Does teaching movement transfer to cognition, achievement, or lifelong activity?

Verdict

This is the database's first clear no-effect verdict, and it is well earned: the transfer claims used to justify physical education are contradicted by the two highest-graded pieces of evidence in the cluster — an umbrella review with bias correction, and a large preregistered cluster RCT.

To be precise about what is and is not being denied: teaching movement skills demonstrably teaches movement skills. What fails is every claimed spillover — to cognition, to academic achievement, and to lifelong physical activity.

What the evidence shows

The strongest evidence is the most deflationary. Ciria's umbrella review is the cleanest summary of the exercise–cognition literature and the effect does not survive bias correction. Fit to Study is the trial that should have settled it favourably if the effect were real — 104 schools, 18,261 adolescents, preregistered — and it returned nulls across the board, including on fitness, which means the intervention did not even move its own proximal target.

The Stodden cascade does not hold. The theory that building motor competence in childhood produces lifelong physical activity is supported in about a quarter of the relevant longitudinal analyses (and 3% for object control), with only two intervention studies testing causality at all. What does survive is capacity-to-capacity linkage — motor competence ↔ fitness, and adiposity → motor competence.

Ludyga's "coordinative exercise is best" moderator is the steelman for a positive reading, and it is precisely the class of estimate that Ciria's bias correction dissolves.

Hereditarian-lens assessment

Risk: medium, and Zi & de Geus supply the likely explanation for the whole pattern. Motor competence is 55–58% heritable, fitness 65–67%, physical activity 29–37% — and there is no genetically informative study in the entire motor-competence→activity literature. The association is stronger cross-sectionally than longitudinally, which is the signature of familial confounding rather than a trainable causal cascade: children who are good at moving and children who stay active share causes.

One scoping correction the adversarial pass required: "no genetically informative study" applies to the cascade, not to motor skill generally — genetically informative motor-skill work does exist (see talent & trainability). The cascade is untested, not the domain.

Boundaries & what critics say

• This is not an argument against PE. Movement skill, enjoyment, and health are sufficient justifications; they simply have to be the actual justification.
• Acute vs chronic: some acute post-exercise cognitive effects exist; they are transient and do not constitute an achievement lever.
• FMS effects are acquisition-phase on the trained battery, mostly quasi-experimental, with delayed retention essentially unmeasured across the literature.

Practical guidance

• Teach movement skills directly and explicitly — they respond well to teaching.
• Stop justifying PE by academic spillover. It will not survive contact with a serious evaluation, and staking the programme on it makes PE vulnerable when the claim fails.
• Don't buy "brain-boosting" movement programmes (coordinative-exercise-for-cognition packages); that is the exact estimate bias correction removes.
• If lifelong activity is the goal, target it directly (habits, access, enjoyment) rather than assuming competence will cascade into it.

Open questions

• Whether any dose or modality of childhood movement produces durable activity habits is genuinely untested with adequate designs.
• Delayed retention of taught motor skills is unmeasured almost everywhere.