Science
Ars Technica - All content

Putting mice into hibernation causes a major loss of synapses

Source Entity

Jacek Krywko

August 24, 2026
Putting mice into hibernation causes a major loss of synapses

Researchers at the Okinawa Institute of Science and Technology found that inducing a hibernation-like state in mice causes a significant loss of neural synapses. Despite this dramatic structural turnover, the mice demonstrated an ability to retain long-term memories, challenging current understanding of synaptic plasticity.

The Paradox of Synaptic Plasticity and Memory Retention

Recent research conducted by neuroscientists at the Okinawa Institute of Science and Technology Graduate University has unveiled a startling phenomenon regarding the biological foundations of memory. By inducing a hibernation-like state in mice, researchers observed a dramatic reduction in synaptic connections, with more than half of the neural hardware being essentially erased or remodeled. This study, published in Science, pushes the boundaries of our understanding of how the brain preserves information despite constant structural flux.

The Challenge of Persistent Hardware

For decades, the leading neuroscientific hypothesis has posited that memories are stored through the strengthening and physical enlargement of synaptic connections between neurons. However, this model faces an inherent logical problem: the brain is highly plastic. As noted by lead researcher Kazumasa Tanaka, the arrangement of these connections changes significantly over just a few days. If memories are tethered to specific physical structures, the rapid turnover of these structures should logically lead to the degradation of stored information, yet humans and animals maintain memories for years.

Hibernation as a Stress Test for Memory

To investigate this discrepancy, the team subjected mice to a state of induced hibernation. This process acted as a extreme stress test for the brain's storage mechanisms. By forcing the mice into this state, the team was able to observe the physical destruction or significant alteration of over 50% of existing synapses. This provided a unique "clean slate" scenario that allowed the researchers to observe whether the underlying memory traces remained intact despite the loss of the physical connections previously thought to house them.

Redefining Memory Storage

Surprisingly, despite the massive loss of these neural connections, the mice exhibited the ability to retain their memories. This finding suggests that the traditional view of memory being strictly tied to the permanence of specific synaptic structures may be incomplete. It implies that memories might be encoded through more resilient mechanisms that can withstand significant structural turnover, or that the brain possesses a self-organizing capacity to reconstruct memory patterns once the hibernation state dissipates.

Broader Implications for Neuroscience

This study has profound implications for the field of neuroscience and potential future therapies. If memories are not exclusively dependent on the static maintenance of synaptic connections, it could change how we approach neurodegenerative diseases, where synaptic loss is a hallmark of cognitive decline. Understanding how the brain preserves information during periods of extreme metabolic or structural change could lead to breakthroughs in treating conditions like Alzheimer’s or dementia, where the "hardware" of the brain is severely compromised.

Future Research Directions

Moving forward, the scientific community must determine if this resilience is a byproduct of hibernation-specific biological processes or a fundamental property of mammalian brains. Future research will likely focus on the mechanisms that allow for the recovery of these neural pathways and whether similar "memory resilience" exists in humans. By continuing to bridge the gap between synaptic plasticity and long-term memory, researchers hope to solve the mystery of how our brains maintain a coherent sense of self amidst a constantly shifting physical substrate.

Verification Required?

Read the full report from the primary source

Go to Ars Technica - All content