How do we know that the quanta of a quantized Dirac field describe elementary spin-1/2 fermions (or point particles of spin-1/2) and not composite spin-1/2 fermions (or extended structure of spin-1/2) such as a proton or a neutron?
Response to comment Surely Dirac equation doesn't describe the composite spin-1/2 particles. This is because, if it were so, the QED Feynman rule (derived assuming that the interacting charged fermions are described by Dirac theory) for the proton-proton-photon vertex in the case electron-proton scattering would be $-ie\gamma_\mu$. But it is not the case.
Response to the answers I was reading Halzen and Martin. There they said that the vertex factor $ie\gamma^\mu$ cannot be used for protons because unlike electrons it is an extended structure. It is an effective vertex proton-photon vertex $ie \Gamma^\mu$ which contains the information that the proton is non-elementary. You may find the same thing written in the paragraph above equation (345), in the notes here.
But if I understand the existing answers correctly, they are suggesting that $ie\gamma^\mu$ is replaced by the effective vertex $ie\Gamma^\mu$ not because the proton is an extended object but because we are taking loop corrections into account.
I am confused now. What is the correct reason?