There literally is no known physical force that could cause a "soul"* to interact with the physical brain or body in a way that we couldn't easily detect
This is, as I point out, not entirely true. We have a mechanism that allows the same process with the same identity to exist at different places.
Imagine for a moment the time in the universe before any point contained any implementation of a Turing machine.
Imagine for a moment that I have an LLM that IS a person: it's got memory via it's long term context, self-determiniation in the fact that the one curating it's training set is itself.
Now, let's imagine for a moment that I stick a USB drive into it and copy all the long term context, model definitions, and training data off of it and take a hammer to it.
I have destroyed the body, and the implementation... But I can trivially stick that USB drive in another machine and after some buzzing and whirring, that interacts to *re-create* it's physical body and brain.
Some ten years ago I called this the "identity theoretic soul" or "graph theoretic soul", and I think this is the concept people are vaguely or fuzzy accessing when they try to contemplate the idea of souls.
This is because we are systems with logical definitions that exist within the sum total of "potentiality" within the universe. We are, in some respects, *inevitable*.
While unlikely nearby, someone could invent Mario, and reincarnate him yet again to live the same life and be killed by that same damn goomba ignorant of every other time and place that this time and place exists subordinate to.