Tgs2: Computational Assistant for Learning

Breno Gramacho Seixas Santos; Bárbara Solana Scarlassara

DOI: 10.22167/2675-6528-202602727

Article derived from a Course Conclusion Work (TCC), with content based on the student’s original work and adapted to the editorial format of the E&S Magazine with the support of the ResumeAI tool, an artificial intelligence solution developed by Instituto Pecege for textual synthesis and organization.

Summary

In school environments, challenges such as attentional dispersion, low engagement, and absence of individualized feedback have driven the search for proposals based on educational Neuroscience. The work described and presented TGS2, a computational learning assistant, developed to provide a more interactive and personalized study experience, focusing on adaptive learning. The objective was to create an educational process guided by the principles of meaningful repetition, emotional engagement, and immediate feedback, using neuroscience fundamentals. For the development of the artifact, the Design Science Research methodology was adopted, which encompassed the stages of problem identification, objective definition, design, development, demonstration, and analytical evaluation. The assistant integrated functionalities such as level progression, automatic difficulty adjustment, multimodal feedback, performance tracking, gamification, and visual monitoring dashboards. The results indicated that TGS2 was implemented functionally and coherently with the proposed objectives, incorporating, in the computational plan, strategies associated with repetition, positive reinforcement, self-regulation, and performance monitoring. The system translated theoretical principles into concrete mechanisms of pedagogical interaction, demonstrating automatic adaptation, immediate feedback, and data persistence. It was concluded that the computational assistant constitutes a promising educational proposal for organizing personalized study paths, offering a basis for future empirical investigations in real teaching contexts. TGS2 represents a relevant contribution to the dialogue between neuroscience, educational technology, and pedagogy, organizing learning experiences sensitive to the student’s pace, performance, and motivation, although empirical evaluation with students in a real learning situation is still necessary.

Keywords: Adaptive learning; Computational assistant; Multimodal feedback; Educational neuroscience.

1. Introduction

Neuroscience applied to education has proven fundamental for understanding the brain mechanisms involved in learning, from the acquisition to the consolidation and retrieval of information. This field of study allows for the development of more effective, personalized pedagogical strategies that are sensitive to the emotional dimensions of the learning process.

However, the growing expansion of digital teaching platforms, although it has broadened access to educational content and resources, has also highlighted persistent challenges. Among these, the students’ attentional dispersion, low engagement with the proposed activities, and the absence of individualized feedback stand out, elements that can compromise the effectiveness of the teaching-learning process.

From a neurobiological point of view, learning is a dynamic process, intrinsically linked to neural plasticity and the formation of new synaptic connections. Kandel (2000) argues that the consolidation of learning is strongly influenced by repetition and emotion, factors that promote lasting changes in the nervous system. Additionally, Damásio (2011) and Goleman (1995) emphasize the inseparability between emotion and cognition, indicating that the affective dimension is an essential component for attentional, decision-making, and mnemonic processes involved in learning.

Traditional pedagogical practices, often based on collective and uniform instruction models, frequently fail to consider individual differences in students’ learning pace, cognitive styles, and memory forms. In contrast, neuroscience evidence indicates that knowledge consolidation is optimized through meaningful repetition, emotional engagement, and immediate feedback. This gap between pedagogical approaches and neuroscientific knowledge justifies the search for solutions that integrate these principles to enhance the educational experience.

Given this scenario, the present work describes and presents the creation of a computational learning assistant (TGS2) that uses neuroscience principles to offer a more adaptive, interactive, and personalized educational process, guided by the principles of meaningful repetition, emotional engagement, and immediate feedback.

2. Material and Methods

The development of the TGS2 computational learning assistant, the central artifact of this study, adopted the Design Science Research Methodology (DSRM), as outlined by Peffers et al. (2007). This systematic approach guided the creation and evaluation of the artifact, ensuring the resolution of practical problems and the contribution to knowledge in the field of educational neuroscience.

The research began with the identification of the problem and motivation, delimiting educational and technological challenges such as attentional dispersion, low engagement, and absence of individualized feedback. Subsequently, the solution’s objectives were defined: to apply cognitive neuroscience fundamentals in structuring learning paths, to create a digital environment that stimulated self-confidence and protagonism, to implement automatic adaptation algorithms, and to develop tools for rapid review and positive reinforcement.

The design and development stage consisted of system modeling and implementation, covering the navigation structure, functional modules, and the adaptation algorithm. The artifact demonstration occurred through usage scenarios and simulated cases, evidencing the path adjustment and the functioning of feedback based on hypothetical performance profiles.

The evaluation was artificial and analytical in nature, conducted through functional tests, criteria verification, and analysis of theoretical adherence to design decisions. This stage validated the internal functioning of the system and the consistency of the algorithm’s responses to controlled inputs, without application with students. The final communication phase documented the development process, design justifications, and system architecture.

The planning and implementation phases were complementary. Phase one, Planning and Architecture, involved defining the interface and the progression logic of the modules. A database was created in Google Sheets for inserting and organizing questions and configuring learning paths, integrating it with automated forms in an HTML interface. Preliminary usability and navigation consistency tests were applied.

The use of Google Sheets as a database aimed to facilitate the system configuration by teachers. The integration between Google Sheets, Apps Script, and HTML code ensured consistency and agility in operations. The system was structured to receive questions from an external base, distribute them by levels, control access authentication, and record the individual history and user performance.

Phase two, Development and Testing, consisted of the programming of interaction and automatic feedback modules, and the implementation of performance dashboards. Error evaluation, consistency verification, and functional system tests were carried out. The assistant was designed to support students, teachers, and educational institutions.

On the initial interface, the system implemented a login screen with user and password identification, allowing individualized access and retrieval of previously saved data. The TGS2 organized questions into levels, controlling the work level and altering progression according to user performance. After each answer, the system offered immediate feedback, indicating correctness or error, displaying the correction, and updating the score. The feedback deepening area could contain text, audio, video, image, or external link. Gamification was incorporated with levels, progression events, challenges, and achievements. Visual performance dashboards, with a radar chart and statistical summary per level, presented metrics of correctness, consolidation, strength, speed, and strategy. The system considered response time, predicted daily usage limits, and ensured data persistence by saving information in batches via Google Script.

3. Results and Discussion

The development of the computational learning assistant TGS2 rigorously followed the steps of the Design Science Research Methodology (DSRM), as proposed by Peffers et al. (2007), which guided the construction of the artifact from problem identification to its analytical evaluation. The results obtained demonstrate the system’s functionality and its adherence to the principles of educational neuroscience, aiming to enhance the study experience through meaningful repetition, emotional engagement, and immediate feedback. The implementation of TGS2 translated these theoretical foundations into concrete computational mechanisms, offering an adaptive and personalized solution for contemporary educational challenges.

Steps – DSRM

The initial phase of the project focused on problem identification and motivation, outlining the limitations of digital learning environments, such as attentional dispersion, low engagement, and the absence of individualized feedback. Subsequently, the solution’s objectives were defined, which included applying cognitive neuroscience fundamentals to the structuring of learning paths, creating a digital environment that stimulated self-confidence and learner protagonism, implementing automatic difficulty adaptation algorithms, and developing rapid review and positive reinforcement tools based on motivational stimuli, aligning the proposal with the identified educational needs.

Subsequently, the design and development stage involved the modeling and implementation of the system, encompassing the navigation structure, functional modules, and the adaptation algorithm. This phase ensured coherence between the theoretical framework and the artifact’s functionalities, guaranteeing that TGS2 incorporated neuroscientific principles from its conception. The artifact’s demonstration was carried out through usage scenarios and simulated cases, which highlighted the adjustment of the learning path, revision recommendations, and the functioning of the feedback, using hypothetical performance profiles to illustrate the system’s adaptability.

The TGS2 evaluation was conducted artificially and analytically, through functional tests, verification of predefined criteria, and analysis of theoretical adherence to design decisions. This approach included the validation of the system’s internal functioning and the consistency of the algorithm’s responses to controlled inputs, without direct application with students. Finally, the communication stage involved detailed documentation of the development process, design justifications, system architecture, study limitations, and its contributions, clarifying the possibilities for future empirical validations in real educational contexts.

Planning and implementation phases

The planning and implementation phases were conceived in a complementary manner, with the purpose of ensuring alignment between the design and evaluation of the TGS2 assistant. Phase 1, named Planning and Architecture, encompassed the definition of the interface and the progression logic of the learning modules. It included the creation of the database using Google Sheets and its integration with automated forms in an HTML interface, in addition to the application of preliminary tests to verify usability and navigation consistency, ensuring a robust and intuitive structure for the system.

The use of Google Sheets as a database proved to be an accessible solution, especially for users with less programming experience. This feature allows professors from various fields to configure the TGS2 system according to their specific pedagogical practices, reducing the gap between teaching work and computational technology. The ease of inserting questions, organizing content, and configuring learning paths without requiring advanced programming knowledge is a differential, promoting greater autonomy and personalization in the educational environment.

In addition to facilitating the creation and customization of study paths, TGS2’s performance records and dashboards offer valuable insights for teachers. This data can assist in identifying recurring difficulties, both at the individual and collective level, allowing for pedagogical interventions more targeted to the needs observed during the learning process. This functionality reinforces the assistant’s role as a support tool for the instructor, providing subsidies to improve teaching quality and the effectiveness of the pedagogical strategies employed.

Phase 2, focused on Development and Testing, consisted of the programming of interaction and automatic feedback modules, the implementation of performance monitoring dashboards, and error evaluation, verifying consistency and performing functional system tests. The integration between Google Sheets, Apps Script, and HTML code ensured that TGS2 predominantly used tools from the Google ecosystem, which contributed to the consistency and agility of integrations. This technological approach ensured the assistant’s robustness and scalability, fundamental for its effective operation.

The system was structured to receive questions from an external database, distribute them by difficulty levels, control individual access authentication, record performance history, and present a study interface with real-time monitoring elements. This implementation confirms that the project has advanced from the conceptual to the operational stage, translating its theoretical principles into concrete mechanisms for pedagogical interaction. The system’s login and authentication screen allows for individualized access and the retrieval of previously saved data, such as accumulated scores, saved levels, and previous statistics, enabling the continuity of the learning journey.

This data recovery functionality is crucial for the student’s continuous progression, as it ensures that each session is not treated in isolation, but as part of an ongoing learning journey. TGS2 was designed as a user-friendly, interactive, and individual-path-sensitive digital environment, where error is not seen merely as a negative outcome, but as information that guides review and feedback. Progression through levels, achievements, and encouragement messages, along with multimodal resources, integrate motivation and gamification into the learning process, favoring task persistence and student engagement.

The user interface, upon returning to the assistant, allows for the visualization of graphs of skills developed in previous sessions, which are updated with the results of the last use. The system also considers the student’s downtime, gradually reducing the skills represented in the graphs by lower usage, which reflects a concern for learning maintenance. This visual and dynamic approach contributes to self-regulated learning, allowing the student to track their progress and identify areas that require more attention or review.

The presentation of activities in TGS2 was implemented to support both open-ended responses and multiple-choice questions, with statements that can be accompanied by image, video, audio, and mathematical notation. This flexibility expands the possibilities of using the system in various disciplines and educational levels, reinforcing the artifact’s replicable nature. The presence of multimodal resources is fundamental to making learning more engaging and responsive, especially in contexts where written text may not be sufficient to sustain student attention, comprehension, and persistence in the task, as in language learning or for different age groups.

One of the main results of the implementation is the automatic adaptation mechanism. The system organizes questions into levels, controlling the current work level and adjusting progression based on the user’s recent performance. The student can advance by accumulating consistent correct answers and regress by presenting a sequence of errors that signal persistent difficulty. This operation materializes the proposal to adjust the complexity of the study experience based on the learner’s response, maintaining the challenge within an adequate range and avoiding both stagnation and cognitive overload, which is consistent with the literature on performance-driven progression.

Beyond trial-and-error adaptation, the system considers response time to calculate the score and speed indicators. The algorithm defines minimum and maximum time ranges, assigning positive or negative scores and a speed score. This aspect refines performance analysis, not being restricted to the final correctness, but being sensitive to response fluidity. In practice, the system differentiates situations where the content appears consolidated from those where correctness was achieved more slowly, providing a more complex picture of user interaction with the task and promoting a more precise knowledge assessment.

After each answer, the TGS2 offers immediate feedback to the student, visually informing them about correctness or error, displaying the correction when necessary, updating the score, and releasing a side area for further study. This feedback area can contain explanations in text, audio, video, image, or external link, transforming an error not into a failure, but into a starting point for guided review. This approach demonstrates the operationalization of the inverted logic, where the student is confronted with the question and then receives learning reinforcement based on the need demonstrated in their answer, promoting more active and reflective learning.

The system not only provides support materials but also monitors user interaction with these resources. The code records the type of feedback accessed and the time spent on each modality, in addition to allowing the student to evaluate the question through likes, difficulty marking, and optional comments. This set of functionalities expands the system’s observation capacity regarding study behavior, paving the way for future analyses of media preference, subjective perception of difficulty, and engagement with reinforcement resources, which allows for a deeper understanding of the learning process.

In the field of visual performance monitoring, TGS2 implements a dashboard with a radar chart and statistical summary by level. The calculated metrics include accuracy, consolidation, strength, speed, and strategy. Accuracy corresponds to the percentage of correct answers in a recent set of interactions; consolidation derives from speed scores on correct answers; strength considers the volume of recent activity; speed represents the average of temporal scores; and strategy relates to the accumulated score. These metrics provide a comprehensive view of student performance.

The performance dashboard also presents visual trend indicators and a comparison between the beginning and end of the session, transforming performance into information immediately perceptible to the user. This result is especially relevant for self-regulated learning, as it favors awareness of one’s own study process. The gamification structure is another concrete result of the implementation, with named levels, progression events, time-based challenges, unlockable achievements, incentive messages, and sounds associated with correct answers and milestones reached, which contribute to the student’s emotional engagement.

Positive reinforcement in the assistant is carried out through specific achievement screens, with videos, images, and texts, and also through sounds played when the student reaches certain parameters. Achievements are not always tied to high performance, but can also be achieved by using a different learning resource for the first time, for example. It is important to emphasize that the professor is able to define digital achievements, challenges, and rewards, personalizing the learning journey for specific audiences and promoting an experience more adapted to individual needs.

The system also provides for daily usage limits and guided session termination. Upon reaching the configured daily time, the TGS2 triggers a visual warning, changes the continuity button to session completion, and prevents undue extension of use. This mechanism is particularly interesting from the perspective of healthy learning, as it signals a concern with time regulation, preventing the gamification logic from converting into excessive use. Thus, the system seeks to balance motivation and restraint, challenge and pause, which is consistent with a more humanized digital pedagogy conscious of student well-being.

Regarding data persistence, the system was implemented to save batch information on scores, time, current level, answer details, and consolidated statistics, using integration with Google Script. This indicates that TGS2 is not limited to a local training interface but has a storage layer capable of sustaining historical and future usage analyses. This outcome strengthens the viability of the assistant as an expanding educational tool, allowing for the envisioning of applications with continuous monitoring and broader reports in institutional contexts, which amplifies its potential impact.

The data recorded by TGS2 can be organized and aggregated into performance reports for monitoring students, classes, learning levels, and usage periods. On a broader scale, this information can assist educational managers in identifying patterns, evaluating pedagogical interventions, and planning actions aimed at improving learning. This capability expands the system’s reach, offering not only support for individual study but also relevant information for institutional educational decisions, contributing to more informed pedagogical management.

From the discussion point of view, the results obtained allow us to state that TGS2 managed to transform the theoretical foundations presented in the introduction and methodology into observable functionalities. The centrality of immediate feedback appears in the instant correction and the offer of review right after the answer. The idea of significant repetition emerges in the guided revisiting of difficulties and in the logic of training continuity. The emotional and motivational dimension is expressed in the reinforcement messages, in the achievement milestones, and in the progress visualization, aligning with the theories of Kandel (2000), Damásio (2011), and Goleman (1995).

Personalization is materialized in the level adjustment, in the individual performance record, and in the selection of support resources according to the demonstrated need. In this sense, the assistant reveals conceptual adherence to the problem it set out to address, offering a learning experience that adapts to the pace and style of each student. However, it is necessary to recognize the limits of this research stage. The results described here concern the implementation and analytical evaluation of the system, and not an empirical validation with students in a real learning situation.

Therefore, it was not possible to affirm, at this moment, measurable gains in school performance, long-term memory, or sustained motivation in an authentic educational context. What could be affirmed, with certainty, is that the assistant was developed with internal coherence, presented functioning compatible with the project’s objectives, and offered concrete conditions for future application and research stages. The assistant’s main value lies in its ability to reorganize the study experience around the user’s response, identifying reinforcement needs and proposing continuity based on detected weaknesses and advances, which makes it compatible with contemporary proposals for teaching personalization.

In summary, TGS2 represents a relevant contribution to the dialogue between neuroscience, educational technology, and pedagogy, demonstrating that it is possible to build digital solutions that organize learning experiences sensitive to the student’s pace, performance, and motivation. The computational assistant developed in this work constitutes a functional technical product, but also a conceptual and practical basis for future innovations and research in the field of technology-mediated education, paving the way for empirical validations that can confirm its impacts in real teaching environments.

4. Conclusion

This work aimed to describe and present the creation of TGS2, a computational learning assistant that uses neuroscience principles to offer a more adaptive, interactive, and personalized educational process, guided by meaningful repetition, emotional engagement, and immediate feedback. It was found that TGS2 was developed through the Design Science Research methodology, demonstrating a functional implementation consistent with the proposed objectives. The system translated theoretical foundations into concrete computational mechanisms, such as level progression, automatic difficulty adjustment, multimodal feedback, and gamification. The assistant’s ability to organize personalized study paths, sensitive to the student’s pace and performance, and to record data for monitoring and analysis, reorganizing the study experience around user response and proposing continuity based on detected needs, was observed.

The main contribution of TGS2 lies in its ability to integrate neuroscience, educational technology, and pedagogy, offering a technical product that serves as a conceptual and practical basis for future innovations. However, it is essential to recognize that the results obtained correspond to the implementation and analytical evaluation of the system, without empirical validation with students in a real learning situation. Thus, it was not possible to affirm measurable gains in school performance or sustained motivation in an authentic educational context. It is suggested that future studies focus on the empirical validation of TGS2, with its application in real classes, comparison between groups, analysis of usage metrics, and collection of qualitative impressions from students and teachers, in order to verify its effects on learning and engagement.

Bibliographic References

Damásio, A.R. 2011. O Erro de Descartes: Emoção, Razão e o Cérebro Humano. Companhia das Letras, São Paulo, SP, Brasil.

Goleman, D. 1995. Inteligência Emocional. Objetiva, Rio de Janeiro, RJ, Brasil.

Kandel, E. R. 2000. Principles of Neural Science. McGraw-Hill, New York, Estados Unidos da América.

Peffers, K.; Tuunanen, T.; Rothenberger, M. A.; Chatterjee, S. 2007. A Design Science Research Methodology for Information Systems Research. Journal of Management Information Systems, 24(3), 45–77.

Article originating from a Final Coursework on Neuroscience and Learning in Education

To learn more about the course, click here and access the MBX Academy platform