Optimizing Students' Cognitive Load – SIP 2023, André Tricot

The discussion commences by framing the challenge of educational heterogeneity, acknowledging that while student differences are natural and not inherently problematic, the assumption that uniform teaching suits all students in a diverse classroom can generate significant learning difficulties or even failure for portions of the cohort, including both high and low performers. This reality contrasts sharply with social discourse advocating for mixed-ability settings and the rejection of streaming or tracking practices.

A solid finding in international literature indicates that the effect of multilevel classes—the very definition of heterogeneity—on student performance is absolutely null; it is neither positive nor negative. Furthermore, sociological surveys reveal that parents from higher socioeconomic backgrounds often express significant anxiety regarding heterogeneity, fearing contamination or time loss for their children, sometimes contradicting the egalitarian discourse adopted by educators themselves.

Studies involving large samples of teachers highlight heterogeneity as a major source of professional difficulty. Parental obsession with this issue often stems from a fear of negative influence, overlooking the 'big fish in a small pond' effect which suggests that high achievers might actually benefit from being among less proficient peers. Current educational devices in France have not adequately addressed the challenges posed by this complex, multidimensional issue.

For approximately fifty years, heterogeneity has been a significant research object, investigating whether homogeneous groups outperform heterogeneous ones, and identifying optimal levels of heterogeneity. Classic results, compiled by Marcel Cahen, show that while the average effect of homogeneity (level-based classes) is zero, this average masks strong negative effects for weaker students and slightly positive effects for the most proficient students.

More recent work by Julien Grenet and Élise Hilery, synthesized in a CEN note from April 2023, analyzed experiments from 2015 onwards involving changes to school zoning maps in cities like Paris and Brest. These extraordinarily interesting results, derived from real-world settings, demonstrated no significant effect on academic performance for either the highest or lowest achieving students. However, a positive effect was observed regarding personal and social well-being across all socioeconomic backgrounds.

When objectives are adapted, meaning higher learning goals for high performers and lower goals for low performers, this amplifies the differences rather than reducing heterogeneity.

Adapting teaching methods presents a complex area. Simply giving more learning time to fragile students sometimes yields inverse results if the extra time is spent repeating the same unsuccessful activity. Conversely, individualized strategies based on solid diagnosis, such as the work done by the former RAZ system, show positive outcomes, as does Benjamin Bloom's Mastery Learning, which benefits the entire class through high-quality instruction for all.

The perspective offered by Cognitive Load Theory (CLT) posits that academic learning is secondary—it is not intuitive and does not arise merely from environmental adaptation. Instead, school learning is specific to the educational context, often difficult to generalize without instructional support, and inherently costly. These processes necessitate significant effort, time investment, and strong cognitive engagement, often requiring motivational support from the environment beyond intrinsic student motivation.

To succeed in academic learning, students must overcome four distinct challenges sequentially. The first is engagement, requiring a readiness to expend effort. The second is concentration, demanding the mobilization of voluntary attention, which is essential for learning. The third involves realizing a specific task, such as solving a problem or reading a text, which relies heavily on metacognitive knowledge about how to execute that task type.

A central tenet of CLT is the critical distinction between the task—the activity performed (e.g., studying a text or solving a problem)—which serves as the means of learning, and the knowledge construction, which is the ultimate goal. This distinction is vital because the activity itself, while necessary, can consume cognitive resources intended for the actual modification or reinforcement of the student's knowledge base.

In any teaching-learning situation, cognitive resources are allocated across three distinct loads. The essential load refers to the cognitive resources dedicated directly to the learning process itself—the effort required for knowledge modification. The intrinsic load relates to the resources needed to manage the complexity inherent in the task being performed. Finally, extraneous load involves processing all superfluous information embedded in the instructional materials or presentation format that does not directly contribute to the learning objective.

The core insight of CLT is that a competition frequently occurs between the resources needed to successfully execute the task (intrinsic/extraneous load) and the resources required for actual learning (essential load). If the task execution demands too many resources, the learning process suffers, meaning the means mobilizes resources at the expense of the end goal.

Sometimes realizing the task is so demanding that it is at the expense of the resources for learning.

Examples outside of strict CLT, such as research on semantic priming, illustrate how task presentation affects demand. If students are asked to solve a multiplication problem presented abstractly versus one embedded in a concrete story about cakes, success rates differ significantly, even though the mathematical operation is identical. This difficulty arises when the concrete, real-world semantic representation conflicts with the abstract mathematical representation (e.g., 14 - 2 = 12).

The primary application of CLT over the last four decades involves developing techniques to reduce extraneous and unnecessarily high intrinsic cognitive load. The objective is always to free up the maximum amount of cognitive resources so they can be dedicated to the essential load—the actual learning. Research validates numerous effects through randomized controlled trials, comparing an optimized presentation against a standard control condition based on pre- and post-test gains.

One of the most robust findings is the effect of providing solved problems instead of requiring immediate problem-solving, especially early in the learning process. For students struggling with the targeted learning objective, studying worked examples leads to significantly better learning outcomes than attempting the problems cold. This finding has been replicated across hundreds of studies and various disciplines, moving beyond mathematics into areas like foreign language acquisition.

Physically integrating related information, such as placing figure commentary directly onto the figure rather than separately, proves powerful. Furthermore, when information is complex, presenting it progressively part-by-part is generally better, although a general objective must be established first to maintain the overall sense of the learning goal. The progressive fading of guidance—reducing support as mastery increases—is another well-documented, effective principle.

The integration of information that students must relate to each other is very powerful.

Beyond CLT techniques, Richard Mayer discovered the Pre-learning effect, which is remarkably simple yet highly impactful. This involves dedicating a minute or two at the start of a lesson to recall and present definitions, notions, and objects that will be discussed. This activation step brings necessary knowledge from long-term memory into working memory, resulting in an absolutely impressive boost to subsequent learning gains.

In conclusion, the persistent gap between the lowest and highest performing students in France continues to widen throughout schooling. Attempts to manage this heterogeneity by adapting learning objectives based on student level often exacerbate inequality. The effective path forward, as illustrated by the various CLT effects, involves temporarily lowering the cognitive demand of the task execution while preserving the high learning objective for everyone.