Context: As I am writing my first paper, I cannot determine what, when and how I should cite scientific literature. It is because I don't know why I should cite (or not cite) other work.
My fist 'newbie' approach was "Every assertion you write should be proven by a relevant citation, so that other can believe you". Given that "I know nothing", no one would then take my word for it and thus I must prove that what I'm saying is true.
However, this lead to overcitation (and, I must admit, over-generalization of results), and frustration since I must find a paper corroborating what I want to say for every sentence, or so
Then if I'm leaving this strict rule, I tend to think I am writing triviality, or affirming things I am not 100% sure. I also fear to unintentionally plagiarize, when I say something someone else already stated.
Question: In a scientific writing, why do we [have to/need to/should] cite?
Clarifying this might help answering corollary questions: what to cite? and when should we cite?
What to cite? Should I cite a paper (i.e. give credit) for ideas that are in its literature review, and that are totally not related to the core/added value of the paper? Should I cite every paper related to one field? Can I cite only not-that-much-cited source and not classic ones? etc.
When to cite? If I am writing an assertion that is general (e.g. "Polar bears have mostly white fur"), should I justify it?(!) Which is the criteria to determine when an assertion need to be proven by a citation, and when it is admited? When I define a key term for my paper, should I always rely on past definitions?
Disclaimer: Of course, I am aware of some obvious reasons for why we cite, such as giving credit and proving assertion, and I've read Arno's answer (who lists giving credit, for proof/evidence, and providing context as reasons) on a similar question. The originality of this question would be, to detail the link between the theory (why) and the practice (what, and when) of citing.
Furthermore, I am conscious that there are multiple questions in this post, especially in the bullet point list. However, these are just here for explanation purpose, and should of course be asked separately to get a specific and detailed answer.
* I know this should be done the other way (i.e. basing what I want to say on literature, and not looking for someone who could corroborate what I want to say)