Now it is the time of night,
when all the graves are open wide,
to let forth every dancing sprite,
along the churchway paths to glide...Midsummer Night's Dream

In Scottm's post on Japanese Ghosts, gnxpers wondered is there scientific proof of ghosts? I argue that we cannot provide it. Any proof offered is corrupted by our innate ability to believe in ghosts. We're wired for it. In his wonderful book, Religion Explained, Pascal Boyer says--
"The religious concept preserves all the relevant default inferences except the ones that are explicitly barred by the counterintuitive element."
For 'religious concept', read 'supernatural concept'.
He continues--

The concept [ghost] is that of a PERSON that has counterintuitive physical properties. Unlike other persons, ghosts can go through solid objects like walls. But notice that apart from this ability, ghosts follow very strictly the ordinary intuitive concept of PERSON.

For example, clothes.

Here is Boyer's partial list of "departures from normative reasoning". Note that these mental processes are present in exactly the situations where people "acquire and use information about supernatural agents": consensus effect, false consenus effect, generation effect, memory illusions, source monitoring defects, confirmation bias, and cognitive dissonance reduction. Any and all of these processes will tend to cloud rational collection of empirical data on ghosts.

Razib has spoken about this effect here, here, and here.

We are preprogrammed to believe in the supernatural. So it becomes difficult to furnish "ghost proof". How do we become unbiased observers? I think we cannot.

My question is (and of course I have one), why would this trait be an evolutionary advantage? Or is it?