Learning vs. Mastery: Rethinking "Smart" Learning Systems for Optimal Growth

By Anastasia Betts, Ph.D., Executive Director, Learnology Labs

Introduction: A Learning Scientist’s Journey

As a learning scientist and researcher, I’ve spent over a decade at the intersection of educational theory, technology, and practice. I’ve had the privilege of designing smart learning systems that serve millions of children, each with their unique strengths, challenges, and opportunities for growth. Along the way, I’ve grappled with a question at the heart of this work: What does it truly mean to optimize learning?

This question isn’t just theoretical for me—it’s shaped every design decision I’ve made working in the field of EdTech. In some cases, I’ve leaned heavily on mastery-based approaches, ensuring learners demonstrate independent proficiency before advancing (Betts, 2019). In others, I’ve experimented with learning-focused systems that prioritize challenging learners at the edge of their potential, leveraging insights from Vygotsky’s Zones of Development theories (Betts et al., 2024; Zaretskii, 2009).

Through these experiences, I’ve come to see the tension between verifying mastery and maximizing learning growth not as a binary choice, but as a dynamic balancing act—one that requires us to rethink how we use time, technology, and human expertise in learning systems. This article represents my latest thinking around that tension and proposes recommendations for future directions, drawing on insights from both theory and practice.

How We Use Time in Learning: A Critical Choice

Consider a four-year-old working with her teacher on an early math activity. She eagerly recites "one, two, three, four, five!" and confidently shows five fingers when asked. But when prompted to count and create a set of five objects, she hesitates, needing her teacher’s guidance to carefully count each item and keep track. With this support, she begins to succeed, demonstrating growth in her understanding of numbers and counting.

This moment highlights a fundamental question in learning design: Are we optimizing for what learners can already do independently, or for what they can achieve with support? Traditionally, learning systems have focused on the former, emphasizing mastery—what Benjamin Bloom (1968) defined as a learner’s ability to demonstrate proficiency without assistance. This definition of mastery closely aligns with Lev Vygotsky’s Zone of Actual Development (ZAD), which also emphasizes what learners can do on their own, without the help of a more knowledgeable other (MKO; Zaretskii, 2009). Both frameworks highlight the importance of ensuring learners have solid, independent foundations before moving forward.

However, Vygotsky’s theory pushes beyond conceptions of mastery to propose the Zone of Proximal Development (ZPD), where the most learning occurs. In this space, learners are challenged at the edge of their current abilities, stretching into new understanding with the help of an MKO, such as a teacher, parent, or peer (Zaretskii, 2009). When time is precious—especially for students working below grade level—this distinction becomes critical. Every minute spent verifying what a learner can already do is a minute not spent helping them grow into what they could achieve next.

This tension between focusing on mastery (the ZAD) and maximizing growth (the ZPD) lies at the heart of how we design learning systems. Take early mathematics for example. It sets the foundation for later academic success, yet over 60% of fourth-grade students are not meeting grade-level expectations in math (NAEP, 2022). Research shows these gaps often begin before kindergarten and widen with each passing year (Duncan et al., 2007). It's true that mastery-based approaches grounded in Bloom’s theories provide a systematic way to secure foundational understanding. However, I've recently begun to think that they may inadvertently limit opportunities for accelerated growth--primarily by focusing too heavily on verifying what students can already do, rather than leveraging every single moment to push the learning further.

Two Perspectives: Bloom and Vygotsky

Benjamin Bloom and Lev Vygotsky, while separated by geography and historical context, both sought to define how individuals accumulate learning and their work continues to inform educational practices and system design today. However, their approaches emphasize fundamentally different aspects of learning.

Bloom: Mastery Through Independent Demonstration

Benjamin Bloom, an American educational psychologist, is best known for his framework of mastery learning. He argued that all students can achieve high levels of learning if given enough time and the right conditions. Central to his approach is the idea that learners must demonstrate independent proficiency in one concept or skill before progressing to the next. This focus on mastery ensures a systematic approach to learning, where gaps in understanding are minimized, and misconceptions are addressed early--an approach particularly useful in learning early mathematics, where much learning is sequential and hierarchical.

Mastery, in this sense, provides clear evidence of what a learner knows and can do independently, forming the foundation for future learning. However, the process of achieving and demonstrating mastery requires time—a resource that is often in short supply, particularly for students who are behind. While mastery-based approaches provide security and confidence in foundational skills, they may inadvertently slow the pace of learning by emphasizing repeated demonstrations of independence.

Vygotsky: Stretching the Learner through Scaffolded Support

On the other hand, Russian psychologist Lev Vygotsky approached learning from a different perspective. His theory of the various zones of development characterized different kinds of learning -- some that a learner could do on their own, and other kinds of learning that one could do in partnership with an MKO. Vygotksy described the Zone of Proximal Development (ZPD) which emphasizes not what learners can already do independently, but what they could learn through more difficult tasks in partnership with and support from a more knowledgeable other (MKO), such as a teacher, peer, or caregiver.

Unlike Bloom, Vygotsky’s focus was not on mastery as an endpoint but on the process of learning itself. He theorized that the most meaningful and efficient learning occurs at the farthest edges of the ZPD, where learners are challenged enough to spur active engagement, and to push through with some help--but not so much challenge that tasks become insurmountable (which Vygotsky called the "Zone of Insurmountable Difficulty"). Importantly, Vygotsky acknowledged that the ZPD is not uniform—two learners at the same Zone of Actual Development may have vastly different capacities for growth (i.e., ZPD).

This insight has profound implications. Systems that rely solely on demonstrations of mastery may hinder those with broader (i.e., "stretchier") ZPDs, effectively limiting their learning potential and pace. For these learners, progressing to the very next step or activity in a learning progression may underutilize their capacity to stretch further, leaving valuable learning opportunities on the table.

Rethinking Smart Learning Systems: Maximizing Learning Potential

These two perspectives present a productive tension for educators and designers of learning systems. Bloom’s approach emphasizes confidence—ensuring that learners have a solid foundation before moving forward. Vygotsky’s approach emphasizes efficiency—maximizing learning by focusing on the most learning that learners can achieve with support. Both are valuable, but they lead to fundamentally different design choices.

For students who are behind, mastery learning ensures that gaps are addressed systematically, preventing future struggles. But when time is limited, and the need for acceleration is urgent, focusing solely on mastery may slow progress unnecessarily. Conversely, Vygotsky’s focus on guided stretch may offer a pathway to faster, more efficient learning, but it raises questions about how and when to validate student mastery.

Designing for ZAD vs. ZPD

Current systems often treat learners at the same Zone of Actual Development (ZAD) as having the same level of readiness for new material. For example, a learner who has just demonstrated independent proficiency in counting to five might next encounter counting to ten. While logical in a mastery focused system, this approach overlooks the variability in learners’ Zones of Proximal Development (ZPD).

The learning efficiency approach offers a fundamentally different solution by optimizing for the Zone of Proximal Development (ZPD) rather than the Zone of Actual Development (ZAD). It prioritizes identifying and targeting the learner’s growing edge—the furthest point at which they can be challenged beyond their current capabilities, and still succeed with help. This approach ensures that every moment is used to maximize growth.

Learning at the furthest edge of the ZPD is not about ease; it is about effortful engagement. The MKO plays a vital role in maintaining this balance, providing just the right level of support to help the learner succeed without removing the challenge. This dynamic process fosters deeper understanding and accelerates progress, ensuring learners spend their time actively growing rather than consolidating or demonstrating what they already know (Betts et al., 2024).

The Role of Predictive Analytics and AI

Optimizing for the ZPD requires a nuanced understanding of each learner’s readiness for new challenges—a task that grows exponentially complex as the number of learners and potential learning pathways increases. This is where advances in predictive analytics and AI offer transformative possibilities. By analyzing patterns in learner performance, these technologies can identify the learner’s growing edge--what Vygotsky termed the "point of difficulty" (POD)--and dynamically adjust learning pathways to ensure every moment is maximized for learning growth.

Inferring Readiness and Targeting the ZPD

Predictive analytics enables systems to go beyond simple assessments of what a learner has mastered. Instead, readiness for tasks that stretch further into the learner’s ZPD can be inferred, based on observed behaviors and patterns of success and struggle collected and analyzed over time. For example, the system might detect that a learner who has successfully mastered certain precursor concepts and skills may have the capacity to explore much more advanced concepts with support, even if they have not yet demonstrated independent mastery of interim objectives. Conversely, if patterns of errors suggest a foundational misconception, the system can flag this for targeted review, ensuring that gaps are addressed before they hinder future growth. This ability to infer readiness allows systems to reduce the time spent on unnecessary verification while focusing on tasks that offer the greatest potential for the accumulation of more learning.

Dynamic Learning Pathways

Unlike traditional systems that follow rigid, linear progressions,?AI-powered systems designed to optimize for the ZPD create dynamic learning pathways tailored to each learner’s unique needs. These thoughtfully engineered systems have the potential to dynamically adjust tasks to:

• Push learners to the furthest Point of Difficulty (PoD): Presenting challenges that require vigorous effort but are still achievable with MKO support, ensuring each activity stretches the learner’s understanding.
• Balance efficiency with confidence: Periodically validate mastery to ensure foundational skills remain solid, without interrupting the momentum of new learning.

It’s important to note that not all AI-powered systems operate in this way. Many (in fact most) systems prioritize linear progressions or focus solely on independent mastery without leveraging ZPD insights. However, systems specifically designed to integrate predictive analytics, ZPD elasticity, and MKO scaffolding have the potential to fundamentally transform the amount of learning that can be acquired in the shortest amount of time (i.e., Learning Efficiency).

Over time, as these systems refine their understanding of not just individual learners, but hundreds of thousands of learners through performance data and interaction feedback, they become increasingly precise in targeting the growing edge of each individual's ZPD. This precision minimizes wasted time on unnecessary repetition and maximizes opportunities for meaningful growth in the most efficient ways possible.

Supporting the More Knowledgeable Other

While technology provides critical insights, the human MKO remains central to the learning process. Predictive analytics and AI enhance this partnership by:

• Offering data-driven recommendations: Systems suggest tasks that align with the learner’s readiness, giving MKOs a clear starting point for scaffolding.
• Highlighting areas for observation: AI flags moments where the learner may need additional support or where misconceptions might arise, enabling proactive intervention.

This integration creates a dynamic feedback loop: the MKO observes and interacts with the learner, refining the system’s predictions and ensuring that each task is both challenging and achievable.

Implications for Educational Design

The integration of predictive analytics and AI into ZPD-optimized systems has the potential to address some of the most persistent challenges in education, particularly for learners who are behind or face systemic barriers to success. By rethinking how we balance mastery verification with learning efficiency and acceleration, these systems offer new pathways for designing equitable and effective learning experiences.

Equity Through Personalization

In traditional smart learning systems, learners assessed at the same level of development often receive identical tasks, regardless of their individual readiness for growth. This one-size-fits-all approach—even within adaptive systems—can unintentionally limit learners with greater capacity to stretch further into their ZPD while overwhelming those who need more incremental steps.

Adaptive ZPD-optimized systems address this disparity by tailoring learning pathways to each individual, but their potential becomes even more transformative when combined with AI and predictive analytics. By analyzing patterns across thousands of learners, and hundreds of thousands of activities, these systems can begin to predict not only what a learner is ready to tackle next but also the elasticity of their ZPD—how far they can stretch into more complex tasks with appropriate support. Over time, as more learner data is collected and analyzed, these predictions become increasingly accurate and effective. The system can model not only what a learner knows but how far they can be stretched, empowering educators to provide challenges that are optimally aligned with each child’s potential. Moreover, better predictive analytics may potentially make the need for demonstrations of mastery less frequent, or in some cases altogether unnecessary.

This personalization has profound implications for equity. By leveraging data-driven insights, ZPD-optimized systems ensure that all students—regardless of starting point—receive tasks that maximize their learning opportunities. For learners from historically marginalized communities, who often face systemic barriers and variability in prior educational experiences, these systems offer a pathway to more equitable and meaningful engagement.

A New Role for Technology: ZPD Optimizer

Ultimately, ZPD-optimized systems represent a fundamental shift in how we design technology in education. Instead of merely assessing what learners can do independently--essentially requiring them to "prove" that they know the content--these systems dynamically create and leverage the optimal conditions for learning and growth. By enhancing the partnership between technology and human MKOs, they enable more responsive, relational, and effective learning environments.

This shift moves technology from a passive role—focused on assessment and verification (i.e., mastery)—to an active collaborator in fostering learning. It transforms smart systems into tools that not only track progress but also catalyze it, ensuring that every learner has the opportunity to thrive in the most efficient ways possible.

Conclusion: A Call to Action for Smarter Learning

The stakes for rethinking smart learning systems couldn’t be higher. Today, millions of children face widening gaps in foundational skills, particularly in mathematics, where early struggles often set the stage for lifelong challenges. These are not abstract numbers—they are real children, each with untapped potential, waiting for opportunities to thrive.

As learning scientists, educators, and technologists, we carry a profound responsibility to design systems that don’t just track progress but actively create the conditions for meaningful growth in the most efficient ways possible. By leveraging learning science principles, predictive analytics, and insights from both Bloom and Vygotsky, we can transform the way we think about learning. These systems must balance confidence in foundational understanding with the ability to accelerate learning—empowering every child to reach beyond their current limits.

The tools to achieve this exist. The challenge lies in how we choose to use them. In a world where time is precious and every moment counts, shouldn’t we design systems that make every learner’s potential a priority? The future of smart learning isn’t just about innovation—it’s about responsibility, equity, and the urgent need to unlock learning for all.

Additional resources...

To read more about ZPD Elasticity, and some experiments we ran to test this theory, see this paper: https://papers.iafor.org/wp-content/uploads/papers/iice2024/IICE2024_77367.pdf

Or, watch a presentation we gave last year on the same topic and experiment: https://vimeo.com/899123140

To better understand Vygotsky's theories on Zones of Development, check out this paper: https://www.researchgate.net/profile/Viktor-Zaretsky/publication/247887086_The_Zone_of_Proximal_Development_What_Vygotsky_Did_Not_Have_Time_to_Write/links/6198e4d73068c54fa505f45d/The-Zone-of-Proximal-Development-What-Vygotsky-Did-Not-Have-Time-to-Write.pdf

References

Betts, A. (2019). Mastery learning in early childhood mathematics through adaptive technologies. In The IAFOR International Conference on Education–Hawaii 2019 Official Conference Proceedings. Paper Presented at the IAFOR International Conference on Education: Independence and Interdependence, Hawaii (pp. 51-63).

Betts, A., Hughes, D., Plache, L., & Smith, K. (2024). Stretching the Zone of Proximal Development: Accelerating Learning through ZPD Elasticity. The IAFOR International Conference on Education – Hawaii 2024 Official Conference Proceedings.

Bloom, B. S. (1968). Learning for mastery. In J.H. Block (Ed.), Mastery learning: Theory and practice (pp. 47-63). Holt, Rinehart, & Winston.

Bloom, B. S. (1984). The 2-sigma problem: The search for methods of group instruction as effective as one-to-one tutoring. Educational Researcher, 13(6), 4-16.

Duncan, G. J., Dowsett, C. J., Claessens, A., Magnuson, K., Huston, A. C., Klebanov, P., ... & Japel, C. (2007). School readiness and later achievement. Developmental Psychology, 43(6), 1428-1446.

National Assessment of Educational Progress. (2022). NAEP long-term trend assessment results: Reading and mathematics. Retrieved from https://www.nationsreportcard.gov/highlights/ltt/2022/

Vygotsky, L. S. (1978). Mind in society: The development of higher psychological processes. Harvard University Press.

Vygotsky, L. S. (1986). Thought and language (A. Kozulin, Ed.). MIT Press.

Zaretskii, V. K. (2009). The zone of proximal development: What Vygotsky did not have time to write. Journal of Russian & East European Psychology, 47(6), 70-93.