i don't think you can make such a generality.


although there is some dissent with the effect of subtypes on most forms of relations, it should be at least fairly clear that conflicting relations are least strained when both types are producing or accepting subtypes rather than a mix of those, to minimize polr bashing (or maybe subtype has no effect on this at all; who really knows?) most people, however, seem to agree that duality is "best" with subtypes of the same axis, whatever that means.

if true, these would be an example that the subtype theory does not apply uniformly to intertype relations, at least in the way you describe.