Predictable Failure Patterns in Non-Habitual Movement: Evidence for Undertrained Degrees-of-Freedom Control

Abstract

Predictable Failure Patterns in Non-Habitual Movement: Evidence for Undertrained Degrees-of-Freedom Control

Ksenia Shcherbakova, PhD — Baseworks

Healthy adults coordinate movements with apparent ease yet fail predictably in a specific class of tasks: those requiring control of degrees of freedom (DOF) typically assigned to the uncontrolled manifold. These failures are invisible in ordinary movement contexts but consistently observable under explicit DOF constraints, regardless of general movement experience.

We document these patterns using data from Baseworks—a movement methodology whose decade-long optimization for movement communicability across 10,000+ learners identified specific tasks where failures cluster predictably: maintaining multi-point trunk organization through weight transfer, holding body segment alignment while executing distal movements, and reproducing unfamiliar configurations without visual feedback.

Our observations are consistent with recent evidence that repositioning accuracy is independent of peripheral receptor input, pointing to a central process whose nature remains uncharacterized. We propose this capacity is systematically undertrained in healthy adults; the predictability of these failure patterns suggests they are experimentally tractable for investigating the underlying mechanism.

References

Proske U, Weber BM. Measures of human position sense do not always include contributions from peripheral sensory receptors. Eur J Neurosci. 2026;63:e70444.

Scholz JP, Schöner G. The uncontrolled manifold concept: identifying control variables for a functional task. Exp Brain Res. 1999;126:289-306.

Note for PDF

I am an independent researcher in my 10th year of systematic observation within the Baseworks methodology. The current work grew directly from questions raised during my presentation at the Brenda Milner Neuropsychology Day last year — specifically, why certain apparently simple movement tasks produce consistent failure in healthy adults. I have since had the opportunity to discuss the underlying mechanisms with Paul Cisek (Université de Montréal), whose feedback significantly sharpened the framing. Related work has been accepted for presentation at the 8th Annual BRNet Meeting in Padua this June. Presenting at Brenda Milner Neuropsychology Day again would allow me to test these refined ideas with the Montreal neuroscience community, and to continue developing the research relationships that may enable future experimental investigation of these questions.

Notes

• Accepted — poster presentation confirmed
• Submitted March 26 (deadline was March 23 — form still open, response confirmed via Google Form)
• Poster format; visual plan: 3 movement examples with joint-marker diagrams (possibly footage from Montreal study group first day)
• Candidate movements: Star Tilt, squat (ribcage/pelvis stack), single-leg weight transfer (pelvis parallel), lunge-to-stand (front knee), T-arms twist/tilt
• Citations omitted from submitted text (150-word limit) but included in PDF version
• Related: Neuro 2025 Abstract, BRNet 2026 Abstract

---

Poster Text

Introduction

• Studies of motor synergies address how the CNS organizes redundant degrees of freedom (DOF) to stabilize a performance variable (PV), with the implicit assumption: the PV is already established as a goal in the CNS. The research question is usually how the CNS manages redundancy around that PV, not whether the PV is represented as a goal at all.
• We document a class of tasks where this assumption is challenged. These tasks (1) are explicitly instructed and understood by learners, (2) require no exceptional strength, speed, or range of motion, and (3) produce consistent, large-magnitude failure in a large proportion of healthy adults.
• These tasks were identified in Baseworks, a movement methodology refined over 10+ years across 10,000+ learners, specifically optimized for communicability of movement (CoM). CoM can be limited by: (1) insufficient pressure to improve it; (2) under-specification across DOF; or  (3) learner capacity. The Baseworks refinement process systematically eliminated factors (1) and (2)—both PVs and instructions were iterated until failures could not be attributed to ambiguity. Residual persistent  failure therefore reflects factor (3), learner’s capacity limitation.
• Within the UCM framework (Scholz & Schöner, 1999; Latash, 2010), these tasks use geometric coupling relationships as PVs that are not habitual end-effectors. We propose that synergy formation requires as a precondition that the PV must first be established as a representable goal in the CNS. Our data suggest this precondition is systematically absent for this class of tasks in healthy adults.

Methods

• •Movement data were sampled from existing video recordings of pre-recorded teaching content and multi-session Baseworks training programs. Still frames were selected from available footage based on clear, unobstructed frontal or sagittal view and extracted at maximal displacement by a single experienced rater (K.S., 10+ years of systematic movement observation).
• •Groups: Trained (3 instructors across multiple videos; each data point is 1 individual per task per video) and Untrained (each data point is 1 individual per task per video, sampled during first 1-3 task exposures; tilt: n=8; lunge: n=4; isolate: n=4; individuals are not matched across tasks). Untrained participants had no history of neurological conditions (confirmed by participation waiver). *All participants received detailed instruction, confirmed task comprehension verbally, and were corrected multiple rounds by experienced instructors. Persistent failure under these conditions is unlikely to reflect misunderstanding.
• •Task Selection. Three tasks were selected to represent distinct common constraint classes: Support Configuration Invariance (Lunge), Shape Invariance (Isolate, Tilt), and Inter-Segmental Coupling (Tilt). 
• •Image Analysis. Lime green markers were manually placed at task-relevant anatomical landmarks (3–6 per task). Custom Python scripts detected marker centroids via HSV thresholding and computed a geometric deviation metric per task: change in shin angle between starting and final position (Lunge); shear angle between shoulder and hip lines (Isolate, Tilt); arm-trunk decoupling angle (Tilt). All metrics are reported in degrees; 0° indicates ideal performance. Statistical analysis: one-tailed Mann-Whitney U test.
• •Tilt performance. Across 7 sessions with the same group (n=15), Tilt task performance (correct vs incorrect) was assessed based on video review as percentage of clearly visible participants who correctly performed the task.

Results

• Based on visual cues routinely used by Baseworks instructors to assess performance in three representative tasks, we developed a system to quantify the failure. This analysis revealed large effect sizes. 

• Many untrained participants showed PV coupling error 8-15× that of trained participants across all three tasks, despite detailed instruction, confirmation of comprehension and repeated correction.

• Across 7 group sessions (the same group, 10-15 participants per session), under 10% of participants were able to perform Tilt task correctly during the first 3 sessions, indicating these tasks require targeted training.

• The failure pattern was documented across qualitatively different movement dynamics. Support configuration invariance (Lunge), trunk shape invariance (Isolate, Tilt), and multi-segment coordination (Tilt) involve distinct kinematics.

• In many cases, failures were not just deviations but rather near-complete breakdown of the PV. (e.g., Tilt untrained shear: 30.8° ± 37.82° vs. trained: 3.55° ± 3.24° (U = 4.00, p = .010); arm decoupling: 16.77° ± 15.39° vs. 2.2° ± 1.39° (U = 1.00, p = .003))

• Motor failure is accompanied by perceptual failure. Learners report being unable to detect whether their body is executing the instructed task. (E.g. “I found it very difficult and confusing. It is really hard to tell if my body is doing what is being described.” / participant feedback)

Discussion

• Motor failure (can’t maintain the PV)  and perceptual failure (can’t detect whether the PV is maintained) co-occur consistently. We propose these tasks are a form of “joint repositioning” task, whose accuracy depends on a “central memory mechanism” (Proske & Weber, 2026) independent of muscle spindle input.
• Synergy formation may require a precondition the UCM framework doesn’t address. Practice improves mean performance (CV1) or redistributes variance toward the UCM (CV2), but both assume the PV is already a CNS goal. We extend Latash’s (2010) learning model with a prior stage, CV0: establishment of perceptual and motor access to the PV. CV0 is a necessary condition for the CV1/CV2 sequence to begin (see Figure →).
• A principle of Distributed Activation (co-contracting as many muscles as possible before and during any movement) emerged spontaneously through Baseworks’ iterative CoM refinement. Our observations suggest it facilitates motor and perceptual skill acquisition. Pre-contraction may support CV0, CV1, and CV2 through two mechanisms: amplifying afferent signal strength (facilitating perceptual access to the novel PV) and raising the cost of habitual synergy solutions (reducing their interference with novel PV formation), which is distinct from Bernstein’s novice freezing, which eliminates DOF complexity rather than scaffolding new PV access.
• Progressive improvement in Tilt performance (8.3% → 36.4% over 7 sessions) confirms CV0 is acquirable, but the slow trajectory indicates it requires dedicated practice not provided by existing training frameworks.
• The Baseworks CoM optimization distinguished instruction inadequacy from learner capacity, preserving failures invisible to standard lab paradigms. This class of motor deficit has no distinct name in movement science, rehabilitation, or physical education. The predictability and quantifiability documented here make it experimentally tractable.

Figures

Graphic: A panel with three images per task showing marker placements.

Task|Metric|Group|Mean ± SD|N Lunge|Foundation Integrity|Trained|1.34 ± 1.63|4 ||Untrained|11.74 ± 3.28|4 Isolate|Shear|Trained|1.49 ± 1.41|5 ||Untrained|9.02 ± 5.68|4 Tilt|Shear|Trained|3.55 ± 3.24|5 ||Untrained|30.8 ± 37.82|8 Tilt|Arm Dec|Trained|2.2 ± 1.39|5 ||Untrained|16.77 ± 15.39|8

Y axis label: PV Coupling Error (°)

Session # | %% of group participants performing Tilt correctly S1 8.3% S2 0.0% S3 9.1% S4 (video corrupted) S5 20.0% S6 30.0% S7 36.4%

---

References

Bernstein NA. The co-ordination and regulation of movements. Oxford: Pergamon Press; 1967.

Cisek P, Pezzulo G. Navigating the affordance landscape: feedback control as a process model of behavior and cognition. Trends Cogn Sci. 2024.

Latash ML. Motor Synergies and the Equilibrium-Point Hypothesis. Motor Control. 2010 July ; 14(3): 294–322.

Proske U, Weber BM. Measures of human position sense do not always include contributions from peripheral sensory receptors. Eur J Neurosci. 2026;63:e70444.