Education loves its theories. We love them with a kind of earnest devotion that would be charming if it were not so historically destructive. A theory emerges in the literature, carefully tested in small, controlled studies involving volunteers who sit quietly, complete tidy tasks and never once throw a pencil across the room. Then, almost immediately, the theory becomes a solution to the chaos of real classrooms, which have never been quiet, tidy or predictable. A theory begins as a precise explanation of something very specific. By the time it arrives in a school, it has been repackaged as a teaching method, a policy platform and, occasionally, a belief system. Cognitive Load Theory is the most recent example of this migration. It is a solid theory with a real research pedigree, but the enthusiasm with which it has been taken up says far more about education’s longing for certainty than it does about how students actually learn.
Cognitive Load Theory was never intended to be a teaching method. It was not even designed to be a broad theory of learning. CLT is a theory of cognitive architecture. Sweller’s original work in the late 1980s examined the limits of working memory, the role of long-term memory and the way novice problem solvers use means-end analysis. The worked example effect was one product of this research. Novices learned complex procedures more effectively by studying a worked example than by solving a similar problem from scratch. The mechanism was simple. Problem-solving requires cognitive search. Cognitive search requires working memory. Working memory is limited. Replace the search with a clear model, and novices can spend their time forming schemas rather than wondering what they are meant to do next. This basic idea has been replicated many times, particularly in mathematics and physics, where the tasks can be defined with surgical precision.
The research base is impressive. Paas and van Merriënboer (1994) developed methods for measuring cognitive load, including subjective ratings and dual-task performance. Kalyuga, Chandler and Sweller (1999) identified the redundancy and split-attention effects. The modality effect demonstrated that distributing information across different sensory channels can improve learning. By the late 1990s these strands had been combined into a coherent instructional design framework. Any educator who takes the time to read the literature will see that the confidence placed in CLT is not unwarranted. The problem lies elsewhere. The problem is not the theory. The problem is what happens when any theory leaves the careful world of research and enters the much less careful world of practice.
Something always gets lost in translation. We have been through this before. Multiple Intelligences is the classic example. Gardner’s (1983) theory was a cultural and psychological argument about the variety of ways humans reason about the world. It was never a learning style theory, and Gardner (1999) said so repeatedly. There is no evidence to support the idea that students should be taught through their supposed dominant intelligence. Yet schools did exactly that. MI became a poster, a personality profile, a set of colour-coded stations and, at its worst, a commercial program pitched to principals who wanted to be progressive. Gardner (2011) spent years distancing himself from the versions of MI that existed in actual classrooms. The more he explained the misuse, the more entrenched the misuse became.
CLT is now at risk of following the same trajectory. The issue is not that CLT is weak. It is that education has a tendency to turn every theory into a universal method. Researchers offer cautious interpretations. Schools convert them into doctrines. Social media distils them into slogans. Somewhere between the article and the staff meeting, CLT shifts from an explanation of how novices learn structured material to a sweeping claim about how all students learn at all times. The leap is breathtaking. It is also predictable. A profession desperate for certainty will cling to any theory that feels scientific, especially if it allows us to believe that learning can be made clean, orderly and measurable.
The problem is that classrooms are nothing like the environments in which CLT research takes place. Laboratory studies involve quiet rooms, short tasks and participants whose prior knowledge can be controlled. The classroom involves students who arrive tired, distracted, anxious, overconfident, underconfident, hungry, bored or preoccupied with the video game they were playing on their way to class. Prior knowledge varies wildly. The content may be structured or interpretive. A teacher begins with a lesson plan, but within ten minutes the plan is wrestling with clarifying questions, pastoral issues and the general hormonal unpredictability of adolescents. Cognitive load exists in this environment, but it exists inside a hurricane of competing demands and shifting intentions.
The humanities make this point particularly clear. CLT is excellent for the structural parts of English and history. Students benefit from explicit vocabulary instruction. They benefit from modelled writing. They benefit from clarity about text structure and from targeted guidance through complex texts. There is robust evidence for this. Reducing irrelevant detail helps novice readers focus on the central ideas. Modelling reduces the cognitive burden of planning. Breaking down a text supports comprehension. None of this is controversial.
The issue arises when the task moves beyond structure. Interpretation does not sit neatly within the boundaries of CLT. A poem does not offer a single schema waiting to be lodged into long-term memory. Meaning is indeterminate. Students must generate possibilities, test them, reject them and replace them. Wineburg’s research on historical thinking shows that expert historians do not simply retrieve schemas. They interrogate sources, question motives, and evaluate evidence with scepticism. Argumentation studies show that students develop claims through cycles of drafting, refining and reassessing, not through the linear application of a model. Creativity sits even further outside CLT’s reach. Divergent thinking involves expanding the space of possibilities, not narrowing it. Beaty’s (2014; 2016) work in neuroscience links creativity to spontaneous thought, associative processing and even cognitive disorganisation. These are not the kinds of processes that benefit from reducing extraneous load. Ethical reasoning is further again. Moral judgment involves empathy, values, social norms and personal experience. None of this can be predicted by cognitive load equations.
CLT is not wrong about these tasks. It is simply irrelevant to their core mechanisms. CLT shines when novices are learning structured material. It does not shine when learners must confront ambiguity. It does not help students wrestle with competing interpretations or form original judgment. These forms of learning are not examples of poor practice. They are examples of humanistic inquiry. They require cognitive freedom, not cognitive reduction. CLT works beautifully to get students to the starting line. It does not run the race.
The broader lesson is that theories rarely translate cleanly into practice. This is not an indictment of theory. Research plays a crucial role. It sharpens our understanding. It exposes assumptions. It gives us language to describe what we observe. But research is conducted in carefully controlled environments. Teaching is conducted in the real world. The difference matters. Teachers rely on judgement, not just knowledge. They interpret theory in the context of students who change daily, content that varies in complexity, and classrooms shaped by culture, policy, and individual need. Good practice is never the pure application of a theory. It is the intelligent blending of ideas into something that fits the moment.
CLT deserves a place in that blend. It tells us why clarity helps novices. It tells us why too much information can overwhelm learners. It tells us why modelling can accelerate schema formation. These are important insights. But the moment CLT becomes a complete pedagogy, we begin to distort it. The theory becomes a list of commandments. It becomes a way to dismiss approaches that do not fit its limited frame. It becomes a weapon in debates that should be grounded in nuance rather than ideology.
There is no need for this. The pendulum does not need to swing from unstructured progressivism to CLT fundamentalism. It can rest somewhere in the middle, where theories are tools and not doctrines. Teachers can use CLT to build firm foundations. They can use other methods when students are ready for ambiguity and complexity. They can recognise that different tasks require different approaches. They can hold multiple ideas in view without demanding that one of them be the final solution.
If CLT is the answer to every question, then the wrong questions are being asked. Classrooms will always be larger than any single theory. Theories describe. Teachers decide. That is the work. That is the profession.
Maximising Student Potential 