Eye tracking gear is pretty cheap I think. All you need is a webcam capable laptop and github account. They could have done this with a video playing instead of a steady photo, so that you can see where the participant would focus on at realtime response. But make sure the experience is immensive like 5 mins long video or so. Also please do temporal analysis instead of averaging everything out. Because one with a normal mind could have figured out some people will be nervous at first looking at those dark places but then feel safe afterwards. You don't see this if you ignore time as a variable.