Morphodynamic Meta-Cognition: A Framework for Self-Regulating Intelligence Systems

Morphodynamic Meta-Cognition: A Framework for Self-Regulating Intelligence Systems | 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 Method Article Morphodynamic Meta-Cognition: A Framework for Self-Regulating Intelligence Systems Ioannis Mavroudis, Alin Ciobica, Bogdan Gurzu, Ecaterina Tomaziu-Todosia Anton, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9641646/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Large language models (LLMs) have demonstrated strong performance in natural language reasoning and knowledge synthesis, yet they remain limited by a lack of robust mechanisms for self-regulation. In high-stakes domains, such as healthcare and law, these systems frequently exhibit overconfidence, incomplete uncertainty representation, and inconsistent handling of conflicting evidence. These limitations reflect the absence of a structured meta-cognitive layer capable of monitoring and controlling the reasoning process. In this study, we introduce a novel computational framework for meta-cognitive control in artificial intelligence, conceptualizing cognition as a dynamic process within a high-dimensional state space and meta-cognition as the regulation of movement within this space. We implement this framework as a meta-cognitively controlled architecture that augments a standard LLM with mechanisms for monitoring uncertainty, detecting contradictions, and dynamically modulating reasoning behaviour. We evaluate the proposed system against a baseline single-pass LLM on tasks involving complex, ambiguous clinical narratives. Performance was assessed using metrics reflecting real-world requirements, including uncertainty coverage, contradiction detection, evidence grounding, confidence calibration, and governance quality. The meta-cognitive system demonstrated improved uncertainty coverage (0.88 vs 0.75), enhanced governance signals (0.96 vs 0.94), and more conservative and calibrated confidence estimates (0.57 vs 0.84), while maintaining comparable performance in evidence grounding and contradiction detection. These findings indicate that improvements in reliability are achieved not through scaling model capacity, but through architectural control of the reasoning process. By embedding meta-cognitive regulation within the system, it is possible to produce outputs that are better aligned with evidence, more transparent in their limitations, and more appropriate for high-stakes decision-making. This work provides a principled framework for integrating meta-cognition into artificial intelligence systems and suggests that future progress in trustworthy AI will depend on developing architectures that regulate how reasoning is performed, rather than solely improving what is generated. Bioinformatics Meta-cognition Large language models Uncertainty modelling Bayesian inference Explainable AI Trustworthy AI Figures Figure 1 Figure 2 Figure 3 Figure 4 Full Text Additional Declarations The authors declare no competing interests. Cite Share Download PDF Status: Posted 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-9641646","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Method Article","associatedPublications":[],"authors":[{"id":636209721,"identity":"2b7feed7-633f-4032-a00b-64a098cde82a","order_by":0,"name":"Ioannis Mavroudis","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABDUlEQVRIiWNgGAWjYBACNjBpAOcz8zCwN4BELEjRwnMAJCJBtK3MDAwSCSAGbi187M3PHhcU1DHotp8x+/CxzVpGfubzqxt+FEgw8Ld3J2B1GM8xc+MZBocZzM7kGM+c2ZbOY3A7p+xmD9BhEmfObsCqRSLBTJrH4ACD2YG0ZGbetsM8BtI5aTd4gFoMJHJxaEn/BtRSx2B2/hlEi/zMM2k3/+DVkgOyhZnB7EbyYbAWhhvsx27jtYXnTBlQy2EesxuPDzPOOAf0y5kcttsyBhI8uPwi396+TZrnT52c2fnEZoYPZdb28u3Hn91888dGjr+9F6sWGOBBZhugixAE7A9IUT0KRsEoGAXDHwAAnEdXDu5FsGwAAAAASUVORK5CYII=","orcid":"","institution":"Leeds University","correspondingAuthor":true,"prefix":"","firstName":"Ioannis","middleName":"","lastName":"Mavroudis","suffix":""},{"id":636209722,"identity":"1c7654c1-5714-43d7-97f2-b7d04c0b63f8","order_by":1,"name":"Alin Ciobica","email":"","orcid":"","institution":"Department of Biology, Faculty of Biology, “Alexandru Ioan Cuza” University of Iasi, 700505 Iasi,","correspondingAuthor":false,"prefix":"","firstName":"Alin","middleName":"","lastName":"Ciobica","suffix":""},{"id":636209723,"identity":"cca125d7-49a6-470f-8316-424308fb6968","order_by":2,"name":"Bogdan Gurzu","email":"","orcid":"","institution":"Grigore T. Popa University of Medicine and Pharmacy, University Street, No. 16, 700115 Iasi,Romania","correspondingAuthor":false,"prefix":"","firstName":"Bogdan","middleName":"","lastName":"Gurzu","suffix":""},{"id":636209724,"identity":"2cee2f87-3ea9-4c74-8810-5cdf9b1880fd","order_by":3,"name":"Ecaterina Tomaziu-Todosia Anton","email":"","orcid":"","institution":"Department of Obstetrics and Gynecology, Clinical Hospital of Obstetrics and Gynecology Cuza Voda, Cuza Voda Street, No. 34, 700038 Iasi, Romania","correspondingAuthor":false,"prefix":"","firstName":"Ecaterina","middleName":"Tomaziu-Todosia","lastName":"Anton","suffix":""},{"id":636209725,"identity":"6f35e220-4973-4400-bad5-c948a049a1d6","order_by":4,"name":"Bogdan Novac","email":"","orcid":"","institution":"Grigore T. Popa University of Medicine and Pharmacy, University Street, No. 16, 700115 Iasi, Romania","correspondingAuthor":false,"prefix":"","firstName":"Bogdan","middleName":"","lastName":"Novac","suffix":""}],"badges":[],"createdAt":"2026-05-07 11:22:37","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-9641646/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9641646/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":109124994,"identity":"c3c2f42e-ddc5-4508-81df-42ac17b2b41b","added_by":"auto","created_at":"2026-05-12 18:29:04","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":5071773,"visible":true,"origin":"","legend":"\u003cp\u003eMeta-Cognitive Control Loop\u003c/p\u003e","description":"","filename":"Figure1Metacognition.png","url":"https://assets-eu.researchsquare.com/files/rs-9641646/v1/044622a546221c5adf8c7886.png"},{"id":109124995,"identity":"e35dcd39-d6db-4f0a-872f-928f4cb8edb6","added_by":"auto","created_at":"2026-05-12 18:29:04","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":5737195,"visible":true,"origin":"","legend":"\u003cp\u003eMorphospace Representation\u003c/p\u003e","description":"","filename":"Figure2metacognition.png","url":"https://assets-eu.researchsquare.com/files/rs-9641646/v1/1bc5a3eabea24c39d926478c.png"},{"id":109205006,"identity":"09e9d94c-74eb-4b5a-934f-a4de0f48fa5f","added_by":"auto","created_at":"2026-05-13 15:03:11","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":5534822,"visible":true,"origin":"","legend":"\u003cp\u003ePerformance Comparison\u003c/p\u003e","description":"","filename":"Figure3Metacognition.png","url":"https://assets-eu.researchsquare.com/files/rs-9641646/v1/385a53d25af387ac7bf11438.png"},{"id":109124996,"identity":"cfc6b593-e169-403f-af84-0c7f3e2ed637","added_by":"auto","created_at":"2026-05-12 18:29:04","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":5713,"visible":true,"origin":"","legend":"\u003cp\u003eConfidence Calibration\u003c/p\u003e","description":"","filename":"placeholderimage.png","url":"https://assets-eu.researchsquare.com/files/rs-9641646/v1/a7d7e2233522645246f286b6.png"},{"id":109208408,"identity":"7d1af5a0-a3c1-4d92-9f97-29f0fb444782","added_by":"auto","created_at":"2026-05-13 15:24:44","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1384810,"visible":true,"origin":"","legend":"","description":"","filename":"TitleMorphodynamicMetaCognitionAFrameworkforSelfRegulatingIntelligenceSystems.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9641646/v1_covered_a8614775-00c1-4b87-8e62-6985b6b12c29.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003e\u003cstrong\u003eMorphodynamic Meta-Cognition: A Framework for Self-Regulating Intelligence Systems\u003c/strong\u003e\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Alexandru Ioan Cuza University","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Meta-cognition, Large language models, Uncertainty modelling, Bayesian inference, Explainable AI, Trustworthy AI","lastPublishedDoi":"10.21203/rs.3.rs-9641646/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9641646/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eLarge language models (LLMs) have demonstrated strong performance in natural language reasoning and knowledge synthesis, yet they remain limited by a lack of robust mechanisms for self-regulation. In high-stakes domains, such as healthcare and law, these systems frequently exhibit overconfidence, incomplete uncertainty representation, and inconsistent handling of conflicting evidence. These limitations reflect the absence of a structured meta-cognitive layer capable of monitoring and controlling the reasoning process.\u003c/p\u003e \u003cp\u003eIn this study, we introduce a novel computational framework for meta-cognitive control in artificial intelligence, conceptualizing cognition as a dynamic process within a high-dimensional state space and meta-cognition as the regulation of movement within this space. We implement this framework as a meta-cognitively controlled architecture that augments a standard LLM with mechanisms for monitoring uncertainty, detecting contradictions, and dynamically modulating reasoning behaviour.\u003c/p\u003e \u003cp\u003eWe evaluate the proposed system against a baseline single-pass LLM on tasks involving complex, ambiguous clinical narratives. Performance was assessed using metrics reflecting real-world requirements, including uncertainty coverage, contradiction detection, evidence grounding, confidence calibration, and governance quality. The meta-cognitive system demonstrated improved uncertainty coverage (0.88 vs 0.75), enhanced governance signals (0.96 vs 0.94), and more conservative and calibrated confidence estimates (0.57 vs 0.84), while maintaining comparable performance in evidence grounding and contradiction detection.\u003c/p\u003e \u003cp\u003eThese findings indicate that improvements in reliability are achieved not through scaling model capacity, but through architectural control of the reasoning process. By embedding meta-cognitive regulation within the system, it is possible to produce outputs that are better aligned with evidence, more transparent in their limitations, and more appropriate for high-stakes decision-making.\u003c/p\u003e \u003cp\u003eThis work provides a principled framework for integrating meta-cognition into artificial intelligence systems and suggests that future progress in trustworthy AI will depend on developing architectures that regulate how reasoning is performed, rather than solely improving what is generated.\u003c/p\u003e","manuscriptTitle":"Morphodynamic Meta-Cognition: A Framework for Self-Regulating Intelligence Systems","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-12 18:29:00","doi":"10.21203/rs.3.rs-9641646/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"bcf3abb9-c6c0-4095-9093-50832a4900c0","owner":[],"postedDate":"May 12th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":67708193,"name":"Bioinformatics"}],"tags":[],"updatedAt":"2026-05-12T18:29:00+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-12 18:29:00","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9641646","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9641646","identity":"rs-9641646","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}