[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-574960-105":59,"doc-detail-574960-en":131},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":124,"head_meta":126,"extra_data":128,"updated_unix":130},105,"en","how-biological-synapses-self-assemble-gradient-learning","How biological synapses self-assemble gradient learning","","Existing models explain how given brain circuits learn, but not how biology can assemble those circuits in the first place. The work introduces Self-Assembling Learning and one mechanism, the Self-Assembling Motif (SAM), showing that it emerges from initially random connectivity under heterosynaptic plasticity rules. Hierarchically composed SAM networks self-organize dynamics that provably approximate stochastic gradient descent and backpropagation-level performance, grounded in local rules rather than fixed global design.",{"@graph":69,"@context":123},[70,84,106],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/how-biological-synapses-self-assemble-gradient-learning/574960/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/how-biological-synapses-self-assemble-gradient-learning/574960.png","ImageObject",300,407,{"name":92,"@type":93},"Olivia Brown","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-11-16","2026-10-14",true,{"@type":102,"interactionType":103,"userInteractionCount":105},"InteractionCounter",{"@type":104},"ViewAction",15,{"@type":107,"mainEntity":108},"FAQPage",[109,115,119],{"name":110,"@type":111,"acceptedAnswer":112},"What gap in brain learning models does the paper address?","Question",{"text":113,"@type":114},"It addresses how biological systems could assemble learning-capable circuits, not just how fixed circuits learn.","Answer",{"name":116,"@type":111,"acceptedAnswer":117},"What is the Self-Assembling Motif (SAM)?",{"text":118,"@type":114},"SAM is a minimal circuit building block that emerges from random connectivity under heterosynaptic plasticity rules.",{"name":120,"@type":111,"acceptedAnswer":121},"How does SAM relate to gradient descent performance?",{"text":122,"@type":114},"Networks composed hierarchically from repeated SAMs self-organize dynamics that provably approximate stochastic gradient descent and backpropagation-level performance.","https://schema.org",{"og:url":83,"og:type":125,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":127,"canonical":83},"index,follow",{"doc_id":129,"site_id":62},574960,1792210237,{"code":4,"msg":5,"data":132},{"doc_id":129,"user_id":133,"nickname":92,"user_avatar":134,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":105,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":140,"language":141,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":142,"faqs":143,"seo_title":144,"seo_description":67,"update_tm":145,"read_time":41,"preview_page_count":19,"preview_urls":146},16904993612988,"https://ap-avatar.wpscdn.com/davatar_a8503ba1806abce46bf441b54a3ca4cd","Downloaded from [https://www.pnas.org by HANYANG UNIVERSITY MEDICAL LIBRARY on October 9](https://www.pnas.org by HANYANG UNIVERSITY MEDICAL LIBRARY on October 9), 2026 from IP address [89.116.89.204](89.116.89.204).  \nHow biological synapses self-assemble gradient learning  \nQianli Liaoa,b,c,1,2, Liu Ziyind,1, Yulu Ganb,e,1, Brian Cheungb,e, Mark T. Harnetta,c,f, and Tomaso Poggioa,b,c,e,f,g,2 Aﬀiliations are included on p. 11.  \nEdited by Terrence J. Sejnowski, Salk Institute for Biological Studies, La Jolla, CA; received February 20, 2026; accepted August 18, 2026  \nExisting models of learning in the brain explain how given circuits learn, but not how biology could assemble those circuitsin the ﬁrst place. To address this gap, we formulate Self-Assembling Learning—the study of how learning systems can emerge from lowerlevel interactions—and introduce one example mechanism, the Self-Assembling Motif (SAM) . SAM is self-assembling at two scales: The motif emerges from initially random connectivity under heterosynaptic plasticity rules, and networks of SAMs, composed hierarchically, self-organize into dynamics that provably approximate a generalized form of stochastic gradient descent—matching backpropagation-level performance. This suggests that biological learning need not be prescribed but can emerge from local rules—and to a far greater extent than previously thought.  \nneural networks j biological learning j gradient descent j synaptic plasticity j self-assembly  \nDeep neural networks have revolutionized computational neuroscience, where Artiﬁcial Neural Networks (ANNs) are increasingly used to model various aspects of information processing in the brain ( 1–8). These models demonstrate remarkable success in capturing neural responses and cognitive functions. Yet a model’s value as an account of the brain rests on more than reproducing such responses. A model of learning in the brain is biologically plausible only if biology could actually produce it—a property we call biological realizability. Realizability is not a verdict to be read off surface resemblance, nor a checklist of forbidden mechanisms to avoid; it is a demonstration that a mechanism could come to exist. And because biological structure has no external designer, any mechanism that realizes it must, at bottom, organize itself. Realizability is therefore fundamentally a question of self-organization—and, crucially, a matter of degree.  \nThis reframing turns biological plausibility from a binary verdict into a graded axis (Fig. 1): Learning models differ in how much of their structure is stipulated in advance vs. shown to emerge from local interactions. The axis runs deep. One may ask only that the weights self-organize, while the circuit that adjusts them is designed; or, going further, that the learning circuit itself assembles; further still, the scaffold that hosts it; and in principle, the developmental program that lays the scaffold down. The further down a model reaches, the more of what others simply assume it must instead derive—to move down the axis is to replace assumption with assembly. By this measure, most existing learning algorithms sit very near the top: They self-organize their inference weights, but under machinery that is otherwise ﬁxed in advance. This is true even when training yields rich, interpretable structure—reusable dynamical motifs (9) or inhibitory microcircuits supporting working memory maintenance ( 10)—because such structure emerges in the inference weights being trained, while the learning rule that adjusts them remains designed.  \nNowhere is this stipulation more evident than in backpropagation, the standard method for computing gradients in stochastic gradient descent (SGD) . It presupposes a precisely aligned architecture—forward and backward pathways tied together by weight transport and neuron-level correspondence—that is difﬁcult to reconcile with biology (5, 6, 11) . Biologically plausible alternatives loosen parts of t","cbCaioAHMj1lOiqR","https://ap.wps.com/l/cbCaioAHMj1lOiqR","pdf",6183942,12,"English","# Significance\n## Self-Assembling Learning and SAM overview\n## Author contributions\n## Competing interest statement","[{\"question\":\"What gap in brain learning models does the paper address?\",\"answer\":\"It addresses how biological systems could assemble learning-capable circuits, not just how fixed circuits learn.\"},{\"question\":\"What is the Self-Assembling Motif (SAM)?\",\"answer\":\"SAM is a minimal circuit building block that emerges from random connectivity under heterosynaptic plasticity rules.\"},{\"question\":\"How does SAM relate to gradient descent performance?\",\"answer\":\"Networks composed hierarchically from repeated SAMs self-organize dynamics that provably approximate stochastic gradient descent and backpropagation-level performance.\"}]","How biological synapses self-assemble gradient learning | PDF",1792012367,[147,151,155,159],{"page_no":8,"url":148,"width":149,"height":150},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-18/a44074b919d868a7fd5a4a9318e5c282.webp",983,1314,{"page_no":14,"url":152,"width":153,"height":154},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-18/8b24fa18262177740c7b052d580d4485.webp",982,1313,{"page_no":81,"url":156,"width":157,"height":158},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-18/aa23b0d4455bc66b9e1c278961ccec34.webp",981,1312,{"page_no":19,"url":160,"width":161,"height":162},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-18/b58c64956e5e859569905e918b8b2df8.webp",980,1310]