Eye movements: the past 25 years

Abstract

This article reviews the past 25 years of research on eye movements (1986-2011). Emphasis is on three oculomotor behaviors: gaze control, smooth pursuit and saccades, and on their interactions with vision. Focus over the past 25 years has remained on the fundamental and classical questions: What are the mechanisms that keep gaze stable with either stationary or moving targets? How does the motion of the image on the retina affect vision? Where do we look - and why - when performing a complex task? How can the world appear clear and stable despite continual movements of the eyes? The past 25 years of investigation of these questions has seen progress and transformations at all levels due to new approaches (behavioral, neural and theoretical) aimed at studying how eye movements cope with real-world visual and cognitive demands. The work has led to a better understanding of how prediction, learning and attention work with sensory signals to contribute to the effective operation of eye movements in visually rich environments.

Figure 1

(a) Eye movements during fixation of a stationary point target recorded with the Dual Purkinje Eyetracker. (b) Retinal montages of the foveal cone mosaic for three subjects. The black square represents the foveal center; the dashed black line is the isodensity contour line representing a 5% increase in cone spacing, and the solid black line is the isodensity contour line representing a 15% increase in cone spacing. Red dots are individual fixation locations. Scale bar is 50 μm. NM Putnam, HJ Hofer, N Doble, L Chen, J Carroll, DR Williams (2005) The locus of fixation and the foveal cone mosaic. Journal of Vision, 17: 5(7), 632–639. Figure 1. (c) Eye movements during fixation while the head is rotating. Traces show movements of head, right eye, left eye, and vergence (right eye – left eye). Image velocities were the same as the eye traces. From R. Steinman & H Collewijn (1980) Binocular retinal image motion during active head rotation. Vision Research, 20, 415–429; Fig. 1.

Figure 2

Smooth pursuit of moving shapes. Mean horizontal and vertical pursuit speeds averaged over trials for 1 observer. Rightward and upward values are positive and up. Colors indicate direction of motion of the stimulus. The moving square produced either purely horizontal or purely vertical eye movements depending on the direction of motion of the square. The tilted diamonds produced a more complex response, generating pursuit along both horizontal and vertical meridians for about 100–200 ms, showing the dependence of the response on the motion vectors of the individual edges of the shape. From G.S. Masson & L.S. Stone (2002) From following edges to pursuing objects. Journal of Neurophysiology, 88, 2869–2873. (Fig. 1)

Figure 3

Illustration of effects of learning on smooth pursuit. Subjects tracked two cycles of constant velocity (40 deg/s) motion (traces show target and eye velocity). After that, velocity decreased to 5 deg/s. Pursuit, however, continued to be influenced by the learned velocity. From G.W. Kao & M.J. Morrow (1994) The relationship of anticipatory smooth eye movement to smooth pursuit initiation. Vision Research, 34, 3027–3036. ( Fig. 10)

Figure 4

Examples of anticipatory smooth eye movements, showing pursuit (lower traces) beginning before horizontal motion of the target (upper traces). Stimuli (shown on the right) were discs moving downward within a Y-shaped tube. The disc entered either the right or left arm of the Y. The horizontal path was indicated by the visual barrier that blocked the untraveled arm. From E. Kowler (1989). Cognitive expectations, not habits, determine anticipatory smooth oculomotor pursuit. Vision Research, 29, 1049–1057. (Figs. 1 and 2.)

Figure 5

Illustration of a comparison of eye fixation positions (dashed lines, middle panel) with computed salience levels (numbers next to fixation locations, middle panel) during inspection of the image shown in the left panel. The right panel shows average computed salience values for all locations in the image. The dashed line is the average computed salience level of the fixated locations. From R.J. Peters, A. Iyer, L. Itti & C. Koch (2005) Components of bottom-up gaze allocation in natural images. Vision Research, 45, 2397–2416. (fig. 7)

Figure 6

Comparison of 3 model searchers to human performance. Najemnik & W. Geisler (2009). Each panel shows average spatial distribution of fixated locations. The Bayes Ideal and ELM (entropy limit minimization) visibility models generate predictions more similar to the human than the MAP searcher, whose winner-take-all strategy is closer to the predictions of saliency models (section 3.1). From J. Najemnik & W.S. Geisler (2009) Simple summation rule for optimal fixation selection in visual search. Vision Research, 49, 1286–1294. Fig. 4.

Figure 7

(a) Attentional operating characteristics, showing performance tradeoff between saccadic and perceptual performance when making saccades to one target and reporting the identity of another. Location of the saccadic target was either constant in a block (open symbols) or cued before each trial (filled symbols). The 3 data points in each function show performance under instructions to give different relative weights to the two tasks. The intersection of the dashed lines is the independence point, showing performance expected if there are no shared resources between saccadic and perceptual tasks. From E Kowler, E. Anderson, B Dosher & E Blaser (1995) The role of attention in the programming of saccades. Vision Research 35, 1897–1916. (fig. 11) (b) Orientation discrimination during pauses between saccades made to look along a color cued (green) path. Only a portion of the saccadic path is shown. Current eye position is shown by the dashed circle. Ahead: Eye position was at the top; the remaining two green cued locations are saccadic targets. Behind: Eye position reached the bottom. The remaining two green locations were previously fixated. All red cued locations are outside the saccadic path. From: TM Gersch, E Kowler, B Schnitzer & B Dosher (2009) Attention during sequences of saccades along marked and memorized paths. Vision Research, 49, 1256–1266 (fig. 3).

Figure 8

Evidence for concurrent planning of pairs of saccades. The sample eye traces show an incorrect saccade to a distractor followed by a second saccade to the target. From: R.M. McPeek, A.A. Skavenski & K. Nakayama (1999) Concurrent processing of saccades in visual search. Vision Research, 40, 2499–2516 (fig. 3)

Figure 9

Example of performance in the block-copying task, where the objective is to move blocks from the resource to the workspace in order to copy the pattern shown in the model. Thin lines show eye traces; thick lines show movements of the visible cursor used to move the blocks. From M. M. Hayhoe, D.G. Bensinger, D.H. Ballard (1998) Task constraints in visual working memory. Vision Research, 38, 12–137. (fig. 1)

Figure 10

(a) Cartoon showing subject tapping a set of rods placed on a worktable. Rotations of head and eye are recorded by the revolving field monitor. Translations of the head are recorded by an acoustic tracking system. (b) Sample traces of gaze, 3 targets and head. Rotational motions of the target are produced by translations of the head, which occurred both during and between gaze shifts. (c) Average speeds of head, eye, gaze and retinal image during both tapping and looking-only tasks. “Search” refers to speeds before the locations of the tapped rods were learned; “sequence” after the locations were learned. From J. Epelboim (1998), Gaze and retinal-image-stability in two kinds of sequential looking tasks, Vision Research, 38, 3773–3784, Figs. 1, 3, and 5.

Figure 11

Eye movements while stacking a set of blocks (a, c, e) or observing someone else stack the blocks (b, d, f). a,b, show positions at end of saccades (diameter of dot proportional to fixation duration); lines show path of hand. c,d show eye positions over time; e,f show eye velocities. From J.R. Flanagan, R.S. Johansson (2001) Action plans used in action observation. Nature, 424, 769–771 (fig. 1).