Muscle growth slows itself down by design
In short
A multiscale model that couples tissue mechanics to intracellular signalling found that hypertrophy does not run away because of feedback between protein synthesis and muscle size. As a muscle grows, the same external load produces less mechanical stimulus per unit of tissue, so signalling through the IGF1-AKT-mTOR-FOXO pathway falls and growth converges. Running identical load protocols on real human muscle geometries produced different amounts of hypertrophy, and the coupling between shape and signalling buffered local signalling heterogeneity so growth stayed smooth.
A stall usually gets blamed on the program or on effort. This model gives a different answer: growth is structured to stop itself. Protein synthesis makes the muscle bigger, and the bigger muscle turns the signal back down.
Why does the same weight stop working?
As a muscle grows, the same external load spreads across more tissue and the internal mechanical stimulus thins out. The model wires that stimulus into the IGF1-AKT-mTOR-FOXO pathway, so a thinner stimulus means less synthesis signalling and a rebalancing against the degradation arm that FOXO governs. The authors describe this feedback as what prevents unbounded growth.
The practical version is familiar — hold the load constant and the response converges. What is different is the reason: not exhausted adaptability, but the ratio of load to size. Raising the load or changing the character of the stimulus is what moves the curve again.
What does the model actually compute?
At tissue level it treats muscle as a transversely isotropic hyperelastic material, separating along-fibre mechanics from the rest. At cell level it solves the signalling pathway as a system of ordinary differential equations. The link between them is a growth tensor: signalling dynamics set the tensor, and the tensor drives changes in muscle cross-sectional area. That is what lets the model simulate long-term adaptation rather than a single session.
Why do identical programs give different results?
The team ran it on real human muscle geometries taken from the Visible Human dataset. With the same load protocol, differences in muscle geometry produced different amounts of hypertrophy. Part of why two people on the same program diverge is not effort — it is the shape they started with.
There is a second result worth keeping. The coupling between geometry and signalling buffered local signalling heterogeneity: a spike in signalling at one point in the muscle still produced a smooth change in overall shape. It is a mechanical explanation for why muscle does not grow in lumps.
This is a computational model, not a training trial in humans. Read it as a framework that generates predictions, not as a set of predictions that have been validated.
What this means for your Big 3 log
Systems with feedback do not climb in straight lines. Fast early and flattening later is the normal shape, which means comparing your Muscle Index week to week will nearly always disappoint. The same data read across quarters tells a different story.
And size and strength never move at the same speed. Put this alongside mTORC1 is not a volume knob and strength gains outrun muscle mass, and this model is the piece that explains why the size curve flattens on its own.
Frequently asked questions
Why does muscle growth not continue indefinitely?
Because of feedback between protein synthesis and muscle size. As the muscle grows, the same load produces less mechanical stimulus per unit of tissue, so growth signalling through the IGF1-AKT-mTOR-FOXO pathway falls and hypertrophy converges instead of running away.
How does the model calculate hypertrophy?
Tissue mechanics are handled with a transversely isotropic hyperelastic model and intracellular signalling with a system of ordinary differential equations for the IGF1-AKT-mTOR-FOXO pathway. A growth tensor links the two: signalling dynamics set the tensor, and the tensor drives changes in muscle cross-sectional area.
Why do two people on the same program grow differently?
Simulations run on real muscle geometries from the Visible Human dataset produced different amounts of hypertrophy from identical load protocols. Differences in muscle geometry are one source of the difference in growth.
So should I just add weight when I stall?
That follows from the model's logic — as the muscle grows, stimulus density under the same load drops, so raising the load or changing the stimulus is what restores the growth signal. It is a prediction from a computational model, not a result from a human training trial.
Can this be used as a training prescription?
No. It is a multiscale computational model rather than a training trial in humans. It provides a framework for forming and testing hypotheses about training design; it does not deliver a validated prescription.
Source: PubMed