Summary of Understanding Epistemic Language with a Bayesian Theory Of Mind, by Lance Ying et al.
Understanding Epistemic Language with a Bayesian Theory of Mind
by Lance Ying, Tan Zhi-Xuan, Lionel Wong, Vikash Mansinghka, Joshua B. Tenenbaum
First submitted to arxiv on: 21 Aug 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper presents a cognitive model called LaBToM (Language-augmented Bayesian theory-of-mind) that helps people understand and evaluate claims about others’ beliefs, even when these beliefs can’t be directly observed. The model is grounded in Bayesian inferences about other agents’ goals, beliefs, and intentions. It translates natural language into an epistemic “language-of-thought” and evaluates these translations against probabilistic generative models of rational action and perception. The authors validate their model through an experiment where participants rate sentences about an agent’s beliefs while watching it navigate a maze. The results show that LaBToM correlates highly with human judgments for various expressions, including modal language, uncertainty, knowledge claims, likelihood comparisons, and attributions of false belief. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about how people understand what others think or believe, even if they can’t see it. The authors created a special model called LaBToM that helps figure out what others might be thinking based on what they say and do. It’s like trying to read someone’s mind! They tested this model by having people watch an agent try to find some keys in a maze, then rate how likely they thought the agent believed certain things. The results show that LaBToM is really good at guessing what people think based on their words and actions. |
Keywords
* Artificial intelligence * Likelihood
