LifeBench: A Benchmark for Long-Horizon Multi-Source Memory

2026-03-04

AI Summary

This work addresses the limitations of current AI memory evaluation frameworks, which predominantly focus on explicit memory while neglecting implicit forms such as procedural and habitual memory, and fail to emulate real-world memory tasks that are multi-source, long-term, and require behavioral inference. To bridge this gap, we propose LifeBench—a novel benchmark that integrates both explicit and implicit memory through a long-horizon, multi-source design. LifeBench is the first to incorporate non-declarative memory into AI evaluation, leveraging real-world priors—including anonymized social surveys, map APIs, and public holiday calendars—and employing a hierarchical parallel generation strategy grounded in cognitive science’s part-whole structural theory to construct dense, consistent, and diverse long-term memory events. Experiments reveal that state-of-the-art memory systems achieve only 55.2% accuracy on LifeBench, underscoring the significant challenges in cross-temporal, multimodal memory integration.

Technology Category

• Machine Learning: Life-Long and Continual Learning
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Long-term memory is fundamental for personalized agents capable of accumulating knowledge, reasoning over user experiences, and adapting across time. However, existing memory benchmarks primarily target declarative memory, specifically semantic and episodic types, where all information is explicitly presented in dialogues. In contrast, real-world actions are also governed by non-declarative memory, including habitual and procedural types, and need to be inferred from diverse digital traces. To bridge this gap, we introduce Lifebench, which features densely connected, long-horizon event simulation. It pushes AI agents beyond simple recall, requiring the integration of declarative and non-declarative memory reasoning across diverse and temporally extended contexts. Building such a benchmark presents two key challenges: ensuring data quality and scalability. We maintain data quality by employing real-world priors, including anonymized social surveys, map APIs, and holiday-integrated calendars, thus enforcing fidelity, diversity and behavioral rationality within the dataset. Towards scalability, we draw inspiration from cognitive science and structure events according to their partonomic hierarchy; enabling efficient parallel generation while maintaining global coherence. Performance results show that top-tier, state-of-the-art memory systems reach just 55.2\% accuracy, highlighting the inherent difficulty of long-horizon retrieval and multi-source integration within our proposed benchmark. The dataset and data synthesis code are available at https://github.com/1754955896/LifeBench.

Problem

• long-horizon memory
• non-declarative memory
• multi-source memory
• memory benchmark
• personalized agents

Innovation

• long-horizon memory
• non-declarative memory
• multi-source integration
• partonomic hierarchy
• event simulation