Computers in Human Behavior: Artificial Humans最新文献

{"title":"The danger in easy answers: Conceptualizing subjective processing fluency for healthcare chatbot information presentation in an inductive reflexive thematic analysis of user experience and perceptions","authors":"Samuel N. Koscelny,&nbsp;Sara Sadralashrafi,&nbsp;David M. Neyens","doi":"10.1016/j.chbah.2026.100370","DOIUrl":"10.1016/j.chbah.2026.100370","url":null,"abstract":"<div><div>The rapid advancement of AI technologies has introduced complex challenges for designers, particularly in understanding the theoretical foundations that guide AI design across different contexts. To bridge this gap, this study investigated the impact of communication style (conversational vs. informative) in a healthcare chatbot setting. Participants interacted with the chatbot and subsequently engaged in semi-structured interviews to discuss their perceptions and ideate future chatbot designs. Using inductive reflexive thematic analysis, we identified distinct user preferences for communication style and conceptualized how these preferences shaped users' subjective processing fluency (i.e., the felt ease of processing health information) in healthcare chatbot interactions. Our findings reveal a key tension in AI design, where the preference for easy, concise interactions may risk prioritizing simplicity over deeper cognitive engagement, underscoring the need to balance ease of use with critical thinking for informed decision-making. The analysis contributes a theoretical framework centered on subjective processing fluency as a fundamental factor shaping user interactions with AI, and we argue that designing chatbots to maximize this fluency may give rise to a processing fluency bias that reduces users' scrutiny of AI-generated information. These findings suggest that healthcare chatbots and other conversational agents should calibrate, rather than maximize, processing fluency, balancing ease of use with the preservation of users' critical engagement with health information. This research emphasizes the importance for healthcare chatbots to align information presentation with users’ individual cognitive abilities. Future research should further explore how information presentation styles can optimize chatbot design to enhance user engagement while fostering accurate information consumption.</div></div>","PeriodicalId":100324,"journal":{"name":"Computers in Human Behavior: Artificial Humans","volume":"9 ","pages":"Article 100370"},"PeriodicalIF":0.0,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148738664","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Simulated self-assessment in large language models: A psychometric approach to AI self-efficacy","authors":"Daniel I. Jackson,&nbsp;Emma L. Jensen,&nbsp;Syed-Amad Hussain,&nbsp;Emre Sezgin","doi":"10.1016/j.chbah.2026.100389","DOIUrl":"10.1016/j.chbah.2026.100389","url":null,"abstract":"<div><div>Large language models (LLMs) are frequently perceived to have human-like comprehension capabilities. However, their proficiency in evaluating and quantifying these capacities remains uncertain. We conducted a controlled psychometric measurement study adapting the 10-item General Self-Efficacy Scale (GSES) to evaluate simulated self-assessment across 10 contemporary LLMs. Models completed the GSES under a no-task control condition and after three task-priming conditions: computational reasoning, social reasoning, and summarization. We examined between-model variation, task versus no-task differences, within-model stability, item-order robustness, internal consistency, and qualitative reasoning patterns. LLM-generated GSES responses were highly stable across repeated administrations, with nearly all model-task-item scores remaining identical across three runs. Internal consistency was high across task conditions, with Cronbach's alpha, and item-order effects were minimal, with intraclass correlation coefficients. Despite this stability, simulated self-efficacy differed significantly between models across all task and no-task conditions. Composite LLM GSES scores were lower and more variable than human norms (compared to the literature), and the latent response structure did not replicate human self-efficacy patterns. Qualitative analysis suggested that models differed in how they interpreted GSES items involving effort, coping, agency, and persistence. Some models rejected these constructs as inapplicable to artificial systems, while others translated them into task-oriented capability language. These findings indicate that LLMs can generate internally consistent psychometric self-assessments, but these outputs appear to reflect model-specific and context-sensitive communication behavior rather than human-like self-efficacy or metacognition. Psychometric prompting may help characterize how LLMs express capability, limitation, and agency under different prompting contexts, but generative responses should be interpreted cautiously to avoid anthropomorphic conclusions.</div></div>","PeriodicalId":100324,"journal":{"name":"Computers in Human Behavior: Artificial Humans","volume":"9 ","pages":"Article 100389"},"PeriodicalIF":0.0,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148935063","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"LLM-enabled humanoid interaction in care settings: A two-phase study with Ameca","authors":"Charlotte Larmuseau,&nbsp;Jens Krijgsman","doi":"10.1016/j.chbah.2026.100360","DOIUrl":"10.1016/j.chbah.2026.100360","url":null,"abstract":"<div><div>Loneliness among older adults is a growing public health concern, particularly in residential care settings where opportunities for sustained one-on-one interaction are limited. Recent advances in large language models (LLMs) enable socially assistive humanoid robots to engage in more natural dialogue, potentially supporting interaction-level experiences related to social presence and engagement, which are conceptually relevant to loneliness. However, evidence from real-world care deployments remains limited. This study evaluated the feasibility and user experience (UX) of interactions with <em>Ameca</em> (Engineered Arts, UK) in a Flemish residential-care setting and examined how baseline psychosocial factors relate to perceived interaction quality.</div><div>We conducted a two-session, mixed-methods feasibility study. Thirty-one older adults (63–93 years) completed a 15–20 min semi-structured Dutch conversation with <em>Ameca</em> (Phase 1); twenty participants returned approximately one week later for Phase 2 after a prompt-level re-finement of the conversational configuration. Pre-interaction measures assessed loneliness and well-being. Post-interaction measures captured user experience (UX): perceived enjoyment, empathy, social comfort, trust, and comprehensibility complemented by open-ended feedback. Conversation transcripts were available for Phase 2 and were analyzed for turn-taking balance, response patterns, and thematic content.</div><div>UX ratings were generally positive and stable across sessions. Higher baseline loneliness was associated with less favorable UX evaluations, most notably reduced comprehensibility. Phase 2 transcript analyses indicated balanced turn-taking but predominantly question–answer interaction patterns, with conversational depth varying across participants and themes centering on everyday life. Overall, findings support the feasibility and acceptability of LLM-enabled humanoid inter-action in residential care and highlight design priorities for future systems, including improved pacing, personalization, and context handling to better support social presence and engagement.</div></div>","PeriodicalId":100324,"journal":{"name":"Computers in Human Behavior: Artificial Humans","volume":"9 ","pages":"Article 100360"},"PeriodicalIF":0.0,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148770998","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Cybersophy: A framework for human cognition, identity, and values in the age of artificial agents","authors":"Huansheng Ning,&nbsp;Jianguo Ding","doi":"10.1016/j.chbah.2026.100385","DOIUrl":"10.1016/j.chbah.2026.100385","url":null,"abstract":"<div><div>Artificial agents—algorithms, large language models, and AI companions—now mediate human cognition, social identity, and moral development at scale. Yet the conceptual frameworks guiding behavioural research into human–AI interaction remain fragmented across disciplines, and no unified architecture spans the cognitive, ethical, existential, and axiological consequences of living alongside artificial agents. We propose Cybersophy—a fourth guiding principle—defined as the systematic wisdom through which humans understand and orient their relationship with digitally mediated environments and artificial agents. Building on prior foundational work that established Cybersophy's philosophical architecture ([Author], 2025b), the present paper makes a distinct empirical contribution: we operationalise each dimension through established psychological theory—epistemic (distributed cognition, metacognitive adaptation), ethical (algorithmic justice, procedural fairness), existential (virtual–physical identity fusion), and axiological (digital well-being)—and derive falsifiable hypotheses with specified mediators, moderators, and recommended designs. Crucially, integrating these dimensions generates four emergent cross-dimensional predictions unavailable from any single anchor theory: habitual AI reliance predicts reduced sensitivity to algorithmic procedural injustice; identity fragmentation attenuates autonomy-supportive well-being interventions; repeated algorithmic injustice erodes identity self-efficacy and triggers resistance-identity formation; and AI-reliance-induced cognitive delegation undermines the autonomy need satisfaction that mediates digital well-being. Cybersophy is advanced as a conceptual framework that generates an empirical research agenda rather than as a completed empirical theory: the hypotheses and cross-dimensional predictions are testable implications rather than validations of it, and we specify a staged strategy for evaluating them, including the conditions under which the integrative claims would require revision.</div></div>","PeriodicalId":100324,"journal":{"name":"Computers in Human Behavior: Artificial Humans","volume":"9 ","pages":"Article 100385"},"PeriodicalIF":0.0,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148843745","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Erratum regarding corrections related to Ethical approval and informed consent in articles published in [Computers in Human Behavior: Artificial Humans 2/2 (2024)]","authors":"","doi":"10.1016/j.chbah.2026.100352","DOIUrl":"10.1016/j.chbah.2026.100352","url":null,"abstract":"","PeriodicalId":100324,"journal":{"name":"Computers in Human Behavior: Artificial Humans","volume":"9 ","pages":"Article 100352"},"PeriodicalIF":0.0,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148934819","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Erratum regarding corrections related to ethical approval and informed consent in articles published in [Computers in Human Behavior: Artificial humans 2/1 (2024)]","authors":"","doi":"10.1016/j.chbah.2026.100341","DOIUrl":"10.1016/j.chbah.2026.100341","url":null,"abstract":"","PeriodicalId":100324,"journal":{"name":"Computers in Human Behavior: Artificial Humans","volume":"9 ","pages":"Article 100341"},"PeriodicalIF":0.0,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148934821","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A recognition-based framework for human–robot relationships","authors":"Junichi Hoshino","doi":"10.1016/j.chbah.2026.100382","DOIUrl":"10.1016/j.chbah.2026.100382","url":null,"abstract":"<div><div>As physically embodied AI becomes increasingly integrated into everyday life, understanding the diversity of human–robot relationships has become an important challenge. Previous studies have examined anthropomorphism, mind perception, moral attribution, and long-term relationships, providing important but complementary perspectives on how humans understand and relate to robots. However, a unified perspective for understanding the diversity of human–robot relationships remains limited. This study proposes a recognition-based framework consisting of four alternative modes: Tool, Decision Maker, Partner, and Social Personhood. The framework was developed by synthesizing complementary perspectives from previous research with recurring recognition patterns identified through comparative analysis of fictional human–robot interaction episodes. Rather than representing developmental stages or intrinsic properties of robots, the four modes describe alternative patterns of human recognition that emerge under different interaction contexts. The proposed framework provides a unified perspective for understanding how different forms of human recognition are reflected in characteristic patterns of human response and ultimately give rise to diverse forms of human–robot relationships, thereby providing a conceptual foundation for future research on human–robot relationships and the design of embodied AI.</div></div>","PeriodicalId":100324,"journal":{"name":"Computers in Human Behavior: Artificial Humans","volume":"9 ","pages":"Article 100382"},"PeriodicalIF":0.0,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148935064","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Can AI deliver appropriate support for diverse student profiles?: A large-scale evaluation","authors":"Sonsoles López-Pernas,&nbsp;Eduardo Oliveira,&nbsp;Kamila Misiejuk,&nbsp;Fitsum Gizachew Deriba,&nbsp;Rogers Kaliisa,&nbsp;Mohammed Saqr","doi":"10.1016/j.chbah.2026.100357","DOIUrl":"10.1016/j.chbah.2026.100357","url":null,"abstract":"<div><div>Large language models (LLMs) have the potential to offer a novel approach to student support through learning analytics (LA)–informed advising. This study examines such potential by exploring whether current LLMs can adaptively recommend appropriate support across student needs, characteristics, and contexts. To do that, we generated 4500 synthetic student vignettes using three LLMs (GPT-5-mini, Mistral-Medium-2508, and Qwen-Plus). Each vignette included a single behavioral student trait, classified as either high- or low-performing, and one positive or negative LA indicator describing student learning behavior. These were randomly selected from pre-defined literature-informed lists to provide the LLM with contextual information about the student. The LLMs then generated recommendations for support, including the level of support needed and timing. We examined whether the three LLMs behaved consistently with each other and whether they responded differentially to LA indicators and other learner and context characteristics. For this purpose, we analysed the generated data using Pearson correlations with False Discovery Rate correction and analyses of variance conducted separately for each LLM. Findings revealed limited sensitivity to student characteristics as indicated by LA indicators and the recommended support. In addition, we detected considerable inconsistencies in LLM-generated support recommendations across tested LLMs. These results suggest that current LLMs are not yet reliable as prescriptive models for providing student support at scale in an ethical, consistent, and reliable manner.</div></div>","PeriodicalId":100324,"journal":{"name":"Computers in Human Behavior: Artificial Humans","volume":"9 ","pages":"Article 100357"},"PeriodicalIF":0.0,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148636651","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Trust in autonomous human–robot collaboration: Effects of responsive interaction policies","authors":"Shauna Heron,&nbsp;Meng Cheng Lau","doi":"10.1016/j.chbah.2026.100358","DOIUrl":"10.1016/j.chbah.2026.100358","url":null,"abstract":"<div><div>Trust plays a central role in human–robot collaboration, yet its formation is rarely examined under the constraints of fully autonomous interaction. This pilot study investigated how interaction policy influences trust during in-person collaboration with a social robot operating without Wizard-of-Oz control or scripted repair. Participants completed a multi-stage collaborative task with a robot that autonomously managed spoken-language dialogue, affect inference, and task progression. Two interaction policies were compared: a responsive policy, in which the robot proactively adapted its dialogue and assistance based on inferred interaction state, and a neutral, reactive policy, in which the robot provided only direct, task-relevant responses when prompted. Responsive interaction was associated with significantly higher post-interaction trust on both trust instruments under viable communication conditions, despite no reliable differences in overall task accuracy. As language-mediated interaction degraded, this advantage attenuated, and ratings became less clearly interpretable as calibrated evaluations of collaborative competence. A tentative pattern in which experienced trust (TI-HRC) appeared more sensitive to communication breakdown than perceived trust (TPS-HRI) is reported descriptively as the small sample size and exploratory design preclude formal tests of differential sensitivity across trust dimensions. These preliminary findings are consistent with the interpretation that trust in autonomous human–robot interaction is shaped by process-level interaction dynamics and operates within constraints imposed by communication viability, highlighting the importance of evaluating trust under real autonomy conditions when designing interactive robotic systems.</div></div>","PeriodicalId":100324,"journal":{"name":"Computers in Human Behavior: Artificial Humans","volume":"9 ","pages":"Article 100358"},"PeriodicalIF":0.0,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148636654","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Chatbot appearance and loneliness trajectories in women with high autistic traits","authors":"Xiaohan Feng,&nbsp;Jie Fu","doi":"10.1016/j.chbah.2026.100369","DOIUrl":"10.1016/j.chbah.2026.100369","url":null,"abstract":"<div><div>Chatbots are increasingly explored as tools to support social and emotional well-being. However, little is known about how chatbot visual appearance influences user outcomes over time, particularly among individuals with elevated autistic traits. This study examined whether chatbot appearance shapes loneliness trajectories among women with high autistic traits who used a chatbot at least three times per week. A randomized longitudinal experiment compared two conditions that were identical in function, dialogue strategy, and usage structure. Seventy-two participants interacted for eight weeks with either a chatbot with a non-human visual appearance or a chatbot with a human-like visual appearance. Loneliness was measured at four time points and analyzed using linear mixed-effects models. We predicted that loneliness would decline in the non-human condition and that the decline would be steeper than in the human-like condition. Both predictions were supported. The Group × Time interaction was significant, and loneliness decreased in the non-human appearance condition, whereas the slope was bounded near zero in the human-like condition, 95% CI [−0.92, 0.99]. Baseline loneliness did not differ between groups. These findings suggest that chatbot appearance can influence longer-term interaction outcomes rather than only immediate impressions. For neurodiverse users, appearance is a core design dimension and warrants longitudinal evaluation.</div></div>","PeriodicalId":100324,"journal":{"name":"Computers in Human Behavior: Artificial Humans","volume":"9 ","pages":"Article 100369"},"PeriodicalIF":0.0,"publicationDate":"2026-08-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148738633","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}