Huh. I wouldn't expect unmodified humans to be able to resolve value conflicts in a confluent way; insomuch as my understanding of neurology is accurate, holding strong beliefs involves some level of self-modification. If prior states influence the direction of self-modification (which I would think they must), confluence goes out the window. That is, moral arguments don't just shift value estimations, they shift the criteria by which future moral arguments are judged. I think this is the same sort of thing we see with halo effects.