Cognitive Load Theory (CLT) is a theory of learning that has played an important role in recent debates about teaching math.1 At the core of CLT is an attempt to show how learning is constrained by the limits of the human mind. CLT researchers have argued that these limits doom many instructional approaches to failure. The doomed pedagogies often include discovery math, problem-based learning and progressive education more broadly.
None of this has happened without controversy. In educational circles, attention is most commonly directed at disagreements between CLT researchers and advocates of these “doomed pedagogies.” While those debates are important, too often we neglect the differences of opinion within the circle of scientists who fully accept CLT’s premises. From the substance of their debates, we can learn how hard it is to study teaching and learning. From the fact of their disagreements, we can learn how individual human judgement impacts the direction of science.
In recent years, CLT theorists have disagreed as to how much complexity their work should encompass. Learning depends on so many factors — everything from a student’s home life to their personal interests — that no theory can encompass it all. To do their work, scientists need to find the proper balance between careful control (limit the factors) and relevance (embrace messiness). There is no recipe for finding this balance, and some of the most fascinating disagreements in educational research come down to this one issue: what complexities need to be included in research if the results are to be relevant for teaching?
Some researchers want to include student motivation in the work of CLT. Others disagree, instead arguing that motivation falls outside the scope of the theory. A fascinating aspect of these internal struggles is that the inventor of CLT, John Sweller, has at different times advocated for each of these positions. This essay is about how John Sweller came to invent CLT, how he expanded the theory to embrace more complexity, and eventually restricted the boundaries of CLT to exclude this complexity.