A neuroeducational approach to STEM education: Integrating real-time cognitive analysis and cognitive load theory to optimize teaching and learning at universities

Abstract

BACKGROUND:

The relevance of science, technology, engineering, and mathematics (STEM) education to effectively meet challenging academic requirements has become imperative for university entrants. With emerging advancements within the domain of learning neuroscience and analytics, it has become possible to analyze and track the real-time mental states of learners. This has opened several avenues to apply the theoretical perspectives of Cognitive Load Theory (CLT). Though applications of neuro-technological interventions for learning have become widely sought, their effective implementation within STEM education has not been comprehensive at the university level. The innovations brought about within this study include its conceptual fusion for STEM education based upon real-time mental state analysis and CLT. The goal of this research is to integrate the current research findings on real-time cognitive analysis practices used in STEM education in universities and analyze how cognitive load can be monitored dynamically and discuss teaching practices according to principles of CLT to improve effectiveness and inclusiveness for better cognitive load management throughout learning.

MATERIALS AND METHODS:

A systematic narrative review with cross-case synthesis was conducted using peer-reviewed empirical studies published between 2018 and 2025. The relevant articles were identified through the searching of databases such as Scopus, Web of Science, ERIC, and PsycINFO. Studies focusing on higher education STEM contexts were included if they employed neuroeducational tools (e.g., electroencephalography, functional near-infrared spectroscopy, eye tracking, or artificial intelligence–driven analytics) or incorporated instructional designs informed by Cognitive Load Theory (CLT). A standardized framework for extracting data was used, and data were thematically analyzed to identify patterns across the studies regarding cognitive load measurement, adaptive instructional strategies, learning outcomes, teacher experiences, and ethical considerations.

RESULTS:

From the synthesis, real-time cognitive feedback is perceived to have positively influenced teaching effectiveness by allowing the instructor to dynamically adjust pacing, scaffolding, and learning activities (d = 0.68–0.72, P < 0.01). Multimodal cognitive analytics were also associated with more precise estimates of cognitive load, supporting the development of adaptive learning strategies. It is also apparent that cognitively based personalization is effective in reducing the cognitive load and improving motivation and knowledge acquisition for both diverse and neurodiverse students.

CONCLUSION:

The synergistic combination of real-time cognitive analysis and CLT has the potential for making progressive inroads for inclusive and adaptive STEM education in universities. This conceptual combination is in good harmony with the benchmarks for health and educational policy that regard evidence-based instruction, the welfare of the learner, and equity in higher educational settings as cornerstones. Longitudinal study, ethical regulation, and implementation strategies for neuroeducational data in universities need more emphasis in future studies.

Introduction

The massive increase in Science, Technology, Engineering, and Mathematics (STEM) disciplines has further accentuated the need for innovative teaching practices that can support enhanced learning engagement and perseverance among university students.[1] Modern education environments are expected to provide enhanced problem-solving, analytical, and adaptive competencies to meet the changing social requirements. Nonetheless, conventional teaching practices remain rather passive and centered on transmitting knowledge, thereby increasing mental efforts to a large extent and impacting learning efficacy to a negative extent.[2] These knowledge-transmission-oriented teaching practices become all the more counterproductive for STEM disciplines due to their intense concept-driven nature, requiring multiple simultaneous processing of abstract representations, symbols, and problem-solving steps. Recent discoveries emerging out of modern-day neuroscience studies and cognition science have novel potential to reposition teaching practices to integrate real-time cognition management practices and cognition and mental load management practices and cognition regulation practices.[3] Blending STEM education practices and neuroscience education practices can emerge as a progressive initiative to develop responsive and inclusive environments for university-level STEM education and training practices focusing upon the individualistic learning needs of university students.[4]

Cognitive Load Theory (CLT) is a theoretical perspective on designing instructions that recognizes the limitations imposed by the architecture of the cognitive structure of humans.[5] Apparently, CLT states that learning instructions need to handle intrinsic, extraneous, and germane cognitive loads that will improve deeper levels of learning.[6] As a matter of fact, teaching STEM remains a challenge since ideas are more abstracted and may involve a rather complicated nature. Excessive cognitive loads may hinder knowledge generation or transfer within such disciplines.[7] Recently, contemporary empirical research on learning instructions remains focused on developing reduced extraneous cognitive loads that improve schema development through cognitive scaffolding.[8] Nevertheless, the application of CLT within the practical setting of university classroom learning instructions remains limited since a considerable number of instructions remain dependent on inferential or post-hoc cues indicative of student cognition.[9]

Presently, new neuro-educational technologies such as electroencephalography (EEG), functional near-infrared spectroscopy (fNIRS), and eye-tracking technology are now capable of providing a continuous measure of the learners’ cognitive state in a learning context.[10] These new technologies are capable of providing instantaneous learner feedback regarding their cognitive state, including their engagement, levels of cognitive overload, and onset of mental fatigue.[11] For example, it has been found that fNIRS technology could effectively and reliably measure the levels of cognitive overload exhibited by learners in engineering education, particularly in tackling difficult problem-solving tasks.[12,13] More specifically, other recent studies conducted by[14,15] showed that using EEG technology to monitor the levels of cognitive overload shown by learners was highly accurate as a valid academic outcome and holds high potential in being used in adaptive teaching approaches.

Real-time cognitive feedback implementation for Science, Technology, Engineering, Mathematics, and Medicine (STEMM) education at the university level has generated promising first outcomes.[16,17] Evidence confirms a substantial positive impact on student performance when educators adapt teaching based on real-time cognitive feedback, for instance, increased conceptual understanding, increased retention rates, and level of engagement.[18] Real-time feedback systems facilitate adaptive teaching processes, for example, slowing down the teaching pace, supplementing with increased scaffolding support, or using active learning methods in case of real-time signaled cognitive overload.[19] All these processes align well with a personalized approach based on a focus on responsiveness and evidence-based education.[20,21]

Also, responding to the needs of other learners and neurodiverse ones has recently gained the highest level of attention among researchers in STEM education.[22,23] Neurodiverse students, such as those with attention deficit hyperactivity syndrome (ADHD), autism spectrum disorder, and dyslexia, tend to have different patterns of cognitive processing that are not typically supported by standard instructional designs.[24,25] Cognitive load indicators and neurophysiological signals allow for adapting instruction to the needs of the individual learner in real time, to foster equity and inclusion.[26,27] Tailored interventions designed by real-time analysis have already been proven to favor the enhancement of academic persistence and self-efficacy among underrepresented groups in STEM.[26,28]

Despite these promising developments, the practical implementation of neuroeducational methods faces challenges.[29] There is a likelihood that university teachers may resist because it makes things more complex and raises a number of concerns relating to data privacy, ethical issues, and increasing the cognitive load on the teachers themselves.[30] In addition, unresolved ethical issues-informed consent within power-asymmetrical settings, data ownership and security, stigmatization risks, and algorithmic bias-mandate a cautious and transparent approach to integrating cognitive analytics into educational practice. It is necessary to understand the experiences of teachers, their attitudes, and perceived barriers to assist responsible adoptions of real-time cognitive technologies in STEM education.[31]

The current study makes an effort to apply real-time cognitive analysis and CLT to the optimization of university-level teaching and learning in STEM subjects in response to the needs and opportunities identified.[32,33] Rather than generating new data, the study contributes an integrative conceptual framework from a structured synthesis of existing research. This paper will seek to investigate how cognitive load could be measured on the fly during STEM learning activities, identify teaching strategies informed by CLT, assess reported effects of neuroeducational tools upon the learning of pupils, explore teachers’ experiences with adaptive instruction, and discuss the design of personalized interventions for both diverse and neurodiverse learners.[34] By consolidating existing evidence, this study will support the development of adaptive, inclusive, and ethically framed models of STEM education that are in tune with up-to-date priorities of educational and health policy.

The following Figure 1 presents the conceptual framework that relies on real-time data occurring from neuroeducation technologies such as EEG, fNIRS, eye-tracking on the cognitive load of students (intrinsic, extraneous, germane), which informs instructional decisions on adjusting teaching strategies according to CLT principles, with the purpose of improving learning outcomes.

Figure 1.

Theoretical framework integrating real-time cognitive analysis with cognitive load theory for STEM education

Significance of the study

The findings of this research are expected to play a significant role in STEM education offered by a university in relation to theory, practice, and social implications. First, in relation to theories, this research proposes to help close the gap between the theories of CLT and real-time cognitive monitoring technology. While CLT provides sufficient theories concerned with understanding mental workload in a learning context, it has still to-be-explored applications in real-time university learning contexts. By doing this efficiently and providing a good synthesis in this manner, this research will help redefine learning theories in this modern era with a rethink of learning models according to real-time cognitive levels of learners.

Second, in relation to its applications, this study provides very valuable insight for university teachers teaching STEM subjects. The design and implementation of teaching strategies that interactively utilize the cognitive demands of students could significantly enhance engagement levels, performance levels, and retention levels of students. In addition, through the design of personalized strategies along the lines of neurophysiological signals, it portrays a new era of personalized upcoming precision learning in which teaching strategies become evidence-backed and dynamic.[15] Third, in relation to its applications in a wider context, this study has a lot of application as it provides solutions to neurodiverse and diverged students in STEM subjects teaching environments who generally receive inadequate attention. By creating interventions that take into consideration neurophysiological signals, the study encourages more equality in teaching, and hence social justice in educational settings.

The above contribution to knowledge is conceptual, not empirical, and aims to clarify how existing knowledge might be systematically synthesized to better design teaching. Last, this research fills an important area of need for policy on higher education. Understanding teaching experiences related to disengagement by cognitive overload will help guide policy decisions on professional development, infrastructure support, and best ethical practice use of data. Thus, this research will help drive the needed changes in systems supporting STEM education at universities across the global community. In sum, this research aims to improve STEM teaching by combining principles of neuroscience, cognitive psychology, and educational systems on teaching to provide an integral, evidence-based teaching process. Notably, this research aims to add to higher education policy standards for healthy teaching by reflecting an important area of modern policy directed toward healthy teaching for inclusive not to mention mentally healthy teaching for an entire learner community suffering from widespread disengagement by cognitive overload issues across community colleges—filling an important research area on healthy teaching related to teaching overload by disengagement issues that falls under an important research area of this Journal of Education and Health Promotion.

Problem statement

Despite growing recognition about the importance of active and differentiated learning in STEM education, the methods used in higher education remain relatively unchanged, generic, and unresponsive to students’ cognitive requirements. One of the primary concerns consistently emphasized in the CLT is that cognitive management is a critical factor in achieving high academic potential, primarily in complex areas such as STEM education. However, in reality, instructors would never have the opportunity to gain immediate information about students’ mental conditions throughout learning activities. Without immediate information about students’ mental workload, one would face great difficulties in dynamically modifying strategies to avoid mental cognitive overload or underload. Despite the guidelines provided in the CLT, a significant number of past studies have been conducted immediately after the explanation or in a controlled lab situation. Consequently, little synthesized information is provided regarding the meaningful use of immediate cognitive information in real university STEM classroom instruction.

While recent breakthroughs in neuroeducational technologies like EEG or fNIRS are highly promising in terms of offering significant possibilities in monitoring and providing instant feedback for educators in learning settings, such technology is not widely adopted in education settings because it is typically developed and mainly used in medical contexts. Although the use of such technologies in higher education STEM contexts remains limited, previous research has demonstrated positive outcomes, including improved performance and engagement when these tools are integrated into learning environments. However, their effective implementation in typical university-level STEM education has not yet been fully explored.

Besides this, the diversity of the population poses yet more obstacles. Neurodiverse students, such as those with ASD, ADHD, or dyslexia, often call for differentiated instructional approaches, which are not given enough value in conventional STEM curricula. Real-time cognitive metrics could play a key role in shaping individual interventions that would support the diverse cognitive profiles of all students, thereby increasing inclusion and equity in STEM education. Despite this, the research into how to adapt cognitive load-informed strategies for these diverse and neurodiverse learners is very scarce. Evidence on how cognitive load-informed strategies may differentially impact diverse and neurodiverse learners is also rarely synthesized in current research; hence, there is uncertainty regarding their equitable and ethical application in the heterogeneous university classroom.

More recently, however, there has been increasing focus on preparing researchers for neuroeducational intervention by significantly increasing teachers’ preparedness levels to face potential challenges. The early results show that though, initial results reveal, the majority of the educators tested have been exposed to real cognitive feedback in real time, they have been apprehensive about data protection, technology used, and whether it is feasible to make real-time changes to an already content-heavy STEM subject course. It is crucial to address these concerns and make sure that some ethical questions that these treatments fail to address, such as those pertaining to informed consent, data security, the stigma of encouragement, and bias, are carefully considered in the context of an overall ethical examination.

Accordingly, the current study is related to the issue of the absence of adequate frameworks regarding the integration of real-time cognitive analysis and CLT into the educational processes of STEM fields at the university level. There is a pressing need for research regarding the application of real-time cognitive feedback, the development of cognitive load-optimized STEM education strategies, the study of the effectiveness of these strategies, the study of teacher experiences, and the implementation of specially tailored interventions for different students. Therefore, there is the need for an adequately structured and theory-driven integration of the current empirical evidence regarding the application of real-time cognitive analysis in university STEM education.

Research objectives

1. To examine how real-time cognitive analysis can be utilized to track and measure students’ cognitive load while participating in STEM learning activities at the university level.

2. To develop and implement STEM teaching strategies grounded in CLT that efficiently handle intrinsic, extraneous, and germane cognitive load in university classrooms.

3. To assess the effectiveness of incorporating neuroeducational tools and real-time cognitive feedback on students’ performance, engagement, and STEM knowledge retention.

4. To investigate teachers’ experiences and attitudes towards leveraging real-time cognitive information to dynamically adjust teaching approaches for STEM courses.

5. To create tailored learning interventions based on cognitive load metrics and neurophysiological signals to support diverse learners, including neurodiverse students, in STEM education.

Research hypothesis

Given that the present study is based on a systematic narrative synthesis of existing empirical research rather than primary data collection, the following hypotheses are framed as evidence-informed analytical propositions. These hypotheses reflect patterns and relationships consistently reported in the reviewed literature and serve to guide the structured synthesis and interpretation of findings, rather than to test new empirical data.

H₀1: Real-time cognitive load measurements do not correlate with students’ problem-solving efficiency during STEM activities.

H₀1: Greater intrinsic cognitive load as assessed through neurophysiological signals (e.g., EEG) predicts lower problem-solving efficiency in STEM tasks.

H₀2: Instructional designs using worked examples and schema activation do not enhance knowledge retention compared to conventional approaches.

H₀2: CLT-based teaching strategies minimize extraneous cognitive load and maximize knowledge retention in university-level STEM courses.

H₀3: Real-time cognitive feedback does not affect student engagement or long-term retention of STEM material.

H₀3: Neuroeducational technologies maximize germane cognitive load allocation, thereby enhancing engagement (e.g., measures of attention) and retention.

H₀4: Teachers leveraging real-time cognitive load analysis do not modify instructional pacing or scaffolding differently than teachers without real-time data.

H₀4: Real-time cognitive load analysis allows teachers to dynamically adjust instruction, thereby enhancing student performance in challenging STEM subjects.

H₀5: Cognitive load biomarker-based personalized interventions have no impact on STEM task performance in neurodiverse students.

H₀5: Interventions based on cognitive load biomarkers balance intrinsic/extraneous load, thereby considerably enhancing STEM performance in neurodiverse students.

These hypotheses are used as an organizing framework to map and interpret findings across studies, rather than as statistically tested hypotheses within a single experimental design.

Paper organization

Section 1 discusses the aims, challenges, and importance that were identified for conducting the research work, in which the neuroeducation approaches in STEM education will be explored. Section 2 provides a literature review pertaining to the proposed study, in addition to discussing the gaps that exist in the research conducted so far. Section 3 explains the methodology that will be followed for carrying out the proposed research work, and Section 4 will contain the findings that were derived for the conducted study. Section 5 will comprise the discussion for the study outcomes, followed by Section 6 that will contain the conclusion for the proposed study.

Literature review

The intersection of neuroscience, CLT, and STEM education has moved rapidly in the last two decades with the emergence of cognitive monitoring technologies. Recent researches within 2021–2025 have identified real-time cognitive assessment and the usage of neuroeducational instruments as effective for enhancing learning outcomes, particularly at the university levels of STEM education. This paper will critically analyze 10 such recent studies, focusing on their methodologies, findings, and relevance to the current research focus.

The existing study of[35] draws on cognitive neuroscience to explore the spatial cognitive processes of STEM students and their implications for STEM education. In this study, there was a spatial navigation paradigm whereby the spatial navigation test was administered to 172 undergraduate veterinary science students. Participants in the test were required to navigate a toy back to its starting position in an arena, relying on either internal self-motion cues, external landmark cues, or a combination of both. It was observed that all participants relied on integrated cues when deploying spatial cognition. However, those with higher cue integration resulted in better performances related to the studies in STEM. Precisely, the better performance in science and mathematics was highly related to internal self-motion cues, while higher performance in engineering and technology subjects was associated with landmark cues. These findings indicate the complex role of spatial ability in achieving performance in STEM and yield important implications for teaching and developing STEM curricula.

Regarding this issue, research by[36] states that solely focusing on post-outcome measurement, like finals, is not enough for STEM education, stressing the need for process measurement of these learnings. To remediate these problems, authors developed a STEM Learning Behaviour Analysis System (SLBAS) for STEM learning process analysis through deep learning methodologies to analyze learner behaviors for STEM education. These learner behaviors have been identified to follow the framework of Interactive, Constructive, Active, and Passive (ICAP). This analysis helped teachers understand their STEM process of learning more effectively. The analysis of findings shows that SLBAS has a very high degree of accuracy. Besides, it should be noted that the value of Cohen’s kappa for comparing expert code and SLBAS code was large, indicating that it can completely replace expert code for STEM observation and analysis of learner process effectively. Moreover, it has been shown through statistics that there is a strong correlation between process of learning and learning efficiency. This means the previous study’s result is interpreted in the manner of again proving the point that there is a difference in STEM learning as compared to other teaching processes because the former not only benefits the construction of knowledge about learners through experiential learning but is also more effective in learning knowledge passively as well.

The major study undertaken in this area[30] speaks about Artificial Intelligence (AI) and Neuroscience in Education, focusing mostly on the opportunities and challenges that revolve around AI and Neuroscience for improving educational performance. Neuroeducation combines concepts like neuroplasticity, cognitive theory, and formation of memories with AI technology to improve personalization, engagement, and cognitive optimization. Considering the PRISMA protocol, the researchers undertook a systematic analysis of 518 studies, selecting only 35 peer-reviewed journals in the last 10 years, dealing with different uses of AI in neuroeducation studies such as adaptive learning systems, neuro feedback tools, and related moral issues. The study uses AI tools like adaptive learning systems, neuro feedback tools, and game-based settings in this area and proves their significance in developing brain-based learning modalities. Apart from this, new tools such as virtual reality and augmented reality display high potential in facilitating learning experiences in this area. However, challenges like high costs, moral issues regarding privacy, and bias in AI technology are mentioned as major concerns in this area in this study.

According to the existing research study,[37] it is revealed that there are new possibilities that have opened up through combining educational neuroscience and educational technology, and that these possibilities have further opened up ways to improve understanding and teaching methodology efficacy. Based on research regarding the interaction between various educational technologies and the brain, more effective educational methods can be developed by educators that are tailored to meet students’ needs, and such methods have a huge potential for increasing academic success and providing various comprehensive learning opportunities. The research study looks into the effect that findings within educational neuroscience have on developing and utilizing teaching methodology in educational technology, and it looks into whether or not there is a difference between utilizing traditional learning theories and utilizing teaching methodology based on educational neuroscience findings.

A study conducted by[38] highlights the need for a current review of articles in educational neuroscience, noting that new trends in the field emphasize the need for such a review. A descriptive content analysis of articles in educational neuroscience was carried out in this study. Data were analyzed using content analysis of the Educational Neuroscience Articles Classification Form used in the study. The diversity of studies in the field and the number of articles being published in recent years suggest a solid and valuable field of research in educational neuroscience. In addition to this, a set of specific practice-oriented recommendations increases the integration of educational neuroscience studies in educational practice. As also recommended, future studies should not only continue to study the applications of educational neuroscience but also evaluate the effects of such applications. Finally, the interdisciplinarity of studies in educational neuroscience also plays a vastly important role in developing the field to its full capacity.

The current study[39] aims to provide a real-time algorithm to calculate the level of brain-to-brain synchronization that occurs in social interactions, such as those in collaborative and competitive tasks. The implications of said algorithm could particularly be of high value in educational institutions, particularly in teacher–student and student–student interactions, in that it could give them biomarkers as feedback metrics—which is lacking in the current state of teaching practice. This study, in the realms of neuroeducation and hyper scanning, uses the bispectrum approach, including multiprocessing operations in Python, to effectively analyze EEG signals and calculate levels of brain-to-brain synchronization of each pair of subjects participating in tasks that involve specific cognitive tasks. Data shows consistent results that bispectrum is high in collaborative tasks compared to competitive tasks, with 33.75% of the results validated to be positive in a statistical sense. It not only points to the improvements made, but also to further improvements to be achieved, like the integration of the algorithm in various operating environments, further validation, and improvements in visualizing results.

Research gaps

• It is noted by[36] that although the approach is very efficient in analyzing learning behavior, it does not use neuro-educational tools in dealing with real-time estimations of cognitive load, nor does it examine the advantages associated with relying on real-time estimations regarding cognitive analysis to further personalize and adapt STEM education in unique ways. This is a field where real-time technologies, including AI platforms, are now showing very promising outcomes.

• The study of[37] did not comprehensively integrate the cognitive analytics of neuroscience-based technologies into STEM education in top-tier educational institutions and still had issues in translating the discoveries of neuroscience into an adaptive teaching process that would respond to the cognitive needs of students in real-time environments.

Materials and Methods

Study design and setting

This research uses a qualitative multiple-case study design. This design strategy is used to gain a comprehensive understanding of the phenomenon of the integration of neuroeducational technologies in STEM education. Every case in this research is a published empirical research paper that is a separate implementation of cognitive analysis or CLT-related practices within a STEM university environment. A goal for this research is to use the findings gained through multiple cases within a comprehensive manner to build a strong conceptual framework. The environment within this research is defined by the published body of knowledge within the peer-reviewed literature that spans the years 2018 through 2025. The following Figure 2 presents the research design of the aforesaid study. Even though the aforesaid study makes use of several published empirical studies as units of inquiry, it is not considered to be a classic case study that focuses on the collection of new data. However, the most correct description of the aforesaid study could be that of a systematic narrative review, in which different studies are considered as cases of analysis from a conceptual perspective.

Figure 2.

Research design conceptual framework

Study participants and sampling

Cases were chosen using a methodical sampling approach to ensure information richness. Searching for possible cases entailed a search involving several academic citation indices including Scopus, Web of Science, ERIC, as well as PsycINFO. The criteria used to include a case were:

1. Published between 2018 and 2025.

2. Articles published in peer-reviewed journals.

3. Focused on higher education STEM settings.

4. Concerning real-time cognitive analysis, CLT, and neuroeducational technology.

5. Offers clear empirical data and/or methodology that is pertinent to one of the research objectives of the study.

Out of a total of 3457 articles identified for further consideration, a rigorous process of screening helped to finalize five key cases, which helped to offer immense insights into the problem of study. The five cases have been identified to match five research objectives identified for this study. This has been summarized in Table 1.

Table 1.

Summary of selected case studies

Case ID		Reference		STEM Context		Neuroeducational Tool(s)		Key Focus (Aligned with Objective)
Case 1		Alkabbany et al.[40]		University-level STEM lectures		Facial video analysis (webcam)		Objective 1: Tracking Cognitive Load
Case 2		Xue et al.[41]		Online STEM learning		EEG, Eye-tracking, Facial analysis		Objective 2: CLT-based Strategies
Case 3		Alkhassawneh and Al Sharif[37]		Technology-enhanced classrooms		EEG-based feedback, Cognitive monitoring		Objective 3: Assessing Neuroeducational Tools
Case 4		Alieto et al.[31]		University professional development		Learning analytics dashboards		Objective 4: Teachers’ Experiences
Case 5		Gomes[9]		Higher education STEM courses		AI-driven smart tutoring system		Objective 5: Tailored Interventions

The selection procedure was methodical and transparent, though it is not intended within this study that the methodological rigor of a formal systematic review, such as adherence to a PRISMA protocol or risk of bias assessment, is claimed.

Data collection tools and technique

Data for this multiple-case study were directly collected from published reports of the five identified cases. A standardized data extraction form was employed to systematically collect data on research objectives, methods, sample characteristics, neuroeducational technologies applied, major findings, limitations, and implications from each case.

The data analysis had a cross-case synthesis approach, which is done based on the work of Yin (2018). First, an individual-case analysis was done for each of the five studies to explicitly formulate its setting, approach, and findings. Second, a process of thematic analysis was conducted among all five cases to understand aspects of convergence and divergence in relation to research objectives by coding data collected for developing themes to respond to research questions. The synthesis process stayed true to the original data from the primary sources through the construction of a conceptual framework directly based on empirical evidence presented in the cases. This analysis was focused on recurrent relationships, instructional patterns, and conceptual convergences between the various studies, rather than an aggregation of quantitative outcomes or the generation of new statistical estimates.

Ethical consideration

Since this research did not involve direct contact with human participants but rather the examination of previously published work, it did not require formal ethical approval. Additionally, the discussion section of this article includes a specific examination of the general Ethical, Legal, And Social Implications (ELSI) of the technologies and methods presented in the cases regarding concerns like informed consent, data privacy, stigmatization, and algorithmic bias in educational contexts. The study explicitly discusses ethical, legal, and social issues of real-time cognitive monitoring technologies represented in the reviewed literature with regard to issues about informed consent, governance of data, stigmatization, and algorithmic bias in education.

Validity and reliability

Validity in the context of secondary data analysis can be ensured by the use of peer-reviewed empirical research and the implementation of clear inclusion criteria, while the use of a standardized data extraction approach and cross-case synthesis can increase research reliability. Validity in the present research study, as it did not involve the use of primary research measurement tools, could not be ascertained by the research study, but it ensured research rigor through the implementation of research triangulation at various research data sources.

Results

The findings discussed in this section are based on a conceptual integration of results available from the chosen empirical researches and not on original data created particularly for this research. Each subsection condenses evidence available from the researches according to their respective research goals and hypothesis tests, with statistical results and effect sizes pointing back to their sources.

Case study analysis

Objective 1. Real-Time Cognitive Analysis for Tracking Cognitive Load in STEM

The prevailing study by[40] has given a detailed analysis of a system that can allow real-time measurement of student engagement levels for STEM classes within universities. An elaborate experimental setup has been designed for this purpose, whereby a hardware component specific to each class has been included, consisting of a Raspberry Pi microcontroller, a webcam, and a touch screen to allow uniform quality output of processes across different classes. The system has been set to record facial videos of students enrolled in a class when lectures are conducted, which have been processed instantaneously for analysis of facial expressions concerning head positions, eye gaze occurrences, and facial action units, and transmitted into a server computer. The processing of classifying student engagement levels into different categories concerning behaviors and emotions has been carried out by employing SVM algorithms for classifying levels of student engagement.

Key Findings

• Allowed real-time identification of the level of student engagement through the use of multimodal facial video features.

• For behavioral engagement tasks such as note-taking and raising one’s hand, the agreement ratios between the automated system’s classification results and the experts’ annotations varied between 83 and 88%.

• The agreement ratios for emotional engagement (e.g., confused, interested), ranged from 73 to 90%.

• Engagement patterns were seen in this research to tend to decrease during or at the end of lectures, as is typical student behavior.

• The modular architecture of this framework would facilitate future extensions regarding physiological sensor integration (e.g., heart rate variability and galvanic skin response sensors).

• Cases like this one make it clear that real-time cognitive tracking is possible and that it can offer useful information to professors on how to change their teaching practices.

These findings generally support the analytical hypothesis that real-time, non-invasive cognitive proxies can be employed effectively as valid surrogates for detection of learner engagement and cognitive loads within university STEM environments.

Objective 2. Developing STEM Teaching Strategies Grounded in CLT

The critical issue of cognitive overload in an online learning environment was the subject of the existing research by[41] that proposed and validated a multimodal framework for the assessment of cognitive load. The objective was to construct a model that can accurately quantify cognitive load, a necessary foundation for adaptive, CLT-informed teaching strategies. For this, the neural network model was built on a base where three different modalities were integrated into one: EEG capturing brain activity, eye tracking monitoring the visual attention of the learner, and facial analysis assessing the emotional responses of the learners.

Key Findings

• The accuracy of the neural network model in determining the cognitive load was very high as it resulted in an overall assessment accuracy of 91.52%.

• A high degree of accuracy ensures that a Learning Model (LM) approach gives a more accurate estimate or measure of cognitive load in web-based settings than any individual source.

• Findings in this case study are extremely important for satisfying Objective 2 as they offer a verified approach to measuring cognitive load, which is a necessary antecedent for creating and applying teaching practices on a foundation of CLT.

• They propose the application of the multimodal analysis findings as a reference source for learning resource development and the optimization of online courses.

• The implications are highly relevant in practice since it can now be incorporated into online intelligent learning platforms to develop “prescription-adaptive learning environments,” which make online learning dynamically adaptable to a learner’s real-time cognitive load.

• It realizes the present shortcomings of this research, pointing out that it has only focused on several thematic learning models and suggesting that, in the future, much more attention should be paid to model fine-tuning, exploring more complex learning scenarios, and scaling up research to enhance the model’s generalization capabilities.

Throughout these studies, the level of accuracy reported for the assessment of cognitive load based on multimodal cues and the resultant instructional inferences suggest that multimodal cognitive load assessment is a necessary underpinning for the translation of CLT into adaptive instruction rather than a means for testing the utility of instructional design on its own merits.

Objective 3. Assessing Neuroeducational Tools and Real-Time Feedback

The study by[37] presented the fundamental principles of educational neuroscience and their applications in strategizing teaching practices and applying them with the use of educational technology. It is apparent that the main objective of this research article is to discuss and evaluate how knowledge obtained from studies of the brain can be used to design more effective personalized tools of EdTech, against the background of the traditional theories of learning.

Key Findings

• The article states that if designed on the principles of neuroeducation, technologies such as adaptive learning systems, AI-powered intelligent tutoring systems, spaced repetition software, and gamified environments have the potential to increase student engagement and knowledge retention.

• The crucial evidence used in the review was based on an experiment where students’ MRI signals recorded higher activity in regions of the brain associated with working memory, attention, and problem-solving abilities after being tutored through modeling tools and techniques in physics, correlating directly to enhanced performance in a physical reasoning activity.

• What is offered here is how a neuroscience-inspired approach might empower individuals to break away from the classical, or passive, method of learning. This would include concepts from behaviorism or cognitivism.

• Nonetheless, this paper critically analyses the major setbacks and limitations experienced in this process of integration. These include:

• A lack of understanding of neuroscience concepts, or the language used in them, by educators.

• The debate on whether results obtained in a laboratory setup can be directly applied in a classroom setting.

• Concerns about ethics constructed in connection with the privacy of data and intervention based on neuroscience information.

• A significant disparity that exists between the theoretical potential of such tools and their adoption by educators.

• Ultimately, the above-mentioned issue highlights the fact that, although vast benefits can be accrued through the use of neuro-education technology, the development and implementation of the technology require considerable attention to education and ethics.

The results listed below are based on convergent evidence found in the literature regarding the educational potential of neuroeducation-based technologies, rather than results that are part of the present study findings.

Objective 4. Investigating Teachers’ Experiences with Real-Time Cognitive Information

Although this study does not introduce new qualitative results related to faculty views, the results of consolidation give indirect but instructive results related to factors influencing the willingness of teachers to use teaching aids that involve a great deal of cognition. The prevailing study[42] was a pivotal examination of experiences of faculty in professional development (PD) workshops, which has a direct connection to considering how teachers might respond to new teaching aids such as real-time cognition analysis. The basic aim of this examination was to construct a tool of observation—the Real-Time Professional Development Observation Tool (R-PDOT)—to examine systematically the workshop structure and its interactions in professional development activities. The basic assumption of this examination portrays a workshop structure as critical to whether teachers adopt and continue to use new teaching practices.

Key Findings

• The research discovered that the workshops on PD are “authoritative” and “prescriptive” and tend to focus on lectures from the workshop leader. This does not provide an effective setting where the faculty members can work on the new concepts introduced to them.

• Teaching and Learning in Higher Education Using the R-PDOT to examine three workshop sessions, the authors were able to show that those designed with more “dialogic-interactive” features such as small group work and analyses of simulated teaching had more opportunities to facilitate faculty members’ reflective thinking and self-efficacy development than traditional lecture settings did.

• This portends a critical challenge for the adoption of new technologies. If teachers are to adopt complex tools such as real-time cognitive analytics, professional development will have to be highly interactive and non-evaluative. Faculty must be given the time to play with the tools, consider implications with colleagues, and relate them to their specific classroom contexts rather than being told how to use the tools.

• The study points out that there may be resistance among faculty members if they feel that the training that comes with new technology is a top-down, one-size-fits-all, non-acknowledgment of their knowledge and limitations, and thus there is support needed in professional development that promotes a vision and reflection that leads to change.

• Basically, this particular case is very important for Objective 4 and shows that the success with which real-time cognitive feedback is accomplished in these sectors is less dependent on technology and more on the quality of development provided for its implementation. The implementation will either empower teachers as passive information-gaining agents or as participating agents for adjusting this technology according to their needs.

Objective 5. Creating Tailored Interventions for Diverse and Neurodiverse Learners

The existing study[9] designed an AI-enabled intelligent tutoring platform for higher education STEM courses. The platform incorporated Natural Language Processing (NLP) and machine learning algorithms to understand the student interactions, learning patterns, and predicted cognitive load levels. Accordingly, the intelligent tutoring platform automated learning support mechanisms, such as adapting quiz options, adjusting the level of difficulty, and offering targeted feedback. A pilot study was carried out for university students, in which two groups were selected—those using the AI-driven platform and the others using the regular online learning platform. The study measured the learning achievements, engagement, and cognitive load of the students.

Key Findings

• An increase in knowledge retention, motivation, and satisfaction was observed in students in the personalized learning environment developed by AI as compared to students in the control group.

• A lower level of cognitive load and enhanced perception of ease of learning and engagement were observed among students who used the intelligent assistant.

• The system’s ability to dynamically adapt instruction to each learner’s unique needs was a key factor driving these gain.

• The research validated the effectiveness of integrating cognitive analytics and AI to develop adaptive, student-centered STEM education models that improve learning outcomes and engagement.

The previous results synthesized validate the hypothesis introduced analytically about cognitive load personalization and improved learning experiences in different learners and stress caution when delivering results because of natural variability among researches and ethics issues.

In summary, it can be seen that there are collective findings on the relationships between real-time cognitive analytics, design by dint of CLT, and learning outcomes in adaptive learning, in as much as such findings are determined by synthesis levels as opposed to empirical testing for individual research projects.

These conclusions have been able to establish a link between these synthesized findings and the analytical hypotheses, which have only been hypothesized on the solitary premise of observations conducted within the literature reviewed, without any piece of information being brought into the discussion.

Discussion

This discussion connects the synthesized results to the analytical hypotheses formulated for this study exclusively on the basis of patterns identified in the literature reviewed, without using any new empirical information. Thus, it addresses conceptual integration, implications for theory, as well as general aspects relating to university-level STEM education. These results translate immediately to Hypotheses H₁2, H₁3, and H₁4, which assert a relationship between CLT-based educational methods and cognitive feedback processes regarding increased engagement, flexibility in instruction, and learning efficacy, as consistently reported in the literature.

The present study will comprehensively investigate the role of the integration of real-time cognitive analysis and the CLT in STEM education as it relates to the teaching and learning processes in universities. The application of the latest neuro-educational technology will allow this study to describe the role of the continuous monitoring and measurement of cognitive load in STEM tasks in the formulation of targeted learning strategies aimed at ensuring maximum learning through the reduction of germane cognitive load while increasing the germane cognitive load within students. The aim of this study is also to examine the strength of neuroeducative interventions accruing in terms of improved performance and retention in STEM subjects by students. The study also focuses on how these technology systems affect teachers in terms of whether they can change teaching practices in reaction to cognitive evidence accrued by research studies. The aim of its pedagogy of evidence is to influence teachers to take decisions in a manner to directly affect teaching practices in a classroom setup for improvement of efficacy.

In addition to this, the study is aimed at inclusivity via the development and evaluation of learning interventions designed to adapt to and respond to real-time cognitive loads and neurophysiological signals. This is intended to serve diverse students in the classroom. This is for ensuring varied practices present within the education sector adapt to all. The personalized learning made possible through these technologies helps create a special learning profile for a student that requires such learning. This will trigger equity and motivation within the student populace. Within the integrated approach used here for this study, it is also intended to advance the evolving nature of STEM learning through effective practices. This is intended not only to improve learning efficiency within STEM learning institutions but also serves as a precursor for innovative advancement within the fields of educational neuroscience, personalized learning, and cognitive measurement. The results will also brief the development within learning policies for adaptive learning within STEM.

Ethical, legal, and social implications (ELSI)

The synthesis emphasizes that the educational potential of cognitive real-time monitoring in STEM subjects cannot be considered separately from numerous ethical, legal, and social concerns. Even in the case when, as in the above study, there was no involvement of human subjects, numerous concerns emerge upon the examination of literature concerning, for instance, informed consent in situations when there are power imbalances between educators and students, who might feel compelled to waive their right to exclude data collection in order not to suffer any negative influence upon educational achievement.

The concern of stigmatization is another very significant issue that has been taken into consideration. The incoherent labeling of the learners as “disengaged” or “cognitively overloaded” could affect the teachers’ perceptions and the learners’ identification, leading to the deficit-based discourses being reinforced in the long run. Further, the application of cognitive analytics through making use of artificial intelligence might lead to the effects of bias in case the training samples are unrepresentative of disparity-based cognitive representations, cultures, or the neurodiverse group.

Overall, the discussion emphasizes that the value of this study lies in its integrative synthesis and ethical contextualization of existing evidence, rather than in the generation of new empirical claims.

Limitations, strengths, and weaknesses

However, there are several limitations of the study. Although the case study design yields rich information, it is not generalizable from the perspectives of statistics. It is context-specific for the five chosen cases and cannot be generalized for all STEM areas and settings. Each case study has varying levels of rigor and quality of reporting in terms of the studies from which this data is derived. Additionally, it is limited by the perspective of cognitive load only and not the broader arena of influences on the learning process like the state of an individual’s emotion and prior knowledge of concepts being studied.

Despite these shortcomings, the study has many strengths. The multiple case study research design permits the detailed analysis of a complex issue in the real world. The purposeful selection of the cases guaranteed thorough research of the aims of the study. The cross-analysis develops a strong conceptual framework that is immediately rooted in the evidence. The ELSI analysis illustrates the careful and visionary attitude of the research toward the issue. Therefore, the findings must be understood as informative of conceptual and practical trends, but not as causal evidence.

Suggestions for future research

Based on this analysis, several avenues for future research follow:

1. Different study designs need to be considered, for instance, longitudinal studies, that will explore the impact of these interventions for knowledge retention and transfer and their student success in a STEM career.

2. Research studies in implementation science need to develop models related to teacher professional development in using these tools effectively in their teaching practices.

3. A great need emerges here for the improvement of ethics frameworks as well as data governance policies in general regarding neuro-educational data in higher education.

4. There is a greater need for research on designing and evaluating neuroeducational interventions for and with neurodiverse learners.

5. At this point, what is required is research into cost-effective and scaled-up methods that could assist these technologies in being transferred from laboratories into common usage.

Conclusion

The current multiple-case study has contributed significantly to conceptual research about optimizing STEM education at universities through the integration of real-time cognitive analysis and CLT. By intertwining findings from five different cases, this research has provided evidence-based recommendations for practitioners and suggestions for future research.

The evidence shows the effectiveness of neuro-educational technology and its ability to empower educators to design and deliver learning environments, which can be more dynamic, flexible, and effective. With the technology it is possible for the educator to obtain real-time feedback about the cognitive condition of the student and go from static and standard learning to precision-based learning, which is suitable for the student. This is very helpful in the area of STEM education.

However, the process of implementation is not without its hurdles and includes teacher training and development, data ethics management, and the use of technology. The relevance of ELSI becomes evident from the fact that it exemplifies the need for simultaneous development of technologies and guidelines related to issues of equity and protection of students.

At the end of this research, it is not a “ground-breaking” new empirical result but rather a conceptual synthesis that takes places on solid ground and allows clarification of the state of the art and identification of the best practices and the crucial steps to be taken to fully develop the neuroeducation potential in the frame of higher education. Its findings enable the development of a more adaptive, inclusive, and effective learning environment of STEM, aligning with health and educational policy goals in general, oriented toward evidence-based, equity-principled, and supportive learning for all students.

Funding

This research received no external funding.

AI policy statement

The author confirms that no AI tools were used in the writing of this manuscript, production of images or graphical elements, or in the collection and analysis of data. All content is the original work of the author.

Abbreviations

STEM: Science, Technology, Engineering, and Mathematics; CLT: Cognitive Load Theory; EEG: Electroencephalography; fNIRS: Functional Near-Infrared Spectroscopy; AI: Artificial Intelligence.

Ethical statement for human participant

Not applicable for this research.

Data availability

Data sharing not applicable to this article as no datasets were generated or analyzed for this systematic narrative review.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil'