Yep, just did a little testing on that. Looks like if any spatially overlapping place (whether or not there is a parent-descendant relationship) has a species marked as Introduced, it will display Introduced for the first such place it finds for a relevant observation. When I go in and manually change the displayed place to Native or Unknown for the species, then it just finds the next spatially overlapping place with Introduced status for the species, and displays that for the observation instead.
In other words, for an observation to display anything other that Introduced status, none of the places that spatially overlap it can have Introduced status for that species, regardless of parent-descendant place relationships. (My test observation: https://www.inaturalist.org/observations/27787778)
Makes logical sense with respect to how we define Introduced for a place. But it does mean that addition or removal of Introduced status on large places (currently possible by any user) can cause wild swings in the status displayed for large numbers of observations.
And it also means that any validation or error checking should be based on spatial intersection of place polygons, not on parent-descendant relationships, which involves a lot more computing resources. Ugh.