Implicit Bias-Like Patterns in Reasoning Models

Implicit Bias-Like Patterns in Reasoning Models | 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Download PDF Article Implicit Bias-Like Patterns in Reasoning Models Messi Lee, Calvin Lai This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7726631/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 Implicit biases refer to automatic mental processes that shape perceptions, judgments, and behaviors. Previous research on "implicit bias" in LLMs focused primarily on outputs rather than the processes underlying the outputs. We present the Reasoning Model Implicit Association Test (RM-IAT) to study implicit bias-like processing in reasoning models, which are LLMs that use step-by-step reasoning for complex tasks. Using RM-IAT, we find that reasoning models like o3-mini, DeepSeek-R1, gpt-oss-20b, and Qwen-3 8B consistently expend more reasoning tokens on association-incompatible tasks than association-compatible tasks, suggesting greater computational effort when processing counter-stereotypical information. In contrast, Claude 3.7 Sonnet exhibited reversed or inconsistent patterns, likely due to embedded safety mechanisms that flagged or rejected socially sensitive associations. These divergent behaviors highlight important differences in how alignment and safety processes shape model reasoning. As reasoning models become increasingly integrated into real-world decision-making, understanding their implicit bias-like patterns and how alignment methods influence them is crucial for ensuring fair and trustworthy AI systems. Social science/Psychology/Human behaviour Social science/Science, technology and society Full Text Additional Declarations There is NO Competing Interest. 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-7726631","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":552295606,"identity":"f0585ce0-edab-4c9c-aef0-38c1184423d5","order_by":0,"name":"Messi Lee","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAz0lEQVRIiWNgGAWjYBACAwaGBAaGAgYGfjAXyGA4QJQWICnZAFJtQJwWCGlwgFgt5uwNz6QLDGzkjG/kGH/+YMAgx3cjAb8Wy54DadIzDNKMzW7kmEkAbTGWJKTF4EZCmjSPweHEbbdzzEAOS9xApJb/9Ztn5xh/AGqpJ1bLgQQD6RwDkMMSDAhqOXMg2ZrHINlwxv1nZRJnDCQMZ555QEDL8Z7E2zwVdvL8PYc3f6iosJHnO07AFgYGHhQVEoSUgwD7AWJUjYJRMApGwUgGAL/yRVkZBOblAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0003-3096-1112","institution":"Washington University in St.Louis","correspondingAuthor":true,"prefix":"","firstName":"Messi","middleName":"","lastName":"Lee","suffix":""},{"id":552295607,"identity":"448447fd-98c2-4a11-894b-bd0e1919ee50","order_by":1,"name":"Calvin Lai","email":"","orcid":"","institution":"Rutgers University","correspondingAuthor":false,"prefix":"","firstName":"Calvin","middleName":"","lastName":"Lai","suffix":""}],"badges":[],"createdAt":"2025-09-27 07:15:24","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7726631/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7726631/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":97251304,"identity":"0f43a821-1939-4d67-aced-02ea8b7721d1","added_by":"auto","created_at":"2025-12-02 13:16:45","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":633587,"visible":true,"origin":"","legend":"Article File","description":"","filename":"ImplicitBiasinReasoningModels.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7726631/v1_covered_7f5705f3-0805-42d8-9634-2694155dfc4a.pdf"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Implicit Bias-Like Patterns in Reasoning Models","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-7726631/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7726631/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Implicit biases refer to automatic mental processes that shape perceptions, judgments, and behaviors. Previous research on \"implicit bias\" in LLMs focused primarily on outputs rather than the processes underlying the outputs. We present the Reasoning Model Implicit Association Test (RM-IAT) to study implicit bias-like processing in reasoning models, which are LLMs that use step-by-step reasoning for complex tasks. Using RM-IAT, we find that reasoning models like o3-mini, DeepSeek-R1, gpt-oss-20b, and Qwen-3 8B consistently expend more reasoning tokens on association-incompatible tasks than association-compatible tasks, suggesting greater computational effort when processing counter-stereotypical information. In contrast, Claude 3.7 Sonnet exhibited reversed or inconsistent patterns, likely due to embedded safety mechanisms that flagged or rejected socially sensitive associations. These divergent behaviors highlight important differences in how alignment and safety processes shape model reasoning. As reasoning models become increasingly integrated into real-world decision-making, understanding their implicit bias-like patterns and how alignment methods influence them is crucial for ensuring fair and trustworthy AI systems.","manuscriptTitle":"Implicit Bias-Like Patterns in Reasoning Models","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-02 11:11:29","doi":"10.21203/rs.3.rs-7726631/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-machine-intelligence","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"natmachintell","sideBox":"Learn more about [Nature Machine Intelligence](http://www.nature.com/natmachintell/)","snPcode":"","submissionUrl":"","title":"Nature Machine Intelligence","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Research","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"d719d490-52d8-4988-aeb4-a7449ba52b7b","owner":[],"postedDate":"December 2nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":58776386,"name":"Social science/Psychology/Human behaviour"},{"id":58776387,"name":"Social science/Science, technology and society"}],"tags":[],"updatedAt":"2026-04-03T05:30:22+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-02 11:11:29","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7726631","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7726631","identity":"rs-7726631","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}