and another thing. Going to PubMed to find some cite to support my logic chain because I don't have it in my reference manager database already doesn't break the flow either.
It enhances and reinforces the thrust of the point you are contemplating, and working on making concretely in academese, because you refresh yourself on all the relevant and irrelevant titles/abstract, perhaps find a new paper or two and remember an old one. That allows you to wrap up that paragraph or page in a trice so that you can move on.
Instead of spinning your wheels writing up a bunch of sheist that you later have to "revise" because your memory of the paper(s) that you thought supported a particular point was erroneous.