Eye Tracking and the Internet
An interview with cognitive psychologist Dr. Albrecht Inhoff  

What is "eye tracking"? How does it work?

Most eye tracking systems monitor reflections of infrared light that is reflected from the surface of the eye (cornea). Reflection of this light may differ as a function of where the eye looks. Most retinal reflection systems are pretty good in detecting the presence of an eye movement and perhaps even its size. Determining exactly where the eyes look (i.e., high spatial accuracy) and keeping the signal drift-free are, however, major challenges. There are some systems that monitor the full eye or that monitor more than one reflection from the eye, e.g., the cornea and the top of lens, that provide a relatively drift-insensitive high-resolution record.

What does eye tracking tell us about human eye movements?

It specifies parameters of eye movements, including typical movement sizes, movement velocities, and binocular coordination. My sense is that eye tracking can also be useful in discerning movement abnormalities.

Briefly describe what eye tracking tells about human cognitive processing.

There is converging evidence --at least from the area of reading research--that the planning of saccades and the duration of fixations that intervene between successive saccades are determined by on-line computations. The duration of fixation durations, in particular, has been linked to the ease of a wide range of ongoing cognitive computations. What eye movements inform us about cognition is, of course, a function of our experimental setup.

How has eye tracking been used to study how people read?

In a typical experiment participants are asked to read sentences or passages of text while eye movements are recorded. The record of eye movements is then used to compute where readers looked and for how long they looked at selected segments of text. Say you want to know whether readers access more than one meaning of homographic words. You may then use sentences with homographs and control sentences without homographs and measure the time spent the target words in the two conditions.

Has eye tracking been used to study how people interact with online documents, such as Web pages?

I am not familiar with this field. I think it imperative to first formulate a question and a hypothesis before one goes and measures eye movements for this type of document. What exactly is the question? Without such guidance, eye movement research would be reduced to a fishing expedition.

How do you think eye tracking research could be used to make Web sites easier to use?

I am not sure there is anything specific about Web sites. You could use eye movements to look at basic visual layout questions. Are fixations, for instance, a lot longer when print is very small or when the page is too much cluttered with flashing advertisements? Are people spending less time looking at Web text than at other types of text because the web is generally a more superficial medium...?

How will eye tracking be used in the future?

Little do I know. My sense/hope is that the measure has a terrific future. People are relatively unaware of their eye movements and the measure is thus unlikely to be distorted by the participant. The measure is also quite sensitive to a wide range of cognitive and visual processes.

Thank you Dr. Inhoff for your time and interesting responses.

(This interview was conducted via email by John S. Rhodes)

Revised: 1 October 1998