Desirable difficulties and cognitive load: Three authors solve a non-existent problem

When I began to read the article I was enthusiastic and thought that I’d begin 2025 with an enthusiastic blog. After reading it I thought: Another worthless article presenting and then solving a non-existent problem.
What follows is a critical discussion of Does difficulty moderate learning? A comparative analysis of the desirable difficulties framework and cognitive load theory by Pyke et al.[1]

In the quest to improve educational outcomes, the question arises: How do we create optimal learning environments? On the face of it, one could say that there are two schools of thought, both grounded in theory-driven evidence from multitudes of empirical studies.[2] Should we make tasks harder to strengthen memory and retention, or should we simplify them to prevent cognitive overload?

Disclaimer 1: I have never seen the two approaches to be antagonistic or incompatible,

A recent study by Wesley Pyke, Johan Lunau, and Amir-Homayoun Javadi delves into this question by comparing the two prominent theories at the basis of this seeming antagonism, namely the Desirable Difficulties and Cognitive Load. By exploring these frameworks and their practical implications, the authors say that they propose a ‘new’ model that integrates the strengths of both approaches.

Contrasting Two Learning Theories

Desirable Difficulties

The concept of desirable difficulties, introduced by Robert Bjork (1994)[3] and later expanded in his collaboration with is wife Elisabeth Bjork[4], suggests that creating certain types of difficulty during learning can enhance long-term memory. It emphasises that tasks requiring more cognitive effort—like spaced learning, interleaving, or retrieval practice[5]—promote deeper encoding and stronger memory traces. Mirjam Neelen and I discuss this in our blog Demystifying desirable difficulties 1: What they are. These strategies may initially hinder performance, but they yield substantial long-term benefits.

Key components of desirable difficulties include:

• Spacing: Distributing study sessions over time rather than cramming.
• Variation: Varying tasks within study sessions rather than blocking them,
• Retrieval: Testing learners rather than merely repeatedly exposing them to information.

One could simplify this by saying that all three (actually five) are based upon what can be called the retrieval effort hypothesis which suggests that more difficult retrieval, such as retrieval involving differentiation between tasks or expanded intervals between study, reinforce memory by demanding the learner expend more cognitive resources.

While desirable difficulties has been supported by robust evidence, critics point out that tasks that are too difficult can demotivate learners or lead to errors that hinder progress.

Disclaimer 2: Mirjam Neelen and I discuss this in our blog Demystifying desirable difficulties 2: What they’re NOT and they’re definitely not tasks that are too difficult to answer nor tasks that supposedly lead to productive failure!

Cognitive Load Theory

Cognitive load theory, developed by John Sweller (1988), offers a different – what some might see as contrasting – perspective by focusing on working memory’s limitations. It posits that effective learning happens when cognitive load is optimised—not too light to fail engagement, nor too heavy to overwhelm the learner.

According to Pyke et al, cognitive load theory breaks cognitive load into three types:

• Intrinsic Load: The complexity inherent in the material (e.g., solving a quadratic equation vs. memorizing vocabulary).
• Extraneous Load: Unnecessary mental effort caused by poorly designed instruction (e.g., confusing layouts or irrelevant details).
• Germane Load: The mental effort allocated to schema-building, which enhances learning.

Disclaimer 3: This is where I began to have second and third thoughts about Pyke et al. CLT didn’t have germane load at it’s inception. It was added at a certain point by fellow researchers Paas and Van Merriënboer (and was originally embraced by John Sweller) but has since for various reasons been rejected by him and more resent research by those who really understand the theory. Intrinsic load is determined by the complexity of the task and the expertise of the learner. Extraneous load is all mental effort caused by the instruction. If it is unnecessary for or impedes learning (e.g., discovery, seductive details), then it should be avoided. If it is necessary and facilitates learning (e.g., worked examples, effective interaction of verbal and pictorial information) then it is germane to learning.

In light of this distinction, read the following as it reverts to a discussion of the Pyke et al. article as they wrote it.

Reducing extraneous load while balancing intrinsic and germane loads is critical for effective instruction. For instance, beginners benefit from worked examples that reduce intrinsic load, while advanced learners may thrive on independent problem-solving.

Bridging the Gap: A New Integrative Model

The authors argue that neither desirable difficulties nor cognitive load alone sufficiently addresses the complexities of learning. Their proposed model synthesises principles from both, incorporating insights from perceptual load theory (Lavie & Tsal, 1994[6]).

Disclaimer 4: Perceptual load theory has nothing to do with learning but rather with discerning between stimuli (the so-called cocktail party problem: how do people at a cocktail party select the conversation they are listening to and ignore the others?). It is a theory of attention and not of information processing and is imho worthless here.
Interesting here is that Lavie’s PhD supervisor, Yehoshua Tsal, the senior author on the original paper, distanced himself from the original article in 2013[7].

For Pyke et al., tis framework tailors task difficulty based on two factors: the complexity of the material (element interactivity) and the learner’s expertise and leads to the following key recommendations:

1. Low-Complexity Tasks:
   • Increase difficulty to enhance focus and retention.
   • Use testing and spaced retrieval to deepen cognitive engagement and strengthen memory traces.
2. High-Complexity Tasks:
   • Reduce difficulty to prevent cognitive overload.
   • Provide worked examples or scaffolding to guide learners.
3. Learner Expertise:
   • Novices benefit from structured, guided tasks with minimal intrinsic load.
   • Experts can handle—and benefit from—higher difficulty, as they rely on pre-existing schemas.

By adjusting difficulty dynamically (yes, no modern article is complete without adding the computer and AI!), this model ensures that learning is both effective and personalised.

My Conclusion

DUH! Anyone with any understanding of both desirable difficulties and cognitive load theory knows this. Furthermore, they also know that there’s no conflict between the two and that they fit together like a hand in a glove (O.J. Simpson notwithstanding).

---

[1] Pyke, W., Lunau, J., & Javadi, A. H. (2024). Does difficulty moderate learning? A comparative analysis of the desirable difficulties framework and cognitive load theory. Quarterly journal of experimental psychology (2006), doi: 10.1177/17470218241308143.

[2] The basis for this blog and the discussion of the Pyke et al. article is that evidence, and thus that explicit instruction is the basis and not ideology.

[3] Bjork, R. A. (1994). Memory and metamemory considerations in the training of human beings. In J. Metcalfe & A. Shimamura (Eds.), Metacognition: Knowing about knowing (pp. 185–205). MIT Press.

[4] Bjork, R. A., & Bjork, E. L. (2020). Desirable difficulties in theory and practice. Journal of Applied Research in Memory and Cognition, 9(4), 475–479. https://doi.org/10.1016/j.jarmac.2020.09.003

[5] Pyke et al. as is the case with many others, limit themselves to these three desirable difficulties, ignoring contextual interference and reduced feedback.

[6] Lavie, N., & Tsal, Y. (1994). Perceptual load as a major determinant of the locus of selection in visual attention. Perception & Psychophysics, 56, 183–197. doi: 10.3758/BF03213897

[7] Benoni, H., & Tsal, Y. (2013). Conceptual and methodological concerns in the theory of perceptual load. Frontiers in Psychology, 4, 53005. https://doi.org/10.3389/fpsyg.2013.00522