A neural network model of free recall learns multiple memory strategies
Nature Machine Intelligence, Published online: 20 July 2026; doi:10.1038/s42256-026-01274-0 Li et al. show that recurrent neural networks optimized for free recall discover diverse, human-like memory strategies beyond classical temporal context models, with top models using an index-based mechanism resembling the memory palace technique.
Neural networks trained for free recall exhibit several distinct memory strategies, including one that resembles the human "memory palace" technique. These models learn to emphasize the position of each list item, rather than its temporal context. This index code emerges when networks are prompted to recall all items and not prioritize recent ones.
This demonstrates that multiple computational mechanisms can yield human-like recall patterns, with stimulus-invariant index codes being the optimal strategy for expert-level recall performance.
Written by urgent.news from Nature Machine Intelligence's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
