Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Investigating and Extending Homans' Social Exchange Theory with Large Language Model based Agents

Lei Wang, Zheqing Zhang, Xu Chen

TL;DR

The paper addresses validating Homans' SET in a controllable virtual setting using LLM-based agents. It builds a virtual society of $M=3$ agents and $N=3$ resource types, with a value function where $r_1=1$, $r_2=4$, and $r_3=9$, running for $T=10$ rounds. Through a two-phase Negotiation and Exchange framework, the authors validate the six SET propositions and observe human-like phases of exploration, cooperation, and endgame dynamics. Building on this foundation, they extend the theory via cognitive style and social value orientation and examine resilience, including real-world human–AI experiments to test corollaries. These results demonstrate a feasible interdisciplinary paradigm bridging social science and computer science and provide open-source code for replication.

Abstract

Homans' Social Exchange Theory (SET) is widely recognized as a basic framework for understanding the formation and emergence of human civilizations and social structures. In social science, this theory is typically studied based on simple simulation experiments or real-world human studies, both of which either lack realism or are too expensive to control. In artificial intelligence, recent advances in large language models (LLMs) have shown promising capabilities in simulating human behaviors. Inspired by these insights, we adopt an interdisciplinary research perspective and propose using LLM-based agents to study Homans' SET. Specifically, we construct a virtual society composed of three LLM agents and have them engage in a social exchange game to observe their behaviors. Through extensive experiments, we found that Homans' SET is well validated in our agent society, demonstrating the consistency between the agent and human behaviors. Building on this foundation, we intentionally alter the settings of the agent society to extend the traditional Homans' SET, making it more comprehensive and detailed. To the best of our knowledge, this paper marks the first step in studying Homans' SET with LLM-based agents. More importantly, it introduces a novel and feasible research paradigm that bridges the fields of social science and computer science through LLM-based agents. Code is available at https://github.com/Paitesanshi/SET.

Investigating and Extending Homans' Social Exchange Theory with Large Language Model based Agents

TL;DR

The paper addresses validating Homans' SET in a controllable virtual setting using LLM-based agents. It builds a virtual society of agents and resource types, with a value function where , , and , running for rounds. Through a two-phase Negotiation and Exchange framework, the authors validate the six SET propositions and observe human-like phases of exploration, cooperation, and endgame dynamics. Building on this foundation, they extend the theory via cognitive style and social value orientation and examine resilience, including real-world human–AI experiments to test corollaries. These results demonstrate a feasible interdisciplinary paradigm bridging social science and computer science and provide open-source code for replication.

Abstract

Homans' Social Exchange Theory (SET) is widely recognized as a basic framework for understanding the formation and emergence of human civilizations and social structures. In social science, this theory is typically studied based on simple simulation experiments or real-world human studies, both of which either lack realism or are too expensive to control. In artificial intelligence, recent advances in large language models (LLMs) have shown promising capabilities in simulating human behaviors. Inspired by these insights, we adopt an interdisciplinary research perspective and propose using LLM-based agents to study Homans' SET. Specifically, we construct a virtual society composed of three LLM agents and have them engage in a social exchange game to observe their behaviors. Through extensive experiments, we found that Homans' SET is well validated in our agent society, demonstrating the consistency between the agent and human behaviors. Building on this foundation, we intentionally alter the settings of the agent society to extend the traditional Homans' SET, making it more comprehensive and detailed. To the best of our knowledge, this paper marks the first step in studying Homans' SET with LLM-based agents. More importantly, it introduces a novel and feasible research paradigm that bridges the fields of social science and computer science through LLM-based agents. Code is available at https://github.com/Paitesanshi/SET.

Paper Structure

This paper contains 18 sections, 3 theorems, 1 equation, 9 figures.

Table of Contents

  1. Introduction
  2. Preliminary
  3. The Constructed Agent Society
  4. Single Agent Framework
  5. Multi-Agent Society
  6. Validation of Homans' SET
  7. Extensions of Homans' SET
  8. Cognitive Style
  9. Social Value Orientations
  10. Social System Resilience
  11. Real-World Experiments on Evaluating the Corollaries of Homans' SET
  12. Related Work
  13. Homans Social Exchange Theory
  14. LLM-driven Agent-based Modeling
  15. Conclusion
  16. ...and 3 more sections

Key Result

Corollary 1

In the real world, rational individuals are more likely to achieve higher benefits, but they also face the risk of greater losses. In contrast, individuals with an experiential thinking style tend to achieve more stable but moderate benefits over time.

Figures (9)

  • Figure 1: Illustration and examples of the six propositions in Homans' Social Exchange Theory.
  • Figure 2: Overview of the agent society: single agent framework and multi-agent exchange pipeline.
  • Figure 3: Key metrics of agent behavior over 10 rounds.
  • Figure 4: Experimental validation of key propositions in Social Exchange Theory.
  • Figure 5: Analysis of affinity and exchange value between Rational and Experiential agents.
  • ...and 4 more figures

Theorems & Definitions (3)

  • Corollary 1: The Stability Corollary
  • Corollary 2: The Reciprocity Corollary
  • Corollary 3: The Resilience Corollary