[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-203751-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-203751-en":131},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","from-language-to-cognition-how-llms-outgrow-the-human-language-network","From Language to Cognition - How LLMs Outgrow the Human Language Network","","Large language models (LLMs) mirror neural activity in the human language network, yet the linguistic properties that drive this alignment are unclear. This study benchmarks 34 training checkpoints across 8 model sizes and 300B tokens to connect brain alignment with linguistic competence. Results show stronger tracking of formal linguistic competence (rules) than functional competence (world knowledge and reasoning). As models surpass human-level proficiency, correlations with next-word prediction, behavioral alignment, and brain alignment fade, while benchmarks remain unsaturated, guiding future modeling improvements.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/from-language-to-cognition-how-llms-outgrow-the-human-language-network/203751/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/from-language-to-cognition-how-llms-outgrow-the-human-language-network/203751.png","ImageObject",300,407,{"name":42,"@type":43},"McQueen","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-10-08","2026-09-04",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",13,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"What does the study investigate about LLMs and the human language network?","Question",{"text":63,"@type":64},"It analyzes which linguistic properties and training dynamics cause LLMs to align with neural responses in the human language network, using controlled large-scale benchmarking across many training checkpoints.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"How does brain alignment relate to formal versus functional linguistic competence?",{"text":68,"@type":64},"Brain alignment tracks the development of formal linguistic competence more closely than functional linguistic competence. Functional competence continues improving later in training but aligns more weakly with neural activity.",{"name":70,"@type":61,"acceptedAnswer":71},"When models become more capable than humans, what happens to alignment signals?",{"text":72,"@type":64},"The correlation between next-word prediction, behavioral alignment, and brain alignment fades once models surpass human language proficiency, indicating a decoupling at later stages.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},203751,1788563199,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,115,120,123,127],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":112,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":113,"slug":114},8,30,"research-report",{"id":116,"doc_module":4,"doc_module_name":25,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":25,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":25,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":25,"category_name":129,"show_sort_weight":97,"slug":130},19,"General","general",{"code":4,"msg":82,"data":132},{"doc_id":79,"user_id":133,"nickname":42,"user_avatar":134,"doc_module":4,"category_id":112,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":128,"language":140,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":12,"update_tm":80,"read_time":144},5909890329169,"https://ap-avatar.wpscdn.com/davatar_9964176cb1d06d4a9deccf72a44ae3dc","From Language to Cognition:  \nHow LLMs Outgrow the Human Language Network  \nBadr AlKhamissi 1 Greta Tuckute2 Yingtian Tang 1 Taha Binhuraib3  \nAntoine Bosselut∗ , 1 Martin Schrimpf∗ , 1  \n1EPFL 2MIT 3 Georgia Institute of Technology  \narXiv :2503 .01830v2 [ cs .CL] 20 Sep 2025  \nAbstract  \nLarge language models (LLMs) exhibit remarkable similarity to neural activity in the human language network. However, the key properties of language underlying this alignment—and how brain-like representations emerge and change across training—remain unclear. We here benchmark 34 training checkpoints spanning 300B tokens across 8 different model sizes to analyze how brain alignment relates to linguistic competence. Specifically, we find that brain alignment tracks the development of formal linguistic competence—i.e., knowledge of linguistic rules—more closely than functional linguistic competence. While functional competence, which involves world knowledge and reasoning, continues to develop throughout training, its relationship with brain alignment is weaker, suggesting that the human language network primarily encodes formal linguistic structure rather than broader cognitive functions. Notably, we find that the correlation between next-word prediction, behavioral alignment, and brain alignment fades once models surpass human language proficiency. We further show that model size is not a reliable predictor of brain alignment when controlling for the number of features. Finally, using the largest set of rigorous neural language benchmarks to date, we show that language brain alignment benchmarks remain unsaturated, highlighting opportunities for improving future models. Taken together, our findings suggest that the human language network is best modeled by formal, rather than functional, aspects of language.1  \n1 Introduction  \nDeciphering the brain’s algorithms underlying our ability to process language and communicate is a  \n∗ Equal Supervision  \n[1](1Project Page: language-to-cognition.epfl.ch)[Project Page:](1Project Page: language-to-cognition.epfl.ch)[ language-to-cognition.epfl.ch](1Project Page: language-to-cognition.epfl.ch)  \nFigure 1: Model Alignment with the Human Language Network is Primarily Driven by Formal than Functional Linguistic Competence. (a) Average brain alignment across five Pythia models and five brain recording datasets, normalized by cross-subject consistency, throughout training. (b) Average normalized accuracy of the same models on formal linguistic competence benchmarks (two benchmarks) . (c) Average normalized accuracy on functional linguistic competence benchmarks (six benchmarks) . The x-axis is logarithmically spaced up to 16B tokens, capturing early training dynamics, and then evenly spaced every 20B tokens from 20B to ~300B tokens.  \ncore goal in neuroscience. Human language processing is supported by the brain’s language network (LN), a set of left-lateralized fronto-temporal regions in the brain (Binder et al., 1997 ; Bates et al., 2003 ; Gorno-Tempini et al., 2004 ; Price, 2010 ; Fedorenko, 2014 ; Hagoort, 2019) that respond robustly and selectively to linguistic input (Fedorenko et al., 2024a) . Driven by recent advances in machine learning, large language models (LLMs) trained via next-word prediction on large corpora of text are now a particularly promising model family to capture the internal processes of the LN. In particular, when these models are exposed to the same linguistic stimuli (e.g., sen-  \ntences or narratives) as human participants during neuroimaging and electrophysiology experiments, they account for a substantial portion of neural response variance (Schrimpf et al., 2021 ; Caucheteux and King, 2022 ; Goldstein et al., 2022 ; Pasquiou et al., 2022 ; Aw et al., 2023 ; Tuckute et al., 2024a ; AlKhamissi et al., 2025 ; Rathi et al., 2025) .  \n1.1 Key Questions and Contributions  \nThis work investigates four key questions, all aimed at distilling why LLM aligns to brain responses. Specif","cbCaioAIMxlT6xJ3","https://ap.wps.com/l/cbCaioAIMxlT6xJ3","pdf",22312852,"English","# Abstract\n# Introduction\n## Key Questions and Contributions","[{\"question\":\"What does the study investigate about LLMs and the human language network?\",\"answer\":\"It analyzes which linguistic properties and training dynamics cause LLMs to align with neural responses in the human language network, using controlled large-scale benchmarking across many training checkpoints.\"},{\"question\":\"How does brain alignment relate to formal versus functional linguistic competence?\",\"answer\":\"Brain alignment tracks the development of formal linguistic competence more closely than functional linguistic competence. Functional competence continues improving later in training but aligns more weakly with neural activity.\"},{\"question\":\"When models become more capable than humans, what happens to alignment signals?\",\"answer\":\"The correlation between next-word prediction, behavioral alignment, and brain alignment fades once models surpass human language proficiency, indicating a decoupling at later stages.\"}]","From Language to Cognition - How LLMs Outgrow the Human Language Network | PDF",48]