Computational modeling of theory of mind: A Bayesian analysis of demonstrative use

Date

2025-3

Author

Aydın, Alaz

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

180
views

43
downloads

Cite This

Human communication involves theory of mind—the ability to infer mental states of others—which becomes especially evident in the use of referring expressions. Demonstratives like this and that require coordination of attention and perspective, making them ideal probes for studying social-cognitive processes. In this thesis, we investigate Turkish demonstrative use in situated dialog, focusing on differences between adults with Autism Spectrum Disorder (ASD) and non-clinical controls (NCC). Building on studies showing that Turkish demonstrative use is temporally linked to gaze, and that individuals with ASD exhibit reduced gaze recurrence, we integrate these findings into a unified framework. We apply a Rational Speech Act (RSA) model to compare nested pragmatic inferences across groups, and combine this with eye-tracking data to investigate joint attention during both the production and comprehension of demonstratives. Specifically, we quantify gaze recurrence and visual search patterns within a window centered on demonstrative utterances. Consistent with prior work, we find that the demonstrative şu reliably modulates attention and is used similarly across groups. We also observe that context dependence and the depth of nested inferences, as quantified by the RSA model, are largely similar between groups. However, the demonstrative bu is used differently by participants with ASD, indicating group-level differences in perspective-taking and referential strategy. These findings provide new insight into the cognitive mechanisms underpinning demonstrative use and demonstrate the value of integrating behavioral and computational methods in the study of pragmatic language.

Subject Keywords

Theory of Mind, Bayesian modeling, Demonstratives, Joint attention, Eye tracking

URI

https://hdl.handle.net/11511/114059

Collections

Graduate School of Informatics, Thesis