Generally speaking, scientists formulate a hypothesis, model some feature of the world, and check the hypothesis against the model. The model could be adjusted to better accommodate the feature of the world, the hypothesis could be modified in light of the model, and so the complexity of explanatory theories goes. I do not pretend to be an expert about the sciences, and perhaps some people would disagree with this characterization of ideas in the sciences. Nevertheless, I think it is true, at least in rough form.
Likewise, philosophers make their own models in the capacity of what are typically called thought experiments. As far as thought experiments go, philosophers construct a model, the thought experiment, which is some idealized version of the world, and then check a hypothesis against it. This process is in no wise as decisive or clear as with the sciences because the whole theory construction process here takes place within the mind of the philosopher as part of his or her own internal thought process, in discussion with other people, or perhaps in the course of written or spoken argument (in the non-pejorative sense, hopefully).
Regarding the theory of knowledge, the hypothesis that what should count as knowledge is "justified true belief" was tested by Edmund Gettier. He proposed two scenarios, two thought experiments, basically, where a person would be said to have, typically, justified true belief but not knowledge. I will just give the first scenario. In the first scenario, two men, Smith and Jones, are up for a job interview. Smith knows that Jones has ten coins in his pocket, which seems like a random fact but is relevant in just a moment. Smith also heard from the president of the company that the two men are doing job interview for that Jones was going to get the job. Smith thinks out loud, right there in the waiting room, while Jones is doing the interview, "The man with ten coins in his pocket is going to get the job."
It turns out, however, that Smith gets the job and he happens to have ten coins in his pocket. So regarding the thought, "The man with ten coins in his pocket is going to get the job," it's true and Smith has a rational explanation for why it's true, namely that he knows Jones has the coins in his pocket, and he heard from the president that Jones was going to get the job. In spite of Smith's having these reasons for his beliefs, the thought experiment demonstrates, contrary to the hypothesis, that knowledge can't be formulated in terms of justified true belief.
Gettier's thought experiments demonstrate both the point that knowledge cannot be formulated in terms of justified true belief but it also hints at the possibility that definitions for ordinary language concepts are not sufficient to capture all the features people want when they use the concept. In this case, the definition is not adequate to account for the concept KNOW.