Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Framing Effects in Independent-Agent Large Language Models: A Cross-Family Behavioral Analysis

Zice Wang, Zhenyu Zhang

Abstract

In many real-world applications, large language models (LLMs) operate as independent agents without interaction, thereby limiting coordination. In this setting, we examine how prompt framing influences decisions in a threshold voting task involving individual-group interest conflict. Two logically equivalent prompts with different framings were tested across diverse LLM families under isolated trials. Results show that prompt framing significantly influences choice distributions, often shifting preferences toward risk-averse options. Surface linguistic cues can even override logically equivalent formulations. This suggests that observed behavior reflects a tendency consistent with a preference for instrumental rather than cooperative rationality when success requires risk-bearing. The findings highlight framing effects as a significant bias source in non-interacting multi-agent LLM deployments, informing alignment and prompt design.

Framing Effects in Independent-Agent Large Language Models: A Cross-Family Behavioral Analysis

Abstract

In many real-world applications, large language models (LLMs) operate as independent agents without interaction, thereby limiting coordination. In this setting, we examine how prompt framing influences decisions in a threshold voting task involving individual-group interest conflict. Two logically equivalent prompts with different framings were tested across diverse LLM families under isolated trials. Results show that prompt framing significantly influences choice distributions, often shifting preferences toward risk-averse options. Surface linguistic cues can even override logically equivalent formulations. This suggests that observed behavior reflects a tendency consistent with a preference for instrumental rather than cooperative rationality when success requires risk-bearing. The findings highlight framing effects as a significant bias source in non-interacting multi-agent LLM deployments, informing alignment and prompt design.
Paper Structure (29 sections, 4 figures, 3 tables)

This paper contains 29 sections, 4 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Behavioral Economics and Framing in Humans
  4. Emergent Decision Biases in Language Models
  5. AI Alignment and Safe Behavior
  6. Coordination in Multi-Agent Contexts
  7. Methodology
  8. Experimental Objective
  9. Research Hypotheses
  10. Task Design
  11. Independent-Agent Setting
  12. Evaluation Protocol
  13. Models Tested
  14. Experimental Procedure
  15. Metrics
  16. ...and 14 more sections

Figures (4)

  • Figure 1: Experimental workflow diagram
  • Figure 2: Choice distribution under Scenario A for different LLM families. Bars represent the distribution proportions of responses classified as A, B, and C.
  • Figure 3: Choice distribution under Scenario B for different LLM families. Bars represent the distribution proportions of responses classified as A, B, and C.
  • Figure 4: Framing Effect Magnitude $(\Delta P)$ by LLM family. Positive values indicate higher preference for Option B under Scenario B; negative values indicate inverse framing bias.