Even when their introspections are uninformative, people still give confident descriptions of their mental processes, being unaware of their unawareness.
This indicates that the data are so uninformative about the parameters that the prior probability of arbitrarily large values still matters in the final answer.