Edupunk Eye-Tracking = DIY Research

One of my favorite presentations at the 2011 Ohio University CALL Conference was made by Jeff Kuhn who presented a small research study he’d done using the above eye-tracking device that he put together himself.

If you’re not familiar with eye-tracking, it’s a technology that records what an person is looking at and for how long.  In the example video below, which uses the technology to examine the use of a website, the path that the eyes take is represented by a line.  A circle represents each time the eye pauses, with larger circles indicating longer pauses.  This information can be viewed as a session map of all of the circles (0:45) and as a heat map of the areas of concentration (1:15).

This second video shows how this technology can be used in an academic context to study reading.  Notice how the reader’s eyes do not move smoothly and that the pauses occur for different lengths of time.

Jeff’s study examined the noticing of errors.  He tracked the eyes of four ESL students as they read passages with errors and found that they spent an extra 500 milliseconds on errors that they noticed.  (Some learners are not ready to notice some errors.  The participants in the study did not pause on those errors.)

The study was interesting, but the hardware Jeff built to do the study was completely captivating to me.  He started by removing the infrared filter from a web cam and mounting it to a bike helmet using a piece of scrap metal, some rubber bands and zip ties.  Then he made a couple of infrared LED arrays to shine infrared light towards the eyes being tracked.  As that light is reflected by the eyes, it is picked up by the webcam, and translated into data by the free, open-source Ogama Gaze Tracker.

So, instead of acquiring access to a specialized eye-tracking station costing thousands of dollars, Jeff has built a similar device for a little over a hundred bucks, most of which went to the infrared LED arrays.  With a handful of these devices deployed, almost anyone could gather a large volume of eye-tracking data quickly and cheaply.

Incidentally, if you are thinking that there are a few similarities between this project and the wii-based interactive whiteboard, a personal favorite, there are several: Both cut the price of hardware by a factor of at least ten and probably closer to one hundred, both use free open-source software, both use infrared LEDs (though this point is mostly a coincidence), both have ties to gaming (the interactive whiteboard is based on a Nintendo controller; eye-tracking software is being used and refined by gamers to select targets in first-person shooters), and both are excellent examples of the ethos of edupunk, which embraces a DIY approach to education.

Do you know of other interesting edupunk projects?  Leave a comment.

Google Yourself

Can you find the needle? Google can.

This is going to sound a bit like one of those motivational books targeted at business managers, but I was struck by a couple of points in a recent article in Wired magazine on Google’s search algorithm (“How Google’s Algorithm Rules the Web“).  It’s got me wondering how I can Google myself: not in the sense of searching for my phone number and website, but in the sense of approaching my work in the way that people at Google have approached theirs.

Many people know the story of Google’s original innovation in web search, namely ranking pages by the number of links to them.  But this article details many tweaks that have been made since the original 1997 version.  These tweaks include weighting links from experts, personalizing results, and universalizing the search across many media including blog and Twitter posts.

In addition to some of interesting linguistic challenges Google is presented with in its search queries (note the differences in meaning in each word in New York, New York Times, and New York Times Square, for example), Google is using the data it gathers in searches to tweak its algorithm and constantly make improvements.  If someone searches for dogs and then searches for puppies, the algorithm learns that these words have a similar meaning.  If these words are found along with leash, fetch, and train on enough pages, the algorithm learns from that association as well.  Even more impressive is that Google is working on making many of these improvements all at the same time without shutting down.  One of Google’s coders likens this to changing “the engines on a plane that is flying at 1,000 kilometers an hour, 30,000 feet above Earth.”

Granted, few of us have the technical expertise or vast resources of a corporation like Google.  But, and this is the business-book-like part I promised, what are we doing in our personal spheres of influence to assess and improve what we are doing?  Is there data we can gather about our students’ experience?  How can we manipulate that data and what might it reveal to us?  How are we acting on the information we find?

I was recently talking to a student about the perception that students’ time is better spent on preparing for standardized tests than classwork.  My explanation that the best way to improve test scores is to do the classwork often falls on deaf ears.  But the good news is, we have the data to determine if that’s true.   If I can pull those numbers together and present them to these students, will I change their minds?  Maybe not, but it’s worth a shot.