AI

Memory Persists Even After Half the Synapses Disappear: A New Brain Preservation Mechanism

Researchers at Yaeyama University induced artificial hibernation in mice and confirmed that memories were retained even when over half of the synapses were eliminated.

4 min read Reviewed & edited by the SINGULISM Editorial Team

Memory Persists Even After Half the Synapses Disappear: A New Brain Preservation Mechanism
Photo by Robina Weermeijer on Unsplash

The Mystery of Memory Retention Revealed by

Research at Okinawa Institute of Science and Technology

Based on an article reported by Jacek Krywko in Ars Technica.

The leading hypothesis for how memories are stored is that learning strengthens and physically enlarges the connections between the neurons involved, forming the memory. The problem is that these connections change dramatically over time and are plastic. “If you compare the connectivity on day one versus day four or day five, it is completely, completely different,” says Kazumasa Tanaka, a neuroscientist at the Okinawa Institute of Science and Technology Graduate University.

To reveal how memories that can last for years could exist on hardware that changes its appearance every few days, Tanaka’s team made the change more dramatic. In a recent study published in Science, they induced a hibernation-like state in mice, which essentially erased over half the state of the synapses. Despite this, the mice appeared to retain their memories.

The Artificial Hibernation Technology QIH,

Controllable at Will

Hibernation is something squirrels, hamsters, and bears excel at, but the neural circuits that induce it are conserved across all mammals and exist even in species that do not hibernate in the wild (e.g., mice). A technique developed in 2020 by neuroscientist Takeshi Sakurai and colleagues at the University of Tsukuba involves artificially activating a group of cells in the hypothalamus called Q neurons. This induces a state called “Q neuron-induced hypothermia and hypometabolism (QIH),” where body temperature drops to about 20 degrees Celsius, and heart rate and respiration decrease significantly.

Tanaka explains, “Some species of squirrels enter deep hibernation where their body temperature drops near freezing, while bears keep theirs around 36–37 degrees. Artificial hibernation lies in the middle of that spectrum.” The most crucial aspect of QIH is that it can be switched on and off at the desired time. In the experiments, mice were kept in this state for 48 hours before being revived.

Experimental Results:

70% Decrease in Activity and Large-Scale Synapse Loss

To determine the extent of synapse loss during QIH, Tanaka’s team implanted microelectrode arrays (tetrodes) into the hippocampus of freely moving mice and recorded the firing of individual neurons. Neural activity decreased by about 70% upon entering the hibernation state. Additionally, brain tissue from some animals was examined using serial block-face scanning electron microscopy before hibernation, during hibernation, and several days after returning to the normal state.

Editorial Opinion

Short-Term Impact This discovery deepens the fundamental neuroscience understanding of the basis of memory. It may also provide implications for research into neurodegenerative diseases where synapse loss progresses, such as Alzheimer’s disease. It could accelerate testing of the hypothesis that memory is distributed through synaptic redundancy or the structural patterns of entire circuits, potentially leading to the development of new intervention strategies. Long-Term Perspective Over a 1-to-3-year span, this finding could redefine the boundary between brain and computation. The extremely robust memory retention mechanism demonstrated by the biological brain will directly influence the design of fault-tolerant neural networks and AI architectures with long-term memory. Furthermore, the use of hypometabolic states could open new frontiers in applied research, such as maintaining brain function during space exploration or prolonged medical procedures. Question from the Editorial Office The fact that memory persists even after losing half the synapses strongly suggests that the “location” of memory does not depend on the strength of specific connections, but rather on higher-order circuit structures and activity patterns.

References

Source: Ars Technica

Comments

← Back to Home