ANNUAL

REVIEWS Further
Quick links to online content
Ann. Rev. Neurosci. 1989. 12,' 33--45
Copyright © 1989 by Annual Reviews Inc. All rights reserved

INTEGRATING WITH NEURONS

Access provided by 2806:106e:15:754d:d90d:36ad:c2dc:bd1f on 05/25/20. For personal use only.

D. A. Robinson
Annu. Rev. Neurosci. 1989.12:33-45. Downloaded from www.annualreviews.org

Department of Ophthalmology and Biomedical Engineering, The Johns

Hopkins University, School of Medicine, Baltimore, Maryland 21205

INTRODUCTION

Integration has two meanings in neurophysiology: One indicates some

vaguely specified combination as in " . . . an integration of visual and
vestibular signals may lead to a perception of . . . "; the other comes from
calculus. If x(t) is one variable in time and yet) another, then

yet) = L x(r) dr 1.

says that y is the time integral of x. This mathematical operation occurs

in the central nervous system and is the subject of this review. The review
concentrates on the integrator of the vestibulo-ocular reflex as a prominent
example, offers a model of how integration might be done by neurons, and
speculates about the extent to which neural integrators occur elsewhere in
motor control.
Integration describes physical processes all around us: The volume of a
fluid (blood) in a container (ventricle) is the integral of the inflow (venous
return); the position of the shaft of a d.c. motor is the integral of the current
applied to its armature. These examples, however, are just statements of
physics. They are not examples of devices deliberately constructed by
nature or technology to integrate a signal to achieve some desired end.
Such devices are not very visible in our world. One exception is attached
to the back of our houses, it measures the energy we use by integrating
our power consumption, but most integrators hide in boxes that operate
cranes, fly airplanes, orient satellites, and so on.
In these examples, the integrators are usually located in negative feed­
back control systems. Their value there, to oversimplify, is that integrators
have very large gains at low frequencies, making controllers very accurate
33
014 7-006X/89/0301-0033$02.00
 34 ROBINSON

in the steady state. On the other hand, integrators have low gains at high
frequencies, helping to prevent oscillations. If a control system does not
contain an integrator naturally (such as a motor), the design engineer
will probably add one, if not two, to the system to get the desired perfor­
mance.
Consequently, when engineers became interested in biological control
systems, they took it for granted that integrators were everywhere-how
else could all these control systems possibly work? The oculomotor system
Access provided by 2806:106e:15:754d:d90d:36ad:c2dc:bd1f on 05/25/20. For personal use only.

offers an especially clear example: The retina senses the error between the
eye (fovea) and the target, and the system turns the eye until the error is
Annu. Rev. Neurosci. 1989.12:33-45. Downloaded from www.annualreviews.org

zero--a simple negative feedback scheme (e.g. Young & Stark 1963).
Moreover, when the goal is reached, a constant eye deviation (output) is
maintained while the error (input) is zero. But that is just what an integrator
does-it holds signals in the absence of new information. Indeed, the only
way its output can be constant is if the input error is zero-the desired
condition. So obviously the oculomotor system had an integrator and to
the bioengineer this idea was so obvious as to be trivial.
To the neurophysiologist a neural integrator seemed exotic, but, so long
as it only appeared in top-down, black-box models, it could be relegated
to "higher centers" and ignored. This was not, however, the case for the
vestibulo-ocular reflex. By the early 1960s, it was clear that the signal from
the semicircular canals, coded in the rate modulation of the primary
afferents, was head velocity. The discharge rate of motoneurons, on the
other hand, largely determined eye position. If a constant head velocity is
to make the eyes move at a constant velocity, the motoneurons must
respond to the time integral of the canal signal. This integrator was not
hidden under a feedback loop-it was the major signal-processing element
in a short, well-defined, pontine reflex. It could not be ignored without
also ignoring the main function of the reflex.
Still, it was not until 1968 that this obvious observation first appeared,
however briefly, in print (Robinson 1968). Soon thereafter neuro­
physiological evidence appeared. Cohen & Komatsuzaki ( 1972) found that
electrical stimulation of the reticular formation caused the eyes of monkeys
to move at a constant velocity-the time integral of the step of excitation.
We know now that they were stimulating an input to the integrator (from
saccadic burst neurons), not the integrator itself, but no matter; integration
was clearly occurring. It was also confirmed that motoneurons were
responsible for determining the position of the eye (Robinson 1970), and
by recording from them during sinusoidal rotations of monkeys we
demonstrated the requisite 90 deg phase lag between vestibular and oculo­
motor motoneurons created by the integrator (Skavenski & Robinson
1973).
 INTEGRATING WITH NEURONS 35
Although the integrator's existence was not open to question simply on
theoretical grounds, these findings lent a sort of respectability to the idea
and helped in making clear the integrator's essential role. The concept
was readily accepted in neuro-ophthalmology; after all, it is the neural
integrator that generates the slow phases of nystagmus and holds the eye
eccentrically after a saccade. Disorders of these basic operations could
now be attributed to the integrator.
Access provided by 2806:106e:15:754d:d90d:36ad:c2dc:bd1f on 05/25/20. For personal use only.

THE NEURAL INTEGRATOR IN THE

VESTIBULO-OCULAR REFLEX
Annu. Rev. Neurosci. 1989.12:33-45. Downloaded from www.annualreviews.org

Figure 1 shows the signal processing involved in the reflex. On the right, the
canal produces, in the frequency range of physiological head movements, a
signal proportional to head velocity, H(t), coded in the discharge-rate
modulation. Rv" of primary vestibular afferents. The background rate
(90 spikesjsec) and sensitivity [0.4 (spikesjsec)j(degjsec)] are taken from
Fernandez & Gol dberg (197 1). On the left is shown the well-established
relationship between the modulation of motoneuron discharge rate, llRm

flocculus
i'"_ �
y vermis

E�-_�_ ��'
mn pursuit

R VI =90+0AH
/ '/
I y; I"'n�
I

\ _J_�d�)
.
H )
NI
conol
Figure 1 The final common integrator. On the right the canals transduce head velocity, H,
and report it, coded as the modulation of the discharge rate, R", of primary vestibular
afferents to the vestibular nucleus, vn. This signal becomes an eye velocity command for
vestibular movements, Ev, which is sent directly to the motoneurons, mn, and to the neural
integrator,NI, to provide the needed position signal E. These signals provide those needed
by the motoneurons modulating by /':,.Rm. The pursuit and saccadic signals also arrive as eye
velocity commands, Ep and E" from the cerebellum (flocculus, vermis) and paramedian
pontine reticular formation (pprf), respectively, and are also sent to the motoneurons and
the neural integrator. The latter is contained in the nucleus prepositus hypoglossi and the
vestibular nuclei (dashed lines).
 36 ROBINSON

(around a background rate of typically 100 spikes/sec), and eye position

and velocity. The coefficient k is the neural reflection of the elasticity of
the orbital tissues while r represents orbital viscosity. Typical values are
4 (spikes/sec)/deg and 1.0 (spikes/sec)/(deg/sec), respectively.
For a vestibular command, the motoneurons need a signal proportional
to desired eye velocity (Ev in this case) and desired eye position E(l). The
former can obviously be obtained directly from the vestibular nucleus by
the direct path shown. This agrees, fortunately, with the anatomical fact
Access provided by 2806:106e:15:754d:d90d:36ad:c2dc:bd1f on 05/25/20. For personal use only.

that the neurons in the vestibular nucleus project directly to the moto­
neurons to form the well-known three-neuron arc.
Annu. Rev. Neurosci. 1989.12:33-45. Downloaded from www.annualreviews.org

The signal E, on the other hand, is the integral of E, and, no matter

how it is done, the process can be given its mathematical name: integration,
labeled NI in Figure I. Since this operation occurs in the dark when no
other sense modality can help, it would appear that a network of neurons
in the pons must perform this mathematical operation.
Interestingly, bioengineers themselves felt uncomfortable with the idea
of integrating just with neurons and suggested an alternative familiar
to engineers-velocity feedback. If the output of a control system is
differentiated before being fed back, the overall system behaves like an
integrator. So it was proposed that velocity feedback from muscle pro­
prioception could do the trick (e.g. Fender & Nye 1961). We were able to
eliminate this hypothesis by showing that there was no mono- or pauci­
synaptic stretch reflex for the eye muscles of the rhesus monkey (Keller &
Robinson 1971). This indicated that integration was done somehow by a
network of neurons.
Figure I is greatly simplified to emphasize the signal processing. The
integrator must be a bilateral structure with its halves coupled across the
midline. It receives a push-pull signal from a pair of canals, one canal
modulation decreasing the other increasing. It sends a push-pull signal
to the motoneurons to modulate the agonist and antagonist muscles in
reciprocal innervation.
The integrator does its job well. Figure 1 shows, as discussed below,
that a single integrator is shared by all the conjugate oculomotor sub­
systems. When, for example, the burst neurons create a saccade by sending
a pulse of activity directly to the motoneurons, the eye is held in its new
position by the step of innervation produced by integrating the pulse. The
integrator is not perfect; it leaks and, in the dark, the eye begins sliding
back toward the center with a time constant, Too of about 25 sec (Becker
& Klein 1973). This is so much longer than the interval between most
normal eye movements that this imperfection can be largely ignored.
Lesions can greatly decrease Tn and create a failure of gaze-holding called
gaze-paretic nystagmus.
 INTEGRATING WITH NEURONS 37

LOCATION OF THE INTEGRATOR

For a long time the neural integrator was thought to lie in the paramedian
pontine reticular formation because lesions there affected eye movements
profoundly and the reticular formation seemed a good place for anything
mysterious. This idea was finally disproved when Renn and his colleagues
made neurotoxin lesions there. Ipsilateral saccades were abolished but not
gaze holding (Renn et aI1984). The cerebellum is important in minimizing
Access provided by 2806:106e:15:754d:d90d:36ad:c2dc:bd1f on 05/25/20. For personal use only.

leak rate. Total cerebellectomy (Robinson 1974), ablation of the flocculus

in particular (Zee et al 1981), reduces the time constant Tn to about 1 .3
Annu. Rev. Neurosci. 1989.12:33-45. Downloaded from www.annualreviews.org

sec. This might tempt one to put the integrator entirely in the cerebellum
(Carpenter 1972) except that 1.3 sec is not negligible. During vestibular
nystagmus, the integrator need only integrate well from one quick phase
to the next (roughly 0.3 sec), and even after total cerebellectomy only close
examination will detcct such lcakiness in nystagmus recordings. Moreover,
during ice-water stimulation of one ear, simulating a vestibular lesion, Tn
in humans is deliberately decreased to about 2.4 sec within 80 sec of the
onset of the inappropriate nystagmus (Robinson et al 1984). The cere­
bellum may well be responsible for parametric adjustments with this time
scale. It is shown below that lesions of the vestibulo-prepositus complex
abolish all integrator action, so the current thinking is that the integrator
is basically in this complex, but the cerebellum has a powerful influence in
adjusting its time constant.
Studies of cells in the vestibular nuclei of alert monkeys discovered that
a large proportion of cells in the superior and rostral-medial subdivisions
carried, among others, the eye position signal E. Many cells are purely
oculomotor in that their activity reflects eye movements whether or not of
vestibular origin. Thus, significant subdivisions of these nuclei form an
eye-movement nucleus. This led Tomlinson & Robinson (1984) to propose
this region as the site of the neural integrator. Meanwhile, the nucleus
prepositus hypoglossi, right next door, had also been suggested because
its cells also carried the eye-position signal and projected directly to moto­
neurons. Finally, Cheron et al (1986) showed integrator failure after electro­
lytic lesions of either region in the cat, and Cannon & Robinson (1987)
showed in the monkey total loss of the neural integrator after bilateral
neurotoxin lesions of both the prepositus and medial vestibular nuclei.
The latter study showed that, as one would predict from Figure 1,
without the integrator the eye velocity commands would pass directly to
the motoneurons and produce an eye position that was proportional to
desired eye velocity. For a step of head velocity, for example, thc step of
Ev, in the absence of the ramp normally generated by the integrator, simply
causes a step change in eye position without the usual slow phases of
 38 ROBINSON

nystagmus. Similar results occurred with pursuit and optokinetic stimu­

lation. For the saccadic pulse, the eyes moved quickly to one side but,
without the usual step from the integrator, the eyes returned rapidly to
straight ahead with the time constant of the orbital mechanics (about 200
msec). Consequently, the time constant of the integrator, if any, was
significantly less than 0.2 sec. Thus, in addition to locating the integrator,
this study showed, as had long been proposed, that a single integrator was
used by all conjugate oculomotor systems.
Access provided by 2806:106e:15:754d:d90d:36ad:c2dc:bd1f on 05/25/20. For personal use only.

MODELS OF NEURAL INTEGRATORS

Annu. Rev. Neurosci. 1989.12:33-45. Downloaded from www.annualreviews.org

If an engineer wants to build an integrator, either rate feedback, mentioned

above, or positive feedback are the two usual choices. The former is risky:
It requires large gains around the feedback loop so that if, by a lesion,
the loop is broken, the output would try to rise to very large values.
Positive feedback is more failsafe. One starts with a very leaky integrator
with a time constant r. Positive feedback of gain k around it raiscs the
effective time constant to r/ ( l-k). The integrator becomes perfect when
k is 1.0; no large gains are involved, and if feedback is lost, the integrator
simply returns to being very leaky. Put another way, if cells excite their
neighbors and are excited by them, then cells excite themselves and this
perseverates activity, once started, by a "system of reverberating colla­
terals." To date, this has been the only model seriously considered.
On the other hand, one could, for example, hypothesize a neuro­
transmitter we could calJ integratide. When released into the subsynaptic
cleft, it binds to the subsynaptic cell membrane and depolarizes it (so the
cell fires faster) indefinitely until integratase is released into the cleft to
inactivate the integratide and decrease depolarization and firing rate, again
indefinitely, until a new signal comes along. One cannot object to this idea
because the time constant of the neural integrator can be decreased from
25 to 2.4 sec within 80 sec of caloric stimulation; one need only hypothesize
another neuromodulator that breaks down both integratide and inte­
gratase with the same suitable rate constant. Unfortunately, such neuro­
modulatory behavior has not yet been observed. Long-term synaptic
changes have been observed in Aplysia (Frost et al 1985) but these are
changes in sensitivity or gain, a multiplicative operation, quite unlike the
linear operation of integration.
Returning to reverberating collaterals or positive feedback, we find two
problems arising. The first is that all the signals to be integrated (except
the saccadic pulse) ride on a background discharge rate. From Figure I,
the vestibular background rate is 90 spikes/ sec. Recordings from other
neurons in the region of the integrator (e.g. Tomlinson & Robinson 1984)
 INTEGRATING WITH NEURONS 39
show that 100 spikes/sec is typical. We don't want to integrate this back­
ground rate, that would be disastrous, just the modulation riding on it.
Shamma & Cannon hit upon lateral inhibition as an exceedingly simple
way to solve this problem (Cannon et aI 1983). In Figure2A, each neuron
inhibits its neighbor with strength w and is inhibited by it. Thus, each cell
excites itself by disinhibition-positive feedback. In this scheme, the initial
lag, is the membrane time constant of an individual neuron, about 5 msec.
Analysis shows that when both inputs, Ul(t) and U2(t), change together,
Access provided by 2806:106e:15:754d:d90d:36ad:c2dc:bd1f on 05/25/20. For personal use only.

the outputs, x let) and xlt), respond with a time constant of r/ ( l + w) or,
since w is close to 1.0, about 2.5 msec. Thus, the background rates are
Annu. Rev. Neurosci. 1989.12:33-45. Downloaded from www.annualreviews.org

simply passed through the system unchanged, as shown initially in Figure

2A. But when Ul and Uz change in push-pull, by l1u in Figure2A, Xl and
X2 differ, the feedback loop starts to operate, and the cells respond with a

time constant of ,/ ( 1- w). If w is very close to 1.0, this time constant can
be very large, and integration takes place. It is remarkable that lateral
inhibition, a work horse of neural network modelers, solves this problem
as well.
The second problem is that to increase the time constant from 5 msec
to 20 sec, the value of w must be 0.99975. Of course, if the membrane time
constant of these particular neurons was 50 instead of 5 msec, the value
of w becomes 0.9975, but this is still hardly robust. On the other hand, no
one supposes that two neurons are adequate. Cannon et al (1983) examined
a ring of 32 neurons connected as in Figure2B. Figure2C shows that the
Bode diagram (log gain vs log frequency, OJ, if the input were a sinusoid)
depends on the spatial frequency, P, of the inputs. If all the cells received
the same signal (e.g. the background rate), its spatial frequency would be
zero and the network would act like a wide-band, low-gain system with a
time constant of2.5 msec (front edge in Figure2C). If the inputs from left
and right canals (or pursuit cells or saccadic burst neurons) are interleaved
as shown, one has the highest spatial frequency (n/2 or one half cycle/
neuron), and the system has the Bode diagram of an integrator with a time
constant of 20 sec (back edge in Figure 2C).
We were able to show that adding spatial noise to such a model (changing
synaptic strengths randomly throughout) did little to perturb its integrating
behavior. Nothing depended heavily on only a few synapses. Killing one
cell (out of 32, 3%) caused the impulse response (a saccade) to drift back
quickly by about 15%, because the dead neuron broke local feedback
loops with nearby cells, but then to recover and drift toward zero at a rate
appropriate to a 20 sec time constant. This result indicated that the major
property of the network did not depend critically on any one parameter
in the model. The model was robust.
Integration still requires a critical adjustment in a global sense. If Xes, P)
 40 ROBINSON

X X
I 2

t::= ,
� t
X
I
X
2
C
A

W
Access provided by 2806:106e:15:754d:d90d:36ad:c2dc:bd1f on 05/25/20. For personal use only.

I I
Annu. Rev. Neurosci. 1989.12:33-45. Downloaded from www.annualreviews.org

u, u, u. u.

� �

Figure 2 A model of the neural integrator. A: Lateral inhibition. Each cell excites itself
(positive feedback) by self-disinhibition through its neighbor. A common input, UI and U2,

is not integrated but is simply repeated by the outputs x I and x2• A differential input, on the
other hand, such as flu, is integrated as shown. B: A fuller model with 32 neurons. The
spatial variablek runs from I to 32. The neuron at ko inhibits itself and neighboring cells
with the strength w(k-ko) shown by the double Gaussian curve at the top. The inputs enter
from push-pull sources and alternate on the neurons to achieve the highest spatial frequency:
nl2 or one half cycle per neuron. C: The Bode diagram of the gain XI U of this network with
the spatial frequency of the input P as a third dimension. As P approaches n12, the system
behaves like an integrator with a time constant T of 20 sec.
 INTEGRATING WITH NEURONS 41

and U(s,P) are the transforms of the outputs and inputs, where s is the
Laplace transform frequency in the temporal frequency domain and P is
the Fourier transform frequency in the spatial frequency domain, then

X(s,P)
2.
U(s, P) S7: + [1- W(P)]

where W(P) is the Fourier transform of the spatial function of the pattern
Access provided by 2806:106e:15:754d:d90d:36ad:c2dc:bd1f on 05/25/20. For personal use only.

of lateral inhibition w(k-ko) shown in Figure 2B. The effective network

time constant is r/[ I - W(P)] when P is near the highest spatial frequency
Annu. Rev. Neurosci. 1989.12:33-45. Downloaded from www.annualreviews.org

of one half cycle/neuron. Thus, W(P), encompassing hundreds of cells and

hundreds of thousands of synapses, must have a global value of 0.99975.
Of course, any network must be able to have its function monitored and
its parameters adjusted. In the case of this integrator one thinks of the
cerebellum. Integrator failure results in eye motion during attempted,
eccentric fixation. The resulting image motion is sensed by the retina and
reported to the inferior olive by way of brainstem visual pathways via the
nucleus of the optic tract. The olive reports this signal to the cerebellum,
particularly the flocculus, via climbing fibers. The flocculus in turn is
well connected to the prcpositus-vestibular complex and is known to be
involved in several ways in decreasing retinal image motion (Zee et al
1981). Thus, a reasonable hypothesis is that the flocculus monitors the
error (retinal slip) and adjusts the connectivity of the integrator network
to fine tune its ability to hold eye position. How the latter might be done
is quite unknown. The main point here is that in the model of Figure 2,
such visual feedback and parametric adaptation should be able to adjust
and maintain the time constant of the integrator. Note that moderate
changes of many individual synapses still contribute only infinitesimal
changes in the global property W(P).
The model in Figure 2B is, of course, much too simple. All the neurons
are identical and inhibitory and all the inputs are identical. To show that
these conditions could be relaxed, Cannon & Robinson (1985) extended
the model. The background discharge rates of the incoming fibers and
cells of the network could be allowed to vary over the population in a
realistic manner. Most important is that almost all the cells in the pre­
positus-vestibular complex carry not just the eye position signal E but a
variety of combinations of En Bv, and Bp as well (Figure 1). The model in
Figure 2 is not dynamically rich enough to do this; it integrates too well
and all the cells just carry E. To provide velocity terms, we used a doub1e­
layer model; a row of excitatory and a row of inhibitory cells that talk, in
certain ways, to both types of neighbors. This is, of course, a minimal step
in the right direction since the integrator output to downstream cells,
 42 ROBINSON

especially the motoneurons, must be excitatory as well as inhibitory.

Finally, we showed that the integrated signal could spread almost instantly
through the whole network, even if the velocity input entered only part of
it, so that neurons could carry E while carrying En Ev, and Ep at quite
different strengths, as observed experimentally.
The time is ripe, now that we know the location of the integrator, to
test the hypothesis in Figure 2 and to modify it or replace it with a
better one. For example, it depends heavily on fibers crossing between the
Access provided by 2806:106e:15:754d:d90d:36ad:c2dc:bd1f on 05/25/20. For personal use only.

bilateral prepositus-vestibular complexes, similar to a proposal by Galiana

& Outerbridge (1984). Preliminary results by T. 1. Anastasio and myself,
Annu. Rev. Neurosci. 1989.12:33-45. Downloaded from www.annualreviews.org

trying to make midline lesions to interrupt these fibers, have met with
peculiar results. In only one monkey so far, electrolytic lesions in the
midline, in the region 0 to 4 mm caudal to the abducens nuclei and 4 mm
deep, did decrease the integrator time constant to 0.6 sec after each lesion,
but recovery was significant in just a few hours and more so after two
days. Two nagging questions arise: What is the anatomical course of these
crossing fibers-did we interrupt them? What pathways and mechanisms
mediate this incredible capacity for repeated recuperation?
Equally important, new models should be proposed by theoreticians to
provoke further experimental tests. The operation we are considering is
not trivial.

INTEGRATORS ELSEWHERE

The integrator we have considered so far is for horizontal, conjugate

eye movements. Vertical eye movements are probably combinations of
movements created by two integrators, in part in the caudal mesen­
cephalon, with planes of action near those of the vertical semicircular
canals and the vertical recti and oblique muscles. Current findings in
the vergence system suggest another integrator there (Mays et al 1986).
That's four. Current models of gaze saccades suggest integrating the head
velocity signal, iI, to obtain head position, H, needed for internal cal­
culations of eye in space, as opposed to eye in head (e.g. Laurutis &
Robinson 1986). That could add three more, one for each degree of free­
dom. Thus, in the oculomotor system, we can see that there is nothing
special about the integrator shown in Figure 1; integrators crop up every­
where.
Can we then conclude that integrators are everywhere in motor control?
Before extrapolating too quickly, it is important to appreciate some
unusual features of the oculomotor system. Phylogenetically, the vestibulo­
ocular reflex has dominated its development. In lateral-eyed, afoveate
animals, this reflex and its visual extension, the optokinetic system, rep-
 INTEGRATING WITH NEURONS 43
resent almost their entire oculomotor capability. In frontal-eyed, foveate
animals, the saccadic, pursuit, and vergence systems were added. The
vestibulo-ocular reflex is dominated by the semicircular canals, which are
unique sense organs. Although each contains a sheet of sensory neurons
(the hair cells), all the receptors report the same signal to the brainstem­
that component of the head rotation vector that is perpendicular to the
plane of that canal. There are no maps as in the visual, auditory, and
somatosensory systems-the signal is entirely one-dimensional. More­
Access provided by 2806:106e:15:754d:d90d:36ad:c2dc:bd1f on 05/25/20. For personal use only.

over, the canal signal rides on a high, stable, resting background rate so
that modulation both up and down, in push-pull, can occur, thereby
Annu. Rev. Neurosci. 1989.12:33-45. Downloaded from www.annualreviews.org

avoiding the most severe nonlinearity in the nervous system: Discharge

rate cannot be negative. The result is linearity. It has also been shown that
the canals have even imposed their planes upon the pulling directions of
the extraocular muscles, even in frontal-eyed animals. Thus, the canals
have imposed on the oculomotor system (a) the pulling directions of the
muscles, (b) a high background discharge rate, (c) push-pull operation
(expressed at the motor end by strict reciprocal innervation), (d) linearity.
These features are exhibited not only by the motoneurons (Figure 1) but
by all the interneurons in this reflex.
In the spinal cord, on the other hand, high background rates, reciprocal
innervation, and especially linearity, are not the order of the day. Because
of cocontraction and the need to handle a wide variety of loads utilizing
proprioceptive feedback (unnecessary and not found in the eye), one would
not even hope to write an equation, as in Figure 1, between motoneuron
discharge rate and load position that was even unique, let alone linear.
Nevertheless, this does not mean that it is too soon to think about inte­
grators in the spinal cord. If I point straight ahead with extended arm and
hand, what keeps my arm from succumbing to gravity? To say that "tonic
signals" form "higher centers" are responsible is simply to avoid the
question.
In the cord one can, of course, use position feedback via proprioception
to hold a limb still, turning it on only during limb fixation. This is, however,
not an integrator; a tonic, central signal is still needed, if only to the y
motoneurons, and where does that come from? One could try to build an
integrator by proprioceptive rate feedback, as mentioned above. The major
problem with all these speculations is that because recording from single
units in the spinal cords of alert, behaving animals is technically difficult,
we have almost no idea of the signal processing that goes on there. Working
out anatomical pathways and some properties of basic spinal reflexes in
immobile animals helps almost not at all in fathoming what the cord does
to the signals we see descending from the cerebellum and motor cortex
before they reach the motoneurons. Without seeing the signals in the
 44 ROBINSON

incredibly rich internuncial networks in the cord, it is difficult to even guess

about premotor signal processing.
In the saccadic system, the signal most often seen in supramesencephalic
structures is transient, indicating an impending change in eye position,
more resembling movement velocity than position. Only when these signals
descend to the caudal pons (for horizontal movements) are they converted
by an immediately premotor network-the integrator-from velocity to
position commands. Similarly, the majority of intracranial signals related
Access provided by 2806:106e:15:754d:d90d:36ad:c2dc:bd1f on 05/25/20. For personal use only.

to limb movements also seem to be phasic or velocity-related, although

this is certainly not always so. If there is any analogy to the saccadic (and
Annu. Rev. Neurosci. 1989.12:33-45. Downloaded from www.annualreviews.org

pursuit system as well), these phasic signals would be converted to limb

position commands at the last minute, namely in the cord close to the
motor nuclei. Of course, identification of such a process will be con­
siderably complicated by the wealth of proprioceptive feedback signals,
but for those who see value in the bottom-up approach, we will never even
start to unravel such mysteries unless we record from premotor neurons
in behaving animals and find out what they are telling the motoneurons.

Literature Cited

Becker, W., Klein, H. M. 1973.Accuracy of tigation of the mechanisms of eye move­

saccadic eye movements and maintenance ment control. Kybernetik I: 81-88
of eccentric eye positions in the dark. Pernandez, c., Goldberg, J. M. 1971.Physi­
Vision Res. 13: 1021-34 ology of peripheral neurons innervating
Cannon, S. C., Robinson, D. A. 1985. An semicircular canals of the squirrel mon­
improved neural-network model for the key. II. Response to sinusoidal stimula­
neural int eg ra tor of the oculomotor sys­ tion and dynamics of peripheral ves­
tem: More realistic neuron behavior. Bioi. tibular sy stem . J. Neurophysiol. 34: 661-
Cybern.53:93-108 75
Cannon, S. C., Robinson, D. A. 1987. Loss Frost, W. N., Castellucci, R. D., Hawkins,
of the neural integrator of the oculomotor R. D., Kandel, E. R. 1985. The mono­
system from brain stem lesions in monkey. synaptic connections from the sensory
J. Neuraphysiol. 57:1383-1409 neurons participate in the storage of long­
Cannon, S. c., Robinson, D. A., Shamma, term memory for sensitization of the gill­
S. 1983. A proposed neural network for and siphon-withdrawal reflex in Aplysia.
the integrator of the oculomotor system. Proc.Natl. Acad. Sci. USA 82: 8266-69
Bioi. Cybern. 49: 127-36 Galiana, H. L., Outerbridge, J. S. 1 984. A
Carpenter, R. H. S. 1972. Cerebellectomy bilateral model for central neural path­
and the transfer function of the vestibulo­ ways in vestibuloocular reflex. J. Neura­
ocular reflex in the decerebrate c at. Proc. physiol.51: 210--41
R. Soc. London Ser. B 181 : 353-74 Henn, V., Lang, W., Hepp, K., Re isine, H.
Cheron, G., Godaux, E., Laune, J. M., van­ 1984. Experimental gaze palsies in mon­
Derke1en, B. 1986. Lesions in the cat pre­ keys and their relation to human path­
positus: Effects on the vestibulo-oeular ology. Brain 107:619-36
reflex and saccades. J. Physiol. London Keller, E. L., Robinson, D. A. 1971. Absence
372: 75-94 of a stretch reflex in extraocular museles
Cohen, B., Komatsuzaki, A. 1972. Eye of the monkey. J. Neurophysial. 34: 908-
movements induced by stimulation of the 19
pontine reticular formation: Evidence for Laurutis, V. P., Robinson, D. A. 1986. The
integration in oculomotor pathways. Exp. vestibulo-ocular reflex during human sac­
Neural. 36: 101-7 cadic eye movements. J. Physiol. London
Fender, D. H., Nye, P. W. 1961. An inves- 373: 209-33
 INTEGRATING WITH NEURONS 45
Mays, L. E., Porler, J. D., Gamlin, P. D. Skavenski, A. A., Robinson, D. A. 1973.
R., Tello, C. A. 1986. Neural control of Role of abducens neurons in the ves­
vergence eye movements: Neurons encod­ tibuloocular reflex. J. Neurophysiol. 36:
ing vergence velocity. J. Neurophysiol. 56: 724-38
1007-21 Tomlinson, R. D., Robinson, D. A. 1984.
Robinson, D. A. 1968. Eye movement con­ Signals in vestibular nucleus mediating
trol in primates. Science 161: 1219-24 vertical eye movements in the monkey. J.
Robinson, D. A. 1970. Oculomotor unit Neurophysiol. 51: It21-36
behavior in the monkey. J. Neurophysiol. Young, L. R., Stark, L. 1963. Variable feed­
33: 393-404 back experiments testing a sampled data
Robinson, D. A. 1974. The effect of cere­ model for eye tracking movements. IEEE
bellectomy on the eat's vestibulo-ocular Trans. Prof Tech. Group on Human Fac­
Access provided by 2806:106e:15:754d:d90d:36ad:c2dc:bd1f on 05/25/20. For personal use only.

integrator. Brain Res. 71: 195-207 tors Electron. HFE-4: 38-51

Robinson, D. A., Zee, D. S., Hain, T. C., Zee, D. S., Yamazaki, A., BUller, P., Giicer,
Holmes, A., Rosenberg, L. F. 1984. Alex­ G. 1981. Effec t of ablation of flocculus
Annu. Rev. Neurosci. 1989.12:33-45. Downloaded from www.annualreviews.org

ander's law: Its behavior and origin in and paraflocculus on eye movement in pri­
the human vestibulo-ocular reflex. Ann. mate. J. Neurophysiol. 46: 878-99
Neurol. 16: 714-22