[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-84816-en":3,"doc-seo-84816-105":29,"detail-sidebar-cat-0-en-105":90},{"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":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},84816,2336464648322,"Aria","https://ap-avatar.wpscdn.com/avatar/2200025388227c56fec?_k=1778556882303663488",8,"Research & Report","How Personas Can Influence Agents to Play Split or Steal","Personas are often used to steer large language model (LLM) agents, yet their ability to reliably shape strategic behavior in social dilemmas is not well established. This study tests persona prompts in an iterated Split or Steal game where persona-driven agents interact with a Virtual Human governed by a fixed prompt. Across 160 sessions of 15 rounds in European Portuguese, mutual Split dominates (~74%). Model choice matters: phi4 and Ministral 3:3b stay cooperative, while Gemma models vary. Big Five analysis links Prosocial/Principled personas to cooperation and Analytical personas to more exploitation; dialogue topics correlate with Split vs Steal outcomes.","arXiv :2607 .05398v1 [ cs .CL] 3 May 2026  \nHow Personas Can Influence Agents to Play Split or Steal  \nCarlos Leon 1 ,2 ,3 ,4 Alexandre Rodrigues 1 ,2 ,3 ,4 Pedro Gamito 1 Thomas D. Parsons4  \n1 Universidade Lusófona, Portugal  \n2 Universitat de Barcelona, Spain  \n3 Université Paris Cité, France  \n4 Computational Neuropsychology and Simulation (CNS) Laboratory, Arizona State University, USA  \n[cleonmar7@alumnes.ub.edu](cleonmar7@alumnes.ub.edu) [alex26andre09@gmail.com](alex26andre09@gmail.com)  \n 0009-0002-8220-1561  0009-0009-7204-986X  0000-0003-0585-8447  \n 0000-0003-0331-5019  \nAbstract   \nPersonas are often employed to guide large language model agents, yet their effectiveness in shaping strategic behavior in social dilemma settings remains uncertain. To address this, we examined the impact of persona prompts in an iterated Split or Steal game where persona-driven agents interacted with a Virtual Human (VH) controlled by a fixed prompt. Agents were instantiated from four open models (Ministral 3:3b, phi4:14b, Gemma3:12b, and Gemma4:e4b) at two temperature settings (0.3 and 0.7) and deterministic decision with zero temperature, while the VH was powered by GPT 4.1 mini. Across 160 sessions of 15 rounds each conducted in European Portuguese, mutual Split outcomes dominated (roughly 74 percent of rounds), with exploitation occurring in fewer than 11 percent of rounds. Model choice significantly influenced behavior: phi4 and Ministral 3:3b remained consistently cooperative across temperatures, whereas Gemma3:12b and Gemma4:e4b exhibited more varied strategies and outcomes. Analyses based on Big Five personality traits indicated that Prosocial and Principled personas were most consistently cooperative, while Analytical personas were more likely to exploit the VH. Topic analysis revealed that friendship-related dialogue aligns with Split decisions, whereas money and vengeance-related content is more prevalent in Steal outcomes; sentiment labels were predominantly neutral or happy and provided limited additional explanatory value. These findings characterize the interaction between persona prompts and model differences in repeated trust games and serve as a baseline for planned virtual reality studies involving human participants interacting with an embodied VH.  \nKeywords: large language models; persona prompting; split or steal; social dilemma games; cooperation; virtual humans  \n1. Introduction  \nPersona prompting is widely used to steer large language model agents, but its impact on strategic behaviour in repeated social dilemmas remains unclear.  \nWe examine this question in an iterated Split or Steal game where persona steered agents interact with a Virtual Human (VH) driven by a fixed prompt. Using four open models for the agents and varying temperature, we test how outcomes and inferred strategies shift with model choice, persona group, and conversational content. Our goal is to characterise when persona cues meaningfully change cooperation, exploitation, and strategy signatures, and to provide a baseline for a follow-up virtual reality study in which a visually embodied VH will interact with human participants. The chosen approach for measuring trust iteratively is a simplified version of the Prisoner’s Dilemma, implemented here as a Split or Steal game. We chose this format because it preserves the basic cooperation versus defection structure while being easier to embed in a conversational setting Vanden Assem et al. (2012) . This makes it suitable for studying how persona cues influence trust, exploitation, and repeated decision patterns across rounds. The iterated design also lets us observe whether players remain consistent or adapt their strategy over time.  \nTo make repeated play interpretable, we summarize each session using cooperation rate and switch rate. Cooperation rate captures the overall tendency to Split, while switch rate captures short-term responsiveness across rounds, making them useful descriptors o","cbCaipQaAbgvh86I","https://ap.wps.com/l/cbCaipQaAbgvh86I","pdf",2082419,1,25,"English","en",105,"# Introduction\n## LLMs in Repeated Trust Games\n## System Prompts and Persona Induction","[{\"question\":\"How does the study evaluate whether persona prompts influence strategic behavior?\",\"answer\":\"It runs an iterated Split or Steal game in which persona-steered agents interact with a Virtual Human controlled by a fixed prompt, then measures outcomes across rounds and sessions.\"},{\"question\":\"What experimental conditions were used for the persona-driven agents and the Virtual Human?\",\"answer\":\"Four open agent models (Ministral 3:3b, phi4:14b, Gemma3:12b, Gemma4:e4b) are tested under temperature settings 0.3 and 0.7 (and deterministic zero temperature), while the Virtual Human is powered by GPT 4.1 mini.\"},{\"question\":\"Which persona traits and conversation topics are most associated with cooperative vs exploitative outcomes?\",\"answer\":\"Prosocial and Principled personas show the most consistent cooperation, while Analytical personas exploit the Virtual Human more often. Friendship-related dialogue aligns with Split decisions, whereas money and vengeance-related content is more prevalent in Steal outcomes.\"}]",1784198476,63,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":85,"head_meta":87,"extra_data":89,"updated_unix":27},"how-personas-can-influence-agents-to-play-split-or-steal","",{"@graph":35,"@context":84},[36,53,67],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/how-personas-can-influence-agents-to-play-split-or-steal/84816/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":61,"encodingFormat":60,"isAccessibleForFree":62,"interactionStatistic":63},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-16",true,{"@type":64,"interactionType":65,"userInteractionCount":4},"InteractionCounter",{"@type":66},"ViewAction",{"@type":68,"mainEntity":69},"FAQPage",[70,76,80],{"name":71,"@type":72,"acceptedAnswer":73},"How does the study evaluate whether persona prompts influence strategic behavior?","Question",{"text":74,"@type":75},"It runs an iterated Split or Steal game in which persona-steered agents interact with a Virtual Human controlled by a fixed prompt, then measures outcomes across rounds and sessions.","Answer",{"name":77,"@type":72,"acceptedAnswer":78},"What experimental conditions were used for the persona-driven agents and the Virtual Human?",{"text":79,"@type":75},"Four open agent models (Ministral 3:3b, phi4:14b, Gemma3:12b, Gemma4:e4b) are tested under temperature settings 0.3 and 0.7 (and deterministic zero temperature), while the Virtual Human is powered by GPT 4.1 mini.",{"name":81,"@type":72,"acceptedAnswer":82},"Which persona traits and conversation topics are most associated with cooperative vs exploitative outcomes?",{"text":83,"@type":75},"Prosocial and Principled personas show the most consistent cooperation, while Analytical personas exploit the Virtual Human more often. Friendship-related dialogue aligns with Split decisions, whereas money and vengeance-related content is more prevalent in Steal outcomes.","https://schema.org",{"og:url":51,"og:type":86,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":88,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":91},[92,96,100,104,109,114,119,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":93,"show_sort_weight":94,"slug":95},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":97,"show_sort_weight":98,"slug":99},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":101,"show_sort_weight":102,"slug":103},"Exam",70,"exam",{"id":105,"doc_module":4,"doc_module_name":45,"category_name":106,"show_sort_weight":107,"slug":108},5,"Comic",60,"comic",{"id":110,"doc_module":4,"doc_module_name":45,"category_name":111,"show_sort_weight":112,"slug":113},6,"Technology",50,"technology",{"id":115,"doc_module":4,"doc_module_name":45,"category_name":116,"show_sort_weight":117,"slug":118},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":120,"slug":121},30,"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":45,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":45,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":45,"category_name":136,"show_sort_weight":105,"slug":137},19,"General","general"]