{"paper_id":"3fa6acfb-d42c-4dfd-aa7b-cf3ecc34885b","body_text":"Metacognitive Reflection in the Era of Generative AI | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Metacognitive Reflection in the Era of Generative AI Kim Uittenhove, Andrew Ellis, Fabian Mumenthaler, Ioana Gatzka, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6973046/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Metacognitive reflection is a crucial transversal skill, especially in an era where generative AI transforms how we teach and learn. As well as being a driver of the need to develop metacognitive reflection, generative AI is also a tool that can be used to enhance metacognitive reflection, such as chatbots that act as coaches to guide students in metacognitive reflective practice. In this study, we examined the potential of LLM-powered chatbots to promote metacognitive reflection across three distinct educational contexts. Our results show that the chatbot successfully constructed a metacognitive dialogue and delivered relevant, evidence-based recommendations. However, student engagement levels were generally low, with limited active participation observed across all studies. Notably, metacognitive self-regulation, and other individual differences, did not consistently predict engagement levels, suggesting that learners with higher reported self-regulation were not inherently more likely to use the tool. We also found no evidence that metacognitive engagement levels led to improved learning outcomes. However, these findings must be interpreted with caution, as engagement levels may be a limited metric for capturing how students benefit from chatbot-assisted reflection. We conclude by raising key design questions around how to develop chatbot systems that not only deliver metacognitive content and feedback but also encourage active student participation. While system prompts can help LLMs maintain focus on metacognitive reflection, hybrid designs that add an additional layer of scripting or multi-agent systems may be necessary to support an active learner role and ensure that important metacognitive checkpoints are met by the learner. chatbot generative AI metacognitive reflection education Figures Figure 1 Figure 2 Introduction Self-regulated learning, comprising metacognitive self-regulation (Credé & Phillips, 2011), the awareness and regulation of one’s own thinking processes (Flavell, 1979), is widely recognized as crucial for effective learning and academic success (Broadbent & Poon, 2015; Jansen et al., 2019; Richardson et al., 2012). In practice, this entails many aspects, including motivational aspects, beliefs such as academic self-efficacy, goal orientation, cognitive aspects such as elaboration and organization, and metacognitive aspects such as planning, monitoring, self-regulation, and evaluation (Pintrich et al., 1991a; Schraw & Dennison, 1994), all contributing to the effective use of high-utility learning strategies (Bjork et al., 2013; Donoghue & Hattie, 2021; Dunlosky et al., 2013; Tormey & Hardebolle, 2017). For details on effective learning techniques, see appendix Section 6. This set of skills is transversal and supports lifelong learning and adaptability across disciplines and changing circumstances. The necessity of cultivating such adaptability has become increasingly important in the 21st century, where knowledge and technology evolve constantly and rapidly, changing how we teach and learn. The rise of generative AI in education in particular is challenging us to examine and potentially reimagine our methods of teaching and learning. Tools like ChatGPT or Gemini can effortlessly and instantly generate solutions and content across academic disciplines, and without guidance, students might over-rely on AI assistance and forgo the mental effort of reflection. Recent studies warn that such over-reliance can worsen learning outcomes (Bastani et al., 2024, Kosmyna et al., 2025) and lead to “metacognitive laziness”, where learners become dependent on AI and neglect their own self-monitoring and critical thinking (Fan et al., 2024). Therefore, metacognitive reflection must be actively taught and stimulated, to navigate a learning environment saturated with AI assistance, and ensuring that learners engage in effective learning unharmed by the presence of AI tools often used to bypass the learning process. At the same time that generative AI creates challenges for student learning, it can also serve as a tool to address these same challenges. For instance, chatbots based on generative AI can identify gaps in self-regulated learning and engage in metacognitive reflective dialogue. Thus, instead of steering students away from using chatbots, we can steer them towards using these tools in a way that is helpful for their learning. This dual role of generative AI, as both a driver of the increased need for metacognitive reflection, and a tool for stimulating it, sets the stage for re-imagining how we coach students in reflective practices. Chatbots for Metacognitive Reflection Given the importance of metacognitive reflection and self-regulation, a crucial question is how we concretely help students develop these skills. Unfortunately, many students do not pick up these skills autonomously (Pintrich, 2002) and struggle with effective learning (Dunlosky et al., 2013; Morehead et al., 2016). Among minority and low socio-economic status groups, this issue is exacerbated as documented by a metacognitive skills gap (McGuire, 2021). Therefore, researchers have argued for the explicit teaching of metacognitive reflection and learning strategies. Students can be directly instructed on metacognitive strategies and the plan–monitor–evaluate cycle for learning. For example, Amzil (2014) provided college students with five sessions of explicit instruction that focused on metacognitive processes for reading, leading to improved reading comprehension and metacognitive awareness. Likewise, Maftoon and Alamdari (2020) tested a 10-week intervention with metacognitive strategy lessons on planning, monitoring, and self-evaluation, leading to improved listening comprehension. These interventions suggest that explicit instruction can change academic behaviors and improve learning outcomes (Cook et al., 2013). However, rather than teaching generic metacognitive skills in isolation, embedding reflective activities within regular coursework may improve transfer and application to authentic learning situations. For example, reflective assignments such as learning journals, diaries, or structured peer discussions allow students to engage with metacognitive reflection in context, and prompt students to articulate what they learned, the strategies they used, and the challenges they faced. These activities make the invisible process of learning more visible to the learner, reinforcing the habit of reflection. Digital tools have expanded the range of options to support these activities, for example through E-portfolios and features in Learning Management Systems, such as personalized learning dashboards. The Learning Companion at EPFL is one such initiative that promotes student self-monitoring and reflection (Hardebolle et al., 2019; https://companion.epfl.ch/]. However, a persistent challenge remains: motivating students to engage meaningfully and consistently with these tools. Among digital technologies, chatbots present a promising approach to scaffolding metacognitive reflection and engaging students through conversational interaction. Emulating the role of a teacher or mentor, chatbots can prompt students with reflective questions and provide adaptive feedback. Early implementations (Carbonell, 1970; Graesser et al., 1999) typically relied on rule-based or decision-tree architectures, guiding learners through predefined pathways. For instance, these chatbots have been used to guide students through a series of reflective questions to deepen understanding of a course topic (Britos Cavagnaro, 2022; Carbonell, 1970; Graesser et al., 1999), to support self-regulated learning and receive metacognitive feedback (Martins et al., 2024; Yin et al., 2024), and to offer coaching for challenges such as exam anxiety (Mai et al., 2022). Some of these systems integrated basic natural language processing (NLP) to interpret student input. A well-known example is AutoTutor (Graesser et al., 2004), which uses a hybrid architecture that integrates rule-based scripts with semantic analysis. Similarly, Sáiz-Manzanares et al. (2023) integrated a hybrid chatbot into the Moodle environment to support students’ metacognitive and self-regulation strategies. The advantage of such systems is their ability to scaffold reflection through a semi-structured conversation, prompting students to articulate their thinking, elaborate on shallow responses, and iteratively refine their understanding. Chatbots can be deployed at optimal moments, such as immediately after a lesson or upon receiving feedback, ensuring that reflection occurs when it is most timely and contextually relevant. Moreover, unlike human instructors, who may be limited in providing individualized support at scale, chatbots can engage in parallel, one-on-one reflective conversations with many students simultaneously, ensuring that every student receives individualized guidance. Several studies report positive outcomes from such chatbot-assisted interventions. These include gains in metacognitive awareness (Yin et al., 2024) and in some cases, improved learning outcomes (Graesser et al., 2004; Yin et al., 2024). The emergence of generative AI and large language models (LLMs) has greatly accelerated these developments. LLM-powered chatbots can interpret and respond to free-form student input with remarkable fluency, enabling natural, personalized interactions across a wide range of contexts. Unlike rule-based or hybrid systems, designing a system where users directly and continuously interact with a single LLM requires less authoring effort. Chatbot behavior can be steered through a well-designed system prompt rather than extensive hard-coded dialogue trees, making them highly scalable and generalizable across domains. However, while LLM-based chatbots excel at flexibly generating contextually appropriate responses, they lack inherent mechanisms for ensuring that pedagogical scaffolding is achieved. This raises a critical design question: to what extent can LLM-based chatbots achieve effective scaffolding solely through prompt engineering? Student Engagement A central challenge in the success of interventions aimed at promoting metacognitive reflection is sustaining student engagement. Chatbots based on rules or decision trees often struggle to stimulate student responsiveness. Martins et al. (2024) analyzed sessions with a self-regulation coaching chatbot and reported that most frequently, conversations were composed of only four messages. The authors noted that such a brief interaction is insufficient for developing complex abilities like those required for self-regulated learning. Moreover, whereas many students tried the chatbot on the first day it was introduced, usage quickly dropped off as the semester went on. Similarly, Graesser et al. (2004), with their course concept tutoring chatbot, found that students initially provide only minimal responses to the chatbot questions, and that many dialogue turns are needed to elicit deeper explanations. In contrast, chatbots powered by LLMs may offer a more compelling and responsive conversational experience, potentially improving engagement. However, simply deploying a capable system does not guarantee active participation in chatbot-assisted metacognitive reflection. To fully develop the potential of LLM-based chatbots for metacognitive reflection, it is essential to understand students’ engagement patterns and the factors that influence them. Student engagement in this context is likely shaped and influenced by multiple interrelated factors. The Technology Acceptance Model (TAM), highlights perceived usefulness and ease of use as key determinants of digital engagement (Davis, 1989). Supporting this, a scoping review conducted by Schei et al. (2024) on how students in higher education perceive AI chatbots found that students generally find AI chatbots to be highly useful and motivating, particularly appreciating the immediacy of feedback provided. In addition to usability, digital competence could influence students’ engagement with chatbots. Students with higher digital skills might find it easier to integrate chatbot interactions into their learning, while those with lower competence could be more hesitant. For instance, AI-related competence has been positively linked to student engagement in digital learning environments (Hidayat-ur-Rehman, 2024). However, given the intuitive design of most LLM-based chatbots (i.e., low AI-related competence is needed), attitudes toward AI may be an even stronger predictor. Students with favorable attitudes may demonstrate deeper and more frequent interactions with chatbot-based tools, whereas skepticism toward AI may result in superficial engagement or resistance. Finally, a student’s existing level of metacognitive self-regulation may influence the effectiveness of reflection activities, as suggested by a meta-analysis by Jansen et al. (2019), which found that preexisting self-regulated learning skills partially mediate the impact of self-regulated learning interventions aimed at improving academic outcomes. This underscores the importance of considering preexisting metacognitive competencies. Those who already plan, monitor, and evaluate their learning may be more receptive to chatbot prompts that encourage reflection, while those with less-developed self-regulation habits may be less responsive. In sum, engagement in chatbot-assisted metacognitive reflection depends on a constellation of learner-specific factors, ranging from perceived usefulness and digital skills to attitudes toward AI and prior self-regulatory skills. Identifying and addressing these factors is critical for understanding how to design interventions that maximize educational benefits by reaching all students and actively engaging them. The Present Study Given the increased need for metacognitive reflection engendered by the emergence of generative AI, and the potential of this technology to support such reflection, our study employed a state-of-the-art large language model (LLM) to power a chatbot with the goal of supporting metacognitive reflection among students. The chatbot was instructed through a system prompt to co-construct a dialogue aimed at stimulating metacognitive reflection in students, deepening their understanding of learning strategies, and provide evidence-based recommendations to improve study preparation practices. We conducted three interconnected studies designed to explore the chatbot’s application across different educational contexts: Study 1 examined chatbot-supported metacognitive reflection for general course preparation throughout a full semester at the University for Teacher Education in Bern. Study 2 focused on the chatbot’s application specifically for exam preparation within a single session at Bern University of Applied Sciences during the semester’s conclusion. Study 3 investigated chatbot interactions in a diverse sample recruited through an online platform, varying the order of chatbot engagement and self-report collection. The sample was balanced between social sciences and STEM fields. We analyzed the outcomes of the three studies centered around three primary questions: Can an LLM-powered chatbot, guided solely by system-level instructions that define the metacognitive goals, effectively co-construct a dialogue that support students in metacognitive reflection? This includes helping students articulate their study strategies, reflect on their learning processes, and receive evidence-based recommendations for improvement. What are the engagement patterns of students with chatbot-assisted metacognitive reflection? How do individual differences—specifically self-regulated learning capabilities, digital competencies, and attitudes toward AI—predict or influence these engagement patterns? Understanding these relationships helps clarify for whom and under which conditions chatbot-based interventions are most effective. Does more engagement with the chatbot increase subsequent academic performance as measured by exam outcomes, potentially demonstrating tangible benefits of improved metacognitive reflection on study behaviors and academic achievement? Methods Participants across the three studies are described in detail in Table 1. All participants provided electronic consent prior to their participation in the study, and the study was approved by the EPFL Human Research Ethics Committee (approval number HREC000466). In Study 1, conducted at PHBern, 159 students were given the opportunity to participate and engage with the chatbot. Of these students, 34 (21.4%) chose to engage with the chatbot, and 48 provided self-report data. The final analyzed sample comprised the 48 students who provided self-report data, which included 20 out of the 34 individuals who actively engaged with the chatbot. Fourteen students used the chatbot but did not complete the self-reports. Study 2 included 49 students from BFH who had the opportunity to interact with the chatbot. Out of these, 30 students (61.2%) engaged with the chatbot. However, the analyzed sample included only 19 participants who provided both self-report data (including SRL measures) and chatbot engagement data. Study 3 involved 131 student participants recruited through the Prolific platform. These participants held active student status at undergraduate or graduate levels (see Table 1 for their countries of residence). Participants were equally distributed between social sciences and STEM disciplines. Nine participants were excluded from the analysis due to reported technical issues. Table 1 Participant Characteristics Study, N (sample size), Sex (M/F), Age (mean ± standard deviation), Residence (country), Education Level, Study Field Apparatus The study comprised two complementary parts: an interactive chatbot system designed to support students during their reflective and learning processes, and self-report data collected through secure online platforms, specifically RedCap and Moodle. Additionally, in Study 1 and 2, following the exam session, we collected the grades of the participating students. Chatbot System Chatbot Design. We designed an LLM-based chatbot capable of conducting a coaching conversation, guided by a semi-structured interview plan that includes the collection of relevant information from the student and accompanying them to reflect on their learning processes and/or apply effective study strategies. Within its context window, the LLM was provided with information from the literature on metacognitive practices and effective study techniques. Reinforced with this evidence-based approach, the chatbot encouraged students’ self-reflection and encouraged the adoption of proven approaches to learning. In Study 1, the chatbot script (see Section 7) guided the students through reflection about course material, understanding, the learning process, and preparation of the next lecture. In studies 2 (see Section 8) and 3 (see Section 9), the chatbot helped students to prepare for an exam by guiding them through reflection on study strategies, exam or project preparation, or self-regulated learning, without explaining or asking to explain the course content itself. The chatbot suggested improved learning strategies based on learning science (see Section 6). The system prompt for Study 3 was slightly revised, to have the chatbot ask about challenges experienced by the students, and to help students make concrete study schedules. Authentication and Access. In Studies 1 and 2, students logged in and were authenticated through their Learning Management System (LMS) and subsequently directed to a Moodle course hosted by the Bern University of Applied Sciences (BFH). This course employed Learning Tools Interoperability (LTI) to ensure secure, SSL-encrypted access to the chatbot’s frontend. The student’s unique identifier and the course ID were transmitted to the chatbot. In Study 3, authentication occurred via a valid Study ID provided by Prolific. In addition, a unique Prolific Personal ID was passed along to the chatbot. These unique identifiers were saved with the interaction logs and permitted us to link interaction data with participant self-reports. Once authenticated, students were able to view the chatbot’s messages and send their responses, with the ongoing dialogue displayed in a conversation history. Technical Infrastructure. The chatbot interface (frontend) was a web application hosted on a server provided by Exoscale in Switzerland. Access to this interface was restricted to authenticated users, either via LMS logins (Study 1 and 2) or Prolific (Study 3). A Python-based LLM orchestration framework coordinated communication with an Azure OpenAI GPT-4 model hosted by Microsoft in Switzerland by sending secure API calls. The Azure OpenAI GPT-4 model processed user inputs and generated responses in accordance with the system instructions that we provided to customize behavior. The data processing agreement between BFH and Microsoft ensured Swiss data storage and that user data were neither used to further train the model nor employed to improve other Microsoft products. The Azure OpenAI GPT-4 Model was set up in a resource group of a BFH subscription with the geolocation of Switzerland North. Self‐Report Measures Self‐Regulated Learning (SRL) and Elaboration (ELA). Students’ self-regulatory learning processes were assessed using an instrument that mainly focuses on metacognitive self-regulation, as a key aspect of self-regulated learning. The instrument mainly draws items from the Motivated Strategies for Learning Questionnaire (MSLQ; Pintrich et al. (1991b)), the Metacognitive Awareness Inventory (MAI; Schraw and Dennison (1994)), and the Learning and Study Strategies Inventory (LASSI; Weinstein et al. (2002)). It also includes additional items designed to expand scales related to planning, monitoring, regulation, and evaluation, which are the main processes involved in metacognitive self-reflection. The use of elaboration strategies during learning was treated as a separate scale. The instrument has been replicated and has shown measurement invariance in English, French, and German using Prolific samples (Uittenhove et al., in preparation), making it suitable for use in German (Study 1) and English (Studies 2 and 3). Items were rated on a 7-point Likert scale from 1 (“Not at all me”) to 7 (“Totally me”). See Section 10 to view the items of the instrument. General Academic Self‐Efficacy (GASE). Students’ beliefs in their academic capabilities were measured using the General Academic Self-Efficacy Scale (Zyl et al., 2022), a five-item instrument (see Section 11) with demonstrated longitudinal invariance and criterion validity. Each item is rated on a 5-point Likert scale ranging from 1 (“Strongly disagree”) to 5 (“Strongly agree”), and the final score is the average across items, resulting in a range from 1 to 5. Mastery (MAST) and Performance (PERF) Orientation. Achievement goal orientations were assessed using the Achievement Goal Questionnaire–Revised (AGQ-R; Elliot and Murayama (2008)), which distinguishes between mastery-approach, mastery-avoidance, performance-approach, and performance-avoidance goals. This instrument is widely used and validated for capturing both the content (i.e., focus on learning versus outperforming others) and the valence (i.e., approach versus avoidance) of students’ achievement motivations. In the present study, only the approach-oriented subscales for mastery and performance were used (see Section 12). Items were rated on a 5-point Likert scale from 1 (“Strongly disagree”) to 5 (“Strongly agree”), and scores for mastery and performance were computed separately as the average of their respective items, yielding a range from 1 to 5. Digital Competency (DIGI). Participants in Studies 1 and 2 completed the DigiCompEdu questionnaire (Ghomi & Redecker, 2019), a framework developed to assess the digital competence of educators. We focused on the problem-solving domain, which evaluates how educators identify needs, creatively use technology to address them, and recognize gaps in digital competence (see Section 13). Participants rated items on a 4-point scale ranging from 0 to 3, with higher values indicating greater perceived competence. The final score represents the average of the item responses and ranges from 0 to 3. AI Attitude (AI_ATT). Attitudes toward artificial intelligence were measured using the AI Attitude Scale developed by Stein and colleagues (Stein et al., 2024). This instrument captures cognitive, affective, and behavioral responses to AI as a general-purpose technology and has shown good reliability and construct validity in capturing both positive and negative sentiments toward AI. It consists of 12 items (see Section 14) rated on a 5-point Likert scale from 1 (“Strongly disagree”) to 5 (“Strongly agree”). The total score is calculated as the sum of all items, resulting in a score range of 12 to 60. Procedure See Table 2 for an overview of the self-report measures and chatbot interactions collected across the three studies. Study 1 was conducted during the spring semester of 2024 at PHBern, where students were informed about the chatbot’s availability and using it was optional but encouraged as a beneficial learning aid. Participants provided self-report measures at the beginning of the semester using REDCap. Throughout the semester, students had unrestricted access to the reflection chatbot. All students in the participating courses were permitted to interact with the chatbot, regardless of formal participation in the study. Study 2 took place in a single session at the end of the semester at BFH. Students completed self-report measures at both the beginning and the end of the session, which lasted approximately one hour. Within this timeframe, students engaged with the chatbot and submitted their self-reports via the Moodle platform. Study 3 was conducted at the conclusion of the autumn semester 2024 through the Prolific platform. Participants were compensated for a one-hour session, during which they provided self-report data and interacted with the chatbot. This study employed a counterbalanced design: one group first interacted with the chatbot before completing self-report measures, while the other group completed self-reports first, followed by chatbot interaction. Self-report data in Study 3 were collected using REDCap. Table 2 Organization of Chatbot Interaction and Self-Reports Chatbot interaction and collection of self-reports (SRL, ELA, DIGI, AI_ATT, GASE, MAST, PERF) in study 1 (S1), study 2 (S2), the study 3 group who interacted with the chatbot first (S3C), and the study 3 group who provided self-reports first (S3Q) Note . In Study 1, we initially planned to assess SRL, ELA, and DIGI a second time following the chatbot interaction. However, only a few participants (n = 5) completed the follow-up questionnaire, so these data were not included in the analysis. Data Analysis Identifying Metacognitive Interactions We analyzed chatbot interaction logs and identified student messages that aligned with the chatbot’s intended purpose, specifically metacognitive interactions, and to identify non-metacognitive interactions, such as requests for explanations of course content or to generate text for assignments. Each message was analyzed together with the context of the preceding conversation, by sending it via Microsoft Azure OpenAI API to a GPT-4 model hosted by Microsoft, and asking the model to classify the message and provide both a response code and a justification (see Section 15). In Study 1, the chatbot was instructed (see Section 7) to avoid providing theoretical explanations of the course content. Instead, its mission was to facilitate dialogue with students about the course content and encourage them to reflect on their understanding and learning process. Therefore, metacognitive interactions included discussions about course material, understanding of the course material, the learning process, and preparation for the next lecture (see Section 15). In studies 2 and 3, the chatbot was instructed to focus on exam preparation and learning strategies specifically (see Section 8; Section 9). Therefore, metacognitive messages focused on study strategies, exam or project preparation, and self-regulated learning, but did not involve explaining or requesting explanations of course content. To assess the reliability of coding these interactions as metacognitive or non-metacognitive, we compared the classifications of GPT-4 to human classifications across 180 randomly selected messages of which half were classified as metacognitive by GPT-4. In 80 cases, both the human and GPT-4 assigned the non-metacognitive label (e.g., “ Thanks. What would you suggest is a good dissertation for a criminology dissertation on domestic Abuse) , and in 69 cases, both assigned the metacognitive label (e.g.,” Sure, I get your point but since a specific seminar session is always going to focused [sic ]on a specific set of texts mixing them up may have a beneficial learning effect but it would reduce ‘synergy’ with the seminar itself, if you know what I mean.”) , resulting in 149 agreements. This corresponds to an observed agreement rate of 82.8%. To adjust for chance agreement, Cohen’s Kappa was computed, yielding a value of approximately 0.66, which indicates substantial agreement between the label attributed by human and GPT-4, according to Landis and Koch (1977) interpretation scale. Disagreements occurred in 31 cases: in 21 instances, the human assigned the non-metacognitive label while GPT-4 assigned the metacognitive label; and vice versa in 10 other instances. Given that the classification agreement was substantial, we subsequently used the classification decisions from the GPT-4 model in our analysis. In addition to identifying metacognitive and non-metacognitive interactions in Study 2 and Study 3, we also analyzed the entire conversation to capture which moderate- and high-utility learning strategies students reported using initially, and which were recommended by the chatbot, whether students initially had a study schedule and whether the chatbot contributed to creating one. We also extracted exam readiness (how ready are you for the exam on a scale from 1 to 10) (see Section 15). Descriptive Analyses We conducted descriptive analyses to examine the distributions of self-report variables (SRL, ELA, GASE, MAST, PERF, AI_ATT) in each sample and to assess student engagement with the chatbot across all three studies. Engagement was measured using several metrics: the total number of messages exchanged, the total character count, and the character count of metacognitive and non-metacognitive student messages. In Studies 2 and 3, we also analyzed students’ initial reports of using moderate- and high-utility learning strategies, as well as the strategies recommended by the chatbot. Study 3 further included students’ usefulness ratings of the chatbot conversation, offering insight into their perceptions of its helpfulness for various aspects of studying. Bayesian Structural Equation Modeling To examine the complex relationships among demographic characteristics, self-reported measures, chatbot engagement, and academic performance, we applied Bayesian Structural Equation Modeling (BSEM). This methodological approach was selected for its capacity to simultaneously model multiple interrelated variables and facilitate multi-group analyses across our studies. The Bayesian framework was particularly appropriate for our exploratory analyses due to modest sample sizes. Because chatbot engagement metrics (i.e., metacognitive and non-metacognitive student message character count) were inherently positive and right‑skewed, we applied a log transformation, then standardized all quantitative variables to improve model fitting and interpretation. Exam scores were standardized separately by exam, to account for differences in exam difficulty or teacher rating biases. We estimated models using the bsem() function from the blavaan package (Version 0.5.8; Merkle & Rosseel, 2018) in R. All variables were standardized prior to estimation. Standard normal priors (μ = 0, σ = 1) were applied to all regression coefficients, including intercepts. All BSEM models were sampled using the No-U-Turn Sampler (NUTS) algorithm with four chains, each consisting of a 1,000-iteration burn-in phase followed by 2,000 post-burn-in iterations. Structural Model for Studies 1 and 2. For Studies 1 and 2, our structural model (see Figure 1) included the effects of SRL, ELA, and DIGI on students’ chatbot engagement (metacognitive and non-metacognitive). We further modeled how SRL, ELA, and metacognitive chatbot engagement influenced academic performance as measured by exam scores. Given the procedural differences between Studies 1 and 2, specifically the timing of chatbot engagement relative to self-report completion and contextual variations (semester-long interaction versus single-session engagement), we fitted a multi‑group BSEM that maximizes a single joint likelihood while estimating all parameters independently for each study. Structural Model for Study 3. The structural model for Study 3 (see Figure 2) incorporated demographic predictors (sex, education level, and field of study) as influences on self-reported measures (SRL, ELA, GASE, MAST, PERF, AI_ATT). We then examined how these self-reported measures, together with demographic variables, predicted chatbot engagement. We fitted a multi‑group BSEM that maximized a single joint likelihood and represented our counterbalanced design, where one group used the chatbot before completing the self‑reports and the other completed the self‑reports first. Paths from self‑reports to chatbot engagement, where order effects were expected, were estimated independently in each group. Paths from demographic variables to self‑reports and from demographics to engagement were not allowed to vary between groups, so a single coefficient was estimated from the combined data of both groups. Each demographic variable was entered as a binary indicator and the estimated coefficients indicated the difference compared to the reference category: Education Level (reference = undergraduate vs. graduate), Study Field (reference = STEM vs. social sciences), and Sex (reference = male vs. female). Although initially included, we omitted age as a predictor due to consistently poor model fit and high residuals that persisted despite various adjustments. Results Descriptive Statistics Table 3 Distribution of Self-Report Variables Across the Three Studies Median, first quartile (Q1), and third quartile (Q3) are reported for each variable Variable Group Median Q1 Q3 SRL S1 4.20 3.74 4.66 S2 4.71 4.12 5.14 S3C 4.87 4.13 5.53 S3Q 4.88 3.98 5.98 ELA S1 5.20 4.64 5.86 S2 5.11 4.61 5.94 S3C 5.50 4.80 6.31 S3Q 5.60 4.53 6.39 GASE S1 NA NA NA S2 4.00 3.90 4.50 S3C 4.00 3.90 4.60 S3Q 4.00 3.60 4.60 MAST S1 NA NA NA S2 4.00 3.67 4.00 S3C 4.00 3.67 4.67 S3Q 4.00 3.67 4.67 PERF S1 NA NA NA S2 4.00 3.33 4.00 S3C 3.67 3.17 4.17 S3Q 4.00 3.33 4.33 DIGI S1 1.82 1.55 2.18 S2 2.27 1.95 2.45 S3C NA NA NA S3Q NA NA NA AI_ATT S1 NA NA NA S2 42.00 39.00 47.50 S3C 48.00 41.00 54.00 S3Q 46.00 43.00 52.00 In Study 1, 48 participants provided self-regulated learning (SRL) data, and 20 engaged with the chatbot. Whether participants engaged with the chatbot did not differ between high-SRL participants (SRL range [4.21, 5.91]), of whom 41.7% engaged, and low-SRL participants (SRL range [2.49, 4.19]), who had an identical engagement rate of 41.7%. Participants’ chatbot engagement measures varied across the three studies (see Table 4 for detailed engagement metrics). In Study 1, students exchanged a median of 27 messages (5’001 characters) with the chatbot over a semester, whereas Study 2 featured a single-session exchange of 21 messages (4’764 characters). Study 3, involving the Prolific sample, revealed very similar engagement levels irrespective of whether students chatted first (3C) or completed self-reports first (3Q). Both groups exchanged around 23 messages (10,125 and 10,144 characters). In Studies 1, 2, and 3, the chatbot consistently focused on metacognitive reflection, dedicating a median of 95.3%, 90.0%, and 100% of its characters to this topic, respectively. In Study 1, students contributed 1’209 characters (~ 200 words), comprising 1’138 (94.0%) characters on metacognition. Student contributions in Study 2 were shorter at 536 characters (~ 90 words), comprising 406 (75.7%) characters on metacognition. Although Study 3 conversations were over twice as long as those in Study 2, student contributions were only slightly longer, contributing 788 and 713 characters (~ 120–130 words), comprising 598 (75.9%) and 565 (79.2%) characters on metacognition. Concerning the initial representation of student study methods, the chatbot effectively collected information in 100% of cases in Study 2, while in Study 3, this information was successfully collected in 98.5% of cases. Participants reported median exam readiness scores of 5 in both Study 2 (IQR [3, 7]) and Study 3 (IQR [3, 6]). Participants in Studies 2 and 3 initially revealed limited use of effective study strategies (see Table 5 ). Only 16–17% of students reported using a structured study schedule, while distributed practice was reported by 16–22%, interleaved practice by 5–6%, and elaborative strategies by 9–11%. Practice testing usage differed notably between studies, with only 21% reporting this strategy in Study 2 compared to 48% in Study 3. After discussing the student’s current study methods, the chatbot recommended high-utility learning strategies as instructed, prioritizing distributed practice (84–91%), followed closely by practice testing (79–87%), elaborative strategies (68–75%), and interleaved practice (68–70%). These recommendations occurred more frequently in Study 3. Additionally, Study 3 uniquely assisted students in creating a structured study schedule in 89% of cases, an intervention which was not specifically added to the chatbot instructions for Study 2. A large majority (96%) of participants in Study 3 found the chatbot useful because it recommended learning strategies, whereas 53% also found it useful because it helped them reflect, and 55% found it useful because it helped them feel prepared for the exam. Table 4 Chatbot Engagement Metrics Across the Three Studies This table presents the distribution of engagement indicators, including total messages exchanged, total characters exchanged, metacognitive characters contributed by both the mentor (chatbot) and the student, and non-metacognitive characters by each party. Values reported for each variable include the median, first quartile (Q1), and third quartile (Q3) Variable Group Median Q1 Q3 Total Messages S1 27 19 48 S2 21 18 26 S3C 23 19 31 S3Q 23 19 31 Total Characters S1 5001 3144 9946 S2 4764 3672 5885 S3C 10125 6730 15104 S3Q 10144 7168 14994 Metacognitive Mentor Characters S1 3802 2207 8438 S2 3496 2560 4144 S3C 7848 5847 11522 S3Q 8287 6252 11187 Non-metacognitive Mentor Characters S1 186 0 409 S2 390 228 628 S3C 0 0 304 S3Q 0 0 362 Metacognitive Student Characters S1 1138 584 2214 S2 406 218 504 S3C 598 347 992 S3Q 565 408 1223 Non-metacognitive Student Characters S1 72 22 162 S2 130 79 196 S3C 190 97 472 S3Q 148 80 370 Table 5 Initial and Recommended Learning Strategies in Studies 2 and 3 Percentage of participants reporting each learning strategy at the beginning (initial) and chatbot-endorsed learning strategies (recommended) in Studies 2 and 3. Strategies included creating a study schedule , distributed practice , interleaved practice , practice testing , and elaboration Bayesian Structural Equation Model Results Study 1 and 2 The model, comprising 42 parameters, was applied to data from both Study 1 (n = 20) and Study 2 (n = 19). The model estimation process completed successfully and yielded a marginal log-likelihood of -308.312 and a posterior predictive p-value (PPP) of 0.442, demonstrating an acceptable model fit. While the full set of regression paths is detailed in Table A3 , the most statistically informative paths—those with 95% credible intervals (CI) not containing zero—are highlighted in Fig. 1. As the majority of the fitted model’s regression paths had 95% CIs that included zero, indicating inconclusive evidence, they will not be analyzed in detail. However, two specific associations deserve more detailed attention. The first is the link between self-regulated learning (SRL) and metacognitive engagement, which yielded a statistically informative positive link in Study 2, despite being negative and inconclusive in Study 1. Secondly, we will assess the subsequent, theoretically important path from metacognitive engagement to exam performance, for which negative but inconclusive links were found in in both studies. In Study 2, evidence for a positive association was observed between SRL and metacognitive engagement, with a mean posterior coefficient estimate of 0.542 (95% CI [0.01, 1.04]). With 97.7% of posterior samples supporting a positive coefficient and a CI that excludes zero, we can state with reasonable certainty that higher SRL was associated with increased metacognitive engagement. However, the width of the CI suggests considerable uncertainty regarding the magnitude of this effect; it could be small or even negligible. Conversely, in Study 1, the relationship between SRL and metacognitive engagement tended to be negative, with a mean posterior estimate of -0.115 (95% CI [-0.686, 0.467]). However, with the CI being so wide and centered on zero and with only 65% of posterior samples supporting a negative coefficient, there was insufficient evidence to conclude a directional effect. Regarding the association between metacognitive engagement and exam performance, the findings also pointed to negative but inconclusive links across both studies. In Study 1, the mean posterior estimate was − 0.089 (95% CI [-0.485, 0.304]), and in Study 2, it was − 0.497 (95% CI [-1.048, 0.072]). Although a high percentage of posterior samples supported a negative relationship in Study 2 (95.8%), the inclusion of zero in the CIs for both studies precluded us from confidently supporting a negative link between metacognitive engagement and exam performance. Study 3 Our structural equation model, comprising 126 parameters and 24 equality constraints, was applied to the two experimental groups (S3C: n = 63; S3Q: n = 59). The model estimation process completed successfully and yielded a marginal log-likelihood of -1416.133, and the posterior predictive p-value (PPP) of .504 indicated good model fit. As with the first model, the full set of regression paths for the second model is provided in Table A4 , with the most statistically informative paths visible in Fig. 2. The model identified several statistically informative paths. This included some evidence that female students reported lower academic self-efficacy (GASE) than males, alongside convincing evidence of a more negative attitude toward AI (AI_ATT), an effect of likely moderate to large size. However, here we focus on the association between SRL and chatbot engagement by examining the tendency for negative relationships with both metacognitive and non-metacognitive engagement. Two additional engagement-related effects, which appeared only in the chatbot-first group, are detailed at the end. Regarding the link between SRL and metacognitive engagement, the strength of evidence for a negative relationship differed between the groups. Evidence for a negative effect emerged in the chatbot-first group (mean posterior estimate = − 0.446, 95% CI [–0.842, − 0.053]), with 98.5% of posterior samples supporting this trend. However, the interval’s width suggests uncertainty about the magnitude of the effect, which could be very small. In contrast, while the second group also pointed toward a negative trend (mean estimate = − 0.079), the evidence remained inconclusive, as its 95% credible interval [–.313, 0.156] overlapped with zero and was supported by only 75.1% of posterior samples. The model also provided consistent evidence for a negative relationship between SRL and non-metacognitive engagement across both groups. This effect was robustly supported in the chatbot-first group (mean estimate = − 0.485, 95% CI [–0.836, − 0.146]) and to some extent in the chatbot-second group (mean estimate = − 0.317, 95% CI [–0.638, − 0.006]), with 99.7% and 97.7% of posterior samples supporting the negative trend, respectively. As both 95% CIs excluded zero, we can conclude that higher SRL was associated with lower non-metacognitive engagement. The effect size is likely not negligible in the chatbot-first group, where the interval’s upper bound was further from zero. Finally, two additional paths emerged in the chatbot-first group: a negative association between performance orientation (PERF) and metacognitive engagement (mean estimate = − 0.362, 95% CI [–0.709, − 0.004]), and a positive association between academic self-efficacy (GASE) and non-metacognitive engagement (mean estimate = 0.516, 95% CI [0.174, 0.843]), supported by 97.6% and 99.8% of posterior samples, respectively. Discussion We examined the potential of LLM-powered chatbots to support metacognitive reflection through three complementary studies in different educational settings. Drawing on the outcomes and insights from these studies, we explored (1) whether an LLM-powered chatbot can effectively co-construct metacognitive dialogue with students using only system-level instructions; (2) patterns of student engagement and whether they relate to individual characteristics, such as self-regulated learning ability, academic self-efficacy and attitudes towards AI; and (3) the extent to which chatbot-assisted metacognitive reflection translates into improved academic performance. From System Instructions to Metacognitive Dialogue Our findings indicate that the chatbot consistently maintained a strong focus on metacognitive dialogue (median 90–100% of characters), even when students were somewhat less consistent in doing so (median 75.7–94.0%). In Studies 2 and 3, the chatbot systematically (~ 100% of cases) queried students about their study preparation and learning strategies. Consistent with prior research (Dunlosky et al., 2013 ; Morehead et al., 2016 ; Pintrich, 2002 ), students initially reported limited use of high-utility strategies such as distributed practice, interleaved practice, elaboration, and practice testing. The chatbot responded by offering evidence-based recommendations for learning techniques and co-constructing concrete study plans that incorporated these techniques. In Study 3, students rated the chatbot as particularly useful for suggesting effective learning strategies. Student Engagement: A Passive Role? Despite the chatbot’s effectiveness in structuring conversations around metacognitive reflection and evidence-based learning strategies, overall student engagement was low. This aligns with emerging evidence that students prefer using generative AI for direct learning support, such as requesting explanations, rather than for the more indirect process of metacognitive self-regulation (Spirgi and Seufert ( 2025 )). Low interest in metacognitive reflection was evident in our data. Participation was low (21.4% uptake in Study 1), and interactions were brief, with students exchanging a median of just 27 messages over an entire semester. Across all studies, students’ contributions were consistently short (~ 90–200 words), suggesting a largely passive role. Interestingly, our data also tentatively suggest that students used the chatbot in a way that matched their existing learning habits. In Study 3, for instance, we found a positive association between students who reported using elaboration in their learning and their non-metacognitive engagement with the chatbot, such as asking it to explain concepts. However, when forced to engage in a metacognitive dialogue rather than receiving direct learning support, students may be reluctant to adopt a more active role. The engagement challenges we observed are not unique and echo findings from other metacognitive chatbot systems (Martins et al., 2024 ). They highlight a critical design imperative: chatbots should be designed to scaffold student reflection, not replace it, thereby aligning with the growing consensus on collaborative human-AI partnerships (Fan et al., 2024 ). Future work should explore how to adapt chatbot behavior to elicit, rather than provide, extensive content, and shift the interaction from passive consumption to the active reflection that is essential for learning. The Inconsistent Link with Self-Regulated Learning Contrary to previous research suggesting that preexisting metacognitive skills are crucial for the success of interventions designed to stimulate them (Jansen et al. ( 2019 )), we found that metacognitive engagement levels were not systematically positively associated with students’ self-regulated learning (SRL) or other individual difference variables such as digital competence or attitudes toward AI. This lack of a consistent, positive relationship aligns with other recent work which also did not find a positive relationship between students’ metacognitive skills and chatbot usage frequency in their Moodle-integrated chatbot for metacognitive reflection (Sáiz-Manzanares et al. ( 2023 )). Concerning SRL, while Study 2 showed a positive association with metacognitive engagement, Studies 1 and 3 tended towards negative associations. This negative association was particularly evident when students completed self-reports after interacting with the chatbot, raising the possibility that the conversation made them more acutely aware of gaps in their SRL competencies. Given that SRL measurement relied on self-reports, which are sensitive to timing and may not capture actual behaviors, we need to be very cautious when interpreting these results. While Study 3 attempted to control for this, the results were not definitive. More Talk, No Gains? Contrary to expectations, we found no evidence that talking with the chatbot improved learning outcomes. In fact, we observed inconclusive negative trends between chatbot-assisted metacognitive engagement and exam scores in Studies 1 and 2. This is at odds with prior research showing that structured metacognitive interventions, including chatbot-assisted interventions (Graesser et al., 2004 ; Yin et al., 2024 ), can improve academic outcomes by fostering self-regulated learning (Amzil, 2014 ; Cook et al., 2013 ; Maftoon & Alamdari, 2020 ). However, we need to consider our sample size limitations and grading variability across courses. Another consideration is that quantity of engagement may be a limited metric for assessing how students benefit from chatbot-assisted metacognitive reflection. For instance, Hobert et al. ( 2023 ) found that increasing time spent conversing with a chatbot, through a more constructive and interactive dialogue, led to greater perceived learning but did not improve objective learning outcomes. Therefore, simply increasing message volume or time on task is not a guarantee for better outcomes. Moreover, Sáiz-Manzanares et al. ( 2023 ) noted that students tended to ask primarily low-level questions focused on basic clarification of metacognitive concepts, rather than truly leveraging the potential for metacognitive reflection. These findings suggest that the quality of interaction, rather than merely the amount of time or message volume, determines the benefits of metacognitive reflection. Identifying what constitutes a high-quality dialogue and what the minimum effective dose of engagement might be remain key open questions for future research. Future Directions: Metacognitive Checkpoints Our findings raise a central design question: can system-level prompts alone guide LLM-based chatbots to support metacognitive reflection effectively? While our results suggest that the chatbot was able to construct a dialogue around metacognitive reflection, student engagement remained limited. The inherent flexibility of LLMs, while advantageous for generating context-sensitive dialogue, may also lead to the chatbot interpreting its instructions too loosely, undermining the pedagogical goals. Future work should explore multi-agent or hybrid systems that combine the adaptability of LLMs with structured instructional scaffolds. For example, adding an additional layer of scripting, such as reflection checkpoints, could help structure interactions and ensure that students meet metacognitive goals. Such hybrid systems may offer the best of both worlds: fluid, natural conversations grounded in instructional design that enforces deeper student engagement. Moreover, addressing barriers to initial and sustained engagement remains critical. As Sáiz-Manzanares et al. ( 2023 ) pointed out, students may need explicit instruction on how to use chatbots effectively. Integration into course workflows, such as timely prompts or reflection windows, could also support continued use. In parallel, chatbot-supported metacognitive reflection could be embedded into evolving assessment practices that focus on learning as a process. For instance, AI-assisted e-portfolios could help students document, reflect on, and refine their learning strategies over time, making metacognitive reflection an ongoing, contextualized part of their academic journey. Conclusion Our findings suggest that LLM-powered chatbots, when guided by well-crafted system prompts, can effectively structure dialogue around metacognitive reflection and deliver relevant content. However, promoting active student engagement remains a key challenge. Future research should move beyond measuring engagement by quantity alone and develop approaches to assess the quality and structure of learner interactions. Increasing meaningful engagement may require hybrid designs that combine the flexibility of LLMs with lightweight instructional scaffolds to ensure students meet metacognitive goals. Ultimately, the successful integration of AI-driven reflection tools will depend not only on technological sophistication but also on thoughtful design that emphasizes the learner’s active role in the reflective process. Declarations Competing Interests : The authors have no conflicts of interest to disclose. Funding : This study was supported by a grant (Project #7: https://belearn.swiss/en/projekt/learning-companion/) from BeLEARN (https://belearn.swiss/). Author Contributions: Classified using the Contributor Role Taxonomy (CRediT; https://credit.niso.org/) as follows: Kim Uittenhove : writing – original draft, data curation, formal analysis, project administration, visualization, and investigation. Andrew Ellis : conceptualization, methodology, resources, software, and writing – review & editing. Fabian Mumenthaler : conceptualization, methodology, project administration, investigation, and writing – review & editing. Ioana Gatzka : funding acquisition, resources, and conceptualization. Patrick Jermann : conceptualization, funding acquisition, methodology, project administration, supervision, and writing – review & editing Data Availability : The datasets comprising self-reports, chatbot engagement metrics, and mention of learning strategies are available on zenodo: https://doi.org/10.5281/zenodo.15676346 Acknowledgement The authors express their gratitude to BeLEARN (https://belearn.swiss/) for the grant that supported this study. The authors are moreover grateful to Thierry Schluchter, Meike Lietz, and Christoph Till from the University of Teacher Education Bern, to Jörg Berkel and Markus Tiede from the Bern University of Applied Sciences. We are also grateful to Caroline Sahli Lozano from PHBern (Professor at the Institute for Research, Development, and Evaluation) and Michael Eckhart from PHBern (Professor at the Institute of Special Education) for their letters of support. References Amzil, A. (2014). The effect of a metacognitive intervention on college students’ reading performance and metacognitive skills. Journal of Educational and Developmental Psychology , 4 (1), 27–45. https://doi.org/10.5539/jedp.v4n1p27 Bastani, H., Bastani, O., Sungu, A., Ge, H., Kabakcı, Ö., & Mariman, R. (2024). Generative AI can harm learning . https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4895486 Bjork, R. A., Dunlosky, J., & Kornell, N. (2013). Self-regulated learning: Beliefs, techniques, and illusions. Annual Review of Psychology , 64 (1), 417–444. https://doi.org/10.1146/annurev-psych-113011-143823 Britos Cavagnaro, L. (2022). Chatbot guides students to learn and reflect . https://digitaleducation.stanford.edu/chatbot-guides-students-learn-and-reflect Broadbent, J., & Poon, W. L. (2015). Self-regulated learning strategies and academic achievement in online higher education learning environments: A systematic review. Internet and Higher Education , 27 , 1–13. https://doi.org/10.1016/j.iheduc.2015.04.007 Carbonell, J. R. (1970). AI in CAI: An artificial-intelligence approach to computer-assisted instruction. IEEE Transactions on Man-Machine Systems , 11 (4), 190–202. Cook, E., Kennedy, E., & McGuire, S. Y. (2013). Effect of teaching metacognitive learning strategies on performance in general chemistry courses. Journal of Chemical Education , 90 (8), 961–967. https://doi.org/10.1021/ed300686h Credé, M., & Phillips, L. A. (2011). A meta-analytic review of the motivated strategies for learning questionnaire. Learning and Individual Differences , 21 (4), 337–346. https://doi.org/10.1016/j.lindif.2011.03.002 Davis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly , 13 (3), 319–340. https://doi.org/10.2307/249008 Donoghue, G. M., & Hattie, J. A. C. (2021). A meta-analysis of ten learning techniques. Frontiers in Education , 6 , 581216. https://doi.org/10.3389/feduc.2021.581216 Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Improving students’ learning with effective learning techniques: Promising directions from cognitive and educational psychology. Psychological Science in the Public Interest , 14 (1), 4–58. https://doi.org/10.1177/1529100612453266 Elliot, A. J., & Murayama, K. (2008). On the measurement of achievement goals: Critique, illustration, and application. Journal of Educational Psychology , 100 (3), 613–628. Fan, Y., Tang, L., Le, H., Shen, K., Tan, S., Zhao, Y., & Gašević, D. (2024). Beware of metacognitive laziness: Effects of generative artificial intelligence on learning motivation, processes, and performance. British Journal of Educational Technology , Advance online publication . https://doi.org/10.1111/bjet.13544 Flavell, J. H. (1979). Metacognition and cognitive monitoring: A new area of cognitive–developmental inquiry. American Psychologist , 34 (10), 906–911. Ghomi, M., & Redecker, C. (2019). Digital competence of educators (DigCompEdu): Development and evaluation of a self-assessment instrument for teachers’ digital competence. Proceedings of the 11th International Conference on Computer Supported Education (CSEDU) , 1 , 541–548. Graesser, A. C., Lu, S., Jackson, G. T., Mitchell, H. H., Ventura, M., Olney, A., & Louwerse, M. M. (2004). AutoTutor: A tutor with dialogue in natural language. Behavior Research Methods, Instruments, & Computers , 36 , 180–192. https://doi.org/10.3758/BF03195563 Graesser, A. C., Wiemer-Hastings, K., Wiemer-Hastings, P., & Kreuz, R. (1999). AutoTutor: A simulation of a human tutor. Cognitive Systems Research , 1 (1), 35–51. https://doi.org/10.1016/S1389-0417(99)00005-4 Hardebolle, C., Tormey, R., Jermann, P., Pereira Correia Pinto, F., & Di Vincenzo, M. C. (2019). Impact on the practice of students and teachers of a learning analytics dashboard. Sciences Po International Teaching and Learning Workshop 2019: \"Integrating Pedagogical Diversity: Blended Learning and Educational Impact\" . Hidayat-ur-Rehman, I. (2024). Examining AI competence, chatbot use, and perceived autonomy as drivers of students’ engagement in informal digital learning. Journal of Research in Innovative Teaching & Learning , 17 (2), 196–212. https://doi.org/10.1108/jrit-05-2024-0136 Hobert, S., Følstad, A., & Law, E. L.-C. (2023). Chatbots for active learning: A case of phishing email identification. International Journal of Human-Computer Studies , 179 , 103108. https://doi.org/10.1016/j.ijhcs.2023.103108 Kosmyna, N., Hauptmann, E., Yuan, Y. T., Situ, J., Liao, X. H., Beresnitzky, A. V., ... & Maes, P. (2025). Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task. arXiv preprint arXiv:2506.08872 . Jansen, R. S., van Leeuwen, A., Janssen, J., Jak, S., & Kester, L. (2019). Self-regulated learning partially mediates the effect of self-regulated learning interventions on achievement in higher education: A meta-analysis. Educational Research Review , 28 , 100292. https://doi.org/https://doi.org/10.1016/j.edurev.2019.100292 Landis, J. R., & Koch, G. G. (1977). The measurement of observer agreement for categorical data. Biometrics , 33 (1), 159–174. Maftoon, P., & Alamdari, E. F. (2020). Exploring the effect of metacognitive strategy instruction on metacognitive awareness and listening performance through a process-based approach. International Journal of Listening , 34 (1), 1–20. https://doi.org/10.1080/10904018.2016.1250632 Mai, V., Neef, C., & Richert, A. (2022). Developing an AI-based coaching chatbot: A study on disclosure as effectiveness factor in human-machine-coaching. Proceedings of the International Conference on Applied Human Factors and Ergonomics (AHFE) 2022 , 651–657. https://doi.org/10.54941/ahfe100917 Martins, L., Fernández-Ferrer, M., & Puertas, E. (2024). Analysing conversation pathways with a chatbot tutor to enhance self-regulation in higher education. Education Sciences , 14 (6), 590. https://doi.org/10.3390/educsci14060590 McGuire, S. Y. (2021). Close the metacognitive equity gap: Teach all students how to learn. Journal of College Academic Support Programs , 4 (1), 69–72. https://doi.org/10.36896/4.1ep1 Merkle, E. C., & Rosseel, Y. (2018). blavaan: Bayesian structural equation models via parameter expansion . Journal of Statistical Software, 85 (4), 1–30. https://doi.org/10.18637/jss.v085.i04 Morehead, K., Rhodes, M. G., & DeLozier, S. (2016). Instructor and student knowledge of study strategies. Memory , 24 (2), 257–271. Pintrich, P. R. (2002). The role of metacognitive knowledge in learning, teaching, and assessing. Theory Into Practice , 41 (4), 219–225. https://doi.org/10.1207/s15430421tip4104_3 Pintrich, P. R., Smith, D. A. F., Garcia, T., & McKeachie, W. J. (1991a). A manual for the use of the motivated strategies for learning questionnaire (MSLQ) . National Center for Research to Improve Postsecondary Teaching; Learning. Pintrich, P. R., Smith, D. A., Garcia, T., & McKeachie, W. J. (1991b). A manual for the use of the motivated strategies for learning questionnaire (MSLQ) . University of Michigan, National Center for Research to Improve Postsecondary Teaching; Learning. Richardson, M., Abraham, C., & Bond, R. (2012). Psychological correlates of university students’ academic performance: A systematic review and meta-analysis. Psychological Bulletin , 138 , 353–387. https://doi.org/10.1037/a0026838 Sáiz-Manzanares, M. C., Marticorena-Sánchez, R., Martín-Antón, L. J., González Díez, I., & Almeida, L. (2023). Perceived satisfaction of university students with the use of chatbots as a tool for self-regulated learning. Heliyon , 9 (1), e12843. https://doi.org/10.1016/j.heliyon.2023.e12843 Schei, O. M., Møgelvang, A., & Ludvigsen, K. (2024). Perceptions and use of AI chatbots among students in higher education: A scoping review of empirical studies. Education Sciences , 14 (8), 922. https://doi.org/10.3390/educsci14080922 Schraw, G., & Dennison, R. S. (1994). Assessing metacognitive awareness. Contemporary Educational Psychology , 19 (4), 460–475. https://doi.org/10.1006/ceps.1994.1013 Spirgi, L., & Seufert, S. (2025). GenAI as a learning assistant, an empirical study in higher education. Proceedings of the 17th International Conference on Computer Supported Education (CSEDU 2025) - Volume 2 , 27–34. https://doi.org/10.5220/0013199300003932 Stein, J.-P., Messingschlager, T., Gnambs, T., Hutmacher, F., & Appel, M. (2024). Attitudes towards AI: Measurement and associations with personality. Scientific Reports , 14 , 2909. Tormey, R., & Hardebolle, C. (2017). Apprendre à étudier: Guide à l’usage des étudiants en sciences et en ingénierie . PPUR. https://infoscience.epfl.ch/handle/20.500.14299/147097 Weinstein, C. E., Palmer, D. R., & Schulte, A. C. (2002). Learning and study strategies inventory (LASSI): Second edition . H&H Publishing Company. Yin, J., Zhu, Y., Goh, T.-T., Wu, W., & Hu, Y. (2024). Using educational chatbots with metacognitive feedback to improve science learning. Applied Sciences , 14 (20), 9345. https://doi.org/10.3390/app14209345 Zyl, L. E. van, Klibert, J., Shankland, R., See-To, E. W. K., & Rothmann, S. (2022). The general academic self-efficacy scale: Psychometric properties, longitudinal invariance, and criterion validity. Journal of Psychoeducational Assessment , 40 (6), 777–789. Additional Declarations No competing interests reported. Supplementary Files Appendix.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {\"props\":{\"pageProps\":{\"initialData\":{\"identity\":\"rs-6973046\",\"acceptedTermsAndConditions\":true,\"allowDirectSubmit\":true,\"archivedVersions\":[],\"articleType\":\"Research Article\",\"associatedPublications\":[],\"authors\":[{\"id\":514415400,\"identity\":\"e31ff3fa-7cb1-4a4b-9789-4cb2c9cbf8d7\",\"order_by\":0,\"name\":\"Kim Uittenhove\",\"email\":\"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA6ElEQVRIiWNgGAWjYBACCWYQaQAieECEDQMD8+EGBsYGgloMJKBa0hgY2BIJaIFQcC2HCWuRbOdOYOYp+FPHL9178DFPxXk5czag+p87cGuRZubdwMwDdJjknHPJxjxnbhtbtjE2MPaewa1FDqbF4EaOmeTMttuJG+43NjAzthGhxR6s5d+5+g3HGPFrgTvMQCLHTOJjw4EEA0JaJJt5NxycY2AsOeNGjrHBh2PJhjuBfjnYi0eLxPmzGx+8+SPHzz8jx/BBQo2dvDkb88EHP/FoAYEDKDwDDBGCwIA05aNgFIyCUTACAAD9DkoOuQePBQAAAABJRU5ErkJggg==\",\"orcid\":\"\",\"institution\":\"Ecole Polytechnique Federale de Lausanne\",\"correspondingAuthor\":true,\"prefix\":\"\",\"firstName\":\"Kim\",\"middleName\":\"\",\"lastName\":\"Uittenhove\",\"suffix\":\"\"},{\"id\":514415401,\"identity\":\"91b45792-c000-4985-9872-fd4b87c33774\",\"order_by\":1,\"name\":\"Andrew Ellis\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Bern University of Applied Sciences\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Andrew\",\"middleName\":\"\",\"lastName\":\"Ellis\",\"suffix\":\"\"},{\"id\":514415402,\"identity\":\"5e490531-9b7d-4e0c-b857-04565fba8fe9\",\"order_by\":2,\"name\":\"Fabian Mumenthaler\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Ecole Polytechnique Federale de Lausanne\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Fabian\",\"middleName\":\"\",\"lastName\":\"Mumenthaler\",\"suffix\":\"\"},{\"id\":514415403,\"identity\":\"2a8cf5a8-bf9e-4117-8ef8-4bc28c3d3030\",\"order_by\":3,\"name\":\"Ioana Gatzka\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Bern University of Applied Sciences\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Ioana\",\"middleName\":\"\",\"lastName\":\"Gatzka\",\"suffix\":\"\"},{\"id\":514415404,\"identity\":\"59a07932-c925-4442-b01a-323778140a17\",\"order_by\":4,\"name\":\"Patrick Jermann\",\"email\":\"\",\"orcid\":\"\",\"institution\":\"Ecole Polytechnique Federale de Lausanne\",\"correspondingAuthor\":false,\"prefix\":\"\",\"firstName\":\"Patrick\",\"middleName\":\"\",\"lastName\":\"Jermann\",\"suffix\":\"\"}],\"badges\":[],\"createdAt\":\"2025-06-25 09:38:20\",\"currentVersionCode\":1,\"declarations\":\"\",\"doi\":\"10.21203/rs.3.rs-6973046/v1\",\"doiUrl\":\"https://doi.org/10.21203/rs.3.rs-6973046/v1\",\"draftVersion\":[],\"editorialEvents\":[],\"editorialNote\":\"\",\"failedWorkflow\":false,\"files\":[{\"id\":92476115,\"identity\":\"fce9990d-2210-4c83-8757-470e7ea8ae4f\",\"added_by\":\"auto\",\"created_at\":\"2025-09-30 07:19:58\",\"extension\":\"docx\",\"order_by\":0,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"acdc-reference\",\"size\":979662,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"chatbotsmetacognitivereflectionv1.docx\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6973046/v1/067fc99bd4403eb12154fc94.docx\"},{\"id\":92478260,\"identity\":\"62e35756-25de-4b13-a732-6f94addd9511\",\"added_by\":\"auto\",\"created_at\":\"2025-09-30 07:28:20\",\"extension\":\"json\",\"order_by\":1,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"acdc-reference\",\"size\":7779,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"3f64b166add74106ba0ecca62c4e783d.json\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6973046/v1/eecd9948bd5aed10c9f4b9d2.json\"},{\"id\":92478202,\"identity\":\"ab057091-d211-4921-b5fc-aa5de0838220\",\"added_by\":\"auto\",\"created_at\":\"2025-09-30 07:28:12\",\"extension\":\"xml\",\"order_by\":2,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"acdc-reference\",\"size\":256013,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"3f64b166add74106ba0ecca62c4e783d1enriched.xml\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6973046/v1/26060cb1519eaa950b7c7912.xml\"},{\"id\":92478188,\"identity\":\"119b29a1-f66b-4f56-8e0e-82c79053bcb7\",\"added_by\":\"auto\",\"created_at\":\"2025-09-30 07:27:58\",\"extension\":\"jpeg\",\"order_by\":3,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"acdc-reference\",\"size\":232456,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"floatimage1.jpeg\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6973046/v1/2049e8b5ee14bdeb86b0d9c5.jpeg\"},{\"id\":92476114,\"identity\":\"8eeda022-a1a1-4157-9081-5233c6fe8142\",\"added_by\":\"auto\",\"created_at\":\"2025-09-30 07:19:58\",\"extension\":\"png\",\"order_by\":4,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"acdc-reference\",\"size\":159364,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"floatimage2.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6973046/v1/8ac201093527050f099bdff6.png\"},{\"id\":92476121,\"identity\":\"59aacd70-2161-48fe-9daf-d9099bc40384\",\"added_by\":\"auto\",\"created_at\":\"2025-09-30 07:19:59\",\"extension\":\"png\",\"order_by\":5,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"acdc-reference\",\"size\":397460,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"floatimage3.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6973046/v1/dab2b81a7258c5da93ae0dc9.png\"},{\"id\":92476119,\"identity\":\"0bd0bc60-0577-4eec-a19c-bad3d7daaaff\",\"added_by\":\"auto\",\"created_at\":\"2025-09-30 07:19:59\",\"extension\":\"png\",\"order_by\":6,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"acdc-reference\",\"size\":35025,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"Onlinefloatimage1.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6973046/v1/3fb66d6eb4204b44a1eca059.png\"},{\"id\":92478189,\"identity\":\"30cf1bf1-ae45-4c19-ab7c-0b542ea3a365\",\"added_by\":\"auto\",\"created_at\":\"2025-09-30 07:27:59\",\"extension\":\"png\",\"order_by\":7,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"acdc-reference\",\"size\":45401,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"Onlinefloatimage2.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6973046/v1/6981085721f82f2b1bb98d65.png\"},{\"id\":92476116,\"identity\":\"9e3f3312-442a-40b7-8f66-46c07e0ec662\",\"added_by\":\"auto\",\"created_at\":\"2025-09-30 07:19:58\",\"extension\":\"png\",\"order_by\":8,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"acdc-reference\",\"size\":109285,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"Onlinefloatimage3.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6973046/v1/e20591c350fb27fcbcff8760.png\"},{\"id\":92476123,\"identity\":\"45606455-455a-48ef-89de-0dfc13364f89\",\"added_by\":\"auto\",\"created_at\":\"2025-09-30 07:19:59\",\"extension\":\"xml\",\"order_by\":9,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"acdc-reference\",\"size\":253327,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"3f64b166add74106ba0ecca62c4e783d1structuring.xml\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6973046/v1/531a5f16d524ffe88e274c2b.xml\"},{\"id\":92476125,\"identity\":\"d711e5fb-71bb-4a83-ae35-363532552062\",\"added_by\":\"auto\",\"created_at\":\"2025-09-30 07:19:59\",\"extension\":\"html\",\"order_by\":10,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"acdc-reference\",\"size\":274720,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"earlyproof.html\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6973046/v1/f5b360473a8ec3b170cbaa9b.html\"},{\"id\":92478187,\"identity\":\"1bec6a4f-f81c-435a-91cb-3a3846f8ee2a\",\"added_by\":\"auto\",\"created_at\":\"2025-09-30 07:27:58\",\"extension\":\"png\",\"order_by\":1,\"title\":\"Figure 1\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":50242,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003eStructural Equation Model for Study 1 and 2 with Indication of Statistically Informative Paths\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"1.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6973046/v1/3812bb6ffc008680896ec070.png\"},{\"id\":92476112,\"identity\":\"192f9500-fe2e-4f46-8e8c-599b0f0ef7e4\",\"added_by\":\"auto\",\"created_at\":\"2025-09-30 07:19:58\",\"extension\":\"png\",\"order_by\":2,\"title\":\"Figure 2\",\"display\":\"\",\"copyAsset\":false,\"role\":\"figure\",\"size\":136758,\"visible\":true,\"origin\":\"\",\"legend\":\"\\u003cp\\u003eStructural Equation Model for Study 3 with Indication of Statistically Informative Paths\\u003c/p\\u003e\",\"description\":\"\",\"filename\":\"2.png\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6973046/v1/c7d7a6fe7f5dd6b6932db2f6.png\"},{\"id\":93633095,\"identity\":\"2c92586a-9ffc-487e-abd9-e74053c90db2\",\"added_by\":\"auto\",\"created_at\":\"2025-10-15 23:31:32\",\"extension\":\"pdf\",\"order_by\":0,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"manuscript-pdf\",\"size\":1307410,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"manuscript.pdf\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6973046/v1/894fc470-de8d-432f-9f37-e5765b1c62a8.pdf\"},{\"id\":92476110,\"identity\":\"4c2d4fcd-17ba-42a3-a4ed-5d0a17506f76\",\"added_by\":\"auto\",\"created_at\":\"2025-09-30 07:19:58\",\"extension\":\"docx\",\"order_by\":1,\"title\":\"\",\"display\":\"\",\"copyAsset\":false,\"role\":\"supplement\",\"size\":55671,\"visible\":true,\"origin\":\"\",\"legend\":\"\",\"description\":\"\",\"filename\":\"Appendix.docx\",\"url\":\"https://assets-eu.researchsquare.com/files/rs-6973046/v1/c59efbc0083fc0ee42fc3996.docx\"}],\"financialInterests\":\"No competing interests reported.\",\"formattedTitle\":\"Metacognitive Reflection in the Era of Generative AI\",\"fulltext\":[{\"header\":\"Introduction\",\"content\":\"\\u003cp\\u003eSelf-regulated learning, comprising metacognitive self-regulation (Credé \\u0026amp; Phillips, 2011), the awareness and regulation of one’s own thinking processes (Flavell, 1979), is widely recognized as crucial for effective learning and academic success (Broadbent \\u0026amp; Poon, 2015; Jansen et al., 2019; Richardson et al., 2012). In practice, this entails many aspects, including motivational aspects, beliefs such as academic self-efficacy, goal orientation, cognitive aspects such as elaboration and organization, and metacognitive aspects such as planning, monitoring, self-regulation, and evaluation (Pintrich et al., 1991a; Schraw \\u0026amp; Dennison, 1994), all contributing to the effective use of high-utility learning strategies (Bjork et al., 2013; Donoghue \\u0026amp; Hattie, 2021; Dunlosky et al., 2013; Tormey \\u0026amp; Hardebolle, 2017). For details on effective learning techniques, see appendix Section 6.\\u003c/p\\u003e\\n\\u003cp\\u003eThis set of skills is transversal and supports lifelong learning and adaptability across disciplines and changing circumstances. The necessity of cultivating such adaptability has become increasingly important in the 21st century, where knowledge and technology evolve constantly and rapidly, changing how we teach and learn. The rise of generative AI in education in particular is challenging us to examine and potentially reimagine our methods of teaching and learning. Tools like ChatGPT or Gemini can effortlessly and instantly generate solutions and content across academic disciplines, and without guidance, students might over-rely on AI assistance and forgo the mental effort of reflection. Recent studies warn that such over-reliance can worsen learning outcomes (Bastani et al., 2024, Kosmyna et al., 2025) and lead to “metacognitive laziness”, where learners become dependent on AI and neglect their own self-monitoring and critical thinking (Fan et al., 2024). Therefore, metacognitive reflection must be actively taught and stimulated, to navigate a learning environment saturated with AI assistance, and ensuring that learners engage in effective learning unharmed by the presence of AI tools often used to bypass the learning process.\\u003c/p\\u003e\\n\\u003cp\\u003eAt the same time that generative AI creates challenges for student learning, it can also serve as a tool to address these same challenges. For instance, chatbots based on generative AI can identify gaps in self-regulated learning and engage in metacognitive reflective dialogue. Thus, instead of steering students away from using chatbots, we can steer them towards using these tools in a way that is helpful for their learning. This dual role of generative AI, as both a driver of the increased need for metacognitive reflection, and a tool for stimulating it, sets the stage for re-imagining how we coach students in reflective practices.\\u003c/p\\u003e\\n\\u003ch2\\u003eChatbots for Metacognitive Reflection\\u003c/h2\\u003e\\n\\u003cp\\u003eGiven the importance of metacognitive reflection and self-regulation, a crucial question is how we concretely help students develop these skills. Unfortunately, many students do not pick up these skills autonomously (Pintrich, 2002) and struggle with effective learning (Dunlosky et al., 2013; Morehead et al., 2016). Among minority and low socio-economic status groups, this issue is exacerbated as documented by a metacognitive skills gap (McGuire, 2021). Therefore, researchers have argued for the explicit teaching of metacognitive reflection and learning strategies.\\u003c/p\\u003e\\n\\u003cp\\u003eStudents can be directly instructed on metacognitive strategies and the plan–monitor–evaluate cycle for learning. For example, Amzil (2014) provided college students with five sessions of explicit instruction that focused on metacognitive processes for reading, leading to improved reading comprehension and metacognitive awareness. Likewise, Maftoon and Alamdari (2020) tested a 10-week intervention with metacognitive strategy lessons on planning, monitoring, and self-evaluation, leading to improved listening comprehension. These interventions suggest that explicit instruction can change academic behaviors and improve learning outcomes (Cook et al., 2013).\\u003c/p\\u003e\\n\\u003cp\\u003eHowever, rather than teaching generic metacognitive skills in isolation, embedding reflective activities within regular coursework may improve transfer and application to authentic learning situations. For example, reflective assignments such as learning journals, diaries, or structured peer discussions allow students to engage with metacognitive reflection in context, and prompt students to articulate what they learned, the strategies they used, and the challenges they faced. These activities make the invisible process of learning more visible to the learner, reinforcing the habit of reflection. Digital tools have expanded the range of options to support these activities, for example through E-portfolios and features in Learning Management Systems, such as personalized learning dashboards. The Learning Companion at EPFL is one such initiative that promotes student self-monitoring and reflection (Hardebolle et al., 2019; https://companion.epfl.ch/]. However, a persistent challenge remains: motivating students to engage meaningfully and consistently with these tools.\\u003c/p\\u003e\\n\\u003cp\\u003eAmong digital technologies, chatbots present a promising approach to scaffolding metacognitive reflection and engaging students through conversational interaction. Emulating the role of a teacher or mentor, chatbots can prompt students with reflective questions and provide adaptive feedback. Early implementations (Carbonell, 1970; Graesser et al., 1999) typically relied on rule-based or decision-tree architectures, guiding learners through predefined pathways. For instance, these chatbots have been used to guide students through a series of reflective questions to deepen understanding of a course topic (Britos Cavagnaro, 2022; Carbonell, 1970; Graesser et al., 1999), to support self-regulated learning and receive metacognitive feedback (Martins et al., 2024; Yin et al., 2024), and to offer coaching for challenges such as exam anxiety (Mai et al., 2022).\\u003c/p\\u003e\\n\\u003cp\\u003eSome of these systems integrated basic natural language processing (NLP) to interpret student input. A well-known example is AutoTutor (Graesser et al., 2004), which uses a hybrid architecture that integrates rule-based scripts with semantic analysis. Similarly, Sáiz-Manzanares et al. (2023) integrated a hybrid chatbot into the Moodle environment to support students’ metacognitive and self-regulation strategies. The advantage of such systems is their ability to scaffold reflection through a semi-structured conversation, prompting students to articulate their thinking, elaborate on shallow responses, and iteratively refine their understanding. Chatbots can be deployed at optimal moments, such as immediately after a lesson or upon receiving feedback, ensuring that reflection occurs when it is most timely and contextually relevant. Moreover, unlike human instructors, who may be limited in providing individualized support at scale, chatbots can engage in parallel, one-on-one reflective conversations with many students simultaneously, ensuring that every student receives individualized guidance. Several studies report positive outcomes from such chatbot-assisted interventions. These include gains in metacognitive awareness (Yin et al., 2024) and in some cases, improved learning outcomes (Graesser et al., 2004; Yin et al., 2024).\\u003c/p\\u003e\\n\\u003cp\\u003eThe emergence of generative AI and large language models (LLMs) has greatly accelerated these developments. LLM-powered chatbots can interpret and respond to free-form student input with remarkable fluency, enabling natural, personalized interactions across a wide range of contexts. Unlike rule-based or hybrid systems, designing a system where users directly and continuously interact with a single LLM requires less authoring effort. Chatbot behavior can be steered through a well-designed system prompt rather than extensive hard-coded dialogue trees, making them highly scalable and generalizable across domains. However, while LLM-based chatbots excel at flexibly generating contextually appropriate responses, they lack inherent mechanisms for ensuring that pedagogical scaffolding is achieved. This raises a critical design question: to what extent can LLM-based chatbots achieve effective scaffolding solely through prompt engineering?\\u003c/p\\u003e\\n\\u003ch3\\u003eStudent Engagement\\u003c/h3\\u003e\\n\\u003cp\\u003eA central challenge in the success of interventions aimed at promoting metacognitive reflection is sustaining student engagement. Chatbots based on rules or decision trees often struggle to stimulate student responsiveness. Martins et al. (2024) analyzed sessions with a self-regulation coaching chatbot and reported that most frequently, conversations were composed of only four messages. The authors noted that such a brief interaction is insufficient for developing complex abilities like those required for self-regulated learning. Moreover, whereas many students tried the chatbot on the first day it was introduced, usage quickly dropped off as the semester went on. Similarly, Graesser et al. (2004), with their course concept tutoring chatbot, found that students initially provide only minimal responses to the chatbot questions, and that many dialogue turns are needed to elicit deeper explanations.\\u003c/p\\u003e\\n\\u003cp\\u003eIn contrast, chatbots powered by LLMs may offer a more compelling and responsive conversational experience, potentially improving engagement. However, simply deploying a capable system does not guarantee active participation in chatbot-assisted metacognitive reflection. To fully develop the potential of LLM-based chatbots for metacognitive reflection, it is essential to understand students’ engagement patterns and the factors that influence them.\\u003c/p\\u003e\\n\\u003cp\\u003eStudent engagement in this context is likely shaped and influenced by multiple interrelated factors. The Technology Acceptance Model (TAM), highlights perceived usefulness and ease of use as key determinants of digital engagement (Davis, 1989). Supporting this, a scoping review conducted by Schei et al. (2024) on how students in higher education perceive AI chatbots found that students generally find AI chatbots to be highly useful and motivating, particularly appreciating the immediacy of feedback provided.\\u003c/p\\u003e\\n\\u003cp\\u003eIn addition to usability, digital competence could influence students’ engagement with chatbots. Students with higher digital skills might find it easier to integrate chatbot interactions into their learning, while those with lower competence could be more hesitant. For instance, AI-related competence has been positively linked to student engagement in digital learning environments (Hidayat-ur-Rehman, 2024). However, given the intuitive design of most LLM-based chatbots (i.e., low AI-related competence is needed), attitudes toward AI may be an even stronger predictor. Students with favorable attitudes may demonstrate deeper and more frequent interactions with chatbot-based tools, whereas skepticism toward AI may result in superficial engagement or resistance.\\u003c/p\\u003e\\n\\u003cp\\u003eFinally, a student’s existing level of metacognitive self-regulation may influence the effectiveness of reflection activities, as suggested by a meta-analysis by Jansen et al. (2019), which found that preexisting self-regulated learning skills partially mediate the impact of self-regulated learning interventions aimed at improving academic outcomes. This underscores the importance of considering preexisting metacognitive competencies. Those who already plan, monitor, and evaluate their learning may be more receptive to chatbot prompts that encourage reflection, while those with less-developed self-regulation habits may be less responsive.\\u003c/p\\u003e\\n\\u003cp\\u003eIn sum, engagement in chatbot-assisted metacognitive reflection depends on a constellation of learner-specific factors, ranging from perceived usefulness and digital skills to attitudes toward AI and prior self-regulatory skills. Identifying and addressing these factors is critical for understanding how to design interventions that maximize educational benefits by reaching all students and actively engaging them.\\u003c/p\\u003e\\n\\u003ch2\\u003eThe Present Study\\u003c/h2\\u003e\\n\\u003cp\\u003eGiven the increased need for metacognitive reflection engendered by the emergence of generative AI, and the potential of this technology to support such reflection, our study employed a state-of-the-art large language model (LLM) to power a chatbot with the goal of supporting metacognitive reflection among students. The chatbot was instructed through a system prompt to co-construct a dialogue aimed at stimulating metacognitive reflection in students, deepening their understanding of learning strategies, and provide evidence-based recommendations to improve study preparation practices.\\u003c/p\\u003e\\n\\u003cp\\u003eWe conducted three interconnected studies designed to explore the chatbot’s application across different educational contexts:\\u003c/p\\u003e\\n\\u003cul type=\\\"disc\\\"\\u003e\\n\\u003cli\\u003e\\u003cstrong\\u003eStudy 1\\u003c/strong\\u003e examined chatbot-supported metacognitive reflection for general course preparation throughout a full semester at the University for Teacher Education in Bern.\\u003c/li\\u003e\\n\\u003cli\\u003e\\u003cstrong\\u003eStudy 2\\u003c/strong\\u003e focused on the chatbot’s application specifically for exam preparation within a single session at Bern University of Applied Sciences during the semester’s conclusion.\\u003c/li\\u003e\\n\\u003cli\\u003e\\u003cstrong\\u003eStudy 3\\u003c/strong\\u003e investigated chatbot interactions in a diverse sample recruited through an online platform, varying the order of chatbot engagement and self-report collection. The sample was balanced between social sciences and STEM fields.\\u003c/li\\u003e\\n\\u003c/ul\\u003e\\n\\u003cp\\u003eWe analyzed the outcomes of the three studies centered around three primary questions:\\u003c/p\\u003e\\n\\u003col start=\\\"1\\\" type=\\\"1\\\"\\u003e\\n\\u003cli\\u003eCan an LLM-powered chatbot, guided solely by system-level instructions that define the metacognitive goals, effectively co-construct a dialogue that support students in metacognitive reflection? This includes helping students articulate their study strategies, reflect on their learning processes, and receive evidence-based recommendations for improvement.\\u003c/li\\u003e\\n\\u003cli\\u003eWhat are the engagement patterns of students with chatbot-assisted metacognitive reflection? How do individual differences—specifically self-regulated learning capabilities, digital competencies, and attitudes toward AI—predict or influence these engagement patterns? Understanding these relationships helps clarify for whom and under which conditions chatbot-based interventions are most effective.\\u003c/li\\u003e\\n\\u003cli\\u003eDoes more engagement with the chatbot increase subsequent academic performance as measured by exam outcomes, potentially demonstrating tangible benefits of improved metacognitive reflection on study behaviors and academic achievement?\\u003c/li\\u003e\\n\\u003c/ol\\u003e\"},{\"header\":\"Methods\",\"content\":\"\\u003cp\\u003eParticipants across the three studies are described in detail in Table 1. All participants provided electronic consent prior to their participation in the study, and the study was approved by the EPFL Human Research Ethics Committee (approval number HREC000466).\\u003c/p\\u003e\\n\\u003cp\\u003eIn Study 1, conducted at PHBern, 159 students were given the opportunity to participate and engage with the chatbot. Of these students, 34 (21.4%) chose to engage with the chatbot, and 48 provided self-report data. The final analyzed sample comprised the 48 students who provided self-report data, which included 20 out of the 34 individuals who actively engaged with the chatbot. Fourteen students used the chatbot but did not complete the self-reports.\\u003c/p\\u003e\\n\\u003cp\\u003eStudy 2 included 49 students from BFH who had the opportunity to interact with the chatbot. Out of these, 30 students (61.2%) engaged with the chatbot. However, the analyzed sample included only 19 participants who provided both self-report data (including SRL measures) and chatbot engagement data.\\u003c/p\\u003e\\n\\u003cp\\u003eStudy 3 involved 131 student participants recruited through the Prolific platform. These participants held active student status at undergraduate or graduate levels (see Table 1 for their countries of residence). Participants were equally distributed between social sciences and STEM disciplines. Nine participants were excluded from the analysis due to reported technical issues.\\u003c/p\\u003e\\n\\u003cp\\u003eTable\\u0026nbsp;1\\u003c/p\\u003e\\n\\u003cp\\u003eParticipant Characteristics\\u003c/p\\u003e\\n\\u003cp\\u003eStudy, N (sample size), Sex (M/F), Age (mean \\u0026plusmn; standard deviation), Residence (country), Education Level, Study Field\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cimg src=\\\"data:image/png;base64,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\\\" width=\\\"746\\\" height=\\\"187\\\"\\u003e\\u003c/p\\u003e\\n\\u003ch2\\u003eApparatus\\u003c/h2\\u003e\\n\\u003cp\\u003eThe study comprised two complementary parts: an interactive chatbot system designed to support students during their reflective and learning processes, and self-report data collected through secure online platforms, specifically RedCap and Moodle. Additionally, in Study 1 and 2, following the exam session, we collected the grades of the participating students.\\u003c/p\\u003e\\n\\u003ch3\\u003eChatbot System\\u003c/h3\\u003e\\n\\u003ch4\\u003eChatbot Design.\\u0026nbsp;\\u003c/h4\\u003e\\n\\u003cp\\u003eWe designed an LLM-based chatbot capable of conducting a coaching conversation, guided by a semi-structured interview plan that includes the collection of relevant information from the student and accompanying them to reflect on their learning processes and/or apply effective study strategies. Within its context window, the LLM was provided with information from the literature on metacognitive practices and effective study techniques. Reinforced with this evidence-based approach, the chatbot encouraged students\\u0026rsquo; self-reflection and encouraged the adoption of proven approaches to learning. In Study 1, the chatbot script (see Section 7) guided the students through reflection about course material, understanding, the learning process, and preparation of the next lecture. In studies 2 (see Section 8) and 3 (see Section 9), the chatbot helped students to prepare for an exam by guiding them through reflection on study strategies, exam or project preparation, or self-regulated learning, without explaining or asking to explain the course content itself. The chatbot suggested improved learning strategies based on learning science (see Section 6). The system prompt for Study 3 was slightly revised, to have the chatbot ask about challenges experienced by the students, and to help students make concrete study schedules.\\u003c/p\\u003e\\n\\u003ch4\\u003eAuthentication and Access.\\u0026nbsp;\\u003c/h4\\u003e\\n\\u003cp\\u003eIn Studies 1 and 2, students logged in and were authenticated through their Learning Management System (LMS) and subsequently directed to a Moodle course hosted by the Bern University of Applied Sciences (BFH). This course employed Learning Tools Interoperability (LTI) to ensure secure, SSL-encrypted access to the chatbot\\u0026rsquo;s frontend. The student\\u0026rsquo;s unique identifier and the course ID were transmitted to the chatbot. In Study 3, authentication occurred via a valid Study ID provided by Prolific. In addition, a unique Prolific Personal ID was passed along to the chatbot. These unique identifiers were saved with the interaction logs and permitted us to link interaction data with participant self-reports. Once authenticated, students were able to view the chatbot\\u0026rsquo;s messages and send their responses, with the ongoing dialogue displayed in a conversation history.\\u003c/p\\u003e\\n\\u003ch4\\u003eTechnical Infrastructure.\\u0026nbsp;\\u003c/h4\\u003e\\n\\u003cp\\u003eThe chatbot interface (frontend) was a web application hosted on a server provided by Exoscale in Switzerland. Access to this interface was restricted to authenticated users, either via LMS logins (Study 1 and 2) or Prolific (Study 3). A Python-based LLM orchestration framework coordinated communication with an Azure OpenAI GPT-4 model hosted by Microsoft in Switzerland by sending secure API calls. The Azure OpenAI GPT-4 model processed user inputs and generated responses in accordance with the system instructions that we provided to customize behavior. The data processing agreement between BFH and Microsoft ensured Swiss data storage and that user data were neither used to further train the model nor employed to improve other Microsoft products. The Azure OpenAI GPT-4 Model was set up in a resource group of a BFH subscription with the geolocation of Switzerland North.\\u003c/p\\u003e\\n\\u003ch3\\u003eSelf‐Report Measures\\u003c/h3\\u003e\\n\\u003ch4\\u003eSelf‐Regulated Learning (SRL) and Elaboration (ELA).\\u0026nbsp;\\u003c/h4\\u003e\\n\\u003cp\\u003eStudents\\u0026rsquo; self-regulatory learning processes were assessed using an instrument that mainly focuses on metacognitive self-regulation, as a key aspect of self-regulated learning. The instrument mainly draws items from the Motivated Strategies for Learning Questionnaire (MSLQ; Pintrich et al. (1991b)), the Metacognitive Awareness Inventory (MAI; Schraw and Dennison (1994)), and the Learning and Study Strategies Inventory (LASSI; Weinstein et al. (2002)). It also includes additional items designed to expand scales related to planning, monitoring, regulation, and evaluation, which are the main processes involved in metacognitive self-reflection. The use of elaboration strategies during learning was treated as a separate scale. The instrument has been replicated and has shown measurement invariance in English, French, and German using Prolific samples (Uittenhove et al., in preparation), making it suitable for use in German (Study 1) and English (Studies 2 and 3). Items were rated on a 7-point Likert scale from 1 (\\u0026ldquo;Not at all me\\u0026rdquo;) to 7 (\\u0026ldquo;Totally me\\u0026rdquo;). See Section 10 to view the items of the instrument.\\u003c/p\\u003e\\n\\u003ch4\\u003eGeneral Academic Self‐Efficacy (GASE).\\u0026nbsp;\\u003c/h4\\u003e\\n\\u003cp\\u003eStudents\\u0026rsquo; beliefs in their academic capabilities were measured using the General Academic Self-Efficacy Scale (Zyl et al., 2022), a five-item instrument (see Section 11) with demonstrated longitudinal invariance and criterion validity. Each item is rated on a 5-point Likert scale ranging from 1 (\\u0026ldquo;Strongly disagree\\u0026rdquo;) to 5 (\\u0026ldquo;Strongly agree\\u0026rdquo;), and the final score is the average across items, resulting in a range from 1 to 5.\\u003c/p\\u003e\\n\\u003ch4\\u003eMastery (MAST) and Performance (PERF) Orientation.\\u0026nbsp;\\u003c/h4\\u003e\\n\\u003cp\\u003eAchievement goal orientations were assessed using the Achievement Goal Questionnaire\\u0026ndash;Revised (AGQ-R; Elliot and Murayama (2008)), which distinguishes between mastery-approach, mastery-avoidance, performance-approach, and performance-avoidance goals. This instrument is widely used and validated for capturing both the content (i.e., focus on learning versus outperforming others) and the valence (i.e., approach versus avoidance) of students\\u0026rsquo; achievement motivations. In the present study, only the approach-oriented subscales for mastery and performance were used (see Section 12). Items were rated on a 5-point Likert scale from 1 (\\u0026ldquo;Strongly disagree\\u0026rdquo;) to 5 (\\u0026ldquo;Strongly agree\\u0026rdquo;), and scores for mastery and performance were computed separately as the average of their respective items, yielding a range from 1 to 5.\\u003c/p\\u003e\\n\\u003ch4\\u003eDigital Competency (DIGI).\\u0026nbsp;\\u003c/h4\\u003e\\n\\u003cp\\u003eParticipants in Studies 1 and 2 completed the DigiCompEdu questionnaire (Ghomi \\u0026amp; Redecker, 2019), a framework developed to assess the digital competence of educators. We focused on the problem-solving domain, which evaluates how educators identify needs, creatively use technology to address them, and recognize gaps in digital competence (see Section 13). Participants rated items on a 4-point scale ranging from 0 to 3, with higher values indicating greater perceived competence. The final score represents the average of the item responses and ranges from 0 to 3.\\u003c/p\\u003e\\n\\u003ch4\\u003eAI Attitude (AI_ATT).\\u0026nbsp;\\u003c/h4\\u003e\\n\\u003cp\\u003eAttitudes toward artificial intelligence were measured using the AI Attitude Scale developed by Stein and colleagues (Stein et al., 2024). This instrument captures cognitive, affective, and behavioral responses to AI as a general-purpose technology and has shown good reliability and construct validity in capturing both positive and negative sentiments toward AI. It consists of 12 items (see Section 14) rated on a 5-point Likert scale from 1 (\\u0026ldquo;Strongly disagree\\u0026rdquo;) to 5 (\\u0026ldquo;Strongly agree\\u0026rdquo;). The total score is calculated as the sum of all items, resulting in a score range of 12 to 60.\\u003c/p\\u003e\\n\\u003ch2\\u003eProcedure\\u003c/h2\\u003e\\n\\u003cp\\u003eSee Table 2 for an overview of the self-report measures and chatbot interactions collected across the three studies.\\u003c/p\\u003e\\n\\u003cp\\u003eStudy 1 was conducted during the spring semester of 2024 at PHBern, where students were informed about the chatbot\\u0026rsquo;s availability and using it was optional but encouraged as a beneficial learning aid. Participants provided self-report measures at the beginning of the semester using REDCap. Throughout the semester, students had unrestricted access to the reflection chatbot. All students in the participating courses were permitted to interact with the chatbot, regardless of formal participation in the study.\\u003c/p\\u003e\\n\\u003cp\\u003eStudy 2 took place in a single session at the end of the semester at BFH. Students completed self-report measures at both the beginning and the end of the session, which lasted approximately one hour. Within this timeframe, students engaged with the chatbot and submitted their self-reports via the Moodle platform.\\u003c/p\\u003e\\n\\u003cp\\u003eStudy 3 was conducted at the conclusion of the autumn semester 2024 through the Prolific platform. Participants were compensated for a one-hour session, during which they provided self-report data and interacted with the chatbot. This study employed a counterbalanced design: one group first interacted with the chatbot before completing self-report measures, while the other group completed self-reports first, followed by chatbot interaction. Self-report data in Study 3 were collected using REDCap.\\u003c/p\\u003e\\n\\u003cp\\u003eTable\\u0026nbsp;2\\u003c/p\\u003e\\n\\u003cp\\u003eOrganization of Chatbot Interaction and Self-Reports\\u003c/p\\u003e\\n\\u003cp\\u003eChatbot interaction and collection of self-reports (SRL, ELA, DIGI, AI_ATT, GASE, MAST, PERF) in study 1 (S1), study 2 (S2), the study 3 group who interacted with the chatbot first (S3C), and the study 3 group who provided self-reports first (S3Q)\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cimg src=\\\"data:image/png;base64,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\\\" width=\\\"746\\\" height=\\\"241\\\"\\u003e\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cem\\u003eNote\\u003c/em\\u003e. In Study 1, we initially planned to assess SRL, ELA, and DIGI a second time following the chatbot interaction. However, only a few participants (n = 5) completed the follow-up questionnaire, so these data were not included in the analysis.\\u003c/p\\u003e\\n\\u003ch2\\u003eData Analysis\\u003c/h2\\u003e\\n\\u003ch3\\u003eIdentifying Metacognitive Interactions\\u003c/h3\\u003e\\n\\u003cp\\u003eWe analyzed chatbot interaction logs and identified student messages that aligned with the chatbot\\u0026rsquo;s intended purpose, specifically metacognitive interactions, and to identify non-metacognitive interactions, such as requests for explanations of course content or to generate text for assignments. Each message was analyzed together with the context of the preceding conversation, by sending it via Microsoft Azure OpenAI API to a GPT-4 model hosted by Microsoft, and asking the model to classify the message and provide both a response code and a justification (see Section 15).\\u003c/p\\u003e\\n\\u003cp\\u003eIn Study 1, the chatbot was instructed (see Section 7) to avoid providing theoretical explanations of the course content. Instead, its mission was to facilitate dialogue with students about the course content and encourage them to reflect on their understanding and learning process. Therefore, metacognitive interactions included discussions about course material, understanding of the course material, the learning process, and preparation for the next lecture (see Section 15). In studies 2 and 3, the chatbot was instructed to focus on exam preparation and learning strategies specifically (see Section 8; Section 9). Therefore, metacognitive messages focused on study strategies, exam or project preparation, and self-regulated learning, but did not involve explaining or requesting explanations of course content.\\u003c/p\\u003e\\n\\u003cp\\u003eTo assess the reliability of coding these interactions as metacognitive or non-metacognitive, we compared the classifications of GPT-4 to human classifications across 180 randomly selected messages of which half were classified as metacognitive by GPT-4. In 80 cases, both the human and GPT-4 assigned the non-metacognitive label (e.g., \\u0026ldquo;\\u003cem\\u003eThanks. What would you suggest is a good dissertation for a criminology dissertation on domestic Abuse)\\u003c/em\\u003e, and in 69 cases, both assigned the metacognitive label (e.g.,\\u0026rdquo;\\u003cem\\u003eSure, I get your point but since a specific seminar session is always going to focused [sic ]on a specific set of texts mixing them up may have a beneficial learning effect but it would reduce \\u0026lsquo;synergy\\u0026rsquo; with the seminar itself, if you know what I mean.\\u0026rdquo;)\\u003c/em\\u003e, resulting in 149 agreements. This corresponds to an observed agreement rate of 82.8%. To adjust for chance agreement, Cohen\\u0026rsquo;s Kappa was computed, yielding a value of approximately 0.66, which indicates substantial agreement between the label attributed by human and GPT-4, according to Landis and Koch (1977) interpretation scale. Disagreements occurred in 31 cases: in 21 instances, the human assigned the non-metacognitive label while GPT-4 assigned the metacognitive label; and vice versa in 10 other instances. Given that the classification agreement was substantial, we subsequently used the classification decisions from the GPT-4 model in our analysis.\\u003c/p\\u003e\\n\\u003cp\\u003eIn addition to identifying metacognitive and non-metacognitive interactions in Study 2 and Study 3, we also analyzed the entire conversation to capture which moderate- and high-utility learning strategies students reported using initially, and which were recommended by the chatbot, whether students initially had a study schedule and whether the chatbot contributed to creating one. We also extracted exam readiness (how ready are you for the exam on a scale from 1 to 10) (see Section 15).\\u003c/p\\u003e\\n\\u003ch3\\u003eDescriptive Analyses\\u003c/h3\\u003e\\n\\u003cp\\u003eWe conducted descriptive analyses to examine the distributions of self-report variables (SRL, ELA, GASE, MAST, PERF, AI_ATT) in each sample and to assess student engagement with the chatbot across all three studies. Engagement was measured using several metrics: the total number of messages exchanged, the total character count, and the character count of metacognitive and non-metacognitive student messages. In Studies 2 and 3, we also analyzed students\\u0026rsquo; initial reports of using moderate- and high-utility learning strategies, as well as the strategies recommended by the chatbot. Study 3 further included students\\u0026rsquo; usefulness ratings of the chatbot conversation, offering insight into their perceptions of its helpfulness for various aspects of studying.\\u003c/p\\u003e\\n\\u003ch3\\u003eBayesian Structural Equation Modeling\\u003c/h3\\u003e\\n\\u003cp\\u003eTo examine the complex relationships among demographic characteristics, self-reported measures, chatbot engagement, and academic performance, we applied Bayesian Structural Equation Modeling (BSEM). This methodological approach was selected for its capacity to simultaneously model multiple interrelated variables and facilitate multi-group analyses across our studies. The Bayesian framework was particularly appropriate for our exploratory analyses due to modest sample sizes.\\u003c/p\\u003e\\n\\u003cp\\u003eBecause chatbot engagement metrics (i.e., metacognitive and non-metacognitive student message character count) were inherently positive and right‑skewed, we applied a log transformation, then standardized all quantitative variables to improve model fitting and interpretation. Exam scores were standardized separately by exam, to account for differences in exam difficulty or teacher rating biases.\\u003c/p\\u003e\\n\\u003cp\\u003eWe estimated models using the\\u0026nbsp; bsem() function from the \\u003cem\\u003eblavaan\\u003c/em\\u003e package (Version 0.5.8; Merkle \\u0026amp; Rosseel, 2018) in R. All variables were standardized prior to estimation. Standard normal priors (\\u0026mu; = 0, \\u0026sigma; = 1) were applied to all regression coefficients, including intercepts. All BSEM models were sampled using the No-U-Turn Sampler (NUTS) algorithm with four chains, each consisting of a 1,000-iteration burn-in phase followed by 2,000 post-burn-in iterations.\\u003c/p\\u003e\\n\\u003ch4\\u003eStructural Model for Studies 1 and 2.\\u0026nbsp;\\u003c/h4\\u003e\\n\\u003cp\\u003eFor Studies 1 and 2, our structural model (see Figure 1) included the effects of SRL, ELA, and DIGI on students\\u0026rsquo; chatbot engagement (metacognitive and non-metacognitive). We further modeled how SRL, ELA, and metacognitive chatbot engagement influenced academic performance as measured by exam scores.\\u003c/p\\u003e\\n\\u003cp\\u003eGiven the procedural differences between Studies 1 and 2, specifically the timing of chatbot engagement relative to self-report completion and contextual variations (semester-long interaction versus single-session engagement), we fitted a multi‑group BSEM that maximizes a single joint likelihood while estimating all parameters independently for each study.\\u003c/p\\u003e\\n\\u003ch4\\u003eStructural Model for Study 3.\\u0026nbsp;\\u003c/h4\\u003e\\n\\u003cp\\u003eThe structural model for Study 3 (see Figure 2) incorporated demographic predictors (sex, education level, and field of study) as influences on self-reported measures (SRL, ELA, GASE, MAST, PERF, AI_ATT). We then examined how these self-reported measures, together with demographic variables, predicted chatbot engagement.\\u003c/p\\u003e\\n\\u003cp\\u003eWe fitted a multi‑group BSEM that maximized a single joint likelihood and represented our counterbalanced design, where one group used the chatbot before completing the self‑reports and the other completed the self‑reports first. Paths from self‑reports to chatbot engagement, where order effects were expected, were estimated independently in each group. Paths from demographic variables to self‑reports and from demographics to engagement were not allowed to vary between groups, so a single coefficient was estimated from the combined data of both groups.\\u003c/p\\u003e\\n\\u003cp\\u003eEach demographic variable was entered as a binary indicator and the estimated coefficients indicated the difference compared to the reference category: Education Level (reference = undergraduate vs. graduate), Study Field (reference = STEM vs. social sciences), and Sex (reference = male vs. female). Although initially included, we omitted age as a predictor due to consistently poor model fit and high residuals that persisted despite various adjustments.\\u003c/p\\u003e\"},{\"header\":\"Results\",\"content\":\"\\u003cdiv id=\\\"Sec18\\\" class=\\\"Section2\\\"\\u003e\\n \\u003ch2\\u003eDescriptive Statistics\\u003c/h2\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eTable 3\\u003c/strong\\u003e\\u003c/p\\u003e\\n\\u003c/div\\u003e\\n\\u003cdiv id=\\\"Sec19\\\" class=\\\"Section2\\\"\\u003e\\n \\u003cp\\u003eDistribution of Self-Report Variables Across the Three Studies\\u003c/p\\u003e\\n \\u003cdiv id=\\\"Sec20\\\" class=\\\"Section3\\\"\\u003e\\n \\u003cp\\u003eMedian, first quartile (Q1), and third quartile (Q3) are reported for each variable\\u003c/p\\u003e\\n \\u003cdiv class=\\\"gridtable\\\"\\u003e\\n \\u003cdiv align=\\\"left\\\" class=\\\"colspec\\\"\\u003e\\u003cbr\\u003e\\u003c/div\\u003e\\n \\u003ctable id=\\\"Tabc\\\" border=\\\"1\\\"\\u003e\\n \\u003cthead\\u003e\\n \\u003ctr\\u003e\\n \\u003cth align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eVariable\\u003c/p\\u003e\\n \\u003c/th\\u003e\\n \\u003cth align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eGroup\\u003c/p\\u003e\\n \\u003c/th\\u003e\\n \\u003cth align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eMedian\\u003c/p\\u003e\\n \\u003c/th\\u003e\\n \\u003cth align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eQ1\\u003c/p\\u003e\\n \\u003c/th\\u003e\\n \\u003cth align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eQ3\\u003c/p\\u003e\\n \\u003c/th\\u003e\\n \\u003c/tr\\u003e\\n \\u003c/thead\\u003e\\n \\u003ctbody\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\" rowspan=\\\"4\\\"\\u003e\\n \\u003cp\\u003eSRL\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.20\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e3.74\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.66\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS2\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.71\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.12\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e5.14\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3C\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.87\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.13\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e5.53\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3Q\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.88\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e3.98\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e5.98\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\" rowspan=\\\"4\\\"\\u003e\\n \\u003cp\\u003eELA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e5.20\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.64\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e5.86\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS2\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e5.11\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.61\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e5.94\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3C\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e5.50\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.80\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e6.31\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3Q\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e5.60\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.53\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e6.39\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\" rowspan=\\\"4\\\"\\u003e\\n \\u003cp\\u003eGASE\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS2\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e3.90\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.50\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3C\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e3.90\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.60\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3Q\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e3.60\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.60\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\" rowspan=\\\"4\\\"\\u003e\\n \\u003cp\\u003eMAST\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS2\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e3.67\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3C\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e3.67\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.67\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3Q\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e3.67\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.67\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\" rowspan=\\\"4\\\"\\u003e\\n \\u003cp\\u003ePERF\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS2\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e3.33\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3C\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e3.67\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e3.17\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.17\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3Q\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e3.33\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e4.33\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\" rowspan=\\\"4\\\"\\u003e\\n \\u003cp\\u003eDIGI\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e1.82\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e1.55\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e2.18\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS2\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e2.27\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e1.95\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e2.45\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3C\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3Q\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\" rowspan=\\\"4\\\"\\u003e\\n \\u003cp\\u003eAI_ATT\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eNA\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS2\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e42.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e39.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e47.50\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3C\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e48.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e41.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e54.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3Q\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e46.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e43.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003e52.00\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003c/tbody\\u003e\\n \\u003c/table\\u003e\\n \\u003c/div\\u003e\\n \\u003cp\\u003eIn Study 1, 48 participants provided self-regulated learning (SRL) data, and 20 engaged with the chatbot. Whether participants engaged with the chatbot did not differ between high-SRL participants (SRL range [4.21, 5.91]), of whom 41.7% engaged, and low-SRL participants (SRL range [2.49, 4.19]), who had an identical engagement rate of 41.7%.\\u003c/p\\u003e\\n \\u003cp\\u003eParticipants\\u0026rsquo; chatbot engagement measures varied across the three studies (see Table\\u0026nbsp;4 for detailed engagement metrics). In Study 1, students exchanged a median of 27 messages (5\\u0026rsquo;001 characters) with the chatbot over a semester, whereas Study 2 featured a single-session exchange of 21 messages (4\\u0026rsquo;764 characters). Study 3, involving the Prolific sample, revealed very similar engagement levels irrespective of whether students chatted first (3C) or completed self-reports first (3Q). Both groups exchanged around 23 messages (10,125 and 10,144 characters). In Studies 1, 2, and 3, the chatbot consistently focused on metacognitive reflection, dedicating a median of 95.3%, 90.0%, and 100% of its characters to this topic, respectively. In Study 1, students contributed 1\\u0026rsquo;209 characters (~\\u0026thinsp;200 words), comprising 1\\u0026rsquo;138 (94.0%) characters on metacognition. Student contributions in Study 2 were shorter at 536 characters (~\\u0026thinsp;90 words), comprising 406 (75.7%) characters on metacognition. Although Study 3 conversations were over twice as long as those in Study 2, student contributions were only slightly longer, contributing 788 and 713 characters (~\\u0026thinsp;120\\u0026ndash;130 words), comprising 598 (75.9%) and 565 (79.2%) characters on metacognition.\\u003c/p\\u003e\\n \\u003cp\\u003eConcerning the initial representation of student study methods, the chatbot effectively collected information in 100% of cases in Study 2, while in Study 3, this information was successfully collected in 98.5% of cases. Participants reported median exam readiness scores of 5 in both Study 2 (IQR [3, 7]) and Study 3 (IQR [3, 6]). Participants in Studies 2 and 3 initially revealed limited use of effective study strategies (see Table \\u003cspan class=\\\"InternalRef\\\"\\u003e5\\u003c/span\\u003e). Only 16\\u0026ndash;17% of students reported using a structured study schedule, while distributed practice was reported by 16\\u0026ndash;22%, interleaved practice by 5\\u0026ndash;6%, and elaborative strategies by 9\\u0026ndash;11%. Practice testing usage differed notably between studies, with only 21% reporting this strategy in Study 2 compared to 48% in Study 3.\\u003c/p\\u003e\\n \\u003cp\\u003eAfter discussing the student\\u0026rsquo;s current study methods, the chatbot recommended high-utility learning strategies as instructed, prioritizing distributed practice (84\\u0026ndash;91%), followed closely by practice testing (79\\u0026ndash;87%), elaborative strategies (68\\u0026ndash;75%), and interleaved practice (68\\u0026ndash;70%). These recommendations occurred more frequently in Study 3. Additionally, Study 3 uniquely assisted students in creating a structured study schedule in 89% of cases, an intervention which was not specifically added to the chatbot instructions for Study 2. A large majority (96%) of participants in Study 3 found the chatbot useful because it recommended learning strategies, whereas 53% also found it useful because it helped them reflect, and 55% found it useful because it helped them feel prepared for the exam.\\u003c/p\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eTable\\u0026nbsp;4\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003c/div\\u003e\\n\\u003c/div\\u003e\\n\\u003cdiv id=\\\"Sec21\\\" class=\\\"Section2\\\"\\u003e\\n \\u003cp\\u003eChatbot Engagement Metrics Across the Three Studies\\u003c/p\\u003e\\n \\u003cp\\u003e\\u003cem\\u003eThis table presents the distribution of engagement indicators, including total messages exchanged, total characters exchanged, metacognitive characters contributed by both the mentor (chatbot) and the student, and non-metacognitive characters by each party. Values reported for each variable include the median, first quartile (Q1), and third quartile (Q3)\\u003c/em\\u003e\\u003c/p\\u003e\\n \\u003cdiv class=\\\"gridtable\\\"\\u003e\\n \\u003ctable id=\\\"Tab1\\\" border=\\\"1\\\"\\u003e\\n \\u003ccaption language=\\\"En\\\"\\u003e\\n \\u003cdiv class=\\\"CaptionContent\\\"\\u003e\\u003cbr\\u003e\\u003c/div\\u003e\\n \\u003c/caption\\u003e\\n \\u003cthead\\u003e\\n \\u003ctr\\u003e\\n \\u003cth align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eVariable\\u003c/p\\u003e\\n \\u003c/th\\u003e\\n \\u003cth align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eGroup\\u003c/p\\u003e\\n \\u003c/th\\u003e\\n \\u003cth align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eMedian\\u003c/p\\u003e\\n \\u003c/th\\u003e\\n \\u003cth align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eQ1\\u003c/p\\u003e\\n \\u003c/th\\u003e\\n \\u003cth align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eQ3\\u003c/p\\u003e\\n \\u003c/th\\u003e\\n \\u003c/tr\\u003e\\n \\u003c/thead\\u003e\\n \\u003ctbody\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\" rowspan=\\\"4\\\"\\u003e\\n \\u003cp\\u003eTotal Messages\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e27\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e19\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e48\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS2\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e21\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e18\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e26\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3C\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e23\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e19\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e31\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3Q\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e23\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e19\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e31\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\" rowspan=\\\"4\\\"\\u003e\\n \\u003cp\\u003eTotal Characters\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e5001\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e3144\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e9946\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS2\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e4764\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e3672\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e5885\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3C\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e10125\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e6730\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e15104\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3Q\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e10144\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e7168\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e14994\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\" rowspan=\\\"4\\\"\\u003e\\n \\u003cp\\u003eMetacognitive Mentor Characters\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e3802\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e2207\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e8438\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS2\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e3496\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e2560\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e4144\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3C\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e7848\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e5847\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e11522\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3Q\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e8287\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e6252\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e11187\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\" rowspan=\\\"4\\\"\\u003e\\n \\u003cp\\u003eNon-metacognitive Mentor Characters\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e186\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e409\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS2\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e390\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e228\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e628\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3C\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e304\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3Q\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e0\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e362\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\" rowspan=\\\"4\\\"\\u003e\\n \\u003cp\\u003eMetacognitive Student Characters\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e1138\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e584\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e2214\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS2\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e406\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e218\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e504\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3C\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e598\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e347\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e992\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3Q\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e565\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e408\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e1223\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\" rowspan=\\\"4\\\"\\u003e\\n \\u003cp\\u003eNon-metacognitive Student Characters\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS1\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e72\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e22\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e162\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS2\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e130\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e79\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e196\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3C\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e190\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e97\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e472\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003ctr\\u003e\\n \\u003ctd align=\\\"left\\\"\\u003e\\n \\u003cp\\u003eS3Q\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e148\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e80\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003ctd align=\\\"char\\\"\\u003e\\n \\u003cp\\u003e370\\u003c/p\\u003e\\n \\u003c/td\\u003e\\n \\u003c/tr\\u003e\\n \\u003c/tbody\\u003e\\n \\u003c/table\\u003e\\n \\u003c/div\\u003e\\n\\u003c/div\\u003e\\n\\u003cdiv id=\\\"Sec22\\\" class=\\\"Section2\\\"\\u003e\\n \\u003cp\\u003eTable\\u0026nbsp;5\\u003c/p\\u003e\\n \\u003cp\\u003e\\u003cstrong\\u003eInitial and Recommended Learning Strategies in Studies 2 and 3\\u003c/strong\\u003e\\u003c/p\\u003e\\n \\u003cp\\u003ePercentage of participants reporting each learning strategy at the beginning (initial) and chatbot-endorsed learning strategies (recommended) in Studies 2 and 3. Strategies included \\u003cem\\u003ecreating a study schedule\\u003c/em\\u003e, \\u003cem\\u003edistributed practice\\u003c/em\\u003e, \\u003cem\\u003einterleaved practice\\u003c/em\\u003e, \\u003cem\\u003epractice testing\\u003c/em\\u003e, and \\u003cem\\u003eelaboration\\u003c/em\\u003e\\u0026nbsp;\\u003c/p\\u003e\\n \\u003ch2\\u003e\\u003cimg src=\\\"data:image/png;base64,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\\\" width=\\\"508\\\" height=\\\"510\\\"\\u003e\\u003c/h2\\u003e\\n \\u003cdiv id=\\\"Sec23\\\" class=\\\"Section3\\\"\\u003e\\n \\u003ch2\\u003eBayesian Structural Equation Model Results\\u003c/h2\\u003e\\n \\u003cdiv id=\\\"Sec24\\\" class=\\\"Section4\\\"\\u003e\\n \\u003ch2\\u003eStudy 1 and 2\\u003c/h2\\u003e\\n \\u003cp\\u003eThe model, comprising 42 parameters, was applied to data from both Study 1 (n\\u0026thinsp;=\\u0026thinsp;20) and Study 2 (n\\u0026thinsp;=\\u0026thinsp;19). The model estimation process completed successfully and yielded a marginal log-likelihood of -308.312 and a posterior predictive p-value (PPP) of 0.442, demonstrating an acceptable model fit. While the full set of regression paths is detailed in Table \\u003cspan class=\\\"InternalRef\\\"\\u003eA3\\u003c/span\\u003e, the most statistically informative paths\\u0026mdash;those with 95% credible intervals (CI) not containing zero\\u0026mdash;are highlighted in Fig.\\u0026nbsp;1.\\u003c/p\\u003e\\n \\u003cp\\u003eAs the majority of the fitted model\\u0026rsquo;s regression paths had 95% CIs that included zero, indicating inconclusive evidence, they will not be analyzed in detail. However, two specific associations deserve more detailed attention. The first is the link between self-regulated learning (SRL) and metacognitive engagement, which yielded a statistically informative positive link in Study 2, despite being negative and inconclusive in Study 1. Secondly, we will assess the subsequent, theoretically important path from metacognitive engagement to exam performance, for which negative but inconclusive links were found in in both studies.\\u003c/p\\u003e\\n \\u003cp\\u003eIn Study 2, evidence for a positive association was observed between SRL and metacognitive engagement, with a mean posterior coefficient estimate of 0.542 (95% CI [0.01, 1.04]). With 97.7% of posterior samples supporting a positive coefficient and a CI that excludes zero, we can state with reasonable certainty that higher SRL was associated with increased metacognitive engagement. However, the width of the CI suggests considerable uncertainty regarding the magnitude of this effect; it could be small or even negligible.\\u003c/p\\u003e\\n \\u003cp\\u003eConversely, in Study 1, the relationship between SRL and metacognitive engagement tended to be negative, with a mean posterior estimate of -0.115 (95% CI [-0.686, 0.467]). However, with the CI being so wide and centered on zero and with only 65% of posterior samples supporting a negative coefficient, there was insufficient evidence to conclude a directional effect.\\u003c/p\\u003e\\n \\u003cp\\u003eRegarding the association between metacognitive engagement and exam performance, the findings also pointed to negative but inconclusive links across both studies. In Study 1, the mean posterior estimate was \\u0026minus;\\u0026thinsp;0.089 (95% CI [-0.485, 0.304]), and in Study 2, it was \\u0026minus;\\u0026thinsp;0.497 (95% CI [-1.048, 0.072]). Although a high percentage of posterior samples supported a negative relationship in Study 2 (95.8%), the inclusion of zero in the CIs for both studies precluded us from confidently supporting a negative link between metacognitive engagement and exam performance.\\u003c/p\\u003e\\n \\u003c/div\\u003e\\n \\u003c/div\\u003e\\n \\u003cdiv id=\\\"Sec26\\\" class=\\\"Section3\\\"\\u003e\\n \\u003ch2\\u003eStudy 3\\u003c/h2\\u003e\\n \\u003cp\\u003eOur structural equation model, comprising 126 parameters and 24 equality constraints, was applied to the two experimental groups (S3C: n\\u0026thinsp;=\\u0026thinsp;63; S3Q: n\\u0026thinsp;=\\u0026thinsp;59). The model estimation process completed successfully and yielded a marginal log-likelihood of -1416.133, and the posterior predictive p-value (PPP) of .504 indicated good model fit. As with the first model, the full set of regression paths for the second model is provided in Table \\u003cspan class=\\\"InternalRef\\\"\\u003eA4\\u003c/span\\u003e, with the most statistically informative paths visible in Fig.\\u0026nbsp;2.\\u003c/p\\u003e\\n \\u003cp\\u003eThe model identified several statistically informative paths. This included some evidence that female students reported lower academic self-efficacy (GASE) than males, alongside convincing evidence of a more negative attitude toward AI (AI_ATT), an effect of likely moderate to large size. However, here we focus on the association between SRL and chatbot engagement by examining the tendency for negative relationships with both metacognitive and non-metacognitive engagement. Two additional engagement-related effects, which appeared only in the chatbot-first group, are detailed at the end.\\u003c/p\\u003e\\n \\u003cp\\u003eRegarding the link between SRL and metacognitive engagement, the strength of evidence for a negative relationship differed between the groups. Evidence for a negative effect emerged in the chatbot-first group (mean posterior estimate = \\u0026minus;\\u0026thinsp;0.446, 95% CI [\\u0026ndash;0.842, \\u0026minus;\\u0026thinsp;0.053]), with 98.5% of posterior samples supporting this trend. However, the interval\\u0026rsquo;s width suggests uncertainty about the magnitude of the effect, which could be very small. In contrast, while the second group also pointed toward a negative trend (mean estimate = \\u0026minus;\\u0026thinsp;0.079), the evidence remained inconclusive, as its 95% credible interval [\\u0026ndash;.313, 0.156] overlapped with zero and was supported by only 75.1% of posterior samples.\\u003c/p\\u003e\\n \\u003cp\\u003eThe model also provided consistent evidence for a negative relationship between SRL and non-metacognitive engagement across both groups. This effect was robustly supported in the chatbot-first group (mean estimate = \\u0026minus;\\u0026thinsp;0.485, 95% CI [\\u0026ndash;0.836, \\u0026minus;\\u0026thinsp;0.146]) and to some extent in the chatbot-second group (mean estimate = \\u0026minus;\\u0026thinsp;0.317, 95% CI [\\u0026ndash;0.638, \\u0026minus;\\u0026thinsp;0.006]), with 99.7% and 97.7% of posterior samples supporting the negative trend, respectively. As both 95% CIs excluded zero, we can conclude that higher SRL was associated with lower non-metacognitive engagement. The effect size is likely not negligible in the chatbot-first group, where the interval\\u0026rsquo;s upper bound was further from zero.\\u003c/p\\u003e\\n \\u003cp\\u003eFinally, two additional paths emerged in the chatbot-first group: a negative association between performance orientation (PERF) and metacognitive engagement (mean estimate = \\u0026minus;\\u0026thinsp;0.362, 95% CI [\\u0026ndash;0.709, \\u0026minus;\\u0026thinsp;0.004]), and a positive association between academic self-efficacy (GASE) and non-metacognitive engagement (mean estimate\\u0026thinsp;=\\u0026thinsp;0.516, 95% CI [0.174, 0.843]), supported by 97.6% and 99.8% of posterior samples, respectively.\\u003c/p\\u003e\\n \\u003c/div\\u003e\\n\\u003c/div\\u003e\"},{\"header\":\"Discussion\",\"content\":\"\\u003cp\\u003eWe examined the potential of LLM-powered chatbots to support metacognitive reflection through three complementary studies in different educational settings. Drawing on the outcomes and insights from these studies, we explored (1) whether an LLM-powered chatbot can effectively co-construct metacognitive dialogue with students using only system-level instructions; (2) patterns of student engagement and whether they relate to individual characteristics, such as self-regulated learning ability, academic self-efficacy and attitudes towards AI; and (3) the extent to which chatbot-assisted metacognitive reflection translates into improved academic performance.\\u003c/p\\u003e\\u003cdiv id=\\\"Sec29\\\" class=\\\"Section2\\\"\\u003e\\u003ch2\\u003eFrom System Instructions to Metacognitive Dialogue\\u003c/h2\\u003e\\u003cp\\u003eOur findings indicate that the chatbot consistently maintained a strong focus on metacognitive dialogue (median 90\\u0026ndash;100% of characters), even when students were somewhat less consistent in doing so (median 75.7\\u0026ndash;94.0%). In Studies 2 and 3, the chatbot systematically (~\\u0026thinsp;100% of cases) queried students about their study preparation and learning strategies. Consistent with prior research (Dunlosky et al., \\u003cspan citationid=\\\"CR11\\\" class=\\\"CitationRef\\\"\\u003e2013\\u003c/span\\u003e; Morehead et al., \\u003cspan citationid=\\\"CR29\\\" class=\\\"CitationRef\\\"\\u003e2016\\u003c/span\\u003e; Pintrich, \\u003cspan citationid=\\\"CR30\\\" class=\\\"CitationRef\\\"\\u003e2002\\u003c/span\\u003e), students initially reported limited use of high-utility strategies such as distributed practice, interleaved practice, elaboration, and practice testing. The chatbot responded by offering evidence-based recommendations for learning techniques and co-constructing concrete study plans that incorporated these techniques. In Study 3, students rated the chatbot as particularly useful for suggesting effective learning strategies.\\u003c/p\\u003e\\u003c/div\\u003e\\n\\u003ch3\\u003eStudent Engagement: A Passive Role?\\u003c/h3\\u003e\\n\\u003cp\\u003eDespite the chatbot\\u0026rsquo;s effectiveness in structuring conversations around metacognitive reflection and evidence-based learning strategies, overall student engagement was low. This aligns with emerging evidence that students prefer using generative AI for direct learning support, such as requesting explanations, rather than for the more indirect process of metacognitive self-regulation (Spirgi and Seufert (\\u003cspan citationid=\\\"CR37\\\" class=\\\"CitationRef\\\"\\u003e2025\\u003c/span\\u003e)). Low interest in metacognitive reflection was evident in our data. Participation was low (21.4% uptake in Study 1), and interactions were brief, with students exchanging a median of just 27 messages over an entire semester. Across all studies, students\\u0026rsquo; contributions were consistently short (~\\u0026thinsp;90\\u0026ndash;200 words), suggesting a largely passive role.\\u003c/p\\u003e\\u003cp\\u003eInterestingly, our data also tentatively suggest that students used the chatbot in a way that matched their existing learning habits. In Study 3, for instance, we found a positive association between students who reported using elaboration in their learning and their non-metacognitive engagement with the chatbot, such as asking it to explain concepts. However, when forced to engage in a metacognitive dialogue rather than receiving direct learning support, students may be reluctant to adopt a more active role.\\u003c/p\\u003e\\u003cp\\u003eThe engagement challenges we observed are not unique and echo findings from other metacognitive chatbot systems (Martins et al., \\u003cspan citationid=\\\"CR26\\\" class=\\\"CitationRef\\\"\\u003e2024\\u003c/span\\u003e). They highlight a critical design imperative: chatbots should be designed to scaffold student reflection, not replace it, thereby aligning with the growing consensus on collaborative human-AI partnerships (Fan et al., \\u003cspan citationid=\\\"CR13\\\" class=\\\"CitationRef\\\"\\u003e2024\\u003c/span\\u003e). Future work should explore how to adapt chatbot behavior to elicit, rather than provide, extensive content, and shift the interaction from passive consumption to the active reflection that is essential for learning.\\u003c/p\\u003e\\u003cdiv id=\\\"Sec31\\\" class=\\\"Section2\\\"\\u003e\\u003ch2\\u003eThe Inconsistent Link with Self-Regulated Learning\\u003c/h2\\u003e\\u003cp\\u003eContrary to previous research suggesting that preexisting metacognitive skills are crucial for the success of interventions designed to stimulate them (Jansen et al. (\\u003cspan citationid=\\\"CR22\\\" class=\\\"CitationRef\\\"\\u003e2019\\u003c/span\\u003e)), we found that metacognitive engagement levels were not systematically positively associated with students\\u0026rsquo; self-regulated learning (SRL) or other individual difference variables such as digital competence or attitudes toward AI. This lack of a consistent, positive relationship aligns with other recent work which also did not find a positive relationship between students\\u0026rsquo; metacognitive skills and chatbot usage frequency in their Moodle-integrated chatbot for metacognitive reflection (S\\u0026aacute;iz-Manzanares et al. (\\u003cspan citationid=\\\"CR34\\\" class=\\\"CitationRef\\\"\\u003e2023\\u003c/span\\u003e)).\\u003c/p\\u003e\\u003cp\\u003eConcerning SRL, while Study 2 showed a positive association with metacognitive engagement, Studies 1 and 3 tended towards negative associations. This negative association was particularly evident when students completed self-reports \\u003cem\\u003eafter\\u003c/em\\u003e interacting with the chatbot, raising the possibility that the conversation made them more acutely aware of gaps in their SRL competencies. Given that SRL measurement relied on self-reports, which are sensitive to timing and may not capture actual behaviors, we need to be very cautious when interpreting these results. While Study 3 attempted to control for this, the results were not definitive.\\u003c/p\\u003e\\u003c/div\\u003e\\u003cdiv id=\\\"Sec32\\\" class=\\\"Section2\\\"\\u003e\\u003ch2\\u003eMore Talk, No Gains?\\u003c/h2\\u003e\\u003cp\\u003eContrary to expectations, we found no evidence that talking with the chatbot improved learning outcomes. In fact, we observed inconclusive negative trends between chatbot-assisted metacognitive engagement and exam scores in Studies 1 and 2. This is at odds with prior research showing that structured metacognitive interventions, including chatbot-assisted interventions (Graesser et al., \\u003cspan citationid=\\\"CR16\\\" class=\\\"CitationRef\\\"\\u003e2004\\u003c/span\\u003e; Yin et al., \\u003cspan citationid=\\\"CR41\\\" class=\\\"CitationRef\\\"\\u003e2024\\u003c/span\\u003e), can improve academic outcomes by fostering self-regulated learning (Amzil, \\u003cspan citationid=\\\"CR1\\\" class=\\\"CitationRef\\\"\\u003e2014\\u003c/span\\u003e; Cook et al., \\u003cspan citationid=\\\"CR7\\\" class=\\\"CitationRef\\\"\\u003e2013\\u003c/span\\u003e; Maftoon \\u0026amp; Alamdari, \\u003cspan citationid=\\\"CR24\\\" class=\\\"CitationRef\\\"\\u003e2020\\u003c/span\\u003e). However, we need to consider our sample size limitations and grading variability across courses.\\u003c/p\\u003e\\u003cp\\u003eAnother consideration is that quantity of engagement may be a limited metric for assessing how students benefit from chatbot-assisted metacognitive reflection. For instance, Hobert et al. (\\u003cspan citationid=\\\"CR20\\\" class=\\\"CitationRef\\\"\\u003e2023\\u003c/span\\u003e) found that increasing time spent conversing with a chatbot, through a more constructive and interactive dialogue, led to greater perceived learning but did not improve objective learning outcomes. Therefore, simply increasing message volume or time on task is not a guarantee for better outcomes. Moreover, S\\u0026aacute;iz-Manzanares et al. (\\u003cspan citationid=\\\"CR34\\\" class=\\\"CitationRef\\\"\\u003e2023\\u003c/span\\u003e) noted that students tended to ask primarily low-level questions focused on basic clarification of metacognitive concepts, rather than truly leveraging the potential for metacognitive reflection. These findings suggest that the quality of interaction, rather than merely the amount of time or message volume, determines the benefits of metacognitive reflection. Identifying what constitutes a high-quality dialogue and what the minimum effective dose of engagement might be remain key open questions for future research.\\u003c/p\\u003e\\u003cdiv id=\\\"Sec33\\\" class=\\\"Section3\\\"\\u003e\\u003ch2\\u003eFuture Directions: Metacognitive Checkpoints\\u003c/h2\\u003e\\u003cp\\u003eOur findings raise a central design question: can system-level prompts alone guide LLM-based chatbots to support metacognitive reflection effectively? While our results suggest that the chatbot was able to construct a dialogue around metacognitive reflection, student engagement remained limited. The inherent flexibility of LLMs, while advantageous for generating context-sensitive dialogue, may also lead to the chatbot interpreting its instructions too loosely, undermining the pedagogical goals.\\u003c/p\\u003e\\u003cp\\u003eFuture work should explore multi-agent or hybrid systems that combine the adaptability of LLMs with structured instructional scaffolds. For example, adding an additional layer of scripting, such as reflection checkpoints, could help structure interactions and ensure that students meet metacognitive goals. Such hybrid systems may offer the best of both worlds: fluid, natural conversations grounded in instructional design that enforces deeper student engagement.\\u003c/p\\u003e\\u003cp\\u003eMoreover, addressing barriers to initial and sustained engagement remains critical. As S\\u0026aacute;iz-Manzanares et al. (\\u003cspan citationid=\\\"CR34\\\" class=\\\"CitationRef\\\"\\u003e2023\\u003c/span\\u003e) pointed out, students may need explicit instruction on how to use chatbots effectively. Integration into course workflows, such as timely prompts or reflection windows, could also support continued use. In parallel, chatbot-supported metacognitive reflection could be embedded into evolving assessment practices that focus on learning as a process. For instance, AI-assisted e-portfolios could help students document, reflect on, and refine their learning strategies over time, making metacognitive reflection an ongoing, contextualized part of their academic journey.\\u003c/p\\u003e\\u003c/div\\u003e\\u003c/div\\u003e\"},{\"header\":\"Conclusion\",\"content\":\"\\u003cp\\u003e Our findings suggest that LLM-powered chatbots, when guided by well-crafted system prompts, can effectively structure dialogue around metacognitive reflection and deliver relevant content. However, promoting active student engagement remains a key challenge. Future research should move beyond measuring engagement by quantity alone and develop approaches to assess the quality and structure of learner interactions. Increasing meaningful engagement may require hybrid designs that combine the flexibility of LLMs with lightweight instructional scaffolds to ensure students meet metacognitive goals. Ultimately, the successful integration of AI-driven reflection tools will depend not only on technological sophistication but also on thoughtful design that emphasizes the learner\\u0026rsquo;s active role in the reflective process.\\u003c/p\\u003e\"},{\"header\":\"Declarations\",\"content\":\"\\u003cp\\u003e\\u003cstrong\\u003eCompeting Interests\\u003c/strong\\u003e: The authors have no conflicts of interest to disclose.\\u0026nbsp;\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eFunding\\u003c/strong\\u003e: This study was supported by a grant (Project #7: https://belearn.swiss/en/projekt/learning-companion/) from BeLEARN (https://belearn.swiss/).\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eAuthor Contributions:\\u003c/strong\\u003e Classified using the Contributor Role Taxonomy (CRediT; https://credit.niso.org/) as follows: \\u003cem\\u003eKim Uittenhove\\u003c/em\\u003e\\u003cstrong\\u003e:\\u0026nbsp;\\u003c/strong\\u003ewriting \\u0026ndash; original draft, data curation, formal analysis, project administration, visualization, and investigation. \\u003cem\\u003eAndrew Ellis\\u003c/em\\u003e\\u003cstrong\\u003e:\\u0026nbsp;\\u003c/strong\\u003econceptualization, methodology, resources, software, and writing \\u0026ndash; review \\u0026amp; editing. \\u003cem\\u003eFabian Mumenthaler\\u003c/em\\u003e\\u003cstrong\\u003e:\\u0026nbsp;\\u003c/strong\\u003econceptualization, methodology, project administration, investigation, and writing \\u0026ndash; review \\u0026amp; editing. \\u003cem\\u003eIoana Gatzka\\u003c/em\\u003e\\u003cstrong\\u003e:\\u0026nbsp;\\u003c/strong\\u003efunding acquisition, resources, and conceptualization. \\u003cem\\u003ePatrick Jermann\\u003c/em\\u003e\\u003cstrong\\u003e:\\u0026nbsp;\\u003c/strong\\u003econceptualization, funding acquisition, methodology, project administration, supervision, and writing \\u0026ndash; review \\u0026amp; editing\\u003c/p\\u003e\\n\\u003cp\\u003e\\u003cstrong\\u003eData Availability\\u003c/strong\\u003e: The datasets comprising self-reports, chatbot engagement metrics, and mention of learning strategies are available on zenodo: https://doi.org/10.5281/zenodo.15676346\\u003c/p\\u003e\\u003ch2\\u003eAcknowledgement\\u003c/h2\\u003e\\u003cp\\u003eThe authors express their gratitude to BeLEARN (https://belearn.swiss/) for the grant that supported this study. The authors are moreover grateful to Thierry Schluchter, Meike Lietz, and Christoph Till from the University of Teacher Education Bern, to J\\u0026ouml;rg Berkel and Markus Tiede from the Bern University of Applied Sciences. We are also grateful to Caroline Sahli Lozano from PHBern (Professor at the Institute for Research, Development, and Evaluation) and Michael Eckhart from PHBern (Professor at the Institute of Special Education) for their letters of support.\\u003c/p\\u003e\"},{\"header\":\"References\",\"content\":\"\\u003col\\u003e\\n\\u003cli\\u003eAmzil, A. (2014). The effect of a metacognitive intervention on college students\\u0026rsquo; reading performance and metacognitive skills. \\u003cem\\u003eJournal of Educational and Developmental Psychology\\u003c/em\\u003e, \\u003cem\\u003e4\\u003c/em\\u003e(1), 27\\u0026ndash;45. https://doi.org/10.5539/jedp.v4n1p27\\u003c/li\\u003e\\n\\u003cli\\u003eBastani, H., Bastani, O., Sungu, A., Ge, H., Kabakcı, \\u0026Ouml;., \\u0026amp; Mariman, R. (2024). \\u003cem\\u003eGenerative AI can harm learning\\u003c/em\\u003e. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4895486\\u003c/li\\u003e\\n\\u003cli\\u003eBjork, R. A., Dunlosky, J., \\u0026amp; Kornell, N. (2013). Self-regulated learning: Beliefs, techniques, and illusions. \\u003cem\\u003eAnnual Review of Psychology\\u003c/em\\u003e, \\u003cem\\u003e64\\u003c/em\\u003e(1), 417\\u0026ndash;444. https://doi.org/10.1146/annurev-psych-113011-143823\\u003c/li\\u003e\\n\\u003cli\\u003eBritos Cavagnaro, L. (2022). \\u003cem\\u003eChatbot guides students to learn and reflect\\u003c/em\\u003e. https://digitaleducation.stanford.edu/chatbot-guides-students-learn-and-reflect\\u003c/li\\u003e\\n\\u003cli\\u003eBroadbent, J., \\u0026amp; Poon, W. L. (2015). Self-regulated learning strategies and academic achievement in online higher education learning environments: A systematic review. \\u003cem\\u003eInternet and Higher Education\\u003c/em\\u003e, \\u003cem\\u003e27\\u003c/em\\u003e, 1\\u0026ndash;13. https://doi.org/10.1016/j.iheduc.2015.04.007\\u003c/li\\u003e\\n\\u003cli\\u003eCarbonell, J. R. (1970). AI in CAI: An artificial-intelligence approach to computer-assisted instruction. \\u003cem\\u003eIEEE Transactions on Man-Machine Systems\\u003c/em\\u003e, \\u003cem\\u003e11\\u003c/em\\u003e(4), 190\\u0026ndash;202.\\u003c/li\\u003e\\n\\u003cli\\u003eCook, E., Kennedy, E., \\u0026amp; McGuire, S. Y. (2013). Effect of teaching metacognitive learning strategies on performance in general chemistry courses. \\u003cem\\u003eJournal of Chemical Education\\u003c/em\\u003e, \\u003cem\\u003e90\\u003c/em\\u003e(8), 961\\u0026ndash;967. https://doi.org/10.1021/ed300686h\\u003c/li\\u003e\\n\\u003cli\\u003eCred\\u0026eacute;, M., \\u0026amp; Phillips, L. A. (2011). A meta-analytic review of the motivated strategies for learning questionnaire. \\u003cem\\u003eLearning and Individual Differences\\u003c/em\\u003e, \\u003cem\\u003e21\\u003c/em\\u003e(4), 337\\u0026ndash;346. https://doi.org/10.1016/j.lindif.2011.03.002\\u003c/li\\u003e\\n\\u003cli\\u003eDavis, F. D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. \\u003cem\\u003eMIS Quarterly\\u003c/em\\u003e, \\u003cem\\u003e13\\u003c/em\\u003e(3), 319\\u0026ndash;340. https://doi.org/10.2307/249008\\u003c/li\\u003e\\n\\u003cli\\u003eDonoghue, G. M., \\u0026amp; Hattie, J. A. C. (2021). A meta-analysis of ten learning techniques. \\u003cem\\u003eFrontiers in Education\\u003c/em\\u003e, \\u003cem\\u003e6\\u003c/em\\u003e, 581216. https://doi.org/10.3389/feduc.2021.581216\\u003c/li\\u003e\\n\\u003cli\\u003eDunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., \\u0026amp; Willingham, D. T. (2013). Improving students\\u0026rsquo; learning with effective learning techniques: Promising directions from cognitive and educational psychology. \\u003cem\\u003ePsychological Science in the Public Interest\\u003c/em\\u003e, \\u003cem\\u003e14\\u003c/em\\u003e(1), 4\\u0026ndash;58. https://doi.org/10.1177/1529100612453266\\u003c/li\\u003e\\n\\u003cli\\u003eElliot, A. J., \\u0026amp; Murayama, K. (2008). On the measurement of achievement goals: Critique, illustration, and application. \\u003cem\\u003eJournal of Educational Psychology\\u003c/em\\u003e, \\u003cem\\u003e100\\u003c/em\\u003e(3), 613\\u0026ndash;628.\\u003c/li\\u003e\\n\\u003cli\\u003eFan, Y., Tang, L., Le, H., Shen, K., Tan, S., Zhao, Y., \\u0026amp; Ga\\u0026scaron;ević, D. (2024). Beware of metacognitive laziness: Effects of generative artificial intelligence on learning motivation, processes, and performance. \\u003cem\\u003eBritish Journal of Educational Technology\\u003c/em\\u003e, \\u003cem\\u003eAdvance online publication\\u003c/em\\u003e. https://doi.org/10.1111/bjet.13544\\u003c/li\\u003e\\n\\u003cli\\u003eFlavell, J. H. (1979). Metacognition and cognitive monitoring: A new area of cognitive\\u0026ndash;developmental inquiry. \\u003cem\\u003eAmerican Psychologist\\u003c/em\\u003e, \\u003cem\\u003e34\\u003c/em\\u003e(10), 906\\u0026ndash;911.\\u003c/li\\u003e\\n\\u003cli\\u003eGhomi, M., \\u0026amp; Redecker, C. (2019). Digital competence of educators (DigCompEdu): Development and evaluation of a self-assessment instrument for teachers\\u0026rsquo; digital competence. \\u003cem\\u003eProceedings of the 11th International Conference on Computer Supported Education (CSEDU)\\u003c/em\\u003e, \\u003cem\\u003e1\\u003c/em\\u003e, 541\\u0026ndash;548.\\u003c/li\\u003e\\n\\u003cli\\u003eGraesser, A. C., Lu, S., Jackson, G. T., Mitchell, H. H., Ventura, M., Olney, A., \\u0026amp; Louwerse, M. M. (2004). AutoTutor: A tutor with dialogue in natural language. \\u003cem\\u003eBehavior Research Methods, Instruments, \\u0026amp; Computers\\u003c/em\\u003e, \\u003cem\\u003e36\\u003c/em\\u003e, 180\\u0026ndash;192. https://doi.org/10.3758/BF03195563\\u003c/li\\u003e\\n\\u003cli\\u003eGraesser, A. C., Wiemer-Hastings, K., Wiemer-Hastings, P., \\u0026amp; Kreuz, R. (1999). AutoTutor: A simulation of a human tutor. \\u003cem\\u003eCognitive Systems Research\\u003c/em\\u003e, \\u003cem\\u003e1\\u003c/em\\u003e(1), 35\\u0026ndash;51. https://doi.org/10.1016/S1389-0417(99)00005-4\\u003c/li\\u003e\\n\\u003cli\\u003eHardebolle, C., Tormey, R., Jermann, P., Pereira Correia Pinto, F., \\u0026amp; Di Vincenzo, M. C. (2019). Impact on the practice of students and teachers of a learning analytics dashboard. \\u003cem\\u003eSciences Po International Teaching and Learning Workshop 2019: \\u0026quot;Integrating Pedagogical Diversity: Blended Learning and Educational Impact\\u0026quot;\\u003c/em\\u003e.\\u003c/li\\u003e\\n\\u003cli\\u003eHidayat-ur-Rehman, I. (2024). Examining AI competence, chatbot use, and perceived autonomy as drivers of students\\u0026rsquo; engagement in informal digital learning. \\u003cem\\u003eJournal of Research in Innovative Teaching \\u0026amp; Learning\\u003c/em\\u003e, \\u003cem\\u003e17\\u003c/em\\u003e(2), 196\\u0026ndash;212. https://doi.org/10.1108/jrit-05-2024-0136\\u003c/li\\u003e\\n\\u003cli\\u003eHobert, S., F\\u0026oslash;lstad, A., \\u0026amp; Law, E. L.-C. (2023). Chatbots for active learning: A case of phishing email identification. \\u003cem\\u003eInternational Journal of Human-Computer Studies\\u003c/em\\u003e, \\u003cem\\u003e179\\u003c/em\\u003e, 103108. https://doi.org/10.1016/j.ijhcs.2023.103108\\u003c/li\\u003e\\n\\u003cli\\u003eKosmyna, N., Hauptmann, E., Yuan, Y. T., Situ, J., Liao, X. H., Beresnitzky, A. V., ... \\u0026amp; Maes, P. (2025). Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task. \\u003cem\\u003earXiv preprint arXiv:2506.08872\\u003c/em\\u003e.\\u003c/li\\u003e\\n\\u003cli\\u003eJansen, R. S., van Leeuwen, A., Janssen, J., Jak, S., \\u0026amp; Kester, L. (2019). Self-regulated learning partially mediates the effect of self-regulated learning interventions on achievement in higher education: A meta-analysis. \\u003cem\\u003eEducational Research Review\\u003c/em\\u003e, \\u003cem\\u003e28\\u003c/em\\u003e, 100292. https://doi.org/https://doi.org/10.1016/j.edurev.2019.100292\\u003c/li\\u003e\\n\\u003cli\\u003eLandis, J. R., \\u0026amp; Koch, G. G. (1977). The measurement of observer agreement for categorical data. \\u003cem\\u003eBiometrics\\u003c/em\\u003e, \\u003cem\\u003e33\\u003c/em\\u003e(1), 159\\u0026ndash;174.\\u003c/li\\u003e\\n\\u003cli\\u003eMaftoon, P., \\u0026amp; Alamdari, E. F. (2020). Exploring the effect of metacognitive strategy instruction on metacognitive awareness and listening performance through a process-based approach. \\u003cem\\u003eInternational Journal of Listening\\u003c/em\\u003e, \\u003cem\\u003e34\\u003c/em\\u003e(1), 1\\u0026ndash;20. https://doi.org/10.1080/10904018.2016.1250632\\u003c/li\\u003e\\n\\u003cli\\u003eMai, V., Neef, C., \\u0026amp; Richert, A. (2022). Developing an AI-based coaching chatbot: A study on disclosure as effectiveness factor in human-machine-coaching. \\u003cem\\u003eProceedings of the International Conference on Applied Human Factors and Ergonomics (AHFE) 2022\\u003c/em\\u003e, 651\\u0026ndash;657. https://doi.org/10.54941/ahfe100917\\u003c/li\\u003e\\n\\u003cli\\u003eMartins, L., Fern\\u0026aacute;ndez-Ferrer, M., \\u0026amp; Puertas, E. (2024). Analysing conversation pathways with a chatbot tutor to enhance self-regulation in higher education. \\u003cem\\u003eEducation Sciences\\u003c/em\\u003e, \\u003cem\\u003e14\\u003c/em\\u003e(6), 590. https://doi.org/10.3390/educsci14060590\\u003c/li\\u003e\\n\\u003cli\\u003eMcGuire, S. Y. (2021). Close the metacognitive equity gap: Teach all students how to learn. \\u003cem\\u003eJournal of College Academic Support Programs\\u003c/em\\u003e, \\u003cem\\u003e4\\u003c/em\\u003e(1), 69\\u0026ndash;72. https://doi.org/10.36896/4.1ep1\\u003c/li\\u003e\\n\\u003cli\\u003eMerkle, E. C., \\u0026amp; Rosseel, Y. (2018). \\u003cem\\u003eblavaan: Bayesian structural equation models via parameter expansion\\u003c/em\\u003e. \\u003cem\\u003eJournal of Statistical Software, 85\\u003c/em\\u003e(4), 1\\u0026ndash;30. https://doi.org/10.18637/jss.v085.i04\\u003c/li\\u003e\\n\\u003cli\\u003eMorehead, K., Rhodes, M. G., \\u0026amp; DeLozier, S. (2016). Instructor and student knowledge of study strategies. \\u003cem\\u003eMemory\\u003c/em\\u003e, \\u003cem\\u003e24\\u003c/em\\u003e(2), 257\\u0026ndash;271.\\u003c/li\\u003e\\n\\u003cli\\u003ePintrich, P. R. (2002). The role of metacognitive knowledge in learning, teaching, and assessing. \\u003cem\\u003eTheory Into Practice\\u003c/em\\u003e, \\u003cem\\u003e41\\u003c/em\\u003e(4), 219\\u0026ndash;225. https://doi.org/10.1207/s15430421tip4104_3\\u003c/li\\u003e\\n\\u003cli\\u003ePintrich, P. R., Smith, D. A. F., Garcia, T., \\u0026amp; McKeachie, W. J. (1991a). \\u003cem\\u003eA manual for the use of the motivated strategies for learning questionnaire (MSLQ)\\u003c/em\\u003e. National Center for Research to Improve Postsecondary Teaching; Learning.\\u003c/li\\u003e\\n\\u003cli\\u003ePintrich, P. R., Smith, D. A., Garcia, T., \\u0026amp; McKeachie, W. J. (1991b). \\u003cem\\u003eA manual for the use of the motivated strategies for learning questionnaire (MSLQ)\\u003c/em\\u003e. University of Michigan, National Center for Research to Improve Postsecondary Teaching; Learning.\\u003c/li\\u003e\\n\\u003cli\\u003eRichardson, M., Abraham, C., \\u0026amp; Bond, R. (2012). Psychological correlates of university students\\u0026rsquo; academic performance: A systematic review and meta-analysis. \\u003cem\\u003ePsychological Bulletin\\u003c/em\\u003e, \\u003cem\\u003e138\\u003c/em\\u003e, 353\\u0026ndash;387. https://doi.org/10.1037/a0026838\\u003c/li\\u003e\\n\\u003cli\\u003eS\\u0026aacute;iz-Manzanares, M. C., Marticorena-S\\u0026aacute;nchez, R., Mart\\u0026iacute;n-Ant\\u0026oacute;n, L. J., Gonz\\u0026aacute;lez D\\u0026iacute;ez, I., \\u0026amp; Almeida, L. (2023). Perceived satisfaction of university students with the use of chatbots as a tool for self-regulated learning. \\u003cem\\u003eHeliyon\\u003c/em\\u003e, \\u003cem\\u003e9\\u003c/em\\u003e(1), e12843. https://doi.org/10.1016/j.heliyon.2023.e12843\\u003c/li\\u003e\\n\\u003cli\\u003eSchei, O. M., M\\u0026oslash;gelvang, A., \\u0026amp; Ludvigsen, K. (2024). Perceptions and use of AI chatbots among students in higher education: A scoping review of empirical studies. \\u003cem\\u003eEducation Sciences\\u003c/em\\u003e, \\u003cem\\u003e14\\u003c/em\\u003e(8), 922. https://doi.org/10.3390/educsci14080922\\u003c/li\\u003e\\n\\u003cli\\u003eSchraw, G., \\u0026amp; Dennison, R. S. (1994). Assessing metacognitive awareness. \\u003cem\\u003eContemporary Educational Psychology\\u003c/em\\u003e, \\u003cem\\u003e19\\u003c/em\\u003e(4), 460\\u0026ndash;475. https://doi.org/10.1006/ceps.1994.1013\\u003c/li\\u003e\\n\\u003cli\\u003eSpirgi, L., \\u0026amp; Seufert, S. (2025). GenAI as a learning assistant, an empirical study in higher education. \\u003cem\\u003eProceedings of the 17th International Conference on Computer Supported Education (CSEDU 2025) - Volume 2\\u003c/em\\u003e, 27\\u0026ndash;34. https://doi.org/10.5220/0013199300003932\\u003c/li\\u003e\\n\\u003cli\\u003eStein, J.-P., Messingschlager, T., Gnambs, T., Hutmacher, F., \\u0026amp; Appel, M. (2024). Attitudes towards AI: Measurement and associations with personality. \\u003cem\\u003eScientific Reports\\u003c/em\\u003e, \\u003cem\\u003e14\\u003c/em\\u003e, 2909.\\u003c/li\\u003e\\n\\u003cli\\u003eTormey, R., \\u0026amp; Hardebolle, C. (2017). \\u003cem\\u003eApprendre \\u0026agrave; \\u0026eacute;tudier: Guide \\u0026agrave; l\\u0026rsquo;usage des \\u0026eacute;tudiants en sciences et en ing\\u0026eacute;nierie\\u003c/em\\u003e. PPUR. https://infoscience.epfl.ch/handle/20.500.14299/147097\\u003c/li\\u003e\\n\\u003cli\\u003eWeinstein, C. E., Palmer, D. R., \\u0026amp; Schulte, A. C. (2002). \\u003cem\\u003eLearning and study strategies inventory (LASSI): Second edition\\u003c/em\\u003e. H\\u0026amp;H Publishing Company.\\u003c/li\\u003e\\n\\u003cli\\u003eYin, J., Zhu, Y., Goh, T.-T., Wu, W., \\u0026amp; Hu, Y. (2024). Using educational chatbots with metacognitive feedback to improve science learning. \\u003cem\\u003eApplied Sciences\\u003c/em\\u003e, \\u003cem\\u003e14\\u003c/em\\u003e(20), 9345. https://doi.org/10.3390/app14209345\\u003c/li\\u003e\\n\\u003cli\\u003eZyl, L. E. van, Klibert, J., Shankland, R., See-To, E. W. K., \\u0026amp; Rothmann, S. (2022). The general academic self-efficacy scale: Psychometric properties, longitudinal invariance, and criterion validity. \\u003cem\\u003eJournal of Psychoeducational Assessment\\u003c/em\\u003e, \\u003cem\\u003e40\\u003c/em\\u003e(6), 777\\u0026ndash;789.\\u003c/li\\u003e\\n\\u003c/ol\\u003e\"}],\"fulltextSource\":\"\",\"fullText\":\"\",\"funders\":[],\"hasAdminPriorityOnWorkflow\":false,\"hasManuscriptDocX\":true,\"hasOptedInToPreprint\":true,\"hasPassedJournalQc\":\"\",\"hasAnyPriority\":true,\"hideJournal\":true,\"highlight\":\"\",\"institution\":\"\",\"isAcceptedByJournal\":false,\"isAuthorSuppliedPdf\":false,\"isDeskRejected\":\"\",\"isHiddenFromSearch\":false,\"isInQc\":false,\"isInWorkflow\":false,\"isPdf\":false,\"isPdfUpToDate\":true,\"isWithdrawnOrRetracted\":false,\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"identity\":\"researchsquare\",\"isNatureJournal\":false,\"hasQc\":true,\"allowDirectSubmit\":true,\"externalIdentity\":\"\",\"sideBox\":\"\",\"snPcode\":\"\",\"submissionUrl\":\"/submission\",\"title\":\"Research Square\",\"twitterHandle\":\"researchsquare\",\"acdcEnabled\":true,\"dfaEnabled\":false,\"editorialSystem\":\"\",\"reportingPortfolio\":\"\",\"inReviewEnabled\":false,\"inReviewRevisionsEnabled\":true},\"keywords\":\"chatbot, generative AI, metacognitive reflection, education\",\"lastPublishedDoi\":\"10.21203/rs.3.rs-6973046/v1\",\"lastPublishedDoiUrl\":\"https://doi.org/10.21203/rs.3.rs-6973046/v1\",\"license\":{\"name\":\"CC BY 4.0\",\"url\":\"https://creativecommons.org/licenses/by/4.0/\"},\"manuscriptAbstract\":\"\\u003cp\\u003eMetacognitive reflection is a crucial transversal skill, especially in an era where generative AI transforms how we teach and learn. As well as being a driver of the need to develop metacognitive reflection, generative AI is also a tool that can be used to enhance metacognitive reflection, such as chatbots that act as coaches to guide students in metacognitive reflective practice. In this study, we examined the potential of LLM-powered chatbots to promote metacognitive reflection across three distinct educational contexts. Our results show that the chatbot successfully constructed a metacognitive dialogue and delivered relevant, evidence-based recommendations. However, student engagement levels were generally low, with limited active participation observed across all studies. Notably, metacognitive self-regulation, and other individual differences, did not consistently predict engagement levels, suggesting that learners with higher reported self-regulation were not inherently more likely to use the tool. We also found no evidence that metacognitive engagement levels led to improved learning outcomes. However, these findings must be interpreted with caution, as engagement levels may be a limited metric for capturing how students benefit from chatbot-assisted reflection. We conclude by raising key design questions around how to develop chatbot systems that not only deliver metacognitive content and feedback but also encourage active student participation. While system prompts can help LLMs maintain focus on metacognitive reflection, hybrid designs that add an additional layer of scripting or multi-agent systems may be necessary to support an active learner role and ensure that important metacognitive checkpoints are met by the learner.\\u003c/p\\u003e\\u003cp\\u003e\\u003c/p\\u003e\",\"manuscriptTitle\":\"Metacognitive Reflection in the Era of Generative AI\",\"msid\":\"\",\"msnumber\":\"\",\"nonDraftVersions\":[{\"code\":1,\"date\":\"2025-09-30 07:19:54\",\"doi\":\"10.21203/rs.3.rs-6973046/v1\",\"editorialEvents\":[{\"type\":\"communityComments\",\"content\":0}],\"status\":\"published\",\"journal\":{\"display\":true,\"email\":\"info@researchsquare.com\",\"identity\":\"researchsquare\",\"isNatureJournal\":false,\"hasQc\":true,\"allowDirectSubmit\":true,\"externalIdentity\":\"\",\"sideBox\":\"\",\"snPcode\":\"\",\"submissionUrl\":\"/submission\",\"title\":\"Research Square\",\"twitterHandle\":\"researchsquare\",\"acdcEnabled\":true,\"dfaEnabled\":false,\"editorialSystem\":\"\",\"reportingPortfolio\":\"\",\"inReviewEnabled\":false,\"inReviewRevisionsEnabled\":true}}],\"origin\":\"\",\"ownerIdentity\":\"210823a2-8ee5-4ac1-af28-605547e21e22\",\"owner\":[],\"postedDate\":\"September 30th, 2025\",\"published\":true,\"recentEditorialEvents\":[],\"rejectedJournal\":[],\"revision\":\"\",\"amendment\":\"\",\"status\":\"posted\",\"subjectAreas\":[],\"tags\":[],\"updatedAt\":\"2025-10-15T23:23:25+00:00\",\"versionOfRecord\":[],\"versionCreatedAt\":\"2025-09-30 07:19:54\",\"video\":\"\",\"vorDoi\":\"\",\"vorDoiUrl\":\"\",\"workflowStages\":[]},\"version\":\"v1\",\"identity\":\"rs-6973046\",\"journalConfig\":\"researchsquare\"},\"__N_SSP\":true},\"page\":\"/article/[identity]/[[...version]]\",\"query\":{\"redirect\":\"/article/rs-6973046\",\"identity\":\"rs-6973046\",\"version\":[\"v1\"]},\"buildId\":\"8U1c8b4HqxoKbykW_rLl7\",\"isFallback\":false,\"isExperimentalCompile\":false,\"dynamicIds\":[84888],\"gssp\":true,\"scriptLoader\":[]}","source_license":"CC-BY-4.0","license_restricted":false}