Perspective-taking behavior as the probabilistic weighing of multiple domains

Area of Science:

• Psycholinguistics
• Cognitive Science
• Computational Linguistics

Background:

• Conflicting findings exist regarding whether listeners use common ground or private knowledge when interpreting referring expressions.
• Previous research has not fully explained how these two perspectives interact during language comprehension.

Purpose of the Study:

• To propose and test a model where listeners simultaneously integrate common ground and private knowledge for interpreting referring expressions.
• To reconcile contradictory results in psycholinguistic literature concerning perspective-taking in language processing.

Main Methods:

• Developed a Bayesian model of reference resolution.
• Focused on model predictions for two key prior studies (Keysar et al., 2000; Heller et al., 2008).
• Conducted a visual-world eye-tracking experiment to empirically test the model's predictions.

Main Results:

• Demonstrated that experimental results previously attributed to differing perspective-taking strategies can arise from a single, integrated strategy.
• Eye-tracking data supported the model's prediction of simultaneous integration of common ground and private knowledge.
• The proposed model successfully accounts for previously contradictory findings in the literature.

Conclusions:

• Referring expression interpretation arises from the simultaneous integration of common ground and private knowledge, not solely one or the other.
• This integrated perspective-taking model offers a unified explanation for diverse findings in psycholinguistic research.
• The findings advance our understanding of the cognitive mechanisms underlying human language comprehension and reference resolution.