1. RNP
  2. Motor Learning
  3. Repeat or vary: what leaves a trace
  4. Blocked practice: why grouped sessions create the illusion of learning

Motor LearningSeries 8/10

SeriesRepeat or vary: what leaves a trace8 of 10

Blocked practice: why grouped sessions create the illusion of learning

Blocked practice boosts in-session performance and misleads you about what actually sticks. Why it happens, the two competing hypotheses, and when grouped order is still the right call.

LabO RNP·14 min read·reviewed on

Blocked practice: trial order, the parameter nobody sets

Hello to you, movement professional,

Blocked practice groups all the trials of one task before moving on to the next. It drives performance up during the session, and yet, in the lab, it produces poorer results on delayed tests than the same trials mixed together. This is the contextual interference effect, first demonstrated by Shea and Morgan in 1979. The grouped session looks successful because what it mainly measures is the ability to replay what was just played.

Take a session sheet like the many that circulate. Forty trials, four footwork situations, ten trials per situation, the four blocks back to back. The coach who wrote it knows exactly why: they want each situation to be settled before moving on, and they're right about the comfort that brings. The sheet also says something else, unintentionally. Volume has been set, and so have load and density. Order, on the other hand, was never treated as a parameter. It just ended up that way.

The objection comes before you've even finished saying it, and it deserves to be addressed right away: mixing tasks to make a session harder is exactly the kind of advice that lets you sabotage training and call it neuroscience. The objection is valid: making a session harder proves nothing in itself. It comes down to a single condition, to be set before you change anything: a delayed test, decided before the block, capable of proving you wrong. Without it, nobody can tell a demanding session from a poorly built one, least of all the end-of-session feeling.


Blocked, serial, random: three ways to order exactly the same trials

In blocked practice, the sequence looks like this: ten change-of-direction cuts to the right, then ten to the left, then ten backpedaling. Serial practice rotates the tasks in a fixed, predictable order, one trial each, and starts over. In random practice, the same tasks come up without the person being able to anticipate which one is next.

Total volume is identical in all three cases. So is the technical content. The only thing that changes is each trial's position in the sequence. Since Shea and Morgan (1979), the literature has called this contextual interference, and it is a training variable just like load, with the particularity that it costs nothing to manipulate and appears on no programming sheet.

The perspective offered by ecological motor learning shifts the unit of analysis from the outset: what gets trained is a system's ability to produce a fitting solution under changing conditions. The broader framework of this question, the dissociation between what you see in a session and what remains, and the way transfer of learning is measured, is laid out elsewhere. Here, the question is narrower and more operational: what exactly does blocking do to the system of the person you work with, and why is the session result such a poor indicator of it.


The RNP training space, free for life

Get access to RNP training, no strings attached.

Welcome! Introduce yourself

SecureNo cardNo password

By continuing, you agree to our Terms of Use and our Privacy Policy.

The illusion of competence: why the grouped session looks successful

During a block, session performance rises fast. The third trial is better than the first, the eighth is clean, and the set ends on a feeling of mastery that everyone in the room shares. The person feels the movement settle. You see a curve going up and verbal cues becoming unnecessary. Nothing in this scene looks like a problem.

What this rise measures is the system's ability to replay a plan it has just played. Between two closely spaced trials of the same task, the motor solution is still available, the parameters are still loaded, and producing the next movement costs very little. In the most common reading, the improvement seen within the block is largely due to the disappearance of that cost, and that cost is exactly what the person will have to pay on the field, where no task announces its arrival. This gap between what practice shows and what remains afterward goes beyond the question of trial order alone: Schmidt and Bjork (1992) found it across several training paradigms.

The discomfort of this finding hits the professional first. A grouped session produces satisfaction in the person you work with, and that satisfaction is what they use to judge the quality of your work. A mixed session produces visible misses, hesitations, a feeling of regression in the legs. You pay immediately for what you gain later, and the gap between those two moments is what makes this setting so hard to hold to in the real life of coaching someone.

What remains is to understand what mixing does to the system to justify that price.


Rebuilding the plan or contrasting it: two hypotheses that don't agree

Two explanations have been competing since the early 1980s. Neither has won, and honesty requires presenting them for what they are: two competing hypotheses consistent with the same results.

The first, proposed by Lee and Magill (1983) and known as the reconstruction hypothesis or forgetting hypothesis, starts from working memory. When another task is interleaved, the previous action plan is partly lost, and the system has to rebuild it instead of replaying it. This repeated reconstruction would engage the very processes that will be needed later, in real situations, when nobody announces what's coming.

The second, formulated by Shea and Zimny (1983) and known as the elaboration hypothesis, relies on the contrast between tasks rather than on forgetting one of them. When several tasks coexist in memory, the system compares them and builds representations that are more distinct from one another. The block offers nothing to compare and would produce less differentiated traces, which are therefore harder to retrieve when the context changes.

The first has a direct consequence in a sensorimotor reading. Rebuilding a plan means re-engaging the motor decision instead of short-circuiting it. It means re-issuing a command along with its efference copy, and therefore formulating a fresh prediction of the expected sensory consequences, then checking it against the actual feedback. Reafference only carries information if a prediction was made before it. A system that replays the same plan ten times makes a prediction that becomes accurate within a few trials, and an accurate prediction corrects nothing. The mismatch is the working material, and the block removes it.

These two hypotheses predict the same result in almost every published protocol. That's precisely why they still coexist, and why no practical decision should rest on one rather than the other.


What the block really trains: execution rather than selection

The movement the person produces is only the visible output of a longer chain. Before it, there's a situation that is read, a choice among several possible solutions, and a plan parameterized on the map of the body available at that moment. This complete chain, sensory input included, is what the sensorimotor loop refers to, and it cannot be cut into independent slices.

The block trains the end of the chain. Selection, on the other hand, is made once on the first trial, then kept. In the LabO RNP glossary, the motor decision refers to the pre-reflexive act that decides between solutions before consciousness has anything to say, and that is the level a predictable sequence puts to rest.

On a team-sport pitch or in a one-on-one duel, execution rarely falls short. The person knows how to make the cut; they've done it a thousand times. What's missing is choosing the right cut early enough, with the right pelvic orientation and the opposite shoulder available. That choice relies on the body schema, the implicit map of the body's position and possibilities, which is constantly updated and has to be reread at every change of situation. In a block, it is read once and the rest of the set relies on it. A mixed sequence makes it be reread on every trial.

Mixing is costly in the session because it works the most expensive part of the chain. That's also why it's so tempting to skip it.


Where grouped order is still the right choice, and there are such cases

None of this justifies mixing on principle. The contextual interference effect is robust in the lab, on sequencing, aiming and timing tasks, replicated since the late 1970s. In applied settings, it becomes smaller and far more scattered, to the point of disappearing in some protocols. The work that separates these two worlds, Brady's review (1998), his 2004 meta-analysis, and the review by Barreiros, Figueiredo and Godinho (2007) devoted to applied settings, finds a clear gap between them. No effect size is quoted here, deliberately: published values vary too much depending on the task, the delay and the participants' level for any single one to be transferable to your session.

Three conditions limit its use. The tasks being compared must be different enough in their organization, a moderator already noted by Magill and Hall (1990); otherwise there is nothing to discriminate and the effect fades. The person must already have a rough version of each of the patterns involved; otherwise mixing adds a load that produces nothing. The volume of practice must be sufficient for the effect to have time to appear, which rules out testing it over a handful of sessions.

The challenge point framework, proposed by Guadagnoli and Lee (2004), provides the general safeguard. It states that the functional difficulty of a task depends on the person's level, and that there is a zone in which a trial generates the most usable information. Below that zone, the trial teaches the system little. Above it, the person receives more information than they can process, and the surplus is lost. It is a theoretical framework, debated, that nobody knows how to quantify in a session, and it remains a good protection against the quickest reading of the topic, the one that consists of making everything harder for everyone.

With a beginner, or on a task whose basic structure isn't in place yet, grouped order is the correct starting point. The useful question then becomes when to move out of it.


Setting trial order without breaking the session

The switch from grouped to mixed is decided on an observable criterion, not on a calendar. One possible marker, to be taken as a field reading and never as a validated protocol: when the person produces the general shape of the pattern without verbal feedback, and that shape holds from one session to the next, the basic structure is in place, and the block has done its job. What follows is a matter of tuning selection.

Three levers act on this dimension, and they combine. The constraints-led approach changes the space of available solutions instead of dictating the solution, which forces a search on every attempt. Representative learning design ensures that the information available during practice is the information that will be available in the real situation. Otherwise you train a selection process built on cues that will no longer exist. Affordances are the possibilities for action that the situation offers to this particular person's body, and they are what variation leads the learner to explore.

Attentional focus remains the cheapest setting to adjust and the one most often forgotten. Two protocols that look identical on paper do not necessarily produce the same learning, depending on what the instruction points to. Wulf's 2013 review documented this effect, and its size has been debated since. Running a mixed sequence with a detailed internal-focus instruction risks losing part of the benefit, though this remains a field hypothesis rather than an established finding.

Then there is variation within a task, which is a different matter from the order between tasks. Repetition without repetition, a phrase borrowed from Bernstein (1967), means that no execution reproduces the previous one, and that this dispersion is part of learning. The glossary distinguishes Repetition without repetition from simple repeatability, which aims to reduce the dispersion of the output. A well-built block can be highly repeatable and still poor in learning. You don't read the two qualities in the same place.

One last decision, and the most uncomfortable one: announce the drop in performance before it happens. A mixed phase feels worse than a blocked phase. If the person discovers that drop without warning, they will read it as a mistake on your part. Announced in advance, it becomes an expected signal.


What a residual reflex changes in the solution space

In RNP, a primitive reflex that remains active imposes a tone pattern tied to a specific trigger. The ATNR (Asymmetric Tonic Neck Reflex) is triggered by head rotation, which engages both cervical proprioception and the labyrinth. The STNR (Symmetric tonic neck reflex) responds to cervical flexion and extension. The TLR (Tonic labyrinthine reflex) is triggered by the position of the head in space.

In every case, keep two things apart: what triggers a reflex, and the full set of systems that activate when it occurs. A reflex is defined by what triggers it. Its response can recruit far more than its input.

What this adds here fits in one sentence. If a tone pattern takes over as soon as the head turns, the solution space that variation can explore is restricted before the first attempt. A mixed sequence will then have the person circling inside a space that is already closed, and it will produce fatigue rather than learning. Read where it's dark, switch the light back on, then make it hold: the order of attempts is a tool for the third step, and it never replaces the first two. This is a reading framework consistent with known mechanisms. It is not an established experimental result.


What no one can promise you yet

The following are established facts: the contextual interference effect exists in the laboratory, blocked practice is superior during acquisition, and the ranking reverses on delayed tests in the reference protocols.

The following sit at an intermediate level of evidence: the size of the effect in applied settings, the choice between the reconstruction and elaboration explanations, and the operational definition of the challenge point.

The following are field extrapolations, and we present them as such: the criterion proposed above for moving from blocked to mixed practice, and the idea that a residual reflex restricts the solution space available to variation. Neither has been validated as a protocol.

A word on causality. Athletes from varied environments may transfer better, but that observation does not show that variation produces transfer. Total practice volume, the diversity of activities in childhood, and the selection carried out by clubs act on both sides at once. That common origin is enough to create the association even if no causal link exists.

To keep the claim testable, here is a falsifiable marker. Inserting an unrelated activity between two attempts of a blocked sequence forces the system to lose its plan without giving it another task to compare against. If the benefit of mixed practice comes from reconstruction, this interrupted blocked sequence should recover part of the gain on delayed tests. If it recovers nothing, then the contrast between tasks is what matters and elaboration is the better explanation. That would change how sets should be composed.

RNP is not a medical procedure. It complements health care and does not replace it.

Above all, a well-built block teaches you how well your athlete can replay what they just played.

👉 I want to understand why, not just what.


Frequently asked questions about blocked practice and contextual interference

What is blocked practice?

Blocked practice is a way of organizing a session that groups all attempts at one task before moving on to the next. It contrasts with serial practice, which rotates tasks in a fixed order, and with random practice, which rotates them in an order the person cannot anticipate. Volume and content are the same in all three cases. Only the position of the attempts changes.

Why does blocked practice give better immediate results?

When two attempts at the same task follow each other closely, the action plan stays available, so producing the next movement costs little. The improvement seen during the block largely reflects the disappearance of that cost. The cost returns as soon as situations follow one another without warning.

Should blocked training be abandoned?

No. It remains the right starting point as long as the basic structure of the pattern has not been acquired, and for tasks whose organization is still costly for the person. The decision is about when to move on from it, never about banning it.

What is the difference between random practice and variability within a task?

Random practice concerns the order between distinct tasks. Variability within a task concerns the dispersion of executions of the same task. Both influence learning through different pathways, and they combine. Confusing them leads you to believe you have mixed a session when you have only added dispersion to a blocked series.

How do you know whether mixing produced anything?

Before the block, set an assessment situation that has never been practiced and a delay without practice, then stick to both. The non-negotiable point is that the situation is chosen beforehand. Otherwise it will be chosen after the fact from among those that confirm what you hoped for.


Sources

Shea J.B., Morgan R.L. (1979). Contextual interference effects on the acquisition, retention, and transfer of a motor skill. Journal of Experimental Psychology: Human Learning and Memory, 5(2), 179-187.

Lee T.D., Magill R.A. (1983). The locus of contextual interference in motor-skill acquisition. Journal of Experimental Psychology: Learning, Memory, and Cognition, 9(4), 730-746.

Shea J.B., Zimny S.T. (1983). Context effects in memory and learning movement information. In R.A. Magill (Ed.), Memory and Control of Action (Advances in Psychology, vol. 12, pp. 345-366). Amsterdam: North-Holland.

Magill R.A., Hall K.G. (1990). A review of the contextual interference effect in motor skill acquisition. Human Movement Science, 9(3-5), 241-289.

Schmidt R.A., Bjork R.A. (1992). New conceptualizations of practice: common principles in three paradigms suggest new concepts for training. Psychological Science, 3(4), 207-218.

Brady F. (1998). A theoretical and empirical review of the contextual interference effect and the learning of motor skills. Quest, 50(3), 266-293.

Brady F. (2004). Contextual interference: a meta-analytic study. Perceptual and Motor Skills, 99(1), 116-126.

Barreiros J., Figueiredo T., Godinho M. (2007). The contextual interference effect in applied settings. European Physical Education Review, 13(2), 195-208.

Guadagnoli M.A., Lee T.D. (2004). Challenge point: a framework for conceptualizing the effects of various practice conditions in motor learning. Journal of Motor Behavior, 36(2), 212-224.

Wulf G. (2013). Attentional focus and motor learning: a review of 15 years. International Review of Sport and Exercise Psychology, 6(1), 77-104.

Bernstein N.A. (1967). The Co-ordination and Regulation of Movements. Oxford: Pergamon Press.


By the LabO RNP team


€0Your free RNP PathsStart