The interplay between sleep and memory has long been a focal point in neuroscience. While extensive past research indicates that sleep facilitates memory consolidation, it remains unclear whether the reverse is also true: specifically, how memory reactivation—a key component of brain activity during sleep—might influence sleep architecture and adaptively participate in the regulation of sleep homeostasis. Addressing this question requires capturing causal relationships between memory-related neural activities and changes in sleep states from multimodal, long-term data on both sleep and memory, thereby validating their potential regulatory roles.
On June 4, 2026, researchers from the Beijing Academy of Artificial Intelligence (BAAI) and Tsinghua University announced new progress on this issue. Their findings, titled “Memory Reactivation Underlies Experience-Dependent Adaptive Regulation of Sleep,” were published in the international academic journal Science. Dr. Lei Bo, head of the WuJie·Brainμ Model Team at BAAI, and Professor Zhong Yi from Tsinghua University’s School of Life Sciences served as co-corresponding authors for the study.
The study demonstrates that memory reactivation during sleep participates in regulating sleep dynamics, providing new experimental evidence for understanding the bidirectional mechanisms between “memory and sleep.” As a technical support for data analysis in this research, Brainμ0, a multimodal foundation model for brain science developed by BAAI’s AI + Neuroscience team, supported key analytical steps such as multimodal memory-sleep data analysis, assisting scientists in hypothesis verification, and sleep state identification. This highlights the potential of AI-for-neuroscience foundation models to contribute to complex basic life sciences research.
1 BAAI’s Self-Developed Neuroscience Foundation Model Brainμ: An AI Analysis Base for Multimodal Neurodata
Modern neuroscience data has entered an era characterized by multimodality, high throughput, and long-term recording. Consequently, the strong heterogeneity of multi-source neural data and the difficulty in achieving unified representation and joint analysis have become common challenges in basic research. To address this need, BAAI developed Brainμ0, a multimodal foundation model for brain science. Its core module, Brainμ Tokenizer, converts various types of neural signals—such as EEG, two-photon calcium imaging data, and Neuropixels recordings—into aligned neural activity representation tokens. This enables the analysis of multimodal data within a unified framework. Coupled with its accompanying foundation model decoder, Brainμ0 supports tasks essential to basic neuroscience research, including cross-subject and cross-scenario data annotation, identification of specific neural activity events, prediction of neural activity, and cross-modal alignment.
Figure 1: Schematic structure of Brainμ Tokenizer (Mouse)
(Image source: AI-generated)
2 From “Sleep Promotes Memory” to “Memory Regulates Sleep”: AI Foundation Models Assist in Hypothesis Verification
Leveraging the multimodal neural signal encoding and decoding capabilities of the Brainμ model, the research team proposed a new “AI + Basic Research” approach for assisting hypothesis verification and causal inference of signals. In this Science-published study on memory-regulated sleep activity, Brainμ0 was used to process and model sleep EEG signals alongside memory-related single-cell two-photon calcium imaging signals. The model not only helped researchers confirm that neural signals of memory activity can effectively predict the occurrence of sleep phase changes but also assisted in distinguishing between “Memory Reactivation Sleep” (MRS) and “Non-Memory Reactivation Sleep.” This provided support for data-driven hypothesis verification in neuroscience. Throughout the study, Brainμ achieved cross-subject and cross-scenario data analysis and verification in zero-shot scenarios, demonstrating the generalization potential of foundation models in neuroscience data analysis.
The team was the first to confirm that negative memory reactivation during sleep exacerbates sleep fragmentation and increases organismal alertness, whereas positive memory reactivation significantly enhances sleep continuity and resistance to interference. This discovery advances our understanding of sleep regulation: sleep is not merely a passive recovery process but may also be dynamically influenced by past experiences and memory content. This establishes a new scientific framework for the bidirectional regulatory mechanisms between sleep and memory, offering novel mechanistic perspectives and therapeutic approaches for sleep disorders associated with mental illnesses such as depression and anxiety.
Figure 2: Brainμ model assists neuroscientists in verifying the dynamic relationship between memory activity and sleep.
3 From Mechanism Verification to Automated Analysis: Brainμ Establishes a New “AI + Neuroscientist” Paradigm
Beyond hypothesis verification and neural signal analysis in this study, the Brainμ0 model has been applied in real-world research scenarios across multiple cutting-edge neuroscience laboratories. It assists neuroscientists in cross-species, cross-modal data analysis and scientific hypothesis verification, covering research directions related to memory, emotion, and brain diseases.
In a joint study between BAAI and the National Institute of Biological Sciences (NIBS), Brainμ0 was applied to the automated analysis of sleep neural activity data in mice. Existing automatic analysis algorithms for mouse sleep often suffer from performance degradation when facing new experimental paradigms or new transgenic strains, limiting their stable application in real research settings. Unlike previous small models designed for single tasks or single data types, Brainμ0’s training data covers over 70,000 nights of sleep records and incorporates data from different genetic backgrounds, task paradigms, brain regions, and other modalities. Consequently, it possesses stronger generalization capabilities across subjects, tasks, and modalities. In collaboration with Professor Liu Qinghua’s team at NIBS, Brainμ0 was used for the automated analysis of long-term sleep data in mice of different transgenic strains, passing a “model + human expert” bidirectional verification involving over 3,000 nights of sleep data. Additionally, the research team collaborated with Huawei to perform deep inference adaptation and optimization on the Brainμ model using Ascend super-nodes and full-stack AI4S capabilities. This has continuously supported automated analysis for over 10 months. The relevant analysis not only achieved zero-shot cross-strain generalization but also maintained high consistency with the results analyzed by professional doctoral students in sleep neuroscience over the 10-month period.
Figure 3: Brainμ assists in cross-scenario, cross-subject automated classification of mouse sleep.
(Image source: AI-generated)
Based on its multimodal foundation model architecture, Brainμ features flexible adaptability for downstream tasks, providing unified neural signal representations and reasoning analysis based on large language models for different experimental paradigms and data models. Looking ahead, BAAI will continue to advance the research and implementation of scientific intelligence foundation models like Brainμ. By addressing the high complexity, multimodality, and cross-scale data challenges in brain science, the institute aims to explore new paradigms for the deep integration of AI and basic neuroscience research, promoting artificial intelligence as a crucial tool for resolving complex life science problems and achieving breakthroughs in basic research.
Original Link:
Comments
Sign in to join the discussion and leave a comment.
Sign in with Google