I'm interested in seeing if there's a physio/psycho/neurological differences between people who find her smile mysterious and people who don't, because to me, the only mysterious thing about her smile is that the media makes such a big deal of it.
I didn't get it either. Finally, this article at least explains the mystery, and now that I'm looking for it, I see the point. I still process it as a "small smile", though, where the ambiguity is moved into the representation of the emotion and not the visual processing. "Smile" is not an atomic face state, there are many gradations within it.
I suspect the people finding it "mysterious" are suffering from an aliasing error, basically, binning a complicated thing (the human face) into far too few bins.
Personally, I've never seen the Mona Lisa's smile as that complicated. Her mouth is smiling, her eyes are not. When you look at the bottom half of her face, you see a smile. When you look at the top half, you don't.
When we fake a smile, we smile with our lips but not our eyes. The Mona Lisa is doing the reverse: her eyes are smiling but her mouth is actually quite neutral.
There are also lines in her cheeks that make it look like her mouth curving upwards, but the lips are actually quite flat.
Try looking directly at her eyes, then down at her mouth. Her expression seems to morph.
Reminds me a little of the SIGGRAPH paper on hybrid images, where the distance at which you view an image determines, for example, whether you see a smiling or frowning person:
So did Leonardo intend to sow so much confusion in the brains of viewers, not to mention scientists? Absolutely, Otero Martinez contends. "He wrote in one of his notebooks that he was trying to paint dynamic expressions because that's what he saw in the street."