ArticlePDF Available

Eyes Pursue Moving Objects, Not Retinal Motion Signals



For smooth-pursuit eye movements, a moving target is necessary, but we show that it is not sufficient. Observers pursued targets that appeared to move in one direction even though they really moved in another. Changes in perceived direction did not disrupt pursuit eye movements, but motion-based failures in object parsing did.
Pursuit eye movements are the smooth rolling movements of the eyeball that lock
a moving target onto the high-acuity fovea (Kowler 1990). Is this a simple tracking servo
system, driven by error signal from retinal slip? Or is it a sophisticated, cognitive system,
receiving the same visual inputs as does motion perception? (Beutter and Stone 2000;
Krauzlis 2004; Barnes 2008). We measured attempts to pursue objects that moved in one
direction but, under the influence of three separate motion illusions (flying bugs, chop-
sticks, and rings: Anstis 1990, 2003; Anstis and Casco 2006), appeared to move in a
different direction. We argued that a retinal-slip servo would track the physical motion,
whereas a perception-based top^ down system would track the perceived motion.
small spot circled clockwise on a large background that was circling counterclockwise
without rotating. Induced movement (Duncker 1929/1938; Reinhardt-Rutland 1988) from
the background made the bug's circular path look like an illusory elongated ellipse,
tilted at 458. Three observers either attempted to pursue the moving bug with their eyes,
or else adjusted the Lissajous path of a separate matching spot, to match the illusory
trajectory. Figure 1 shows that they perceived the bug's path as elliptical (dotted ellipse
shows the best fit to their mean settings). Their pursuit eye movements accurately
tracked the bug's circular orbit (the circle in figure 1 shows the best fit to their eye
movements that are shown in figures 2a and 2b), but when questioned, they claimed
that their eyes were moving around an elliptical path rather than the actual circular
path. Conclusion : They were not aware of what their own eyes were doing. Evidently
awareness of eye movements, even if based on efference copy (Holst and Mittelstaedt
1950; Perrone and Krauzlis 2008), is substantially modulated by retinal feedback.
The accuracy of the smooth pursuit despite strong reports of perceptual illusion has
been reported before, even, recently, in monkeys (Zivotofsky et al 2005). Moreover,
an earlier paper showed that the eye tracking was unaffected by the illusion but head
movement was (Zivotofsky et al 1995).
In the chopstick illusion (Anstis 1990, 2003, 2007), a horizontal and a vertical line
overlapped to form a cross (figure 2c). The lines moved along counterphase clockwise
paths at 0.88 revolutions s
without rotating. The centre of the cross, where the
lines intersect, actually moves along a counterclockwise Lissajous circle, but observers
nearly always perceive it as moving clockwise, with the lines sliding over each other.
Eyes pursue moving objects, not retinal motion signals
Perception, 2010, volume 39, pages 1408 ^ 1411
Stuart Anstis
Department of Psychology, University of California San Diego (UCSD), 9500 Gilman Drive, La Jolla,
CA 92093-0109, USA; e-mail:
Hiroyuki Ito
Department of Visual Communication Design, Kyushu University, 4 -9-1, Shiobaru, Minami-ku,
Fukuoka-shi, 815-8540 Japan; e-mail:
Received 5 April 2009, in revised form 3 September 2010
Abstract. For smooth-pursuit eye movements, a moving target is necessary, but we show that it is
not sufficient. Observers pursued targets that appeared to move in one direction even though they
really moved in another. Changes in perceived direction did not disrupt pursuit eye movements,
but motion-based failures in object parsing did.
We believe that the clockwise motion of the line tips propagates along the lines and is
blindly assigned to the central intersection (see McDermott and Adelson 2004). Five
observers made large errors when they attempted to pursue this sliding intersection
(figure 2c). This illusion disappears instantly, and accurate pursuit movements are
restored, if a floating square frame is added that just touches the ends of the lines
(figure 2d). Now the lines are correctly parsed as a rigid cross rotating counterclockwise,
with no sliding, because, even though in figure 2d the circling tips are always visible,
they are interpreted as being occluded by the square frame (Shimojo et al 1989), so are not
eye movement
Figure 1. Perceived and pursuit paths in the flying-bug
illusion (movie versions are in the supplementary
materials). The bug, and the pursuit movements
(also shown in figures 2a and 2b), actually moved in
a circle (Lissajous phase shift 908). But observers
believed that the bug and their eyes followed the
illusory tilted ellipse (phase shift 558).
Eye movements t
Eye movements
(a) (b)
(c) (d)
(e) (f)
Figure 2. Illusory motion stimuli and typical tracking eye movements. Movie versions are in the
supplementary materials. (a) and (b) A bug flying in counterclockwise circles was accurately tracked
with circular eye movements. When the background jittered randomly (b), the observers correctly
saw the bug and their eyes as moving in circles. When the background circled clockwise (a),
they incorrectly believed that the bug and their eyes were following a right-oblique ellipse.
Conclusion : They did not know how their eyes were moving. (c) Circling chopsticks appeared to
slide and apparently circle clockwise. Eye tracking was very poor. (d) The same chopsticks with
a moving frame were parsed as a rigid cross circling counterclockwise. Eye tracking was good.
(e) Rotating rings with vertically aligned gaps or spots appeared to slide. Observers could not track
sliding intersection (arrowed). (f) Rings with painted-on spots were parsed as a solid figure of eight.
Now observers could track rigid intersection (arrowed).
Eyes pursue moving objects, not retinal motion signals 1409
taken as a reliable guide to the movements of the lines. We argue that the changed status
of the line terminators makes the intersection easier to track. With the counterrotating
terminators out of the picture, both remaining elements, the intersection and the frame,
rotate in the same direction. However, the frame moves 1808out of phase with the inter-
section, being at 6 o'clock when the intersection is at 12 o'clock, so any attempt to track
the square frame itself would disrupt rather than improve tracking of the intersection.
Root-mean-square tracking errors were calculated as the SD of the eye movement
radii, divided by the mean radius. Removing the frame from the chopsticks increased
these errors by 50%, from 0.30 to 0.433 (F
We now modified the chopstick illusion into a sliding-rings version. We removed
the line terminators (tips) by bending the lines around into rings that rotated at 0.45
revolutions s
. Two gaps or spots on each ring acted like terminators. When these
spots rotated with the rings (figure 2f), three observers reported a single rigid rotating
figure of eight, and they could successfully pursue the X-shaped junction (arrowed
in figures 2e and 2f ) where the two rings intersected. But when the two spots on each
ring always remained vertically aligned, somewhat like a floating compass needle (fig-
ure 2e), the percept was radically reorganized into two separate rings that slid over each
other. Ability to pursue the sliding X-shaped intersection with the eyes fell off markedly
and the rms tracking errors doubled from 0.147 to 0.291 (tÿ6:156,p0:025).
The chopstick/ring illusion has a different mechanism from the bugs illusion.
Although both illusions comprise overlapping interacting circular orbits, the bugs illu-
sion involves motion contrast, in which a moving spot seems to go in the opposite
direction to other objects. The chopstick/ring illusion involves motion assimilation, in
which a moving intersection seems to go in the same direction as other parts of the
same assembly (Spering and Gegenfurtner 2007). Crucially, the chopstick and ring
illusions disrupted pursuit eye movements where the bugs illusion did not. Thus, in
the bugs illusion, observers accurately tracked the circular orbits of the bugs, but they
misperceived both the bug paths and their own eye movements as being elliptical.
In the chopstick illusion, framed intersections (figure 2d) were parsed as rigid objects
and could be successfully tracked, but without the frame (figure 2c) the very same inter-
sections were parsed not as objects but as incoherent sliding junctions, and they could
no longer be well tracked. Likewise, rigidly intersecting rings could be tracked but sliding
ring junctions could not (figures 2e and 2f). The local retinal signals from intersecting
lines or rings were the same whether they were rigid or sliding, so clearly these pursuit
movements were not simply driven by retinal slip.
So what disrupts the pursuit eye movements? Not just illusory changes in direction,
since the bugs are perceived to deviate from their true direction but can be tracked,
whereas the sliding ring intersections appear to follow their true, circular, path and yet
cannot be tracked. It is the failure of object parsing that upsets pursuit. Bugs are seen as
objects and can be pursued, and so can perceptually rigid intersections, whereas the
sliding intersections of rings and lines are not parsed as objects and cannot be pursued.
We conclude that pursuit eye movements are not a simple tracking servo, but receive
sophisticated top ^ down control from object parsing (Beutter and Stone 2000; Krauzlis
2004; Barnes 2008). A visual control system that can dismiss intersecting tree twigs as
noise, but can correctly parse and track even the stealthiest moving predator, is likely
to survive.
Guide to the movies in supplementary materials
Six movies are attached, each about 6 Mb in size:
1. BugStim.mpg shows a circling yellow bug, first against a rotating background
(figure 2b: the bug's path appears to be an ellipse), then against a randomly jittering
background (figure 2a: the bug's path is correctly seen as circular).
1410 S Anstis, H Ito
2. BugEM.mpg shows the same stimuli as #1, with typical eye movements superimposed
as a red trail.
3. ChopsticksStim.mpg shows the chopsticks circling counterclockwise, first on their own
(figure 2c: the central intersection appears to move clockwise), then with a floating square
frame (figure 2d: intersection correctly seen as circling counterclockwise).
4. ChopstickEM.mpg shows the same stimuli as #3, with typical eye movements super-
imposed as a red trail. For convenience a blue disk (never seen by the observers) indicates
the target intersection they were trying to track.
5. RingsStim.mpg shows the sliding rings, first floating (figure 2e: hard to track) then
painted on the rings (figure 2f: easy to track).
6. RingsEM.mpg shows the same stimuli as #5, with typical eye movements super-
imposed as a red trail. Blue disk indicates the target intersection they were trying to track.
See also the GuideToVideoMaterials.ppt in the supplementary materials.
Acknowledgments. Supported by Grants-in-Aid for Scientific Research (20300048, 19653083 and
19103003) to HI from the Ministry of Education, Culture, Sports, Science and Technology
Japan. SA thanks UCSD for granting sabbatical leave.
Anstis S, 1990 ``Imperceptible intersections: The chopstick illusion'', in AI and the Eye Eds A Blake,
T Troscianko (London: John Wiley) pp 105 ^ 117
Anstis S, 2003 ``Levels of motion perception'', in Levels of Perception Eds L Harris, M Jenkin
(New York: Springer) pp 75^ 99
Anstis S M, 2007 ``The flash-lag effect during illusory chopstick rotation'' Perc eption 36 1043 ^ 1048
Anstis S, Casco C, 2006 ``Induced movement: The flying bluebottle illusion'' Journal of Vision
6(10):8, 1087 ^ 1092;,doi:10.1167/6.10.8
Barnes G R, 2008 ``Cognitive processes involved in smooth pursuit eye movements'' Brain and
Cognition 68 309 ^ 326
Beutter B R, Stone L S, 2000 ``Motion coherence affects human perception and pursuit similarly''
Visual Neuroscience 17 139^153
Duncker K, 1929/1938 ``U
ëber induzierte Bewegung (Ein Beitrag zur Theorie optisch wahrgenommener
Bewegung)'', in Source Book of Gestalt Psychology Editor and translator W D Ellis (London:
Routledge & Kegan Paul) pp 161 ^ 172 (Reprinted from Psychologische Forschung 1929 12 18 0 ^ 259)
Holst E von, Mittelstaedt H, 1950 ``Das Reafferenzprinzip'' Naturwissenschaften 37 464 ^ 476
KowlerE(Ed.),1990Eye Movements and Their Role in Visual and Cognitive Processes (Amsterdam
and New York: Elsevier)
Krauzlis R J, 2004 ``Recasting the smooth pursuit eye movement system''Journal of Neurophysiology
91 591^603
McDermott J, Adelson E H, 2004 ``Junctions and cost functions in motion interpretation'' Journal
of Vision 4(7):3, 552 ^ 563;,doi:10.1167/4.7.3
Perrone J A, Krauzlis R J, 2008 ``Vector subtraction using visual and extraretinal motion signals:
A new look at efference copy and corollary discharge theories'' Journal of Vision 8(14):24,
Reinhardt-Rutland A H, 1988 ``Induced movement in the visual modality: an overview'' Psychological
Bulletin 103 57^71
Shimojo S, Silverman G H, Nakayama K,1989 ``Occlusion and the solution to the aperture problem
for motion'' Vision Research 29 619 ^ 626
Spering M, Gegenfurtner K R, 2007 ``Contrast and assimilation in motion perception and smooth
pursuit eye movements'' Journal of Neurophysiology 98 1355 ^ 1363
Zivotofsky A Z, Averbuch-Heller L, Thomas C W, Das V E, Discenna A O, Leigh R J, 1995
``Tracking of illusory target motion: differences between gaze and head response'' Vision Research
35 3029 ^ 3035
Zivotofsky A Z, Goldberg M E, Powell K D, 2005 ``Rhesus monkeys behave as if they perceive
the Duncker Illusion'' Journal of Cognitive Neuroscience 17 1011 ^ 1017
ß 2010 a Pion publication
Eyes pursue moving objects, not retinal motion signals 1411
ISSN 0301-0 06 6 (print)
Conditions of use. This article may be downloaded from the Perception website for personal research
by members of subscribing organisations. Authors are entitled to distribute their own article (in printed
form or by e-mail) to up to 50 people. This PDF may not be placed on any website (or other online
distribution system) without permission of the publisher.
ISSN 1468-423 3 (electronic)
... Early visual areas responded equally under the two conditions, however a region including area V5/MT was activated during the perception of apparent motion (Muckli et al., 2002). Finally, Anstis and Ito (2010) have shown that smooth pursuit eye-movements are guided by real stimuli and not by retinal signals. Therefore, LM's difficulty with visually-guided smooth pursuit eye-movements is more likely of central origin, caused by her cerebral motion blindness, and not by dysfunction of her peripheral visual system. ...
Full-text available
The significance of early and sporadic reports in the 19th century of impairments of motion vision following brain damage was largely unrecognized. In the absence of satisfactory post-mortem evidence, impairments were interpreted as the consequence of a more general disturbance resulting from brain damage, the location and extent of which was unknown. Moreover, evidence that movement constituted a special visual perception and may be selectively spared was similarly dismissed. Such skepticism derived from a reluctance to acknowledge that the neural substrates of visual perception may not be confined to primary visual cortex. This view did not persist. First, it was realized that visual movement perception does not depend simply on the analysis of spatial displacements and temporal intervals, but represents a specific visual movement sensation. Second persuasive evidence for functional specialization in extrastriate cortex, and notably the discovery of cortical area V5/MT, suggested a separate region specialized for motion processing. Shortly thereafter the remarkable case of patient LM was published, providing compelling evidence for a selective and specific loss of movement vision. The case is reviewed here, along with an assessment of its contribution to visual neuroscience.
... Some motion illusions disable observers in tracking a target by smooth eye movements. The Chopsticks illusion and the Rotating Rings illusion (Anstis 2003) change the perceptual property of the intersections, resulting in a failure to track the intersection visually, although induced motion did not affect the tracking motion in spite of the perceived distortion of the motion path (Anstis and Ito 2010). Conversely, Tomimatsu et al (2010) showed that smooth pursuit eye movement greatly reduces the effect of the Rotating Snakes illusion. ...
Full-text available
A circular object placed in the centre of a radial pattern consisting of thin sectors was found to cause a robust motion illusion. During eye-movement pursuit of a moving target, the presently described stimulus produced illusory background-object motion in the same direction as that of the eye movement. In addition, the display induced illusory stationary perception of a moving object against the whole display motion. In seven experiments, the characteristics of the illusion were examined in terms of luminance relationships and figural characteristics of the radial pattern. Some potential explanations for these findings are discussed.
... All rights reserved. doi:10.1016/j.visres.2012.02.001 variants of the Duncker illusion, steady-state pursuit performance is not affected by illusory motion induced by motion of a small target and a large background in humans (Anstis & Ito, 2010;Zivotofsky et al., 1995) and also in monkeys (Zivotofsky, Goldberg, & Powell, 2005). ...
To pursue a small target moving in front of a drifting background, motion vectors from the target need to be integrated and segmented from those belonging to the background. Smooth pursuit eye movements typically integrate target and background directions initially and after some time shift towards the veridical target direction. The perceived target direction on the other hand is generally stable over time: the target is perceived to move in the same direction as long as the motion information maintains the same properties over time. If illusory target motion is observed, this tends to be shifted away from the background. Here we investigated how initial motion integration and segmentation of such stimuli are modulated by direction cues. We presented a small pursuit target moving along a straight path, in front of a background moving in a different direction. Without a direction cue, initial pursuit was biased towards the background direction before shifting towards the veridical target direction. The target's perceived direction on the other hand was near veridical. A cue in the background direction increased initial pursuit integration but also caused perception to behave in a similar way: the target initially had an illusory motion component in the background direction and after about 200 ms it was perceived to curve towards its veridical direction. This illusion shows that during the initial process of segmenting the direction of a pursuit target from irrelevant background motion, both pursuit and perception can be erroneously influenced by a direction cue and integrate the cued background motion. Both modalities corrected this initial integration error as more information about the target became available.
... Early visual areas responded equally under the two conditions, however a region including area V5/MT was activated during the perception of apparent motion (Muckli et al., 2002). Finally, Anstis and Ito (2010) have shown that smooth pursuit eye-movements are guided by real stimuli and not by retinal signals. Therefore, LM's difficulty with visually-guided smooth pursuit eye-movements is more likely of central origin, caused by her cerebral motion blindness, and not by dysfunction of her peripheral visual system. ...
A spot moved vertically up and down across a background grating that was tilted at 45 degrees. In foveal vision this was seen accurately, but when viewed peripherally the spot's path was perceptually attracted toward the grating orientation, and at large eccentricities (>20 degrees) the spot appeared to move at 45 degrees, parallel to the grating. The intersections between the grating and the moving spot drive this illusion, revealing profound differences between fovea and periphery in processing visual motion.
Full-text available
I shall present some new, or newish, illusions to show that motion signals in the early parts of the visual system are profoundly altered by stimulus luminance and contrast. I shall show that contrast affects: 1. Motion strength in Time till breakdown 2. Motion strength in Crossover motion 3. Speed in The Footsteps illusion 4. Direction in The Plaid-motion illusion 5. Direction: Split dots I shall then consider how it is that higher perceptual processes massage these neural motion signals into the perception of moving objects. For instance, moving line terminators help to solve the aperture problem. But these solutions are modified by stimulus contrast in the Plaid-motion illusion and in the Peripheral-oblique illusion. In the Chopstick and Sliding Rings illusion, the motion of terminators propagates along straight lines and is blindly (and incorrectly) assigned to the motion of the central intersection. Finally, a new display of moving dots alternates perceptually between two radically different perceptual interpretations. Usually the Local percept (trees) is seen first, but the Global interpretation (forest) gradually takes over in the course of time.
Full-text available
The question as to how the visual motion generated during eye movements can be 'canceled' to prevent an apparent displacement of the external world has a long history. The most popular theories (R. W. Sperry, 1950; E. von Holst & H. Mittelstaedt, 1950) lack specifics concerning the neural mechanisms involved and their loci. Here we demonstrate that a form of vector subtraction can be implemented in a biologically plausible way using cosine distributions of activity from visual motion sensors and from an extraretinal source such as a pursuit signal. We show that the net result of applying an 'efference copy/corollary discharge signal' in the form of a cosine distribution is a motion signal that is equivalent to that produced by vector subtraction. This vector operation provides a means of 'canceling' the effect of eye movements. It enables the extraretinal generated image motion to be correctly removed from the combined retinal-extraretinal motion, even in cases where the two motions do not share the same direction. In contrast to the established theories (efference copy and corollary discharge), our new model makes specific testable predictions concerning the location (the MT-MST/VIP areas) and nature of the eye-rotation cancellation stage (neural-based vector subtraction).
Full-text available
The primate visual system faces a difficult problem whenever it encounters the motion of an object moving over a patch of the retina. Objects typically contain a number of edges at different orientations and so a range of image velocities are generated within the receptive field of a neuron processing the object movement. It is still a mystery as to how these different velocities are combined into one unified and correct velocity. Neurons in area MT (V5) are considered to be the neural substrate for this motion integration process. Some MT neurons (pattern type) respond selectively to the correct global motion of an object, whereas others respond primarily to the individual components making up the pattern (component type). Recent findings from MT pattern cells tested with small patches of motion (N. J. Majaj, M. Carandini, & J. A. Movshon, 2007) have put further constraints on the possible mechanisms underlying MT pattern motion integration. We tested and refined an existing model of MT pattern neurons (J. A. Perrone, 2004) using these same small patch stimuli and found that it can accommodate these new findings. We also discovered that the speed of the test stimuli may have had an impact on the N. J. Majaj et al. (2007) results and that MT direction and speed tuning may be more closely linked than previously thought.
The “aperture problem” indicates that a local reading of the velocity of an oriented contour is inherently ambiguous, insufficient by itself to recover the velocity of image points. In Wallach's “barber pole” display consisting of moving diagonal lines within an elongated rectangular aperture, it has been suggested that the unambiguous motion of edge-terminators along the longer edges of the aperture propagates towards the motion-ambiguous center part of drifting stripes. This results in the perception of a surface moving in the direction of the longer axis of the aperture. By manipulating the stereoscopic disparity of a striped pattern relative to the aperture plane, we found that the disambiguating effects of terminators could be abolished if the striped pattern was in uncrossed disparity relative to the aperture plane. Also, the motion in 3 separate horizontally oriented, and vertically aligned apertures which would otherwise be seen as moving horizontally, was seen as “linked” together and moving vertically. This occurred only when the horizontally oriented segments separating these apertures were stereoscopically coded so that they appeared as occluders in front. These findings suggest that accidental or “extrinsic” terminators created by occluding edges are treated differently from real or “intrinsic” terminators, and that the real-world constraint of occlusion is thus implemented in the ambiguity-solving processes for motion.
Ocular pursuit movements allow moving objects to be tracked with a combination of smooth movements and saccades. The principal objective is to maintain smooth eye velocity close to object velocity, thus minimising retinal image motion and maintaining acuity. Saccadic movements serve to realign the image if it falls outside the fovea, the area of highest acuity. Pursuit movements are often portrayed as voluntary but their basis lies in processes that sense retinal motion and can induce eye movements without active participation. The factor distinguishing pursuit from such reflexive movements is the ability to select and track a single object when presented with multiple stimuli. The selective process requires attention, which appears to raise the gain for the selected object and/or suppress that associated with other stimuli, the resulting competition often reducing pursuit velocity. Although pursuit is essentially a feedback process, delays in motion processing create problems of stability and speed of response. This is countered by predictive processes, probably operating through internal efference copy (extra-retinal) mechanisms using short-term memory to store velocity and timing information from prior stimulation. In response to constant velocity motion, the initial response is visually driven, but extra-retinal mechanisms rapidly take over and sustain pursuit. The same extra-retinal mechanisms may also be responsible for generating anticipatory smooth pursuit movements when past experience creates expectancy of impending object motion. Similar, but more complex, processes appear to operate during periodic pursuit, where partial trajectory information is stored and released in anticipation of expected future motion, thus minimising phase errors associated with motion processing delays.
Induced movement, illusory movement in a stationary stimulus resulting from adjoining movement, has received steady experimental investigation over the last 70 years or so. It is observed under different viewing conditions in a wide variety of displays that differ considerably in overall size and in form of inducing and induced stimuli. Explanations have been diverse, some being based on relations within the display and others invoking mediation by other aspects of the observer’s perception. Probably, no one explanation can account for all forms of induced movement. Current knowledge about induced movement may have important implications for visual perception of object motion.
We compared ocular and eye-head tracking responses to an illusion of diagonal motion produced when vertical movement of a small visual target was synchronized to horizontal movement of a background display. In response to sinusoidal movement, smooth ocular pursuit followed vertical target motion, with only a small horizontal component. In response to regular stepping movement, all anticipatory saccades were in the direction of the illusion; these erroneous oblique movements were followed by corrective horizontal saccades. When the head was free to move, it usually showed a diagonal trajectory that, for both sinusoidal and stepping target motion, was always in the direction of the illusion; no corrective movements were present. Thus, for our illusory stimuli, eye and head tracking showed qualitative differences that imply that ocular tracking was ultimately controlled by actual target motion but head tracking was controlled by illusory target motion.