Teaching physical skills and athletic development

The program (what a school should actually do)

Physical skill acquisition is a domain where the evidence is humbler than the coaching literature suggests, and where this project's premise shows up unusually clearly: capacity is substantially heritable, practice amplifies rather than compresses individual differences, and transfer beyond the trained skill is close to absent.

Teach movement skills for their own sake.

• Explicit teaching of fundamental movement skills works on the skills taught (gross motor SMD ≈ 1.42, object control ≈ 0.63) — but that is acquisition-phase performance on the trained battery, and delayed retention is essentially unmeasured across the entire literature.
• Do not justify PE by claimed spillovers. Transfer to cognition is ~zero at scale (bias-corrected d ≈ 0.05, CI crossing zero; a preregistered 104-school RCT of 18,261 adolescents found nulls on maths, executive function, and fitness itself), and the motor-competence → lifelong-activity cascade is not supported. (evidence: motor competence & transfer)

Structure practice for retention, not for how the session looks.

• The core insight is real: schedules that depress performance during practice (varied/random order, spacing, faded feedback) often produce better durable learning. Anyone judging a schedule by in-session performance systematically picks the one that harms retention.
• But apply it with care — the contextual-interference effect is laboratory-bound and null in under-18s (applied settings −0.01 after outlier correction; under-18s −0.39). Spacing is real but shrinks sharply as tasks get complex. (evidence: practice scheduling)

Feedback: less is probably more, but the famous versions have deflated.

• Fade augmented feedback rather than giving it every trial — the demonstration is clean, though the wider literature does not reproduce it. Self-controlled feedback and the OPTIMAL-theory motivational claims collapse under publication-bias correction and should not be built on. Mental practice is real at roughly half its advertised size and does not replace physical practice. (evidence: feedback & mental practice)

Youth sport: delay specialization; distrust early selection.

• Early specialization predicts junior success while a later main-sport start plus more other-sport practice predicts adult world-class status — near-opposite profiles, so selecting on junior performance selects the wrong athletes.
• Specialization carries a real but modest injury cost that survives adjustment for training volume (ORs ≈ 1.4–1.9).
• The relative age effect is a calendar artifact of the cut-off date — an argument against chronological age-grouped talent identification, not a measure of ability. (evidence: youth specialization)

What to refuse to spend money or time on

• PE justified by academic or cognitive spillover — the effect is ~zero on rigorous testing.
• Fundamental-movement programs sold as producing lifelong physical activity — the cascade is unsupported.
• Self-controlled feedback / OPTIMAL-theory motivational packages — collapse under bias correction.
• Genetic talent-identification testing — expert consensus (co-signed by the HERITAGE principal investigator) is that it has no role and no child should be screened.
• Early talent selection on current performance — it measures birth month, maturity, and load.

The hereditarian bottom line for a founder

Physical capacity is heritable at roughly the level of cognitive ability — measured VO₂max h² ≈ 60% in twin designs, youth fitness and motor tests 52–79%, with no detectable shared-environment component anywhere. Household does not explain aerobic capacity.

Two consequences a school builder should internalise:

1. Practice amplifies differences rather than erasing them. In twins reared apart, heritability of a motor skill was already high at trial 1 and rose with practice, and the rate of learning was itself heritable — the same signature found in reading. Good coaching lets genetic potential express; it does not equalise athletes.
2. Heritable is not predictable. The molecular record is null (1,520 elite endurance athletes, zero replicated variants). Heritability describes variance in a population; it licenses no prediction about an individual child, and no screening programme. (evidence: talent & trainability)

Open questions a founder should watch

• Delayed retention is barely measured anywhere in the movement-skill literature — nearly all headline effects are acquisition-phase.
• Whether "trainability" (heritable differences in response to training) is a real trait at all is now contested by control-arm meta-analyses.