Unexplored brain cells could be the key to understanding our expansive memory capacity
MIT researchers propose new hypothesis for astrocytes' role in memory storage
A new study from MIT suggests that astrocytes - star-shaped cells in the human brain - play a more active role in memory storage than previously believed. Astrocytes, which were long thought to be mainly supportive cells, may contribute to the brain's massive storage capacity through a network of interactions with millions of neurons.
According to the study, published in the Proceedings of the National Academy of Sciences, astrocytes can detect neural activity and alter their calcium levels in response. These changes in calcium may trigger the release of gliotransmitters - signaling molecules similar to neurotransmitters - into synapses, creating a feedback loop with neurons.
The architecture suggested by the MIT team's model could explain the brain's memory storage capacity, which is much larger than would be expected using neurons alone. Astrocytes, which can interact with millions of neurons through long extensions, could potentially be used for computation, according to Jean-Jacques Slotine, an MIT professor of mechanical engineering and brain and cognitive sciences, and an author of the study.
Dmitry Krotov, a research staff member at the MIT-IBM Watson AI Lab and IBM Research, is the senior author of the open-access paper. Leo Kozachkov PhD '22 is the paper's lead author.
Memory capacity
Astrocytes have a variety of support functions in the brain, including cleaning up debris, providing nutrients to neurons, and helping to ensure an adequate blood supply. Astrocytes also send out many thin tentacles, known as processes, which can each wrap around a single synapse to create a tripartite (three-part) synapse.
Recent studies have shown that disrupting the connections between astrocytes and neurons in the hippocampus - a brain region critical for memory - leads to impaired memory storage and retrieval. Unlike neurons, astrocytes cannot fire action potentials, the electrical impulses that carry information throughout the brain. However, they can use calcium signaling to communicate with other astrocytes and coordinate their activity with neurons in the synapses that they associate with.
The MIT team's neuron-astrocyte associative memory model, based on Hopfield networks, can store significantly more information than a traditional Hopfield network, potentially accounting for the brain's massive memory capacity.
Significance
The model treats each astrocyte as a collection of processes, each functioning as an independent computational unit. Because of the high information storage capabilities of dense associative memories, the system is not only high-capacity but also energy-efficient. By conceptualizing tripartite synaptic domains - where astrocytes interact dynamically with pre- and postsynaptic neurons - as the brain’s fundamental computational units, the researchers argue that each unit can store as many memory patterns as there are neurons in the network.
The extensive biological connections between neurons and astrocytes offer support for the idea that this type of model might explain how the brain's memory storage systems work, the researchers say. They hypothesize that within astrocytes, memories are encoded by gradual changes in the patterns of calcium flow. This information is conveyed to neurons by gliotransmitters released at synapses that astrocyte processes connect to.
The study could have implications for both neuroscience and artificial intelligence research. By varying the connectivity of the process-to-process network, researchers could generate a huge range of models that could be explored for different purposes, such as creating a continuum between dense associative memories and attention mechanisms in large language models.
"While neuroscience initially inspired key ideas in AI, the last 50 years of neuroscience research have had little influence on the field, and many modern AI algorithms have drifted away from neural analogies," Slotine says. "In this sense, this work may be one of the first contributions to AI informed by recent neuroscience research."
- The MIT study suggests that astrocytes, star-shaped cells in the human brain, play a more active role in memory storage by interacting with millions of neurons.
- The new hypothesis proposes that memory encoding within astrocytes can occur through gradual changes in the patterns of calcium flow.
- The MIT team's neuron-astrocyte associative memory model, based on Hopfield networks, can store significantly more information than a traditional Hopfield network, accounting for the brain's massive memory capacity.
- The architecture suggested by the model could potentially be used for computation, according to Jean-Jacques Slotine, an MIT professor of mechanical engineering and brain and cognitive sciences.
- The study could have implications for both neuroscience and artificial intelligence research, as it may lead to the creation of a continuum between dense associative memories and attention mechanisms in large language models.
- The extensive biological connections between neurons and astrocytes offer support for the idea that this type of model might explain how the brain's memory storage systems work, contributing to health-and-wellness and medical-conditions related to memory.
