Have you ever had that creepy feeling when you go to the gym and you catch someone glancing at you in the mirror? Have you ever noticed how professional poker players were sunglasses? Have you ever secretly given a cue to someone with your eyes? What about when people roll their eyes or look away in shame? Our eyes make over about two to four movements a second and over 100,000 movements per day. We rely on our eyes so much but are often unaware of its activity. It is often said that the eyes are the windows to the souls. At the Rochester Institute of Technology (RIT), eye tracking research is providing windows into how people think about tasks and goals.
Two professors at RIT have been conducting research on eye tracking. Dr. Roxanne Canosa is an assistant professor of the RIT Computer Science Department who recently earned her PhD at the Carlson Center for Imaging Science at RIT in September of 2003. She completed her dissertation under the direction of Dr. Jeff Pelz, an associate professor of the College of Imaging Science who received his PhD from the University of Rochester in Brain and Cognitive Sciences. Although both professors have done research in eye tracking, their ultimate goals are very different. Dr. Pelz’s research with eye tracking is to understand how the human brain and its cognitive processes – how we gather visual information and use it to plan and guide our actions. On the contrary, Dr. Canosa’s research involves the same goal but modeling it on an artificial system.
A lot of information is contained in a picture. As the saying goes “a picture is worth a thousand words.” However, not all words are important. For example, go up to some flyer posted somewhere and focus on it. Now try reading another flyer close by without moving your eyes. Your eyes focus on what is right in the center of your vision, but you can barely read what the other flyer says. Your brain interprets the surrounding information as non-essential to your task much like the sensation of your clothes against your skin. It just simply ignores it. Even though this process seems passive, our eyes are very active and tightly tied to the planned behavior and action.
Canosa’s purpose to extend this human process to an artificial brain requires the same strategy. The computer need not digest all the information it is receiving. Creating this artificial brain requires data that heavily relies on how our eyes move and what they see. To emulate the brain and its visual process requires creating a model that can interpret an environment. Some models are able to distinguish colors, brightness, and edges of objects. However, these attempts do not accurately portray how an eye moves and what it sees. Combining that model with eye tracking can provide a more intelligent module that can be applied to a wide variety of fields.
The purpose of Canosa’s experiments is to gain insight to the interaction between vision and action. Rather than conducting research in the confines of a lab, Canosa decided to capture data in natural environments. One main problem was that the giant eye tracker in the lab could not be carried outside (much less worn by the subject) to perform tasks. This led Canosa to build a portable eye tracking system from scratch. The subject carries the equipment in a backpack and dons the headgear to wander around campus and proceed with experiments almost like some blatantly obvious secret agent.
The eye tracking headgear contains a module that tracks the eye. A near-infrared light source is used to illuminate the eye. After it hits the eye, the light is reflected back into a mirror and into a sensor that records where the center of pupil is. There is also a camera on the front of the headgear that records a video of the scene where the subject is. The control unit then takes the information from the sensor and computes a line of gaze. The line of gaze is displayed as a cursor or crosshair as an overlay on the video scene to indicate what the subject is looking at.
Canosa’s subjects proceeded to carry out tasks such as having a conversation with someone, sorting or counting items, and reading posters and forms. The data gathered reveals a general trend for each of the different tasks. Visual activity ranged from low to high depending on the task. Tasks that are very clear, such as reading and counting, require lots of visual attention. Tasks like these cause the mind being “restricted” to that task because the goal is so defined. In contrast, tasks like walking down a hall or having a conversation on the phone do not necessarily require a lot of visual attention. This allows the mind to wander and spend its time looking at other things. Sorting cards or following Lego instructions lies somewhere in between. Both visual engagement and contemplative thought is required at the same time. These tasks require more of a strategy to complete which gives more freedom in the ways to complete them.
What all the results boil down to is how much our eyes fixate on certain objects during certain tasks. Our eyes tend to fixate on objects that are task-relevant for longer times. There are also instances when our eyes tend to fixate on other objects in the environment that are not necessary in attaining the goal. In Canosa’s experiment, subjects’ eyes fixated on the floor while trying to find a restroom even though the floor is not relevant or particularly conspicuous. The fixation on objects depends heavily on the task and how much they stand out in an environment.
Research in eye tracking and movements can lead to developments in all sorts of applications. We can potentially learn how the brain works on a pre-conscious level. We can understand how deaf students divide their attention between the instructor and the interpreter and what impediments they encounter. If we can understand how some people use visual skills, it is possible to teach those skills to others. Using a computer model, applications could include those in the military such as helping soldiers locate objects in the environment. There are even some people who think that eye tracking will be a future user interface for computers, similar to the hand-tracking computer in the movie Minority Report. There are so many things that eye tracking can be used for; it is only limited to whatever the imagination can conceive.
