Do we track what we see? Common versus independent processing for motion perception and smooth pursuit eye movements: a review.
ABSTRACT Many neurophysiological studies in monkeys have indicated that visual motion information for the guidance of perception and smooth pursuit eye movements is - at an early stage - processed in the same visual pathway in the brain, crucially involving the middle temporal area (MT). However, these studies left some questions unanswered: Are perception and pursuit driven by the same or independent neuronal signals within this pathway? Are the perceptual interpretation of visual motion information and the motor response to visual signals limited by the same source of neuronal noise? Here, we review psychophysical studies that were motivated by these questions and compared perception and pursuit behaviorally in healthy human observers. We further review studies that focused on the interaction between perception and pursuit. The majority of results point to similarities between perception and pursuit, but dissociations were also reported. We discuss recent developments in this research area and conclude with suggestions for common and separate principles for the guidance of perceptual and motor responses to visual motion information.
Do we track what we see? Common versus independent processing
for motion perception and smooth pursuit eye movements: A review
Miriam Speringa, Anna Montagninib,⇑
aDepartment of Psychology & Center for Neural Science, New York University, NY, United States
bInstitut de Neurosciences Cognitives de la Méditerranée, CNRS and Aix-Marseille University Marseille, France
a r t i c l ei n f o
Received 23 June 2010
Received in revised form 9 October 2010
Available online 20 October 2010
Smooth pursuit eye movements
a b s t r a c t
Many neurophysiological studies in monkeys have indicated that visual motion information for the guid-
ance of perception and smooth pursuit eye movements is – at an early stage – processed in the same
visual pathway in the brain, crucially involving the middle temporal area (MT). However, these studies
left some questions unanswered: Are perception and pursuit driven by the same or independent neuronal
signals within this pathway? Are the perceptual interpretation of visual motion information and the
motor response to visual signals limited by the same source of neuronal noise? Here, we review psycho-
physical studies that were motivated by these questions and compared perception and pursuit behavior-
ally in healthy human observers. We further review studies that focused on the interaction between
perception and pursuit. The majority of results point to similarities between perception and pursuit,
but dissociations were also reported. We discuss recent developments in this research area and conclude
with suggestions for common and separate principles for the guidance of perceptual and motor responses
to visual motion information.
? 2010 Elsevier Ltd. All rights reserved.
Imagine riding a bicycle on a busy four-lane street in Manhat-
tan. Moving cars, delivery vans, and garbage trucks surround you.
You have to avoid falling into unmarked construction sites and hit-
ting pedestrians that randomly cross the street. In order to plan
your movements and to survive in such a complex and dynamic vi-
sual environment, it is important to obtain a veridical percept of
the visual scene. You have to estimate the direction and speed of
vehicles around you and use this information fast to decide where
to move next.
In primates, two types of voluntary eye movements critically
support vision by centering and stabilizing the image of a visual
object of interest on the fovea, the region on the retina where vi-
sual acuity is highest. Saccadic eye movements are discrete, ballistic
movements that direct the eyes quickly toward a visual target;
smooth pursuit eye movements are continuous, slow rotations of
the eyes that compensate for motion of the visual target. These vol-
untary eye movements are not simply sensorimotor reflexes, but
depend on the sophisticated sensory and cognitive processing
capabilities that characterize our central nervous system (for re-
views see Krauzlis, 2004, 2005).
Smooth pursuit eyemovementsare closely related to sensory in-
puts from the motion processing system. A large number of neuro-
physiological studies in awake behaving monkeys have linked
neuronal activity in the primate’s middle temporal area (MT) and
the adjacent middle superior temporal area (MST) to both the per-
ception of visual motion (e.g., Newsome, Britten, & Movshon,
1989; Newsome & Paré, 1988; Pasternak & Merigan, 1994; Rudolph
& Pasternak, 1999; Salzman, Murasugi, Britten, & Newsome, 1992;
but see Ilg & Churan, 2004) and the control of smooth pursuit eye
movements (e.g., Dürsteler & Wurtz, 1988; Ilg & Thier, 2003;
Komatsu & Wurtz, 1988, 1989). These findings are paralleled in hu-
early stage. This common pathway links direction-selective retinal
ganglion cells to areas MT and MST and to higher-order motion pro-
cessing areas in the occipital, parietal and frontal cortex (e.g., Brem-
mer, Distler,& Hoffmann, 1997; Culham,He, Dukelow,& Verstraten,
2001; Fukushima, 2003; Ilg & Churan, 2004; Orban, Sunaert, Todd,
Van Hecke, & Marchal, 1999) through the lateral geniculate nucleus
(LGN) and primary visual cortex (areas V1, V2).
However, these studies leave a number of questions open. Are
perception and pursuit driven by the same or independent neuro-
nal signals? Given their response variability, are perceptual and
pursuit performance limited by the same source of neuronal noise?
How do pursuit eye movements affect the perception of visual
motion? In four parts, we review psychophysical studies in
0042-6989/$ - see front matter ? 2010 Elsevier Ltd. All rights reserved.
⇑Corresponding author. Address: Institut de Neurosciences Cognitives de la
Méditerranée, UMR 6193, CNRS-Université de la Méditerranée, 31, chemin Joseph
Aiguier, 13402 Marseille cedex, France.
E-mail address: Anna.Montagnini@incm.cnrs-mrs.fr (A. Montagnini).
Vision Research 51 (2011) 836–852
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/visres
humans that have addressed these issues, with a focus on those
published in the last 10 years. We begin with a brief overview of
the pursuit system (Section 2) before addressing each of the ques-
tions stated above. Section 3 provides a summary of behavioral evi-
dence for common and independent processing of visual motion
information for perception and pursuit as well as a discussion of
noise sources. Section 4 focuses on the interaction between per-
ception and pursuit. In the final Section 5, we review model ideas
that have emerged recently in this active field of research and con-
clude with a summary of common and separate principles of mo-
tion processing for the guidance of perception and pursuit.
The research reviewed here is generally motivated by the ques-
tion how sensory information is transformed into motor actions so
that we can successfully interact with our dynamic visual environ-
ment. Pursuit eye movements are a relatively simple and well-
understood motor behavior – we largely know how the sensory
input and motor events related to pursuit are represented in the
brain – and therefore provide an excellent model system to study
this sensorimotor transformation. We continue a tradition of re-
views on the visual signals that drive pursuit eye movements
(Krauzlis & Stone, 1999; Lisberger, 2010; Lisberger, Morris, &
Tychsen, 1987) and provide the first systematic review on the
question whether motion information processing for perception
and pursuit is common or independent. Due to limited space, we
mention studies on other motion-guided eye movement systems
(optokinetic nystagmus, ocular following response) only briefly,
and we do not review studies on saccades (for reviews on pursuit
and saccades see Krauzlis & Stone, 1999; Orban de Xivry & Lefèvre,
2007). We further do not include neurophysiological studies, mod-
eling studies, studies that focus on aspects of development, or
those on pathologies in motion processing.
2. Characterizing smooth pursuit eye movements
2.1. Response latency, open-loop and closed-loop pursuit
Human smooth pursuit eye movements in response to a moving
visual target usually have a latency of about 80–120 ms (Carl &
Gellman, 1987; Krauzlis, 2004; see Merrison and Carpenter (1991)
for a report on express pursuit). In monkeys, pursuit latencies can
be as low as 65 ms (Lisberger & Westbrook, 1985; Lisberger et al.,
1987). The latency of the pursuit response generally depends on
properties of the visual target such as luminance, size, velocity
and position (e.g., Tychsen & Lisberger, 1986). Latency further de-
pends on properties of the context such as the number of potential
targets (Ferrera & Lisberger, 1995; Krauzlis, Zivotofsky, & Miles,
1999; Spering & Gegenfurtner, 2008) and is influenced by cognitive
factors such as the predictability of the target trajectory (e.g., Bahill
& McDonald, 1983). Human observers are usually able to track a
target moving up to a velocity of 100?/s (Meyer, Lasker, & Robinson,
1985). However, pursuit is often too slow with respect to the target,
especially when target velocity exceeds 30?/s. To compensate for
retinal image slip, smooth pursuit eye movements are substituted
by catch-up saccades (De Brouwer, Yuksel, Blohm, Missal, & Lefèvre,
The pursuit response is separated into an open-loop (initiation)
and a closed-loop or steady-state (maintenance) phase (Lisberger
the first ?100 ms of the eye movement, pursuit is primarily driven
by visual motion (the retinal image velocity) of the target. The eye
initially accelerates in the direction of the target, and later adjusts
to target velocity. During the closed-loop phase, the velocity error
of the eyes seems to be minimized by a negative feedback loop,
i.e., to stabilize the image of the target on the fovea the efference
copy and the retinal target motion signal are compared. Moreover,
ing pursuit (Lisberger et al., 1987; Morris & Lisberger, 1987).
Other types of slow eye movements that are driven by visual
motion are the ocular following response (OFR) and the optokinetic
nystagmus (OKN), a subsystem of ocular following. Whereas pur-
suit is voluntary, these responses are low-level reflexes resembling
the properties of early visual processing (Miles, Kawano, & Optican,
1986). The OFR is usually best elicited by brief, unexpected motion
of a large part of the visual scene (e.g., a full-field random-dot pat-
tern). It has a brief execution time and a considerably shorter la-
tency than pursuit, 50–85 ms in monkeys and humans (Gellman,
Carl, & Miles, 1990; Miles et al., 1986). Although pursuit and reflex-
ive OKN and OFR differ in important aspects such as latency, both
types of eye movements are driven by similar neural systems (see
Dürsteler & Wurtz, 1988; Konen, Kleiser, Seitz, & Bremmer, 2005;
Takemura, Murato, Kawano, & Miles, 2007) that have been re-
1991; Krauzlis, 2004, 2005; Leigh & Zee, 2006; Thier & Ilg, 2005).
2.2. Pursuit stimulus and mechanisms for its selection
Despite anecdotal reports to the contrary, primatesare unable to
smoothly track a purely imaginary object. Pursuit eye movements
require visual motion signals (e.g., Lisberger et al., 1987; Rashbass,
1961; Robinson, 1965). Although most laboratory studies have
background, pursuit can be elicited by a variety of motion stimuli,
such as random-dot kinematograms (e.g., Heinen & Watamaniuk,
1998) or line figures (e.g., Masson & Stone, 2002). It can further be
evoked in the absence of a physical motion stimulus, i.e., without
retinal image motion, for instance by perceived or apparent image
1976; Wyatt, Pola, Fortune, & Posner, 1994) or by predicted motion
(Barnes, 2008; Kowler & Steinman, 1979a, 1979b, 1981).
In our natural visual environment, we rarely encounter motion
of a single isolated object. Rather, an object of interest is usually
surrounded by a dynamic visual context (see Section 3.2.1), requir-
ing the observer to actively select a visual target. Target selection is
one of the central requirements of the pursuitsystem. While we can
perceptually keep track of multiple moving objects (for a review,
see Cavanagh & Alvarez, 2005), we can only actively track one mo-
tion trajectory with our eyes. When confronted with multiple mov-
ing objects with equal salience at the same time, pursuit eye
movements initially follow the direction of the average of all the
available motion vectors (e.g., Ferrera & Lisberger, 1995, 1997;
see Section 3.2.1). Vector averaging (VA) is usually a transient re-
sponse reflecting conflicting motion inputs, and corrected in the
direction of one selected object (winner-take-all response, WTA)
during the later stage of the pursuit response. The time course for
this shift from VA to WTA is reflected in the activity of neurons in
areas MT and MST (Recanzone & Wurtz, 1999). The target selection
process can be modulated by stimulus properties such as lumi-
nance contrast and color (Spering, Montagnini, & Gegenfurtner,
2008; see Section 3.3.2), as well as visual spatial or feature-based
attention (Ferrera, 2000; Ferrera & Lisberger, 1995; Garbutt &
Lisberger, 2006; Krauzlis et al., 1999; Recanzone & Wurtz, 2000).
3. Common versus independent motion signals for perception
3.1. Precision of motion discrimination and sources of noise for
perception and pursuit
Visual motion information is the main determinant of pursuit
control. Pursuit eye movements represent a fast and direct linear
M. Spering, A. Montagnini/Vision Research 51 (2011) 836–852
readout of the direction and speed-tuned population activity of
neurons in area MT (Groh, Born, & Newsome, 1997; Lisberger &
Movshon, 1999; Newsome, Wurtz, & Komatsu, 1988). Similarly,
perceptual judgments of motion are reflected by MT neuronal
activity (Britten, Shadlen, Newsome, & Movshon, 1992; Newsome
et al., 1989; Salzman et al., 1992). These findings have brought
up two major questions. First, do perceptual and oculomotor ac-
counts of motion information provide the same amount of infor-
mation and are they affected by the same neuronal noise? This
question addresses the similarity between the two systems with
regard to their accuracy and noise source (e.g., Gegenfurtner, Xing,
Scott, & Hawken, 2003; Stone & Krauzlis, 2003). Second, how faith-
fully do oculomotor and perceptual outputs reflect the sensory in-
put with regard to motion direction, speed and acceleration? This
question focuses on the quantitative efficiency of either system
independently, compared to sensory noise (e.g., Osborne, Lisberger,
& Bialek, 2005; Rasche & Gegenfurtner, 2009).
3.1.1. Same or different accuracy–same or different noise source?
The studies reported here used detection and discrimination
tasks for the visual motion features direction, speed, and accelera-
tion. Binary perceptual and continuous pursuit responses were di-
rectly compared on the same trial by converting raw pursuit traces
into probabilistic binary pursuit decisions about motion direction,
speed or acceleration. Binary pursuit decisions could then be plot-
ted as a function of the variable of interest, yielding oculometric
functions (Kowler & McKee, 1987). These could directly be com-
pared to standard psychometric functions for the analysis of detec-
tion or discrimination thresholds and response precision, as
indicated by the slope of the functions (Fig. 1a and b). When both
responses are equally sensitive to changes in physical stimulus
parameters, the psychometric and oculometric functions should
largely overlap (Fig. 1a). If both responses differ with regard to
their response precision, the response with higher precision should
yield a steeper function (Fig. 1b). The question whether both re-
sponses share a common noise source that limits their perfor-
mance was usually addressed by testing whether the response
errors were correlated on a trial-by-trial basis.
220.127.116.11. Motion direction and speed. Two studies published at the
same time (Gegenfurtner et al., 2003; Stone & Krauzlis, 2003) have
tested perceptual and pursuit motion discrimination performance
simultaneously – Gegenfurtner and colleagues (2003) compared
the sensitivity of both systems to small changes in stimulus speed
and Stone and Krauzlis (2003) used a direction discrimination task
– while keeping the retinal motion input to the perceptual and the
pursuit system the same. Both studies report similarities in sensi-
tivity for perception and pursuit, but come to different conclusions
regarding the source of noise underlying perception and pursuit.
Gegenfurtner et al. (2003) presented a small visual target that
either moved at a constant speed of 4?/s, or increased or decreased
in speed (perturbation range: ±0.75?/s) during steady-state pursuit.
Observers were asked to report whether the target speed during
perturbation was faster or slower than during the rest of the trial
while smoothly tracking target motion with their eyes. In Stone
and Krauzlis (2003), the target moved at a constant speed but
along different directions, either straight horizontal or vertical
(cardinal), or diagonally (up to ± 6? off the cardinal axes). Observ-
ers had to indicate whether target motion direction was above or
below (right or left) with respect to the horizontal (vertical) axis
while smoothly tracking targets.
Both studies showed similar accuracy in perception and pursuit
in discriminating target speed changes (for an example see Fig. 1c
and d) and directions, and disagree with an older study
(Watamaniuk & Heinen, 1999), where a primacy of perceptual
accuracy had been reported, probably due to noisy oculomotor
measurements. Both studies also propose a similar model (Fig. 1e)
where an initial common stage of motion processing (with a com-
mon source of noise) is followed by two independent modules for
perception and pursuit control, respectively, that are affected by
two independent sources of additive noise. However, only Stone
and Krauzlis (2003) found evidence for a significant trial-by-trial
correlation between errors in perception and in pursuit, especially
for low signal values where errors were most frequent, whereas
Gegenfurtner and colleagues (2003) did not. These two studies pro-
vide strong behavioral evidence for common processing of motion
information for perception and pursuit and therefore fit well with
physiological reports, but leave the question open whether both
systems are limited by the same source of noise.
A recent study came to a different conclusion and reported that
the pursuit system was the more faithful analyzer of visual motion:
Tavassoli and Ringach (2010) used a task similar to the one in
Gegenfurtner et al. (2003) and applied a single-cycle sinusoidal
perturbation to target motion during steady-state pursuit. At the
end of each trial, observers had to discriminate whether the peak
(speed increase) or the trough (speed decrease) of the sinusoidal
perturbation had occurred first. Pursuit discrimination perfor-
mance was generally superior to perceptual performance and, in
some cases, even sensitive to speed changes that remained unper-
ceived. Interestingly, one out of four observers in Gegenfurtner
et al. (2003) showed a similar pursuit performance benefit.
Tavassoli and Ringach (2010) suggested that perceptual judgments
might have been corrupted by larger amounts of noise and/or more
heavily filtered, especially at low signal amplitude, where the mo-
tion input to the perceptual system is integrated across a larger
temporal window (i.e., filtered with a low-pass temporal filter).
The studies by Gegenfurtner et al. (2003) and Tavassoli and
Ringach (2010) have another result in common: perceptual and
pursuit errors were uncorrelated on a trial-by-trial basis. One
explanation for this discrepancy from the Stone and Krauzlis
(2003) study could be the use of speed perturbations during stea-
dy-state pursuit in the Gegenfurtner et al. and in the Tavassoli and
Ringach papers. Speed perturbations are a good method to probe
the strength (gain) of the pursuit response by testing whether
and how it is affected by changes in the visual input. Schwartz
and Lisberger (1994) found that the response to a speed perturba-
tion was modulated by both visual stimulus and motor features: a
perturbation had a stronger effect during steady-state pursuit
than during fixation. Gain-control mechanisms, as described in
Lisberger (2010; his Fig. 8), are held responsible for this enhance-
ment. We could then speculate that the neuronal signal acting as
a modulatory gain of the pursuit signal could introduce a form of
independent (multiplicative) noise in the oculomotor response to
the perturbation in the studies by Gegenfurtner et al. and Tavassoli
and Ringach, thereby disrupting the correlation between percep-
tual and pursuit responses without a large cost for pursuit discrim-
Two recent studies directly compared the initial OFR and
perception and reported differences between both. Boström and
Warzecha (2010) found higher sensitivities in perception than in
the OFR and no correlation between responses; Hayashi, Sugita,
Nishida, and Kawano (2010) found higher sensitivities in the OFR
for high temporal frequencies. Because of the differences between
pursuit and OFR (see Section 2.1) it is unclear how these results
compare to those obtained from comparisons between pursuit
and perception and whether the close connection between percep-
tion and pursuit holds for perception and OFR.
18.104.22.168. Oblique effect. The oblique effect – a performance asymme-
try (anisotropy) for the discrimination of motion direction and
orientation in human observers with lower discrimination thresh-
olds for cardinal than for diagonal directions (Ball & Sekuler, 1982;
M. Spering, A. Montagnini/Vision Research 51 (2011) 836–852
Furmanski & Engel, 2000) – is a well-known phenomenon in mo-
tion perception. Two psychophysical studies have compared per-
ceptual and oculomotor accounts of the oblique effect – one
reported an absence of the oblique effect in pursuit, the other
found an oblique effect in perception and pursuit.
In one of the first reports of a dissociation between perception
and pursuit observers had to discriminate between the directions
of two objects that were presented in succession and either moved
along horizontal or vertical (cardinal) axes or diagonal axes
(Churchland, Gardner, Chou, Priebe, & Lisberger, 2003). In a 2-IFC
task, observers reported whether the two target motions were
the same or different. Perceptual discrimination performance was
better if objects moved along the cardinal axes (oblique effect).
Interestingly, this asymmetry was not reported for pursuit as
(stimulus strength > reference)
Fig. 1. (a and b) Schematic results for a comparison between psychometric (blue) and oculometric (red) functions plotted as the probability of a response in a particular
direction over stimulus strength in arbitrary coordinates. In (a), both functions overlap, indicating similarities in response precision. In (b), the slope of the oculometric
function is steeper, indicating higher sensitivity of pursuit. (c and d) Example data from Gegenfurtner et al. (2003, p. 869; to be printed with permission) for two observers.
Psychometric and oculometric functions have similar slopes. (e) Noise model from Stone and Krauzlis (2003, p. 731; to be printed with permission). In response to a given
stimulus direction h, a motion signal is generated and transformed into output signals driving perception and pursuit, respectively. At each stage, noise (g) is added to the
signal. Because of their shared noise (gv), the two output signals are partially correlated. Because of the independent noise for perception (gp) and pursuit (gm) the correlation
is not perfect.
M. Spering, A. Montagnini/Vision Research 51 (2011) 836–852
assessed in a separate experiment. At the end of the open-loop
interval – here at ?175 ms after eye movement onset – eye move-
ment direction discriminated equally well between target motion
directions, regardless of direction. An analysis of neuronal re-
sponses in anaesthetized monkeys to stimuli moving in different
directions revealed no signature of directional anisotropies in area
A follow-up study by Krukowski and Stone (2005), however,
found a reliable oblique effect of comparable magnitude in percep-
tion and pursuit. These authors compared perception and pursuit
responses on the same trial. In a 2-IFC paradigm, observers had
to track stimulus motion and report the interval that contained
the more clockwise direction. The oblique effect in pursuit was
found for the open-loop interval, as well as for the closed-loop
interval (here defined as the time period 350–500 ms after eye
movement onset). The absence of an oblique effect in open-loop
pursuit in the Churchland et al. (2003) study could have been
due to methodological choices (e.g., the use of only three reference
directions) resulting in a lack in statistical power. The study by
Krukowski and Stone (2005) addressed this problem and used a
full set of cardinal and oblique reference directions. Given these
similarities in directional asymmetries in perception and pursuit
and the previous observation that MT neurons do not show an ob-
lique effect (Churchland et al., 2003), Krukowski and Stone (2005)
proposed a separation of information processing in or downstream
from area MST.
In contrast to the claim by Churchland et al. (2003), other stud-
ies, using functional magnetic resonance imaging (fMRI) in humans
(Furmanski & Engel, 2000), optical imaging in monkeys (Xu, Collins,
Khaytin, Kaas, & Casagrande, 2006), or modeling (Rokem & Silver,
2009) indicate that neuronal responses in areas V1 and MT carry
a signature of the oblique effect. However, none of the studies re-
ported in this section provide direct evidence for a separation of
motion processing information for perception and pursuit. Church-
land et al. (2003) merely reported an effect in one domain and no
effect in the other, and Krukowski and Stone (2005) reported obli-
que effects in both domains. Given that the neuronal basis for the
oblique effect has only been studied in early visual areas including
area MT, hypotheses about where a possible separation occurs –
beyond the almost trivial assumption that this must happen down-
stream from area MT – are difficult to draw from these studies.
22.214.171.124. Acceleration. Many studies have found that humans are less
sensitive to continuous changes in stimulus speed (acceleration)
than to step changes in stimulus speed (Snowden & Braddick,
1991; Werkhoven, Snippe, & Toet, 1992). Given the similarities be-
tween perception and pursuit with regard to direction and speed
discrimination, the pursuit system can be expected to show a sim-
ilar lack of sensitivity to acceleration changes as the perceptual
system. Watamaniuk and Heinen (2003) asked human observers
to discriminate the acceleration of stimuli that started at an initial
speed between 4 and 8?/s and accelerated at a constant rate (range
0–30?/s2). After each trial, observers had to judge whether the
stimulus accelerated more or less than an implicit standard – the
average acceleration of all the stimuli presented in the set. Percep-
tual and pursuit acceleration discrimination – based on the vari-
ability of eye acceleration during the interval 40–140 ms after
eye movement onset – were compared to perceptual and pursuit
discrimination of initial stimulus speed. Discrimination thresholds
(Weber fractions, the proportional change in target speed or accel-
eration producing responses that differed from the mean stimulus
75% of the time) in perception and pursuit were generally larger for
relative acceleration than for relative speed – replicating the origi-
nal finding that humans are better in processing visual information
about speed than acceleration –, and discrimination performance
for perception and pursuit was similar. These findings again sup-
port the notion of common processing for perception and pursuit.
On the neuronal level, the perceptual and behavioral insensitivity
to acceleration changes is reflected in the finding that individual
MT neurons are not tuned to acceleration; acceleration information
is obtained from a population response of speed-sensitive neurons
in area MT (Lisberger & Movshon, 1999; Price, Ono, Mustari, &
3.1.2. Are perception and pursuit faithful read-outs of noisy visual
Some of the studies discussed above (see Section 126.96.36.199) have
analyzed the efficiency of pursuit in response to visual motion dur-
ing the steady-state phase of the system, i.e. when eye velocity clo-
sely matches target velocity (Gegenfurtner et al., 2003; Stone &
Krauzlis, 2003; Tavassoli & Ringach, 2010). However, these ac-
counts are limited by a possible confound, because steady-state
pursuit is affected by both the retinal motion signal and extra-
retinal oculomotor feedback – an efference copy signaling ongoing
eye velocity (Robinson, 1965; Robinson, Gordon, & Gordon, 1986;
for a review see Lisberger et al., 1987). To characterize pursuit re-
sponses to visual motion signals alone, Osborne et al. (2005) ana-
lyzed the variability of monkeys’ pursuit eye movements during
the initiation phase – here, the first 125 ms after pursuit onset.
They concluded that pursuit variability was mostly due to sensory
errors in estimating target motion parameters such as time of on-
set, direction and speed, accounting for ?92% of the pursuit vari-
ability. In a follow-up study, Osborne, Hohl, Bialek, and Lisberger
(2007) estimated the time course of the pursuit system’s sensitiv-
ity to small changes in target direction, speed and time of onset.
This analysis was based on pursuit variability during the first
300 ms after target motion onset. Thresholds decreased rapidly
during open-loop pursuit and, in the case of motion direction, fol-
lowed a similar time course to the one obtained from the analysis
of neuronal activity in area MT (see Osborne, Bialek, & Lisberger,
2004). These studies suggest that the pursuit response – even in
the initiation phase – provides a faithful, and almost on-line ac-
count of motion information that is as efficient as perceptual judg-
ments. However, this last assumption is weak, as it is based on a
comparison of monkeys’ pursuit direction discrimination with
human perceptual direction discrimination data. Due to the differ-
ences in species and experimental paradigms, a direct, simulta-
neous evaluation of oculomotor and perceptual performance on a
trial-by-trial basis is not possible.
To address this concern, Rasche and Gegenfurtner (2009) tested
human observers on a speed discrimination task and directly com-
pared perception and pursuit. These authors followed a similar
mathematical analysis as described in Osborne et al. (2007) and,
importantly, also took temporal correlations in the variability of
pursuit velocity at different moments in time into account; the
analysis was done for short (open-loop phase only, first 300 ms
of pursuit) and long time intervals (open- and closed-loop pursuit,
first 400–500 ms of pursuit). In contrast to findings by Osborne and
colleagues, motor variability consistently affected pursuit in the
initiation phase, thereby leading to higher speed discrimination
thresholds in pursuit than in perception. Motor errors during this
phase accounted for ?50% of the pursuit variability. Interestingly,
when estimating oculometric thresholds on the basis of a longer
time interval spanning both the initiation and steady-state phase,
pursuit outperformed perceptual judgments, indicating that the
temporal integration of motion information plays a major role in
this kind of analysis. Differences between species (humans versus
monkeys) and the amount they were trained might explain the dis-
crepancy between the studies by Osborne and colleagues and by
Rasche and Gegenfurtner.
The neural substrate responsible for the differences in motion
discriminability between perception and pursuit, as revealed by
M. Spering, A. Montagnini/Vision Research 51 (2011) 836–852
Rasche and Gegenfurtner (2009), is not clear. Following the logic of
the studies by Osborne and colleagues, one would expect to see
very little noise added to the system downstream from cortical
motion processing areas V1 and MT, i.e., towards areas controlling
motor output. If pursuit variability is mostly due to motor noise,
however, substantial amounts of noise should be present in the
oculomotor control system. A direct comparison of the variability
in pursuit and neuronal responses in brain areas involved in trans-
forming sensory signals into eye movement commands showed
that little noise seems to be added in FEFsem and the floccular
complex in the cerebellum (Medina & Lisberger, 2007; Schoppik,
Nagel, & Lisberger, 2008; for a review see Lisberger (2010), his
Fig. 7). These results seem to provide direct evidence in support
of the idea that motion processing for pursuit initiation does not
imply any dramatic degradation of the visual signal. However, they
are based on trial-by-trial comparisons between pursuit responses
and neural activity in the respective areas, and do not involve a di-
rect comparison with perception. The finding by Rasche and
Gegenfurtner (2009) that pursuit discriminability outperformed
perception over longer time intervals is clearly inconsistent with
the idea that motor noise is added at later stages. Further studies
are needed to help explain the discrepancies between different
behavioral studies and differences found for pursuit initiation
and maintenance. In particular, a true direct comparison between
neural activity, pursuit and perception, with well-matched time
intervals for analysis, is still missing.
3.2. Motion segmentation and object motion integration
Traditionally, it has been assumed that pursuit eye movements
depend solely on the retinal motion input and on later feedback
from the oculomotor system (Robinson, 1965; Robinson et al.,
1986). However, there is evidence that higher-level visual and per-
ceptual mechanisms (for reviews, see Hafed & Krauzlis, 2010;
Krauzlis & Stone, 1999; Masson, Montagnini, & Ilg, 2010) as well
as cognitive factors (for a review see Barnes, 2008) play a major
role in pursuit control. In order to reflect this, some studies have
compared perception and pursuit in more complex visual scenes
where different motion signals are present at different positions
of the visual field. In such situations, an accurate interpretation
of the scene requires a reconstruction of the correspondence be-
tween different motion signals and different locations (motion seg-
mentation), and the integration of spatially different motion signals
into a coherent visual object (object motion integration).
3.2.1. Contextual effects on perception and pursuit
In our natural environment, visual objects are usually embed-
ded in and sometimes partially occluded by a richly structured, dy-
namic visual context. In order to track such a visual object with the
eyes, its motion signals have to be integrated into a coherent pat-
tern and spatially segregated from other motion signals in the
visual context. When the pursuit system has to select a target in
the presence of a second moving distractor, the initial pursuit re-
sponse usually follows the vector average of both motion signals
(Lisberger & Ferrera, 1997), i.e., it goes in the direction of the aver-
age motion signal, unless the response is cued towards one of the
two targets (Ferrera & Lisberger, 1995; Garbutt & Lisberger, 2006),
or the observer receives information about one target’s motion
trajectory (Recanzone & Wurtz, 1999; Spering, Gegenfurtner, &
Kerzel, 2006). Similarly, pursuit follows the vector average when
a visual target is surrounded by a dynamic visual context. A con-
text moving along with the pursuit target increases pursuit veloc-
ity; a stationary context or a context moving opposite to the
pursuit target decreases pursuit velocity (e.g., Masson, Proteau, &
Mestre, 1995; Spering & Gegenfurtner, 2007a).
However, these studies have not examined perceptual re-
sponses. The phenomenon of induced motion reveals that the per-
ceptual system can be unable to segregate motion signals from
spatially different sources: a moving object in the periphery can
induce motion so that a stationary, fixated target object appears
to move in the opposite direction to the physically moving object
(Duncker, 1929; Zivotofsky, 2005; see also Anstis & Casco, 2006;
Nawrot & Sekuler, 1990). How do perception and pursuit respond
to speed perturbations in a central target that is surrounded by a
dynamic visual context? Spering and Gegenfurtner (2007b) were
the first to report profoundly differential effects in perception
and pursuit. They compared both responses in a speed discrimi-
nation task in which changes in target speed had to be segregated
from speed changes in the visual context. A small target sur-
rounded by a spatially separated moving visual context either
moved at a constant speed of 11?/s or briefly increased or de-
creased in speed (perturbation range ± 6?). The context moved
at a constant speed or changed speed simultaneously with the
target, but in an independent direction (e.g., in a given trial, the
target could increase in speed while the context decreased in
speed). Observers had to track the target and indicate whether
it had increased or decreased in speed. Pursuit eye movements
clearly followed the vector average of target and context motion
(Fig. 2a; see also Kodaka, Miura, Suehiro, Takemura, & Kawano,
2004; Lindner, Schwarz, & Ilg, 2001; Miura, Kobayashi, & Kawano,
2009; Schwarz & Ilg, 1999; Spering & Gegenfurtner, 2007a;
Suehiro et al., 1999). Perceptual responses, however, went in the
direction of the induced motion: Observers systematically under-
estimated target velocity when context velocity increased, and
overestimated target velocity when context velocity decreased
(Fig. 2b). In trials with constant target velocity, a brief increase
in context speed alone therefore produced a transient increase
in eye velocity and a decrease in perceived target velocity (for
similar findings in perception, see Schweigart, Mergner, & Barnes,
These opposite effects of context perturbations on perception
and pursuit (Fig. 2c) resemble the structure of receptive fields in
area MT. Parafoveal and peripheral MT neurons have large recep-
tive fields (>10?, Pack & Born, 2001; see also Komatsu & Wurtz,
1988) and integrate motion signals over space, sometimes taking
into account information from outside the receptive field (Allman,
Miezin, & McGuinness, 1985; Born, Groh, Zhao, & Lukasewysc,
2000; Frost & Nakayama, 1983). One type of MT neurons responds
best to wide-field motion stimuli, which extend the area of the
classical receptive field. The other type of MT neurons does not re-
spond to these stimuli; it inhibits information from the surround.
The perceptual responses reported in Spering and Gegenfurtner
(2007b) might have been controlled by motion-sensitive neurons
with inhibitory surrounds whereas pursuit responses might have
been based on activity in neurons with excitatory receptive-field
structures that spatially sum over larger regions of the visual field.
The dissociation reported in Spering and Gegenfurtner (2007b) can
therefore be reconciled with the idea of common pathways – it
merely indicates that motion information for perception and pur-
suit can, under some circumstances, be processed differently or
by different populations of neurons in areas MT/MST.
There is also evidence that perturbations in the visual context
can affect perception and pursuit similarly. Debono, Schütz,
Spering, and Gegenfurtner (2010, this issue) asked observers to
indicate the motion direction of a RDK that moved horizontally
or in a diagonal direction slightly off the horizontal axis, while con-
currently recording eye movements. In some trials, a portion of the
dots in an extrafoveal location, at different positions relative to the
observer’s center of gaze, were perturbed in direction. Both percep-
tion and pursuit were most influenced if the perturbation angle
was similar to the motion direction angle of the RDK, and if the
M. Spering, A. Montagnini/Vision Research 51 (2011) 836–852
perturbation was close to the fovea. Both systems can therefore
integrate motion signals over the same spatial range.
3.2.2. Object motion integration for perception and pursuit
188.8.131.52. Aperture problem. Most studies presented so far have used
single dots or dot patterns as the pursuit target. For more complex
moving objects, an important question is whether perception and
pursuit follow global object motion, or a combination of retinal
motion signals of each local object component. Under particular
stimulus conditions, such as those described by the aperture prob-
lem (Wallach, 1935), motion segmentation and motion integration
can be difficult, leading to an ambiguous interpretation of the vi-
sual scene and to multistable percepts. The aperture problem oc-
curs for instance when a moving object is occluded and only one
edge is visible through a small aperture. The moving edge then ap-
pears to move orthogonally to its orientation (Fig. 3a, red arrow),
irrespective of the global object motion (Fig. 3a, green arrow).
Aperture-problem like situations have been used to study the
ability of the motion processing system to integrate local motion
signals into a coherent global percept. Some studies directly com-
pared perception and pursuit (Beutter & Stone, 2000), others stud-
ied both systems independently but in parallel, using similar
stimuli and comparing a wide range of visual features (e.g., Castet,
Lorenceau, Shiffrar, & Bonnet, 1993; Lorenceau, Shiffrar, Wells, &
Castet, 1993; Masson & Stone, 2002; Wallace, Stone, & Masson,
Beutter and Stone (2000) analyzed motion direction perception
and the direction of steady-state pursuit eye movements in re-
sponse to line-figure parallelograms moving behind stationary
rectangular apertures, such that only the motion of a segment of
each side of the object was visible (Fig. 3b, left). The motion trajec-
tory of the stimulus was either tilted to the right (?10?) or left
(+10?). When the apertures were visible (Fig. 3b, middle), percep-
tion and pursuit followed the true (physical, global) right- or left-
ward tilted motion of the parallelogram (see red and blue eye
traces, respectively, in Fig. 3b, middle), indicating that motion sig-
nals of individual components were integrated into a coherent ob-
ject. Directional judgments in perception and pursuit were
correlated on a trial-by-trial basis, indicating common processing.
In contrast, when the apertures were not distinguishable from the
background (Fig. 3b, right), responses followed the incoherent mo-
tion direction of local line segments along the orientation of the
aperture (see eye traces in Fig. 3b, right; for an earlier report of
similar perceptual results, see Lorenceau & Shiffrar, 1992). Overall
performance (when compared to the true motion direction) was
lower in perception and pursuit than with visible apertures. Direc-
tional judgments in perception and pursuit were compared with
three model predictions for motion integration (for a review, see
Bradley & Goyal, 2008; Weiss, Simoncelli, & Adelson, 2002): The
intersection of constraints (IOC) rule predicts that responses follow
the direction of true object motion. The vector average (VA) rule
predicts responses in the direction of the average motion signal
(see Section 3.2.1), in this case the orientation of the line segments.
Finally, the terminator motion rule predicts responses in the direc-
tion of corner or endpoint motion, in this case the orientation of
the aperture. With visible apertures, perception and pursuit fol-
lowed the IOC; with invisible apertures, responses were in be-
tween the IOC and terminator motion model. Surprisingly, the
direction of the vector average had only a small effect on perceived
and pursued motion direction, although VA is a robust and com-
mon rule for the integration of motion signals from multiple tar-
gets for open-loop pursuit (see Section 3.2.1), but possibly not for
steady-state pursuit (see also Spering et al., 2006).
When a tilted line is moving horizontally, local motion (line-ori-
entation related) signals (dashed red lines in stimulus orientation
in Fig. 3c, top) are not in agreement with the line’s global (horizon-
tal) motion (solid red lines in Fig. 3c, top). Castet et al. (1993) pre-
sented these stimuli briefly and asked observers to indicate their
motion direction. The perceived direction was biased towards the
strongest local motion signal, the direction orthogonal to line-
orientation (see also Lorenceau et al., 1993). Similarly, pursuit in
response to a horizontally moving tilted line or tilted parallelo-
gram stimulus was initially biased toward the edge-orthogonal
direction – the vector average of the object’s line segments in the
case of the parallelogram – and only later corrected towards the
object’s global motion (Montagnini, Mamassian, Perrinet, Castet,
& Masson, 2007b; Masson & Stone, 2002). Fig. 3c shows pursuit
eye velocity in response to a rightward moving vertical or tilted
line for three different stimulus speeds for one representative ob-
server. For the tilted line, the initial bias in the line-orthogonal
direction is indicated by an early transient positive component in
vertical eye velocity (see raw vertical eye velocity traces in
Fig. 3c, right). The size of the vertical bias depended on target speed
(see unbiased eye velocity traces in Fig. 3c, right), in line with the
general finding that visual stimulus properties (line length, lumi-
nance contrast, speed) affect oculomotor (Wallace et al., 2005,
Montagnini et al., 2007b) and perceptual accounts of the aperture
problem (Castet et al., 1993; Lorenceau et al., 1993) similarly.
These studies show that perception and pursuit integrate local
motion information similarly. The neurophysiological basis of the
local motion direction bias induced by the aperture problem, as
well as the dynamic evolution of the bias, has been clearly demon-
strated in area MT (Born, Pack, Ponce, & Yi, 2006; Pack & Born,
2001), suggesting that the perceptual and pursuit tracking error
-1 -0.5 0 0.5 1
Context speed (deg/s)
68 11 141768 1114 17
Eye speed (deg/s)
Context speed (deg/s)
Fig. 2. Pursuit and perceptual responses to speed perturbations in target and context obtained in Spering and Gegenfurtner (2007b). All data are for five observers. (a) Mean
eye velocity responses to context perturbations at fixed target speed 11.3?/s. Error bars are SEM. (b) Mean perceptual judgments of target speed change. (c) Correlations
between two model predictions (VA and motion contrast) and perception (blue) and pursuit (red). Class boundaries divide the plot into zones in which responses are
classified as VA-type or contrast-type responses. Data points falling in between the boundaries are considered as unclassified, which means that responses are well predicted
by both models as correlation coefficients did not significantly differ from each other (adapted from Spering and Gegenfurtner, 2007b, p. 1358).
M. Spering, A. Montagnini/Vision Research 51 (2011) 836–852
observed in humans and monkeys is a result of the common input
to both responses from area MT. One fundamental advantage of
studying pursuit judgments of motion integration as compared to
perception is that the pursuit response is updated continuously.
Different pursuit phases can therefore reflect the temporal dynam-
ics of motion integration – from local to global motion tracking –,
whereas perceptual judgments may be based either on motion sig-
nals at a specific moment in time or on the averaged signals across
the whole presentation time. This might be the reason why no
study to date has systematically compared oculomotor and percep-
tual accounts of the aperture problem within the same paradigm,
and on a trial-by-trial basis. Montagnini et al. (2007) proposed a
recurrent Bayesian model that mimics perceptual motion integra-
tion and allows qualitative predictions of pursuit responses to
ambiguous motion stimuli. This model has recently been extended
to include two modules for motion perception and eye movements
and now allows quantitative predictions for perceived motion
and pursuit eye movements (Boghadi, Montagnini, Mamassian,
Perrinet, & Masson, 2010, this issue).
184.108.40.206. Biological motion. Another example of complex motion is
that of biological motion – the representation of human or animal
movement patterns (Johansson, 1973). In the laboratory, biological
motion is reproduced by point-light stimuli (often termed
point-like walkers) with each point corresponding to a main joint.
Biological motion is of interest here, because it requires motion
integration of local signals, which (by themselves) provide little
information about the object’s global motion. It has been suggested
that the ability to perceive coherent biological motion – and there-
fore the ability to integrate local biological motion signals into a
global percept – relies on both local and global mechanisms, using
the local motion of individual dots corresponding to body parts, as
well as the global motion of the whole body (Troje, 2008).
A recent study compared perceptual and pursuit responses to
biological motion stimuli. Orban de Xivry, Coppe, Lefèvre, and
Missal (2010) asked observers to discriminate the heading direc-
tion of point-light walkers, and, in a separate experiment, to track
them with their eyes. The net motion of the stimuli was null, as
though the walker was moving on a treadmill. Control stimuli
were built by spatially scrambling the position of the walker-
points so that global motion differed while local motion remained
unchanged. Although the analysis of the psychophysical task was
less rigorous than in other studies comparing perception and pur-
suit, the general result was that both perceptual direction discrim-
ination accuracy and pursuit velocity gain (a standard steady-state
pursuit measure calculated as eye velocity divided by target
Local (1D) motion
Global (2D) motion
0 100 200 300 400
Time from target motion onset [ms]
0100 200 300 400
Tilted 5 deg/s
Tilted 10 deg/s
Tilted 15 deg/s
Vertical 5 deg/s
Vertical 10 deg/s
Vertical 15 deg/s
(tilted minus vertical)
Horizontal eye velocity [deg/s]
Vertical eye velocity [deg/s]
Fig. 3. (a) Schematic illustration of the aperture problem. A moving edge seen through an aperture appears to move orthogonally to its orientation, in this case diagonally
right and up (solid red arrow). The object’s global or true motion path, in this case to the right (green arrow), can be decomposed into two local vector components, one
parallel to the visible edge (dashed red arrow) and one orthogonal (solid red arrow). Only the orthogonal component is visible and drives motion perception. (b) Line-figure
parallelogram as used in Beutter and Stone (2000). Left: Schematic of the parallelogram. Only the solid white lines were visible. Middle: Stimulus with visible apertures and
example eye movement traces from one observer. Right: Invisible apertures and example eye movement traces. Eye traces are from Beutter and Stone (2000, p. 143; to be
printed with permission). (c) Top: Schematic of the visual stimulus used in Montagnini et al. (2007). Red arrows correspond to the global (or 2D, solid line) and local (or 1D,
dashed line) motion signal direction. Bottom: Mean eye velocity (left: horizontal, right: vertical) for one representative observer in response to a tilted (solid lines) or a
vertical (dashed lines) moving line; colors denote responses to three different stimulus speeds. The right-bottom panel shows the ‘‘unbiased’’ pursuit response, obtained by
subtracting vertical eye velocity in response to the vertical line from vertical eye velocity to the tilted line (adapted from Montagnini, Mamassian, Perrinet, & Masson, 2007a).
M. Spering, A. Montagnini/Vision Research 51 (2011) 836–852
velocity) were significantly higher for the biological motion stim-
ulus than for scrambled motion. This finding is particularly inter-
esting considering that area MT, the key area for the processing of
translational motion, seems less crucially involved in the process-
ing of this complex motion: MT lesions in human patients did not
necessarily disrupt biological motion processing (Billino, Braun,
Böhm, Bremmer, & Gegenfurtner, 2009). The neural substrate of
biological motion processing includes areas other than MT, lying
outside the ‘‘classic’’ motion pathway, such as a region on the
superior temporal sulcus as well as occipital and fusiform face
areas (Grossman & Blake, 2002; Michels, Lappe, & Vaina, 2005).
The link between perception and pursuit might therefore extend
to other cortical areas that are usually devoted to the processing
of form rather than motion.
3.3. Visual features modulating motion perception and pursuit
Perception and pursuit of moving objects can be influenced by
visual features unrelated to the object’s motion properties, such
as luminance contrast and color. The studies reviewed in this sec-
tion report similarities in how perception and pursuit process vi-
sual motion when the moving target’s luminance contrast is low,
or when it is modulated by color.
3.3.1. Luminance contrast
Motion perception is strongly influenced by the moving target’s
luminance contrast. A low-contrast stimulus moving at a given
speed is perceived to move slower than a high-contrast stimulus
moving at the same speed (Stone & Thompson, 1992; Thompson,
1982; for a model see Weiss et al., 2002). A corresponding phe-
nomenon to perceptual slowing was demonstrated in the pursuit
response in humans (e.g., Spering, Kerzel, Braun, Hawken, &
Gegenfurtner, 2005) and monkeys (Priebe & Lisberger, 2004). As
a function of contrast, pursuit latency decreased and velocity gain
increased. Although perception and pursuit have not been com-
pared on a trial-by-trial basis, these studies imply that stimulus
contrast influences perception and pursuit similarly. Note that
the finding of perceptual slowing holds for relatively slow speeds
only. For higher speeds, a paradoxical increase of perceived speed
has been observed at low contrast (Thompson, Brooks, & Hammett,
2006). To our knowledge, this effect has not yet been studied in
pursuit eye movements.
When a moving stimulus is isoluminant to the background –
when it has the same luminance than the background and differs
from the background only in color – the stimulus is usually per-
ceived to move up to 50% slower than a luminance-defined stimu-
lus moving at the same physical speed (Cavanagh, Tyler, & Favreau,
1984; Gegenfurtner & Hawken, 1995; Gegenfurtner et al., 1994) or
even perceived to stand still (Lu, Lesmes, & Sperling, 1999). This
effect of perceptual slowing has been found with relatively
slow-moving stimuli (< 4?/s). Impairments have recently also been
reported for pursuit eye movements. Braun et al. (2008) measured
the effects of isoluminant stimuli on pursuit characteristics and
compared speed judgments to luminance- and color-defined tar-
gets during fixation and pursuit (see Section 4.1.2). The strongest
impairments in response to color stimuli were found during the
initiation phase: Pursuit latency was delayed by 50 ms, and initial
eye acceleration was reduced. Interestingly, pursuit slowing and
latency increases were also found for stimuli moving at higher
speeds (up to 10.3?/s). Only a small but significant difference was
observed for steady-state pursuit velocity. Motion signals from col-
or stimuli are therefore weaker in driving both perception and
Spering et al. (2008) come to a similar conclusion in a study
comparing target preferences for color and luminance in pursuit,
saccades, and perception. Observers initially tracked a horizontally
moving target that split into a color- and a luminance-defined
component. After the split, the two stimuli either moved in two
diagonal directions (pursuit task) or reappeared in two peripheral
locations (saccade task). Observers had to choose the more salient
stimulus with their eyes. In separate experiments, we measured
perceptual salience judgments during pursuit/saccades. Perceptual
judgments and early pursuit responses were strongly biased to-
wards the luminance stimulus (whereas saccades, in comparison,
showed a clear preference for color). In about one third of all trials,
pursuit was reversed towards the color stimulus by a saccade
occurring at ?65 ms after pursuit onset. Apart from the interesting
dissociation in target preference between pursuit and saccades –
indicating different processing mechanisms for target selection in
the two types of eye movement – this study reveals a close rela-
tionship between perception and the earliest pursuit response with
regard to the processing of color and luminance information.
While these studies show that color information is less readily
available to the motion perception and pursuit system, there is also
evidence that chromatic motion information can be used as a cue
for the interpretation of visual motion direction, specifically, for
motion integration and segmentation (for reviews, see Dobkins &
Albright, 2003; Gegenfurtner & Hawken, 1996). Dobkins and
Sampath (2008) reported that the influence of color information
as a motion-segmentation cue might be stronger for perception
than for eye movements. However, these authors studied a mixture
of pursuit and OKN and did not differentiate their results based on
the type of eye movement observed. Conclusions regarding color
information as a motion-segmentation cue for either pursuit or
OKN and the finding that perception and eye movements might
use motion information differently therefore have to be treated
3.4. Perception and pursuit of physical versus perceived motion
A particularly interesting type of motion stimulus is illusory
motion, where perceived and physical stimulus motion do not
match. A reliable pursuit response can be elicited and maintained
in the absence of visual motion on the retina, provided the
observer perceives visual motion. This was first demonstrated by
Steinbach (1976), who attached two light-emitting diodes (LEDs)
on the opposite sides of a wheel riding along a horizontal track
in front of an observer sitting in the dark. Observers not only per-
ceived a rolling wheel, but were also able to track its imagined cen-
ter smoothly with their eyes. Recent studies compared perception
and pursuit in response to apparent motion stimuli such as illusory
contours and the motion aftereffect. We briefly review studies on
non-visual motion to address the claim that pursuit can be elicited
in the complete absence of a visual stimulus.
3.4.1. Apparent motion and moving illusory contours
Madelain and Krauzlis (2003) used a directionally ambiguous
multi-stable stimulus – a physical image that evokes two possible
perceptual representations (here, left- and rightward motion) –
and compared perceptual and pursuit reversals in motion direction
with regard to their timing. The stimulus was an array of Kanitza-
style illusory squares defined by illusory contours, which were pro-
duced by circular inducers placed at the corners of the square (see
Fig. 4a). The orientation of the inducers was changed rapidly on
every presentation frame, evoking a bi-directional motion percept
(i.e., left- and rightwards) of the illusory squares. An auditory tone
waspresented at a random timeduring eachtrialand observershad
to indicate in a 2-AFC task whetherthe toneoccurred before or after
the perceptual reversal. Eye movements were tracked concurrently.
M. Spering, A. Montagnini/Vision Research 51 (2011) 836–852
This study provides three interesting results: First, accurate pursuit
(with a steady-state velocity gain close to 1) of illusory motion is
possible (see also Lamontagne, Gosselin, & Pivik, 2002; van der
Steen, Tamminga, & Collewijn, 1983). Second, perceived reversals
were correlated with reversals in pursuit direction. Third, pursuit
reversals consistently followed perceptual reversals by ?50 ms,
suggesting that pursuit can provide a real-time readout of per-
ceived motion. This study and others reported here observed that
pursuit follows the visual percept, rather than retinal motion infor-
mation. Note that this finding does not imply causality with regard
to the perception–pursuit relationship, as causal effects have not
been systematically analyzed in this context.
Ilg and Thier (1999, 2003) compared pursuit and neuronal re-
sponses in monkeys to illusory motion using a stimulus, in which
the only visual information was carried by peripheral contour-
inducers and the target center was only imaginary (see Fig. 4b).
Again, monkeys were perfectly capable of tracking the imaginary
target with a steady-state gain close to 1, whereas pursuit of a sin-
gle, extra-foveal stimulus at the same eccentricity as the inducers
was much less efficient. The activity of MT neurons was signifi-
cantly reduced, but the activity in a subpopulation of neurons in
areas MST and FEF was similar to that observed for pursuit in re-
sponse to real contours under foveal stimulation. These findings
indicate that MT carries signals related to visual image motion,
whereas MST and FEF activity might reflect the extra-retinal eye
movement signal (efference copy), implying that common neuro-
nal processing of motion information for perception and pursuit
is not strictly limited to MT. Interestingly, in a more recent study,
Biber and Ilg (2008) found that pursuit latency increased consider-
ably for illusory contours versus real contours. In contrast to what
has been reported for real contours (see Section 3.2.2), when the
illusory contour was tilted with respect to its motion direction, ini-
tial pursuit responses still followed the true contour motion, and
showed only a small bias in the direction orthogonal to the contour
orientation. These two results suggest that retinal motion informa-
tion might be a more important driving signal for pursuit initiation
than for steady-state pursuit, when the motion percept seems to
3.4.2. Motion aftereffect
Studies on the motion aftereffect (MAE) – a visual illusion that
arises from prolonged adaptation to a moving pattern, causing
observers to subsequently perceive stationary objects as moving in
the opposite direction to the adapting pattern – provide more evi-
stances, the MAE can drive a reliable pursuit response in humans
(Braun, Pracejus, & Gegenfurtner, 2006; Watamaniuk & Heinen,
2007) and monkeys (Gardner, Tokiyama, & Lisberger, 2004). In their
study, Braun et al. (2006) directly compared the effect of a MAE on
perception and pursuit. Observers were first adapted to a vertical
sine-wave grating presented at high (100%) or medium (40%) con-
trast that moved continuously to the left or right at ?8?/s for 30 s
during fixation. Observers were asked to indicate the motion direc-
tion of a subsequently presented test grating (left, right, or station-
ary) while their eye movements were recorded. Gratings at
medium, but not at high contrast, reliably elicited a MAE in pursuit
– a response to a physically stationary test grating that was per-
ceived to move in the opposite direction to the adapting stimulus.
Conversely, the eyes remained stationary in response to a stimulus
ary. The velocities at which the stimulus was subjectively perceived
as stationary were almost identical for perception and pursuit, indi-
cating MAEs of similar magnitude in perception and pursuit. These
findings again support the assumption of a close link between mo-
tion processing for perception and pursuit and suggest a common
neural substrate for adaptation in perception and pursuit in area
MT (see Gardner, Tokiyama & Lisberger, 2004; Kohn & Movshon,
2004; Watamaniuk & Heinen, 2007).
3.4.3. Pursuit and perception of non-visual motion
In his influential study on pursuit in response to perceived mo-
tion, Steinbach (1976) noted: ‘‘[...] the fundamental requirement
for pursuit is the appreciation of an object in motion with respect
to the observer irrespective of retinal stimulation, and [...] irre-
spective of the sense modality through which motion is assessed.’’
(p. 1371). Interestingly, tracking of non-visual objects such as audi-
tory, tactile or proprioceptive targets with pursuit eye movements
has been reported as very poor, with particularly low velocity gains
in response to moving auditory stimuli (Berryhill, Chiu, & Hughes,
2006; Boucher, Lee, Cohen, & Hughes, 2004; Gauthier & Hofferer,
1976). These findings exclude a comparison between perception
and pursuit, even though motion perception of non-visual targets
can be intact. Pursuit can further be affected by non-visual signals
such as cognitive expectations about target motion direction
and there is some evidence that perception is affected similarly
(Krauzlis & Adler, 2001). Motion expectation is also a powerful
drive for anticipatory smooth pursuit, i.e. pursuit performed before
the actual availability of visual target motion information (for a re-
view, see Barnes, 2008), but this phenomenon does not have a per-
ceptual correlate and so again excludes a direct comparison
between perception and pursuit.
4. Interaction of perception and pursuit
4.1. Effects of pursuit on perception
When a visual stimulus is briefly presented before or during a
saccadic eye movement, its perception can be blurred or even sup-
pressed. This loss of sensitivity for visual information during sac-
cades is known as saccadic suppression (e.g., Bridgeman, Hendry,
& Stark, 1975; Burr, Holt, Johnstone, & Ross, 1982) and both retinal
and extra-retinal mechanisms have been proposed to explain this
Fig. 4. Apparent and illusory motion stimuli. (a) Kanitza-style stimulus used in Madelain and Krauzlis (2003; p. 644, to be printed with permission). (b) Hourglass-shaped
stimulus with blanked target center as used in Ilg and Thier (2003). In the main experiment, monkeys were instructed to track the invisible, ‘‘imaginary’’ target center. (c)
Occluded line object (termed a ‘‘chevron’’) from Hafed and Krauzlis (2006, p. 1450; to be printed with permission). Only the white line segments were visible. Left: fixation
condition, right: pursuit condition.
M. Spering, A. Montagnini/Vision Research 51 (2011) 836–852
phenomenon (for reviews see Krekelberg, 2010; Wurtz, 2008).
During a saccade, the eyes can move at speeds up to 1000?/s
(e.g., Carpenter, 1988) and saccadic suppression might help pre-
venting spatial instability due to saccade-induced, rapid large-field
shifts of the retinal image (Ross, Morrone, Goldberg, & Burr, 2001).
As a side note, it is interesting that motion perception can be intact
during saccades (Castet & Masson, 2000). During pursuit eye
movements, where the eyes move at much slower speeds than
during saccades, no systematic suppression has been reported dur-
ing the initiation phase (Schütz, Braun, & Gegenfurtner, 2007).
However, it is well documented that the execution of pursuit eye
movements can cause misperceptions of stationary and moving
objects (e.g., Morvan & Wexler, 2009; Souman, Hooge, & Wertheim,
2005; for a review, see Freeman, Champion, & Warren, 2010). On
the other hand, recent studies have found perceptual benefits dur-
ing pursuit eye movements (e.g., Hafed & Krauzlis, 2006; Spering,
Schütz, Braun, & Gegenfurtner, submitted for publication). Finally,
other studies have reported that pursuit has no effect on percep-
tion (e.g., Krukowski, Pirog, Beutter, Brooks, & Stone, 2003). The
evidence for whether and under which circumstances pursuit
helps or impairs motion perception, and what aspects of vision suf-
fer or benefit, and how, is therefore mixed.
4.1.1. Pursuit impairs motion perception
Tracking a moving object with pursuit eye movements always
produces a motion signal on the retina, induced by the motion of
the stationary background. Because we generally perceive station-
ary objects as stationary and moving objects as moving, even dur-
ing pursuit, this movement-induced retinal motion signal has to be
cancelled to maintain perceptual stability. This cancellation might
be achieved through a comparison of an external (retinal) motion
signal with an internal (extra-retinal) reference signal that reflects
the motor command of the eye movement (von Helmholtz, 1910/
1962; von Holst & Mittelstaedt, 1950). But although the notion of
a system that compensates movement-induced motion signals is
well established and the underlying neural network well studied
(e.g., Thier, Haarmeier, Chakraborty, Lindner, & Tikhonov, 2001),
compensation during pursuit eye movements is usually imperfect.
Misperceptions include stationary objects, objects moving along
with the pursuit target, and objects moving perpendicular to the
In the Filehne illusion, a briefly presented stationary object ap-
pears to move in the direction opposite to the pursuit eye move-
ment (Filehne, 1922; Freeman & Banks, 1998; Haarmeier & Thier,
1996, 1998; Mack & Herman, 1973). The Aubert–Fleischl phenom-
enon describes a case in which a visual object appears to move
slower when it is smoothly tracked than when the observer views
it during fixation (Aubert, 1887; Turano & Heidenreich, 1999; von
Fleischl, 1882; Wertheim & van Gelder, 1990). Finally, objects
that move perpendicularly (Souman et al., 2005) or diagonally
(Festinger, Sedgwick, & Holtzman, 1976; Morvan & Wexler, 2009)
relative to the pursuit trajectory are perceived to move at an angle
rotated further away from the pursuit target. These misperceptions
might be due to an imperfect compensation for eye movement-
induced retinal image motion (e.g., Haarmeier, Thier, Repnow, &
Petersen, 1997). Freeman et al. (2010) recently developed an alter-
native explanation and showed that a simple Bayesian model can
explain various misperceptions during pursuit. This model is based
on the general idea that prior expectations increasingly influence
perceptual decisions as sensory signals become uncertain (e.g.,
Stocker & Simoncelli, 2006). The finding that stimulus speed is
more difficult to discriminate during pursuit than during fixation
could therefore be explained by the assumption that we are
expecting the world to be stationary, or, in other words, that the
prior for motion is centered at zero (Weiss et al., 2002) both for im-
age motion and pursuit-target motion (Freeman et al., 2010).
4.1.2. Pursuit enhances motion perception
Pursuit might impair perception under some circumstances, but
it can also enhance it, in line with the idea that eye movements
generally improve vision (Land, 2006). Hafed and Krauzlis (2006)
asked observers to judge the coherence of a partially occluded line
object (a ‘‘chevron’’) that moved in a circle behind two occluders
(Fig. 4c). Observers either maintained fixation in the center be-
tween the two occluders (Fig. 4c, left), or tracked the fixation spot
that moved circularly along with the occluders (Fig. 4c, right),
while the line object was fixed in space. Both eye movement con-
ditions yield similar retinal image motion. Perceptual coherence
(the perceived alignment of the segments behind the two occlud-
ers) was better during pursuit than during fixation, regardless of
image features such as shape, added noise or presence of a refer-
ence frame. These results show that ongoing motor commands de-
rived from pursuit eye movements can be used to perceptually
disambiguate spatial relationships between sensory features of vi-
Moreover, these motor commands can also inform perceptual
judgments about motion direction. In Spering et al. (submitted
for publication), observers had to judge whether a linearly moving
target (ball) would hit or miss a stationary vertical line (goal). Ball
and goal were presented briefly for 100–500 ms, and disappeared
from the screen together before the perceptual judgment was
prompted. Observers were asked to either pursue the ball or fix-
ate. In one version of the experiment, observers fixated on a sta-
tionary ball while the goal was moving towards fixation. In this
condition, retinal stimulation was similar to the pursuit condition.
In another version of the experiment, observers fixated on the
goal while the ball was moving towards fixation. Results show
that perceptual performance was significantly better during pur-
suit than during fixation, regardless of fixation position. The per-
formance difference between pursuit and fixation has to be due to
extra-retinal motion direction information gained from the pur-
suit response – through an efference copy signal as well as the
pursuit direction error (the angular velocity difference between
eye and ball). Pursuit, even if it is not perfectly accurate, can
therefore aid the prediction of visual motion in space. However,
pursuit does not seem to benefit the temporal predictions of
visual motion (i.e., about the time when a target will reach a cer-
tain position), as demonstrated with a task in which observers
had to judge time-to-contact of an accelerating object (Benguigui
& Bennett, 2010).
A similar advantage of pursuit over fixation was found for the
perception of motion smear, which is generally elicited by a single
moving image when viewed during fixation on a secondary stimu-
lus. Interestingly, perceived motion smear was lower when evoked
by a stationary stimulus and viewed during pursuit than when
evoked by a moving stimulus and viewed during fixation (Bedell
& Lott, 1996). Given the similarity of retinal motion input in both
conditions, the advantage has to be – again – attributed to an ex-
tra-retinal motion signal, which benefits perception.
Braun and colleagues (2008) showed that pursuit improves per-
ceptual judgments of speed when the moving stimulus is modu-
lated by color. Speed judgments of moving isoluminant stimuli
were veridical during pursuit but impaired during fixation. Here,
judgments reflected a substantial slowing up to 30%. This finding
is surprising, given that pursuit initiation and acceleration to isolu-
minant targets is slowed down. Therefore, motion information
about isoluminant stimuli seems to be available during steady-
state pursuit, but not during fixation. Pursuit also influences the
perception of color itself – pursuit can, interestingly and almost
paradoxically improve the sensitivity to color (Schütz, Braun, &
Gegenfurtner, 2009; Schütz, Braun, Kerzel, & Gegenfurtner, 2008).
A detailed description of these studies goes beyond the scope of
the current review.
M. Spering, A. Montagnini/Vision Research 51 (2011) 836–852
4.1.3. Pursuit does not affect perception
Other studies found that pursuit does not always inform percep-
tion (Freeman, Champion, Sumnall, & Snowden, 2009; Krukowski
et al., 2003; Tavassoli & Ringach, 2010). Krukowski et al. (2003)
found no advantage of pursuit over fixation in a perceptual direc-
tion discrimination task. Direction thresholds were similar during
fixation and pursuit, and perceptual performance was not related
to pursuit gain. These authors used a memory task with two inter-
vals in which a visual motion signal had to be compared to an inter-
nal reference. Freeman et al. (2009) also used a 2-IFC task in which
observers had to indicate which interval contained the faster back-
groundmotion,whilepursuinga target thatmovedacrossthe back-
ground. Observers based theirjudgments mostly on relative motion
between target and background and did not take retinal motion
information into account, even if they received feedback on their
own eye velocity after each trial. An important difference to many
studies reported in the previous sections is that here, observers
were asked to judge the speed of the motion surrounding the pur-
suit target and not that of the pursuit target itself. The finding that
pursuit is more sensitive to small speed changes than perception
(Tavassoli & Ringach, 2010; see Section 220.127.116.11) shows that visual
signals can drive eye movements in the absence of a corresponding
conscious visual percept. Generally, the availability of internal mo-
tion signals (speed or direction) to the perceptual system seems to
depend on task requirements.
5. Principles for motion processing for perception and pursuit
Although the perception of visual motion and pursuit eye move-
ments have been shown to be closely linked, pursuit can also be
independent of a concurrent visual percept. This difference is in
line with the more general idea that visual information undergoes
partly independent processing for visual perception and the guid-
ance of motor action (Goodale & Milner, 1992) and might therefore
not surprise some readers. Furthermore, perceptual and pursuit re-
sponses are given through different effectors, implying a necessary
separation of the information streams toward the motor-output
end. Performance differences between both responses also fit well
with the notion that motion perception and pursuit eye move-
ments have inherently different task demands. Perception serves
the visual representation and interpretation of moving objects,
whereas pursuit eye movements imply an interaction with our vi-
sual world through the representation of visual motion signals in
motor responses. Both responses further underlie different tempo-
ral constraints – they differ in response time – and spatio-temporal
constraints: it is possible to covertly track multiple moving objects,
but not to simultaneously track them all with the gaze; pursuit re-
quires the selection of a single target or the construction of an ob-
ject center within a larger moving pattern.
Yet, the evidence in favor of a largely common neural substrate
for motion perception and pursuit control – at least throughout the
visual part of the processing stage – is overwhelming, suggesting
that a functional separation into a perception and an action path-
way is unlikely for an earlier stage of visual motion processing. Dis-
sociations between perception and pursuit should also be
surprising to those who believe that a discrepancy between what
we perceive and what we track with our eyes might be strategi-
cally inefficient and potentially dangerous – a notion that matches
our daily life experience.
We hypothesize that behavioral differences between perception
the perceptual and the oculomotor system with regard to task de-
mands and temporal constraints and are therefore compatible with
an overall agreement of the two modalities. These functional differ-
the processing of visual motion information.
5.1. Different responses to different task demands
When the observer sees a moving object of interest and decides
(or is instructed) to track it, the pursuit control system has to pro-
vide a continuous quantitative estimate of the required force that
has to be applied to the eye muscle in order to move the eye while
minimizing retinal slip (e.g., Robinson, 1965; Robinson et al.,
1986). The initiation of a pursuit eye movement requires the effi-
cient use of time-varying information transmitted by direction-
and speed-tuned neurons in the visual cortex (e.g., Lisberger &
Movshon, 1999). Interestingly, the earliest phase of pursuit – the
first ?40 ms – is relatively unselective for visual stimulus features
other than motion direction (Lisberger et al., 1987; Rashbass,
1961), probably a cost of the relatively short latency of 80–
120 ms (compare to 200–250 ms in saccades). During the pursuit
maintenance phase, the eye velocity feedback signal helps to main-
tain steady-state pursuit. However, in healthy untrained observers,
steady-state pursuit velocity gain is usually smaller than 1, indicat-
ing that the eye lags behind the target (providing a driving signal
for catch-up saccades; De Brouwer et al., 2002). These observations
indicate that the pursuit eye movement system can be surprisingly
imprecise with regard to velocity matching. Initially, it seems to be
the pursuit system’s ‘‘job’’ to program an eye movement fast and
into the correct direction with relatively little time spent on the
extraction of precise visual signals.
The perceptual system, on the other hand, has to detect or dis-
criminate particular visual features and must therefore provide a
precise, deep analysis of visual signals. As an aside, it is not clear
whether these different task demands are inherent to the systems
or whether they are a result of the way pursuit and perception are
usually tested. In all psychophysical studies reported here, observ-
ers had a specific task with regard to perception (e.g., ‘‘report
whether the target became faster or slower’’), but not with regard
to pursuit, where the instruction was to simply track the stimulus
with the eyes. It might be interesting to test whether variations to
the pursuit instruction (e.g., ‘‘track the target’s speed change’’),
thereby matching the level of scrutiny required from both systems,
could influence results.
5.2. Continuous versus discrete responses: differences in temporal
resolution and time course
Pursuit eye movements usually start at around 80–120 ms after
stimulus motion onset and vary dynamically over time, providing a
continuous, analog readout of motion signals. In contrast, motion
perception is typically tested at discrete moments in time and usu-
ally takes the form of a discrete, binary response. However, observ-
ers will most likely base their perceptual judgments on the entire
time period in which a motion signal was presented. The size of the
temporal integration window could therefore be larger for percep-
tion than for pursuit. The uncertainty about the temporal resolu-
tion of the perceptual response is (or should be) a major concern
in studies that directly compare perception and pursuit, because
the use of a larger time window for motion integration essentially
low-pass-filters the perceptual response. One way to address this
problem is to present stimuli for short time periods only and,
where applicable, to use visual masking techniques to enforce a
more rigorous control of presentation time. These measures allow
better control over the amount of time available to accumulate
sensory evidence for a perceptual decision. However, an analysis
of pursuit for different time intervals has to take into account that
pursuit at different times might be dominated by different visual
Wilmer and Nakayama (2007) investigated accuracy differences
in speed estimation between early and late pursuit stages (before
and after the first catch-up saccade). Given the known differences
M. Spering, A. Montagnini/Vision Research 51 (2011) 836–852
between early and late pursuit, this study aimed at identifying the
(potentially separate) mechanisms driving these two pursuit
stages. The correlation between the moment-to-moment pursuit
accuracy across time and the perceptual performance was mea-
sured in two 2-IFC speed discrimination tasks. These involved
two kinds of stimuli, presumably driving the early versus late pur-
suit phase, respectively. Given the relative independence of early
pursuit responses to motion information other than direction, this
phase has been associated with ‘‘low-level’’ motion signals (here: a
drifting luminance-modulated sinusoidal grating), whereas ‘‘high-
level’’ motion information seems to be the more important driving
signal for the pursuit maintenance phase (here: contrast-modu-
lated circularly drifting rings with no net-luminance motion). Re-
sults showed that early pursuit was predicted by the precision of
low-level speed estimation, whereas late pursuit was predicted
by the precision of high-level speed estimation. According to Wil-
mer and Nakayama (2007), catch-up saccades mark the transition
from low-level to high-level motion computation. This study has
methodological and theoretical implications for the direct compar-
ison between perception and pursuit. First, it underlines that stim-
uli and behavioral tasks have to be matched well to the
requirements of the tested pursuit phase. Second, it raises the
question whether there is a corresponding phenomenon in percep-
tion that operates the transition between different motion process-
ing mechanisms. In their commentary on the study by Wilmer and
Nakayama (2007), Krauzlis and Hafed (2007) suggested that the
occurrence of the catch-up saccade could simply be temporally
correlated with the transition, rather than causally related to it.
5.3. Methodological considerations in the comparison between
perception and pursuit
Differences in task demands (Section 5.1) and different re-
sponse times (Section 5.2) are two crucial methodological issues
to consider when comparing binary perceptual judgments and a
dynamic motor response such as pursuit directly, on a trial-by-trial
basis. First, perception and pursuit are not independent and pur-
suit can both impair and enhance perceptual performance, with
the direction of the effect depending on the particular task (see
Section 4.1). Second, perception and pursuit differ with regard to
the time that both systems have available to make a decision. Most
studies controlled for the effect of temporal integration by manip-
ulating target presentation duration as well as the pursuit analysis
interval. However, none of these studies analyzed speed–accuracy
trade-offs to test a possible effect of perceptual reaction time on
perceptual performance and the perception–pursuit comparison.
In other words, temporal integration of motion information for
perceptual judgments could occur beyond the limits of the stimu-
lus presentation interval (especially in the absence of visual mask-
ing) and responses with longer reaction times could therefore lead
to better perceptual performance. Third, most studies required
observers to commit to a dual task – tracking a visual stimulus
and making a perceptual decision simultaneously – imposing a dif-
ferent load on attention and working memory than the separate
testing of both responses. Although this does not necessarily re-
quire a shift of attention away from the pursuit target, there might
have been an overall cost related to higher cognitive load.
5.4. Vector averaging as a distinctive characteristic of pursuit initiation
These differences between perception and pursuit with regard
to task demands and timing might explain why both responses
are sometimes dominated by different computational mechanism.
As described above (see Section 3.2.1), when challenged by a time-
critical task such as target selection, the pursuit system is pro-
grammed to follow the ‘‘quick and dirty’’ solution of the VA, with
the possibility to correct the initial choice later, when target selec-
tion has been accomplished. A VA solution is not only optimal in
minimizing retinal slip, but it is also the ‘‘best guess’’ (leading to
a statistically minimal tracking error) in the case in which there
is no a priori reason to favor one moving object over the other.
The VA is a robust phenomenon often observed during pursuit
For perception, reports of VA are less common (but see Yo &
Wilson, 1992). In our daily life, we do not usually rely on the aver-
age of perceived motion vectors to make a perceptual decision.
Imagine a Lisberger and Ferrera (1997)-like situation, where two
targets move towards the fovea, in our natural environment. If
two cars move towards a common location from different posi-
tions, we would probably get nervous about a possible crash rather
than having the impression of a global motion along the VA direc-
tion – although our eyes might well choose the average motion
path. Many studies have reported that perception integrates mo-
tion information from two sources by following the motion con-
trast – the difference between two motion signals (for a review
see Spering & Gegenfurtner, 2008) – again possibly reflecting dif-
ferences in task requirements and timing.
5.5. Accumulate-to-threshold model
On a mechanistic level, processing differences between percep-
tion and pursuit might be reflected in a model where the same vi-
sual signals, but different internal decision signals, guide both
responses (for a similar idea, developed for the comparison be-
tween pursuit and saccades, see Liston & Krauzlis, 2005). Following
a simple ‘‘accumulate-to-threshold’’ model, differences in behavior
could result from a combination of same or different internal deci-
sion signals driving perception and pursuit to same or different re-
sponse thresholds (Fig. 5).
For example, latency and sensitivity differences between per-
ception and pursuit might be the result of different underlying re-
sponse thresholds for perception and pursuit. A lower response
threshold (as shown in Fig. 5 for pursuit) usually leads to faster re-
sponses but also to a higher number of errors, and therefore to an
overall lower sensitivity. Note that this accumulate-to-threshold
model is a simplified model, designed to visualize some of the
behavioral differences between perceptual and motor decisions
(such as differences in sensitivity or response latency). However,
this model does not allow any predictions for responses in more
complex stimulus environments, e.g. when more than one stimulus
is present and a target has to be selected; the model does not eluci-
date the processes underlying perceptual and visuomotor decision-
making. In showing that perception and pursuit mostly share the
same visual input, the studies discussed in this review provide a
first step towards such a model of perception and pursuit. However,
Fig. 5. Hypothetical and simplified mechanisms for the control of perceptual and
pursuit responses to motion input. Differences in latency and sensitivity between
both systems could be explained by (1) the same decision signal (either D1 or D2)
guiding perception and pursuit to different thresholds, (2) by different decision
signals D1 and D2 guiding to the same threshold, or (3) by different decision signals
guiding to different thresholds.
M. Spering, A. Montagnini/Vision Research 51 (2011) 836–852
testing this model would require a speed–accuracy type of analysis,
which has not been performedyetin studies thatdirectlycompared
perception and pursuit (see Section 3.1).
5.6. Different neuronal substrates beyond MT/MST?
Many studies reviewed here concluded that the similarity be-
tween perception and pursuit is grounded on the largely common
underlying neuronal processing, especially in areas MT and MST.
However, as mentioned throughout this review, a separation must
occur – most likely downstream from MT/MST. Several cortical
areas other than MT/MST partake in the processing of motion
information for pursuit control (e.g., the frontal and supplementary
eye fields and the cerebellum; see Krauzlis, 2004; Lisberger, 2010),
but are probably less involved in motion processing for perception.
Where in the brain the perceptual decisions about visual motion
are taken is much less clear, although an important role has been
assigned to area LIP (Williams, Elfar, Eskandar, Toth, & Assad,
2003). This study reported that neuronal activity in LIP correlated
with the perceived motion direction of an apparent motion stimu-
lus, while activity in area MT did not reflect subjective perception.
This finding underlines the importance of higher parietal areas for
the formation of perceptual decisions. In sum, areas MT/MST are
crucial for the processing of visual motion information for both
perception and pursuit, but they are not the final processing stage.
This review summarized studies from an active field of research
which has developed in the last ?20 years by fusing two tradition-
ally independent research lines – classic psychophysics of visual
motion on the one hand, and the oculomotor control-theory ap-
proach on the other hand – into a unique conceptual framework
and a common set of experimental techniques. Some of the chal-
lenges for future research in this area will be to understand the
involvement of higher-level brain areas beyond MT/MST in motion
processing for perception and action, to generalize the principles
identified here to other action systems, and to test existing model
ideas by analyzing speed–accuracy trade-offs between perception
Supported by German Research Foundation Fellowship SP1172/
1-1 to MS and by EC IP Project FP6-015879 ‘FACETS’ to AM. The
authors would like to thank Guillaume Masson and Alexander
Schütz for helpful comments on an earlier version of the
Allman, J. M., Miezin, F., & McGuinness, E. (1985). Direction- and velocity-specific
responses from beyond the classical receptive field in the middle temporal
visual area (MT). Perception, 14, 105–126.
Anstis, S., & Casco, C. (2006). Induced movement: The flying bluebottle illusion.
Journal of Vision, 6, 1087–1092.
Aubert, H. (1887). Die Bewegungsempfindung (The sense of motion). Pflügers Archiv,
Bahill, A. T., & McDonald, J. D. (1983). Smooth pursuit eye movements in response to
predictable target motions. Vision Research, 23, 1573–1583.
Ball, K., & Sekuler, R. (1982). A specific and enduring improvement in visual motion
discrimination. Science, 218, 697–698.
Barnes, G. R. (2008). Cognitive processes involved in smooth pursuit eye
movements. Brain & Cognition, 68, 309–326.
Bedell, H. E., & Lott, L. A. (1996). Suppression of motion-produced smear during
smooth pursuit eye movements. Current Biology, 6, 1032–1034.
Benguigui, N., & Bennett, S. J. (2010). Ocular pursuit and the estimation of time-to-
contact with accelerating objects in prediction motion are controlled
independently based on first-order estimates. Experimental Brain Research,
Berryhill, M. E., Chiu, T., & Hughes, H. C. (2006). Smooth pursuit of nonvisual motion.
Journal of Neurophysiology, 96, 461–465.
Beutter, B. R., & Stone, L. S. (2000). Motion coherence affects human perception and
pursuit similarly. Visual Neuroscience, 17, 139–153.
Biber, U., & Ilg, U. J. (2008). Initiation of smooth pursuit eye movements by real and
illusory contours. Vision Research, 48, 1002–1013.
Billino, J., Braun, D. I., Böhm, K.-D., Bremmer, F., & Gegenfurtner, K. R. (2009).
Cortical networks for motion processing: Effects of focal brain lesions on
perception of different motion types. Neuropsychologia, 47, 2133–2144.
Bogadhi, A. R., Montagnini, A., Mamassian, P., Perrinet, L. U., & Masson, G. S. (2010).
Pursuing motion illusions: A realistic oculomotor framework for Bayesian
inference. Vision Research.
Born, R. T., Groh, J. T., Zhao, R., & Lukasewysc, S. J. (2000). Segregation of object and
background motion in visual area MT: Effects of microstimulation on eye
movements. Neuron, 2, 725–734.
Born, R., Pack, C., Ponce, C., & Yi, S. (2006). Temporal evolution of 2-dimensional
direction signals used to guide eye movements. Journal of Neurophysiology, 95,
Boström, K. J., & Warzecha, A. K. (2010). Open-loop speed discrimination
performance of ocular following response and perception. Vision Research, 50,
Boucher, L., Lee, A., Cohen, Y. E., & Hughes, H. C. (2004). Ocular tracking as a measure
of auditory motion perception. Journal of Physiology (Paris), 98, 235–248.
Bradley, D. C., & Goyal, M. S. (2008). Velocity computation in the primate visual
system. Nature Reviews Neuroscience, 9, 686–695.
Braun, D. I., Mennie, N., Rasche, C., Schütz, A. C., Hawken, M. J., & Gegenfurtner, K. R.
(2008). Smooth pursuit eye movements to isoluminant targets. Journal of
Neurophysiology, 100, 1287–1300.
Braun, D. I., Pracejus, L., & Gegenfurtner, K. R. (2006). Motion aftereffect elicits
smooth pursuit eye movements. Journal of Vision, 6, 671–684.
Bremmer, F., Distler, C., & Hoffmann, K. P. (1997). Eye position effects in monkey
cortex. II. Pursuit- and fixation-related activity in posterior parietal areas LIP
and 7A. Journal of Neurophysiology, 77, 962–977.
Bridgeman, B., Hendry, D., & Stark, L. (1975). Failure to detect displacement of visual
world during saccadic eye movements. Vision Research, 15, 719–722.
Britten, K. H., Shadlen, M. N., Newsome, W. T., & Movshon, J. A. (1992). The analysis
of visual motion: A comparison of neuronal and psychophysical performance.
Journal of Neuroscience, 12, 4745–4765.
Burr, D. C., Holt, J., Johnstone, J. R., & Ross, J. (1982). Selective depression of motion
selectivity during saccades. Journal of Physiology (London), 333, 1–15.
Carl, J. R., & Gellman, R. S. (1987). Human smooth pursuit: Stimulus-dependent
responses. Journal of Neurophysiology, 57, 1446–1463.
Carpenter, R. H. S. (1988). Movements of the eyes. London: Pion.
Castet, E., Lorenceau, J., Shiffrar, M., & Bonnet, C. (1993). Perceived speed of moving
lines depends on orientation, length, speed and luminance. Vision Research, 33,
Castet, E., & Masson, G. S. (2000). Motion perception during saccadic eye
movements. Nature Neuroscience, 3, 177–183.
Cavanagh, P., & Alvarez, G. A. (2005). Tracking multiple targets with multifocal
attentino. Trends in Cognitive Science, 9, 349–354.
Cavanagh, P., Tyler, C. W., & Favreau, O. E. (1984). Perceived velocità of moving
chromatic gratings. Journal of the Optical Society of America A, 1, 893–899.
Churchland, A. K., Gardner, J. L., Chou, I., Priebe, N., & Lisberger, S. G. (2003).
Directional anisotropies reveal a functional segregation of visual motion
processing for perception and action. Neuron, 37, 1001–1011.
Culham, J., He, S., Dukelow, S., & Verstraten, F. A. (2001). Visual motion and the
human brain: What has neuroimaging told us? Acta Psychologica, 107, 69–94.
De Brouwer, S., Yuksel, D., Blohm, G., Missal, M., & Lefèvre, P. (2002). What triggers
catch-up saccades during visual tracking? Journal of Neurophysiology, 87,
Debono, K., Schütz, A. C., Spering, M., & Gegenfurtner, K. R. (2010). Receptive fields
for smooth pursuit eye movements and motion perception. Vision Research.
Dobkins, K. R., & Albright, T. D. (2003). Merging processing streams: Color cues for
motion detection and interpretation. In L. Chalupa & J. S. Werner (Eds.), The
visual neurosciences (pp. 1217–1228). Cambridge, MA: MIT Press.
Dobkins, K. R., & Sampath, V. (2008). The use of chromatic information for motion
measures. Perception, 37, 993–1009.
Duncker, K. (1929). Über induzierte Bewegung: Ein Beitrag zur Theorie optisch
wahrgenommener Bewegung [On induced motion: A contribution to the theory
of optically perceived motion]. Psychologische Forschung, 12, 180–259.
Dürsteler, M. R., & Wurtz, R. H. (1988). Pursuit and optokinetic deficits following
chemical lesions of cortical areas MT and MST. Journal of Neurophysiology, 60,
Ferrera, V. P.(2000). Task-dependent
transformation for smooth pursuit eye movements. Journal of Neurophysiology,
Ferrera, V. P., & Lisberger, S. J. (1995). Attention and target selection for smooth
pursuit eye movements. Journal of Neuroscience, 15, 7472–7484.
Ferrera, V. P., & Lisberger, S. G. (1997). Neuronal responses in visual areas MT and
MST during smooth pursuit target selection. Journal of Neurophysiology, 78,
Festinger, L., Sedgwick, H. A., & Holtzman, J. D. (1976). Visual perception during
smooth pursuit eye movements. Vision Research, 16, 1377–1386.
Filehne, W. U. (1922). Über das optische Wahrnehmen von Bewegungen [On the
visual perception of movement]. Zeitschrift für Sinnesphysiology, 53, 134–145.
modulation of thesensorimotor
M. Spering, A. Montagnini/Vision Research 51 (2011) 836–852
Freeman, T. C. A., & Banks, M. S. (1998). Perceived head-centric speed is affected by
both extra-retinal and retinal errors. Vision Research, 38, 941–945.
Freeman, T. C. A., Champion, R. A., Sumnall, J. H., & Snowden, R. J. (2009). Do we have
direct access to retinal image motion during smooth pursuit eye movements?
Journal of Vision, 9, 1–11.
Freeman, T. C. A., Champion, R. A., & Warren, P. A. (2010). A Bayesian model of
perceived head-centered velocity during smooth pursuit eye movement.
Current Biology, 20, 757–762.
Frost, B. J., & Nakayama, K. (1983). Single visual neurons code opposite motion
independent of direction. Science, 220, 744–745.
Fukushima, K. (2003). Frontal cortical control of smooth-pursuit. Current Opinion in
Neurobiology, 13, 647–654.
Furmanski, C. S., & Engel, S. A. (2000). An oblique effect in human primary visual
cortex. Nature Neuroscience, 3, 535–536.
Garbutt, S., & Lisberger, S. G. (2006). Directional cuing of target choice in human
smooth pursuit eye movements. Journal of Neuroscience, 26, 12479–12486.
Gardner, J. L., Tokiyama, S. N., & Lisberger, S. G. (2004). A population decoding
framework for motion aftereffects on smooth pursuit eye movements. Journal of
Neuroscience, 24, 9035–9048.
Gauthier, G. M., & Hofferer, J. M. (1976). Eye movements in response to real and
apparent motions of acoustic targets. Perceptual & Motor Skills, 42, 963–971.
Gegenfurtner, K. R., & Hawken, M. J. (1995). Temporal and chromatic properties of
motions mechanisms. Vision Research, 35, 1547–1563.
Gegenfurtner, K. R., & Hawken, M. J. (1996). Interaction of motion and color in the
visual pathways. Trends in Neurosciences, 19, 394–401.
Gegenfurtner, K. R., Kiper, D. C., Beusmans, J. M., Carandini, M., Zaidi, Q., & Movshon,
J. A. (1994). Chromatic properties of neurons in macaque MT. Visual
Neurosciences, 11, 455–466.
Gegenfurtner, K. R., Xing, D., Scott, B. H., & Hawken, M. J. (2003). A comparison of
pursuit eye movement and perceptual performance in speed discrimination.
Journal of Vision, 3, 865–876.
Gellman, R. S., Carl, J. R., & Miles, F. A. (1990). Short-latency ocular following
responses in man. Visual Neuroscience, 5, 107–122.
Goodale, M. A., & Milner, A. D. (1992). Separate visual pathways for perception and
action. Trends in Neurosciences, 15, 97–112.
Groh, J. M., Born, R. T., & Newsome, W. T. (1997). How is a sensory map read out?
Effects of microstimulation in visual area MT on saccades and smooth pursuit
eye movements. Journal of Neuroscience, 17, 4312–4330.
Grossman, E. D., & Blake, R. (2002). Brain areas active during visual perception of
biological motion. Neuron, 35, 1167–1175.
Haarmeier, P., & Thier, P. (1996). Modification of the Filehne illusion by conditioning
visual stimuli. Vision Research, 36, 741–750.
Haarmeier, P., & Thier, P. (1998). An electrophysiological correlate of visual motion
awareness in man. Journal of Cognitive Neuroscience, 10, 464–471.
Haarmeier, P., Thier, P., Repnow, M., & Petersen, D. (1997). False perception of
motion in a patient who cannot compensate for eye movements. Nature, 389,
Hafed, Z. M., & Krauzlis, R. J. (2006). Ongoing eye movements constrain visual
perception. Nature Neuroscience, 9, 1449–1457.
Hafed, Z. M., & Krauzlis, R. J. (2010). Interactions between perception and smooth
pursuit eye movements. In U. J. Ilg & G. S. Masson (Eds.), Dynamics of visual
motion processing:Neuronal, behavioral,
(pp. 189–211). New York: Springer.
Hayashi, R., Sugita, Y., Nishida, S., & Kawano, K. (2010). How motion signals are
integrated across frequencies: Study on motion perception and ocular
following responses using multiple-slit stimuli. Journal of Neurophysiology,
Heinen, S. J., & Watamaniuk, S. J. N. (1998). Spatial integration in human smooth
pursuit. Vision Research, 38, 3785–3794.
Huk, A. C., & Heeger, D. J. (2000). Task-related modulation of visual cortex. Journal of
Neurophysiology, 83, 3525–3536.
Ilg, U. J. (1997). Slow eye movements. Progress in Neurobiology, 53, 293–329.
Ilg, U. J. (2002). Smooth pursuit eye movements: From low-level to high-level
vision. Progress in Brain Research, 140, 279–298.
Ilg, U. J., & Churan, J. (2004). Motion perception without explicit activity in areas MT
and MST. Journal of Neurophysiology, 92, 1512–1523.
Ilg, U. J., & Thier, P. (1999). Eye movements of rhesus monkeys directed towards
imaginary targets. Vision Research, 39, 2143–2150.
Ilg, U. J., & Thier, P. (2003). Visual tracking neurons in primate area MST are
activated by smooth-pursuit eye movements of an ‘‘imaginary’’ target. Journal of
Neurophysiology, 90, 1489–1502.
Johansson, G. (1973). Visual perception of biological motion and a model for its
analysis. Perceptions & Psychophysics, 14, 201–211.
Keller, E. L., & Heinen, S. J. (1991). Generation of smooth-pursuit eye movements:
Neuronal mechanisms and pathways. Neuroscience Research, 11, 79–107.
Kodaka, Y., Miura, K., Suehiro, K., Takemura, A., & Kawano, K. (2004). Ocular tracking
ofmovingtargets: Effects of perturbing
Neurophysiology, 91, 2474–2483.
Kohn, A., & Movshon, J. A. (2004). Adaptation changes the direction tuning of
macaque MT neurons. Nature Neuroscience, 7, 764–772.
Komatsu, H., & Wurtz, R. H. (1988). Relation of cortical areas MT and MST to pursuit
eye movements. I. Localization and visual properties of neurons. Journal of
Neurophysiology, 60, 580–603.
Komatsu, H., & Wurtz, R. H. (1989). Modulation of pursuit eye movements
by stimulation of cortical areas MT and MST. Journal of Neurophysiology, 62,
Konen, C. S., Kleiser, R., Seitz, R. J., & Bremmer, F. (2005). An fMRI study of
optokinetic nystagmus and smooth pursuit eye movements in humans.
Experimental Brain Research, 135, 203–216.
Kowler, E., & McKee, S. P. (1987). Sensitivity of smooth eye movement to small
differences in target velocity. Vision Research, 27, 993–1015.
Kowler, E., & Steinman, R. M. (1979a). The effect of expectations on slow oculomotor
control. I. Periodic target steps. Vision Research, 19, 619–632.
Kowler, E., & Steinman, R. M. (1979b). The effect of expectations on slow
oculomotor control. II. Single target displacements. Vision Research, 19,
Kowler, E., & Steinman, R. M. (1981). The effect of expectations on slow oculomotor
control – III. Guessing unpredictable target displacements. Vision Research, 21,
Krauzlis, R. J. (2004). Recasting the smooth pursuit eye movement system. Journal of
Neurophysiology, 91, 591–603.
Krauzlis, R. J. (2005). The control of voluntary eye movements: New perspectives.
Neuroscientist, 91, 124–137.
Krauzlis, R. J., & Adler, S. A. (2001). Effects of directional expectations on motion
perception and pursuit eye movements. Visual Neuroscience, 18, 365–376.
Krauzlis, R. J., & Hafed, Z. M. (2007). Finding our way around the sensory–motor
corner. Neuron, 54, 852–854.
Krauzlis, R. J., & Stone, L. S. (1999). Tracking with the mind’s eye. Trends in
Neurosciences, 22, 544–550.
Krauzlis, R. J., Zivotofsky, A. Z., & Miles, F. A. (1999). Target selection for pursuit and
saccadic eye movements in humans. Journal of Cognitive Neuroscience, 11,
Krekelberg, B. (2010). Saccadic suppression. Current Biology, 20, 228–229.
Krukowski, A. E., Pirog, K. A., Beutter, B. B., Brooks, K. R., & Stone, L. S. (2003). Human
discrimination of visual direction of motion with and without smooth pursuit
eye movements. Journal of Vision, 3, 831–840.
Krukowski, A. E., & Stone, L. S. (2005). Expansion and direction space around the
cardinal axes revealed by smooth pursuit eye movements. Neuron, 45, 315–323.
Lamontagne, C., Gosselin, F., & Pivik, T. (2002). Sigma smooth pursuit eye tracking:
Constant k values revisited. Experimental Brain Research, 143, 130–132.
Land, M. F. (2006). Eye movements and the control of actions in everyday life.
Progress in Retinal and Eye Research, 25, 296–324.
Leigh, R. J., & Zee, D. S. (2006). The neurology of eye movements (4th ed). Oxford:
Oxford University Press.
Lindner, A., Schwarz, U., & Ilg, U. J. (2001). Cancellation of self-induced retinal image
motion during smooth pursuit eye movements. Vision Research, 41, 1685–1694.
Lisberger, S. G. (2010). Visual guidance of smooth-pursuit eye movements:
Sensation, action, and what happens in between. Neuron, 66, 477–491.
Lisberger, S. G., & Ferrera, V. P. (1997). Vector averaging for smooth pursuit eye
movements initiated by two moving targets in monkeys. Journal of Neuroscience,
Lisberger, S. G., Morris, E. J., & Tychsen, L. (1987). Visual motion processing and
sensory–motor integration for smooth pursuit eye movements. Annual Review
of Neuroscience, 10, 97–129.
Lisberger, S. G., & Movshon, J. A. (1999). Visual motion analysis for pursuit eye
movements in area MT of macaque monkeys. Journal of Neuroscience, 19,
Lisberger, S. G., & Westbrook, L. E. (1985). Properties of visual inputs that initiate
horizontal smooth pursuit eye movements in monkeys. Journal of Neuroscience,
Liston, D., & Krauzlis, R. J. (2005). Shared decision signal explains performance and
timing of pursuit and saccadic eye movements. Journal of Vision, 5, 678–689.
Lorenceau, J., & Shiffrar, M. (1992). The influence of terminators on motion
integration across space. Vision Research, 32, 263–273.
Lorenceau, J., Shiffrar, M., Wells, N., & Castet, E. (1993). Different motion sensitive
units are involved in recovering the direction of moving lines. Vision Research,
Lu, Z. L., Lesmes, L. A., & Sperling, G. (1999). Perceptual motion standstill in rapidly
moving chromatic displays. Proceedings of the Academy of Sciences USA, 96,
Mack, A., & Herman, E. (1973). Position constancy during pursuit eye movements:
An investigation of the Filehne illusion. Quarterly Journal of Experimental
Psychology, 25, 71–84.
Madelain, L., & Krauzlis, R. J. (2003). Pursuit of the ineffable: Perceptual and motor
reversals during the tracking of apparent motion. Journal of Vision, 3, 642–653.
Marcar, V. L., Zihl, J., & Cowey, A. (1997). Comparing the visual deficits of a motion
blind patient with the visual deficits of monkeys with area MT removed.
Neuropsychologia, 35, 1459–1465.
Masson, G. S., Montagnini, A., & Ilg, U. J. (2010). When the brain meets the eye:
Tracking object motion. In U. J. Ilg & G. S. Masson (Eds.), Dynamics of visual
motion processing:Neuronal, behavioral,
(pp. 189–211). New York: Springer.
Masson, G. S., Proteau, L., & Mestre, D. R. (1995). Effects of stationary and moving
textured backgrounds on the visuo-oculo-manual tracking in humans. Vision
Research, 35, 837–852.
Masson, G. S., & Stone, L. S. (2002). From following edges to pursuing objects. Journal
of Neurophysiology, 88, 2869–2873.
Medina, J. F., & Lisberger, S. G. (2007). Variation, signal, and noise in cerebellar
sensory–motor processing for smooth-pursuit eye movements. Journal of
Neuroscience, 27, 6832–6842.
Merrison, A. F. A., & Carpenter, R. H. S. (1991). ‘‘Express’’ smooth pursuit. Vision
Research, 35, 1459–1462.
M. Spering, A. Montagnini/Vision Research 51 (2011) 836–852