When I worried that the correlation between testosterone and politics means that political opinions are hopelessly biased by emotions, it was pointed out to me that it could be valid for emotions to affect values if not probability estimates.

It just so happens that deficits in emotional processing are usually linked to gambling disorders and in the lab are linked to difficulty in distinguishing good bets from bad bets. I suspect that most decisions are less "probability estimates" and more based on approach/avoid emotions. (And incidentally, political orientation is also linked to differences in approach/avoidance behaviors and differences corresponding brain regions. If you thought the testosterone link were bad wait till you read about the amygdala links).

In short, as far as humanity goes emotion basically is totally inextricable from accurate probability estimates, and differences in emotional processing are probably responsible for the variation in viewpoints that cannot be explained by variations in life experience.