The chosen database is Scopus because it has a wide range of journals that are peer-reviewed, a lot of bibliographic information, and is suitable for bibliometric analysis. ²³ These benefits appear to be limited to one database (Scopus), but it is very probable that some relevant studies are not in other databases, which may create a selection bias. The researchers had to overcome this limitation; an extensive search strategy was adopted, and all data extracted was carefully checked for relevance with the research topic and the validity of the data. Furthermore, only the publications included in the Scopus database have been foreseen with the results and interpreted. This has been acknowledged as a methodological constraint of this study that will be elaborated on in the conclusion.

These criteria (see Table 1) ensured that included studies positioned SRL as the primary theoretical framework rather than a supplementary construct and were clearly situated within STEM education, either interdisciplinarily or within specific disciplines. Studies addressing SRL outside STEM contexts were excluded to maintain analytical focus. Eligible articles were peer-reviewed journal publications in English that addressed at least one research question and were published between 2010 and 2026. These restrictions were applied to ensure academic quality, comparability of findings, and relevance to contemporary developments in SRL research within STEM education.

The following search query was set to identify publications relevant to the topic, from the Scopus database: TITLE-ABS-KEY ((“self-regulated learning” OR “self-regulated learning” OR “self-regulation of learning” OR SRL) AND (STEM OR “STEM education” OR science OR mathematics OR engineering OR technology)). This search strategy was used to look for studies that explicitly studied SRL in educational settings that were related to STEM. This search was carried out on 10 ^th January 2026. The search was carried out to see how SRL research in STEM education has progressed over the years; thus, the search was limited to publications from 2010 to 2026. Only peer-reviewed publications from journals were analyzed, in order to ensure that the data set is of high quality, and even in terms of content. Other papers (conference papers, review papers, books section, editorial, note, etc.) were not included. Furthermore, only studies where it was clearly stated that SRL was the primary construct were included. Only research that was done in the context of STEM education (science, technology, engineering, and mathematics) was included. The data set only included studies that addressed SRL and/or those that were conducted in STEM.

This selection and management of the articles was carried out collaboratively, as seen in Figure 1, assisted by the Rayyan software, according to the guidance of the PRISMA 2020. ²⁴ Altogether, 2273 records were initially found in Scopus. Following the titles and abstracts were screened, 445 of these articles were deemed relevant to the study and were retained with a good interrater reliability (κ = 0.693). After an in-depth eligibility check of the full text, 413 articles were assessed. 228 articles were included for in-depth analysis after final selection, and met all of the inclusion criteria. At this time, interrater reliability was good (κ = 0.832). The screening was done in collaboration to ensure consistency and methodological rigor in screening.

Data was analyzed on the final data set post-screening and eligibility. For a complementary analysis of the bibliography, two big software programs for bibliometric analysis were used: VOSViewer and Biblioshiny. For mapping the relationships and the intellectual structure of the research field, co-occurrence analysis of keywords, co-authorship, and co-citation analysis were performed in VOSviewer. Bibliometric analyses, descriptive and performance analyses were carried out using the R-based tool “Biblioshiny” to show an overview of the publications evolution, the most productive authors, countries, and sources, and thematic analysis to identify the research themes and their trend over time. These tools can help to map the entire landscape of research, such as performance indicators and conceptual structures of SRL research in STEM education. ²⁵

The synonymous keywords were standardized to ensure consistency across the dataset using a thesaurus file. Several related terms were merged, while terms unrelated to SRL, as well as overly general terms were removed. This process was conducted to minimize noise and generate a more meaningful and interpretable network structure. The thesaurus file and removed keywords are provided as supplementary materials.

Figure 2 shows that the annual scientific production of SRL research in STEM education from 2010 to 2026 exhibits a fluctuating but overall increasing trend. In the early period (2010–2014), publication output was relatively low and unstable, indicating that research on SRL within STEM education was still in its initial stage. This pattern continued during 2015–2018, where no consistent growth was observed. A more noticeable increase began in 2019 and 2020, marking growing interest in integrating SRL into STEM contexts. Despite a slight decline in 2021, the number of publications increased steadily from 2022 onwards, reaching a peak in 2025. The lower number of publications in 2026 should be interpreted with caution, as it represents an incomplete year at the time of data collection. These findings indicate that SRL research in STEM education has developed progressively over time, with a more pronounced growth in recent years, suggesting an increasing scholarly attention to the role of SRL in STEM learning.

Table 3 indicates that the largest number of articles fall in the category of Meta-cognition and Learning (11), followed by Education and Information Technologies (9) and Computers & Education (8). Other journals also have a number of articles in their collection, albeit not as large, such as Frontiers in Psychology (7 articles), Computers in Human Behavior (6 articles), and, surprisingly, Frontiers in Education (6 articles). Other articles are published by Interactive Learning Environments (5 articles) and other journals such as Educational Technology and Society, European Journal of Engineering Education, International Journal of Science Education, Journal of Science Education and Technology publish four articles each. As shown, the published papers in these journals are the SRL research of STEM education published in either psychology-oriented journals or educational technology-oriented journals. There’s a substantial quantity of journals out there that can be published in the field, with a few having fairly high publishing numbers and many moderate ones.

Table 4 presents the most productive authors based on both the number of published articles and their fractionalized contributions. Greene, J.A., ranks as the most productive author with nine publications and a fractionalized count of 1.79, indicating that his contributions are distributed across multiple collaborative studies. Zheng, J. follows with seven publications and a fractionalized score of 1.53, reflecting a similar pattern of participation in collaborative research. Bernacki, M.L., and Xie, C. each contributed six publications, with fractionalized counts of 1.34 and 1.20, respectively, suggesting that their contributions are also shared across collaborative works. Interestingly, Cleary, T.J., despite having five publications, shows a relatively high fractionalized count of 1.92, indicating a greater proportion of contribution within his publications. In contrast, authors such as Li, S., and Xing, W. have similar publication counts but lower fractionalized scores (1.03), suggesting that their contributions are distributed among larger collaborative teams. Santoso, H.B., with five publications and a fractionalized count of 1.45, demonstrates a moderate level of contribution across his works. Overall, the combination of these metrics indicates that SRL research in STEM education is largely collaborative, with varying degrees of contribution among leading authors.

As illustrated in Table 5, the University of North Carolina at Chapel Hill is the most productive institution, with a substantial contribution of 42 publications. This is followed by the University of Ljubljana with 28 publications and Universitas Negeri Malang with 19 publications. Other notable institutions include McGill University (16 publications), the University of Oulu (14 publications), and Beijing Normal University (12 publications). Additional contributions are observed from the Graduate School of Engineering (11 publications), National Taiwan Normal University (10 publications), and the Education University of Hong Kong (10 publications). The presence of highly productive institutions suggests that SRL research in STEM education is supported by several key academic centers that play a significant role in advancing the field. In particular, the dominance of the University of North Carolina at Chapel Hill indicates a strong research focus and sustained contribution in this area. At the same time, the distribution of other productive institutions across different regions, including North America, Europe, and Asia, reflects the global nature of SRL research in STEM education. This result indicates that while certain institutions serve as major contributors, the field is also shaped by a diverse range of universities worldwide.

Figure 3 shows that the number of publications coming from the United States is 190, which is significantly higher than that of other countries. This is followed by publications in China (112) and Indonesia (76), both developed and emerging countries, playing a part in developing the field. The contributions of Europe are also considerable, with, for instance, 49 publications in Germany, 30 in Slovenia, and 27 in the Netherlands, which points to the role of Europe in the development of SRL research. The field is also spreading around the world, evidenced by the publication of countries such as Canada (26 publications), Australia (24 publications), Turkey (24 publications), and Spain (21 publications). The results indicate that the research field of SRL in STEM education is distributed widely around the world, but also heavily skewed towards a few key nations. The strong dominance of the United States and the increase in Asian countries’ participation could be a new pattern of geographically diversified research.

The publication with the highest number of citations in the dataset received 271 citations, as shown in Table 6. The study investigated the relationship between emotions, self-regulated learning (SRL), and mathematics achievement, emphasizing the influence of affective factors on learners’ self-regulatory processes. ²⁶ The second most cited publication in the dataset is the study by Kistner et al., which received 243 citations and emphasized the promotion of SRL in classroom settings, particularly through instructional practices. ²⁷ This finding is consistent with a broader body of research highlighting the importance of instructional and technological support in fostering SRL. For example, previous studies have examined the integration of achievement goals with adaptive scaffolding and the effectiveness of online learning environments in supporting students’ self-regulatory processes. ^{28 , 29} These studies demonstrate the growing role of educational technology in SRL research. More recent publications have focused on advanced technologies, particularly artificial intelligence (AI), and although they have accumulated fewer total citations due to their recent publication dates, they exhibit remarkably high annual citation rates, indicating rapid scholarly attention and emerging research interest in AI-supported SRL within STEM education. ^{30 , 31}

Several journals are also cited in some of the most cited articles that include the major journals in this field, like Metacognition and Learning and Computers & Education. Overall, most of the cited works focus on the SRL process, SRL instructional method, and technology integration, which shows that SRL studies are multidimensional.

The annual impact of publications is indicated in Table 7. The results indicate that most of the articles that have been published in the past have high citation impact due to their long publication time. For instance, articles published between 2010 and 2015 demonstrate relatively high MeanTCperArt values, with peaks observed in 2010 (68.91), 2013 (67.62), and 2015 (61.00). This shows that the primary literature of the early phases of SRL research in the STEM field has been highly cited over the years. But the picture is different when looking at citation intensity per year (MeanTCperYear). This seems to indicate that the more recent publications have more citations received in shorter time periods, with the highest number of citations received per year occurring for the 2024 publications (6.00), followed by the 2021 publications (6.94) and the 2018 publications (6.32). This shows that there are studies that have a respectable impact and are noticed by academicians rapidly. In comparison, the citation values MeanTCperYear are relatively low for the latest years, specifically for 2025 and 2026, 0.92 and 0.75, respectively. This trend can be noticed as a result of the short length of citable years; namely, these publications have not been established long enough to be cited. Therefore, caution should be exercised in relying upon citations to assess the effect of more recent publications.

Furthermore, it is interesting that citations had varying impacts from year to year. For instance, for 2016, MeanTCperArt (19.25) and MeanTCperYear (1.75) are quite low, whereas the following years (2017 and 2018) are improving the citation performance. It suggests that the impact of the research is not constant, and this could be because of the shift in research focus or new themes within the SRL arena. The resulting list shows that the articles mentioned in the first years of the year have been mainly quoted from the articles that were published as pioneer articles, and the more recent articles mentioned in the list are newer and more intensely cited articles in each year. As illustrated in this trend, SRL research in STEM education is becoming increasingly dynamic, and more recent studies are becoming the focus of academic research.

The analysis of the main research areas of SRL in STEM education was conducted by applying the software VOSviewer. A co-occurrence analysis was used to analyze the relationships between the keywords and to find out the thematic structure of the field. The results of this research are displayed in Figure 4, which represents the interrelated research area of literature.

In the graphic, SRL is represented as a very close and dominant node, in the sense that it holds a central concept within the research areas that it is connected to. The structure of the network is around groups of thematic areas. The first cluster is related to the technology-enhanced learning environment, such as the keywords: online learning, educational technology, generative AI, ChatGPT, virtual reality, and cognitive load, all of which are instances of using SRL in technology and AI. The latter one is instructional design and learning contexts, including engineering education, STEM education, project-based learning, flipped classroom, and learning analytics, suggesting that SRL is broadly utilized in instructional design and classes related to it. The third cluster concerns learning outcomes and cognitive-motivational constructs like academic achievement, mathematics learning, self-efficacy, problem solving, and cognitive strategies, which are related to the relationship between SRL processes and academic performance. Moreover, other clusters that arose were learning motivation and gamification, AR, and science education, showing a trend towards other innovative approaches to engage learners.

Finally, the four thematic areas of SRL related to STEM education are interconnected as illustrated in the network. The SRL clusters are not independent, but interdependent, based on the concept of multidimensionality of SRL. In addition, these cognitive constructs (metacognition and cognitive strategies) and motivational constructs (self-efficacy and learning motivation), with another study of SRL, should not be directed towards the technology-based learning environment. The pattern of SRL studies in the field of STEM education has two aspects: theories and pedagogical applications, as well as new technology innovations, are all under the same roof.

The overlay visualization is represented as a color gradient showing the ‘average year’ keywords were published, as seen in Figure 5. Conceptual underpinning of SRL is primarily related to the previous studies on the SRL foundations, such as metacognition, cognitive strategies, and Academic performance. The developments are in line with the developments of SRL in the context of practical education, such as online learning, engineering education, STEM education, and blended learning. New themes are connected to topics that appeared as a result of the technology diffusion process, for example, generative AI, ChatGPT, virtual reality, gamification, flipped classroom, etc., and are aligned with an increase in innovative and digitally mediated learning environments. The overall evolution is from cognitive-based research to more applied and technologically oriented research.

Figure 6 shows the intensity of the research in the network. It is well recognized that the main focus is on SRL, and this is the case. Academic achievement, learning motivation, mathematics learning, and engineering education are also subjects with high density in the high-density regions, which are well-known and studied subjects. Other themes, such as moderation or metacognition, online learning, and educational technology, are present in moderate density, but still in development. The others (such as generative AI, ChatGPT, VR, and gamification) appear to be less prominent in terms of density and more recent and less mature in their development. The research of SRL begins with the basics of cognition and performance and continues to be a more technology-enhanced and innovation-based learning environment based on this distribution.

The recent incorporation of the topics related to technology in the overlay visualizations and the relatively low density shows there is ample opportunity for future research. There is a particular interest in the role played by the more advanced technologies in the management of the cognitive aspects (e.g., cognitive tests), motivational aspects, and learning outcomes of SRL processes. Besides, some of the themes were rated as moderately important, such as metacognition and the use of educational technology in the learning environment, suggesting further in-depth research into the theoretical dimensions of SRL and integrating these studies into learning environments that are technologically rich. Trends identified here suggest that further research is needed to investigate the effects of technology in complex and technologically-rich STEM learning environments on SRL.

The decrease in publications in 2021 is temporary and appears to be linked to the global disruption and the significant impact of the COVID-19 pandemic on educational systems and research activities. The sudden switch to the emergency remote learning resulted in less support in SRL opportunities as there were more unstructured contexts in which students were expected to learn and fewer opportunities for feedback, both of which are of paramount importance for effective SRL. ³² Furthermore, the swift shift to online learning presented numerous challenges for both students and teachers, such as limited access to learning materials, challenges in the regulation of learning, and the need for new learning strategies. ³³ This is a minor downswing, but not one so significant that it would impact the publishing trend, as SRL research in STEM education remained steady in terms of increases. Based on the pattern, SRL is gaining importance as a model to explain how the SRL process is activated by the learners in complex problem-solving and inquiry-based learning in STEM contexts.

It is noteworthy that the USA is more dominant than other countries in SRL research in STEM education, which has a relationship to the history of SRL as a field of research. The early theories and concepts of SRL were largely inspired by, ³⁴ drawing a theoretical foundation from the study of SRL in the United States. Zimmerman is one of the scholars who provided a theory for SRL study. With a long tradition of studies on educational psychology, the United States has become one of the important research hubs for research on SRL. ³⁵ But, in the past several years, there have been SRL studies that were relatively new and quick in the Chinese, etc., countries. The results of previous research also confirm that most of the major SRL research institutions are located in North America and Europe, with recent trends indicating other countries, such as China and Indonesia, are developing SRL research institutions. Some studies in China mark the beginning of SRL research in this country, ³⁶ while in Indonesia, the interest in SRL research has been enhanced due to a number of recent education changes. Specifically, components of deep learning approaches that are close to SRL are highlighted in the use of a deep learning approach in the curriculum of the Indonesian language. ³⁷

Involvement of research institutions like the University of North Carolina at Chapel Hill (UNC) is also an indication of the tendency of research in high-academic institutions in the United States, particularly in the academic institutions in North Carolina. The pattern of this distribution is consistent with the overall distribution of SRL publications, with the USA making up the largest portion, suggesting that the productivity of the institutions is also closely related to that of the countries, and that SRL research has a long history in the study of educational psychology in the USA. ³⁵ The high score for the strong contribution of UNC suggests that international research has been going on for some time to understand how SRL is an effective strategy to improve students’ motivation, metacognition, and academic performance. ^{38 , 39} There are more general reasons, and this is because of the awareness of its relevance and importance, which is a significant contributor to students’ success in learning, particularly in complex learning settings, as more research is undertaken about SRL. SRL is one of the tools that can be useful to manage one’s own motivation, effective learning strategies, and engage in metacognition activities required for successful learning in STEM education. Furthermore, in the online and self-directed learning setting, the increasing integration of technology in the domain of education has made SRL more relevant in these settings. ⁴⁰

The institutions run by Universitas Negeri Malang are not only found in North America and Europe, but also the most productive affiliation, reflecting the contribution of emerging research contexts, in this case, in Southeast Asia. The tendency can be correlated with the process of educational reform in Indonesia for the past few years as a part of the Merdeka Curriculum that adopted a student-centered and deep learning approach. The changes are consistent with SRL principles, and therefore, they invite further studies on the autonomy development process, meta-cognitive awareness, and meaningful learning processes in STEM education. It’s the same with other institutions and other universities being the most productive affiliations. As educational technology and innovation have made rapid progress, online and personalized learning environments have become an important factor that can greatly promote the rapid development of SRL research in China. ^{41 , 42} Furthermore, SRL is viewed as a competence that is needed in a knowledge-based and technology-based society. ⁴³ This can be observed in the pursuit of academic success in science, technology, engineering, and math (STEM) related fields, where SRL has been reported as one of the biggest factors that impact in-class learning outcomes.

From the distribution of the sources in SRL in STEM education, as shown in Table 2, the research in SRL in STEM education is quite diverse in terms of disciplines. The publications aren’t all within one domain, but spread out in journals of education, psychology, science education, engineering, and technology. This variation implies that SRL is a construct that is not domain-specific, but it is cross-domain and can be applied in various STEM disciplines. The interdisciplinary nature of this spread is strongly associated with the cooperative nature of research in this area. Table 3 shows that most productive writers contribute at a moderate level; most of the publications are produced with other people and not by themselves. This would be observed in a research field like SRL in STEM education, where the SRL pedagogy, cognitive psychology, domain knowledge, and the use of educational technology would be used. Considering the differences between the publishing media and the multi-authorship of the publication, one can imagine that SRL research in STEM education will be multidisciplinary and network-oriented. This underscores the importance of SRL, a cross-domain framework for capturing the role that the learners play in managing their learning processes in complex STEM contexts.

Table 5 demonstrates that three aspects (metacognitive processes, motivational factors, and instructional interventions) are related to the most common documents in the field of SRL studies in STEM education. Studies of emotions and feedback are increasing in foundation studies, especially in mathematics and science, as are studies of the classroom practices that affect students’ self-regulatory processes. Moreover, several articles that were often referred to mentioned technology integration (e.g., intelligent tutoring systems and online learning environments), which led to technology-enriched SRL studies. Newer papers, even those with artificial intelligence, have a very high number of yearly citations; new lines of high-impact research. ³⁰ This is further confirmed by the annual pattern of citations as illustrated in Table 6: older publications received more citations, either because they were cited over a longer period of time or because they received more citations over their life, or because they received more citations per year for a shorter period of time. The change has come in the move from basic theoretical research to more dynamic and ever-changing research fields. The results show that SRL research in STEM education is slowly bringing in higher-level technologies, but it is still related to metacognition and motivation.

Based on the conceptual structure obtained from the conceptual co-occurrence mapping of the keywords, the SRL studies in STEM education can be divided into several clusters, each of which represents one of the key conceptual areas of SRL in STEM education. SRL is in the middle of the network with the SRL construct highlighted as the integrative aspect of SRL. If it is a complex learning process that integrates the cognitive process with the motivations, pedagogical, and technological supports, then SRL is a unified process. The high relationship between cognitive-metacognitive processes and learning outcomes is indicated by one big cluster. Research literature in SRL explains the theoretical background of SRL. For instance, the keywords in the SRL research field, such as metacognition, cognitive strategies, academic achievement, and problem solving, prove that the theoretical basis of SRL is still valid; this means that SRL skill is a part of the ability to regulate cognition in achieving academic goals. The discovery confirms the results of other research, ^{44 , 45} which found that self-regulation plays an important role in students’ academic performance. Highly motivated students and those who have more self-regulatory skills perform better academically and are self-regulated learners. Besides, the motivational constructs represented, such as self-efficacy and learning motivation, indicate that SRL is considered a multi-dimensional process, which involves both cognitive and affective regulation. This is in line with previous studies, ⁴⁶ which showed that it is important to keep the learners engaged and involved in the learning tasks for effective SRL.

The other big group is in recent books on SRL integration in technologically-mediated learning environments. The words used (e.g., online learning, digital learning, educational technology, generative AI, ChatGPT, and virtual reality) signal the move from traditional learning environments to learning-rich and digitally mediated environments. In this case, technology can be used not only to help in learning processes but also to regulate, monitor, and evaluate the learning processes. Through the use of technology, personalization of the learning experience is possible by adapting content and learning pathways to the individual. ⁴⁷ Both adaptive e-learning environments and personal learning environments (PLEs) enable learners to plan, monitor, and reflect on their learning. Adaptive e-learning environment and personal learning environment (PLEs) are a set of concepts that will support learners in planning, monitoring, and reflecting on their learning. This indicates the process by which SRL has been introduced into the dynamic and adaptive learning process. More importantly, in the last few years, the which instrumental nature of technology has started to change, and now, the new trend of technology, which is AI, appears to be more transformative in SRL. It might be useful to have some key SRL skills taught, such as planning, monitoring, and reflection, ^{48 , 49} and scaffolding, suggestions, and feedback can be provided in real time through AI systems. This is a movement that shows that technology can be more than a tool that facilitates learning to become a facilitator in developing and maintaining SRL in complex STEM learning environments.

In addition, PBL, application of domain-specific contexts (engineering education and STEM education), and flipping the classroom show that the SRL is related to active and student-centered learning designs. The use of learning agents, problem solving, co-operators, and practical exercises in learning is one important factor in learning STEM that involves SRL. As students are expected to plan, monitor, and evaluate their learning independently, self-regulatory skills are crucial for students to be effective learners. Students with high SRL skills have been found to have a high level of motivation, better learning strategies, and greater academic success in the past. ⁴⁵ Furthermore, the independent learning, efficient, and flexible performance of complex tasks is more conducive to deep thinking through SRL strategies, and it is helpful to provide children with appropriate guidance in learning. ⁵⁰ The findings of this research prove the research hypothesis of this study, which is that ID can create learning structures, and ID can be used in preserving and facilitating the learning process of SRL in SCE in STEM.

The changes in the research themes of SRL in STEM education over time are visualized using overlay visualization. But its symptoms are that in all its forms, AI/Generative AI, ChatGPT, VR, and gamification are becoming more and more embedded, and that this is leading to more and more digitally mediated and innovative learning spaces. The importance of future research is demonstrated in this time sequence, as it needs to not only focus on the use of the new technologies, but also explore how the technologies can be used effectively to support important SRL processes. AI and immersive technologies have started to be explored in recent studies; however, the use of these technologies to assist in and inform key regulatory processes such as planning, monitoring, and reflecting remains under-investigated. In addition, few studies have focused on motivational aspects of SRL, while metacognitive and cognitive aspects of SRL, such as planning, monitoring, and evaluating the task, have been given more attention by most of the existing AI systems. ^{51 , 52} It is consistent with the previous studies that advocated the development of AI and immersive tools for students to help them during different stages of SRL, particularly in reflecting and revising their SRL. This is a line of research that needs to be continued, and new technologies, pedagogically informed tools that can be used to facilitate meaningful and sustainable SRL processes. ^{53 , 54}

This is also reflected in the density visualization, which shows that research remains a very central focus regarding some key constructs such as SRL, academic achievement, learning motivation, mathematics learning, and engineering education. The themes that would be investigated in the field would be the cognitive-motivational and the performance-related ones, out of which the latter has been a major theme investigated in SRL research for a long time. ⁵⁵ The salience of these dense regions indicates that they are well elaborated, theoretically informed, and are created and sometimes discussed in a number of STEM contexts. As opposed to these more mature topics, the emerging topics in the overlay visualization, especially those related to advanced technologies like generative AI, ChatGPT, virtual reality, and gamification, seem to be in lower-density areas. The topics are well known and have been in the spotlight in the recent past, but are still in relatively early stages of development, and not yet more solidly established in the general framework of the basic SRL research. These themes are somewhat peripheral and indicate a connection with more well-known themes, which have not been explored in-depth, such as meta-cognition and motivation. Last but not least, the results are consistent with the previous research that investigated the concept of outcome orientation in SRL research studies and with technologically fragmented research studies where a comprehensive SRL model that combines the new technologies was not presented.

The emerging patterns also showed a need for further research to relate well-developed, fundamental constructs with new technological innovations. It can be concluded that the future research directions should be: closing the gap between the two, and extending the research dealing with SRL by means of integrative approaches, which would integrate all aspects of SRL (cognitive, motivational, and technological) into one. Moreover, empirical and longitudinal research should be conducted in the future to further critique the developmental aspects of SRL in a technology-mediated learning environment, such as the AI-supported and immersive learning systems. All of these will facilitate the SRL research to create more comprehensive, adaptive, and context-sensitive learning models.

This future research direction is also highly relevant to physics education, where the learning process often involves complex problem solving, abstract conceptual understanding, and intensive cognitive regulation. New technologies such as AI-based learning systems, virtual reality, and immersive simulations have the potential to support students’ self-regulated learning in physics by facilitating planning, monitoring, reflection, and adaptive feedback during inquiry and problem-solving activities.