It's pretty long, but if it interests you, it's worthwhile to read it. It goes into how we now have visual proof that people don't notice continuity errors that much. And how eye gaze can be controlled on a long take, where the viewer is free to look wherever they please for a lengthy time, in addition to scenes using quick cuts with blatant targets of visual interest.
They used a device to track how pupils moved on 11 people while they watched a scene from There Will Be Blood. It's pretty neat to see all of the circles move around, and point out how common it is for completely different people to draw gaze in the same places at the same time (especially in a long, quiet scene like this):
<iframe src="//player.vimeo.com/video/19788132" width="400" height="180" frameborder="0"></iframe><p><a href="http://vimeo.com/19788132">There Will Be Blood with gaze locations of 11 viewers</a> from <a href="http://vimeo.com/visualcognition">TheDIEMProject</a> on <a href="http://vimeo.com">Vimeo</a>.</p>
Then they went one step further and created a "peekthrough" version of the video. This only shows the areas that people are looking at, so that you get an even better idea of what is actually in direct focus for the viewer. It works on a hot/cold color scheme with the red areas concentrating the most gaze:
<iframe src="//player.vimeo.com/video/19677876" width="400" height="180" frameborder="0"></iframe><p><a href="http://vimeo.com/19677876">There Will Be Blood + eye movement peekthrough</a> from <a href="http://vimeo.com/visualcognition">TheDIEMProject</a> on <a href="http://vimeo.com">Vimeo</a>.</p>
You would pick the movie I found least enjoyable of all the films I've seen in the past decade. The pace of it and the dialogue and plot made me want to gnaw my leg off to escape. I don't know that a scan of my gaze would match up because I frequently got bored with the scene and my eys would wander.