If you are doing mathematics, you probably should be discussing your ideas with others, even if you are the sole author of the paper. They can contribute helpful ideas, some of which are wrong and should be ignored, but some of which can save you months.
Create a novel on your own? Go for it. (Though even there, other people can contribute interesting suggestions.) But then, if you're going to publish it, you're going to run it through an editor...
Even the most "solo" proofs rest on a foundation built by others (notations, concepts, lemmas, entire frameworks invented decades or centuries earlier), so nothing really is created in isolation.
It's also right to point out that mathematics can be done in solitude — but in my experience, that solitude is anything but passive. It takes a kind of disciplined internal dialogue—working through examples, forming your own structures, asking endless questions. (I quite like the way Paul Halmos puts this: “Don’t just read it; fight it.”)
This is a common thought. Worth noting that many major literary artists emerged from "salons" or social groups of writers, from Hemingway in Paris to the Beat poets in SF, or even Shakespeare at the Globe, there has always been greater impact from writers who worked and communicated with their peers.
Create a novel on your own? Go for it. (Though even there, other people can contribute interesting suggestions.) But then, if you're going to publish it, you're going to run it through an editor...