The Five Words That Keep Your Posture Stuck

There are five words that run every posture program you have ever tried.

Hold. Fix. Stretch. Strengthen. Correct.

Say them out loud and notice what your body does. You brace. You tighten. You try harder. That reaction is the problem. Not because trying is bad. Because those five verbs point your effort at a system that does not respond to effort.

Your posture is automatic. It is generated by a non-conscious model your brain maintains of your body in space. That model has a name: the body schema [1][2][3]. It has been studied since 1911. It generates your posture every second of every day without asking your opinion.

Those five verbs assume you are the one running your posture. You are not.

If you are trying to hold, fix, stretch, strengthen, or correct your posture. That is why you are stuck.

The Five Verbs and What They Get Wrong

Every posture program you have encountered is built on at least one of these verbs. Most use all five. They sound reasonable. They feel productive. And they all share the same fatal assumption: that you are the conscious agent controlling your posture.

You are not. Your posture is an output. Generated automatically by the body schema. Below the threshold of awareness [2][3]. These verbs talk to the conscious mind. The system generating your posture does not live there.

Hold

“Hold your shoulders back.” “Hold your spine straight.” “Hold this position for thirty seconds.”

Hold assumes posture is a position you maintain through effort. It is not. Posture is a prediction your nervous system generates [10][13]. You can override that prediction for a few minutes. Maybe an hour if you are very focused. Then your attention shifts. The prediction resumes. The pattern returns.

This is not a willpower failure. It is a systems issue. Wolpert and Ghahramani demonstrated that the motor system operates through internal predictive models, not through conscious instruction [4]. Shadmehr, Smith, and Krakauer showed that conscious correction is inherently temporary because it bypasses the predictive model entirely [5]. The model does not update from you holding a position. It waits for you to stop holding. Then it generates the same output it was generating before.

You cannot hold an automatic process. You can interrupt it. Briefly. But holding is not updating.

Fix

“Fix your posture.” “Fix your forward head.” “Fix your rounded shoulders.”

Fix assumes something is broken. That there is a structural problem in need of repair. But in the vast majority of postural conditions, nothing is broken. The spine is intact. The muscles are functional. The joints move within normal range.

Your body is running a program. The program is outdated. The system is doing exactly what it was told to do based on the sensory inputs it has received over your lifetime [10][13]. The output is protective, not pathological. It is the schema’s best prediction given the evidence available.

Fixing a program that is running as designed does not work. You do not fix software by bending the hardware. You update the software.

Stretch

“Stretch your hip flexors.” “Stretch your pecs.” “Stretch your hamstrings.”

Stretch assumes tightness is the cause of your postural pattern. It is not. Tightness is the output. Your nervous system is generating it [11].

The research here is clear. Weppler and Magnusson’s 2010 review found that stretching increases range of motion primarily through changes in sensory tolerance, not through actual tissue lengthening [6]. You feel like you can go further. The muscle has not changed. Konrad and Tilp confirmed this in 2014: static stretching increases ROM with no measurable structural change to muscle or tendon [7].

The system that generated the tightness is still running. Twenty minutes after you stretch, the signal fires again. The muscle tightens again. Not because the stretch did not work mechanically. Because the system producing the tightness was never addressed.

This does not mean stretching has no place. Stretching can feel good. It can create temporary relief. It can be part of a broader movement practice. But as the primary verb of your posture strategy? It addresses the output while the system keeps generating the same signal.

Strengthen

“Strengthen your core.” “Strengthen your back extensors.” “Strengthen your glutes.”

Strengthen assumes the muscles are too weak to support good posture. But strength and organization are different systems. Tsao, Galea, and Hodges showed in 2008 that people with recurrent low back pain have reorganized motor cortex maps for their trunk muscles [14]. The muscles are not weak. Their cortical representation has shifted. The schema is not asking them to fire in the pattern that would produce upright posture.

You can build a very strong muscle that your nervous system does not recruit for postural support. Athletes do this constantly. They can deadlift twice their body weight and still stand with a forward head. The strength is there. The schema is not using it for posture.

Strengthening a muscle that the schema has excluded from its postural program does not change the program. It gives you a stronger muscle running the same pattern.

Correct

“Correct your alignment.” “Correct your curve.” “Correct your head position.”

Correct assumes there is a wrong position and a right one. That posture exists on a spectrum from incorrect to correct, and the job is to move toward the correct end.

But the body schema does not generate “wrong” posture. It generates its best prediction based on the inputs it has received [10][13]. If your nervous system reads threat, the schema generates a protective pattern. If your sensory inputs are limited, the schema generates from limited data. If the pattern has been running for twenty years, the schema treats it as baseline.

There is no wrong position. There is an outdated program still running. Correction implies error. What is actually happening is a system running predictions that no longer match your current reality. You do not correct a prediction. You provide new evidence so the prediction updates.

Why These Verbs Persist

If these five verbs do not work, why does every program use them?

Because they match the mechanical model of the body. That model treats the spine as a structure, muscles as pulleys, and posture as a position. Within that model, hold, fix, stretch, strengthen, and correct are perfectly logical verbs. You would hold a structure in place. You would fix a broken beam. You would stretch a tight cable. You would strengthen a weak support. You would correct a misalignment.

The mechanical model is not wrong in all contexts. It is excellent for fractures. It is excellent for joint replacements. It is excellent for acute injuries where something is genuinely broken.

But posture is not a structure. It is a process. Generated by the body schema. Running automatically. Below awareness. And when you apply structural verbs to a generative process, you get the experience that millions of people share: temporary improvement followed by the same pattern returning.

The villain is not your doctor. Not your physical therapist. Not your yoga teacher. The villain is the model they were trained in. A model that does not account for the system generating the output it is trying to change.

The Verb That Works

If hold, fix, stretch, strengthen, and correct all talk to the wrong system, what is the right verb?

Update.

The body schema is a predictive model [10][13]. It generates posture from the evidence it has accumulated. It does not respond to force. It does not respond to willpower. It does not respond to repetition of the same inputs it already expects. It responds to new evidence that contradicts its current prediction [5][10].

This is not a metaphor. The rubber hand illusion proved in 1998 that the schema updates from sensory evidence [8]. You place a rubber hand in front of someone while their real hand is hidden. Stroke both simultaneously. Within minutes the brain incorporates the rubber hand into its body schema. Not from instruction. Not from effort. From correlated sensory input that created a prediction error the model could not ignore.

Ramachandran’s mirror therapy showed the same principle in clinical practice [9]. Amputees with phantom limb pain watch a mirror reflection of their intact hand where the missing hand should be. The schema receives visual evidence that the missing hand exists and is moving freely. Pain resolves. The schema updated. From visual input alone.

These are not obscure findings. The rubber hand illusion is one of the most replicated experiments in psychology. Mirror therapy is used in hospitals worldwide. The mechanism is established: the schema updates when it receives sensory evidence that contradicts its current prediction.

Update does not mean trying harder. It means providing the type of input the schema needs to revise its model. Novel sensory experiences. New spatial information. Inputs the system does not already predict. The kind of evidence that creates prediction error large enough to trigger model revision [10].

This is what generative posture is built on. Not fixing the output. Updating the model that generates it.

What This Means for You

Listen to the language of whatever posture program you are using. Count the verbs. If the instructions are built on hold, fix, stretch, strengthen, or correct, you now know which system those instructions are talking to. And you know it is not the system generating your posture.

This does not mean you should stop moving. Movement matters. Stretching and strengthening have real value in their own contexts. They build capacity. They maintain tissue health. They feel good. But as the primary strategy for changing your posture, they address the output while the program keeps running [11][12].

The question is not whether you are doing the right exercises. The question is whether any exercise, done with the verb “correct” in your mind, can reach the system that is generating the pattern.

It cannot. Because that system is automatic. It runs below conscious access [2][3]. It does not listen to your intentions. It listens to sensory evidence.

When the right evidence reaches the schema, the output changes. Not because you forced it. Because the model revised its prediction. That is what updating looks like. Quiet. Automatic. Lasting. The opposite of holding something in place through effort.

Your body is not broken. Your posture is not wrong. The program running your posture is outdated. And programs do not need to be fixed. They need to be updated.

The Shift

Five verbs built an industry. Hold, fix, stretch, strengthen, correct. They built every posture program, every spinal correction protocol, every “stand up straight” reminder on your phone. They all assume you are running the show.

One verb changes the game. Update.

Not because it sounds better. Because it talks to the right system. The system that has been generating your posture since before you knew you had one. The system that does not care what you want your posture to be. The system that responds only to evidence.

You are not broken. You are not weak. You are not doing it wrong. You are using the wrong verb.

The body schema was first described in 1911 [1]. The mechanism has been studied for over a century. The science is not new. The application is.

Your posture is generated. The model that generates it is waiting for an update. Not a correction. Not a fix. An update.

The question was never “how do I hold better posture?” The question is: “what evidence does my schema need to generate a different output?”

That question changes everything.

Written by Sam Miller. Eight years of clinical practice working with the body schema in scoliosis, kyphosis, and chronic postural conditions. Founder of Posture Dojo and creator of Syntropic Core.

---

---

Sources

1. Head, H., & Holmes, G. (1911). Sensory disturbances from cerebral lesions. Brain, 34(2-3), 102-254. [T1]
   
   First formal description of the body schema. The brain maintains a postural model that generates motor output automatically. Establishes that posture is not consciously held.
2. 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 (pp. 197-214). Sofia: Academic Publishing House. [T1]
   
   Schema vs image dissociation. Schema operates non-consciously, generating motor output. Conscious correction addresses the image, not the schema.
3. Gallagher, S. (2005). How the Body Shapes the Mind. Oxford University Press. [T1]
   
   Body schema as pre-reflective motor organization. The generative engine runs beneath conscious access. You cannot consciously hold what is automatically generated.
4. Wolpert, D.M., & Ghahramani, Z. (2000). Computational principles of movement neuroscience. Nature Neuroscience, 3(Suppl), 1212-1217. [T1]
   
   Internal models generate motor output through prediction. Conscious correction is temporary because it bypasses the predictive model. The model resumes when attention shifts.
5. Shadmehr, R., Smith, M.A., & Krakauer, J.W. (2010). Error correction, sensory prediction, and adaptation in motor control. Annual Review of Neuroscience, 33, 89-108. [T1]
   
   Motor adaptation requires prediction error, not conscious correction. The motor system updates from sensory mismatch, not from intention.
6. Weppler, C.H., & Magnusson, S.P. (2010). Increasing muscle extensibility: a matter of increasing length or modifying sensation? Physical Therapy, 90(3), 438-449. [T1]
   
   Stretching increases range of motion primarily through sensory tolerance, not tissue lengthening. The system’s set point does not change. Tightness returns because the system is still generating it.
7. Konrad, A., & Tilp, M. (2014). Increased range of motion after static stretching is not due to changes in muscle and tendon structures. Clinical Biomechanics, 29(6), 636-642. [T1]
   
   Static stretching increases ROM without structural change to muscle or tendon. The change is sensory, not mechanical. The system’s output remains unchanged.
8. Botvinick, M., & Cohen, J. (1998). Rubber hands ‘feel’ touch that eyes see. Nature, 391(6669), 756. [T1]
   
   Rubber hand illusion. T1 proof that the body schema is plastic and updates from sensory evidence, not from conscious instruction.
9. Ramachandran, V.S., & Hirstein, W. (1998). The perception of phantom limbs. Brain, 121(9), 1603-1630. [T1]
   
   Mirror therapy updates the schema through visual input alone. The schema does not require physical correction. It requires sensory evidence.
10. Friston, K. (2010). The free-energy principle: a unified brain theory? Nature Reviews Neuroscience, 11(2), 127-138. [T1]
    
    Active inference. The brain generates predictions and updates only from prediction error. Conscious correction does not generate prediction error. The model ignores it.
11. Hanna, T. (1988). Somatics: Reawakening the Mind’s Control of Movement, Flexibility, and Health. Da Capo Press. [T1]
    
    Sensory Motor Amnesia. The cortex loses voluntary control of muscles the schema has locked into chronic contraction. Strengthening those muscles does not restore cortical access.
12. Feldenkrais, M. (1972). Awareness Through Movement: Health Exercises for Personal Growth. Harper & Row. [T1]
    
    Novel movement as schema input. Non-habitual patterns create prediction errors. The system updates from novelty, not from repetition of the same exercises.
13. Clark, A. (2015). Surfing Uncertainty: Prediction, Action, and the Embodied Mind. Oxford University Press. [T1]
    
    Predictive processing. The brain is a prediction machine. Motor output is generated from the model’s best guess. You do not override a prediction. You update the evidence it draws on.
14. Tsao, H., Galea, M.P., & Hodges, P.W. (2008). Reorganization of the motor cortex is associated with postural control deficits in recurrent low back pain. Brain, 131(8), 2161-2171. [T1]
    
    Motor cortex reorganization in postural dysfunction. Muscles are not weak. Their cortical representation has shifted. Strengthening bypasses the actual deficit.