[datameta]

Your suggestion is fully valid without needing the strawman you mention second.

So the proposition is to use images with eye-tracking? Or video?

Edit: In response to the disclaimer, I am trying to develop your idea for improving the study because I agree it could control for more variables.

[flir]

Video would seem to be an improvement over still images.

If we're fleshing it out, I think a wider array of college students (BYU may have its own issues) and adding some "neutral" spaces to the test might yield more information. (Although if being "wary" is habitual, the behaviour would probably still show up in neutral spaces?)