Learning Isn’t Supposed to Hurt

Instructional theories are often considered learning theories. This category error is easy to make. A momentary analysis however reveals the logical error. Let’s look at an example.

Mastery learning even uses the word ‘learning.’ But it has everything to do with instruction and nothing really to do with learning. This is a profound conflation that pervades education. We keep mistaking delivery mechanisms for the actual phenomenon of learning.

Mastery learning actually is a perfect example to dwell on. It’s essentially a quality control system for instruction. “Don’t move on until 80% proficiency” is an administrative decision, not a discovery about how brains acquire knowledge. It tells you when to open the next gate, not what happens in the learner’s mind.

Yet educators treat mastery learning as if it revealed something fundamental about human cognition, when it’s really just industrial grade process management applied to the classroom.

***

Direct Instruction isn’t a learning theory either. It’s a script for teachers. It’s like mistaking a recipe for a theory of nutrition. The script might work, but not because it aligns with how learning happens; it works or not because it standardizes the definition of what learning is and is not. Learning is whatever the objective of the lesson measures.

Differentiated instruction pretends to be about how different students learn differently, but it’s actually just logistics. How to manage multiple tracks in the same classroom. It’s a classroom management strategy dressed up as cognitive science.

Scaffolding gets treated as if it describes learning, but it’s really about instructor behavior, i.e., when to provide support and when to remove it. The metaphor itself reveals this: scaffolding is external to the building. The building (learning) could happen with or without it; scaffolding just makes learning easier.

Even “constructivism” as a broad spectrum instructional method as typically implemented is more pragmatic than learning-focused. Teachers think they’re honoring how students “construct knowledge” by having them do hands-on activities, but they’re really just choosing an instructional format. The actual construction of knowledge happens apart from the instruction regardless of whether it was listening to a lecture or building a diorama.

This confusion has serious consequences. For one thing, teachers optimize for the wrong things. If mastery learning is mistaken for a learning theory, we focus on hitting arbitrary proficiency benchmarks instead of understanding actual learning processes and substrates through noticing, listening, and probing.

We cargo-cult the superficial features. Teachers implement “think-pair-share” because collaborative learning is supposed to be good, without understanding that the collaboration itself isn’t what causes learning. It’s the cognitive processes that collaboration might (or might not) trigger.

We usually ignore actual learning science. While we’re busy arguing about instructional methods, real findings about learning like the importance of sleep for consolidation, the role of prediction and constraints, or how prior knowledge shapes perception, inception, and conception gets ignored because they don’t come packaged as instructional techniques.

We create false dichotomies. The “traditional vs. progressive” education debate is almost entirely about instructional preferences, not learning. Teacher-centered vs student-centered is a thing because it impacts instruction, not learning. Both sides act like they’re defending theories of learning when they’re really just advocating for different classroom management styles.

Learning theories rarely make it into education because they don’t immediately translate into “Monday morning procedures.” They explain what happens in the brain, not what the teacher should do at 9 AM.

It’s like the entire field of education is focused on optimizing the delivery truck rather than how the content of the truck is unpacked and its load put to use in students’ lives.

Until AI, many instructional theories seemed to work despite being based on incorrect assumptions about human learning. Direct instruction might be effective in some circumstances not because learning is hierarchical and sequential (as its proponents claim), but because it produces the behaviors the instructor expects regardless of what was learned.​

AI has changed everything.

***

Cognitive load theory has become potent in the discourse of instruction as it swirls around issues of AI. It’s not a theory of learning. It is actually a theory about how the brain functions during a particular instructional framework, not about how learning occurs.

It describes the constraints of working memory, essentially saying “the cognitive plumbing can only handle so much at once.” Instruction determines learning by exquisitely detailed teacher activity. But people mistake this plumbing manual for a recipe for learning.

The confusion arises because the name itself is misleading. Cognitive load sounds like it’s about thinking and learning broadly, when it’s really about memory system constraints. When learning is reduced to the limits of working memory, all of the richness of creativity and critical thinking evaporates.

Like direct instruction, it makes testable predictions about instruction, which makes it look like a learning theory. When CLT says “worked examples are better than problem-solving for novices,” teachers think it’s explaining how people learn, when it’s actually just explaining how to avoid cognitive traffic jams at industrial scale.

The “productive struggle” anti-AI crowd co-opted the terminology of cognitive load theory to critique student uses of AI during academic assignments. AI lets the pressure out of the teacher’s strategy. Indeed, the core of the argument in favor of disallowing AI derives from its tendency to derail mechanistic instruction protocols that define learning as a product of instruction, not as a product of a motivated, self-regulated learner in a mentored setting.

They heard “germane cognitive load” (the good kind that builds schemas) and thought “ah, so difficulty equals learning!” But CLT actually says the opposite. Minimizing extraneous load so you have room for the germane load is the goal. It’s about efficiency, not struggle.

Learning means struggle, friction, discomfort to the anti-AI crowd. It’s hard. It fits our cultural biases. The “no pain, no gain” mentality maps perfectly onto misreading CLT as “learning requires cognitive strain.” We want learning to feel hard because it validates our effort.

It’s prescriptive in a way that feels like pedagogy but is really behavioral control. CLT tells teachers what to do (segment content, use worked examples, avoid split-attention), which makes it seem like it’s telling teachers how people learn, when it’s really just telling teachers how to work around working memory limits.

The reality is that CLT is more like an ergonomics guide for the mind. It tells teachers how to arrange information so it doesn’t strain the cognitive system, not how the system actually learns and integrates knowledge.

Instructional plans carefully calibrate the exact amount of confusion needed for learning, like chemists measuring reagents. Meanwhile, kids are out there learning languages from cartoons and TikTok without a single germane load calculation.

The irony being that cognitive load theory actually warns against unnecessary difficulty, yet it sometimes gets weaponized to justify making things harder than they need to be under the banner of “desirable difficulties.“​​​​​​​​​​​​​​​​

The student wasn’t struggling because the math was appropriately challenging. She was struggling because we’d created artificial friction where natural complexity should have produced joyful exploration.

We mistake the symptoms of friction-induced cognitive processing for the consciousness that creates learning. And in doing so, we remove something precious: the intrinsic joy that makes learning sustainable.

I argue that schools mistake complexity for a site of necessary struggle, no pain, no gain—and it’s costing learners more than we realize. It’s costing not just engagement and motivation, but the very essence of what makes us natural learners: curiosity, wonder, and the aesthetic pleasure of understanding.

.We’ve collapsed the rich, multidimensional nature of learning into a single, impoverished axis: easy versus hard. Learning is hard. It must be controlled and measured. If it isn’t hard it isn’t learning. And it just keeps getting harder and harder.

AI becomes the villain because it can makes the hard seem manageable. All of this is absurd, of course, because learning is neither easy nor hard. It is to the human what flight is to the bird. The problem is we wait until our children are adults before we let them use their cognitive wings.

***

Real learning doesn’t happen along a linear difficulty scale. It emerges from dynamic tensions between complementary forces. When I finally mapped these tensions on a napkin talking it over with a friend at Starbucks, five key polarities emerged:

Empirical ↔ Logical: Learning oscillates between gathering concrete evidence from the world and organizing it into abstract frameworks. A child learning to read moves between recognizing letter shapes (empirical) and understanding phonetic patterns (logical).

Uncertainty ↔ Certainty: Effective learners pivot between not-knowing and knowing, between questions and temporary answers. I watched for this in my classes: “I need to sit with not understanding this for a while before I try to explain it.”

Complex ↔ Simple: Understanding spirals between encountering sophisticated ideas, breaking them into manageable pieces, then reconstructing richer comprehension. Think of how children naturally move from simple sentences to complex narratives without being forced through arbitrary difficulty progressions.

Predictable ↔ Random: Learning benefits from both reliable patterns (the comfort of familiar routines) and surprising discoveries (the spark of unexpected connections). The best teachers I know create structured environments that leave plenty of room for the joy of serendipity.

Creative ↔ Stable: Growth requires both innovative thinking and consolidation into reliable knowledge. Students need to experiment with new ideas while building stable foundations they can count on.

***

The struggle paradigm flattens these rich dynamics into crude binaries: hard good, easy bad. It treats uncertainty as an obstacle rather than one pole of a generative tension. It sees unpredictability as chaos rather than opportunity for discovery. It sees AI as a thorn in its paw.

When we send signals that learning has to hurt, that AI is bad because it can mitigate confusion, we make learning a chore instead of a privilege—we inadvertently undermine learners’ trust in their own cognitive processes to learn on their own. Students learn to override their intrinsic motivation, to ignore their aesthetic sense, to suppress their curiosity in favor of manufactured difficulty.

Learning is not a struggle. Learning is the joy of being human.