大语言模型中的前景理论：对损失厌恶、敏感性递减与概率加权的元分析
Prospect Theory in Large Language Models: A Meta-Analysis of Loss Aversion, Diminishing Sensitivity, and Probability Weighting
AI 导读
一项元分析汇总 11 项研究中 32 个大语言模型的 102 个拟合估计，检验其是否符合前景理论的三个成分。结果显示 LLM 损失厌恶较弱，合并 λ 为 1.26（95% CI 1.06–1.46），远低于人类的 1.96；价值函数在 Tversky-Kahneman 方法下接近线性（1.01 对人类 0.67）；但概率加权呈类人的倒 S 形（γ 为 0.81 和 0.84）。
正文 · 原文
Abstract:Prospect theory splits risky choice into three parts: loss aversion, measured by the relative weight of losses to gains, $\lambda$ (e.g., $\lambda = 2$ implies that losses weigh twice as much as gains), diminishing sensitivity to the size of outcomes (i.e., the curvature and linearity of the value function), and the nonlinear weighting of probabilities. I examined whether Large Language Models (LLMs) display the same properties as people by meta-analyzing 102 fitted estimates from 11 studies covering 32 language models. The three components emerge differently. Loss aversion is modest: the pooled $\lambda$ is 1.26 (95% CI 1.06 to 1.46), far below the strongest human meta-analytic estimate ($\lambda = 1.96$) and compatible with more conservative ones (1.07 and 1.31). Matched by design, LLMs are compatible with people when gains and losses are symmetric ($\lambda = 1.15$ against 1.07), but well below them when losses are smaller than gains (1.46 against 2.3), where human loss aversion is strongest. Value functions are close to linear rather than concave under the Tversky-Kahneman method (1.01, against a human value of 0.67), but more concave than in people under the Tanaka method (0.52, against 0.67), so diminishing sensitivity depends on how it is measured. Probability weighting shows a human-like inverse S shape ($\gamma = 0.81$ under Tversky-Kahneman and 0.84 under the Tanaka method), though on few studies. In sum, LLMs show little loss aversion and little diminishing sensitivity, but retain human-like probability distortion, yet these generalizations shift when looking at specific model families and fitting methods.
| Subjects: | Artificial Intelligence (cs.AI); Computation and Language (cs.CL); General Economics (econ.GN) |
| Cite as: | arXiv:2610.33103 [cs.AI] |
| (or arXiv:2610.33103v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.33103 arXiv-issued DOI via DataCite |
Submission history
From: Adi Tarabeih [view email]
[v1]
Thu, 22 Oct 2026 15:08:56 UTC (1,184 KB)
来源：arXiv cs.AI · arxiv.org