Prospect Theory Fails for LLMs: Revealing Instability of Decision-Making under Epistemic Uncertainty

2025-08-12

AI Summary

This study investigates whether prospect theory—a foundational model of human decision-making under uncertainty—adequately characterizes the behavioral responses of large language models (LLMs) to cognitive uncertainty, particularly when confronted with natural-language uncertainty markers such as “possible.” Method: We propose a three-stage economic questionnaire framework integrating probabilistic mapping and cognitive marker injection to systematically assess LLMs’ responses to diverse linguistic uncertainty expressions under controlled conditions. Contribution/Results: Our empirical evaluation reveals that prospect theory exhibits significant instability in LLMs: neither the value function nor the probability weighting function is consistently replicable, especially across heterogeneous linguistic uncertainty formulations. This work introduces the first integrated assessment framework linking cognitive uncertainty and decision behavior in LLMs; uncovers how surface-level linguistic form critically undermines standard rationality assumptions; and publicly releases all code and data to establish a reproducible benchmark for AI rationality modeling.

Technology Category

• Reasoning under Uncertainty: Decision/Utility Theory
• Machine Learning: Calibration & Uncertainty Quantification
• Natural Language Processing: Safety and Robustness

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Prospect Theory (PT) models human decision-making under uncertainty, while epistemic markers (e.g., maybe) serve to express uncertainty in language. However, it remains largely unexplored whether Prospect Theory applies to contemporary Large Language Models and whether epistemic markers, which express human uncertainty, affect their decision-making behaviour. To address these research gaps, we design a three-stage experiment based on economic questionnaires. We propose a more general and precise evaluation framework to model LLMs' decision-making behaviour under PT, introducing uncertainty through the empirical probability values associated with commonly used epistemic markers in comparable contexts. We then incorporate epistemic markers into the evaluation framework based on their corresponding probability values to examine their influence on LLM decision-making behaviours. Our findings suggest that modelling LLMs' decision-making with PT is not consistently reliable, particularly when uncertainty is expressed in diverse linguistic forms. Our code is released in https://github.com/HKUST-KnowComp/MarPT.

Problem

• Tests Prospect Theory applicability to LLM decision-making
• Examines epistemic markers' influence on LLM uncertainty behavior
• Reveals instability in LLM decision-making under linguistic uncertainty

Innovation

• Empirical probability values for epistemic markers
• Three-stage experiment with economic questionnaires
• General evaluation framework for decision-making