[Previous Chapter] [Next Chapter]
 [Table of Contents] [Home] [Glossary] [Video] [Help]

Chapter 14

Key words: saccades, latency, main sequence, pulse, step, integration

X Voluntary Tracking eye movements: Visual Grasping and Gaze holding

Higher centers in the brain orchestrate fixation and tracking eye movements by activating the various reflex subroutines built into the brainstem.  The function of these higher centers is to transform afferent sensory signals from the retina into efferent motor commands.  (Afferent signals project towards a central site, and efferent signals project away from a central site, such as the brainstem).  The transformation of signals is a complex process because the afferent visual information contains information about the position and motion of all seen objects in the visual field.  It is a spatial map of depth, direction and motion of all objects we see in space.  This is a parallel process in that all perceptions occur at once.  In contrast the motor systems are serial.  That is they can only make a single response to one of the many stimulus conditions present in the complex visual scene.

Thus we need to select the stimulus to which we wish to respond, ignore or suppress responses to compelling gaze stabilization stimuli such as head motion and optic flow, and at the same time remain prepared to shift the motor response to unexpected urgent stimuli such as large looming objects.  All of this requires selective voluntary attention, the ability of fixation and tracking responses to override stabilization reflexes, and low level reflexes to call our attention to novel stimuli.  All of these activities are carried out at the same time by various high level centers in the brain.  As a result there appear to be multiple pathways controlling the same type of eye movement.

Some of these pathways are concerned with attention and voluntary control of eye movements such as reading and in visual search, and some with reflexive attention shifts to novel stimuli, and some with interactions between tracking and stabilization and fixation eye movements such as the fast phase of OKN and the VOR.

Fig 14.1 (left) Reading graph for a 50 minute selection.  The eyes move from left to right progressive saccades and then return sweep with a larger saccade for the next line.

Fig 14.2 (above) Record of eye movements during free examination of photograph with both eyes for 1 minute.  Notice saccades between fixation points.

The exact purpose of many of the pathways that control eye movement is still not completely understood.  In addition there are species differences, so that the pathways are different in cats, rabbits, monkeys, and human.  This chapter will present the pathways that are most understood from the occipital and frontal eye fields and the superior colliculus, but Leigh and Zee point out anatomical studies of many other regions of the brain that appear to be part of the oculomotor pathways.

This chapter reviews the function and pathways for tracking eye movements.  The functions of tracking eye movements, which include visual grasping and gaze holding, are accomplished by saccades and pursuits respectively.  These classes of eye movements are distinguished by their function, speed, control characteristics, and anatomical sites for their regulation.


Fig 14.3
Origin of the word "saccade". 

The horse's head (eyeball) is turned (saccade) by a jerk of the reins (extraocular muscles), one taught (agonist facilitation), the other relaxed (antagonist inhibition).
Final position is then maintained by tonic activity of the reins (muscles)

Saccades (Ballistic Control)
Saccades are named after the French word describing the rapid turning of a horse's head.  Saccades are very fast yoked eye movements that have a variety of functions.  All functions are related to the presence of our fovea.  Saccades produce the quick phase of vestibular and OKN to avoid turning the eyes to their mechanical limitations.  They reflexively shift gaze in response to novel stimuli that unexpectedly enter our visual field.  Saccades also occur with head movements.  Before we move our head we shift fixation with a saccade to a new direction and let the head and body catch up.  They are also under higher-level voluntary control to allow us to scan our environment in search of new information.


Fig 14.4
Gaze Saccade resulting from unexpected target movement.

Notice that the eye saccades first to shift fixation, and then the head "catches up".  VOR maintains gaze while the head moves.

Saccades get the eye to a direction in the orbit as quickly as possible with little regard for the conservation of energy.  It's a survival tactic.  Consider a baseball player who is stealing second.  He starts off with an incredible effort to speed up to the second base and then breaks as fast as possible to avoid being tagged out.  The effort is much greater than if he had simply walked or jogged over to second.  The saccadic system does the same thing.  It accelerates as quickly as possible to overcome the small mass and large viscosity of the eye and orbital structures, and then adopts a new level of innervation to maintain a new position which is restrained by elastic forces of the ocular muscles.  The rapid acceleration is accomplished by the frequency of the pulse, the amplitude by the duration of the pulse, and the new sustained eye position by the amplitude of the step innervation programmed by the PPRF.  In addition these actions require reciprocal inhibition of antagonists during the movement.  While the agonist constricts, the antagonist is totally inhibited, and at the end of the movement the antagonist  becomes active to help break the saccade when the antagonist innervation is reactivated to its new step level corresponding to the new eye position.


Fig 14.5
Electromyogram activity of agonist (LLR) and antagonist (LMR) during a saccade to the left.

Most of this innervational pattern is preprogrammed before the saccade actually occurs, much like a ballistic missile is preprogrammed.  After the launch, the missile follows its course determined by preprogrammed forces.  However, interactions with unforeseen forces, like shearing winds, can occur.  This is unlike the guided missiles that take in feedback during their flight to make corrections for unforeseen events or for a path which was too complex to be programmed prior to the launch.  Saccades have been referred to as ballistic because their force program of pulse and step appeared to be preprogrammed one latency period prior to the response.


Fig 14.6
Pause, burst, and integration circuit.  Resulting innervation of the oculomotor neuron is a "preprogrammed" pulse-and-step, which moves the eye and holds it in the new position.



Fig 14.7
Components of saccade generation for inferior rectus muscle during downward saccade (A) and upward saccade (B).

Note pulse and step components of motoneuron innervation.  (During upward saccade, IR is the antagonist and is thus inhibited.)

Pulse-Step mismatches
When an error is very small, the saccade ends with a gliding motion similar to the baseball player sliding into second base.  It can be remembered as the “saccadic waltz”(pulse-slide-step).  The slide is called a glissade which results from a mismatch between the sizes of the phasic (pulse) and tonic (step) innervation of the saccade.  The duration of the pulse (fast component of the saccade) is smaller than necessary, and the resulting saccade stops short of the target eye position.  The step innervation is too low in frequency to keep up the fast velocity but it does cause the eye to reach its intended direction.  Some eye anomalies are expressed with glissades.  It is important to distinguish between some of these measures of glissades and eye movement recording artifacts caused by lateral movement of the eye in the orbit (translation artifacts).
Main Sequence diagrams: Evaluation of the pulse component of saccades.
Saccades respond very quickly because of their burst or pulse innervation.  The average duration of a 2 degree saccade is 24 msec.  The largest commonly made saccades have 15 deg amplitudes and 50 msec durations.  Generally, the burst neuron firing frequency (spikes/sec) controls the saccadic velocity, and the burst neuron duration controls the amplitude.  Since the amplitude is proportional to frequency, the burster frequency also affects saccadic amplitude, and burster duration affects saccadic velocity.  Small 2 deg saccades have velocities of nearly 100 deg/sec whereas 15 deg saccades have velocities of nearly 1000 deg/sec.  The acceleration of the saccade follows similar rules because of the burst firing rate.

These observations are summed up in the main sequence diagram which is a continuous function or envelope that describes the duration and velocity of normal saccades based upon their amplitude.

Fig 14.8 and 14.9   The main sequence. 
A) Peak velocity (deg/sec) vs. magnitude of human saccade.
B) Duration (msec) vs. magnitude of human saccade.

In a main sequence diagram, saccadic duration and velocity are plotted as a function of saccadic amplitude.  The main sequence diagram was adopted by Bahill and Stark from the Hertzsprung-Russell diagram of star classification, which plots luminosity versus temperature.  Most stars lie along the diagonal of this plot.  Likewise, most saccades lie diagonally on the duration versus amplitude and velocity versus amplitude plots.  The main sequence diagram reflects the pulse component of the saccade, and it can be used as a diagnostic tool to evaluate the bursters of the midbrain.  If saccades are sluggish, their main sequence values will be low for the amplitude of the resulting saccade.
Latency and Visual Sampling
The normal latency or refractory period for the saccade to respond after a stimulus has been presented is 200 msec.  Questions that arise are 1) is this a fixed refractory period, and 2) can information be taken in to guide the saccade during its refractory period.  If the saccade is ballistic then the answer to the second question should be no.


Fig 14.10
Ballistic nature of saccades.

Notice that the intersaccadic interval remains constant despite the different durations of target jumps.

Westheimer showed that if you presented a target that jumps from the fovea to a new retinal location and promptly jumps back to it in less than 100msec, the eye responds with the expected latency of 200 msec after both displacements had occurred.  The saccade moved the eye away from the target on the fovea.  Then after 200 msec the eye looked back to the target.  The interval between saccades was independent of the interval between the target jumps away and back to the fovea.  This suggested that the saccadic system could only react to one target at a time and there was a fixed latency or refractory period, during which a second saccade could not be initiated after the first.  Finally it suggested that no new information could be taken in during the refractory period.

More recently however there have been experiments demonstrating that the eye does pay attention to the target during the first 110 msec of the latency period so that if multiple displacements in the target do occur they can affect the amplitude of the initial saccade response, and even cause additional short latency supplementary saccades to follow up in 20 msec rather than 200 msec.  Some studies have shown that you can even change the direction of a saccade during flight if a second stimulus is presented late in the refractory period. Therefore the saccade is not really ballistic because it takes in information during its latency period.  A likely alternative explanation is that two separate saccades were programmed in parallel and there times of execution overlapped one another.

There are other circumstances in which the typical saccade latency of 200 msec can be reduced by prediction.  Prediction is essential for almost all motor systems that control balance and locomotion.  For example when we run up a set of stairs we must predict when our leading foot will pass over a step to synchronize the foot landing on the step. Prediction also occurs when the motion of the eye is very repetitive and synchronized with motion of a highly repetitive target.  Suppose you are watching a ping pong or tennis match and you measure the latency of saccadic eye movements.  The eyes track the ball with saccades.  You fixate alternatively between the two players.  Because the timing is so consistent you are able to initiate the saccades within one tenth the normal latency or within 20 msec after the ball has been hit toward the other player.

It is possible that whole volleys of saccades are programmed in advance of moving to a new target site.  There is an initial hypometric saccade that has a normal 200 msec latency and subsequently there are corrective saccades with latencies as short as 20 msec.  This could be the nature of express saccades.  This prediction ability is likely to be a function performed by the cerebellum since it normally predicts the completion of most motor movements in order to coordinate walking running and speech.  Similarly, the cerebellum may be involved in some way with terminating the saccade at the right moment by sensing any errors in amplitude and activating the antagonist as a break in the case of potential overshoots.  To do this we need to predict when the saccade will end.  If we don't make this prediction, we end up with saccade dysmetria and over- and under-shoots.

Saccade Accuracy
Given the brevity of saccades, it is difficult to precisely break the movement, especially when it is large.  To compensate for this problem, the saccadic system behaves like a golfer who usually shoots short of the hole until he gets within close range.  This analogy with saccades is that large saccades usually fall short of their target and are followed up with a series of smaller corrective saccades.  This strategy reduces the uncertainty of the direction of the saccade error and only requires that we assess the amount of undershoot.  These corrective saccades have very short latencies of only 100-130 msec.

Corrective saccades
The saccadic system also has strategies when tracking a moving target.  If the target moves rapidly, the smooth following response can lag behind and produce a position error.  Catch-up saccades are used to refoveate the target during the pursuit.  The astonishing feature is that the catch-up or corrective saccade is initiated long before the positional error is fully manifest.  This indicates that the saccadic control system must estimate and predict from the velocity of the target and eye how large a correction is going to be required.  This is another predictive function that is likely to be controlled by the cerebellum.

Fig 14.11
Pursuit response to a "ramp" motion.  (eye movement in degrees on vertical axis, and time in sec plotted on horizontal axis.  Motion of target plotted with dashed lines.)

Notice the catch-up saccade used to refoveate the rapidly moving object.

Saccadic Suppression and Spatial direction constancy
Whenever we make a saccade, there is an abrupt shift of the entire retinal image.  Objects on the fovea move off of the fovea.  Interestingly, they appear to remain stable or in a fixed direction in space.  This is because our sense of direction relative to our head and body is updated by computing both the position of the eye from a corollary discharge or efference copy signal and position of the retinal image.

Corollary discharge signals are likely to come from the neural integrators of the brain stem.  This computation occurs in the parietal lobe and takes some time, maybe as long as 50 msec.  To avoid transient errors in perceived direction, the visual percept is suppressed momentarily during the saccade.  You can experience this by looking back and forth between the reflected images of your eyes in a mirror.  You never see your eyes move.  The suppression begins about 50 msec before you begin the movement and ends about 30 msec after the movement.  The suppression is not total.  It is due to three mechanisms.  The first mechanism is the mechanical optical smearing of the retinal image during the saccade.  The second process is a 20% reduction of visual sensitivity, and the third and strongest effect is visual backward masking in which the perception of a prior target is suppressed by a later one.  Its as though the stronger stimulus at the end of the saccade overtakes the prior weaker stimulus during the saccade and suppresses it within the visual pathways.

The same suppression effect is seen without a saccade simply by jerking or shifting the retinal image.  Problems in this suppression and the general integration of eye position and retinal image position result in apparent motion of the world during eye movements.  This condition is called oscillopsia and it can affect patients with recently acquired nystagmus.  However, after several weeks or months the oscillopsia usually disappears when the eye-brain learns to anticipate the rapid nystagmoid movements of the eyes.  The oscillopsia probably results from a mismatch in sensed retinal image motion and eye motion sensed via efference copy.  When the perceptual system recalibrates to match retinal and extra-retinal motion, oscillopsia disappears.

Cortical Control of Saccades
The frontal eye fields (FEF) correspond with Brodman's area 8, which is traditionally called the motor cortex.  This area mediates voluntary control of contralateral saccades.  It is active whether saccades occur or not.  The activity is related to visual attention and when saccades occur the related activity in the FEF can precede them by 50 msec.  The surface of the FEF has a vague retinotopic organization so that stimulation of a particular area causes a saccade to change eye position in a specific direction and amplitude.  For example, stimulation of a certain region will always cause an oblique saccade of 5 degrees from wherever the eye is currently pointing.  These stimulated points are called movement fields and they are analogous to receptive fields of sensory neurons in the visual cortex.  The movement field is that part of the visual field to which the will eye move in response to activity of that cell.  The field controls movement relative to the fovea.  Stimulation of a movement field in one hemisphere causes conjugate saccades to the contralateral side.  Vertical saccades require stimulation of both hemispheres of the FEF.  Modalities that can stimulate movement fields include vision, audition, and tactile.


Fig 14.12 (left)
Schematic of pathways involved in control of saccades.

Voluntary saccadic eye movements are mediated by parallel pathways from the frontal eye fields and superior colliculus that converge within the brain stem.

Frontal Eye Fields (FEF)

  1. Stimulation elicits a contralateral saccade
  2. The latency for stimulation to movement is 30 to 45 msec
  3. With stimulation, the size and direction depend on the point stimulated not the intensity or duration (movement fields)
  4. There are many neurons that discharge before saccades. There are 2 major cell types:
    1. movement cells, encoding amplitude and direction
    2. fixation cells, associated with maintaining or releasing fixation
  5. Lesions in FEF impair saccades and voluntary visual search

Supplementary Eye fields (SEF)

  1. Active during saccade tasks
  2. Stimulation elicits saccades
  3. Goal directed mapping, no matter where the eye starts
  4. Important in encoding motor behavior based on previous training.

Dorsolateral prefrontal cortex: Memory of target location

Posterior Parietal Cortex

Related to visual attention, not a motor map like FEF

7a: Encodes position of target with respect to head

Lateral intraparietal area (LIP): Holds a signal of desired position until saccade is made to the remembered location.

Internal Medullary Lamina (IML)

Goal directed Saccades of Fixed Vectors (amplitude and direction)


  1. Receives lots of projects from lots of places
  2. Assumed to be major center for feedback loops with all the connections and projects.
  3. Role not totally understood

Substantia Nigra Pars Reticulata

  1. Decreases activity when voluntary saccades occur
  2. The voluntary saccades can be either memory guided or visually guided
  3. Chief role is to control superior colliculus (somewhat like RIP influences PPRF and riMLF)

Superior Colliculus

  1. Has both sensory and motor maps (visual grasping center)
  2. Orderly Retinal projections on mapped on surface (Dorsal)
  3. Motor map is in the deeper layers (Ventral) and this map projects to PPRF and riMLF.

Fig 14.13 (right)
Physical locations of several cerebral regions.

Note location of frontal eye fields (FEF).


The FEF project two main efferent pathways for the control of saccades, one to the superior colliculus and the other to the PPRF and riMLF.  The fibers from the frontal eye fields descend from the cortical field to ipsilateral superior colliculus and, cross the midline, to the contralateral PPRF.  Gaze palsies or lesions of the PPRF cause drift of the eyes to the contralateral side and an inability to look to the same side.  The FEF specifies to the SC the amplitude and direction of an upcoming saccade.  It also participates in the process of maintaining and releasing fixation.   Neither the colliculus or FEF are exclusively required to generate saccades.  Either one can be ablated and saccades continue.  However if both are ablated, saccades are abolished.  Thus both regions receive some independent afference and the FEF has several pathways to the brainstem to generate saccades.    Lesions of the FEF may prevent patients from making voluntary saccades to remembered target locations without a strong visual stimulus to guide them.  This condition is known as pseudo-ophthalmoplegia.






Fig 14.14
Projection of efferent pathway from FEF to contralateral PPRF for left saccade.

Superior colliculus control of saccades
The principal pre-oculomotor target of the FEF is the superior colliculus (SC).  The purpose of the SC is to represent the retinal error the saccade is to correct.  It can be thought of as a visual grasping center whose function is related to sensory-motor coordination in reorienting the eye and head with saccadic motions that place the retinal image on the fovea.  In addition, the SC has another function which is to hold images on the fovea once they have been grasped.  As mentioned above, the SC does not require an intact FEF to generate saccades.  Neurologists have documented patients who have lost visual afference to the FEF who still can make saccades presumably based on the function of the colliculus.  Thus, these patients don't perceive a visual target but they are able to make saccades to fixate it.  This phenomenon is known as blind sight and is thought to be a vestige of a more primitive visual system where the colliculus served the function now assumed by the cortex.  In fact there is a whole literature on the "two visual systems".  The subcortical one is supposed to be the global system that responds to orientation of the body and location of targets in space.  The newer system is called the local system that deals with structure and form of objects in space.  These are referred to as the where and what systems.  The SC is the where system, and geniculo-striate pathways the what system.  It turns out this is an oversimplification, because what and where functions are both found in the cortex.

The colliculus has several layers.  The superficial layer contains a visual sensory map of the visual field that is derived from both direct retinal projections via the optic tract and cortical projections.  The intermediate and deep layers contains a motor map of the visual field.  It contains cells with movement fields that are much more organized than those of the FEF.  These cells respond to all sensory modalities including vision, audition, and tactile.  The direction or spatial location of all of these sensory stimuli are mapped in the colliculus relative to the fovea.  Like the FEF, stimulation of one side of the SC causes a conjugate saccade to the contralateral side and stimulation of both sides is necessary to evoke purely vertical saccades.  The motor map has a polar organization and is relative to the fovea so that it signals change of fixation from the current eye position and not to a particular orbital location.  The larger the retinal error to be corrected the more caudal is the point of activity.  Thus small saccades are stimulated rostrally and large saccades are stimulated caudally with the accompaniment of head movements.  Medial stimulation causes upward components and temporal stimulation causes downward components of saccades.  Pure vertical only occurs with bilateral stimulation.   A current theory being tested is that during a saccade, as the retinal error is reduced, the activity in the colliculus travels rostrally toward the fixation pole as a traveling wave.  The output of the SC projects to the longlead bursters of the PPRF and the riMLF for horizontal and vertical components of saccades respectively.

Fig 14.15
Motor map for the superior colliculus.

The map indicates sites which control saccades of specific amplitude and direction from the current point of fixation or direction of gaze. Thus the superior colliculus controls change of fixation from any starting direction of gaze.

Superior Colliculus: Fixation
One of the main tasks of the colliculus is to maintain or hold the retinal image at the fovea after grasping it with a saccade.  This is accomplished at the rostral or forward pole of the SC, which contains fixation cells that are tonic and activate pause cells in the brainstem to prevent any saccades from moving the eye away from its current position.  Lesions of the SC cause prolonged latencies for saccades as well as inaccuracies of saccade amplitude.  They also make it difficult for patients to break fixation away from one target and foveally grasp a new stimulus.  These cells act with the neural integrators of the brainstem where velocity signals are transformed to head and eye position signals in the form of tonic cell discharge.  The tonic cells hold fixation that is uninterrupted by saccades if the saccades are suppressed by pause cells driven by fixation neurons at the rostral pole of the colliculus.



Fig 14.16
Saccade abnormalities related to pulse and step.

(Innervation patterns are shown on the left, and resulting eye movements on the right.  Dashed lines indicate the normal response.)


Review Questions:

  1. Why are saccades described as ballistic?
  2. How do burst and tonic cells shape a saccade?
  3. What is the main sequence diagram?
  4. What is a pulse-step mismatch?
  5. What cortical regions initiate saccades?
  6. What is the role of the colliculus in controlling saccades?
  7. What is the function of the pause cell in the colliculus and brainstem?



[Previous Chapter] [Next Chapter]
 [Table of Contents] [Home] [Glossary] [Video] [Help]