The Just-So-Story I tell (while stating up front that it is a Just-So story) is this:

Our tendency is to anthropomorphize. Today, we angrily yell at our cars and computers when they fail us. Rationally we know they are unthinking hulks of metal, but we still ascribe malevolence when the real culprit is a broken, unsentient machine.

There are plausible reasons for humans to have evolved this trait. One of the most complicated tasks a human has to do is predict the actions of other humans. We need to be able to make allies, to identify deceptive enemies, to please lovers. I’m not an evolutionary psychologist and I should be careful when telling this sort of Just-So story, but I can easily imagine selection pressures that resulted in a powerful ability to draw conclusions about sentient creatures similar to ourselves.

And then, there was NOT a whole lot of pressure to NOT use this tool to predict, say, the weather.

We weren’t very good at predicting storms and snowfalls, one way or another. Weather is complicated. Even modern meteorologists get it wrong half the time. So would it matter much if an early human leveraged their empathy and narrative skills to predict natural phenomena? The rain would come, or it wouldn’t, regardless of whether we ascribed it to gods or “emergent complexity.”

So we told stories about gods, with human motivations, and we honestly believed them because there was nothing better.

Assuming this is true, (I don't know how to test for it) then once the basic ideas are in place, I think they'd be subject to memetic selection pressures that include "the person telling the stories becomes higher status for being a good story teller" which means people trust them more and other related things.

Most people (especially in very simple social environments, which is where this whole thing started) hold opinions for approximately good reasons. This is less true of very complex or infrequently occurring issues.

Nevertheless, individual opinions do constitute a certain degree of evidence. It strikes me as very likely that a mechanism for accepting second-hand, unverifiable information would provide a fairly substantial evolutionary benefit to a hunter-gatherer. For that matter, it provides some benefit to us today.

Religion is also a good source of individual motivation, if you don't happen to have an elaborated system of reasoning and ethics and metaethics handy. People tell stories about events (not all of them religious in nature) to contextualize them and predict them. Though that doesn't necessarily explain their tendency to spread in and of itself.