# Rashomon Memory: Towards Argumentation-Driven Retrieval for Multi-Perspective Agent Memory

arXiv: https://arxiv.org/abs/2604.03588 · HTML: https://mlsanity.com/paper/2604.03588/

## Key facts

- Cited by 3 papers in the map (data through 14 Oct 2026).
- Builds on 21 of its 45 references that are in the map.
- First posted on arXiv 4 Apr 2026.

## Builds on (most cited first, 21 of 21)

- [Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks](https://mlsanity.com/paper/2005.11401/) (2020)
- [Overcoming catastrophic forgetting in neural networks](https://mlsanity.com/paper/1612.00796/) (2016)
- [Generative Agents: Interactive Simulacra of Human Behavior](https://mlsanity.com/paper/2304.03442/) (2023)
- [Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection](https://mlsanity.com/paper/2310.11511/) (2023)
- [Improving Factuality and Reasoning in Language Models through Multiagent Debate](https://mlsanity.com/paper/2305.14325/) (2023)
- [Explanation in Artificial Intelligence: Insights from the Social Sciences](https://mlsanity.com/paper/1706.07269/) (2017)
- [Encouraging Divergent Thinking in Large Language Models through Multi-Agent Debate](https://mlsanity.com/paper/2305.19118/) (2023)
- [MemoryBank: Enhancing Large Language Models with Long-Term Memory](https://mlsanity.com/paper/2305.10250/) (2023)
- [RAPTOR: Recursive Abstractive Processing for Tree-Organized Retrieval](https://mlsanity.com/paper/2401.18059/) (2024)
- [AI safety via debate](https://mlsanity.com/paper/1805.00899/) (2018)
- [Zep: A Temporal Knowledge Graph Architecture for Agent Memory](https://mlsanity.com/paper/2501.13956/) (2025)
- [Theory of Mind for Multi-Agent Collaboration via Large Language Models](https://mlsanity.com/paper/2310.10701/) (2023)
- [MemoRAG: Boosting Long Context Processing with Global Memory-Enhanced Retrieval Augmentation](https://mlsanity.com/paper/2409.05591/) (2024)
- [LLMs4OL: Large Language Models for Ontology Learning](https://mlsanity.com/paper/2307.16648/) (2023)
- [Argumentative XAI: A Survey](https://mlsanity.com/paper/2105.11266/) (2021)
- [On Verifiable Legal Reasoning: A Multi-Agent Framework with Formalized Knowledge Representations](https://mlsanity.com/paper/2509.00710/) (2025)
- [On the Evolution of Knowledge Graphs: A Survey and Perspective](https://mlsanity.com/paper/2310.04835/) (2023)
- [Evaluating Theory of Mind and Internal Beliefs in LLM-Based Multi-Agent Systems](https://mlsanity.com/paper/2603.00142/) (2026)
- [Heterogeneous Debate Engine: Identity-Grounded Cognitive Architecture for Resilient LLM-Based Ethical Tutoring](https://mlsanity.com/paper/2603.27404/) (2026)
- [Multi-Agent Dialectical Refinement for Enhanced Argument Classification](https://mlsanity.com/paper/2603.27451/) (2026)
- [Structured Decomposition for LLM Reasoning: Cross-Domain Validation and Semantic Web Integration](https://mlsanity.com/paper/2601.01609/) (2026)

## Built on by (most cited first, 3 of 3)

- [TOKI: A Bitemporal Operator Algebra for Contradiction Resolution in LLM-Agent Persistent Memory](https://mlsanity.com/paper/2606.06240/) (2026)
- [Choosing the Lens: Strategic Perspective Activation in Context-Dependent Argumentation](https://mlsanity.com/paper/2605.31581/) (2026)
- [Interpreting at Write Time: A Policy Ablation for Multi-Goal Agent Memory](https://mlsanity.com/paper/2610.02897/) (2026)

## Abstract (arXiv)

AI agents operating over extended time horizons accumulate experiences that serve multiple concurrent goals, and must often maintain conflicting interpretations of the same events. A concession during a client negotiation encodes as a ``trust-building investment'' for one strategic goal and a ``contractual liability'' for another. Current memory architectures assume a single correct encoding, or at best support multiple views over unified storage. We propose Rashomon Memory: an architecture where parallel goal-conditioned agents encode experiences according to their priorities and negotiate at query time through argumentation. Each perspective maintains its own ontology and knowledge graph. At retrieval, perspectives propose interpretations, critique each other's proposals using asymmetric domain knowledge, and Dung's argumentation semantics determines which proposals survive. The resulting attack graph is itself an explanation: it records which interpretation was selected, which alternatives were considered, and on what grounds they were rejected. We present a proof-of-concept showing that retrieval modes (selection, composition, conflict surfacing) emerge from attack graph topology, and that the conflict surfacing mode, where the system reports genuine disagreement rather than forcing resolution, lets decision-makers see the underlying interpretive conflict directly.