One of the most widely cited findings in cognitive science — that mixing different subjects during study sessions produces better long-term retention than grouping them by topic — has just cleared a significant methodological hurdle. This matters because interleaved practice is embedded in educational curricula, skill training programs, and self-improvement apps worldwide, all built on research whose core measurement technique had never been properly scrutinized.
The original 2008 Kornell and Bjork study, and virtually all direct replications thereafter, measured learning using an identification task: participants were shown a painting and asked to name the artist. The problem is that identification tasks conflate genuine memory strength with response bias — the tendency to favor certain response options regardless of actual recall. A participant who happens to overuse "Seurat" as a guess will appear to know more Seurat than they actually do. In a direct online replication with 288 participants, the authors confirmed the interleaving advantage (d = 0.48, p < 0.001) but simultaneously documented substantial response biases, with some artist categories chosen at rates several times higher than others. They then ran a conceptual replication using a two-alternative forced-choice (2AFC) design — the gold standard for isolating true memory signal from decisional noise — and found the interleaving effect remained robust and comparable in magnitude.
This is genuinely reassuring science rather than paradigm-shifting revision. The finding validates a decade of classroom and laboratory applications built on the interleaving literature, while simultaneously improving the field's measurement standards. However, important caveats apply: the stimuli are artistic categories, not procedural or declarative academic knowledge, and generalizability to mathematics, language learning, or motor skills requires further 2AFC-based investigation. The study also does not address optimal interleaving ratios or whether the effect scales with longer retention intervals. For now, this represents a methodologically important confirmatory result that strengthens — rather than overturns — the evidentiary base for spaced, interleaved learning.