Do you mean guilt-the-emotion, or guilt as a causal relation? Neither of those are beyond the pervue of science. The only troubles you get -- the reasons scientists don't engage in this kind of theorizing -- is that the domain is both complex and culturally sensitive, and thus work doesn't receive funding and appreciation. It's not because it can't be done.
As you note, a large portion of it depends on value judgements. If I wilfully murder 1 person to save 3 from being accidentally killed, am I guilty? What if I take no action and allow 3 people to be accidentally killed?
Without attempting to Asimovly define the value of all possibilities, I'm not sure how science can answer these questions in any meaningful sense.
Without tumbling into a morass of complexity (e.g. a life is worth this because of GDP per capita and birth rate and lifespan, and taking an action is to be valued this way because of neuroscience and experimental psychology) that, should it be fallen into, would start to look more like ethical philosophy than post-Enlightenment science.
But then, that's why they were both the same field before experimental repetition was enshrined...
E.g. Trolley Problem https://en.m.wikipedia.org/wiki/Trolley_problem