Memory Consolidation Flattens the Temporal Shape of User Facts

2026-09-28

AI Summary

This study addresses the loss of temporal cues in existing long-term memory systems when storing user facts, which causes downstream models to misjudge fact validity. Targeting the phenomenon of "aspectual flattening" during memory writing, this work proposes the LAPSE benchmark, integrating mainstream LLM memory pipelines such as mem0 with contrastive behavioral analysis techniques to systematically evaluate how temporal information loss impacts downstream decision-making. The research confirms that prevalent systems exhibit selective loss of progressive aspect, and that this deficiency asymmetrically alters reading models' judgments regarding fact validity. By exposing a core blind spot in current memory architectures, this project provides a critical empirical foundation for building temporally aware and robust memory systems.

Abstract

Long-term memory systems turn conversations into short stored notes. A note can keep a user fact while losing evidence about whether the fact still holds. For example, "I am driving a Peugeot" can become "The user drives a Peugeot," which drops the cue that the activity is ongoing. We call this aspectual flattening and measure it with LAPSE, a benchmark of matched user statements that differ only in temporal form. We find that memory writers flatten aspect selectively. Three writer models flattened the progressive statement but kept its simple-present match in 244 of 381 pairs, never the reverse. The asymmetry holds in all 11 model configurations tested and in the installed pipelines mem0, Graphiti, and Letta. The lost cue matters to later readers. In exploratory tests, changing only the stored verb shifted all three readers' estimates that a fact still holds. When readers could ask the user before acting, two of three acted without asking more often on flattened notes. Our planned memory-use task could not detect this, because readers there acted on almost every stored fact, even expired ones. Memory writing can thus remove evidence that later models use to decide whether to act.

Problem

• memory consolidation
• aspectual flattening
• long-term memory systems
• temporal shape

Innovation

• Aspectual Flattening
• Memory Consolidation
• LAPSE Benchmark
• Long-term Memory Systems
• Temporal Cues