Introduction
A study published in the journal Nature has revealed new details about how the brain processes memories during sleep, highlighting a mechanism that prevents old memories from being erased or distorted by new experiences. This discovery offers valuable clues about the role of sleep in memory consolidation and possible applications for artificial intelligence.
Methods
Researchers at Cornell University conducted experiments with laboratory rats. The animals were trained to perform tasks in mazes and had their brain and pupil activity monitored during sleep. The researchers used brain electrodes and cameras to record neuronal activity and changes in the pupils, allowing a detailed analysis of the different phases of sleep.
Results
1. Memory Processing in Sleep:
• During non-REM sleep, recent memories are reactivated when the pupils are constricted.
• Old memories are processed in a different subphase, when the pupils are dilated.
2. Alternating Memory Cycle:
• The brain alternates between consolidating new memories and reprocessing old memories, avoiding so-called “catastrophic forgetting.”
• This cycle helps to integrate new information without compromising previously stored knowledge.
3. Importance of Temporal Separation:
• Temporal separation in sleep ensures that new learning does not interfere with the storage of old memories, protecting memory integrity.
Discussion
The findings highlight how sleep plays an essential role in memory consolidation. In addition to contributing to the understanding of brain processes, the discoveries may lead to the development of techniques to improve memory in humans and inspire advances in artificial intelligence. For example, separating memories into subphases may help to create more efficient neural networks that are less dependent on energy resources.
Practical Applications
• Human Health: Techniques based on non-invasive tracking of pupils during sleep may be used to treat memory deficits in people with neurological disorders.
• Artificial Intelligence: Models based on the alternation of memory consolidation phases may make machine learning systems more efficient.
Conclusion
The research offers revolutionary insights into the mechanisms that ensure the preservation and enhancement of memory during sleep. In addition to applications in health and technology, the results pave the way for further research into how to improve human cognition and the efficiency of artificial intelligence systems.
Reference
Almeida, CS (2025). Why don’t new memories replace old ones? The science of sleep provides clues. SIC Notícias. Available at: https://sicnoticias.pt
