When Agents "Misremember" Collectively: Exploring the Mandela Effect in LLM-based Multi-Agent Systems

TL;DR

This paper investigates the Mandela effect in LLM-based multi-agent systems, introducing a benchmark and mitigation strategies to reduce collective memory biases, thereby enhancing system reliability and ethical alignment.

Contribution

It presents MANBENCH, a novel benchmark for evaluating the Mandela effect in multi-agent systems, and proposes mitigation strategies that significantly reduce this bias.

Findings

The Mandela effect can be quantitatively evaluated in multi-agent systems.

Mitigation strategies reduced the Mandela effect by an average of 74.40%.

Different factors influence the susceptibility of agents to collective memory biases.

Abstract

Recent advancements in large language models (LLMs) have significantly enhanced the capabilities of collaborative multi-agent systems, enabling them to address complex challenges. However, within these multi-agent systems, the susceptibility of agents to collective cognitive biases remains an underexplored issue. A compelling example is the Mandela effect, a phenomenon where groups collectively misremember past events as a result of false details reinforced through social influence and internalized misinformation. This vulnerability limits our understanding of memory bias in multi-agent systems and raises ethical concerns about the potential spread of misinformation. In this paper, we conduct a comprehensive study on the Mandela effect in LLM-based multi-agent systems, focusing on its existence, causing factors, and mitigation strategies. We propose MANBENCH, a novel benchmark designed to evaluate agent behaviors across four common task types that are susceptible to the Mandela effect, using five interaction protocols that vary in agent roles and memory timescales. We evaluate agents powered by several LLMs on MANBENCH to quantify the Mandela effect and analyze how different factors affect it. Moreover, we propose strategies to mitigate this effect, including prompt-level defenses (e.g., cognitive anchoring and source scrutiny) and model-level alignment-based defense, achieving an average 74.40% reduction in the Mandela effect compared to the baseline. Our findings provide valuable insights for developing more resilient and ethically aligned collaborative multi-agent systems. Code and dataset are available at https://github.com/bluedream02/Mandela-Effect.