Updated
Updated · spacedaily.com · Jun 3
MIT Scientists Propose Astrocytes Let 86 Billion-Neuron Brains Store Arbitrarily Large Memories
Updated
Updated · spacedaily.com · Jun 3

MIT Scientists Propose Astrocytes Let 86 Billion-Neuron Brains Store Arbitrarily Large Memories

2 articles · Updated · spacedaily.com · Jun 3

Summary

  • A 23 May PNAS paper from MIT argues astrocytes may help the brain encode far more memories than neuron-only models can explain, by making tripartite synapses the basic computational unit.
  • The model treats each astrocyte process as a semi-independent partner to 2 neurons, creating the higher-order coupling dense associative memory needs and allowing storage to scale with network size rather than hit a fixed ceiling.
  • Researchers say that architecture could also improve energy efficiency, while recent hippocampus and calcium-imaging studies have already hinted astrocytes play an active role in memory storage and retrieval.
  • The claim remains untested experimentally: the paper offers a mathematical hypothesis about storage capacity, not proof of how human memory works or why some memories persist while others fade.

Insights

Have scientists found a hidden 'second brain' that could explain our vast memory capacity?
Could mimicking the brain's ignored 'star cells' unlock the next generation of energy-efficient AI?

Astrocytes and the Brain’s Limitless Memory: The 2025 Paradigm Shift in Neuroscience, Medicine, and AI

Overview

In May and June 2025, MIT scientists introduced a groundbreaking hypothesis that challenges the traditional neuron-centric view of memory storage. They propose that astrocytes—star-shaped support cells—are actually central to how the brain stores memories. According to this new model, the brain’s basic computational units are tripartite synaptic domains, where astrocytes actively interact with both pre- and postsynaptic neurons. This dynamic interplay is suggested to be the core mechanism for memory storage, allowing each neuron-astrocyte unit to store as many memory patterns as there are neurons in the network. As a result, the brain’s memory capacity could be almost unlimited, constrained only by its physical size.

...