CSCE: Boosting LLM Reasoning by Simultaneous Enhancing of Causal Significance and Consistency

CSCE: Boosting LLM Reasoning by Simultaneous Enhancing of Causal Significance and Consistency | 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 Research Article CSCE: Boosting LLM Reasoning by Simultaneous Enhancing of Causal Significance and Consistency Kangsheng Wang, Xiao Zhang, Juntao Lyu, Tianyu Hu, Huimin Ma This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6224619/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 Chain-based reasoning methods like chain of thought (CoT) play a rising role in solving reasoning tasks for large language models (LLMs). However, the causal hallucinations between a step of reasoning and corresponding state transitions are becoming a significant obstacle to advancing LLMs' reasoning capabilities, especially in long-range reasoning tasks. This paper proposes a non-chain-based reasoning framework for simultaneous consideration of causal significance and consistency, i.e., the Causal Significance and Consistency Enhancer (CSCE). We customize LLM's loss function utilizing treatment effect assessments to enhance its reasoning ability from two aspects: causal significance and consistency. This ensures that the model captures essential causal relationships and maintains robust and consistent performance across various scenarios. Additionally, we transform the reasoning process from the cascading multiple one-step reasoning commonly used in Chain-Based methods, like CoT, to a causal-enhanced method that outputs the entire reasoning process in one go, further improving the model's reasoning efficiency. Extensive experiments show that our method improves both the reasoning success rate and speed. These improvements further demonstrate that non-chain-based methods can also aid LLMs in completing reasoning tasks. Computer Architecture and Engineering Large language model Cause effect analysis Model-based reasoning Inference algorithms 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-6224619","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":428730559,"identity":"01a6de3b-a8a0-4b9c-81f0-d06169b441c6","order_by":0,"name":"Kangsheng Wang","email":"","orcid":"https://orcid.org/0009-0009-8392-4148","institution":"University of Science and Technology Beijing","correspondingAuthor":false,"prefix":"","firstName":"Kangsheng","middleName":"","lastName":"Wang","suffix":""},{"id":428730560,"identity":"3e5a1650-c4fb-473c-9ae8-1c8778587579","order_by":1,"name":"Xiao Zhang","email":"","orcid":"https://orcid.org/0009-0007-2804-6915","institution":"University of Science and Technology Beijing","correspondingAuthor":false,"prefix":"","firstName":"Xiao","middleName":"","lastName":"Zhang","suffix":""},{"id":428730561,"identity":"448fa1cd-4546-4401-9d54-e8842cc931b9","order_by":2,"name":"Juntao Lyu","email":"","orcid":"https://orcid.org/0009-0006-3374-047X","institution":"University of Science and Technology Beijing","correspondingAuthor":false,"prefix":"","firstName":"Juntao","middleName":"","lastName":"Lyu","suffix":""},{"id":428730562,"identity":"1364e8c4-4969-4bcc-833f-8d83c1e1d82b","order_by":3,"name":"Tianyu Hu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAy0lEQVRIiWNgGAWjYDACZjBpA8Q8QMxGhA4eiJY0oGqitUCowyRosWdnfvaYt+184vz5vQcYPpQdZuCf3UDIYWzmxrxttxM3HONLYJxx7jCDxJ0DhLQwmEmDtbDxGDDzth1mMJBIIKSF/RtQy7nE+W1ALX+J08IDsuVAYsMxoBZGorQc5imTnHMu2XjDsbyEgz3n0nkkbhDQwt5/fJvEmzI72fnNZw8++FFmLcc/g4AWEGCCRg7DAQZ4RBEAjD+IUjYKRsEoGAUjFgAAYG87IrOhcPsAAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0001-9903-0696","institution":"University of Science and Technology Beijing","correspondingAuthor":true,"prefix":"","firstName":"Tianyu","middleName":"","lastName":"Hu","suffix":""},{"id":428730563,"identity":"8bddaaa4-189c-4a73-9f3e-d8ba097d1529","order_by":4,"name":"Huimin Ma","email":"","orcid":"https://orcid.org/0000-0001-5383-5667","institution":"University of Science and Technology Beijing","correspondingAuthor":false,"prefix":"","firstName":"Huimin","middleName":"","lastName":"Ma","suffix":""}],"badges":[],"createdAt":"2025-03-14 08:38:19","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-6224619/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6224619/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":79144914,"identity":"72399ef2-d611-4cf5-a2c7-29ac93ebe9c9","added_by":"auto","created_at":"2025-03-25 02:50:37","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":420935,"visible":true,"origin":"","legend":"","description":"","filename":"conferencelatextemplateicmecameraPreprint.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6224619/v1_covered_83d12d9e-c80f-49e3-870c-4fb170fe1bef.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eCSCE: Boosting LLM Reasoning by Simultaneous Enhancing of Causal Significance and Consistency\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"University of Science and Technology Beijing","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":"Large language model, Cause effect analysis, Model-based reasoning, Inference algorithms","lastPublishedDoi":"10.21203/rs.3.rs-6224619/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6224619/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eChain-based reasoning methods like chain of thought (CoT) play a rising role in solving reasoning tasks for large language models (LLMs). However, the causal hallucinations between a step of reasoning and corresponding state transitions are becoming a significant obstacle to advancing LLMs' reasoning capabilities, especially in long-range reasoning tasks. This paper proposes a non-chain-based reasoning framework for simultaneous consideration of causal significance and consistency, i.e., the Causal Significance and Consistency Enhancer (CSCE). We customize LLM's loss function utilizing treatment effect assessments to enhance its reasoning ability from two aspects: causal significance and consistency. This ensures that the model captures essential causal relationships and maintains robust and consistent performance across various scenarios. Additionally, we transform the reasoning process from the cascading multiple one-step reasoning commonly used in Chain-Based methods, like CoT, to a causal-enhanced method that outputs the entire reasoning process in one go, further improving the model's reasoning efficiency. Extensive experiments show that our method improves both the reasoning success rate and speed. These improvements further demonstrate that non-chain-based methods can also aid LLMs in completing reasoning tasks.\u003c/p\u003e","manuscriptTitle":"CSCE: Boosting LLM Reasoning by Simultaneous Enhancing of Causal Significance and Consistency","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-03-25 02:42:30","doi":"10.21203/rs.3.rs-6224619/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":"a6392f3e-4443-42fb-9d76-a069225b2ce8","owner":[],"postedDate":"March 25th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":45674652,"name":"Computer Architecture and Engineering"}],"tags":[],"updatedAt":"2025-03-25T02:42:30+00:00","versionOfRecord":[],"versionCreatedAt":"2025-03-25 02:42:30","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6224619","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6224619","identity":"rs-6224619","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}