Resolving Cognitive Debt : The Role of Learning Communities and Reshaping Motivation
“True learning often unfolds in the seemingly mistaken process of exploration. Through the wise use of AI as a scaffold, we can reclaim the authentic joy of discovery.”
Speaker
Emma Macey
Dean of Academics and Strategy, Winchester College, UK
Today, LLM-based AI is profoundly influencing learning across subjects, with literacy and writing courses being among the earliest and most deeply impacted. In response, we propose the following phased design for literacy education:
Phase 1 (No LLM Involvement):
Building Word Sense. This includes mastering the fundamental "code" of language: phonics, spelling, vocabulary, grammar, tense, and sentence structure.
Phase 2 (Transitional, Subject-Specific):
Developing Fluency in Form and Expression. This stage focuses on genre, style, function, and context. Here, students move beyond memorizing words—they become curious about a word’s history and etymology, which further sparks their interest. They also begin developing a purposeful, independent voice, as writing is how we construct our own perspectives.
Phase 3 (LLM as a Support Tool):
Once a solid foundation is established, LLMs can be used for tasks that do not replace thinking—such as creating study plans, generating practice exercises based on the student’s own work, or assisting with spell-checking and citations. The key is helping students learn how to self-assess, making the metacognitive process more central.
Phase 4 (Assessment):
I believe that in the future, we will have two types of assessments—one that permits AI use and one that does not—much like how we treat calculators today. For project-based assignments, we may revive oral defenses to verify that students have truly internalized what they’ve learned.
Self-Determination Theory:
How AI Impacts Students’ Core Needs
From a motivational perspective, why do adolescents exhibit specific behaviors when using LLMs? Self-Determination Theory states that students thrive when three fundamental psychological needs are met: Autonomy, Competence, and Relatedness.
Yet, many educational systems unintentionally undermine these needs during the critical developmental stage of adolescence.
-
Autonomy is stifled by rigid schedules.
-
Competence is reduced to scores on high-stakes tests rather than mastery-based growth.
-
Relatedness is replaced by assigned classroom groupings instead of organic, meaningful connections.
When schools overlook these basic needs, students increasingly turn to online alternatives. LLMs, too, deeply influence these three areas:
▶ Relatedness
For socially isolated students, AI may seem to offer safe companionship. But real learning depends on human interaction. In one study at a girls’ school, high-achieving students struggled most with the skill of "asking for help."
They would rather seek answers online than appear "unable" in front of peers. Yet explaining concepts, defending viewpoints, and engaging in debate within a learning community are essential for deep understanding. AI interactions tend toward "averaged" responses, reinforcing echo chambers rather than sparking genuine dialogue.
▶ Competence
As discussed earlier, over-reliance on LLMs leads to "cognitive debt," eroding not only subject knowledge but also key academic traits like perseverance, problem-solving, and learning from failure. When LLMs provide instant answers, students miss the essential "struggle" required to develop intellectual resilience and a sense of accomplishment. Real learning often happens where mistakes occur.
▶ Autonomy
Especially in high-achieving classes, overly structured environments may be a root cause of declining motivation during adolescence. Well-meaning adults—teachers, parents, counselors—often exacerbate this through what we call "over-helping." When you do something for a student (especially girls, who may be more sensitive to this) that they are fully capable of doing themselves, the implicit message becomes, "I don’t think you can do this." Over time, the student internalizes that belief.
Alarmingly, high achievers—striving to maintain a perfect image—may turn to LLMs as a "crutch" for tasks they could accomplish independently. A skilled teacher knows when to remove scaffolding and encourage, "You can do this—try again." AI, however, keeps offering help until dependence forms and critical thinking weakens. Counterintuitively, it’s often not the struggling students but the high achievers—those reluctant to show vulnerability—who fall into this dependency.
Types of Motivation:
Why Do We Learn?
To guide children toward developing healthy learning motivation, it is essential to understand the different types of motivation and their actual impact on learning.
Positive Motivation:
-
Intrinsic Motivation (pure joy and curiosity)
-
Integrated Motivation (aligned with personal values)
-
Identified Motivation (goal-oriented)
Negative Motivation:
-
Amotivation (lack of purpose)
-
External Motivation (driven by rewards or avoiding punishment)
-
Introjected Motivation (rooted in guilt and shame)
External motivation—such as pursuing trophies or top grades—can be effective in early stages. However, as students enter adolescence, they often begin to resist: "I don’t care about that trophy; you can’t force me."
More commonly observed among female students is introjected motivation—"to avoid disappointing my parents" or "to not look incompetent in front of my friends." While this type of motivation can drive engagement, it often leads to burnout and anxiety by the later years of high school.
The good news is that we can guide students from negative to positive forms of motivation—an effort that becomes particularly crucial when integrating Large Language Models (LLMs) into learning.
By clarifying the school’s core values—such as emphasizing process, critical thinking, and a genuine love for knowledge—we can help students transform "introjected motivation" into "integrated motivation" that aligns with their personal values.
We can also assist students in setting personal goals, shifting "external motivation" toward "identified motivation." For example, telling a student who dreams of becoming a marine biologist: "If you want to become an expert, this challenging work must ultimately be done by you."
Conversations like these are becoming increasingly important. As for the highest form of motivation—intrinsic motivation (learning for the sheer joy of it)—research shows it is the most difficult to cultivate through direct intervention. Instead, educators can focus on fostering integrated and identified motivation, which more effectively supports sustainable, long-term learning.
Key Takeaways & Actions
In the classroom, disengagement often manifests differently between boys and girls. Boys tend to display overt, behavioral disengagement (e.g., disruption), while girls often show covert, withdrawn disengagement—what we call compliant disengagement. They may sit quietly, physically present yet mentally absent, making their disengagement harder to detect.
Building on these insights, we have taken concrete steps in our partner schools. For example, at Winchester College, we adjusted the digital device policy during evening study hours for younger students, returning to pen-and-paper assignments. This allows teachers to consciously control when and how to introduce or withdraw AI as a scaffolding tool.
To summarize, our core arguments are:
-
Cognitive effort is central to learning and cannot be outsourced. Learning is inherently challenging, and the sense of achievement comes precisely from the effort we invest.
-
Large language models hold the potential to extend thinking, not replace it. Their effective use depends on a solid foundation of core skills.
-
AI cannot replace human dialogue and critical exchange.
The interactions between teachers and students, and among peers, form the essential soil in which critical thinking grows.