For a good example, look at all of the “apparitions” of Jesus or the Virgin Mary on different objects, from toast to tortillas. I remember seeing one where a shadow falling across some rocks supposedly looked like JFK, but at another time it was clearly just a pile of rocks. The infamous “Mars Face” is another example. The point is that we have this built-in face recognition software, and because false positives are much better than false negatives, we tend to see faces where there aren’t any.
Most of those ghost pictures are just shadows falling in a strange way. I find it funny how some people see what is quite easily explained by shadows, grainy film, and a misfire in a our face detection software, and jump to the conclusion that it’s a ghost.
The same can be said of our ability to detect intention. Because we live among agents that have intentions, we have to be able to infer what those intentions are. We can get false positives from nature, infer design or intent where there isn’t any. There are some interesting consequences of this, but that’s for another post.