Perhaps if it helps to clarify this, we might use the example of an adjustment layer in Photoshop, which has been given certain Blend If settings to restrict its action according to luminance values in the layer stack below.
And then consider what happens, if you copy such an adjustment layer from one PS document, into another. You will not get the same 'practical' results from Blend If, unless it is being given the exact same input to operate with. That is part and parcel, of embodying the same 'logical' result.
That kind of entity and that kind of logical operation, is probably the closest PS analogy, to a LrC local adjustment using a luminance mask. There is no mask bitmap involved, and nothing seen in the Channels palette. Just a layer that has been given dynamic rules, making it act in a certain selective way.
Extending this, something that there is no analogy for in PS: a Subject selection is merely an instruction of the same kind. The practical fruition of which must equally await the dynamic application of that instruction, onto this new image's own content. The requirement to approve recalculation is different, but harking back to the BlendIf example, if THAT did not recalculate itself continually and automatically to whatever is below, we would be dissatisfied with that.
In PS when we want a static mask we do that, and when we want a dynamic mask we do that. LrC / ACR is IMO somewhat ahead of PS layers in what you can do dynamically / parametrically and how efficiently; its focus is parametric and even, its brushed masks are resolution independent, because that is how it makes most sense to navigate its particular task landscape.
PS is way ahead of LrC when it comes to static bitmap masks and what you can do with those. That's clearly a very different task landscape being navigated by that. Hence seeking to map or transfer the tactics and solutions from one onto the other, is IMO going to be prove as pointless a quest, as it would be trying to mount oars onto a motorbike.
... View more