Imagine you’re learning something completely new. Do you want to start with clear instructions, or would you rather dive into a problem first and figure things out as you go? This is an ongoing debate in education: Should students first receive direct instruction (instruction-first), or should they struggle with problems before learning the correct approach (problem-solving-first)? A recent study put these two approaches to the test, with a twist—it also considered how much prior knowledge students have.

The researchers, Fiorella and Lemons, conducted two classroom experiments with undergraduates learning about noncovalent interactions in biochemistry. The first experiment involved introductory biology students with relatively low prior knowledge. The second experiment focused on biochemistry students with some background in the topic. Students were divided into two groups: one that received explicit instruction before problem-solving (I-PS) and one that attempted to solve problems before receiving instruction (PS-I). Afterwards, all students took tests measuring how well they could apply their learning to similar (near-transfer) and different (far-transfer) problems.

The results favoured the problem-solving-first approach for the introductory biology students. Even though these students initially struggled more, they performed better on near-transfer tests compared to those who received instruction first. This supports the idea of “productive failure”—the notion that struggling before instruction can make learning more effective. However, for the biochemistry students, the opposite was true. Those who received instruction first outperformed the problem-solving-first group, supporting cognitive load theory, which argues that learners should avoid excessive struggle when encountering complex material.

Interestingly, neither group showed significant differences on far-transfer problems, which require applying knowledge in novel situations. This suggests that while instructional sequences may influence immediate learning outcomes, they may not be enough to ensure deeper understanding that transfers beyond the classroom.

So, what does this mean for educators? The study suggests there’s no one-size-fits-all answer. If students have little prior knowledge, letting them struggle a bit before instruction might help them engage more deeply with the material. However, if they already have some foundation, direct instruction could be the more effective path. Either way, blindly applying one teaching method without considering students’ background knowledge might not yield the best results.

I have to admit this study got me puzzled a bit, because I didn’t expect these results. Maybe worth replicating!

Abstract of the study:

This study tested competing theories about the effectiveness of different instructional sequences for learners with different levels of prior knowledge. Across two classroom experiments, undergraduates learned about noncovalent interactions in biochemistry by either receiving explicit instruction before problem-solving (I-PS group) or engaging in problem-solving before explicit instruction (PS-I group). Then all students completed near- and far-transfer tests on the material. In Experiment 1, participants were introductory biology students (n = 367), who had relatively low prior knowledge of the topic. Results indicated that the PS-I group significantly outperformed the I-PS group on the near-transfer test, providing support for productive failure. In Experiment 2, participants were biochemistry students (n = 138), who had relatively higher prior knowledge of the topic. In contrast to Experiment 1, results indicated that the I-PS group significantly outperformed the PS-I group, providing support for cognitive load theory. Neither experiment showed significant effects of instructional sequences on the far-transfer test. Overall, the findings suggest the effects of instructional sequences on students with different levels of topic-specific prior knowledge may not be as straightforward as existing theories suggest.