EYE MOVEMENT RECORDING

We propose to use two eye tracking systems in our reading experiments. The first of these systems is currently located at the University of Connecticut and the second will be shortly installed at the Department of Psychology at the Hebrew University in Jerusalem.

Eye Tracker No. 1: The Eyelink II is the newest eye-tracking system utilized at Haskins Laboratories. It was developed at SR Research Ltd. The system consists of three miniature cameras mounted on a padded headband. Two eye cameras allow binocular eye tracking. The third, head-tracking camera is integrated into the headband and allows accurate tracking of the participant's point of gaze while providing for natural head motion and speech. Errors caused by headband slippage or environmental vibration are reduced by the system's use of corneal reflections in combination with pupil tracking. Pupil-only tracking is also available, if corneal reflection tracking is not possible. The available dark pupil tracking and off-axis illumination allow tracking of participants with almost any eyeglass prescription. Corneal reflection mode is also usable with eyeglasses.

EyeLink II control software integrates calibration, gaze position collection, head position compensation, and saccade and fixation analysis into one step. This allows the experimenter to focus on stimulus presentation and data analysis. In addition, the participantÕs gaze position is overlaid on experiment graphics in real time, allowing validation of calibration accuracy before recording and monitoring of data quality during data collection. EyeLink II, has high resolution (noise-limited at <0.01¡) and a fast sample rate (500 samples per second).

Eye Tracker No. 2: The eye movement tracker, which is about to be installed at the Hebrew University, is known as a "double-Purkinje-image (DPI) eye tracking system" designed by Crane and Steele (1978) and manufactured by Fourward Technologies, Inc. of Los Angeles, CA. This eyetracker illuminates the eye with a collimated beam of 0.93 um infra-red light and employs a complex combination of lenses and servo-controlled mirrors to continuously locate the positions of the first and fourth Purkinje images. Purkinje images are formed by light reflected from surfaces in the eye. The first reflection takes place at the anterior surface of the cornea while the fourth occurs at the posterior surface of the lens of the eye at its interface with the vitreous humor. Both the first and fourth Purkinje images lie in approximately the same plane in the pupil of the eye and, since eye rotation alters the angle of the collimated beam with respect to the optical axis of the eye, and eye translations move both images by the same amount, eye movement can be obtained from the spatial position and distance between the two Purkinje images.

The 0.93 um collimated beam of incident light is electronically-chopped at a 4 kHz rate and falls on the eye after reflection from a dichroic mirror that is transparent to visible light and offers an uninterrupted view of the stimulus field. This feature serves to reduce noise, permit the use of more stable ac amplifiers and avoid interference from ambient room lighting or light emanating from a display monitor.

The DPI eyetracker can measure eye movements with a frequency response of up to 500 Hz and with an accuracy of the order of 1 min of arc, or double that of the IRIS system described above. Therefore, the device provides ample precision for the purposes of studying eye movement during reading.

Eye Tracker No. 3: The eye tracking system in use at UConn is a system known as IRIS designed by Reulen et al. (1988) and manufactured by Skalar Medical, a Dutch company. The basic phenomenon exploited by the recorder is the difference in infra-red (IR) reflectance between the iris and the sclera of a healthy eye. A horizontal row of nine 950nm IR light emitting diodes (LEDs) illuminate the iris/scleral boundary on both the nasal and temporal sides of the eye at a common chopped frequency of 2.5kHz. A corresponding row of nine photo transistors located above the LEDs receives the reflected signal. Each LED has an output beam width (bounded by its half-power points) of 48 degrees while each photo transistor has a reception beam width of 28 degrees. One transmitter and receiver assembly is available for each eye and is fitted into a transparent Perspex (or Lucite) case attached to an adjustable lightweight metal head mount. Subtraction of the nasal and temporal detector signals from each assembly gives the eye position with respect to head position.

The chopped mode of IR light emission serves to minimize interference from ambient light sources and to enhance the signal-to-noise ratio by permitting a high level of IR incident energy during signal detection while keeping the time-integrated energy down to a safe level. Following subtraction of the nasal from the temporal signals, the eye position signal and its carrier are initially passed through a narrow second-order filter tuned to 2.5kHz. Then demodulation is achieved by convolving the signal with the synchronous chopped signal used to drive the LEDs and low pass filtering the result through a DC-200Hz filter to separate the low frequency eye movement from the higher frequency carrier signal.

The IRIS recorder can track eye rotation in the horizontal plane over a range of 30 degrees and resolve displacements of as little as two minutes of arc. Thus its resolution is about one seventh of the width subtended by the average alphabetic character on this page when held at 35cm from the eyes.