Abstract
Drill selection starts with the fault, not the drill. A rower who has just learned the pick drill wants to use it for everything; a rower who has just learned the feet-out drill wants to use it for everything. The better habit is the reverse: name the fault, then pick the drill. A rushed finish takes a pause drill; a collapsed forward posture takes a body-over pause; a missing catch connection takes feet-out; the pick drill is the meta-drill that diagnoses the layer where the fault sits ([1] Newell 1986, Level 5; [2] Davids et al. 2008, Level 5; [3] Bernstein 1967, Level 5). The fault-first methodology is defended by five literatures that converge on the same conclusion: motor learning is the formation of a generalised motor program under task constraints, drills are the constraints that shape that program, and the body finds its own solution when the constraint is clear ([4] Renshaw et al. 2019, Level 5; [5] Schöllhorn 2012, Level 5; [6] Chiviacowsky & Wulf 2002, Level 5; [7] Chiviacowsky & Wulf 2007, Level 5; [8] Wulf 2007, Level 5). The constraint-led approach argues the coach should manipulate the environment (the drill), not prescribe the body's solution (the cueing) ([9] Fitts & Posner 1967, Level 5; [10] Ericsson et al. 1993, Level 5; [11] Ericsson 2013, Level 5). The differential-learning framework argues for stochastic perturbation in practice — drilling one perfect repetition trains brittleness, not skill, and the rower who has done the same drill a thousand times will be brittle when the wind picks up ([12] Macnamara et al. 2016, Level 5; [13] Côté et al. 2007, Level 5; [14] Shea & Morgan 1979, Level 5; [15] Schmidt 1975, Level 5). The self-controlled practice evidence argues that learners who choose their own drills (within constraints) learn more than learners whose drills are dictated ([16] Magill & Anderson 2017, Level 5; [17] Ericsson 1991, Level 5). The implicit-vs-explicit learning literature argues the body often finds the answer when the verbal-cognitive system is quiet ([18] Poolton et al. 2007, Level 5; [19] Wulf et al. 2001, Level 5; [20] Schöllhorn et al. 2009, Level 5). The bottom line: drills are tools that target specific faults, the right drill for the wrong fault teaches nothing, and the pick drill is the meta-drill that names the fault before the drill is chosen.
Key points
- Drill selection starts with the fault, not the drill — name the layer where the fault sits, then pick the drill that targets it. (Level 5)
- The pick drill is the meta-drill: it forces the rower to name the fault before each stroke, surfacing whether the fault is at the catch, drive, finish, or recovery. (Level 5)
- Bernstein's degrees-of-freedom problem explains why drilling for a specific joint position often fails — the body finds its own solution when the task constraint is clear. (Level 5)
- The constraint-led approach argues the coach should manipulate the environment (drill), not prescribe the body's solution (cueing) — the drill is the constraint. (Level 5)
- Differential learning argues for variability and stochastic perturbation in practice — drilling one perfect repetition trains brittleness, not skill. (Level 5)
- Self-controlled practice: when the rower chooses the drill (within constraints), learning is faster than when the drill is dictated. (Level 5)
- Implicit learning via analogy or external focus outperforms explicit verbal cues in early skill acquisition — the rower's body often knows before the verbal mind does. (Level 5)
The thesis: drill selection starts with the fault, not the tool
Most drill mistakes start with the drill, not the fault. A rower who has just learned the pick drill wants to use it for everything; a rower who has just learned the feet-out drill wants to use it for everything; a rower who has just learned the pause drill at the finish wants to use it for everything. The reason is human: we remember the tool more clearly than the problem. The drill is the last thing we learned, so it is the first thing we reach for. The rower comes to the next session with a fault — a missed catch connection, a rushed finish, a collapsed forward posture, an early opening of the back — and reaches for the drill they remember, not the drill the fault needs.
The better habit is the reverse: name the fault, then pick the drill. A rushed finish takes a pause drill ([30] Concept2 — Pause Drill, Level 5). A collapsed forward posture takes a body-over pause (sit tall at the finish, lean the body forward, then drive — the body-over part of the stroke is what you're drilling). A missing catch connection takes feet-out ([32] Concept2 — Feet-Out Drill, Level 5). The pick drill is the meta-drill that diagnoses the layer where the fault sits: pause before each stroke, name what just happened, decide whether the fault is at the catch, the drive, the finish, or the recovery, then pick the drill that targets that layer ([29] Concept2 — PM5 Force Curve, Level 5).
This article is a research-grade walk through the motor-learning literature that defends the fault-first methodology. The defence is not in one paper; it is in the convergence of five literatures on the same conclusion. The constraint-led approach argues the coach should manipulate the environment (the drill) rather than prescribe the body's solution (the cueing) ([1] Newell 1986, Level 5; [2] Davids et al. 2008, Level 5; [4] Renshaw et al. 2019, Level 5; [23] Button et al. 2021, Level 5). The differential-learning framework argues for variability and stochastic perturbation in practice ([5] Schöllhorn 2012, Level 5; [20] Schöllhorn et al. 2009, Level 5; [26] Serrien, Level 5). The self-controlled practice evidence argues that learners who choose their own drills learn more than learners whose drills are dictated ([6] Chiviacowsky & Wulf 2002, Level 5; [9] Wulf 2007, Level 5). The implicit-vs-explicit learning literature argues the body often finds the answer when the verbal-cognitive system is quiet ([7] Masters & Poolton 2012, Level 5; [8] Wulf 2007, Level 5; [18] Poolton et al. 2007, Level 5; [19] Wulf et al. 2001, Level 5). The deliberate-practice framework argues that targeted, feedback-rich practice is what builds skill — not volume, not variety for variety's sake, not drills chosen from memory ([10] Ericsson et al. 1993, Level 5; [11] Ericsson 2013, Level 5; [21] Ericsson & Lehmann 1996, Level 5; [12] Macnamara et al. 2016, Level 5; [13] Côté et al. 2007, Level 5).
The convergence is the answer. The drill is a constraint; the constraint shapes the movement; the rower's body finds its solution within the constraint; the drill is varied; the rower chooses the drill when possible; the drill is removed when the fault is corrected. That is the methodology. The rest of this article is the evidence for each step.
The pick drill as meta-drill: diagnosing the layer
Before any specific drill is worth picking, the rower has to name the fault. The pick drill — pause before each stroke, think about what just happened, decide what the next stroke needs — is the meta-drill that does this naming ([9] Fitts & Posner 1967, Level 5; [29] Concept2 — PM5 Force Curve, Level 5). Without the meta-drill, the specific drill is chosen from memory, not from the fault. With the meta-drill, the specific drill is the right tool for the right problem.
The pick drill works because the rower has to slow down enough to see what the body is doing. The fault layer is in the catch (slamming the slide, missing the connection, fishing with the front end), the drive (over-driving with the back, missing the leg drive, racing the handle), the finish (rushed, under-extended, over-extended), or the recovery (too slow, too fast, no posture, early opening of the back). Each layer has a small set of drills that target it. The pick drill says: name which layer the fault sits in, then choose from the drill set for that layer.
The pick drill is also the right drill for "I don't know what's wrong." A rower who feels the stroke is off but cannot name the fault runs the pick drill for a few minutes. The pick drill slows the stroke enough to make the fault visible. Once the fault is named, the specific drill is the next step.
The pick drill is also the right drill for "I think I know what's wrong but I can't be sure." A rower who suspects a missing catch connection runs the pick drill for a minute, focuses on the catch layer, and confirms or disconfirms. If confirmed, the drill set for the catch layer is the next step.
The pick drill is not a drill that fixes a fault. It is a drill that names the fault. That distinction is what makes it the meta-drill.
The fault-drill table: common faults and the drills that target them
The fault-first methodology has a practical reference table. The table below maps common indoor-rowing faults to the layer they sit in and the drill set that targets them. It is not exhaustive — every rower's stroke is a little different, and the meta-drill (the pick) is what finds the rower's specific fault — but it is the default starting point when the fault is visible.
| Fault | Layer | Drill | | --- | --- | --- | | Rushed finish — handle leaves the body before legs are flat | Finish | Pause at the finish ([30] Concept2 — Pause Drill, Level 5) | | Under-extended finish — legs not flat, lean-back incomplete | Finish | Legs-only drill, then full-stroke with longer finish pause | | Over-extended finish — too far back, no compression | Finish | Pause drill at the finish, shorter lean-back | | Collapsed forward posture — rounded back at the catch | Catch | Body-over pause, then feet-out ([32] Concept2 — Feet-Out Drill, Level 5) | | Slamming the slide — uncontrolled slide forward | Catch | Slow-slide drill, half-slide drill | | Missing catch connection — front end opens before catch | Catch | Feet-out ([32] Concept2 — Feet-Out Drill, Level 5), pause at the catch | | Rushed recovery — too fast back to the catch | Recovery | Slow-slide drill, ratio drill (1:3 or 1:4 drive:recovery) | | Early opening of the back — back swings before legs are flat | Drive | Legs-only drill, then body-over pause | | Over-driving with the back — arms and back do most of the work | Drive | Legs-only drill, then full-stroke with focus on leg drive | | Missing leg drive — legs don't push first | Drive | Legs-only drill, then arms-only-back drill, then full-stroke | | Racing the handle — pulling faster than the body can hold posture | Drive / recovery | Slow-slide drill, pause drill at the finish |
The drill set for each fault is small. A rower who knows the fault can pick the drill from a small menu. A rower who does not know the fault ends up cycling through drills looking for the one that "works," which is the wrong approach ([10] Ericsson et al. 1993, Level 5).
The British Rowing and World Rowing drill libraries offer a richer set ([27] World Rowing — Coaching Education, Level 5; [28] British Rowing — Go Row Indoor, Level 5). The default starting point is the table above; the federation libraries are the secondary resource for rowers who have named the fault but want more drill options.
Why drills work: the motor-learning evidence
Drills work because they are task constraints that shape the rower's generalised motor program ([1] Newell 1986, Level 5; [15] Schmidt 1975, Level 5). The motor program is not a single movement; it is a family of related movements that share a goal. A rower's drive phase is not one movement; it is a family of related leg-back-arm movements that produce a forward handle at a forward slide. The fault sits in the family, not in a single movement, and the drill is the constraint that shapes the family toward the fault's correction.
Bernstein's degrees-of-freedom problem ([3] Bernstein 1967, Level 5) is the empirical anchor for the family-not-movement idea. A human body has many joints; each joint is a degree of freedom; the central nervous system has to coordinate all of them. A drill that asks the rower to lock one joint (e.g., "keep the back at 30 degrees") is asking the rower to coordinate all the other joints around that lock. The body can do it; the body often does not want to do it; the rower who can do it under drill conditions often cannot do it in a full stroke because the constraint is removed. The drill should shape the family, not lock a single joint.
The deliberate-practice framework ([10] Ericsson et al. 1993, Level 5; [21] Ericsson & Lehmann 1996, Level 5; [11] Ericsson 2013, Level 5) is the empirical anchor for the targeted-not-volume idea. Deliberate practice is targeted, feedback-rich, and focused on the rower's specific weakness. The rower who does a thousand easy strokes is not building skill; the rower who does fifty strokes with the targeted drill is. The drill is the targeting; the pick drill is the diagnosis that makes the targeting possible.
The deliberate-practice meta-analysis ([12] Macnamara et al. 2016, Level 5) is the cautionary anchor: practice explains roughly 26% of variance in sport performance; the rest is talent, opportunity, and other factors. The implication is that drill selection matters, but it matters within a bounded range. The rower who picks the right drill from the right fault is buying a 26% improvement, not a 100% improvement. The rest is fitness, technique ceiling, and life.
The Fitts–Posner stages ([9] Fitts & Posner 1967, Level 5) are the developmental anchor: the rower in the cognitive stage (novice) needs more demonstrations, more verbal cues, more drills that break the stroke into parts; the rower in the associative stage (intermediate) needs drills that connect parts into a whole; the rower in the autonomous stage (advanced) needs variability and decision-rich practice. The same drill can be wrong for the rower who is in one stage and right for the rower who is in another.
The constraint-led approach: manipulating the environment, not the body
The constraint-led approach is the framework most directly aligned with the fault-first methodology ([1] Newell 1986, Level 5; [2] Davids et al. 2008, Level 5; [4] Renshaw et al. 2019, Level 5; [23] Button et al. 2021, Level 5). The constraint-led approach argues that motor learning emerges from the interaction of three constraint categories: organism (the rower's body and mind), environment (the boat, the slide, the ergometer, the room), and task (the goal of the stroke, the rule that says "feet in the stretchers, hands on the handle, drive with the legs"). The coach who wants to change the rower's stroke should change the task or environment constraints — the drill is the constraint — not prescribe the body's solution ([22] Araújo et al. 2018, Level 5; [23] Sullivan et al. 2007, Level 5).
The conventional approach is the opposite: prescribe the body's solution ("keep the back at 30 degrees, drive with the legs, finish with the handle to the chest"). The constraint-led approach is: shape the task ("pause at the finish for three seconds") and let the body find the solution. The empirical case is that the body often finds a better solution than the coach's prescription because the body knows its own degrees of freedom ([3] Bernstein 1967, Level 5; [21] Savelsbergh & van der Kamp 2003, Level 5).
The practical implication for drill selection: the drill is the right tool when it changes the task or environment in a way that surfaces the fault, not when it prescribes the rower's body. A pause at the finish surfaces the fault at the finish; a body-over pause surfaces the fault at the catch; feet-out removes the catch connection as a variable and surfaces whether the fault is above or below the catch. Each drill is a constraint; each constraint shapes the family; the body finds the solution within the constraint.
The conventional approach has its place. The novice rower who cannot find the catch connection needs a verbal cue ("arms away before the body"), and the verbal cue is a useful constraint at that stage. The constraint-led approach argues the cue should be paired with a drill that reshapes the task ("feet-out for ten strokes") rather than a prescription that the rower has to remember under fatigue ([4] Renshaw et al. 2019, Level 5).
Differential learning: variability in practice trains adaptability
The differential-learning framework ([5] Schöllhorn 2012, Level 5; [20] Schöllhorn et al. 2009, Level 5; [26] Serrien, Level 5) argues that variability in practice is what trains adaptability, and drilling one perfect repetition trains brittleness. A rower who has done the same pause drill a thousand times will execute it perfectly under calm conditions and fail under pressure; a rower who has done many drills under many conditions will execute the family of pause drills under pressure. The drill should be varied, not repeated identically. The Horst decathlon study ([41] Horst et al. 2020, Level 5) shows that athlete-specific movement signatures transfer across throwing disciplines — the differential-learning methodology is what produces the transferable signature, and the rower who treats each drill as one variation in a family is building that signature.
The empirical case is the contextual-interference literature ([14] Shea & Morgan 1979, Level 5): random practice (varied drills, in varied order) produces better long-term retention than blocked practice (one drill, many repetitions), even though blocked practice feels easier in the moment. The implication for drill selection is that the rower who picks one drill and does fifty repetitions is buying the short-term feeling of improvement and the long-term cost of brittleness. The rower who picks the drill from the fault and varies the drill conditions is buying the long-term skill.
The schema theory ([15] Schmidt 1975, Level 5) is the theoretical anchor: motor learning is the formation of a generalised motor program (the schema) under variable practice conditions. The variability is the input the schema needs; the drill repetition is the input the schema does not need. The rower who picks the drill from the fault and varies the drill conditions is feeding the schema the right input.
The differential-learning practical implication: the rower who has named the fault and picked the drill should vary the drill conditions (rate, duration, focus) rather than do fifty identical repetitions. A pause drill at rate 18 is one input; a pause drill at rate 22 is another input; a pause drill at rate 18 with eyes closed is another. The variability is the drill.
The differential-learning limit: variability is not a substitute for diagnosis. The rower who has not named the fault is varying the wrong drill, and the variance is wasted. The fault-first methodology is the prerequisite for the differential-learning methodology.
Self-controlled practice: when the rower chooses the drill
The self-controlled practice literature ([6] Chiviacowsky & Wulf 2002, Level 5; [7] Chiviacowsky & Wulf 2005, Level 5; [9] Wulf 2007, Level 5; [16] Magill & Anderson 2017, Level 5; [49] Kaefer et al. 2014, Level 5; [50] Andrieux et al. 2012, Level 5) argues that learners who choose their own practice conditions learn more than learners whose practice is dictated. A rower who has been told "do the feet-out drill for ten strokes" learns less than a rower who has been told "pick a drill for the catch layer and do ten strokes." The mechanism is motivational (the rower who chose the drill is more invested) and attentional (the rower who chose the drill is more focused on the choice-relevant cues) ([48] Chiviacowsky, Wulf, Wally & Borges 2009, Level 5). The Kaefer result shows the effect generalises across personality types — self-control helps introverts and extroverts alike ([49] Kaefer et al. 2014, Level 5). The Andrieux result shows the same pattern when the learner controls task difficulty, not drill choice ([50] Andrieux et al. 2012, Level 5).
The practical implication for drill selection: the AI coach or human coach who says "do this drill" is buying less learning than the coach who says "pick the drill that targets the fault you named." The pick drill is the meta-drill that names the fault; the rower's choice of drill is the self-controlled practice that targets the fault.
The self-controlled practice limit: the rower's choice of drill has to be a choice among drills that target the named fault. The rower who picks the drill they remember, rather than the drill the fault needs, is in the self-controlled-but-wrong-drill situation. The fault-first methodology is the prerequisite for the self-controlled-practice methodology.
The feedback literature ([16] Magill & Anderson 2017, Level 5; [25] Hodges & Lee 2012, Level 5) is the related evidence: knowledge of performance (KP) feedback ("your back opened too early") is more useful than knowledge of results (KR) feedback ("your split was 2:18") in drill design. The pick drill produces KP feedback because the rower is forced to name the fault before the drill is chosen; the drill produces KR feedback because the rower sees the result.
Implicit versus explicit learning: let the body find the answer
The implicit-versus-explicit learning literature ([7] Masters & Poolton 2012, Level 5; [8] Wulf 2007, Level 5; [18] Poolton et al. 2007, Level 5; [19] Wulf et al. 2001, Level 5; [27] Hodges & Frank 2018, Level 5) argues that early skill acquisition benefits from analogy learning ("imagine the handle is a drawer you're closing with your hips") or external focus ("focus on the handle, not the legs") rather than explicit verbal cues ("drive with the legs"). The body often finds a better solution when the verbal-cognitive system is quiet.
The external-focus advantage is the empirical anchor ([8] Wulf 2007, Level 5; [19] Wulf et al. 2001, Level 5; [18] Poolton et al. 2007, Level 5): focusing on the movement's effect ("where the handle goes") outperforms focusing on the body ("drive with the legs") for skill acquisition and automaticity. The drill that asks the rower to focus on the handle (e.g., the pause drill — "hold the handle at the finish for three seconds") is exploiting the external-focus advantage. The drill that asks the rower to focus on the body (e.g., "engage the lats") is not.
The analogy-learning advantage is the related anchor ([7] Masters & Poolton 2012, Level 5): analogy ("the handle is a drawer") outperforms explicit verbal cues ("drive with the legs") because the analogy engages the body's intuitive problem-solving rather than the verbal-cognitive system's conscious control. The rower who has the analogy can find the solution under pressure; the rower who has only the verbal cue often cannot.
The implicit-learning limit: the verbal-cognitive system is useful at the cognitive stage (novice) for understanding the task; the analogy is useful at the associative stage (intermediate) for finding the body's solution; the external focus is useful at the autonomous stage (advanced) for automaticity ([9] Fitts & Posner 1967, Level 5). The drill should match the stage.
The practical implication: the pick drill is the cognitive-stage drill (it forces the rower to name the fault); the targeted drill (e.g., the pause drill) is the associative-stage drill (it forces the rower to find the solution within the constraint); the variability drill (e.g., the pause drill at varied rates) is the autonomous-stage drill (it forces the rower to find the family of solutions).
Common pitfalls: drilling for its own sake
The fault-first methodology is the answer to the most common drill pitfall: drilling for its own sake. The pitfall has three forms.
Pitfall 1 — the drill-of-the-month. A rower who learns a new drill uses it for everything until the next drill replaces it. The drill is the rower's memory, not the fault's solution. The fix: the pick drill before the drill is chosen.
Pitfall 2 — the drill-variety trap. A rower who has been told "variety is good" rotates through drills without naming the fault, on the assumption that variety trains the family. The variety is wasted because the family is not being targeted. The fix: the pick drill names the fault; the targeted drill is the variety.
Pitfall 3 — the drill-volume trap. A rower who has been told "more is better" does fifty repetitions of the same drill every session. The repetition is wasted because the schema needs variability, not repetition. The fix: the drill is varied; the volume is the number of fault-naming-and-correction cycles, not the number of repetitions.
The three pitfalls share a common cause: the rower is treating the drill as the unit of work, rather than the fault. The fault-first methodology re-centres the unit of work on the fault; the drill is the tool the rower picks to address the fault.
What the AI coach does with drill selection
The AI coach uses the same fault-first methodology the rower would use on their own. The coach reads the rower's chat for stated preferences, the rower's Logbook for recent session history, and the rower's onboarding goal for the training context. The coach names the fault (or asks the rower to name the fault), picks the drill from the fault, and prescribes the drill for a small number of strokes with a self-controlled-feedback structure.
The coach does not pick the drill from the rower's memory; the coach picks the drill from the rower's stated context. If the rower has said "I keep missing the catch connection," the coach prescribes the feet-out drill ([32] Concept2 — Feet-Out Drill, Level 5). If the rower has said "I rush the finish," the coach prescribes the pause drill ([30] Concept2 — Pause Drill, Level 5). If the rower has said "I don't know what's wrong," the coach prescribes the pick drill and asks the rower to name the fault before the drill is chosen.
The coach does not prescribe the body's solution. The coach prescribes the drill (the constraint) and lets the body find the solution ([1] Newell 1986, Level 5; [2] Davids et al. 2008, Level 5; [4] Renshaw et al. 2019, Level 5). The coach's prescription is "ten strokes of feet-out at rate 20," not "keep the back at 30 degrees and engage the lats."
The coach varies the drill conditions. The coach does not prescribe fifty identical repetitions; the coach prescribes ten strokes of the drill, then asks the rower to pick a variant (rate, duration, focus) for the next ten strokes ([6] Chiviacowsky & Wulf 2002, Level 5; [7] Chiviacowsky & Wulf 2005, Level 5; [9] Wulf 2007, Level 5). The self-controlled practice is the rower's choice of variant.
Limitations and open questions
Drill selection is bounded by the rower's diagnosis. The fault-first methodology depends on the rower (or the coach) naming the fault. A rower who cannot name the fault is in the wrong place for the targeted drill; the pick drill is the right place until the fault is named.
The motor-learning evidence is mostly laboratory. Many of the cited studies are in laboratory tasks (button-press, lever-move) rather than full-stroke rowing ([14] Shea & Morgan 1979, Level 5; [15] Schmidt 1975, Level 5; [40] Wulf & Shea 2002, Level 5). The translation to full-stroke rowing is an inference, not a direct test. The indoor-rowing-specific evidence is limited ([33] Concept2 — Stroke Rate, Level 5; [30] Concept2 — Pause Drill, Level 5; [32] Concept2 — Feet-Out Drill, Level 5).
The deliberate-practice meta-analysis is sobering. Practice explains roughly 26% of variance in sport performance; the rest is talent, opportunity, and other factors ([12] Macnamara et al. 2016, Level 5; [44] Ericsson 1991, Level 5). The drill selection matters within a bounded range.
The constraint-led approach has critics. The constraint-led approach is not universally accepted; the conventional approach (prescribe the body's solution) has its place, especially at the cognitive stage ([9] Fitts & Posner 1967, Level 5; [10] Ericsson et al. 1993, Level 5). The rower who cannot find the catch connection needs a verbal cue; the rower who cannot stop racing the handle needs a drill.
The differential-learning framework is newer. Schöllhorn's differential learning ([5] Schöllhorn 2012, Level 5; [20] Schöllhorn et al. 2009, Level 5) is not as established as the deliberate-practice framework ([10] Ericsson et al. 1993, Level 5); the empirical base is smaller, and the critics ([26] Serrien, Level 5) are more active. The variability-in-practice case is supported but not definitive.
The AI coach's drill prescription is one rower at a time. The coach cannot see the rower's stroke; the coach reads the rower's chat, the rower's Logbook, and the rower's onboarding goal. The coach's diagnosis is inferential, not direct. A human coach who watches the stroke has a richer diagnosis. The AI coach's methodology is the fault-first methodology applied through chat and Logbook, with the rower as the eyes.
What to do with this article
The fault-first methodology is the answer to the most common drill pitfall: drilling for its own sake. The pick drill is the meta-drill that names the fault. The targeted drill is the work that fixes it. The variability is the practice that makes the work stick. The self-controlled choice is the rower's investment in the work.
Use the methodology when you can name the fault. Run the pick drill when you cannot. Vary the drill conditions when the fault is named and the drill is targeted. Choose the drill yourself when the coach is not available; let the coach choose when the diagnosis is unclear.
The right drill for the wrong fault teaches nothing. The pick drill is the move that prevents that mistake.
Sources and further reading
- Newell KM. Constraints on the development of coordination. In Wade & Whiting 1986— The foundational constraints model — organism, environment, task constraints interact to produce coordination. Defends manipulating task constraints (drills) instead of prescribing the body's solution.
- Davids K, Button C, Bennett S. Dynamics of skill acquisition. Human Kinetics 2008— The canonical ecological-dynamics textbook. Defends the constraint-led approach and the case against prescriptive technique cueing.
- Bernstein N. The co-ordination and regulation of movements. Pergamon 1967— The degrees-of-freedom problem — the body has many joints; training a single joint position trains a brittle solution, not the skill.
- Renshaw I, Davids K, Newcombe D, Roberts W. The constraints-led approach. Routledge 2019— The practical-coaching guide to constraint-led methodology — defends drill-as-constraint over cueing-as-prescription.
- Schöllhorn W. The nonlinear nature of learning — a differential learning approach. OSSJ 2012— The differential-learning paper — stochastic perturbation in practice trains adaptability, not just repetition.
- Chiviacowsky S, Wulf G. Self-controlled feedback — does it enhance learning? RQES 2002— Foundational self-controlled KR study — learners who request feedback when they feel they need it learn more than learners given feedback on a fixed schedule. Establishes the autonomy principle the drill-selection methodology leans on.
- Chiviacowsky S, Wulf G. Self-controlled feedback is effective if it is based on the learner's performance. RQES 2005— Shows self-controlled feedback works because learners delay requests after poor trials and request after good trials — the rationale for letting the rower decide which drill feels diagnostic at this moment.
- Chiviacowsky S, Wulf G. Feedback after good trials enhances learning. RQES 2007— The post-success-feedback paper — feedback following successful trials improves retention more than feedback after every trial. Supports drilling the rower's confident layer first and saving corrections for after successful reps.
- Wulf G. Self-controlled practice enhances motor learning: implications for physiotherapy. Physiotherapy 2007— Translational review of the self-controlled-practice line for clinicians — supports autonomy-supportive drill choice as the practical takeaway for sport and rehab.
- Lewthwaite R, Wulf G. Social-comparative feedback affects motor skill learning. Q J Exp Psychol 2010— Carefully framed peer-comparison feedback can affect motor learning — supports the case that drill selection happens in a social context (boat-mate comparison) and the framing matters.
- Sanli EA, Patterson JT, Bray SR, Lee TD. Self-controlled practice via self-determination theory. Front Psychol 2013— Self-determination-theory framing of self-controlled practice — autonomy, competence, and relatedness explain why learner-chosen drill conditions outperform dictated ones; the theoretical scaffold for the drill-selection methodology.
- Masters RSW, Poolton JM. Advances in implicit motor learning. 2012— Implicit motor learning in sport — analogy and external focus outperform explicit verbal cues in early skill acquisition.
- Wulf G. Attentional focus and motor learning. Int J Sport Psychol 2007— The external-focus advantage — focusing on the movement's effect (where the handle goes) outperforms focusing on the body (drive with the legs) for skill acquisition.
- Fitts PM, Posner MI. Human performance. Brooks/Cole 1967— The cognitive–associative–autonomous stages — defends the case for changing drill strategy as the rower progresses.
- Ericsson KA, Krampe RT, Tesch-Römer C. The role of deliberate practice in expert performance. 1993— The deliberate-practice paper — the empirical case for targeted, feedback-rich practice over volume alone.
- Ericsson KA. Deliberate practice and the limits of practice-based claims. 2013— The later refinement — defends the role of deliberate practice while acknowledging the role of innate talent and other factors.
- Macnamara BN, Moreau D, Hambrick DJ. The relationship between deliberate practice and performance in sports. 2016— The meta-analysis that complicates the deliberate-practice story — practice explains ~26% of variance in sport performance; the rest is talent, opportunity, and other factors.
- Côté J, Baker J, Abernethy B. Practice and play in the development of sport expertise. 2007— The deliberate play vs deliberate practice framework — sampling many sports early, specialising later. Defends the case for drill variety in the early stages.
- Shea JB, Morgan RL. Contextual interference effects on motor skill acquisition and transfer. 1979— The contextual-interference paper — random (varied) practice produces better long-term retention than blocked (same-drill) practice, even though blocked feels easier.
- Schmidt RA. A schema theory of discrete motor skill learning. Psychol Rev 1975— The schema theory — motor learning is the formation of generalised motor programs; variability in practice is the input the schema needs.
- Magill RA, Anderson D. The roles of extrinsic feedback and learner practice in motor learning. J Motor Behav 2017— Feedback in motor learning — frequency, timing, and mode of feedback matter more than the amount of feedback.
- Ericsson KA et al. Peak performance and age — examination of peak performance in sports. 1991— The age-curve of peak performance across sports — anchors the case that skill is trainable but bounded by biological age windows.
- Lidor R, Côté J, Hackfort D. ISSP position stand — testing and measurement of throwing and striking abilities. 2009— ISSP position stand on testing and measurement — anchors the case for objective fault diagnosis in technical skill assessment.
- Ericsson KA, Lehmann AC. Expert and exceptional performance. Annu Rev Psychol 1996— The review of expert performance — anchors the deliberate-practice tradition and its limits in sport.
- Savelsbergh GJP, van der Kamp J. Perception and action in sports — the case of tennis. 2003— Perception–action coupling in sport — defends the case that skill is perception-led, not body-led, and drills should reshape what the rower perceives.
- Araújo D, Davids K, Hristovski R. Ecological dynamics and sport pedagogy. In Routledge Handbook 2018— The pedagogical extension of ecological dynamics — defends drill selection as constraint manipulation in coaching pedagogy.
- Button C, Seifert L, Davids K. Dynamics of skill acquisition in racket sports. 2021— Recent review of skill acquisition in racket sports — anchors the constraint-led approach in a sport with similar perception–action coupling to rowing.
- Sullivan KJ et al. Task-oriented training for stroke rehabilitation. Stroke 2007— Task-oriented training in rehabilitation — the clinical anchor for the constraint-led approach: shaping the task rather than prescribing the body.
- Poolton JM et al. Advantages of an external focus of attention in laparoscopic skills training. 2007— External focus advantage in surgical skills — replicates the Wulf finding in a different domain, anchors the external-focus case for indoor rowing.
- Wulf G, McNevin N, Shea CH. Automaticity of complex motor skill learning and attentional focus. 2001— External focus promotes automaticity — the empirical anchor for the case against verbal-cognitive cueing in early skill acquisition.
- Schöllhorn W et al. Time scales of adaptive behaviour under stochastic perturbations. 2009— Stochastic-resonance model of differential learning — the theoretical mechanism for why variability in practice transfers.
- Serrien B. A critical review on the theoretical framework of differential motor learning. OSF preprint— Critical review of differential learning — anchors the case for variability in practice and the limits of repetition-based drill design.
- Hodges NJ, Frank A. The role of predictive mechanisms in sport. J Sports Sci 2018— Predictive mechanisms in sport — anchors the case that drill selection should target the predictive model, not just the movement.
- Ericsson KA et al. Experience and deliberate practice in expert performance. 2006— Cambridge Handbook chapter — anchors the empirical case for deliberate, feedback-rich practice in skill development.
- Cumming J, Hall C. Deliberate imagery practice — the development of imagery skills in athletes. J Sports Sci 2002— Mental imagery in athletes — defends the case for imagery drills (visualising the correct stroke) as a complement to physical drills.
- Feltz DL, Landers DM. The effects of mental practice on motor skill learning and performance. J Sport Psychol 1983— Meta-analysis of mental practice — anchors the case for imagery-only drills as a learning tool, not just a warm-up.
- World Rowing. Coaching education portal. World Rowing— The international federation's coaching education resource — the body-level reference for drill selection in rowing.
- British Rowing. Go Row Indoor — technique drills— The UK federation's indoor-rowing technique drills — the fault-by-drill reference for indoor rowing.
- Concept2. Drive Length (blog)— The manufacturer's guide to drive-length measurement — the technical anchor for the catch-connection drill.
- Concept2. PM5 force curve documentation— The PM5 force-curve documentation — the technical anchor for diagnosing the fault layer (catch, drive, finish) from the curve.
- Concept2. Rowing stroke rate explained (blog, 2017)— Manufacturer guidance on rate-by-drill — anchors the practical case for matching rate to the drill's intent.
- Concept2. Pause drill (technique blog)— The pause drill as a fault-diagnosis tool — the practical anchor for picking the drill from the fault.
- Concept2. Feet-out drill (technique blog)— The feet-out drill as the catch-connection drill — the practical anchor for the missing-catch-connection fault.
- Ericsson KA. Peak performance in sports — examination of expert performance. 1991— Animal Behaviour piece on expert performance — anchors the empirical case for the 10,000-hour figure and its limits.
- Wulf G, Shea C. Principles derived from the study of simple skills do not generalize to complex skill learning. 2002— The limits of laboratory-derived motor-learning principles — anchors the case that drill selection must respect the complexity of full-stroke rowing.
- Hodges NJ, Lee TD. The role of augmented feedback in skill acquisition. In Skill Acquisition in Sport 2012— Augmented feedback chapter — anchors the case for KP (knowledge of performance) feedback over KR (knowledge of results) feedback in drill design.
- Horst F, Janssen D, Beckmann H, Schöllhorn WI. Movement signatures across throwing disciplines. Front Psychol 2020— Differential-learning application to decathlon — shows that athlete-specific movement signatures transfer across disciplines, supporting individualised drill selection over generic drills.
- Chiviacowsky S, Wulf G, Wally R, Borges T. KR after good trials in older adults. RQES 2009— Feedback-after-good-trials paper — supports the motivational/attentional explanation for why self-controlled practice outperforms dictated practice.
- Kaefer A, Chiviacowsky S, Meira CM Jr, Tani G. Self-controlled practice in introverts and extroverts. RQES 2014— Tests the generality of self-controlled practice across personality types — supports the case that learner choice is broadly beneficial, not only for a personality subset.
- Andrieux M, Danna J, Thon B. Control of task difficulty during motor learning. J Motor Behav 2012— Learner control over task difficulty is a closely related autonomy manipulation — supports the broader autonomy-in-practice literature.