Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Oblivion: Self-Adaptive Agentic Memory Control through Decay-Driven Activation

Ashish Rana, Chia-Chien Hung, Qumeng Sun, Julian Martin Kunkel, Carolin Lawrence

Abstract

Human memory adapts through selective forgetting: experiences become less accessible over time but can be reactivated by reinforcement or contextual cues. In contrast, memory-augmented LLM agents rely on "always-on" retrieval and "flat" memory storage, causing high interference and latency as histories grow. We introduce Oblivion, a memory control framework that casts forgetting as decay-driven reductions in accessibility, not explicit deletion. Oblivion decouples memory control into read and write paths. The read path decides when to consult memory, based on agent uncertainty and memory buffer sufficiency, avoiding redundant always-on access. The write path decides what to strengthen, by reinforcing memories contributing to forming the response. Together, this enables hierarchical memory organization that maintains persistent high-level strategies while dynamically loading details as needed. We evaluate on both static and dynamic long-horizon interaction benchmarks. Results show that Oblivion dynamically adapts memory access and reinforcement, balancing learning and forgetting under shifting contexts, highlighting that memory control is essential for effective LLM-agentic reasoning. The source code is available at https://github.com/nec-research/oblivion.

Oblivion: Self-Adaptive Agentic Memory Control through Decay-Driven Activation

Abstract

Human memory adapts through selective forgetting: experiences become less accessible over time but can be reactivated by reinforcement or contextual cues. In contrast, memory-augmented LLM agents rely on "always-on" retrieval and "flat" memory storage, causing high interference and latency as histories grow. We introduce Oblivion, a memory control framework that casts forgetting as decay-driven reductions in accessibility, not explicit deletion. Oblivion decouples memory control into read and write paths. The read path decides when to consult memory, based on agent uncertainty and memory buffer sufficiency, avoiding redundant always-on access. The write path decides what to strengthen, by reinforcing memories contributing to forming the response. Together, this enables hierarchical memory organization that maintains persistent high-level strategies while dynamically loading details as needed. We evaluate on both static and dynamic long-horizon interaction benchmarks. Results show that Oblivion dynamically adapts memory access and reinforcement, balancing learning and forgetting under shifting contexts, highlighting that memory control is essential for effective LLM-agentic reasoning. The source code is available at https://github.com/nec-research/oblivion.

Paper Structure

This paper contains 27 sections, 1 equation, 2 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Contributions.
  3. Oblivion Overview
  4. Hierarchical Memory Representation
  5. Dynamic Procedural Memory ($L_1$).
  6. Semantic Memory ($L_2$).
  7. Preemptive Episodic Memory ($L_3$).
  8. Read Path: Uncertainty-Gated Retrieval
  9. Decayer.
  10. Activator.
  11. Write Path: Feedback-Driven Updates
  12. Recognizer.
  13. Memory Manager.
  14. Executor: Self-Adaptive Control Loop
  15. Experimental Setup
  16. ...and 12 more sections

Figures (2)

  • Figure 1: Oblivion facilitates memory-augmented agents by decay-driven activation over hierarchical memory traces. The Executor orchestrates the read path for uncertainty-gated retrieval ($\diamond$); and the write path for feedback-driven updates ($\diamond$), enabling dynamic control over memory activation.
  • Figure 2: (a) Ebbinghaus forgetting curve showing decay patterns with reinforcement. (b) Retention distributions at time steps $t\in\{50,100,150\}$ for temperature $T\in\{1,3,5,10,20,50\}$. (c) Reinforcement fraction over time.