DelRec: learning delays in recurrent spiking neural networks

DelRec: learning delays in recurrent spiking neural networks | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article DelRec: learning delays in recurrent spiking neural networks Alexandre Queant, Ulysse Rancon, Benoit COTTEREAU, Timothée Masquelier This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9370190/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Biological neurons transmit information with stereotyped electrical impulses called ``spikes'', sensitive to coincident timings. Spiking Neural Networks (SNNs), introduced in the nineties, have gained popularity in AI for their energy efficiency and competitive performance with deep learning. Among them, Recurrent SNNs (RSNNs) are particularly appealing for their ability to learn long-term dependencies and exhibit rich dynamics. In SNNs, each connection can have a weight and a transmission delay, both plastic in the brain. While theory has long suggested that trainable delays enhance a network's expressivity, practical learning methods emerged only recently and remain mostly limited to feedforward delays. Here, we introduce DelRec, the first method to jointly optimize recurrent delays with synaptic weights in RSNNs via surrogate gradient learning, compatible with any spiking neuron model. DelRec works in discrete time, leveraging differentiable interpolation to handle non-integer delays with well-defined gradients at training time, then rounding them for inference. Using simple neurons, DelRec outperforms all baselines on a chaotic time-series prediction task, and sets new state-of-the-art accuracies on two challenging temporal datasets. Analysis of trained networks reveals structured, depth-dependent spatio-temporal receptive fields and delay-weight co-adaptation reshaping temporal selectivity. This work establishes recurrent delay optimization as a promising framework for both biological circuit modeling and neuromorphic computing. Physical sciences/Mathematics and computing/Computational science Biological sciences/Neuroscience/Computational neuroscience/Learning algorithms Full Text Additional Declarations There is NO Competing Interest. Supplementary Files supplementarynatcom.pdf Supplementary Information for: DelRec: learning delays in recurrent spiking neural networks Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9370190","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":622087314,"identity":"eca40cb7-ccc1-408a-ac5a-40414675069a","order_by":0,"name":"Alexandre Queant","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA/klEQVRIiWNgGAWjYDACdjDJjBDgh1ByuLUwo2mRkGwA08YkaDE4QEALfzPzMYmfO6zz+RmYt0l8qLlTZ3z87LGPPxgM8nFpkTjMlmzYeybdcmYDW5nkjGPPJMzO5CXP5mEwsGzApecwj+ED3rbDBgYHeMykedgOS5jd4DEGuvOPAS4d8of5Pxz8C9RiD9Ly599hCeMZPMaMQIfh1GJwmIfxMdgWBqAWxrbDEgYSPMYMPHi0GB5mMzaWbUs3AHqq2LK377DkjDM5xsw8Bri1yB1vfib5ts3agL+9eeONH98O8/O3nwE6rAK3FgRgZkBWRIQGEpSNglEwCkbBiAMA1GZKOvXvSQUAAAAASUVORK5CYII=","orcid":"","institution":"CerCo CNRS UMR 5549","correspondingAuthor":true,"prefix":"","firstName":"Alexandre","middleName":"","lastName":"Queant","suffix":""},{"id":622087315,"identity":"80beef3b-7a60-4424-b749-620ee7523a18","order_by":1,"name":"Ulysse Rancon","email":"","orcid":"https://orcid.org/0000-0002-3149-4870","institution":"Cerco, UMR 5549 CNRS - Université Toulouse III Paul Sabatier","correspondingAuthor":false,"prefix":"","firstName":"Ulysse","middleName":"","lastName":"Rancon","suffix":""},{"id":622087316,"identity":"81df5fc8-9af4-4b77-99ed-aee3412feac7","order_by":2,"name":"Benoit COTTEREAU","email":"","orcid":"https://orcid.org/0000-0002-2624-7680","institution":"Cerco, UMR 5549 CNRS - Université Toulouse III Paul Sabatier","correspondingAuthor":false,"prefix":"","firstName":"Benoit","middleName":"","lastName":"COTTEREAU","suffix":""},{"id":622087317,"identity":"d162201c-ae31-4ac4-8edf-67a33ec370c4","order_by":3,"name":"Timothée Masquelier","email":"","orcid":"https://orcid.org/0000-0001-8629-9506","institution":"CNRS ","correspondingAuthor":false,"prefix":"","firstName":"Timothée","middleName":"","lastName":"Masquelier","suffix":""}],"badges":[],"createdAt":"2026-04-09 14:57:34","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9370190/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9370190/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":106994354,"identity":"24928840-fdf7-4359-9234-af642ced1c23","added_by":"auto","created_at":"2026-04-15 15:07:55","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2024378,"visible":true,"origin":"","legend":"Article File","description":"","filename":"natcom.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9370190/v1_covered_d77bee0e-e480-41c7-b82c-8daa6ca4d84e.pdf"},{"id":106977544,"identity":"62c47dc0-410f-4477-a1ff-e247dd4a2f73","added_by":"auto","created_at":"2026-04-15 11:05:48","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":508540,"visible":true,"origin":"","legend":"Supplementary Information for: DelRec: learning delays in recurrent spiking neural networks","description":"","filename":"supplementarynatcom.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9370190/v1/591a925134d2cd6041f0de32.pdf"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"DelRec: learning delays in recurrent spiking neural networks","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-9370190/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9370190/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Biological neurons transmit information with stereotyped electrical impulses called ``spikes'', sensitive to coincident timings. Spiking Neural Networks (SNNs), introduced in the nineties, have gained popularity in AI for their energy efficiency and competitive performance with deep learning. Among them, Recurrent SNNs (RSNNs) are particularly appealing for their ability to learn long-term dependencies and exhibit rich dynamics. In SNNs, each connection can have a weight and a transmission delay, both plastic in the brain. While theory has long suggested that trainable delays enhance a network's expressivity, practical learning methods emerged only recently and remain mostly limited to feedforward delays. Here, we introduce DelRec, the first method to jointly optimize recurrent delays with synaptic weights in RSNNs via surrogate gradient learning, compatible with any spiking neuron model. DelRec works in discrete time, leveraging differentiable interpolation to handle non-integer delays with well-defined gradients at training time, then rounding them for inference. Using simple neurons, DelRec outperforms all baselines on a chaotic time-series prediction task, and sets new state-of-the-art accuracies on two challenging temporal datasets. Analysis of trained networks reveals structured, depth-dependent spatio-temporal receptive fields and delay-weight co-adaptation reshaping temporal selectivity. This work establishes recurrent delay optimization as a promising framework for both biological circuit modeling and neuromorphic computing.","manuscriptTitle":"DelRec: learning delays in recurrent spiking neural networks","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-15 11:05:40","doi":"10.21203/rs.3.rs-9370190/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-communications","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"NCOMMS","sideBox":"Learn more about [Nature Communications](http://www.nature.com/ncomms/)","snPcode":"","submissionUrl":"https://mts-ncomms.nature.com/","title":"Nature Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Communications","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"a99832cf-b050-4e45-9506-8d68b7f18969","owner":[],"postedDate":"April 15th, 2026","published":true,"recentEditorialEvents":[{"type":"editorInvitedReview","content":"This content is not available.","date":"2026-05-04T10:47:22+00:00","index":4,"fulltext":"This content is not available."}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":66195891,"name":"Physical sciences/Mathematics and computing/Computational science"},{"id":66195892,"name":"Biological sciences/Neuroscience/Computational neuroscience/Learning algorithms"}],"tags":[],"updatedAt":"2026-04-15T11:05:40+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-15 11:05:40","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9370190","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9370190","identity":"rs-9370190","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}