How Does the Sleeping Brain Support a Day’s Worth of Memories?
Dr. George Dragoi from Yale University discusses his latest work on memory consolidation during sleep.
Complete the form below to unlock access to ALL audio articles.
Sleep became a focus of serious scientific enquiry during the latter half of the 1800s, when researchers started to understand the structure and function of the human nervous system in greater depth. Many years – and countless studies later – our knowledge of sleep architecture and the vital role sleep plays in memory formation and retention is significantly more advanced.
Yet, despite spending one-third of our lives snoozing, there is still so much we don’t know about the behavior. For instance, consider your average day and the sheer number of experiences and information you are exposed to. How does the brain encode and consolidate all of these different occurrences, without getting mixed up?
“For decades, researchers have focused on revealing how sleep improved memories of certain events or facts. But memory can't really be understood on a single-item level – throughout a typical day we experience multiple events and learn numerous facts,” Dr. Eitan Schechtman, assistant professor of neurobiology and behavior at the University of California, Irvine, told Technology Networks.
Associate Professor George Dragoi’s latest research, published in Nature Neuroscience, helps to address this very query.
At Yale University, Dragoi examines the dynamic interplay between externally-driven and preconfigured, internally-generated representations of the external world. He has been studying the hippocampus, a brain structure that is central to learning and memory, for three decades.
The hippocampus comprises ensembles of neurons that replay our experiences and preplay future experiences while we sleep, supporting long-term memory formation. Using rat models, Dragoi’s latest work explored how multiple memory traces are encoded and consolidated in the mammalian brain without major interference. To achieve this, the research team subjected the rats to 15 linear track experiences and recorded the activity of hippocampal neuronal ensembles. They also took recordings as the rats slept before, in between and after tracks exploration.
Dragoi and colleagues’ study reveals several novel principles of information representation that help us to understand the efficiency of the mammalian brain.
One of the key findings was a serial position effect, whereby the rats’ earliest and most recent experiences were more strongly represented by the hippocampal neurons than other experiences. Different experiences were also co-represented at the same time during replay and preplay, which might help the brain process and store more information.
In an exclusive interview with Technology Networks, Dragoi sheds light on these data and other novel insights from the study. He also discusses how the study was set up to explore the process of memory encoding and consolidation under intense cognitive load, and the implications of his teams’ findings for future research on the brain and sleep.
Molly Coddington (MC): Can you explain what we know about the encoding-consolidation process of memory formation during sleep?
George Dragoi (GD): Long-term memory formation goes through two stages. Stage one is encoding, when new information from the external world modifies the internal dynamics of the brain to create a representation or trace of the associated experience the animal had. This trace is vulnerable and will be rapidly forgotten or interfered with unless it is consolidated in stage two into long-lasting changes in the brain, and communicated for long-term storage to areas that are generally different from those involved in encoding.
Both preplay expressed during the preceding sleep and the online activity during the new experience are believed to be critical for encoding, while replay during the following sleep is critical for consolidation.
MC: What inspired you to conduct this research?
GD: The first inspiration for this study came from my desire to better understand the neural network capacity for multiple memories encoding, which is the first stage of the generic two-stage encoding-consolidation process of memory formation.
My previous work estimated that when rats are confronted with encoding of 15 different contexts within a day, the hippocampal network might reach a capacity limit for their rapid and distinct encoding using the pre-existing (preplay) sequential motifs. At or after that point, the network would start exhibiting an interference between the representations of distinct experiences, which computational models predicted could result in “catastrophic interference” and likely disruption in distinct memory formation. However, we experience numerous experiences daily, and that catastrophic interference does not seem to occur so easily.
I decided to study the process of memory encoding and consolidation under intense cognitive load (e.g., 15 experiences/day) – something that has never been attempted before. In our new study, we show several principles by which the hippocampal network avoids this interference.
MC: How did you craft a study that could explore the process of memory encoding and consolidation under intense cognitive load?
GD: We tried to mimic a day in the life of a mammal (rat) naturalistically. We alternated 4 sleep sessions throughout a day with 20-30 minutes of individual track exploration, using 15 different tracks. The tracks were part of 2 mazes (maze 1 had 7 tracks, while maze 2 had 8) that were placed in 2 rooms or compartments. The rats physically (not virtually) explored them in sequence, from track 1 to track 15. Within each rest/sleep epoch (4 in total), we checked how the brain network represented these experiences.
MC: You found that replay of previous experiences displayed a track exposure serial position effect, i.e., the earliest and most recent experiences had the strongest replay. Can you discuss this finding in greater detail?
GD: These representations occurred during sleep (slow-wave sleep or non-REM sleep) and the rats were likely unaware of their occurrence.
These findings could mean that the brain allocates more resources/capacity to the most excitable neurons earlier on, when there is less potential for interference due to fewer prior experiences (primacy), and also to the latest experiences, whose entrained neurons are likely more excitable because they were the most recently activated (recency).
These network dynamics could likely contribute to the increased memory performance for the earliest and latest items in a series to be memorized.
MC: Overall, what does this study teach us about the efficiency of the brain? Can this research help us to understand there any implications from this data for how we understand dreaming and problem solving?
GD: Besides the serial position effect, this study revealed three new principles of information representation:
- The parallel nested co-representation of distinct experiences suggesting an enhanced capacity for parallel processing of information without interference in the hippocampus
- The serial binding of distinct experiences, both of which could support the generative function of the brain underlying insight and problem solving
- The serial representation of excerpts of multiple/all experiences across the day at hyper-compressed rates that are common during dreaming.
MC: Are there any limitations to this work that you wish to overcome in future studies?
GD: This was quite an intense project, and I think we already broke several limitations that past projects encountered. In the future, I would like to add a distinct meaning to the experiences encountered by the animals to see how that changes the organization principles of co-representation.
“Dragoi's work helps us understand how the sleeping brain supports a day's worth of memories. This is a critical step in our journey to understand how sleep helps memories in the real world, and it's likely to inspire research that increasingly acknowledges the complex nature of memory,” said Schechtman, who was not involved in the study.
MC: This study advances our understanding of how the rat brain manages and manipulates memories during sleep. In your opinion, what are the “big questions” that remain in sleep research?
GD: Sleep has a major restorative effect, which was not explored here. Also, different stages of sleep may play different roles in the organization of memories in the brain.
Dr. George Dragoi and Dr. Eitan Schechtman were speaking to Molly Coddington, Senior Science Writer and News Team Lead at Technology Networks.
Reference: Liu K, Sibille J, Dragoi G. Nested compressed co-representations of multiple sequential experiences during sleep. Nat Neuro. 2024. doi: 10.1038/s41593-024-01703-6
About the interviewees:
Dr. George Dragoi is an associate professor of psychiatry and of neuroscience in the department of psychiatry, and a member of the Wu-Tsai Institute, at Yale University School of Medicine in New Haven, CT. He completed his postdoctoral studies and was a research scientist at the Picower Institute for Learning and Memory at the Massachusetts Institute of Technology (MIT), where he revealed the existence of preconfigured cellular assemblies that pre-play in time the spatial sequences occurring during a future novel spatial experience in naive animals. Dr. Dragoi studies the dynamic interplay between externally-driven and preconfigured internally-generated representations of the external world to understand memory formation and spatial navigation. He aims to map the neural circuits and decipher the neuronal codes underlying the formation of these representations across brain development and in adulthood using large-scale high-density electrophysiology and computational methods for data analysis. Recently, he conceptualized the existence of a generative grammar in the brain that could support the brain’s ability to express internally generated representations about the world.
Dr. Eitan Schechtman completed a BA in Psychology and an MSc and PhD in neural computation, all at the Hebrew University of Jerusalem. He then completed his postdoc at Northwestern University. Dr. Schechtman is passionate about understanding the neural underpinnings of human cognition, and specifically the manner in which the sleeping brain supports cognitive function. Outside of the lab, he enjoys spending time with his husband and two kids, hiking, reading and listening to music. His passions include LGBTQ+ advocacy and scientific outreach, especially with school-aged kids.
