Exploring Before Instruction Backfires When Learning Tasks Are Too Complex

For decades, educators have been told that letting students wrestle with a problem before receiving formal instruction can prime them for deeper learning. The idea, variously known as productive failure, exploration before instruction, or inventing to prepare for future learning, has accumulated an impressive body of supporting evidence: students who first attempt to solve problems on their own often develop richer conceptual understanding and greater curiosity than students who sit through a lecture first. But a new large-scale study of undergraduate engineering students suggests that this popular recipe has a hidden ingredient that determines whether it works at all — the sheer complexity of the exploration task itself.

The research, published in Educational Psychology Review by Marci S. DeCaro, Ryan J. Patrick, Campbell R. Bego, and Angela K. Thompson of the University of Louisville, tested a prediction drawn from cognitive load theory: that exploratory activities only pay off when they impose a manageable burden on working memory. When the task forces learners to juggle too many interacting pieces of information at once, the benefits of exploration appear to evaporate — and may even reverse.

The theoretical backbone of the study is the concept of element interactivity, a central construct in cognitive load theory. Learning materials vary enormously in how many distinct elements — facts, rules, relationships, procedures — must be held and coordinated in working memory simultaneously. A single programming error with one obvious cause involves low element interactivity; a task requiring students to diagnose three different error types, each with multiple interacting features, involves high element interactivity. Because working memory can hold only a handful of elements at once, tasks with high interactivity can overwhelm novices before meaningful learning even begins. The researchers reasoned that exploration, which lacks the scaffolding of direct instruction, might be especially vulnerable to this overload.

To test the idea, the team ran three classroom-based experiments with first-year engineering students learning to recognize and fix common programming errors. In Study 1, involving 388 students, half the class received a lecture followed by a practice activity targeting three types of programming errors — the traditional instruct-first sequence. The other half completed the same activity before the lecture, in the explore-first order. Afterwards, all students completed a survey measuring interest and curiosity, along with a posttest assessing both basic facts and deeper conceptual understanding of the material.

The results were striking. Students who explored first actually scored lower on the posttest than those who received instruction first. They reported no more interest or curiosity than their instruct-first peers — a disappointment for advocates of exploration — and, crucially, they reported higher cognitive load. In other words, the exploration activity seemed to have consumed mental resources without delivering its promised motivational or conceptual dividends. The authors interpret this as suggestive evidence that the combined activity, covering three error types at once, exceeded what novices could productively process before any instruction.

Studies 2a and 2b, with 393 and 402 students respectively, then manipulated the key variable directly: the complexity of the exploration activity, operationalized through element interactivity. Study 2a kept the complex format, bundling all three error types into a single activity before the lecture, and essentially replicated Study 1 — explore-first students again underperformed on the posttest and carried a heavier cognitive load. Study 2b, however, split the same content into three simpler activities, each targeting one error type and each interleaved with a mini-lecture. The total content was identical; only the size of the conceptual units changed.

The outcome flipped. In Study 2b, exploring first boosted conceptual knowledge relative to the instruct-first sequence, and equalized performance on basic factual knowledge between the two orders. Explore-first students also reported higher interest and curiosity — the motivational signature that exploration research has long promised — although, notably, they still reported higher cognitive load. The pattern suggests that when exploration is broken into smaller conceptual chunks, the extra mental effort becomes productive rather than paralyzing: learners can discern the important conceptual features of each unit, generate meaningful ideas about them, and then use the subsequent instruction to refine those ideas.

The authors are careful to frame their findings as ecologically valid but suggestive, with effect sizes ranging from small to very small. These were real classrooms, not laboratories, which strengthens the practical relevance of the results but also introduces the noise inherent in authentic educational settings. Even so, the consistency of the reversal across Study 1 and Study 2a, and its disappearance under the simplified format of Study 2b, aligns closely with what cognitive load theory would predict. Prior work by Sweller, Kalyuga, and colleagues had already shown that problem-solving before instruction tends to hurt learning when element interactivity is high; the new study extends that logic to the specific case of exploratory pre-instruction activities, showing that the same task can succeed or fail depending entirely on how its conceptual units are sized.

The implications for instructional design are potentially far-reaching. Active learning has become a rallying cry in STEM education, with large meta-analyses linking it to improved performance in science and engineering courses. Yet this study suggests that not all active learning is created equal: an exploration activity that is too complex may actively harm conceptual learning while simultaneously failing to spark the curiosity it is meant to ignite. Designers of flipped classrooms, invention tasks, and productive-failure curricula may need to audit their activities for element interactivity, breaking multi-concept challenges into smaller exploratory episodes before learners face the full complexity of the domain.

There is also a subtler lesson about cognitive load itself. Explore-first students in Study 2b reported higher cognitive load even while learning more, indicating that load is not inherently bad — what matters is whether that load is spent on sense-making within manageable conceptual units or on futile attempts to coordinate too many unfamiliar elements at once. As the authors conclude, exploration activities with smaller conceptual units may help learners discern important features and deepen understanding, while complex ones may bury those features under a fog of overload. For a field still debating how much guidance to give and when, the message is refreshingly concrete: before you let students explore, count the elements they will have to juggle.

Subject of Research: The role of element interactivity and cognitive load in determining whether exploratory learning before instruction improves or impairs student learning

Article Title: Considering Cognitive Load When Exploring Before Instruction: The Potential Role of Element Interactivity

Article References: DeCaro, M. S., Bego, C. R., Thompson, A. K., & Patrick, R. J. (2026). Considering Cognitive Load When Exploring Before Instruction: The Potential Role of Element Interactivity. Educational Psychology Review, 38(1), Article 127. https://doi.org/10.1007/s10648-026-10223-7

Image Credits: AI Generated

DOI: 10.1007/s10648-026-10223-7

Keywords: exploratory learning, cognitive load theory, element interactivity, productive failure, instructional sequencing, engineering education, programming errors, conceptual knowledge, working memory, active learning, undergraduate students, curiosity