Rounded Shoulders: Why “Pull Them Back” Makes It Worse

Pull your shoulders back.

You have heard this. You have tried this. You pulled them back. They went forward again. You pulled harder. They went forward again. You set reminders on your phone. Stuck a note on your monitor. Asked your partner to tap you when they noticed.

They went forward again.

Every person I have seen with rounded shoulders has been told to pull them back. Every one. And every one’s shoulders went forward again within minutes. The instruction is not wrong. It is aimed at the wrong system.

The Instruction Everyone Gives

Pull your shoulders back. Squeeze your shoulder blades together. Strengthen your rhomboids. Do rows. Do band pull-aparts.

The logic is clean: your shoulders are forward, the muscles in back are weak, strengthen the back and the shoulders will stay.

It makes perfect sense if muscles are the problem.

They are not.

Your rhomboids are not weak. Your brain is overriding them. That is a different problem with a different solution. And no amount of rowing will touch it.

Why It Fails Within Minutes

Here is what happens when you pull your shoulders back.

Your motor cortex sends a command. The muscles between your shoulder blades contract. Your shoulders move back. You feel the correction. It feels right.

And your brain learns absolutely nothing from it.

This is the part nobody explains. When you voluntarily move your shoulders back, your brain already predicted what that movement would feel like [1]. The resulting sensation matches the prediction perfectly. No surprise. No new information. No reason for the brain’s model to update.

The instruction reached your muscles. It never reached your body schema: the brain’s internal model of where your body should be [3]. That model is still predicting rounded shoulders. The moment your conscious effort drops, the prediction reasserts. Forward they go.

This is not a failure of discipline. It is a feature of how the nervous system processes information [2]. A motor command that produces an expected result generates zero prediction error. Zero prediction error means zero model update. You could do this a thousand times. The model would not change once.

Rounded shoulders are not caused by weak upper back muscles. The brain is actively holding the shoulders forward as part of a protective pattern. Research by Thomas Hanna (1988) identified a pattern called the Green Light Reflex, where the anterior muscles (chest, front of shoulder) are chronically contracted below voluntary control. The person cannot release what they cannot feel. The instruction “pull your shoulders back” asks the posterior muscles to overpower an anterior holding pattern that the brain is actively maintaining. The correction lasts seconds to minutes before the pattern reasserts. In the predictive processing framework (Friston 2010, Clark 2015), the brain generates the rounded position as its best available prediction. Correcting the output without updating the prediction produces temporary change that reverts when conscious effort stops. Effective change requires updating the nervous system’s prediction about what the shoulders should be doing, not overpowering the current instruction with muscle effort.

What Is Actually Rounding Your Shoulders

Your brain maintains a model of your body. Where each part is. Where each part should be. What each muscle group should be doing right now [3]. This model runs constantly, below conscious awareness. It generates your posture as its output.

Your rounded shoulders are not a weakness. They are an output.

The model is predicting that your shoulders should be forward. The motor system executes the prediction. The muscles obey. The shoulders round.

Three things are happening simultaneously.

First: the muscles in the front of your body are chronically contracted below voluntary control. Thomas Hanna called this pattern the Green Light Reflex [1]. Your pectorals. Your anterior deltoids. They are locked short. Not because they are tight. Because your brain has lost the ability to release them. This is Sensory Motor Amnesia: the brain has lost voluntary control over these muscles. You cannot release what you cannot feel.

Second: your nervous system has organized this position as part of a protective pattern [5]. The shoulders roll forward. The head pushes forward. The chest elevates. This is not three separate problems. It is one pattern. The nervous system is bracing, and the rounded shoulders are one piece of the brace. For the full picture of why the upper back rounds, see our complete guide to kyphosis.

Third: the muscles between your shoulder blades are not weak. They are being suppressed by a descending signal from your nervous system [6]. The brain has turned down their recruitment. Strengthening a muscle the brain is actively suppressing gives you a stronger muscle that is still suppressed. This is why your rows get heavier and your shoulders stay forward.

You have been treating the printout. The program generating the printout has not changed.

Shoulders round forward and stay forward because the nervous system is generating the position as a prediction, not as a default that muscles can override. The brain maintains an internal model of the body called the body schema (Paillard 1999). This model generates predictions about what each segment should be doing. The motor cortex executes the prediction. When you pull your shoulders back, you briefly override the motor cortex’s output. But the prediction in the parietal cortex has not changed. The moment conscious effort stops, the prediction reasserts and the shoulders round forward again. Porges (2011) explains part of why: the nervous system organizes posture around perceived safety, and the forward rounding is part of a protective bracing pattern. The brain considers this position safer than an open, retracted position. Until the nervous system’s threat assessment shifts, the protective pattern will regenerate the rounding regardless of how strong the rhomboids become.

—

This is what we work on inside the Posture Dojo. Not stronger muscles. Not reminders. Updating the prediction that generates the pattern. If you have been pulling your shoulders back for months and nothing has stuck, the problem is the layer you are working at. Learn the nervous system approach at posturedojo.com.

—

What Changes the Pattern

The prediction changes when the brain receives sensory evidence it did not expect.

Not a motor command. Not a correction. Not an instruction to squeeze something harder.

Sensory input. Novel. Precise. Delivered in a state where the nervous system is actually available to receive it.

This is the inversion. You have been trying to overpower the pattern with effort. The pattern changes when you stop efforting and start attending.

Try something right now. Do not pull your shoulders back. Instead, place one hand flat across the front of your chest, just below your collarbones. Do not push. Do not correct anything. Just feel the tissue under your hand. Notice its temperature. Notice whether it feels contracted or slack. Notice whether you can feel your breath moving there.

That is sensory input. No motor command. No predicted sensation to cancel the evidence. Your brain just received information about a region it may not have clearly felt in years.

“Pull your shoulders back” is a motor instruction. It generates a predicted sensation. Evidence cancelled [2].

“Notice where the front of your chest meets your collarbones” is a sensory invitation. No motor command. No predicted sensation. The evidence arrives intact. The model has something to work with.

The difference between those two sentences is the difference between a brain defending its current model and a brain available to update it. This is the same reason stretching does not fix posture and why posture correctors make the problem worse. They all target the output. None of them target the prediction.

The anterior muscles locked in the Green Light Reflex need cortical access restored, not a stronger pull from behind. That restoration happens through sensation, not force. Through noticing, not correcting. Through giving the brain information it did not predict [2].

The brain’s map of your shoulder girdle may have degraded. The cortical representation of that region becomes blurred when the same pattern runs for years without change. The brain cannot precisely control what it cannot precisely feel. Restoring the map is a sensory event, not a muscular one.

This is also why trying harder makes posture worse, not better. More effort generates more predicted sensation. More predicted sensation means less new information. The system locks tighter the harder you push.

When to See Someone

If your rounded shoulders come with numbness or tingling in the arms or hands, get a professional assessment first. If you have pain that wakes you at night or has been worsening over weeks, see a provider. If you have a diagnosed condition affecting the thoracic spine, work with someone who understands it.

Outside those situations, the rounded position is almost certainly a nervous system pattern, not a structural failure. The shoulders are doing what the brain told them to do. The conversation needs to move from the muscles to the model.

Upper back strengthening does not fix rounded shoulders because the problem is not weakness. The brain has reorganized its motor strategy (Hodges and Moseley 2003) to include the rounded position as part of its protective postural prediction. The rhomboids and lower trapezius are real muscles that respond to training. They get stronger with rows and band pull-aparts. But the nervous system continues to generate the rounded position as its motor output because the prediction driving the output has not changed. The brain is simultaneously maintaining the anterior contraction (Sensory Motor Amnesia, Hanna 1988) that holds the shoulders forward and suppressing the posterior muscles from full recruitment. Strengthening a muscle the brain is actively suppressing produces a stronger muscle that is still suppressed. The instruction must reach the body schema’s prediction (Paillard 1999), not the muscle. The correction needs to update the model, not overpower its output.

The Shoulders Are Listening to Something

Your shoulders are not disobedient. They are not lazy. They are not defective.

They are doing exactly what your nervous system asked them to do. Precisely. Reliably. Every single time.

The instruction is the problem. Not the execution.

You can keep pulling them back. You can set more reminders. You can buy a posture corrector. You can do rows until your rhomboids are the strongest muscles in your body. Your shoulders will return to the position your brain predicted for them. Every time. Because the prediction is the boss. The muscles are just employees.

Change the instruction. The employees follow.

I spent years pulling my own shoulders back. Bracing against my 85-degree scoliosis with everything I had. They went forward again. Every time. Until I stopped talking to the muscles and started talking to the system running them. The shoulders moved on their own. Not because I forced them. Because the prediction updated.

Your neck pain is connected to this. Your forward head is connected to this. Your shoulders are not an isolated problem. They are one line in a longer conversation your nervous system is having with gravity.

The conversation needs to change. Not get louder.

—

Sources

[1] Hanna, T. (1988). Somatics: Reawakening the Mind’s Control of Movement, Flexibility, and Health. Da Capo Press.

[2] Friston, K. (2010). The free-energy principle: a unified brain theory? Nature Reviews Neuroscience, 11(2), 127-138.

[3] 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.

[4] Clark, A. (2015). Surfing Uncertainty: Prediction, Action, and the Embodied Mind. Oxford University Press.

[5] Porges, S.W. (2011). The Polyvagal Theory: Neurophysiological Foundations of Emotions, Attachment, Communication, and Self-Regulation. W.W. Norton.

[6] Hodges, P.W., & Moseley, G.L. (2003). Pain and motor control of the lumbopelvic region: effect and possible mechanisms. Journal of Electromyography and Kinesiology, 13(4), 361-370.

—

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

1. Hanna, T. (1988). Somatics: Reawakening the Mind’s Control of Movement, Flexibility, and Health. Da Capo Press. [T1]
   
   Green Light Reflex. Chronic anterior contraction below voluntary control.
2. Friston, K. (2010). The free-energy principle: a unified brain theory? Nature Reviews Neuroscience, 11(2), 127-138. [T1]
   
   Predictive coding. The brain generates the rounded shoulder position as a prediction.
3. 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 neural model generating the postural prediction.
4. Clark, A. (2015). Surfing Uncertainty: Prediction, Action, and the Embodied Mind. Oxford University Press. [T1]
   
   Predictive processing. Motor outputs generated by internal models.
5. Porges, S.W. (2011). The Polyvagal Theory: Neurophysiological Foundations of Emotions, Attachment, Communication, and Self-Regulation. W.W. Norton. [T1]
   
   Polyvagal theory. Nervous system organizes posture around perceived safety.
6. Hodges, P.W., & Moseley, G.L. (2003). Pain and motor control of the lumbopelvic region: effect and possible mechanisms. Journal of Electromyography and Kinesiology, 13(4), 361-370. [T1]
   
   Motor reorganization. Brain restructures motor strategy under chronic conditions.