How the Mechanical Model of the Body Was Built (and Where It Stops Working)
A field hospital. Northern France. 1944.
A surgeon stabilizes a fractured femur with traction and external fixation. The bone is broken. The tissue is torn. The solution is mechanical: hold the structure in place, let biology heal. It works. It works brilliantly. Soldiers walk again.
This is where the model was born. Not in a research lab thinking about posture. Not in a clinic treating chronic back pain. In a field hospital, under fire, where the body was broken and needed to be put back together.
The mechanical model of the body was built for emergencies. It has saved millions of lives. It is extraordinary at what it was designed for. And it was never designed for the question you are asking when your posture will not change.
That is not a criticism. It is a boundary.
Where the Model Came From
The two World Wars created an enormous body of knowledge about the human body under structural failure. Fractures. Dislocations. Shrapnel wounds. Amputations. The surgeons who treated these injuries developed techniques that were genuinely revolutionary. External fixation. Internal fixation. Traction. Stabilization.
This knowledge did not disappear when the wars ended. It became the foundation of modern orthopedics. The surgeons who trained in field hospitals went home and became the professors who trained the next generation. The model they carried with them was the model they knew: the body is a mechanical structure. When it breaks, you repair it. When it is out of place, you put it back.
The model expanded through the industrial era. Factory workers with repetitive strain. Laborers with spinal compression. The body as a machine that wears down. The treatment: repair the worn parts, brace the weakened structures, cut out the things that do not belong.
Paul Harrington developed his instrumentation system in 1962 [1]. The original application was polio patients whose spines were collapsing from muscular paralysis. The spine was structurally failing. Rods and hooks held it in place. For polio patients, this was life-changing. The spine could not support itself. External structure provided what internal structure could not.
The Harrington rod became the template for scoliosis surgery. A spine with a curve could be straightened with rods. The logic was clean. The curve is the problem. The rod is the solution. Straighten the beam.
The Milwaukee brace appeared in 1945. The Boston brace in 1972. Rigo-Cheneau in Europe. Each one a different engineering approach to the same mechanical premise: the curve is the problem, external force is the solution, contain the shape and the shape will improve [11] [12].
For decades, this model went unquestioned. Because within its domain, it worked.
What It Does Brilliantly
The mechanical model saves lives. That sentence is not diplomatic softening before a critique. It is the truth.
Fracture stabilization. A broken vertebra needs to be held in place while the bone heals. The mechanical model does this with extraordinary precision. Internal fixation. External bracing. Traction when needed. The bone heals. The person walks.
Joint replacement. A hip destroyed by arthritis, bone grinding on bone, every step painful. The mechanical model removes the destroyed joint and replaces it with engineered components. The person walks without pain. Sometimes for decades.
Decompression surgery for neurological compromise. A disc herniation pressing on the spinal cord. Progressive weakness. Bowel and bladder changes. The surgeon removes the material compressing the nerve. Function returns. Sometimes within hours.
Acute disc herniation management. Tumor resection. Infection drainage. Spinal fusion for instability after trauma.
These are not small things. These are people who could not walk, walking. People who were losing bladder control, recovering it. People in neurological crisis, stabilized.
The model does not need to be replaced in these domains. It needs to be honored. The surgeons and physicians who do this work operate at extraordinary levels of skill within a framework that is perfectly designed for what they are treating.
The question is not whether the model works. The question is where it stops working.
What is the mechanical model of posture treatment?
The mechanical model of posture treatment is the framework inherited from wartime and industrial-era orthopedics that treats the body as a mechanical structure. Under this model, postural problems are structural misalignments to be corrected through external force, bracing, strengthening, or surgical realignment. The model traces to World War I and II field surgery, where fracture stabilization and structural repair saved lives (Harrington 1962). It expanded through the development of bracing systems like the Milwaukee brace (1945) and Boston brace (1972). The model is extraordinary for acute trauma, neurological compromise, and structural failure. Its limitation appears in chronic postural conditions, where the body is not structurally broken but is generating a pattern through its nervous system’s predictive model (Friston 2010, Paillard 1999). The mechanical model addresses the output. It does not address the generator.
Where It Reaches Its Limit
Ray is 61. Thoracic kyphosis. Mild stenosis. He has been through physical therapy three times. Each round followed the same pattern: strengthening exercises for the upper back, stretching for the chest, postural cues to stand taller. Each round produced temporary improvement. Each time, the posture returned to where it started within weeks of stopping the exercises.
The exercises were not wrong. Ray’s therapists were not incompetent. They were working within a model that assumes posture is a structural problem with a structural solution. The assumption: strengthen what is underperforming. Stretch what is restricted. Reposition what has shifted.
But posture is not a structural position. Posture is a generated output [3] [9]. Your nervous system maintains an internal model of your body called the body schema [4]. That model generates your posture moment by moment. The model decides where your head sits, how your thoracic spine curves, where your center of mass falls. Every second.
The mechanical model has no framework for this. There is no lever to pull that changes the generator. Strengthening the upper back muscles does not update the schema that is generating the rounded posture. Stretching the pectorals does not change the prediction that is pulling the shoulders forward. Postural cues (“stand up straight”) create temporary motor overrides that fatigue within minutes because the underlying prediction has not changed.
This is where the model reaches its boundary. Not with broken bones. Not with compressed nerves. Not with structural failure. With anything that requires the body to change its own output.
Scoliosis. Kyphosis. Chronic postural patterns. Forward head position. Rounded shoulders. Anterior pelvic tilt. These are generated outputs. The body generates these patterns through the schema. The schema is running a prediction. The prediction was built from sensory inputs, protective responses, and developmental organization.
The BrAIST study, published in the New England Journal of Medicine in 2013, demonstrated that bracing reduces the risk of surgical intervention in adolescent idiopathic scoliosis [2]. This is an important finding. Bracing works within the mechanical model. External force applied to a growing spine can prevent curve progression to the threshold where surgery is indicated.
But what happens when the brace comes off?
If the generator has not been updated, the generation resumes. The brace contained the output. It did not change the prediction producing the output. The BrAIST study showed bracing reduces surgical risk. It did not show that bracing changes the nervous system’s reason for building the curve.
The same pattern repeats across the mechanical model’s approach to chronic posture. Correct the output. The output returns. Correct it again. It returns again. The model has no explanation for why correction does not hold. Because the model does not recognize that the output is being generated.
Why doesn’t physical therapy fix posture permanently?
Physical therapy for posture typically follows the mechanical model: strengthen weak muscles, stretch tight muscles, correct alignment. This approach treats posture as a structural position rather than a generated output. Research in predictive neuroscience (Friston 2010, Clark 2015) shows that posture is generated by the brain’s internal model of the body, the body schema (Paillard 1999). Strengthening exercises and postural cues create temporary motor overrides, but they do not update the prediction that generates the postural pattern. This is why posture often returns to its previous position after physical therapy ends. The exercises address the output. They do not contact the generator. Effective long-term postural change requires updating the body schema through sensory evidence the nervous system did not predict, a mechanism outside the mechanical model’s framework.
Why the Model Persists
If the mechanical model reaches its limit with chronic postural conditions, why is it still the dominant approach?
Three forces hold it in place.
Training pipelines. Medical schools, physical therapy programs, and chiropractic colleges teach anatomy, biomechanics, and kinesiology through the mechanical lens. The body is taught as a system of levers, pulleys, and force vectors. This is not wrong. It is incomplete. But by the time a clinician finishes their training, the mechanical model is the water they swim in. The framework is invisible because it is everywhere.
Residency programs reinforce the model through clinical practice. Surgical training reinforces it through outcomes. The surgeon who performs a successful spinal fusion sees the model confirmed. The physical therapist whose patient improves during a course of strengthening sees the model confirmed. The model works within its domain. Within its domain, it earns its persistence.
Insurance and reimbursement. The billing infrastructure was built for the mechanical model. CPT codes describe mechanical interventions: manual therapy, therapeutic exercise, neuromuscular reeducation, spinal manipulation, surgical procedures. Each code maps to a mechanical action performed on a body.
There is no CPT code for “body schema assessment.” There is no billing pathway for “sensory evidence that updates the nervous system’s predictive model.” The insurance system can reimburse a therapist for strengthening exercises. It cannot reimburse a therapist for guiding a patient through the sensory experience that allows their nervous system to discover a different prediction.
This is not conspiracy. It is infrastructure. The billing system was built to support the mechanical model because the mechanical model was the model. The infrastructure now perpetuates the model because changing infrastructure is harder than changing ideas.
Institutional momentum. Clinical guidelines, professional organizations, and research funding all orbit the mechanical model. The SOSORT guidelines for scoliosis treatment describe a progression from observation to bracing to surgery [8]. Each step is mechanical. Each step addresses the output. The guidelines are evidence-based within the framework they operate in. The framework itself is rarely questioned.
Research funding follows the same pattern. Studies that test mechanical interventions receive funding because they fit existing paradigms. Studies that investigate generative mechanisms struggle for funding because they challenge the framework the funding committees were trained in.
The mechanical model persists not because it is wrong. It persists because the institutions built around it were designed to support it. The model and the institutions grew up together. Changing one requires changing the other.
Why is posture treatment still based on the mechanical model?
The mechanical model persists in posture treatment for three systemic reasons. First, clinical training pipelines in medicine, physical therapy, and chiropractic teach anatomy and biomechanics through a mechanical framework where the body is understood as a system of levers and force vectors. Second, insurance reimbursement infrastructure (CPT codes, billing categories) was built for mechanical interventions. There is no billing code for body schema assessment or nervous system prediction updating (Paillard 1999, Friston 2010). Third, clinical guidelines like the SOSORT scoliosis treatment recommendations (Negrini et al. 2018) describe mechanical progressions (observation to bracing to surgery) that address the output rather than the generator. The model persists not because it is wrong within its domain, but because institutional infrastructure was built to support it and now perpetuates it.
What Was Almost Found
The mechanical model did not go unchallenged. Three approaches came remarkably close to a generative understanding. Each one was captured by the paradigm it could have replaced.
Katharina Schroth, 1920s Germany.
She worked in a sanatorium with scoliosis patients. She used mirrors. Not to show patients what was wrong with their posture. To give the nervous system visual feedback about the body’s current position. She used rotational breathing. Not as an exercise to strengthen respiratory muscles. As a way to send respiratory input into the concavities of the curve. She used proprioceptive cueing. Hands on the body. Attention directed to specific territories. Sensation before correction.
Schroth was doing generative work before the neuroscience existed to describe it. She was not working on the output. She was working on the inputs that could update the system generating the output. Mirrors gave visual evidence. Rotational breathing gave respiratory evidence. Proprioceptive cueing gave somatosensory evidence. Three sensory channels, all aimed at the generator.
Her daughter, Christa Lehnert-Schroth, formalized the approach [7]. The formalization was necessary for clinical adoption. It was also the moment of capture. The intuitive, sensory, generative work was translated into corrective exercises. Auto-elongation. Rotational angular breathing patterns. Motor patterning drills. The protocol became reproducible. Teachable. Billable.
The generative insight was domesticated into the mechanical framework. The most promising early approach to working with the generator was absorbed into the paradigm of correcting the output.
Pavel Kolar and the Prague School.
Dynamic Neuromuscular Stabilization works with developmental positions [5]. The positions an infant moves through in the first year of life. Supine. Prone. Rolling. Crawling. Kneeling. Standing. Each position represents a stage in the nervous system’s original organization of posture.
This is schema-level work. The developmental sequence is the sequence the nervous system used to build the body schema in the first place. Working with these positions is working with the pattern that organized the original prediction. DNS is not strengthening muscles. It is re-engaging the developmental logic that generated upright posture.
But the clinical world needed mechanical language to accept it. DNS is framed as “stabilization.” The deep stabilizing system. Intra-abdominal pressure regulation. Joint centration. Each concept is accurate. Each concept also translates a generative mechanism into mechanical terminology so that clinicians trained in the mechanical model can recognize it.
The generative insight survives inside DNS. It is framed in mechanical language because the clinical world has no other language to receive it.
Ron Hruska and the Postural Restoration Institute.
PRI recognizes something the mechanical model does not: asymmetry is a system pattern [6]. Not a structural defect. A pattern driven by respiratory mechanics and autonomic state. The left diaphragm is different from the right. The liver sits on the right. The heart sits on the left. The body is asymmetrical by design, and the nervous system organizes around that asymmetry.
PRI addresses respiratory input, autonomic state, and sensory perception. It recognizes that the pattern generating the posture is driven by the system, not by individual muscle deficit or restriction. This is close to generative posture. Closer than most clinical approaches.
But PRI operates within clinical framing. It needs insurance-compatible language. It needs clinical credibility within a mechanical paradigm. The generative insight survives, constrained by the container it must fit inside to reach patients.
Three approaches. Three generative seeds. Each one captured, to varying degrees, by the mechanical model they were attempting to transcend. Not because the people who developed them lacked vision. Because the institutional infrastructure had no category for what they were doing.
What is the Schroth method and how does it work for scoliosis?
The Schroth method was developed by Katharina Schroth in 1920s Germany and formalized by her daughter Christa Lehnert-Schroth (Lehnert-Schroth 2007). It uses three primary inputs: mirror feedback for visual awareness of the body’s position, rotational breathing directed into the concavities of the scoliotic curve, and proprioceptive cueing to direct attention to specific body regions. Originally, this approach worked with the sensory inputs that could update the nervous system’s model of the body (Paillard 1999). As the method was formalized for clinical adoption, it was translated into corrective exercise protocols, auto-elongation techniques, and muscle activation patterns. The Schroth method remains one of the most evidence-supported conservative approaches to scoliosis treatment, though its original generative insights are often reduced to mechanical exercise prescription in clinical practice.
What a Different Model Makes Possible
The mechanical model asks: what is wrong with the structure, and how do we correct it?
A generative model asks a different question: what is the nervous system predicting, and what evidence would update that prediction?
The difference is not philosophical. It is practical. It changes what you do in a session. It changes what counts as progress. It changes what the person sitting in front of you is asked to do.
Under the mechanical model, progress is measured by structural change. Degrees of curvature. Range of motion. Muscle strength. These measurements are valid. They measure the output. They do not tell you whether the generator has changed.
Under a generative model, progress includes the output measurements but adds something the mechanical model cannot capture: has the prediction updated? Is the nervous system generating a different pattern, or is a corrected pattern being held in place by effort?
The distinction matters because it predicts what happens when the person stops the exercises. Stops wearing the brace. Stops thinking about their posture. If the generator updated, the new pattern persists without effort. If the output was corrected without touching the generator, the old pattern returns.
David knows this pattern. Forty-eight years old. Former runner. Hip restriction and lower back pain that physical therapy helped temporarily and that returned every time he stopped going. Three rounds of therapy. Three rounds of improvement. Three returns to the same pattern. Not because the therapy was wrong. Because each round corrected the output without updating the prediction running underneath it.
A different model does not reject what came before. It includes it. The mechanical model’s understanding of anatomy is precise. Its surgical techniques save lives. Its imaging technology reveals structural reality. A generative model uses all of this. It adds the question the mechanical model was never designed to ask: why is the body building this pattern?
Your diagnosis describes a shape. Thirty degrees of thoracic curvature. Twelve degrees of lumbar lordosis above normal. Forward head position measured in centimeters. These measurements are accurate. They describe the output. They do not explain why the body is generating it.
When you understand that your body is generating a pattern rather than failing to hold a position, the entire question changes. You stop asking “what is wrong with my structure?” You start asking “what is my nervous system predicting, and what evidence would allow it to predict something different?”
That question has no CPT code. No billing pathway. No clinical guideline. And it is the question that explains why your posture keeps returning to the same pattern despite everything you have tried.
—
The system that tried to help me was extraordinary at measuring my curve and terrible at asking why my body was building it. Both of those things can be true at the same time.
The mechanical model is not the enemy. It is a tool. An extraordinary tool, used outside its domain. The mismatch is not the model’s fault. It was never designed for your question.
The question is generative. What is your nervous system predicting? What evidence would allow it to predict something different? What inputs reach the system that generates your posture?
These are the questions the mechanical model was never built to ask. And they are the questions that explain why 80% of scoliosis cases have no explanation under the current model. The cases are not mysterious. The model is too small for them.
—
This is what we work with inside the Posture Dojo. Not correcting outputs. Not fighting the structure. Working with the nervous system’s prediction. Giving it evidence to update. Learn how at syntropiccore.com.
—
Next in the Generative Posture Series: How Your Diagnosis Became Your Identity (and Why That Keeps the Pattern Locked) (G-4)
What is the difference between mechanical and generative models of posture?
The mechanical model treats posture as a structural position determined by muscle capacity, flexibility, and skeletal position. Treatment involves repositioning the structure through targeted exercises, stretching, bracing, or surgical intervention (Harrington 1962, Negrini et al. 2018). The generative model, supported by predictive neuroscience (Friston 2010, Clark 2015), treats posture as an output generated by the brain’s internal model of the body, the body schema (Paillard 1999). Under the generative model, the postural pattern is a prediction. Treatment involves providing sensory evidence that allows the prediction to update. The practical difference: under the mechanical model, correction must be maintained by effort or external support. Under the generative model, when the prediction updates, the new pattern persists without effort because the generator itself has changed. Approaches like the Schroth method (Lehnert-Schroth 2007), DNS (Kolar 2014), and PRI (Hruska 2006) contain elements of both models.
—
Sources
[1] Harrington, P.R. (1962). Treatment of scoliosis: correction and internal fixation by spine instrumentation. Journal of Bone and Joint Surgery, 44(4), 591-610.
[2] Weinstein, S.L., et al. (2013). Effects of bracing in adolescents with idiopathic scoliosis. New England Journal of Medicine, 369(16), 1512-1521.
[3] Friston, K. (2010). The free-energy principle: a unified brain theory? Nature Reviews Neuroscience, 11(2), 127-138.
[4] Paillard, J. (1999). Body schema and body image: A double dissociation in deafferented patients. In G.N. Gantchev et al. (Eds.), Motor Control, Today and Tomorrow.
[5] Kolar, P., et al. (2014). Clinical Rehabilitation. Alena Kobesova (Ed.). Rehabilitation Prague School.
[6] Hruska, R. (2006). Myokinematic Restoration: An Integrated Approach to Treatment of Patterned Lumbo-Pelvic-Femoral Pathomechanics. Postural Restoration Institute.
[7] Lehnert-Schroth, C. (2007). Three-Dimensional Treatment for Scoliosis: A Physiotherapeutic Method for Deformities of the Spine. Martindale Press.
[8] Negrini, S., et al. (2018). 2016 SOSORT guidelines: orthopaedic and rehabilitation treatment of idiopathic scoliosis during growth. Scoliosis and Spinal Disorders, 13, 3.
[9] Clark, A. (2015). Surfing Uncertainty: Prediction, Action, and the Embodied Mind. Oxford University Press.
[10] Waddell, G. (1987). A new clinical model for the treatment of low-back pain. Spine, 12(7), 632-644.
[11] Blount, W.P., & Schmidt, A.C. (1957). The Milwaukee brace in the operative treatment of scoliosis. Journal of Bone and Joint Surgery, 39(3), 693-708.
[12] Hall, J.E., Miller, M.E., Schwartzmann, J.R., & Griner, P.F. (1975). A refined concept in the orthotic management of idiopathic scoliosis. Prosthetics and Orthotics International, 29, 7-13.
—
About the author: Sam Miller is the creator of Syntropic Core and founder of Posture Dojo. Diagnosed with an 85-degree scoliosis at 18, he spent two decades mapping the nervous system mechanisms that conventional treatment misses. He works with people whose bodies did not respond to the standard playbook. His approach is built on the predictive neuroscience of posture, not the mechanical model that failed him.
Sources
- Harrington, P.R. (1962). Treatment of scoliosis: correction and internal fixation by spine instrumentation. Journal of Bone and Joint Surgery, 44(4), 591-610. [T1]
Harrington rod development for polio patients. Origin of surgical instrumentation for scoliosis. - Weinstein, S.L., et al. (2013). Effects of bracing in adolescents with idiopathic scoliosis. New England Journal of Medicine, 369(16), 1512-1521. [T1]
BrAIST study. Bracing reduces surgical risk but does not address the generator of curvature. - Friston, K. (2010). The free-energy principle: a unified brain theory? Nature Reviews Neuroscience, 11(2), 127-138. [T1]
Predictive coding. Posture as a generated output of a predictive model, not a structural position. - Paillard, J. (1999). Body schema and body image: A double dissociation in deafferented patients. In G.N. Gantchev et al. (Eds.), Motor Control, Today and Tomorrow. [T1]
Body schema. The nervous system’s internal model generates posture. Not addressable by mechanical intervention alone. - Kolar, P., et al. (2014). Clinical Rehabilitation. Alena Kobesova (Ed.). Rehabilitation Prague School. [T1]
DNS and developmental positions. Working with the sequence the nervous system used to organize posture originally. - Hruska, R. (2006). Myokinematic Restoration: An Integrated Approach to Treatment of Patterned Lumbo-Pelvic-Femoral Pathomechanics. Postural Restoration Institute. [T2]
PRI. Asymmetry as a system pattern driven by respiratory mechanics and autonomic state. - Lehnert-Schroth, C. (2007). Three-Dimensional Treatment for Scoliosis: A Physiotherapeutic Method for Deformities of the Spine. Martindale Press. [T1]
Schroth method formalization. Rotational breathing, proprioceptive cueing, mirror feedback. - Negrini, S., et al. (2018). 2016 SOSORT guidelines: orthopaedic and rehabilitation treatment of idiopathic scoliosis during growth. Scoliosis and Spinal Disorders, 13, 3. [T1]
Current clinical guidelines for scoliosis treatment. Observation-bracing-surgery paradigm. - Clark, A. (2015). Surfing Uncertainty: Prediction, Action, and the Embodied Mind. Oxford University Press. [T1]
Predictive processing. The brain generates posture from a model, not from structural alignment. - Waddell, G. (1987). A new clinical model for the treatment of low-back pain. Spine, 12(7), 632-644. [T1]
Early challenge to the purely mechanical model. The biopsychosocial model of pain. - Blount, W.P., & Schmidt, A.C. (1957). The Milwaukee brace in the operative treatment of scoliosis. Journal of Bone and Joint Surgery, 39(3), 693-708. [T1]
Milwaukee brace development. External force model for scoliosis management. - Hall, J.E., Miller, M.E., Schwartzmann, J.R., & Griner, P.F. (1975). A refined concept in the orthotic management of idiopathic scoliosis. Prosthetics and Orthotics International, 29, 7-13. [T1]
Boston brace development. Low-profile bracing as mechanical containment.
Leave a Reply