Unless the test subjects were looking at a blank screen these results are misleading. Most websites are designed with visual focus at the top left (logo) and middle of the page, therefore people look there first <i>because</i> the design leads their eyes there. If there is no visually leading content on the page, I imagine the results would be much more random-looking.
I'm guessing that the study was conducted with readers of a left-to-right language. I wonder what the results of a study on right-to-left or top-to-bottom readers would be.
At art school, we were taught that in movies and paintings, a person's gaze tends to enter in the upper right, and loop around in a C shape, exiting the frame on the lower right. Therefore, it's the job of the artist to compose the image to sweep the viewer back up on the right side, to repeat the cycle.<p>We didn't do eye tracking studies to confirm this, but in analyzing intuitively "good" images vs. "bad" ones, the good ones invariably catered to this kind of eye movement.<p>Websites control the eye so differently (and more rigidly) than how representational images do, that I wouldn't expect the two to generate similar viewer interactions.
<i>>The conclusion we can draw from all this math is that overwhelmingly, people look at the top left of a website before moving on to other features. That’s where they expect navigational information to go; it’s where they expect to orient themselves. It’s also where you can capture their attention; and it’s where you should put your stuff.</i><p>And yet, GazeHawk's navigational info is at the top right. :)
I would be interested to see the same data used with Asian websites where they read right to left to see the differences. I wonder if the print style of reading varies from how they consume websites.
I think there's a chicken-and-egg issue here, as most websites are designed with left or top navigation.<p>It's also dependent on the subject matter and expectations of the subjects. Alfred Yarbus did eye tracking studies in the 60s that show that there is no "default" saccade path: <a href="http://en.wikipedia.org/wiki/Eye_tracking" rel="nofollow">http://en.wikipedia.org/wiki/Eye_tracking</a>