{"id":8751,"date":"2025-04-10T18:51:10","date_gmt":"2025-04-10T18:51:10","guid":{"rendered":"https:\/\/www.europesays.com\/uk\/8751\/"},"modified":"2025-04-10T18:51:10","modified_gmt":"2025-04-10T18:51:10","slug":"digital-mouse-brain-twin-offers-new-window-into-neural-function","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/uk\/8751\/","title":{"rendered":"Digital Mouse Brain Twin Offers New Window Into Neural Function"},"content":{"rendered":"<p><strong>Summary: <\/strong>Researchers have created an AI-powered \u201cdigital twin\u201d of the mouse visual cortex that can accurately simulate neural responses to visual input, including movies. Unlike earlier models, this digital twin generalizes beyond its training data, predicting neuron behavior and structure with remarkable accuracy.<\/p>\n<p>Trained on 900 minutes of brain recordings, the model allows researchers to run limitless experiments quickly and efficiently. This advancement may revolutionize how we study intelligence, brain disorders, and eventually, the human brain.<\/p>\n<p><strong>Key Facts:<\/strong><\/p>\n<ul class=\"wp-block-list\">\n<li><strong>Visual Cortex Modeling:<\/strong> The AI model predicts how tens of thousands of neurons respond to novel visual stimuli.<\/li>\n<li><strong>Beyond Training Data:<\/strong> It generalizes to new inputs and even predicts anatomical features like neuron type and location.<\/li>\n<li><strong>Unlimited Experiments:<\/strong> The digital twin allows researchers to run virtual brain experiments far faster than in living subjects.<\/li>\n<\/ul>\n<p><strong>Source: <\/strong>Stanford<\/p>\n<p><strong>Much as a pilot might practice maneuvers in a flight simulator, scientists might soon be able to perform experiments on a realistic simulation of the mouse brain. <\/strong><\/p>\n<p>In a new study, Stanford Medicine researchers and collaborators used an artificial intelligence model to build a \u201cdigital twin\u201d of the part of the mouse brain that processes visual information.<\/p>\n<p>  <img fetchpriority=\"high\" decoding=\"async\" width=\"1200\" height=\"799\" src=\"https:\/\/www.europesays.com\/uk\/wp-content\/uploads\/2025\/04\/ai-mouse-brain-neuroscience.jpg\" alt=\"This shows a mouse and a digital brain.\"  \/> The digital twin revealed which similarities mattered the most. Credit: Neuroscience News<\/p>\n<p>The digital twin was trained on large datasets of brain activity collected from the visual cortex of real mice as they watched movie clips. It could then predict the response of tens of thousands of neurons to new videos and images.<\/p>\n<p>Digital twins could make studying the inner workings of the brain easier and more efficient.<\/p>\n<p>\u201cIf you build a model of the brain and it\u2019s very accurate, that means you can do a lot more experiments,\u201d said\u00a0Andreas Tolias, PhD, Stanford Medicine professor of ophthalmology and senior author of the\u00a0study\u00a0published April 10 in\u00a0Nature.<\/p>\n<p>\u201cThe ones that are the most promising you can then test in the real brain.\u201d<\/p>\n<p>The lead author of the study is Eric Wang, PhD, a medical student at Baylor College of Medicine.<\/p>\n<p><strong>Beyond the training distribution<\/strong><\/p>\n<p>Unlike previous AI models of the visual cortex, which could simulate the brain\u2019s response to only the type of stimuli they saw in the training data, the new model can predict the brain\u2019s response to a wide range of new visual input. It can even surmise anatomical features of each neuron.<\/p>\n<p>The new model is an example of a foundation model, a relatively new class of AI models capable of learning from large datasets, then applying that knowledge to new tasks and new types of data \u2014 or what researchers call \u201cgeneralizing outside the training distribution.\u201d<\/p>\n<p>(ChatGPT is a familiar example of a foundation model that can learn from vast amounts of text to then understand and generate new text.)<\/p>\n<p>\u201cIn many ways, the seed of intelligence is the ability to generalize robustly,\u201d Tolias said. \u201cThe ultimate goal \u2014 the holy grail \u2014 is to generalize to scenarios outside your training distribution.\u201d<\/p>\n<p><strong>Mouse movies<\/strong><\/p>\n<p>To train the new AI model, the researchers first recorded the brain activity of real mice as they watched movies \u2014 made-for-people movies. The films ideally would approximate what the mice might see in natural settings.<\/p>\n<p>\u201cIt\u2019s very hard to sample a realistic movie for mice, because nobody makes Hollywood movies for mice,\u201d Tolias said. But action movies came close enough.<\/p>\n<p>Mice have low-resolution vision \u2014 similar to our peripheral vision \u2014 meaning they mainly see movement rather than details or color. \u201cMice like movement, which strongly activates their visual system, so we showed them movies that have a lot of action,\u201d Tolias said.<\/p>\n<p>Over many short viewing sessions, the researchers recorded more than 900 minutes of brain activity from eight mice watching clips of action-packed movies, such as\u00a0Mad Max. Cameras monitored their eye movements and behavior.<\/p>\n<p>The researchers used the aggregated data to train a core model, which could then be customized into a digital twin of any individual mouse with a bit of additional training.<\/p>\n<p><strong>Accurate predictions<\/strong><\/p>\n<p>These digital twins were able to closely simulate the neural activity of their biological counterparts in response to a variety of new visual stimuli, including videos and static images. The large quantity of aggregated training data was key to the digital twins\u2019 success, Tolias said.<\/p>\n<p>\u201cThey were impressively accurate because they were trained on such large datasets.\u201d<\/p>\n<p>Though trained only on neural activity, the new models could generalize to other types of data.<\/p>\n<p>The digital twin of one particular mouse was able to predict the anatomical locations and cell type of thousands of neurons in the visual cortex as well as the connections between these neurons.\u00a0<\/p>\n<p>The researchers verified these predictions against high-resolution, electron microscope imaging of that mouse\u2019s visual cortex, which was part of a larger project to map the structure and function of the mouse visual cortex in unprecedented detail.<\/p>\n<p>The results of that project, known as\u00a0MICrONS, was\u00a0published\u00a0simultaneously in\u00a0Nature.<\/p>\n<p><strong>Opening the black box<\/strong><\/p>\n<p>Because a digital twin can function long past the lifespan of a mouse, scientists could perform a virtually unlimited number of experiments on essentially the same animal.<\/p>\n<p>Experiments that would take years could be completed in hours, and millions of experiments could run simultaneously, speeding up research into how the brain processes information and the principles of intelligence.\u00a0<\/p>\n<p>\u201cWe\u2019re trying to open the black box, so to speak, to understand the brain at the level of individual neurons or populations of neurons and how they work together to encode information,\u201d Tolias said.<\/p>\n<p>In fact, the new models are already yielding new insights. In another related\u00a0study, also simultaneously published in\u00a0Nature, researchers used a digital twin to discover how neurons in the visual cortex choose other neurons with which to form connections.<\/p>\n<p>Scientists had known that similar neurons tend to form connections, like people forming friendships. The digital twin revealed which similarities mattered the most. Neurons prefer to connect with neurons that respond to the same stimulus \u2014 the color blue, for example \u2014 over neurons that respond to the same area of visual space.<\/p>\n<p>\u201cIt\u2019s like someone selecting friends based on what they like and not where they are,\u201d Tolias said. \u201cWe learned this more precise rule of how the brain is organized.\u201d<\/p>\n<p>The researchers plan to extend their modeling into other brain areas and to animals, including primates, with more advanced cognitive capabilities.<\/p>\n<p>\u201cEventually, I believe it will be possible to build digital twins of at least parts of the human brain,\u201d Tolias said. \u201cThis is just the tip of the iceberg.\u201d<\/p>\n<p>Researchers from the University G\u00f6ttingen and the Allen Institute for Brain Science contributed to the work.<\/p>\n<p><strong>Funding: <\/strong>The study received funding from the Intelligence Advanced Research Projects Activity, a National Science Foundation NeuroNex grant, the National Institute of Mental Health, the National Institute of Neurological Disorders and Stroke (grant U19MH114830), the National Eye Institute (grant R01 EY026927 and Core Grant for Vision Research T32-EY-002520-37), the European Research Council and the Deutsche Forschungsgemeinschaft.<\/p>\n<p>About this AI research news<\/p>\n<p class=\"has-background\" style=\"background-color:#ffffe8\"><strong>Author: <\/strong><a href=\"http:\/\/neurosciencenews.com\/cdn-cgi\/l\/email-protection#5a3433343b74383b331a292e3b343c35283e743f3e2f\" target=\"_blank\" rel=\"noreferrer noopener\">Nina Bai<\/a><br \/><strong>Source: <\/strong><a href=\"https:\/\/stanford.edu\" target=\"_blank\" rel=\"noreferrer noopener\">Stanford<\/a><br \/><strong>Contact: <\/strong>Nina Bai \u2013 Stanford<br \/><strong>Image: <\/strong>The image is credited to Neuroscience News<\/p>\n<p class=\"has-background\" style=\"background-color:#ffffe8\"><strong>Original Research: <\/strong>Open access.<br \/>\u201c<a href=\"https:\/\/doi.org\/10.1038\/s41586-025-08829-y\" target=\"_blank\" rel=\"noreferrer noopener\">Foundation model of neural activity predicts response to new stimulus types<\/a>\u201d by Andreas Tolias, et al. Nature<\/p>\n<p><strong>Abstract<\/strong><\/p>\n<p><strong>Foundation model of neural activity predicts response to new stimulus types<\/strong><\/p>\n<p>The complexity of neural circuits makes it challenging to decipher the brain\u2019s algorithms of intelligence.<\/p>\n<p>Recent breakthroughs in deep learning have produced models that accurately simulate brain activity, enhancing our understanding of the brain\u2019s computational objectives and neural coding.<\/p>\n<p>However, it is difficult for such models to generalize beyond their training distribution, limiting their utility.<\/p>\n<p>The emergence of foundation models\u00a0trained on vast datasets has introduced a new artificial intelligence paradigm with remarkable generalization capabilities.<\/p>\n<p>Here we collected large amounts of neural activity from visual cortices of multiple mice and trained a foundation model to accurately predict neuronal responses to arbitrary natural videos.<\/p>\n<p>This model generalized to new mice with minimal training and successfully predicted responses across various new stimulus domains, such as coherent motion and noise patterns.<\/p>\n<p>Beyond neural response prediction, the model also accurately predicted anatomical cell types, dendritic features and neuronal connectivity within the MICrONS functional connectomics dataset.<\/p>\n<p>Our work is a crucial step towards building foundation models of the brain. As neuroscience accumulates larger, multimodal datasets, foundation models will reveal statistical regularities, enable rapid adaptation to new tasks and accelerate research.<\/p>\n","protected":false},"excerpt":{"rendered":"Summary: Researchers have created an AI-powered \u201cdigital twin\u201d of the mouse visual cortex that can accurately simulate neural&hellip;\n","protected":false},"author":2,"featured_media":8752,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[3163],"tags":[323,1942,215,3725,3690,219,220,5738,53,16,15],"class_list":{"0":"post-8751","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-artificial-intelligence","8":"tag-ai","9":"tag-artificial-intelligence","10":"tag-brain-research","11":"tag-deep-learning","12":"tag-machine-learning","13":"tag-neurobiology","14":"tag-neuroscience","15":"tag-stanford","16":"tag-technology","17":"tag-uk","18":"tag-united-kingdom"},"share_on_mastodon":{"url":"https:\/\/pubeurope.com\/@uk\/114315177696968248","error":""},"_links":{"self":[{"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/posts\/8751","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/comments?post=8751"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/posts\/8751\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/media\/8752"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/media?parent=8751"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/categories?post=8751"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/uk\/wp-json\/wp\/v2\/tags?post=8751"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}