[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-detail-290351-en":59,"doc-seo-290351-105":80},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":5,"data":60},{"doc_id":61,"user_id":62,"nickname":63,"user_avatar":64,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":66,"doc_content":67,"file_id":68,"file_url":69,"file_type":70,"file_size":71,"view_count":72,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":72,"language":73,"language_code":74,"site_id":75,"html_lang":74,"table_of_contents":76,"faqs":77,"seo_title":78,"seo_description":66,"update_tm":79,"read_time":41,"preview_page_count":4},290351,4398048949847,"Eliana","https://ap-avatar.wpscdn.com/avatar/400002536579ef2da7f?_k=1778318612642679267","Do Machine Learning Models Represent Their Targets?","Sullivan argues that machine learning models in science should be understood as highly idealized toy models. By treating ML models this way, the paper applies established representational and epistemic strategies from toy-model scholarship to explain how ML can still yield epistemic success even when models lack similarity to their targets. The central aim is to address the underexplored representational problem that arises if ML fails to represent real-world targets in any meaningful sense.","Edinburgh Research Explorer  \nDo machine learning models represent their targets?  \nCitation for published version:  \nSullivan, E 2024, 'Do machine learning models represent their targets?', Philosophy of Science, vol. 91, no.  \n5, pp. 1445-1455. [https://doi.org/10.1017/psa.2023.151](https://doi.org/10.1017/psa.2023.151)  \nDigital Object Identifier (DOI):  \n10.1017/psa.2023.151  \nLink:  \nLink to publication record in Edinburgh Research Explorer  \nDocument Version:  \nPublisher's PDF, also known as Version of record  \nPublished In:  \nPhilosophy of Science  \nGeneral rights  \nCopyright for the publications made accessible via the Edinburgh Research Explorer is retained by the author(s) and / or other copyright owners and it is a condition of accessing these publications that users recognise and abide by the legal requirements associated with these rights.  \nTake down policy  \nThe University of Edinburgh has made every reasonable effort to ensure that Edinburgh Research Explorer content complies with UK legislation. If you believe that the public display of this file breaches copyright please [contact openaccess@ed.ac.uk](contact openaccess@ed.ac.uk) providing details, and we will remove access to the work immediately and investigate your claim.  \nDownload date: 27. Jul. 2026  \nPhilosophy of Science (2024), 91, 1445–1455 doi:10.1017/psa.2023.151  \nSYMPOSIA PAPER  \nDo Machine Learning Models Represent Their Targets?  \nEmily Sullivan  \nUtrecht University, The Netherlands Email: [e.e.sullivan@uu.nl](e.e.sullivan@uu.nl)  \n(Received 20 April 2023; revised 22 September 2023; accepted 05 October 2023; first published online 20 October 2023)  \nAbstract  \nI argue that machine learning (ML) models used in science function as highly idealized toy models. If we treat ML models as a type of highly idealized toy model, then we can deploy standard representational and epistemic strategies from the toy model literature to explain why ML models can still provide epistemic success despite their lack of similarity to their targets.  \n1. Introduction  \nMost attention on complex machine learning (ML) models used in science has centered around issues of opacity, such as the nature of opacity (Creel 2020; Boge 2022) and its epistemic consequences for science (Duede 2023) .1 While some have argued that ML models can still provide understanding of phenomena despite their opacity (Meskhidze 2021; Sullivan 2022a), others demur (Räz and Beisbart 2022) . However, before the epistemic consequences of opacity become salient, there is an underexplored prior question of representation. If ML models used in science do not represent real-world targets in any meaningful sense, how can ML models provide understanding in the first place?  \nThe problem is that it seems as though ML models do not represent their targets in any meaningful sense. For example, the similarity view of representation seems to exclude the possibility that ML models can represent phenomena. According to the similarity view (Mäki 2009; Giere 2004; Weisberg 2013), for a model to represent some phenomenon requires that the model be sufficiently similar to its target. However, ML models use methods of finding feature relationships that are highly divorced from  \n1 In this paper, my focus is on deep learning neural network architectures. However, analogous arguments could be made toward random forests or Bayesian nets, or other ML techniques that engage in idealization.  \n© The Author(s), 2023 . Published by Cambridge University Press on behalf of the Philosophy of Science Association. This is an Open Access article, distributed under the terms of the Creative Commons Attribution-NonCommercial-ShareAlike licence ([http://creativecommons.org/licenses/by-nc-sa/4.0/](http://creativecommons.org/licenses/by-nc-sa/4.0/)), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the same Creative Commons licence is used to distribute the re-used or adapted article and the","cbCainKQxedEDzO5","https://ap.wps.com/l/cbCainKQxedEDzO5","pdf",301433,12,"English","en",105,"# Introduction\n## Opacity and the underexplored problem of representation\n## Similarity view of representation and why it threatens ML representation\n## ML representation hypothesis","[{\"question\":\"What is the main question the paper addresses?\",\"answer\":\"Whether machine learning models represent their targets in any meaningful sense, and how they can still support understanding in science if they do not.\"},{\"question\":\"What does the paper argue instead of the similarity view?\",\"answer\":\"It argues that ML models function representationally and epistemically like highly idealized toy models, enabling epistemic success despite lacking similarity to targets.\"},{\"question\":\"What hypothesis does the paper aim to reject?\",\"answer\":\"The ML representation hypothesis: complex or opaque ML models fail to enable understanding of real-world phenomena because they are not similar to and therefore fail to represent their targets.\"}]","Do Machine Learning Models Represent Their Targets? | PDF",1789641444,{"code":4,"msg":81,"data":82},"ok",{"site_id":75,"language":74,"slug":83,"title":65,"keywords":84,"description":66,"schema_data":85,"social_meta":139,"head_meta":141,"extra_data":143,"updated_unix":79},"do-machine-learning-models-represent-their-targets","",{"@graph":86,"@context":138},[87,101,121],{"@type":88,"itemListElement":89},"BreadcrumbList",[90,94,96,99],{"item":91,"name":92,"@type":93,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":95,"name":9,"@type":93,"position":14},"https://docshare.wps.com/document/",{"item":97,"name":40,"@type":93,"position":98},"https://docshare.wps.com/document/research-report/",3,{"item":100,"name":65,"@type":93,"position":19},"https://docshare.wps.com/document/do-machine-learning-models-represent-their-targets/290351/",{"url":100,"name":65,"@type":102,"image":103,"author":108,"headline":65,"publisher":110,"fileFormat":113,"inLanguage":74,"description":66,"dateModified":114,"datePublished":115,"encodingFormat":113,"isAccessibleForFree":116,"interactionStatistic":117},"DigitalDocument",{"url":104,"@type":105,"width":106,"height":107},"https://docshare.wps.com/thumbnails/do-machine-learning-models-represent-their-targets/290351.png","ImageObject",300,407,{"name":63,"@type":109},"Person",{"url":91,"name":111,"@type":112},"DocShare","Organization","application/pdf","2026-11-11","2026-09-17",true,{"@type":118,"interactionType":119,"userInteractionCount":72},"InteractionCounter",{"@type":120},"ViewAction",{"@type":122,"mainEntity":123},"FAQPage",[124,130,134],{"name":125,"@type":126,"acceptedAnswer":127},"What is the main question the paper addresses?","Question",{"text":128,"@type":129},"Whether machine learning models represent their targets in any meaningful sense, and how they can still support understanding in science if they do not.","Answer",{"name":131,"@type":126,"acceptedAnswer":132},"What does the paper argue instead of the similarity view?",{"text":133,"@type":129},"It argues that ML models function representationally and epistemically like highly idealized toy models, enabling epistemic success despite lacking similarity to targets.",{"name":135,"@type":126,"acceptedAnswer":136},"What hypothesis does the paper aim to reject?",{"text":137,"@type":129},"The ML representation hypothesis: complex or opaque ML models fail to enable understanding of real-world phenomena because they are not similar to and therefore fail to represent their targets.","https://schema.org",{"og:url":100,"og:type":140,"og:title":65,"og:site_name":111,"og:description":66},"article",{"robots":142,"canonical":100},"index,follow",{"doc_id":61,"site_id":75}]