechanism Model Architecture

Evaluating the theory of mind (ToM) capabilities of language models (LMs) has recently received a great deal of attention. However, many existing benchmarks rely on synthetic data, which risks misaligning the resulting experiments with human behavior. We introduce the first ToM dataset based on naturally occurring spoken dialogs, Common-ToM, and show that LMs struggle to demonstrate ToM. We then show that integrating a simple, explicit representation of beliefs improves LM performance on Common-ToM.

Similar Work