Just for the information, the Tobbii is recording the glint of an IR light on the retina. So no arousal information from pupil dilatation.
Another information : you can't infer what people are doing just by looking at their mouse trail patterns (because in the end, it's just that). You need to correlate that with what's onscreen at the moment they were looking.
And no computer, available at the right price for the public, can record a fullscreen video (let say something starting from 1440x900) while generating the graphics at the same time, without stuttering (think about fullframe acquisition, storing in RAM, realtime compression and disk access time, while generating graphics...)
And, just for the record, I worked with those, so I don't say they are not useful : for disabled people, hemiplegia/tetraplegia, it's a wonderful tool. It's the difference between locked in your body, and interacting at least with a computer to say and do things to the world...