I wonder if "hardening" to certain stimulus would be an 'answer'? A person might freak out after their first corpse, but after their hundredth they're not likely to care. They might still have a severe problem with being assaulted however. It's one thing for it happen to other people, quite another to have it happen to you personally. If I have it correctly, currently Dwarves care about everything, from bread to their mothers, on a singular scale. Splitting might dull or negate their response to specific stimuli over time (he said putting on his armchair psychologist's hat) while leaving others open. It could also have a knock on affect of differentiating them personality wise a little more. One Urist shakes the mandible of the Giant Cave Spider, other other runs screaming in the other direction.
That said, I imagine splitting their concerns into subcategories is a significantly hefty task and would probably require a a slew of changes to god knows how many files. The other problem is whether there are enough broad categories of trauma for a dwarf to experience over their lifetime to make splitting their concerns like that worthwhile. Currently they're mostly concerned with death and beer and not much else. One person might be able to cope with a dead body, but will utterly breakdown if their lover walks out or cheats on them. Dwarves, at current, don't have the capability to be concerned about their close connections beyond those people dying. Then, I suppose you move on to the problem of psychological/emotional callousing affecting other aspects of a person's psyche. Been through some rough times? You're probably less likely to respond as dramatically as a more "sheltered" person, though the specific circumstance is wildly different from your specific traumatic event. At the very extreme end of the spectrum I suppose it's not entirely dissimilar to the dwarven 'doesn't care about anything anymore'.
I suppose this goes to show that there's a reason people spend years in university for this stuff. it all gets very complicated very quickly. I get the impression that you could spend a year or more just working on psychology simulation alone and still end up with a system that would end up at once recognisable and simultaneously awkward. Either way the implementation of memories alone is, I think, something very few developers, if any, have really done before. As far as solutions to problems goes it's inspired so I'm excited.