Smooth Pursuit
What is smooth pursuit?
¶ Smooth pursuit (SP) means foveal tracking of a small moving target.
SP cares about position of the target in a way that optokinesis does not.
¶ SP is a conjunctive eye movement, but only one eye has to be looking
at target.
¶ Latency of SP is about 140 msec, less than saccade latency, but more
than VOR.
¶ You cannot generate SP without a small moving object to pursue! Asked
to attempt SP while looking at a blank wall, and the subject will make a series
of small saccades.
¶ Smooth pursuit doesn't appear in humans until about 6 weeks of age, about
the time when the first smile appears.
¶ The total tracking response can involve a saccade to bring the target
onto the fovea and a head movement to maintain contact with object over a wide
angle. Gaze is head plus eye movement. Tracking can involve using head
movement and body movement for body rotation.
SP vs optokinesis: concern for position matching
SP vs vergence: is there a smooth pursuit of objects approaching the viewer?
Is there such a thing as SP after-nystagmus? like OKAN?
SP is "evolving": RHSC2 p. 56: "Pursuit can occur in any meridian,
but is on the whole smoother and more precise in the horizontal direction. The
maximal velocities achieved depend on the species, and the kind of target used.
In man and monkeys, estimates range from some 80 to 160_/sec. In the cat, peak
velocities are much smaller, perhaps some 10_/sec at most."
The adequate stimulus for smooth pursuit
Rashbass & Robinson used the step-ramp stimulus and concluded that only
slip error, not position error, stimulates smooth pursuit. Motivation for SP
or fixation can suppress saccade system.
Does the slip have to be on the fovea?
Adequate stimulus: <30 deg/sec (maybe higher speed...see recent publications)
Response to acceleration is SP + saccades.
can be made to global pattern (center of wheel)
can it be made to sound movement in dark?
in dark: Subjects can make smooth pursuit to arm movements...
? size of stimulus...verge into optokinetic
A target does not have to lie on the fovea for smooth pursuit to occur...p 56
Rashbass step/ramp stimulus: What do we learn from it? that velocity error
is more important than position error for tracking...delay too long for saccade
to intervene?
Prediction in foveal tracking. Does it lead to momentum?
Tracking illusions in space or time! RHSC2 page 57.
Smooth pursuit to strobe illuminated motion?
Cinema...apparent motion "So long as the discrete jumps are less than 150
msec apart such a target is tracked smoothly."
Reading:
RHSC2, chpt 3, pp 55-68
Lisberger et al. Ann Rev Neurosci. 1987 pp. 97-129, "Visual
motion processing and sensory-motor integration for smooth pursuit eye movements."
Review of:
80 - 130 msec latency
visual acuity starts to decrease when slip exceeds 3 deg/sec
DILEMMA: Ordinary negative feedback might reduce error to zero, stop the motor
response and have a 100 msec delay starting up again...
see viewgraph #2(?), which, includes an eye velocity positive feedback command...if
error from eye is zero, system cruises on current output...
RPE, RVE and RAE were all found to have influence in open loop error testing...
looked for correlations with eye movement acceleration 90 msec later...
Viewgraphs worked well, esp Fig 1 from Lisberger, showing details of smooth
pursuit eye movements in response to sinusoidal target for monkey : presence
of saccades implies no prediction? tracking with tolerance of position error.
Miniature eye movements (chpt 6) Is drift like smooth pursuit?
Are other "slow" eye movements (VOR) equivalent to smooth pursuit?
RHSC2 has optokinesis and smooth pursuit in same chapter...
Is there a vergence-smooth pursuit? A vergence saccade?
Rolf Eckmiller, Physiol. Rev. 67: 797-857 (1987), "Neural control
of pursuit eye movements."
speed and direction
OKN and VN combined
compensate for VN adaptation?
motion toward and away from...
detection of acceleration
Cut-out demo of fig 3.17, showing a circle moving past a slit in aluminum foil.
what if only the top half of the circle can be see? Can you pursue the circle?
Leigh & Zee
page 72, "Attention, motivation and age influence how well subjects pursue."
page 75. head and eye movements in smooth pursuit. ?Interchangable.
PAGE 82: 5 abnormalities of smooth pursuit (open the book!)
weirdest: inverted following responses
paper Vis Res 22: 1193, "Visual following during stimulation of an immobile
eye (the open loop condition)" (1982).
See clinical material. 62 year old male who presented: "complete left ophthalmoplegia
with ptosis."
Great figure 2, showing gain of 50 at low speed presentation to paralyzed eye.
Actual eye movements are not all that smooth! see figure 3
Feedback model for smooth pursuit
If the visual system ("afferent arm") differentiates position for
speed, then the efferent arm must have an integrator to restore position control.
Smooth pursuit as a feedback system:
positive feedback inside of negative feedback. Say positive feedback has a
gain of +1.000. In fact, for smooth pursuit, it really does need a perfect gain
of 1.000! If slip goes to zero, but target is still moving, then the positive
feedback loop will maintain eye movement. Is there dynamics in the feedback
loop? Notice that the output to the OMN is given in velocity! How does the OMN
deal with that? we'll see later, at the single unit level.
What happens when the target stops moving? Is there SP after response? Depends
on the dynamics of the efferent arm. If the integrator is in the afferent arm,
then we might be OK.
Need to provide H(s) details for the feedback system with efference copy.
Integrating to eliminate tracking error
Consider a way to drive the slip response to zero
what about the differentiation of slip getting canceled by the integrator?
§ 1 / 4 Latency and apparent motion
Delay of 90 - 140 msec for smooth pursuit, vs 200-400 msec for saccades...
200 - 100 msec = 0.1 sec; 30 x 0.1 = 3 deg of tracking before saccade...;
DELAY how much is retinal, how much central? Maybe 80 + 50 msec...
1994: The delay for initiating smooth pursuit is about 100 msec, and so is the
time between strobe flashes for allowing smooth pursuit to "apparent motion."
RHSC2 says that if a pattern has spatial period p meters and the strobe flashes
occur a frequency f per second, then the apparent motion occurs at speed p·f
meters/second. Is that right? Is this good only for marquee lights?
discussion with Ellen Barton:
keeping delayed target on the fovea--open loop after movement starts, because
slip won't be detected until some delay.
stopping time after target stops? 200 msec?
How smooth pursuit is different from slow phase OKN-
OKAfterN duration is related to size of stimulus tested; continuous stimulus
needed for about 10 sec!
Afferent and efferent arms of smooth pursuit
EFFERENT ARM
gives velocity memory...is it like OKaN?
velocity errors cause eye acceleration
cerebellum flocculus-to-OMN integrator
stimulation of flocculus can cause a twitch of eye velocity
problem with positive feedback - cerebellum inhibits...
Lisberger: "We assume that increases in floccular output cause ipsilateral
eye motion by reducing the firing of brain stem cells that cause contralateral
eye motion."
AFFERENT ARM
V1, MT PPC, all projecting down to Pons...
Local lesions of MT cause transient defects in smooth pursuit; "With step
ramp stimulus, deficits were limited to the initiation of pursuit." --using
step/ramp stimulus
and cleared up in a couple of days...
See Newsome J. Neurosci. 5: 825 (March 1985) figures
talk at Brown on 4/18/91
VIEWGRAPH
Initiation=acceleration depends on velocity, position, direction of target.
there is a first 40 msec direction-dependent part
and a 100 msec ("late"), depends on position (near fovea, <5deg)
& speed (like MT...30-90 deg/sec)
Maintenance position, velocity and acceleration errors
correlation with 90 msec delayFrequency response of system: RHSC fig. 3.12
demo of Innisfree for sine wave gratings.
(Fastest and slowest perceptible motion, vs made movement)
Kaufman: 10-20 min arc / sec
a limiter for output
Open the loop
Lesion of area MT: effect on SP
what happens when area MT is lesioned? Assgn 12B
Why does smooth pursuit capability stop at 30_/sec or does it? L&Z say 100_/sec.
Could there be saturation on the "output" of the SP box?
.
RHSC2 says that slip helps select the model which tracks the target. How prediction
can be used in guiding movement.
----------------
Is there an arm movement like smooth pursuit? Playing the violin?
Summary
¶ Phylogony and Ontogony of smooth pursuit. And WHY make smooth pursuit
¶ Lisberger viewgraphs of SP initiation
¶ SP vs optokinesis; need for fovea
¶ The necessary and sufficient stimulus (vs OKN)
¶ small size--foveal--of target.
¶ delay of 100 msec vs 200 msec for saccades
¶ maximum speed of SP = 30-100°/ sec in humans...
¶ model: internal positive feedback loop for "velocity storage"
¶ opening the loop & finding a unity gain integrator
¶ efferent arm: how is the cerebellum involved? as positive feedback element?
¶ afferent arm--visual cortex involved: V1, MT, PPC-What is the error signal?
initiation and maintenance of SP; Newsome viewgraphs.
¶ Deficits in smooth pursuit, due to cerebellar lesions.
¶ A MODEL-Sensory-Motor transformation in ponsVIEWGRAPH: SP and saccades