Eye-tracking

Eye-tracking was probably the most challenging part of this thesis. In this section, I hope to leave behind some useful things I have done, so that I can expedite future eye-tracking studies.

Using the eye-tracker machine in the ACS lab, I discovered that the file format most helpful to me was the Plane Intersection Coordinates (PIC) format. This gave the X-Y locations of the eye in relation to the screen. The analysis I did focused on turning these locations into fixations. I have borrowed algorithms for fixation detection and have created some basic software to aid analysis.

• fixfunc.c - This file is freely available on the Internet. It converts x-y locations into fixations. It requires you to create your own "hook" to feed data.
• fixget.cpp - This file provides that hook! A very simple program that feeds in x-y points and prints out the returned fixations. WILL NOT WORK AS IS, because you must update the pathnames of input/output files. Essentially you can use a .PIC file as input, except you must erase all the header information.
• fixview.cpp - An OpenGL program to play back fixations. You must specify some inputs, to determine the input file and the phone you wish to display the fixations on. WILL NOT WORK AS IS, because you must update the pathnames of the input/output files. Additionally, you will need to set up an OpenGL development environment -- information on this is widely available on the Internet.
• Phone images - The backgrounds you'd want to play telephone dialing fixations over.

I include here a lot of largely raw, sometimes organized data. Some of these data can be played back using fixview.cpp.

• Experiment trials - Actual experiment data, these are ready to be played back using fixview.cpp
• "Aggregate" data - This is what fixget.cpp would produce as output. I would take these files and split them into 10 pieces (as in "experiment trials) and use the smaller pieces as fixview.cpp input.
• Totally disorganized data - Perhaps you can use this to determine a better format than PIC. I don't expect that this is very useful.
• Fixation playback data - Fixation data placed over the phone images.

This CD and documentation created May 3, 2003 by Andrew R. Freed to aid cognitive modelers and friends of the Penn State Applied Cognitive Science lab.