Self-Emotion Blended Dialogue Generation in Social Simulation Agents

TL;DR

This paper investigates how self-emotion in dialogue agents influences their strategies and decision-making, demonstrating that incorporating self-emotion results in more human-like interactions and significantly impacts agent decisions in social simulations.

Contribution

It introduces the concept of self-emotion in dialogue agents and shows its effects on strategy, naturalness, humanness, and decision-making within a large language model framework.

Findings

Self-emotion improves human-likeness of dialogue strategies.

Models with self-emotion exhibit higher naturalness and humanness.

Self-emotion significantly alters agent decision-making, with about 50% change in decisions.

Abstract

When engaging in conversations, dialogue agents in a virtual simulation environment may exhibit their own emotional states that are unrelated to the immediate conversational context, a phenomenon known as self-emotion. This study explores how such self-emotion affects the agents' behaviors in dialogue strategies and decision-making within a large language model (LLM)-driven simulation framework. In a dialogue strategy prediction experiment, we analyze the dialogue strategy choices employed by agents both with and without self-emotion, comparing them to those of humans. The results show that incorporating self-emotion helps agents exhibit more human-like dialogue strategies. In an independent experiment comparing the performance of models fine-tuned on GPT-4 generated dialogue datasets, we demonstrate that self-emotion can lead to better overall naturalness and humanness. Finally, in a virtual simulation environment where agents have discussions on multiple topics, we show that self-emotion of agents can significantly influence the decision-making process of the agents, leading to approximately a 50% change in decisions.