A course structured around mastering one skill type completely before moving to the next — blocked practice — produces smoother, more confident-feeling progress during each individual practice session than an alternative structure mixing several related skill types together within the same session — interleaved practice. Tested on retention and on transfer to new, unfamiliar problems some time after practice ends, learners who practiced with interleaving consistently outperform those who practiced with blocking, despite interleaving feeling harder and producing less encouraging results during the practice sessions themselves.
Why blocked practice feels more effective in the moment
Repeating the same skill or problem type multiple times in a row allows a learner to settle into a consistent approach and see steadily improving performance within that narrow, repeated context, producing a strong, immediate, and genuinely felt sense of mastery and progress — a felt experience that both learners and instructors tend to interpret as evidence the practice method is working well.
Why interleaving produces a worse feeling despite better actual learning
Mixing several related but distinct skill types together requires a learner to repeatedly reorient and select the appropriate approach for each new problem as it's encountered, rather than simply continuing an already-established approach from the previous repetition — this repeated reorientation is genuinely more cognitively effortful, produces more visible errors and slower apparent progress during practice, and consequently feels less effective, even in studies where it later demonstrates measurably better retention and transfer.
Why interleaving actually produces the stronger underlying learning
Interleaved practice requires learners to actively discriminate between different skill or problem types and select the correct approach for each specific case, building a more robust, flexible underlying skill at recognizing which approach applies to a new, unfamiliar situation — precisely the skill that blocked practice, where the correct approach is already known in advance because it's the same as the previous several repetitions, doesn't require and therefore doesn't build. This difference shows up specifically in transfer tasks, where learners are tested on new problem types not seen during practice, and is a core example of a broader category researchers call desirable difficulties — practice conditions that feel harder and produce worse in-session performance, while producing meaningfully better durable learning.
Why course design defaults toward blocked practice anyway
Blocked practice's smoother, more encouraging in-session progress makes it more appealing to both learners, who prefer the more comfortable experience, and instructors and course designers, who can point to visibly improving performance during the course itself as evidence of effective teaching — even though this visible in-session progress is a weaker predictor of genuine long-term retention and transfer than the harder, less encouraging interleaved alternative.
What this means for how courses should actually be structured
- Build interleaved practice — mixing related skill or problem types within a session — into course design, rather than defaulting to blocked practice because it feels more encouraging
- Communicate to learners explicitly that interleaved practice feeling harder is expected and consistent with it working better for long-term retention, to manage the discomfort this creates
- Evaluate course effectiveness based on delayed retention and transfer testing, not just in-session performance, since the two can favor opposite practice structures
- Reserve blocked practice specifically for the earliest stage of learning a genuinely new skill, where some initial repetition helps establish basic competence before interleaving is introduced
Interleaving is one of the more counterintuitive, well-replicated findings in learning science — the practice method that feels like it's working less well during the session itself is frequently the one producing meaningfully better learning once the session ends.