I remember some prototype studies made with fruits, where they tested (I think?) Mexicans and Iranians, and the results were different - one side picking apples as prototypical fruits, another bananas. So prototypes are culture-dependent, it's just that that one is widespread enough to mask the effect.

And the superhero example made me notice that they vary individually, too. For example I don't usually consume American hero media, but I do watch quite a bit of anime, so when the author mentioned superheroes my mind went like

• "superhero" → Goku (arguable, but that's what immediately came to my mind!)
• "costume superhero" → Sailor Moon
• "costume superhero with a cape" → Saitama
• "costume superhero with a cape, can fly" → Superman

The shift from Saitama to Superman was specially interesting because my mind actually backtracked into Goku, only to handle the conflicting info (not wearing costume, not wearing cape) afterwards. That fits rather well the metaphor used by the text, of prototypes as the centres of a web of associations, once you split the web into a bunch of overlapping "territories" - conflicting info forces you to migrate from one territory to another, and the reference to "flying" made me beeline for the prototype of "flying character" first, only then to check the overlap between "flying character" and "costume superhero with a cape".