Instrument considerations in measuring fast eye movements

Abstract

The dynamic limitations of eye movement recorders can distort the measurement of fast eye movements such as saccades and nystagmic quick phases. In this paper, the effects of the bandwidth and noise of recording methods and the problems incurred by digital sampling are discussed theoretically with respect to the measurement of peak velocity and duration offast eye movements. As a practical example, a TV-based infrared corneal reflex system is examined and a method for calibrating it for peak velocity measurement is described.The measurement of eye movements (EMs) is becoming increasingly relevant in fields other than those concerned with the analysis of the oculomotor system itself. For example, correlations between unusual EMs and neurological pathologies have been made by Hamann (1979) and Zee, Optican, Cook, Robinson, and Engel (1976). There is also a large body of literature on the relationship between reading and EMs (e.g., Levy-Schoen & O' Regan, 1979). Furthermore, EM measurement is beginning to be usedin the developmental studyof the human infant (Aslin & Salapatek, 1975;Hainline, 1981). Each application of EM measurements has its own technological priorities. Some research questions require that the measuring system have great speed and resolution, whereas others require that intrusiveness upon the subject be minimized. In most current systems, there is a tradeoff among resolution, intrusiveness, and, of course, cost.The various instruments and their meritshave been discussed elsewhere (e.g., see Young & Sheena, 1975, for a comprehensive review). However, for efficientprogress, it is essential to be able to compare data from different instruments in different settings. To this end, McConkie (1981) stressed the need for investigators to report the quality of their data, such as sampling rate, drift, noise, accuracy, and short-and long-term repeatability. But such reporting necessitates an intimate understanding of one's EM recording instrument.The needfor calibrating an instrument so that the direction of the subject's eyeis precisely mapped onto stimulus space is self-evident, and there are various schemes for doing this (Bullinger & Kaufmann, 1977) Goldman-Rakic, 1981).It is less appreciated, though, that even an accurate mapping procedure does notcalibrate the instrument for all purposes. For example, during measurement of fast EMs such as saccades or nystagmic quick phases, the bandwidth and noise of the recording system can degrade the recording of dynamic responses. The most commonly used parameters for describing fast EMs are peak velocity (PV) and duration. In this paper, wediscuss the general effects of a system's bandwidth and noise on the measurement of these twoparameters of fast EMs. As an example, we present the results of a dynamic calibration of a TV-based, infrared, corneal-reflex system (Hainline, 1981) for the measurement of PV. Although this kind of instrument is certainly less than ideal for recording fast EMs, it has. the advantage of being nonintrusive to the subject, which makes...

Estimating peak velocity of rapid eye movements from video recordings

“…The interpretations reported here stand in stark contrast to the discouraging results reported by Harris et al (1984), but there are conspicuous differences between the two approaches, which presumably can account for the discrepancy. Their recording system involved automated evaluation of distance between the center of a corneal reflection (first Purkinje image) and the pupillary margin, rather than direct measurement of limbus position relative to an electronically imposed reference line in the pictures.…”

“…Theoretical analyses of potential artifacts associated with certain aspects of video eye tracking have recently been published (Haslwanter & Moore, 1995;Moore, Haslwanter, Curthoys, & Smith, 1996). There is, in addition, a single previously published evaluation of the limitations of a video-based system for estimating critical dynamic properties of saccades, such as peak velocity (Harris, Abramov, & Hainline, 1984), and that article painted a very discouraging picture. For example, it indicated that peak velocity of 10°saccades would on average be underestimated by about 40%, with a very large "noise" level; such a conclusion seems intuitively plausible because ofthe obvious mismatch between the time course of eye dynamics and the 50-or 60-Hz sampling rate of video.…”

Human eye-head gaze shifts preserve their accuracy and spatiotemporal trajectory profiles despite long-duration torque perturbations that assist or oppose head motion

“…Off-line, the eye and head channels were further low-pass filtered (MATLAB, zero-phase digital filter, 60 Hz, 10 dB/octave) This is in keeping with the recommendations of Bahill et al (1981) and Harris et al (1984) for the accurate measurement of peak velocity in head-fixed saccades. For our experiments, 60 Hz is a conservative value because for our large (Ͼ50°) gaze shifts the time from saccade start to peak eye deviation was more than four times that of head-fixed saccades of the same amplitude (see Fig.…”

Eye movements in reading and information processing: 20 years of research.

“…Eye-tracking systems are currently in use that rely on (a) surface electrodes (which are fairly good at measuring saccade latency but not good at measuring location), (b) infrared corneal reflections, (c) video-based pupil monitoring, (d) infrared Purkinje image tracking, and (e) search coils attached like contact lenses to the surface of the eyes. Although there has been some discussion concerning the measurement, evaluation, and reporting of eye movement data (Harris, Abramov, & Hainline, 1984; Heller, 1983; Inhoff & Radach, 1998; McConkie, 1981; McConkie, Wolverton, & Zola, 1984; Nodine, Kundel, Toto, & Krupinski, 1992), no measurement standards have been adopted, and many methodological issues remain unaddressed or unresolved (see Inhoff & Radach, 1998, for a good discussion of these issues). Despite this fact, most of the important findings discussed in this review have been replicated across different labs.…”