Pupillometry

[brandonhackney]

The eyetracker spits out pupil size data, and there is a rich literature surrounding pupillometry. So we should extract that data in a meaningful way and include some analysis of it.

[brandonhackney]

Invoking analysis('pupil') will extract the signal for all subjects, but analysis doesn't have a dedicated statistical method for it yet, so I let it fail by not including it in getGraphLabel. Please do the following:

• [ ] Decide how to analyze the pupil size signal
• [ ] Add some sort of code to analysis that handles the fact that it's a vector
• [ ] Add pupil info to getGraphLabel

[brandonhackney]

Turns out SR Research advises against analyzing this signal if it was collected during regular eyetracking, since they do not perform any sort of compensation for head position, eye rotation, etc. It's basically just the raw number of pixels that pass the pupil detection threshold. It would only be usable during a head-stabilized central-fixation-only experiment, which this was definitely not. You also would need to estimate the effects of video luminance, contrast, motion, etc. and remove them from the signal before trying to analyze effects of cognitive processes. I added some code that estimates luminance, but not anything else.

I'm leaving this issue open and the branch as-is, because those preprocessing steps could potentially be added later.