Language Models Trained to do Arithmetic Predict Human Risky and Intertemporal Choice

2024-05-29

arXiv.org

AI Summary

This paper investigates whether large language models (LLMs) can serve as reliable computational models of human cognition—particularly in risk and intertemporal decision-making—addressing concerns that existing LLMs exhibit anthropomorphic behavior primarily due to overfitting to human behavioral data. Method: We introduce the “computational equivalence task” paradigm: training an LLM (Arithmetic-GPT) exclusively on ecologically grounded arithmetic tasks—without any human behavioral data—to probe whether human-like choice patterns emerge spontaneously. Our approach integrates arithmetic competence modeling, expected-value computation, and systematic comparison with canonical cognitive models (e.g., prospect theory, hyperbolic discounting), supplemented by ablation analyses. Contribution/Results: Arithmetic-GPT significantly outperforms most conventional models across both decision domains, demonstrating that foundational arithmetic ability constitutes a critical cognitive prerequisite for human decision-making. This provides a novel, data-free pathway for cognitive modeling—establishing that domain-general reasoning capacities, rather than behavioral data fitting, can yield human-aligned decision behavior.

Technology Category

• Cognitive Modeling & Cognitive Systems: Simulating Human Behavior
• Humans and AI: Other Foundations of Human Computation & AI
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

The observed similarities in the behavior of humans and Large Language Models (LLMs) have prompted researchers to consider the potential of using LLMs as models of human cognition. However, several significant challenges must be addressed before LLMs can be legitimately regarded as cognitive models. For instance, LLMs are trained on far more data than humans typically encounter, and may have been directly trained on human data in specific cognitive tasks or aligned with human preferences. Consequently, the origins of these behavioral similarities are not well understood. In this paper, we propose a novel way to enhance the utility of LLMs as cognitive models. This approach involves (i) leveraging computationally equivalent tasks that both an LLM and a rational agent need to master for solving a cognitive problem and (ii) examining the specific task distributions required for an LLM to exhibit human-like behaviors. We apply this approach to decision-making -- specifically risky and intertemporal choice -- where the key computationally equivalent task is the arithmetic of expected value calculations. We show that an LLM pretrained on an ecologically valid arithmetic dataset, which we call Arithmetic-GPT, predicts human behavior better than many traditional cognitive models. Pretraining LLMs on ecologically valid arithmetic datasets is sufficient to produce a strong correspondence between these models and human decision-making. Our results also suggest that LLMs used as cognitive models should be carefully investigated via ablation studies of the pretraining data.

Problem

• Enhancing LLMs as cognitive models for human-like decision-making
• Investigating task distributions for LLMs to mimic human behavior
• Validating pretraining data's role in LLM-human behavioral correspondence

Innovation

• Leverage computationally equivalent tasks for cognition
• Use ecologically valid arithmetic datasets for pretraining
• Apply ablation studies to investigate pretraining data