Thanks! The idea of multiple umwelts is actually a slightly contentious one, at least between me and my buddy I discuss philosophy with- he holds forth that for any given organism there is only one umwelt, regardless of things like whether it integrates its sensory input or even if the organism is actually a colony organism like a siphonophore, where every organ is actually a different critter all working together. He even holds that an anthill only has a single umwelt, which is actually somewhat mainstream in umwelt related thinking.

Of course, umwelt as a concept is nearly 80 years old, and it starts to break down when it comes in contact with taxonomic questions. Actually defining what the boundaries to a species or a particular organism can somewhat break down some of the underlying assumptions of umwelt.

As for whether non-integrated sensory input would function as a cognitive barrier- I personally think that it depends on the degree of non-integration. If it has a central nervous system analogue that integrates at least some of the input, I think it could achieve higher intelligence even if a great deal of its sensory data isn't integrated. If it's unintegrated to the degree that box jellies are, I'm pretty doubtful.