I’ve almost finished my data exploration tool. Well, finished is a very ambitious term. I’ve got a functioning prototype, a proof of concept with which I can do some rudimentary explorations. The bGaze too provided by SMI Vision is a fine tool, but for group experiments with multiple subjects and multiple trials, I have found it a bit cumbersome to rapidly explore different combinations of groups, subjects, and time variables. The tool I am building is shown below.
It will allow me to import, at once, any number of subject data files. From those, it extracts the point-of-regard data and the pupil diameters. On the display at the top left it shows each point of regard for the selected subject(s) and the selected trial(s). The subject list is shown at far right, and the trials list is shown in the middle. Between these two lists is a time interval list.
My trials, at least the portions in which I collect data, are between 5 and 20-s in duration. The eye tracker then collects 60 lines of data for each second, thus it is important to be able to quickly examine data over various intervals to determine what will be the optimum interval to use for the final data extraction and analysis.
The interval selector will allow me to choose the interval into which I want a trial divided. It can be modified between .5 seconds up to the length of the trial. So, for example, I can select a .5-second interval in the box at the top. If the trial was, for example, 5 seconds, then below the interval adjustment box will appear the 10 intervals available in the selected trial (e.g., 0-.5, .5-1, 1-1.5…). Then, I can go in and scroll between different intervals within the trial rapidly and easily, observing the datapoints in the top left figure update in real time.
The window in the bottom left is very special. I’ll have much more to say about it in the next post. Very interesting things are being found in the data because of that window (which took me all morning to get right).
The next step will be to create an Area of Interest editor that will allow me to define areas of interest to quantify the extent to which attention is directed towards those areas. From my examination of the data patterns I’ve some ideas for some very interesting ways to define areas of interest that are going to be useful for tasks like mine where attention will change, in predictable ways, over time. SMI’s tool doesn’t have this particular functionality that I am planning (perhaps they’d be willing to make some trade arrangements for some of the ideas and functionality I’m developing into this tool…Thorsten… are you reading? )
So, the user (me) can load files, select subjects, trials, and intervals within trials, define areas of interest, and observe how the gaze varies across the scene image rapidly and easily, allowing the essential variables necessary for the final analysis to be quickly determined. If my next project is funded I may completely geek-out and build R into the tool so that analyses can be run without ever leaving it (but, I’ll probably never get to that point)
Next post I’ll show some more screens that show how participants’ attention, and arousal, change over the course of exposure, conditioning, and extinction of stimuli.