Evidence for Variation #

Every substantive claim on the Variation page is checked against current research. Here is each claim, how well today’s evidence supports it, and the sources. The full, de-duplicated source list lives on the references page.

Supported · strong evidence — Practising a skill under varied, jumbled, less predictable conditions (contextual interference) tends to depress performance during practice while improving longer-term retention and transfer relative to constant, blocked practice.

The contextual-interference effect is a canonical desirable difficulty: Schmidt & Bjork’s integrative review documents that variable/random practice typically impairs acquisition-phase performance yet enhances retention and transfer, a pattern repeatedly observed across motor and verbal tasks. The direction is well established; magnitudes vary.

Sources: Schmidt & Bjork (1992), New conceptualizations of practice: Common principles in three paradigms suggest new concepts for training, Psychological Science — https://doi.org/10.1111/j.1467-9280.1992.tb00029.x · Soderstrom & Bjork (2015), Learning Versus Performance: An Integrative Review, Perspectives on Psychological Science — https://journals.sagepub.com/doi/10.1177/1745691615569000 · full reference ›

Supported · moderate evidence — Learners who practise several movement patterns in a random, mixed order perform worse during practice than those who practise them in blocks, yet do better on a later retention test.

Shea & Morgan (1979) is the foundational laboratory demonstration of the contextual-interference effect: random practice depressed acquisition but improved retention and transfer relative to blocked practice. The core finding has been replicated many times in the lab, though it is more fragile in applied/field settings and with some skills and populations.

Sources: Shea & Morgan (1979), Contextual interference effects on the acquisition, retention, and transfer of a motor skill, Journal of Experimental Psychology: Human Learning and Memory — https://doi.org/10.1037/0278-7393.5.2.179 · full reference ›

Supported · strong evidence — Varied and interfered practice is one of a family of ‘desirable difficulties’: conditions that make practice harder and depress in-session performance while strengthening durable learning, so how easy practice feels is a poor guide to how much is being learned.

Treating contextual interference and variable practice as desirable difficulties, and warning that current performance can mislead judgements of learning, is mainstream and well supported across the motor and cognitive literatures.

Sources: Schmidt & Bjork (1992), New conceptualizations of practice, Psychological Science — https://doi.org/10.1111/j.1467-9280.1992.tb00029.x · Soderstrom & Bjork (2015), Learning Versus Performance: An Integrative Review, Perspectives on Psychological Science — https://journals.sagepub.com/doi/10.1177/1745691615569000 · full reference ›

Mixed · moderate evidence — The benefit of varied/interleaved over constant/blocked practice is real but conditional: it is of roughly medium size on average and varies substantially with the task and materials, being large in some domains and small or absent in others.

Brunmair & Richter’s meta-analysis of 59 interleaving studies (the cognitive sibling of contextual interference) found an overall medium benefit with large heterogeneity strongly moderated by stimulus type. Within the motor-learning literature too, the contextual-interference effect is robust in controlled settings but inconsistent in applied contexts. The ‘reliable direction, variable magnitude’ framing is the accurate one; this corrects the original page’s overstated ‘will increase and will stay higher’.

Sources: Brunmair & Richter (2019), Similarity matters: A meta-analysis of interleaved learning and its moderators, Psychological Bulletin — https://doi.org/10.1037/bul0000209 · full reference ›

Supported · moderate evidence — Variation helps most when the variants are similar and confusable—when you must discriminate between cases and select the right response—and gives little advantage when the things mixed together are unrelated.

Stimulus similarity is the key moderator in Brunmair & Richter’s meta-analysis, with the largest benefits for confusable categories and negligible benefits for dissimilar/unrelated items, consistent with a discriminative-contrast account. The same logic underlies the page’s ‘keep it in context’ guidance.

Sources: Brunmair & Richter (2019), Similarity matters: A meta-analysis of interleaved learning and its moderators, Psychological Bulletin — https://doi.org/10.1037/bul0000209 · full reference ›

Mixed · moderate evidence — The advantage of varied/random over constant practice tends to be smaller, absent, or reversed for beginners and very early in learning, where high contextual interference can overwhelm rather than deepen learning; a common recommendation is to establish a basic stable skill first and increase variability as competence grows.

Skill level is a recognised moderator of the contextual-interference effect: benefits are most reliable once learners have some competence, and high interference can impair acquisition for novices or complex skills. The graduated ’low interference first, raise it later’ guidance is widely advised but the boundary conditions are still debated, so this is appropriately qualified rather than presented as settled.

Sources: Schmidt & Bjork (1992), New conceptualizations of practice, Psychological Science — https://doi.org/10.1111/j.1467-9280.1992.tb00029.x · Brunmair & Richter (2019), Similarity matters: A meta-analysis of interleaved learning and its moderators, Psychological Bulletin — https://doi.org/10.1037/bul0000209 · full reference ›

Supported · moderate evidence — Variations and interference should be ‘in context’—drawn from conditions likely to occur during or after training—because the difficulty only helps when it resembles what later performance will demand; arbitrary, unrelated difficulty does not.

That practice variability should sample the range of conditions transfer is wanted for follows directly from the variability-of-practice and transfer-appropriate-processing principles in Schmidt & Bjork and the broader transfer literature; benefit is greatest when practised variations match the target performance context. Direction is well supported.

Sources: Schmidt & Bjork (1992), New conceptualizations of practice, Psychological Science — https://doi.org/10.1111/j.1467-9280.1992.tb00029.x · full reference ›

Supported · moderate evidence — Contextual interference in motor-skill learning and interleaving in cognitive/verbal learning reflect the same underlying principle: less predictable, mixed practice depresses acquisition-phase performance but improves later retention and transfer.

The motor contextual-interference effect and cognitive interleaving are widely treated as expressions of one desirable-difficulty mechanism; both trade poorer acquisition for better delayed performance, though magnitudes vary by task and skill level. This justifies cross-linking the two pages.

Sources: Brunmair & Richter (2019), Similarity matters: A meta-analysis of interleaved learning and its moderators, Psychological Bulletin — https://doi.org/10.1037/bul0000209 · Soderstrom & Bjork (2015), Learning Versus Performance: An Integrative Review, Perspectives on Psychological Science — https://journals.sagepub.com/doi/10.1177/1745691615569000 · full reference ›