It is done. After quite a few months, numerous pilot scans, failed attempts, reconsiderations and some despair: we settled on a final acquisition setup for the phase 2 recordings of the project -- full-brain fMRI at 3-Tesla (Philips Achieva 3T dStream, 32ch) with simultaneous eye-gaze tracking at 1000 Hz (SR Research Eyelink 1000), as well as cardiac and respiratory trace measurements.

In the end, we had to built a custom wooden frame to be able to mount the camera above the projection screen while getting enough light from the infra-red spotlight through the space between the rear projection screen and the scanner tunnel. We also learned that mirrors for consumer rear-projection TVs are an excellent 50 EUR alternative to "professional" ~600 EUR front-reflective IR mirrors.

But I think the investment was worth it. With our setup, we can now do simultaneous eye-tracking with a fairly good quality that is independent of the stimulus lightness and dynamics. With a good subject it possible to get data with a spatial uncertainty of 0.3 visual degrees and virtually no drift over a 15 min recording session. Our only remaining problem are beautifully long eyelashes...

Custom wooden camera mount at the back of the scanner.

Camera and IR light setup with our custom wodden mount at the back of the scanner.

I am also pleased that we were able to increase the number of cables, gender changers, and signal converters in the acquisition setup dramatically. This leaves no doubt that real science is in progress. See for yourselves:

Stimulation and recording setup at the scanner console.

Scanner, stimulation machine and eye-tracker computer connected by a beautiful set cables of all colors, lengths, and widths.

We have now started recording data for the audio-visual "Forrest Gump" from the participants that had already volunteered for the recordings of the audio-movie in the 7-Tesla scanner. Fingers crossed ...