The Blocked–Interleaved Contrast
The standard laboratory version of this experiment is simple to describe. Learners receive several categories of problem — say, three types of mathematical calculation, or four types of bird species — and study them in one of two orders. In blocked practice, all the examples of category A come before category B, which comes before category C. In interleaved practice, the types are shuffled so consecutive problems rarely share a category. Both groups spend identical time with identical material. At test, the interleaved group reliably outperforms the blocked group, often by a substantial margin.
The effect has been replicated across domains: mathematics, natural-science classification, motor learning (such as practising different kinds of tennis shot), and inductive category learning. Its robustness is one reason it earns more confidence than some other results in cognitive science that have shrunk under replication pressure.
Why Blocked Practice Feels Superior
The paradox is that blocked practice generates a strong illusion of competence. When a learner works through ten instances of the same problem type in a row, each successive answer comes faster and more easily. That growing fluency registers as mastery. The judgements of learning people report after blocked practice are higher than after interleaved practice — and they are wrong. The smoothness is real; it is just not learning. The brain is retrieving a warm answer rather than reconstructing one from first principles.
Robert Bjork at the University of California, Los Angeles, frames this in terms of desirable difficulties: conditions that appear to impede performance during acquisition but produce stronger long-term retention and transfer. Interleaving is one of the clearest examples of the category. Short-term fluency and long-term retention are, in a sense, competing currencies.
When the problems within an interleaved set become too similar to distinguish — or too different to usefully compare — the advantage shrinks.
What Interleaving Forces the Brain to Do
The leading explanation is discrimination: because consecutive problems no longer share a category, the learner must identify which approach is appropriate before deploying it. Interleaving restores the selection step, which is precisely the step required in any real-world use of the knowledge.
Research from Washington University in St. Louis by researchers in Henry Roediger’s group, and from Purdue University, supports this discrimination account. The benefit depends on the learner having to make a choice, not simply on random ordering.
A secondary mechanism involves spacing: because interleaving necessarily inserts other material between two instances of the same type, it also spaces each category out over time. Some portion of the interleaving advantage may therefore be the spacing effect under a different experimental name. Disentangling the two is methodologically difficult, and most researchers treat them as partially overlapping rather than cleanly separate.
Where the Evidence Gets Complicated
The interleaving advantage is robust in laboratory studies but the effect size in applied settings — across a full course, with material learned over weeks — is more variable. One large meta-analysis found positive overall effects, but with meaningful heterogeneity: the benefit is larger when the categories are genuinely confusable, and smaller when they are so obviously different that discrimination requires no effort.
There is also a level-of-experience complication. Blocked practice may genuinely serve a complete beginner who has no representation of a category yet; shuffling types before the individual categories are distinct enough to discriminate may produce confusion rather than desirable difficulty. The benefit of interleaving appears to emerge once the learner has at least minimal fluency with each type in isolation. This boundary condition matters: the sequence of some blocking, then interleaving, has practical support even if pure blocking to mastery does not.
The core finding, though, is durable: smoothness during practice is a signal worth distrusting.