This thesis describes the research of Herst et al. (2001) who examined the coordination of head and eye movements while participants searched for targets under natural conditions in which the head was free to move. This investigation was done because, while the technical difficulties in measuring these patterns of head-eye coordination in humans have been overcome, the results obtained over the last 35 years have yet to show what the typical pattern of head-eye coordination looks like. Herst et al. (2001) described the ‘natural’ temporal coordination of head and eye of four participants who tapped a sequence of targets arranged in 3D on a worktable in front of them. The results were not expected based on prior studies of head-eye coordination performed under less natural conditions. This thesis reinterprets the original (2001) findings and draws new conclusions in the light of new research on head-eye coordination conducted since then.
The maintenance of stable gaze relies on effective coordination of head and eye movements. And having control of gaze is necessary for gathering any visual information with high spatial frequency. As a result, understanding how the head and eyes are coordinated as we search our environment is a topic of interest in the oculomotor literature. Of more practical importance, the relationship between eye movements and information processing has been of interest for at least 145 years (Steinman & Levinson, 1990). Normally, the head and eyes move about freely and coordinate together in a richly structured and changing environment containing numerous possible targets. It becomes clear, then, that investigating head-eye coordination under natural conditions would be of great fundamental and practical importance. However, examining how the head and eyes coordinate under natural, unrestrained, conditions is difficult to do and few attempts have been made. This thesis presents the results of one such attempt by Herst, Epelboim and Steinman (2001).
With the understanding of the importance of examining the coordination of the head and eyes under natural conditions, and noting the lack of prior research, Herst et al. (2001) described the patterns head-eye coordination under natural, unrestrained conditions while performing a natural task. I begin with a brief historical overview of some of the more notable attempts at understanding how the head and eyes are coordinated. I present a more historical summary in order to show why the results obtained by Herst et al. (2001), and other experiments examining head-eye coordination under more natural conditions, are noteworthy. I then present the results obtained by Herst et al. (2001) and conclude with a discussion of the theoretical implications of my research. Since 2001, when our experiment was published, there have been investigations of the coordination of the head and eyes under natural conditions while the head free was to move. I reinterpret the original results of Herst et al. (2001) given the more recent experimental findings.
Recording with the Head Free to Move
The problem of studying head-eye coordination under relatively realistic, natural conditions had been solved for the rabbit by Collewijn (1977) even before Fuller’s review (1992). Collewijn solved the problem when he introduced the cubesurface field-coil, phase-detecting, magnetic eye/head recording system. Collewijn’s new method made it possible to record both head and eye rotations accurately while rabbits walked freely about in a relatively large field. Under these novel, rather natural, conditions, rabbits, who did not make saccades when their heads were immobilized, showed themselves capable of relatively stable patterns of saccades and head movements. Collewijn (1981) summarized his observations on the head-eye coordination of the freely-moving rabbit as follows: “It must be concluded that ... most gaze changes are achieved by combined eye and head movements. In many of these, head and eye movements are both saccadic and initiated simultaneously” (p. 19).
Steinman and Collewijn (1980) used this rabbit instrumentation to record human gaze-control as the head was actively oscillated about its vertical axis, while distant objects, seen through a window on the 15th floor of the Medical faculty in Rotterdam, were fixated binocularly. They reported several features of human oculomotor performance that could not have been anticipated from more conventional observations made with the head restrained in a visually-impoverished environment usually used in more conventional laboratory experiments. Fuller (1992) discussed two papers (Collewijn, Steinman, Erkelens, Pizlo & Van der Steen, 1992; Kowler, Pizlo, Zhu, Erkelens, Steinman & Collewijn, 1992) which used Collewijn’s recording technique after it had been implemented in a much larger, and more accurate, phase-detecting instrument called the Maryland Revolving Field Monitor (MRFM). This instrument was scaled-up sufficiently to make it more comfortable for research with human participants. Within these papers are descriptions of the control of gaze during both natural and unnatural visuomotor tasks. Once again, it was shown that oculomotor performance under relatively natural conditions is different from performance under the constraints that were ubiquitous before Collewijn’s important contributions to recording instrumentation.
Measuring Eye Movements While Tapping Targets
Binocular eye/head movements were measured while participants tapped (TAP) sequences of 3-D targets (colored LEDs) located on a worktable in front of them. The angular separation of targets was random, varying between about 1.5o and 35o of visual angle. The distance from the participants' eyes to the targets varied from about 50 to 90 cm, depending on where the targets were and how much each seated subject moved. All targets were arranged before the beginning of each trial and were stationary and visible throughout. Eyes were closed between trials. Each target configuration was tapped 10 times before a new randomly-generated configuration was presented. See Epelboim et al. (1995) for additional procedural details. Herst et al. (2001) examined the temporal relations between the onset- and offset-times of head rotations and saccades (relative to the head) which met the following two criteria for a coordinated head-eye movement: (1) the head and eye moved in the same direction, and (2) the horizontal components of both the head and eye were larger than 10º. The criterion used for saccade and head onset and offset was a horizontal velocity = 20% of its peak.
This criterion was chosen because Smeets et al. (1996), the prior experiment most closely related to ours (see above), had used “a very conservative threshold to detect the onset of movement ... velocity surpassed 50% of its maximum value.” (p. 436). We also desired a conservative criterion, but were able to set it lower (20%) because our temporal resolution was much better, viz., ~2, rather than 16 ms. Head and eye movements were considered to begin simultaneously if their onset occurred within ±8 ms of each other, also a conservative value, i.e., 4 times our resolution limit. In all, 2729 “coordinated” head-eye movements met these criteria (N/Subject: ZP = 637, HC = 649, RS = 720, CE = 723). The MRFM data used in these analyses consist of angular positions measured to 1 minarc with successive samples separated by 2.04 ms. Examples of the different kinds of head-eye coordination
Guitton, D., & Volle, M. (1987). Gaze control in humans: Eye head coordination during orienting movements to targets within and beyond the oculomotor range. Journal of Neurophysiology, 58, 427-459.
Hayhoe, M., Shrivastava, A., Mruczek, R., & Pelz, J.B. (2003). Visual memory and motor planning in a natural task. Journal of Vision, 3(1), 49-63.
Hayhoe, M., Bensinger, D., & Ballard, D. (1998). Task constraints in visual working memory. Vision Research, 38, 125-137.
Dissertation Done by Andrew Neal Herst, University of Maryland