Stephen Law has a post on whether you can learn much about the human condition from literature. (Law seems to consider literature only to involve stories; this is certainly not the case, as we see when we get to, say, Lucretius's De Rerum Natura.
One of the things that strikes me is that parallel claims to Law's claims about literature can be made against the sorts of regimented arguments considered by philosophers (and are made fairly often, if my students are any indication). Arguments can be used to tell lies about the human condition; they can present things in ways that make us feel sympathy for causes we should revile, or make what is wrong seem right and normal. This is just a function of the fact that being an argument doesn't guarantee any particular content, just as being a story doesn't guarantee any particular content.
Likewise, when analytic philosophers discuss arguments, they discuss them in terms of clear-cut premises and conclusions; but real reasoning is almost never this simple, and it can, in fact, be difficult to tell whether something is supposed to be an additional premise, a conclusion, an enthymematic supporting argument, a clarification, etc. So when analytic philosophers deal with arguments, what they are dealing with can seem like it has no connection to real reasoning -- it's just paper-logic, simplified, idealized, and detached from the complexities of real reasoning, sometimes, in straw men cases, purely fictional. Philosophers regularly look at cleaned-up arguments rather than the original arguments; and sometimes in the cleaning essential features drop out. This is just a function of the fact that when philosophers regiment or formalize arguments, they are working with artifacts created for a particular purpose, just as storytellers are.
And, of course, the fact that philosophers deal with the same patterns of argument again and again is even more obvious than in the literary case. And this is just a function of the fact is that one of the things you want to know is what the patterns are and how they work in certain sorts of conjunctions with other patterns.
None of the claims tell us much about what we can learn, whether we are considering arguments or stories: the first just tells us that we can be deceived, and doesn't tell us about cases where we are not; the second just tells us that to understand something we go to a cleaner case to see how things work there; and the third just tells us that there are patterns, not what they are or how they are used. And so on. Literature, of course, only deals with plausibilities and implausibilities, and rarely anything more rigorous than that; but within the limits imposed by that constraint there still seems plenty of room to find worthwhile lessons.
But the question asked is an interesting question. What non-trivial truths can be learned from literature? Here are some thoughts.
(1) Sometimes stories are a good way of laying out a position in such a way that it looks plausible. An example is Eliot's atheistic view of religion, adapted from Feuerbach, in which it involves a failure to reconcile moral activity with the fictional projection of morality onto reality. This is very neatly laid out in her characterization of Savonarola in Romola. Likewise, Ibn Tufayl's Hayy ibn Yaqzan can be seen as in part an argument that philosophical reasoning needs for social purposes to be clothed in religious images. Ibn Tufayl's story is laid out in such a way as to make this plausible, by telling a story of a pure philosopher trying to teach his philosophy to a common mass of people to whom his manner of argument is foreign. And in the course of making it seem plausible, he can't help but touch on reasons why someone might consider it plausible: political resistance, psychological causes, the expenditure of resources (in time and effort) required to do philosophy, etc. That literature has this ability to display positions reached philosophically in ways showing their strengths is argued at greater length by Santayana in his Three Philosophical Poets.
(2) Sometimes stories are an excellent source of distinctions between phenomena that might otherwise be missed. Eliot provides another example of this, which I take from Miriam Burstein, the distinction between hypocrisy and what can be called 'Bulstrodism'.
(3) Sometimes stories are good ways to lay out why certain positions are implausible. It has been noted since Ryle at least that Jane Austen's novels do this fairly regularly, so that, for instance, the Crawfords in Mansfield Park are described in terms that make them come out very favorably (especially in comparison with crying, quiet Fanny), and that the characterization closely fits what would be expected by certain forms of eighteenth-century virtue theory. But, of course, what Austen shows in doing this is that those philosophical positions lead us to sympathize with the wrong people in the story, because they overemphasize social traits and underemphasize constancy.
But, of course, what you mostly gain from literature, as from life, is not a set of general truths, but a collection of analogues. In fact, all three of the points listed above contribute to this exquisitely valuable feature of story. Eliot's Tito Melema, with his easy amiability that in its very nature is the beginning of moral corruption, has real-world cousins; once you've read Romola you begin to recognize that there are many Tito-like things done in the world. And so it is with her Bulstrode, her Dorothea, etc. Reading George Eliot gives you a new and substantive vocabulary for talking about ethics precisely because you can use her characters, her situations, and the like, as a way of talking about the moral matters raised in the course of her story. And so it is with Austen, with Dickens, with Dante, with Homer.
Of course, none of this is the primary reason for reading literature, which is that it, like most contemplation of art, is an enjoyment excellently suited to rational animals, whether you take away much from it or not. But insofar as literature is used, or insofar as in reading it you happen to come away with things that you can use, there seems to be plenty that can be learned from literature.