What the book covers

Chapter 1 reproduces Wilson and colleagues’ derivation and simulations and sets out their assumptions and limits; Chapter 2 discusses deliberate practice, “10,000 hours” and the related meta-analyses and replication; Chapter 3 discusses Hebbian plasticity, long-term potentiation and activity-dependent myelination, with a simplified cable model of g-ratio and conduction velocity; Chapter 4 discusses forgetting curves, the spacing effect and retrieval practice, with the ACT-R activation model and an SM-2 scheduler; Chapter 5 discusses interleaving and “desirable difficulties”; Chapter 6 the timing and grain of feedback. Chapter 7 reads this research beside the Analects, the Record of Learning in the Book of Rites and Zhu Xi’s method of reading, and explains that the classical authors were concerned with moral self-cultivation, not experiments in cognitive psychology. Chapter 8 offers a “four-step deliberate-practice checklist” and a pre-registered randomised controlled trial.

Claims corrected after checking

Several points in the author’s original framework needed correction or qualification after checking. First, the formula for the optimal error rate: the framework gave ½(1 − 1/√2) ≈ 0.1587, but ½(1 − 1/√2) ≈ 0.146; the paper’s result is ER* = ½ erfc(1/√2) = Φ(−1) ≈ 0.1587, an accuracy of about 84.13 %. Second, “85 %” holds only under the paper’s assumptions: with Laplacian noise the optimal error rate is about 18.4 %, with Cauchy noise 25 %; the authors themselves note that the result has yet to be generalised to multi-choice tasks and other learning algorithms, and that not every model has a “sweet spot” of difficulty (an ideal Bayesian learner’s posterior, for example, does not depend on the order of trials). Third, the paper discusses Metcalfe and colleagues’ “Region of Proximal Learning”; Vygotsky’s zone of proximal development is a related intellectual source, not the same theorem. Fourth, the domain figures from the deliberate-practice meta-analysis (games 26 %, music 21 %, sports 18 %, education 4 %, professions under 1 %) were confirmed; but “10,000 hours” is not a rule proposed in the 1993 paper. It is an estimate of the average practice accumulated by the best violin students by age 20, which Gladwell turned into a “rule” and which Ericsson himself explicitly criticised. Fifth, Chomiak and Hu’s 0.77 is a theoretical optimal g-ratio for the central nervous system once volume constraints are included, not a measured value. Sixth, on Zhu Xi’s method of reading: “follow the proper order and advance step by step; read until familiar and reflect with care” comes from Zhu Xi’s Essentials of Reading; the “six principles of Master Zhu’s method of reading” were distilled by later scholars from his sayings: his disciple Fu Guang compiled the first collection, but the six headings were set by Zhang Hong and Qi Xi in 1265–1266, so the framework’s attribution of the six principles to Fu Guang is corrected accordingly. Each chapter states such corrections where they arise, and they are collected in the Editorial Notes released with the book.

How to read the boxes

As in the other books of 5D Library, different kinds of statement are marked separately.

When the book applies concepts from machine learning, cable theory and dynamics (gradient descent, precision, conduction velocity, activation decay) to human learning, these are models and analogies: they help us see structure, but a person is not a perceptron and memory is not a formula.

About this edition

This is version 1.0.0, the first complete edition: a preface and eight chapters, in Chinese and English. The text, boxes and code were drafted with an AI writing tool from the author’s title, framework and notes; the author is responsible for reviewing and revising them and for checking every source. Corrections and suggestions of literature the author has missed are welcome through the 5D Library website.