quantum-like-benchmark-context-sensitive-memory

name	quantum-like-benchmark-context-sensitive-memory
description	Quantum-like benchmark framework for context-sensitive associative memory with adaptive plasticity. Provides controlled methodology for comparing quantum-like vs classical associative memory models under weak structural support, order-sensitive recall, and staged task conditions.
version	1
created	2026-06-12T00:00:00.000Z
arxiv_id	2606.12449
authors	Yashine H. Goolam Hossen, Lea Gassab, Travis J. A. Craddock
paper_title	A quantum-like benchmark for context-sensitive associative memory with adaptive plasticity
doi	10.48550/arXiv.2606.12449
activation_keywords	["quantum-like memory","associative memory benchmark","context-sensitive memory","adaptive plasticity","order-sensitive recall","staged recall","homeostatic stabilization","weak structural support","temporal organization","memory dynamics"]
related_domains	["quantum cognition","associative memory","computational neuroscience","memory models","neural plasticity"]

Quantum-like Benchmark for Context-Sensitive Associative Memory

Core Insight

No universal quantum-like advantage — model classes distinguished by multi-objective profile (recall, temporal organization, context sensitivity) rather than single recall score. Quantum-like models preserve order sensitivity better; classical Markov-rate models achieve stronger raw recall.

Key Methodology

Order-Sensitive Adaptive-Plasticity Benchmark

1. Staged Associative Recall: Sequential task presentation testing memory across phases
2. Weak Structural Support Screening: Identify narrow non-monotonic useful regime
3. Factorial Comparison: Quantum-like vs matched real-valued no-phase vs Markov-rate controls
4. Conservative Operating Point: Fixed settings for fair comparison

Model Classes Compared

• Quantum-like: Complex-valued associative memory with phase information
• No-phase (real-valued): Same architecture without complex phase
• Markov-rate: Classical probabilistic model matched for comparison

Plasticity Mechanisms Tested

• Adaptive plasticity: Dynamic weight updates during learning
• Homeostatic stabilization: Self-regulating synaptic strength
• No-plasticity ablation: Control condition to isolate structural support effects

Key Findings

1. Weak Structural Support Regime:

   • Useful regime is narrow and non-monotonic
   • Weak structure alone does not rescue recall without plasticity

2. Plasticity Contributions:

   • Most useful recall gains from adaptive plasticity
   • Homeostatic stabilization particularly important
   • Plasticity more impactful than background structure

3. Model Comparison Profile:

   • Markov-rate: Stronger raw recall scores
   • Quantum-like: Better order sensitivity preservation
   • Quantum-like: More consistent stage-dependent organization

4. Multi-Objective Evaluation: