Saccadic Eye Movements: Basic Neural Processes
Laurent Goffart
To cite this version:
Laurent Goffart. Saccadic Eye Movements: Basic Neural Processes. Reference Module in Neuroscience
and Biobehavioral Psychology, , 2017, Sensorimotor systems, �10.1016/B978-0-12-809324-5.02576-1�.
�hal-01427818�
HAL Id: hal-01427818
https://hal.science/hal-01427818
Submitted on 5 Mar 2022
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diffusion de documents
scientifiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
In: Reference Module in Neuroscience and Biobehavioral Psychology
https://doi.org/10.1016/B978-0-12-809324-5.02576-1
Saccadic Eye Movements:
Basic Neural Processes
Laurent Goffart
CNRS, Marseille, France
Contact information:
Laurent Goffart, PhD
Institut de Neurosciences de la Timone,
UMR 7289 Centre National de la Recherche Scientifique Aix-Marseille Université,
Marseille,
France.
e-mail: laurent.goffart@univ-amu.fr
KEYWORDS:
Adaptation, Cerebellum, Feedback, Fovea, Gaze, Goal-directed, Interception, Motor
control, Neural basis of behavior, Neurophysiology, Saccade, Space, Subcortical, Time,
Vision.
ABSTRACT:
The sudden appearance of an object in the visual field triggers an orienting shift of gaze
toward its location. This movement consists of an extremely rapid rotation of the two
eyes, the saccade. In this article, instead of describing the paths leading from the targetevoked retinal activity to the changes in muscle tension that rotate the eyes, we take the
reverse path. Starting from the muscle contractions, we proceed upstream and describe
the basic neural and functional mechanisms involved in generating commands that
permit the eyes and, therefore, the foveae to visually capture the location of an event.
The sudden appearance of an object in the visual field triggers an orienting shift
of gaze toward its location. This movement is an extremely rapid, step-like rotation of
the two eyes – the saccade – which can sometimes be accompanied by a head
movement. By positioning the object’s image on the retinal region of highest acuity – the
fovea – the saccade increases the number of neurons activated by the presence of the
object; a recruitment which likely contributes to improve its identification and
localization. But saccades are not exclusively aimed at objects that suddenly appear;
scanning saccades are also generated when one is exploring the environment. In this
case, saccade goals are extracted from the visual scene based on their saliency or their
relevance for subsequent actions (e.g., grasping an object, heading toward a location,
avoiding an obstacle, or reading the next word in a sentence).
Studying the neural control of saccades is a choice research theme for
investigating how the central nervous system controls goal-directed actions. Not only
because the saccade is an essential movement that vertebrates use to spatially
organize their sensory experience, but also because it is a relatively simple model of the
transformations which are applied to sensory signals in order to generate motor
commands for reaching a goal, whether it is stationary or moving. This research is
facilitated by the fact that the eye movements can nowadays be recorded with very
good spatial and temporal resolutions. Figure 1 illustrates the trajectory and the time
course of a saccade aimed at a small static target appearing in the peripheral visual
field. The trajectory is shown by plotting the vertical versus the horizontal angular
deviation (also commonly called position) of the eye (Figure 1(a)). The time course
shows that, after a latency (reaction time) of approximately 150 ms, the change in eye
orientation is extremely rapid (velocity profiles in Figure 1(c)) and ends with the
maintenance of the eye in a new orientation (Figure 1(b)). It also shows that a
secondary correction saccade sometimes follows the primary saccade.
<Figure 1 near here>
This seemingly very simple movement is actually the outcome of complex
muscular and neural activities with a functional organization which is not yet completely
uncovered. In this article, instead of describing a labyrinthine path from the targetevoked retinal signals to the changes in muscle tension that rotate the eyes, we take
the reverse path. Starting from the muscle contractions, we proceed upstream and
succinctly expose the neural mechanisms involved in generating the commands for
saccadic eye movements.
Extraocular Muscles and Rotation of the Eyes
Saccades are elicited by the cooperative action of six muscles attached to the
eyeball. These muscles are called extraocular muscles, to distinguish them from those
located inside the eyeball and which influence the diameter of the pupil or the lens
curvature. Two pairs of muscles are involved during horizontal rotations of the eyes
(i.e., rotations around a vertical axis through the eyeball): the lateral rectus (LR) and
medial rectus (MR) muscles. Orienting the right eye toward the right is accomplished by
the contraction of the LR and the relaxation of the MR (Figure 2(a)) while in the left eye
the MR is contracted and the LR relaxed. Conversely, shifting the gaze toward the left is
produced by the MR contraction and the LR relaxation in the right eye (Figure 2(b))
concomitantly with LR contraction and MR relaxation in the left eye. The muscles which
contract are called the agonist muscles; those which relax are called the antagonist
muscles.
<Figure 2 near here>
The muscle contraction generates a torque which is delivered to the eyeball
through tendinous attachments. Figure 2(b) shows the tension developed by the LR and
MR muscles during a horizontal saccade. The agonist muscle tension waveform first
shows a rapid rise in tension, which is followed by a quasi-exponential decay until a
steady-state tension is achieved and maintains the eye deviated in the orbit. For the
antagonist muscle, the tension waveform shows a symmetrical pattern, with a sudden
drop in tension followed by an exponential increase until the steady-state tension is also
reached. These muscle tension recordings reveal that, after every saccade, the
orientation of the eye is settled well before the agonist and antagonist muscle forces
reach their final equilibrium. They also illustrate the three components of the innervation
signal supplied to the extraocular muscles: a pulse, a slide, and a step component. The
pulse and the slide quickly rotate the eyes and stop them abruptly against the viscosity
of the muscles and orbital tissues, respectively. The step generates the muscle tension
required to hold the eyes deviated in the orbit, overcoming the elastic forces that tend to
drag the eyes toward their more central resting position.
Vertical movements of the eyes (rotations about a horizontal axis through the
eyeball) are slightly more complex than horizontal ones because they are the result of a
synergy between four other extraocular muscles. The joint contraction of the superior
rectus (SR) and inferior oblique (IO) muscles of both eyes, combined with the relaxation
of the inferior rectus (IR) and superior oblique (SO) muscles, rotates the eyes upward.
The bilateral contraction of the IR and SO, combined with the relaxation of SR and IO,
rotates them downward. Oblique saccades are generated by tensions simultaneously
developed by the muscles rotating the eye horizontally and those rotating the eye
vertically.
The Saccade Generators
The force generated by extraocular muscles is driven by the activity of motor
neurons (MNs) located in cranial nerve nuclei III (oculomotor), IV (trochlear), and VI
(abducens). The schema in Figure 3 shows the simplest possible view of the network of
motor and premotor neurons which are involved in generating a rightward horizontal
saccade (leftward saccades involve a similar neuronal network but located in
contralateral territories of the reticular formation). The recruitment of MNs driving the
agonist muscles is ensured by a population of excitatory burst neurons (EBNs) located
in the ipsilateral pontine reticular formation (path p1). These EBNs drive the activity of
abducens MNs innervating the LR of the ipsilateral eye. For contracting the MR in the
contralateral eye, the EBN command is relayed to MNs in the oculomotor nucleus by
internuclear neurons (INs), also located in the abducens nucleus. In parallel, EBNs
activate a population of inhibitory burst neurons (IBNs) located in the ipsilateral
medullary reticular formation and which inhibit the activity of EBNs and MNs in the
contralateral side (path p2). The burst of ipsilateral IBNs thereby participates in the
inhibition of MNs innervating the antagonist muscles (the LR of the left eye and MR of
the right eye). Thus, the antagonist muscles can relax while the agonist muscles
contract. Some neurons in the IBN area also burst during contralateral saccades.
Hence, if these neurons also inhibit contralateral MNs, the motor command sent to the
MNs innervating the agonist muscles must be viewed as the combination of two inputs:
an excitatory input from ipsilateral EBNs (path p1) and an inhibitory input from the
contralateral IBNs (path p3). This bilateral organization of the premotor drive during
horizontal saccades explains the bilateral organization of its control by the medioposterior cerebellum (see below).
<Figure 3 near here>
During upward and downward saccades, different neuronal populations are
involved. They are located in the mesencephalic reticular formation, more specifically in
the rostral interstitial nucleus of the medial longitudinal fasciculus (riMLF). One major
difference between the neural control of horizontal and vertical saccades is that the
former are mostly driven by a unilateral excitatory input (from the ipsilateral pontine
reticular region), whereas the latter are driven by a bilateral excitatory input from the left
and right riMLF.
The burst of excitatory and inhibitory burst neurons in the pontomedullary
(horizontal component) and mesencephalic (vertical component) reticular formations
seems to be gated by the activity of another class of inhibitory neurons located in the
nucleus raphe interpositus: the omnipause neurons (OPNs). These neurons exhibit a
sustained activity which is interrupted every time a saccade is generated, regardless of
the amplitude and direction of saccades. Released from this sustained inhibition, EBNs
and IBNs drive the motoneurons. The resulting saccadic eye movement has a direction
which is determined by the activation ratios between the horizontal and vertical saccade
generators. In healthy subjects, the examination of oblique saccades shows that both
the horizontal and vertical rotations are initiated simultaneously, not sequentially (see
Figure 1). Moreover, some observations show that the horizontal and vertical
components are coupled – when the velocity of one component is reduced, the other
component is also slowed down, thereby reducing the delay between the completion of
horizontal and vertical components. Despite this component-stretching mechanism, the
horizontal and vertical components may end at slightly different times. The analysis of
saccades generated by neurology patients is very instructive in this matter. For
example, in patients suffering from Niemann–Pick type C disease, the horizontal
component precedes the vertical component, resulting in the completion of the vertical
component well after the end of the horizontal component. Because of their inhibitory
influence on burst neurons, the OPNs are mainly considered as involved in visual
fixation (i.e., maintaining still the direction of gaze). Yet, they could also be involved in
coupling the horizontal and vertical saccade generators.
In Figure 3, it is the same network which is involved for a horizontal saccade
aimed at an object located 4° or 16° to the right. Additional mechanisms must be added
to this network to generate saccades of different amplitudes and durations.
The Control of Saccade Execution
The Duration and Velocity of Saccades
In almost all studies, the duration of a saccade is defined as the time elapsed
from when the eye velocity exceeds a certain threshold (e.g., 15° s −1) to when it returns
below it. The duration of saccades increases linearly with saccade amplitude (ranging
from 1° to 50°) at a rate of approximately 2–3 ms deg−1 in humans (1–2 ms deg−1 in
monkeys). For saccades less than approximately 30°, the velocity profile is bell-shaped
(see Figure 1(c)), with a rapid rise to a peak value (acceleration phase) followed by a
symmetrical decline in velocity (deceleration phase). For larger saccades, the velocity
profile becomes skewed, with the duration of deceleration phase being longer than the
acceleration phase. The peak velocity increases linearly with saccade amplitude until a
value of approximately 15°, beyond which the increase in velocity saturates to a
maximum value as large as 700° s−1 (900° s−1 in monkeys). These relationships
between the amplitude and the duration or peak velocity of saccades are called the
main sequences. Because their graphical shapes are relatively similar across healthy
subjects, the main sequences have become a tool for studying the oculomotor deficits
in brain-lesioned patients. Some authors have also suggested that common
physiological systems are responsible for the generation of reactive saccades and
voluntary saccades, as well as for the fast phases of optokinetic and rotatory nystagmi.
However, these duration–amplitude and peak velocity–amplitude relationships
show considerable variability between subjects and even within the same subject. They
are sensitive to alertness, motivation, and several other parameters such as saccade
direction, initial deviation of the eyes in the orbit, and the testing conditions (i.e.,
whether the saccade is memory-guided or immediate). Saccade velocity also depends
on the sensory nature of the target. Saccades toward somatosensory and auditory
targets are much slower than saccades of comparable amplitude and direction aimed at
visual targets. Saccades to a target moving away from the central visual field can also
be slower than saccades of matched amplitude toward a stationary target; some are
characterized by a lower peak velocity and a prolonged deceleration phase.
The Spatiotemporal Transformation
On the motor output side, the duration and velocity of saccades are defined by
the temporal properties of MN discharge. For saccades initiated from the straight ahead
position in the orbit, different saccade amplitudes do not involve different groups of
MNs; all the MNs are recruited. Saccades of increasing amplitude are caused by
increases in the duration and rate of MN discharge, the instantaneous firing rate
dictating instantaneous saccade velocity (rate code). On the other side, at the sensory
level, the code is topographic; targets at different locations in the visual field activate
different zones in the retina. A topographical organization is found in several regions in
the brain, including the intermediate and deep layers of the superior colliculi, the
neurons of which are located two to three synapses from the motor nuclei. From a
macroscopic point of view, the two deep superior colliculi (dSC) form a motor map,
which appears to be in agreement with the retinal map, with the anterior (rostral)
regions of both colliculi representing the foveal and perifoveal visual fields, posterior
(caudal) regions representing the peripheral visual field, medial regions representing
the upper visual field, and the lateral regions representing the lower visual field. In the
dSC, neurons emit a burst of action potentials shortly before a contralateral saccade
within their movement field (i.e., the range of saccade amplitudes and directions that is
preceded by a burst). In the rostral region, the burst occurs during small contralateral
saccades, whereas neurons in more caudal regions burst during saccades of larger
amplitude. Neurons located in the medial border burst prior to upward oblique
saccades, and those located on the lateral border burst before downward oblique
saccades. Interestingly, the electrical microstimulation of a particular site in the dSC
evokes a saccade with amplitude and direction corresponding to those for which the
stimulated neurons would maximally discharge. Thus, the amplitude and direction of
impending saccades seem to be determined by the locus of active neurons in the dSC.
However, this topographical code is not sufficient because a mismatch can be observed
between the amplitude and direction of saccades and those expected from the
topography of active neurons. Indeed, for the same target location, the amplitude and
direction are very different between saccades to a permanent target and saccades to a
memorized target location. The latter systematically end above the target location while
the locus of collicular activity has not changed. Therefore, the topographical code
seems to be completed by adjustments downstream, in the reticular formation.
Another fundamental problem is to understand how the locus of activity in the
dSC is converted into the MN temporal discharge. This transformation was initially
considered to be performed by a putative spatiotemporal translator (STT) located in the
cerebellum. A kind of look-up table mechanism associated a specific temporal pattern
of motor activity with each locus of collicular activity. But, this explanation is not
satisfactory because saccades generated toward the same target and from the same
starting position can be slower in velocity (longer in duration), yet still have the same
amplitude. The hypothesis was then proposed by David A. Robinson that the duration
of saccades is determined by a negative feedback loop. The saccade generators would
be driven by a motor error signal, a command which would result from identical
comparison between two signals: a reference signal specifying the goal or desired state
(e.g., desired eye orientation or desired change in eye orientation), and a feedback
signal estimating the current state (current eye orientation or actual change in eye
orientation). As long as there is a mismatch between these two internal signals, the
motor error would drive the saccade generators until the estimated current state
matches the desired one. In this framework, the motor error is a dynamic command that
changes with the direction of gaze. There is no need for an explicit STT; the duration of
saccades is merely determined by the time taken to zero-out the motor error command.
To date, the negative feedback loop remains the simplest hypothesis proposed to
account for the spatiotemporal transformation. Later, neuromimetic modeling taught us
that the neuronal solutions to the computations involved in saccade generation (e.g.,
spatiotemporal transformation and component stretching) may not be explicitly
performed by separate groups of neurons that neurophysiologists ought to identify.
Instead, these computations would be implemented implicitly, emerging from the
activities of populations of neurons that are massively interconnected and distributed
over several neural territories.
The Feedback Control of Saccades
One of the strengths of the negative feedback control is that it endows the
oculomotor system with a high level of flexibility for the execution of saccades. If for
some reason – pathological (e.g., spinocerebellar degeneration), experimental
(functional perturbation by brief electrical microstimulation or by systemic/local injection
of pharmacological agents), or natural (drop in vigilance, blink, presence of a distractor)
– the saccade is executed with a velocity which is much slower than normal or with an
altered initial trajectory, its accuracy is preserved as long as the feedback signal
conveys information about the changes in horizontal and vertical velocity.
Several sources can contribute to this feedback signal. The intrasaccadic visual
signals (optic flow and/or target-elicited retinal trace) could participate in estimating the
direction of the ongoing saccade. In patients affected by spinocerebellar degeneration,
saccades are so dramatically slowed that they can be redirected in flight toward a new
target presented during their course. However, in healthy subjects, most saccades are
too short in duration for visual signals to influence their trajectory. Inflow signals
originating from mechanoreceptors located in the orbital tissues or from extraocular
muscle proprioceptors could also be involved in the feedback. But their elimination by
sectioning of the ophthalmic branch of the trigeminal nerve does not affect the accuracy
and velocity of saccades, nor does it affect the ability to compensate for unexpected
changes in eye position prior to saccades toward flashed targets. The estimation of the
ongoing oculomotor state could also be based on copies of commands sent to the MNs
(efference copy). These outflow signals could originate from collaterals of burst neurons
which drive the MNs. It has indeed been shown that partial pharmacological inactivation
of the region involved in the generation of horizontal saccades barely affects the
accuracy of the horizontal component of oblique saccades, despite a dramatic
reduction in velocity. These observations, which are consistent with the feedback
control of saccade amplitude, also indicate that the number of active neurons
influences the velocity of saccades. Interestingly, the cerebellar influence on saccade
generation seems to involve recruiting neurons and controlling the size of the active
population in the pontomedullary reticular formation.
The hypothesis of saccade accuracy being controlled by a negative feedback
loop also predicts that a defect in estimating the current state should lead to dysmetric
saccades. For example, an underestimation of the displacement performed by the eyes
from saccade onset is expected to lead to hypermetric saccades (saccades larger in
amplitude than required) because the motor error is canceled later than usually (i.e.,
until the time when the defective estimated displacement matches the desired
displacement). Conversely, an overestimation of the actual displacement is expected to
lead to hypometric saccades (i.e., saccades that are smaller in amplitude than
required). From this perspective, the generation of accurate saccades requires an
adjustment of feedback signals such that they accurately represent the actual
oculomotor output. This adjustment is also required because the feedback neural
signals are not “commensurable” with the neural signals encoding the desired state. In
fact, the neural substrate of this feedback process still remains to be discovered.
Several lines of evidence suggest that the dSC participates in the processes
specifying the goal of saccades, upstream from the neural site where the motor error
command is elaborated. Concerning the feedback signals, the cerebellum is likely
involved in their adjustment. In particular, its medio-posterior region (composed of the
vermal lobules VIc–VII and caudal fastigial nuclei), receives target-related signals from
pontine nuclei, proprioceptive signals from extraocular muscles, and eye
position/displacement signals from the nucleus prepositus hypoglossi/medial vestibular
nucleus complex. In return, the output nuclei of the medial-posterior cerebellum, the
caudal fastigial nuclei, send distributed projections to the territories housing the
premotor neurons of the saccade generators. When the normal function of fastigial
oculomotor region (FOR) is impaired, the horizontal component of saccades becomes
severely dysmetric: it is hypometric for contralesional saccades, hypermetric for
ipsilesional saccades and an ipsipulsion affects the trajectory of vertical saccades. The
different deficits that affect ipsilesional and contralesional saccades indicate a bilateral
control which is consistent with the discharge and the connectivity of FOR neurons.
The medio-posterior cerebellum can be excluded from the node performing the
STT because its dysfunction does not break the spatial-to-temporal mapping: the
saccades become dysmetric but different amplitudes still have different durations. In
the reticular formation, the nucleus reticularis tegmenti pontis (NRTP) is a major target
of both the dSC and the FOR. Because pharmacological inactivation of NRTP
suppresses the generation of ipsilateral saccades, more studies are required to study
its role in the spatiotemporal transformation.
The Adaptive Control of Saccades
Saccade accuracy is maintained by adaptive mechanisms which modify the
visuomotor transformation to reduce the error of the following saccade. For example, a
mismatch between saccade amplitude and target eccentricity characterizes saccades
generated by newborns. When orienting toward a peripheral target, infants as young as
1–2 months of age frequently execute a series of consecutive saccades of
approximately equal size, whether accompanied by the head or not. However, although
it is hypometric, the amplitude of primary saccades increases with the eccentricity of the
target.
These adaptive mechanisms have also been studied in human subjects and in
monkeys with weakened muscles in one eye. When the subject monocularly views a
visual target, saccades are accurate if the normal eye is used but are too small in
amplitude when the weakened eye is the viewing eye. If the normal eye is patched and
the subject is forced to use the weakened eye to look at visual targets, the amplitude of
saccades gradually increases until they are nearly accurate. The fact that the changes
are observed in both the weakened eye and the normal patched eye (which now makes
hypermetric saccades) indicates that the adaptation involves a command which is
issued for conjugate movements of both eyes. This flexibility of the oculomotor system
to adjust saccade accuracy has also been studied in experiments in which the target is
displaced to a new location during the saccade. Initially, saccades land near the first
target location, but gradually they land closer to the second one (see paragraph “Natura
non facit saltus”).
Comparable adaptive processes must also take place during the generation of
saccades toward the changing location of a moving target. Interestingly, chemical
lesions of cortical area MT or of its pontine target, the dorsolateral pontine nucleus
(DLPN), two structures known to be involved in the processing of visual motion, impair
the accuracy of saccades to moving targets without altering that of saccades to
stationary targets. The DLPN being one of the major sources of motion-related visual
signals to the cerebellum, the adaptation therefore involves this network that connects
saccade-related regions of the brainstem and the cerebellum through pre-cerebellar
nuclei.
The Specification of the Desired State
The Varieties of Targets
Saccades are not exclusively aimed at visual targets. They can also be
generated in the dark, toward sounds, and to stimuli applied to the skin surface.
Localizing such stimuli is a much more complicated task. In the case of a
somatosensory stimulus, estimating its location requires the knowledge of the
stimulated body region (e.g., the index fingertip of the right hand) and of the posture of
the limb on which the stimulus is applied. For auditory stimuli, they are localized on the
basis of the interaural differences in intensity and timing of acoustic signals. The dSC
seems to be the interface through which these different sensory signals have access to
the saccade generators. Several saccade-related cells in the dSC are active prior to
saccades to visual, somatosensory, or auditory stimuli, suggesting that the converging
sensory signals are translated into oculocentric coordinates by the level of the dSC. But
the differences in velocity observed between these different groups of saccades
suggest that a signal specifying the desired displacement is not the unique output signal
from the dSC to the saccade generators. The differences could be due to different
levels of firing rate in the population of active neurons or to different population sizes.
Here again, the influence of the size of the active population is consistent with the
observations that regular visual saccades also have a reduced velocity and increased
duration (accuracy relatively spared) during pharmacological inactivation either of small
portions within the dSC or the source of one of its major cortical afferents, the frontal
eye fields.
“Natura non facit saltus”
The specification of the goal of saccades should not be viewed as a discrete
process but a continuous one. Several lines of evidence indeed show relationships
between the amplitude (or direction) and the reaction or response times of saccades,
unraveling a dynamical process. For example, a dependence of saccade endpoints on
the latency has been documented in studies testing saccades aimed at two stimuli that
are simultaneously presented at close locations in the visual field. Under these
conditions, the saccade endpoints are scattered along an axis joining the two target
locations, characterizing an averaging process. This averaging depends on the relative
properties of visual targets (size and energy) – the saccade landing position is closer to
the larger or the brighter target. Moreover, it mostly occurs for short latency saccades
and not when the latency exceeds 300 ms. The time window of the averaging process
can be delimited by presenting the two targets in a successive manner; short-latency
saccades end at the location of the first target, whereas long-latency saccades end at
the second target location. The amplitude transition function defines the averaging
window that accounts for endpoints that proportionately change with saccade latency.
Another instance of dynamic specification has recently been reported in a study which
tested saccades toward a transient moving target. For example, in response to a target
that uniformly accelerates, the landing position of saccades is more eccentric with
longer response time (i.e., the time interval from the target motion onset to the saccade
end).
Saccades toward the Changing Location of a Moving Target
The study of saccades toward a moving target nicely illustrates the fact that the
specification of the saccade goal is continuous process. In primates, the appearance of
an object moving in the visual field elicits a saccadic eye movement that brings the
target image onto the foveae. Such a spatiotemporal coincidence is quite remarkable
when one considers that the retinal signals are transmitted to the motor neurons
through multiple parallel channels connecting separate populations of neurons
(characterized by different integration times) with variable conduction speeds and
delays. Indeed, the flows of neural activity which are initiated by a moving target do not
remain local. They propagate both “vertically” (i.e., across the successive relays
interposed between the retina and the extra-ocular motoneurons) and “horizontally”
(within each relay) in the brain. Because of the delays, the foveal capture cannot be
driven by the current retinal signals as they correspond to past positions of the target.
Yet, when the trajectory of an interceptive saccade is experimentally perturbed,
a correction is made, sometimes after an intersaccadic pause (i.e., with presumably no
premotor activities). These corrections do not bring gaze toward a future location of the
target; they land next to the location where unperturbed saccades would have landed
at about the same time. These observations indicate that the saccade generators are
driven by a command estimating the current spatiotemporal coordinates of the target,
i.e., where it is here-and-now. This estimate is likely built upon antecedent visual
signals combined with adaptive signals memorized during previous experiences. When
the spatiotemporal coincidence is not reached, the saccade falls short of the target and
a correction (catch-up) saccade is made to bring the target image back onto the
foveae. Most observations show indeed that the interceptive saccades do not orient
the foveae toward any “future” location where the gaze would be maintained, waiting
for the target to enter within the foveal field and initiate its pursuit.
Yet, the idea has diffused that the target motion-related signals are used to
predict the “future target position” so as to assure a “spatial lead” of the gaze at the
saccade end, instead of attempting a precise capture of the target. Such an idea is
problematic for two fundamental reasons. Firstly, this view does not specify the
horizon of predictability, i.e., how far ahead the visuo-saccadic system is able to
“predict the future”. Secondly, it neglects the fact that the saccades do not orient a
point without thickness (as schematized in Figure 1 and Figure 2). They orient a
surface of photoreceptors which is rather extended. This extent is likely linked to the
fact that in their “Umwelt”, the animals do not track or chase after tiny objects; their
visual targets have extended sizes. In other words, considering the gaze direction as a
“line” is misleading because it leads us to consider as future a direction which is in fact
present.
The Control of Saccade Initiation
Saccades are triggered on the basis of information gathered during the time
interval that precedes saccade onset. This latency is far longer than would be expected
by adding the synaptic delays and nerve conduction times between the excitation of
sensors and the contraction of extraocular muscles. It reflects the time taken to extract
relevant signals from sensors, to make decisions about whether, and how, to respond
to the stimulus, and to start transmitting the initially-selected goal to the saccade-related
motor system. In response to subsequent incoming signals, a “decision” signal is
supposed to rise from an initial level until it reaches a threshold level, at which point the
saccade is initiated. Thus, a rise that varies randomly between trials would account for
the wide distribution of saccade latencies.
Several factors influence the latency of saccades. On the retinal side, latency
depends, for instance, on the eccentricity of the target, on its energy, and on its contrast
relative to the background. Saccade latency also depends on the presence and location
of distractors and cues, as well as on more cognitive factors such as the probability of
the target location, the presence of cues, the amount of reward associated with the
target, the target valence and salience, and visuospatial attention.
Under certain experimental conditions, the distribution of saccade latencies is
bimodal. The first mode, which can peak as low as 90 ms, designates the express
saccades, whereas the second mode corresponds to regular saccades. The
observation of express saccades is facilitated when a temporal gap separates the
fixation target offset and second target appearance. Express saccades are suppressed
when the deep superior colliculus is lesioned.
Conclusion
Saccadic eye movements are an important component of the animal’s behavior
for exploring and spatially organizing its sensory environment. They provide a useful
tool for understanding how the central nervous system relates the sensorium with the
motor repertoire, how it enables the entire animal to locate objects in its environment.
The aim of the neurophysiological approach, whether it is focused on processes
occurring at microscopic, mesoscopic or macroscopic scales, is to reveal the embedded
neuronal organizations which subtend the behavioral performance in a world of different
complexity. Further empirical research is required to determine how the basic neural
processes underlying orienting movements influence the generation of other types of
goal-directed movements and how they participate in the construction of spatial
memory. However, this bottom-up approach should be made carefully. It is indeed
possible that the brain, in its internal workings, does not support principles that are
similar to those used in Newtonian mechanics.
Further Readings
Barton, E.J., Nelson, J.S., Gandhi, N.J. and Sparks, D.L. (2003). Effects of partial
lidocaine inactivation of the paramedian pontine reticular formation on saccades of
macaques. Journal of Neurophysiology 90, 372-386.
Becker, W. (1989). Saccades In Carpenter R.H.S. (ed) Eye Movements: Vision and
Visual Dysfunction. pp 95-137 London: Macmillan.
Becker, W. and Jürgens, R. (1990). Human oblique saccades: Quantitative analysis of
the relation between horizontal and vertical components. Vision Research 30, 893920.
Cromer, J.A. and Waitzman D.M. (2007). Comparison of saccade-associated neuronal
activity in the primate central mesencephalic and paramedian pontine reticular
formations. Journal of Neurophysiology 98, 835-850.
Davis-López de Carrizosa, M.A., Morado-Díaz, C.J., Miller, J.M., de la Cruz, R.R. and
Pastor A.M. (2011). Dual encoding of muscle tension and eye position by
abducens motoneurons. Journal of Neuroscience 31, 2271-2279.
Fleuriet, J. and Goffart, L. (2012). Saccadic interception of a moving visual target after a
spatiotemporal perturbation. Journal of Neuroscience 32, 452–461.
Fleuriet, J., Hugues, S. and Goffart, L. (2011). Saccadic foveation of a moving visual
target in the rhesus monkey. Journal of Neurophysiology 105, 883–895.
Fuchs, A.F. Kaneko, C.R. and Scudder, C.A. (1985). Brainstem control of saccadic eye
movements. Annual Review of Neuroscience 8, 307-337.
Gandhi, N.J. and Bonnadonna, D.K. (2005). Temporal interactions of air-puff-evoked
blinks and saccadic eye movements: insights into motor preparation. Journal of
Neurophysiology 93, 1718-1729.
Gandhi, N.J. and Katnani, H.A. (2011). Motor functions of the superior colliculus. Annual
Review of Neuroscience 34, 205–231.
Goffart, L., Chen, L.L. and Sparks, D.L. (2004). Deficits in saccades and fixation during
muscimol inactivation of the caudal fastigial nucleus in the rhesus monkey. Journal
of Neurophysiology 92, 3351-3367.
Goffart, L., Hafed, Z.M. and Krauzlis, R.J. (2012). Visual fixation as equilibrium:
evidence from superior colliculus inactivation. Journal of Neuroscience 32, 1062710636.
Hafed, Z.M. and Krauzlis, R.J. (2012). Similarity of superior colliculus involvement in
microsaccade and saccade generation. Journal of Neurophysiology 107, 19041916.
Harris, C.M., Jacobs, M., Shawkat, F. and Taylor, D. (1993). The development of
saccadic accuracy in the first seven months. Clinical Vision Sciences 8, 85-96.
Hopp, J.J. and Fuchs, A.F. (2004). The characteristics and neuronal substrate of
saccadic eye movement plasticity. Progress in Neurobiology 72, 27-53.
Jürgens, R., Becker, W. and Kornhüber H. (1981). Natural and drug-induced variations
of velocity and duration of human saccadic eye movements: Evidence for a control
of the neural pulse generator by local feedback. Biological Cybernetics 39, 87-96.
Kaneko, C.R. and Fuchs, A.F. (2006). Effect of pharmacological inactivation of nucleus
reticularis tegmenti pontis on saccadic eye movements in the monkey. Journal of
Neurophysiology 95, 3698-3711.
Keller, E.L. (1974). Participation of medial pontine reticular formation in eye movement
generation in monkey. Journal of Neurophysiology 37, 316-332.
Kojima, Y., Iwamoto, Y., Robinson, F.R., Noto, C.T. and Yoshida, K. (2008). Premotor
inhibitory neurons carry signals related to saccade adaptation in the monkey.
Journal of Neurophysiology 99, 220-230.
Krauzlis, R.J., Lovejoy, L.P. and Zénon, A. (2013).Superior colliculus and visual spatial
attention. Annual Review of Neuroscience 36, 165-182.
Miller, J.M. and Robins, D. (1992). Extraocular muscle forces in alert monkey. Vision
Research 32, 1099-1113.
Moschovakis, A.K., Scudder, C.A. and Highstein, S.M. (1996). The microscopic
anatomy and physiology of the mammalian saccadic system Progress in
Neurobiology 50, 133-254.
Noto, C.T. and Gnadt, J.W. (2009). Saccade trajectories evoked by sequential and
colliding stimulation of the monkey superior colliculus. Brain Research 1295, 99118.
Nichols, M.J. and Sparks, D.L. (1995). Nonstationary properties of the saccadic system:
new constraints on models of saccadic control. Journal of Neurophysiology 73,
431- 435.
Optican, L.M. and Robinson, D.A. (1980). Cerebellar-dependent adaptive control of
primate saccadic system. Journal of Neurophysiology 44, 1058-1076.
Pellionisz, A. and Llinas R. (1982). Space-time representation in the brain: the
cerebellum as a predictive space-time metric tensor. Neuroscience 7, 2949-2970.
Quinet, J. and Goffart, L. (2015). Cerebellar control of saccade dynamics: contribution
of the fastigial oculomotor region. Journal of Neurophysiology 113, 3323–3336.
Quinet, J. and Goffart, L. (2015). Does the brain extrapolate the position of a transient
moving target? Journal of Neuroscience 35, 11780-11790.
Robinson, D.A. (1973). Models of the saccadic eye movement control system.
Biological Cybernetics 14, 71-83.
Robinson, D.A. (1975). Oculomotor control signals. In Lennerstrand, G. and Bach-YRita P (eds) Basic Mechanisms of Ocular Motility and Their Clinical Implications pp
337-378, Oxford: Pergamon Press.
Robinson, D.A. (1992). Implications of neural networks for how we think about brain
function. Behavioral and Brain Sciences 15, 644–655.
Schlag, J., Pouget, A., Sadeghpour, S. and Schlag-Rey, M. (1998). Interactions
between natural and electrically evoked saccades. III. Is the nonstationarity the
result of an integrator not instantaneously reset? Journal of Neurophysiology 79,
903-910.
Scudder, C.A., Kaneko, C.R.S. and Fuchs A.F. (2002). The brainstem burst generator
for saccadic eye movements: A modern synthesis. Experimental Brain Research
142, 439-462.
Sparks, D.L. and Mays, L.E. (1990). Signal transformations required for the generation
of saccadic eye movements. Annual Review of Neuroscience 13, 309-336.
Sparks, D.L. (2002). The brainstem control of saccadic eye movements Nature Reviews
Neuroscience 3, 952-964.
Taouali, W., Goffart, L., Alexandre, F. and Rougier, N.P. (2015). A parsimonious
computational model of visual target position encoding in the superior colliculus.
Biological Cybernetics 109, 549-559.
van Gisbergen, J.A.M., Robinson, D.A. and Gielen, S. (1981) A quantitative analysis of
generation of saccadic eye movements by burst neurons. Journal of
Neurophysiology 45, 417-442.
Change History
February 2016. L Goffart updated the text and further readings to this entire article.
Figure 1. Oblique saccade: (a) spatial trajectory; (b) time course of eye position; (c) time course of eye
velocity. The oblique saccade is initiated from a fixation stimulus located straight ahead (intersection
between the axes in (a)) and aimed at peripheral static target located 16° to the right and 8° upward (green
diamond). Positive values of horizontal (H) and vertical (V) positions correspond to rightward and upward
eye deviations in the orbit.
Figure 2. (a) Extraocular muscles of the right eye; (b) changes in LR and MR muscles tensions during a
16° horizontal saccade. Forces (black traces) are given as percentage of the maximum force ( Fmax). Upward
deflection corresponds to muscle contraction; downward deflection corresponds to muscle relaxation. H,
horizontal position; IO, inferior oblique; IR, inferior rectus; LR, lateral rectus; MR, medial rectus; SO,
superior oblique; SR, superior rectus; V, vertical position. (b) Reproduced from Miller JM and Robins D
(1992) Extraocular muscle forces in alert monkey. Vision Research 32: 1099–1113, with permission from
Elsevier and the authors.
Figure 3. Neuronal organization involved during a rightward saccade. Arrows indicate excitatory
connections; connecting lines ended by a circle indicate inhibitory connections. The connecting lines have
different colors for clarity. EBN, excitatory burst neuron; IBN, inhibitory burst neuron; IN, interneuron; LR,
lateral rectus; MN, motor neuron; MR, medial rectus; p1, path 1; p2, path 2; p3, path 3.