Where do priors come from?

I thought that this paper, Probabilities in Economic Modeling by Gilboa, Itzhak, Postlewaite, Andrew and Schmeidler, David might provide an answer.
Abstract:
Economic modeling assumes, for the most part, that agents are Bayesian, that is, that they entertain probabilistic beliefs, objective or subjective, regarding any event in question. We argue that the formation of such beliefs calls for a deeper examination and for explicit modeling. Models of belief formation may enhance our understanding of the probabilistic beliefs when these exist, and may also help up characterize situations in which entertaining such beliefs is neither realistic nor necessarily rational.

Unfortunately not. This paper was more a summary of earlier works and if I thought it would provide an answer as to where priors come from, I was wrong. One of the suggestions by the authors was, yikes(!) to use multiple priors. In any case, the idea was intriguing enough that I should probably take a look some time in the future. References are in the paper.