I don’t know. I think blending modes may operate in a different way, and thus create a mental distinction when you think about them, but I have yet to come to a result that was any different or better by using them.
I think they are a computationally sloppy way to get to the same thing. You’re making the machine process 2 copies of the image, just to get more saturated reds in the midtones, or richer shadows or something.
I am only making my case as a chance to be disproven and learn something. But this is where I stand now.