Discovery by Dreaming: Cross-Domain Recombination in Artificial Memory

2026-06-28

AI Summary

Traditional memory consolidation models emphasize preventing forgetting but overlook the role of cross-domain knowledge recombination in driving creative discovery. This work proposes that the core value of memory consolidation lies in offline cross-domain recombination rather than repetitive rehearsal, and introduces a dual-system validation framework: a LoRA-based DREAMS fine-tuning pipeline in neural networks and a structured knowledge object replay engine, SAPIENCE, in symbolic systems. Experiments demonstrate that this mechanism substantially enhances creative discovery—achieving an 85.7% cross-domain linkage discovery rate in symbolic systems (a 21-percentage-point improvement) and yielding a 14.5-percentage-point performance gain on cross-domain tasks such as GSM8K in neural models. These gains stem from model weight restructuring rather than prompt engineering.

Abstract

Dreams splice together people, places, and times that never met. Neuroscience suggests this recombination is not noise, but a function driving insight and creative discovery. This reframes memory consolidation: rather than merely defending against forgetting, its measurable value lies in recombining knowledge across experiences that have not yet co-occurred. We test this directly by isolating the recombinatory-replay mechanism and implementing it in two architecturally unrelated systems: a LoRA fine-tuning pipeline (DREAMS) and a symbolic engine replaying structured knowledge objects (SAPIENCE). Both systems converge on the same finding: cross-domain consolidation creates value, while within-domain rehearsal does not. The symbolic arm surfaces novel cross-domain connections at 85.7%, a +21 percentage point (pp) gain over baseline. The neural arm improves overall by +5.64 pp, but on subtasks explicitly requiring cross-domain transfer (like unseen math reasoning on GSM8K), gains reach +14.5 pp. This effect is a genuine property of the weights--not a prompt artifact--as prepending the same material in-context to a 671B-parameter model actually reverses the gain. We validate this prediction against documented discoveries across 50,000 real papers and state a falsifiable hippocampal-recording prediction to distinguish recombination from rehearsal. Ultimately, this principle is substrate-general, tracking real discovery at scale. Reading the literature teaches a model to recall what it has seen, but producing discovery requires a separate offline phase that recombines knowledge across domains--the computational analog of dreaming. Consolidation is not for remembering, but for discovering.

Problem

• cross-domain recombination
• artificial memory
• creative discovery
• memory consolidation
• dreaming

Innovation

• cross-domain recombination
• artificial memory
• dreaming-inspired learning
• memory consolidation
• creative discovery