Undergraduate Honors Thesis Projects
Date of Award
2023
Document Type
Honors Paper
Degree Name
Psychology-BS
Department
Psychology
Advisor
Dr. Michele Acker
First Committee Member
Dr. Michele Acker
Second Committee Member
Dr. Meredith Meyer
Third Committee Member
Dr. Halard Lescinsky
Keywords
mentalizing, Bayesian computational model
Subject Categories
Cognitive Psychology | Social Psychology
Abstract
How do people reason about other people’s minds? Recent work has shown inferences about others’ mental states (“mentalizing”) relies on Bayesian reasoning over an intuitive theory of others’ minds. Bayesian computational models have allowed researchers to formalize predictions about how people process information during mentalizing, and such models predict human mentalizing with great accuracy. Despite their success, these models are limited to simple inferences, and they do not scale to complex inferences humans can make featuring large sets of possible mental states. I discuss how people can use abstract representations to simplify such large spaces of possibilities. I hypothesized that, in the case of inferring mental states from observed actions, people use abstract dimensions outlined in the ‘3D Mind Model’ and ‘ACTFASTaxonomy’ (e.g., Valence is an abstract dimension referring to how positive or negative a mental state is). To test this hypothesis, I created a computational model that utilizes these abstract representations to make inferences, and examined whether it captured people’s inferences as well as a model not using abstract representations. I collected data on human inferences in an online experiment, in which participants made ratings about how likely some particular mental state was to have caused a given action. These human inferences were compared to the models’ predictions for each trial. The model with abstract representations predicted human inferences just as well as the other model, supporting the main hypothesis. Yet, all models predicted human inferences less accurately than previous studies, weakening the conclusions drawn from the main analyses. I discuss the implications of these findings, and the clues they provide into what abstract representations may scaffold the process of mentalizing.
Licensing Permission
Copyright, all rights reserved. Fair Use
Recommended Citation
Nelson, Logan, "Scaffolding Mental State Inference with Abstract Representations" (2023). Undergraduate Honors Thesis Projects. 162.
https://digitalcommons.otterbein.edu/stu_honor/162
Acknowledgement 1
1
Acknowledgement 2
1