[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-85864-en":3,"doc-seo-85864-105":30,"detail-sidebar-cat-0-en-105":83},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":13,"seo_description":14,"update_tm":28,"read_time":29},85864,1099514068035,"Ezra","https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c",8,"Research & Report","Behavioural Signatures of Risk-Sensitive Decision Making in Large Language Models","Large language models used for decision support must exhibit stable, interpretable behaviour under uncertainty. This study tests whether LLM choices show human-like risk structure using a controlled multi-model setting based on no-limit Texas Hold’em. Behaviour is quantified by Participation (voluntary engagement in uncertain opportunities) and Proactiveness (pre-flop risk escalation). Frontier LLMs form stable, model-specific profiles spanning conservative to aggressive styles, largely robust to opponent composition but diverging in mixed interactions. Under global risk pressure and personal resource constraints, models adapt in structured yet heterogeneous ways—supporting behavioural auditing of risk-sensitive decision-making in interactive environments.","arXiv :2607 . 1025 1v 1 [ cs .AI] 11 Jul 2026  \nBehavioural Signatures of Risk-Sensitive Decision-Making in Large Language Models  \nXuankun Rong 1 , Wenke Huang2 , Bo Du 1 , Dacheng Tao2 , and Mang Ye 1  \n1 School of Computer Science, Wuhan University, Wuhan, China  \n2 College of Computing and Data Science, Nanyang Technological University, Singapore  \nABSTRACT  \nAs large language models (LLMs) are increasingly used in decision support, it is important to understand whether their choices under uncertainty exhibit stable and interpretable behavioural regularities. Human decision-making combines relatively persistent risk preferences with context-dependent adjustment, yet it remains unclear whether analogous behavioural structure can be observed in LLM-based decision systems. Here we examine this question using a controlled multi-model framework based on no-limit Texas Hold’em, where behaviour is quantified by Participation, measuring voluntary engagement in uncertain opportunities, and Proactiveness, measuring pre-flop risk escalation. Across homogeneous self-play and heterogeneous mixedmodel interactions, frontier LLMs exhibit stable, model-specific risk profiles, forming a spectrum from conservative to aggressive decision styles. These profiles remain largely robust under changing opponent composition, while the most conservative and most aggressive models diverge further in mixed settings. Under global risk pressure and personal resource constraint, models adapt in structured but heterogeneous ways, ranging from broad behavioural contraction to selective de-escalation and near-invariant behaviour. These findings suggest that LLMs differ not only in baseline risk disposition, but also in the risk signals they respond to and the flexibility with which they adjust, providing a behavioural basis for auditing risk-sensitive decision-making in interactive settings. Our code is publicly available at: [https://github.com/XuankunRong/AgentTexasPoker](https://github.com/XuankunRong/AgentTexasPoker).  \nDecision-making under uncertainty is a fundamental characteristic of complex systems, shaping both human societies and, increasingly, artificial intelligence (AI) systems 1, 2. In human contexts, individuals and groups continuously make choices, from resource allocation in early cooperative settings to risk evaluation in modern complex environments. As AI systems advance, large language models (LLMs) are increasingly moving beyond passive information processing and becoming embedded in workflows that require repeated choices, actionselection and responses to uncertain outcomes3–6. Across high-stakes domains such as medical diagnosis and financial decision-making, as well as everyday settings such as information selection and interactive assistance, these systems are increasingly involved in processes where decisions carry meaningful consequences.  \nClassical theories often model humans as rational decisionmakers who maximize expected utility under constraints7, 8. However, a large body of empirical evidence shows that realworld decision-making systematically deviates from this rational paradigm. Such deviations often appear as asymmetric risk perception, loss sensitivity and context-dependent behavioural adjustment9, 10. Behavioural decision research further shows that individuals differ in relatively stable risk attitudes, while their observed risk-taking also varies with domain, framing and environmental context 11–13. Together, these findings indicate that human decision-making is shaped by both persistent dispositional tendencies and adaptive responses to changing risk conditions.  \nAs LLM-based systems become part of decision-making processes, evaluating their behaviour requires more than measuring task performance or reasoning accuracy. They may  \nalso differ in how readily they enter uncertain opportunities, how strongly they escalate risk and how flexibly they adjust when the decision environment changes4–6, 14. This raises a beh","cbCaioRHjpVNu3kU","https://ap.wps.com/l/cbCaioRHjpVNu3kU","pdf",2411269,3,1,27,"English","en",105,"# Abstract\n# From human decisions to AI-assisted decisions\n## Behaviour assay","[{\"question\":\"How do the models respond when risk conditions change?\",\"answer\":\"Under global risk pressure and personal resource constraint, models adapt in structured but heterogeneous ways, including broad behavioural contraction and selective de-escalation, with near-invariant behaviour in some cases.\"}]",1784206780,68,{"code":4,"msg":31,"data":32},"ok",{"site_id":25,"language":24,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":78,"head_meta":80,"extra_data":82,"updated_unix":28},"behavioural-signatures-of-risk-sensitive-decision-making-in-large-language-models","",{"@graph":36,"@context":77},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,50],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":20},"https://docshare.wps.com/document/research-report/",{"item":51,"name":13,"@type":43,"position":52},"https://docshare.wps.com/document/behavioural-signatures-of-risk-sensitive-decision-making-in-large-language-models/85864/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":24,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-22","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71],{"name":72,"@type":73,"acceptedAnswer":74},"How do the models respond when risk conditions change?","Question",{"text":75,"@type":76},"Under global risk pressure and personal resource constraint, models adapt in structured but heterogeneous ways, including broad behavioural contraction and selective de-escalation, with near-invariant behaviour in some cases.","Answer","https://schema.org",{"og:url":51,"og:type":79,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":81,"canonical":51},"index,follow",{"doc_id":7,"site_id":25},{"code":4,"msg":5,"data":84},[85,89,93,97,102,107,112,115,120,123,127],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":86,"show_sort_weight":87,"slug":88},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":90,"show_sort_weight":91,"slug":92},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Exam",70,"exam",{"id":98,"doc_module":4,"doc_module_name":46,"category_name":99,"show_sort_weight":100,"slug":101},5,"Comic",60,"comic",{"id":103,"doc_module":4,"doc_module_name":46,"category_name":104,"show_sort_weight":105,"slug":106},6,"Technology",50,"technology",{"id":108,"doc_module":4,"doc_module_name":46,"category_name":109,"show_sort_weight":110,"slug":111},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":113,"slug":114},30,"research-report",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":46,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":98,"slug":130},19,"General","general"]