Key words: pursuits, velocity, latency, OKN, VOR, tracking
Outline
X Voluntary Tracking eye movements (Contí): Pursuits
Functions of Pursuits
Pursuits
are continuous slow conjugate eye movements that have velocities
under 100 deg/sec. They have several functions, the most
universal of which is to suppress or null the VOR and OKN when
tracking a moving target. Pursuits are most developed in
primates but there is evidence of them in any foveate animal that
uses head movements to track or follow objects of interest.
During head tracking of an object, fixation would be disrupted
by a counter-roll of the eyes stimulated by the vestibulo ocular
reflexes. Either the stabilization reflexes are shut off
or pursuits could overcome them and keep the eyes in their primary
position while following a moving target with the head.
Often these cancellation pursuits are more developed than pursuits
with a moving eye while the head is stationary. For example,
the cat can pursue a moving target with its eyes alone up to only
8 degrees/sec. However, the cat can follow a target with
its head and keep the eyes in primary position for velocities
up to 60 deg/sec. In man, the angular velocity of the head
can reach up to 200-300 deg/sec. We are able to compensate
for this with nulling pursuits or by switching off the VOR.
Pursuits suppress or null OKN while the eyes track a moving target. Tracking small moving targets with the fovea is not like OKN where eye movements null the optic flow. Pursuits of small targets generate more optic flow than they eliminate because of the retinal image motion produced by the stationary background. It is clear that the background requires some compensation because pursuits of a small target are 20% slower when the target is presented on a stationary textured background than when it is seen on a stationary untextured background.
Pursuits are also used to hold the eye steady during fixation.
Inspection of miniature eye movements during fixation reveals
many sources of ocular instability in the form of slow drifts.
In darkness this unstable drift velocity is about 1 deg/sec but
in the light it is only several minutes of arc/sec. As mentioned
earlier, pursuit compensation is referred to as slow control.
It compensates for imperfections in stability of the tonic integrators
in the brainstem that control innervation to tonic cells, and
for imbalanced activity of the vestibular canals. For example
an imbalance of only 2% between the horizontal canals can cause
a drift of 2 deg/sec. The fact that the eyes can fixate
without microsaccades attests to the presence and effectiveness
of this slow control system.
Pursuit Dynamics
Pursuits follow targets ranging in velocity from a
several minutes of arc/sec to over 100 deg/sec. The latency
to a sudden or unpredictable change in velocity is about 130 msec
ranging from 100 - 150 msec. This is about 50 msec shorter
than the latency for a saccade. About 50 msec of this is
lost to conduction in the afferent pathways and 25 msec for conduction
in the efferent pathways. The remaining 30-80 msec
is due to computation. Just for comparison remember the
VOR has a latency of only 6-14 msec.
The pursuit has two general phases referred to as open loop and closed loop. Open-loop means that the response is only influenced by the initial view of the target and not by changes in the retinal image produced by the eye movement. During the closed loop state, the eye is responding to alterations of the retinal image produced by the eye movement response. For example the target may be moving in space at 10 deg/sec and the eye may be moving at 9 degrees per second. The resulting retinal error or slip velocity is only 1 degree/sec. The response velocity keeps increasing to reduce the remaining error. This influence of eye motion on the stimulus is called negative feedback.
Fig 15.1 |
The open-loop response is divided into an early and late component.
The early component lasts for only 20 msec in which there is a
rapid acceleration of the eye that is independent of the stimulus
velocity and retinal image position. Typical values range
from 40-100 deg/sec/sec depending on the subject and means of
measurement. It is the same for responses to all pursuit
stimuli. Think of it as an initial shrug by the oculomotor
system to begin a movement. The late component lasts
80 msec during which the velocity of the pursuit is modified to
conform to the initial velocity of the stimulus as well as the
retinal image position or distance from the fovea.
Pursuit velocity increases to reduce fixation errors as well as
velocity errors. After this initial open loop response,
the pursuit response is a refinement of these initial estimates
based upon the remaining velocity errors between target motion
and eye motion. This is the closed loop phase of the response
that uses negative feedback. During this period the eye
corrects its velocity in response to velocity errors of the eye
as well as position errors and perhaps even acceleration errors.
The eye can speed up to reduce a position error or lag or increase
its velocity to match that of the stimulus. It can even
reverse direction from left to right if the stimulus reverses
direction. Sustained tracking of sinusoidal target
motion is typically used to examine the sustained feedback component
of the pursuit response.
Stimuli for Pursuits
Velocity: The open-loop components of
pursuit have been studied with a step-ramp stimulus that displaces
the target away from the fovea and then drifts it back toward
the fovea. This stimulus produces a conflict between the
position and velocity corrections needed to foveate the target.
Because the pursuit system has a shorter latency than the saccadic
system, the direction of the initial response tells us what the
stimulus is for pursuit and it also makes it easier to separate
out the dynamics of saccades and pursuits in the dynamic response.
An illustration shows that the initial response is a pursuit away
from the stimulus during which we can observe the open-loop components
of the response. Then there is a saccade in the reverse
direction to foveate the target and then the pursuit continues
in the original direction. This sequence demonstrates
that it is target velocity and not target position that stimulates
the pursuit when these two cues are in conflict.
Fig15.2 (Target position is indicated with dashed line.) |
Position: When the position and velocity cues are not in conflict, the pursuit system also responds to small position errors. This can be shown by introducing a small fixation error while tracking at the correct velocity and the eye will speed up to catch the target and then slow down again. It is also seen with an afterimage placed near the fovea. The eye attempts to catch the afterimage but can't. The resulting response is an accelerating smooth motion.
Target Motion: Pursuits also respond to perceived motion relative to the head. If eye velocity exactly matches target velocity, there is exact foveal placement of the target. Even thought there are no remaining retinal velocity or position errors, the eye continues accurate tracking. This is because the eye tracks the perceived motion of the target relative to the head. This percept is the result of combining eye position information with retinal image position information. Consider the example in which the eyes are fixating a stationary object and a bird flies across the horizon. At first the bird casts a moving image on the retina. We make a saccade to foveate the bird and begin to follow. If we follow too slowly, the image of the bird leads ahead and stimulates faster pursuit. If we follow too fast the reverse happens. Eventually we follow the bird accurately and its retinal image does not slip. Even while the image is perfectly stabilized on the fovea we continue to follow the bird. What information is used to continue following without retinal image motion? We still see the bird as moving because its motion results from the combination of eye movement efference copy and retinal image motion. This egocentric localization is derived in the parietal cortex and in MST which continue to stimulate following eye movements. This has been illustrated by having subjects view a row of flashing retinally stabilized lights. When the eye is still, the lights appear still. When the eye is moving, the lights appear to move with the eye. This apparent motion is the result of the efference copy signal of eye motion added to the stationary images of the lights on the retina. This perception of motion helps keep the eyes moving. The same thing can be demonstrated with a foveal afterimage.
Fig 15.3 |
Inferential motion: High level inferred motion can also stimulate pursuit eye movements. Steinback had an interesting demonstration whereby subjects viewed two lights mounted on the rim of an invisible wheel. When the wheel rotated, the subjects were asked to follow an imaginary hub of this rotating wheel. They could do this even thought the motion of the visible lights did not equal the velocity of the imagined wheel.
Proprioception: Although it is not a strong cue, subjects are able to smoothly track their moving finger in darkness. In a related example subjects can track a moving sound in darkness using auditory localization. However neither of these non-visual cues is as strong as any to the aforementioned visual cues.
The latency for pursuits that is due to neural transmission
time and computation during tracking and the use of negative feedback
could make it difficult to follow a target whose path of motion
changed. The pursuit response is able to compensate for
these delays by predicting the future course of motion of redundant
target motion. For example it is possible to pursue a repetitive
sinusoidal motion accurately with no phase lag after only one
cycle of the stimulus. It is also possible to continue tracking
for brief periods when a target has been blanked off briefly for
a fraction of a second. It is not clear where in the
brain prediction occurs, but the cerebellum is a likely candidate
because its general function is to predict the ending and time
course of most motor movements in order to avoid overshoots at
the end of movements. Indeed, the tracking response to a
step-ramp change in position stimulus does not overshoot the final
foveal target position. The pursuit system is able to anticipate
when the fovea will reach the desired image point and stop itself
at precisely the right moment without continuing the response
for one reaction time after reaching the target. As we will
see, slow vergence movements also show this same property.
Fig 15.4 |
Control of Pursuits: Parietal Cortex (MT,
MST)
Pursuit eye movements function to stabilize the image
of a moving object on the retina even when the background surrounding
the object is not moving. As mentioned earlier, this process
requires that we select a particular target from many that are
seen and that we suppress reflex stabilization response to the
motion of the background caused by pursuing the isolated moving
target. The motion signals that guide eye movements are
shaped first in the extra striate cortex in areas MT and MST that
lie in the superior temporal sulcus (STS). Cells in the
MST fire in concert with movement of the target with respect to
the head. This is different from the saccade stimulus in
the FEF and SC which is displacement of the target of interest
with respect to the fovea. Pursuits must track targets that
don't always move on the retina. This efference signal related
to eye velocity is also projected to the MST where target velocity
relative to the head is used as a signal to drive pursuits, even
during brief periods of darkness. The same visual
and efference copy information are combined in the cerebellar
flocculus to drive OKN. The FEF may be helpful in programming
predictive smooth pursuit eye movements, but no one knows for
sure.
Fig 15.5 |
Each hemisphere of the MST codes motion to the ipsilateral side. For example stimulation of the right MST causes a rightward following eye movement. These motion fields in a given hemisphere represent both halves of the visual field. This is the same organization as seen in the NOT for the control of OKN. Lesions in one hemisphere of the MST cause unidirectional deficit of horizontal smooth pursuit for targets moving toward the side of the MST lesion, irrespective of the visual hemifield into which the stimulus falls. Thus if a lesion occuring in the right hemisphere also causes hemianopsia, the eyes seem to look away from the hemiplegia.
Efference from MST projects to the NOT to generate OKN stimuli
and to the DLPN for pursuit tracking. The DLPN pathway for
pursuit proceeds to cerebellar flocculus and then vestibular nuclei
where the velocity signal is converted to a position signal for
the motoneurons that control rotation of the eye. Change
in eye position is coded relative to the current position of the
eye. Lesions in the occipital eye fields cause blindness
in striate cortex and motion blindness in extra striate cortex,
and or inability to make either OKN or pursuit tracking eye movements
to the same side as the lesion.
Fig 15.6 Note that motion blindness for targets in contralateral visual field affects pursuits but not saccades. |
Fig 15.7 |
Fig 15.8 (Means and standard deviations are shown. Target was an isolated spot with predictable ramp motion.) |
Fig 15.9 Note that in the elderly, saccadic frequency is increased, consistent with reduced pursuit gain. (More error-correcting or ìcatch-upî saccades are necessary to maintain foveation.) |
Review Questions: