Tải bản đầy đủ - 0trang
3 Subthalamic Nucleus and Primary Processes of Motivation: Consumption
as a result of the extrafusal muscle ﬁbers in parallel with the spindle, or as a result
of the contraction of intrafusal muscles due to gamma neuron activity. It reﬂects the
difference between desired muscle length and actual length—the error signal in the
length controller. The spindle acts as a mechanical comparator, as a stretch detected
by muscle spindle is compared with the net reference signal for muscle length. The
discrepancy or error is then sent to the alpha motor neuron to produce muscle contraction. At the same time, however, the alpha motor neurons can also be commanded directly by corticospinal projections. Such direct adjustment of force
reference seems particularly important for the movement of distal digits.
The descending signals for length reference come from the gamma motor neurons, which in turn receive inputs from the brainstem and the cerebral hemispheres.
When gamma motor neurons are activated, the contractile parts of the spindle (intrafusal ﬁbers) attempt to shorten. A pull is generated at the equatorial region of the
spindle that results in Ia afferent activity, even though the spindle length does not
change signiﬁcantly because it is anchored at both ends. Consequently, alpha motor
neurons are activated.
Position Control: Joint Angle and Body Configuration
To change a joint angle, the lengths of multiple muscles must be changed simultaneously. In turn, a body conﬁguration consists of a set of joint angles. In each case,
the relevant perceptual input is represented as a one-dimensional signal. The controlled variable is a conﬁguration of lower order proprioceptive inputs, and the
output function can reach a group of muscles that work together to produce the
appropriate net effect.
The control of body conﬁgurations is a type of position control. In position control, the controlled variable represents some position coordinate, and output is generated by computing the difference between the reference or desired position coordinate
and the input signal reporting the actual position. For motion with multiple degrees
of freedom, the actual position vector is determined by the action of multiple orthogonal controllers. In the brainstem, for example, there is evidence for distinct position
controllers for vertical and horizontal movements (Deliagina et al. 2012; King et al.
1981; Luschei and Fuchs 1972; Masino 1992; Masino and Knudsen 1990).
The key neural substrates for posture control are the reticulospinal, vestibulospinal, and rubrospinal pathways (Deliagina et al. 2008; Foreman and Eaton 1993;
Peterson et al. 1979). For example, stimulation of the reticulospinal pathway can
produce coordinated changes in joint angles: depending on stimulation location,
ipsilateral ﬂexion and contralateral extension or the opposite pattern of ipsilateral
extension and contralateral ﬂexion can be produced (Sprague and Chambers 1954).
The reticulospinal pathway receives direct and indirect projections from the SNr.
These descending projections are assumed to alter the reference signals for body
20 The Basal Ganglia and Hierarchical Control in Voluntary Behavior
Locomotion and Gait
Although the reticulospinal pathway receives direct projections from the BG, more is
known about the indirect projections via the mesencephalic locomotor region (MLR)
and the pedunculopontine nucleus (PPN) (Garcia-Rill 1986; Sherman et al. 2015;
Shik and Orlovsky 1976). In addition, another area critical for locomotion, the diencephalic locomotor region (DLR), is found in the ventral thalamus in the lamprey and
thought to be analogous to the zona incerta in mammals (Menard and Grillner 2008).
DLR appears to be functionally similar to MLR but their respective roles remain
unclear (Mogenson and Nielsen 1983). In rodents, the substantia innominata or ventral pallidum projects to the zona incerta as well as the pedunculopontine nucleus.
One possibility is that the DLR is primarily concerned with head movements, whereas
the MLR is important for axial and limb movements. This is supported by studies
showing rotational movements of the head and eyes produced by zona incerta stimulation (Hyde and Toczek 1962) and deviation and torsion of the eyes as well as tilting
of the head following damage to this region (Hedges and Hoyt 1982).
The MLR is a primary target of BG outputs (Grillner et al. 2008). Recent work
showed that the core of the MLR is not in the PPN, but the lateral pontine tegmentum, which sends direct glutamatergic projections to the ventral horn of the spinal
cord. The SNr output reaches the midbrain extrapyramidal area, which in turn projects to the lateral pontine tegmentum (Sherman et al. 2015). The ﬁring rates of many
MLR neurons are correlated with speed of running (Lee et al. 2014). Lesions of this
area produce cataplexy and episodic immobility. There appears to be a limb extensor circuit in the ventral MLR that is critical for the standing posture, receiving an
inhibitory projection from the SNc (GABAergic and dopaminergic neurons); the
dorsal MLR contains a ﬂexor-dominant circuit necessary for locomotion, receiving
a projection from the SNr (Sherman et al. 2015).
The PPN and MLR send projections to the reticulospinal pathway, where they
alter the activity of locomotor oscillators (Garcia-Rill et al. 2014; Lee et al. 2014;
Moruzzi and Magoun 1949). But the same region is also part of the reticular activating system that can enhance overall arousal in the forebrain (Garcia-Rill et al. 2014;
Lee et al. 2014; Moruzzi and Magoun 1949). Autonomic functions as well as perceptual functions are adjusted at the same time to accommodate the needs of
In addition to the SNr output, the ventral tegmental area (VTA) also appears to
provide a critical BG output to locomotion-related body conﬁguration controllers
(Swanson and Kalivas 2000; Wang and Tsien 2011). Common to all forms of locomotion is the alternating swing of the body. Locomotion involves not only alternating rhythms in limb movements (e.g., extensor/ﬂexor alternation), but also left right
alternation in the spine conﬁguration in locomotion. Whereas the MLR appears to
be more important for regulation of limb joint angle, the PPN and other related
regions could be critical for controlling torso joint angles, e.g., bending the spine.
The BG are not responsible for generating locomotor rhythm per se, which
depend on lower levels of the hierarchy. But the BG output can determine the rate
of change in body conﬁgurations. In locomotion, the rate of propulsion is proportional to the rate of change in the conﬁguration reference signal.
Orientation Control and the Tectum
Neural position controllers differ in how they sense position. Proprioceptive sensors
from muscles and joints report the state of the body. Senses such as vision and hearing are used to report distal changes in the environment, and the relevant receptors
must be moved toward the relevant part of the environment. This type of signal
acquisition is equivalent to orientation control. Orientation is not a property of the
organism only, but of the relationship between the organism and its external environment. The levels below orientation control cannot act on the environment in any
directed way. Devoid of any distal senses, one cannot move towards some distal
The key neural structure for orientation control is the tectum, a large region in the
midbrain that receives visual, auditory, and somatosensory inputs. Indeed, in some
organisms the primary visual analysis occurs at the level of the tectum (superior
colliculus), which enables adaptive behaviors such as striking at prey (Lettvin et al.
1959). Signals from each tectal sensory map converge on the comparator. The error
signal from the comparator, in turn, has access to a number of body conﬁguration
controllers. The error signal is transformed into the position reference signals that
ultimately specify the movements. For example, to foveate on a moving target, one
can move the eyes, the head, or the whole body. Any of these can acquire the visual
input needed, thus achieving the orientation of “straight ahead,” and normally all
can be engaged at the same time. The head will turn if the body is restrained, and the
eyes will move if the head is restrained.
Foveation is just another example of the control of input. The photoreceptors are
organized retinotopically, and moving up the neural hierarchy this organization is
retained at multiple levels, including the superﬁcial layer of the superior colliculus
(Drager and Hubel 1976; Robinson 1972). Activity of each unit on this map can
have a certain range of values. To orient the sensory receptors towards the distal
stimulus, the value of the relevant units on the map can be matched with a reference
value. Thus the sensory signal acquired at each map location is dictated by the reference. This is effective not only for foveation, which keeps a high level of activation
roughly in the “center” of the visual ﬁeld, but for other types of behaviors, e.g.,
grasping or whisking, that rely on different sensory receptors.
The high ﬁring rates on a map of low ﬁring rates represent the reference location,
the default setting which allows the eyes to be “centered” at rest. Movements will
be produced to reduce the error until each unit returns to the baseline level speciﬁed
by the reference for peripheral units. But as soon as the position changes to the
receptive ﬁeld of the neighboring unit, that unit too will generate a direction-speciﬁc
movement to reduce its error, and so forth until the foveation units are reached.
These units have different reference signals, so that the high perceptual input no
longer creates an error signal.
20 The Basal Ganglia and Hierarchical Control in Voluntary Behavior
The tectal orientation controller can function in two modes. In the bottom-up
mode, any salient stimulus can trigger an orienting movement. Pavlov called this the
“what is it reﬂex,” whose primary function is to orient one towards changes in the
environment, in preparation for possible behavioral engagement, whether to
approach, avoid, or ignore (Konorski 1967). In the top-down mode, the animal can
select any part of its perceptual ﬁeld and orient towards it, regardless of its salience.
Nigrotectal Projections Send Descending Reference
On the tectal map, activation of any unit above its reference level will generate an
error signal from the comparator, resulting in movement. In foveating animals, it
appears that high level of activation is allowed only in the rostral fovea representation, as a result of topographical differences in the reference signals received. To
break foveation, the higher level must reset the reference for the periphery. For
“voluntary” saccades, neurons are activated in a region that represents the location
to which the saccade will be directed. Just prior to a saccade, activity rapidly builds
up at the target location, reﬂecting the reference signal sent to the chosen location in
the peripheral visual space that is not currently foveated. Descending projections to
the tectum select which part of the map is to be activated.
A major source of descending reference signals is the BG, via the massive nigrotectal projections. The SNr sends direct GABAergic projections to the tectum
(Beckstead 1983; Redgrave et al. 1992; Rinvik et al. 1976). The nigrotectal projection is critical for self-initiated or memory-guided saccades (Hikosaka et al. 2000).
When the monkey must move its eyes towards some arbitrary target, SNr neurons
pause transiently, whereas their target neurons in the intermediate layers of the
superior colliculus burst. The intermediate layer receives two sets of inputs, glutamatergic inputs from the superﬁcial layers, where the visual inputs arrive, and
GABAergic inputs from the SNr, which represent descending reference signals
from the BG (Isa and Hall 2009). Thus these neurons can implement a comparator
function to generate the difference between reference and input.
Turning and Steering
Orientation control is critical for steering during locomotion. Unilateral striatal
stimulation produces muscle contraction on the contralateral side of the body.
Unilateral striatal (caudate) lesions produce a posture in which neck and body are
curved towards the side of the lesion. The laterally curved posture results from an
imbalance of the activities of the two striata, and the animal turns away from the
side of the greater activity, presumably due to contralateral contraction and ipsilateral relaxation of different muscle groups along the spine (Ferrier 1876; Jung and
We assume that leftward modules are mainly located in the right striatum, and
rightward modules in the left striatum. Balanced outputs from both striata are needed
for moving straight ahead. Thus, the net effect of stimulation on the right side will be
leftward turning and vice versa. Unilateral muscimol injections into the SNr, by activating GABA-A receptors, can mimic the effects of unilateral striatonigral transmission (Cools 1985). Cools found that the animal repeatedly attempted to bridge the gap
between the current position and some egocentrically deﬁned point. The fully symmetric posture is the point of departure, or neutral position, for the drug-induced
movements. While the initial turning movements are restricted to that of the head,
with time there is progressively more movements with different points of departure
moves from the ears to the eyes, to the midline of the head, and then more caudally,
from the head to the tail (oculus, auriculum, cranium, scapula, and pelvis). The point
of departure lies on the vertical axis of the egocentric coordinate system. These are the
locations where the descending reference signals can alter the body conﬁguration.
According to the present model, the BG implement transition control. By sending
reference signals to position controllers, the BG can reach some desired rate of
change (transition) in different perceptual variables. This insight was anticipated by
Cools, who speculated that the inhibitory output from the SNr represents reference
signals, which contain a “propriotopic code,” and that the BG circuit could implement transition control (Cools 1985). But lacking knowledge of the computational
roles of striatal and nigral neurons and the neural integrator needed to convert velocity error into position reference, Cools could not provide a model of how transition
control is implemented.
The results discussed earlier, which reveal representation of movement kinematics by BG neurons, suggest a control system for movement velocity, in which the
rate of change in body conﬁgurations is a controlled variable (Yin 2014a). In a
velocity controller, the actual velocity will match the desired velocity.
If we only cared about start and end positions, then position control would be
adequate, and the speed with which the position is altered is determined largely by
loop gain. Given a speciﬁc amount of position error, the speed cannot be varied and
the movements may appear jerky. But if we wish to vary how quickly this transition
occurs, it is important to control the rate of change itself.
Velocity control is critical for smooth motion. To achieve it, it is necessary to
sense the rate of change in body conﬁgurations. For example, in an engineered speed
controller, the rotation speed of the motor can be sensed, and this sensed value is
compared with the reference to calculate the error signal. Any deviation is used to
generate the output. In a biological organism, there are proprioceptive inputs that sense
the rate of transitions (Yin 2014a). Movement velocity in this sense is independent
20 The Basal Ganglia and Hierarchical Control in Voluntary Behavior
of movement in space. On the other hand, rate of change in exteroceptive inputs,
such as optic ﬂow, is detected and controlled by a different system (see below).
Cascade Organization and Velocity Control
Engineered velocity controllers and position controllers are similar in design, with
the key difference being how the output is measured or transduced. For example, a
potentiometer changes its voltage output proportional to the position moved,
whereas a tachometer generates a voltage proportional to the rotational speed of the
motor. In a biological organism, these are achieved by the perceptual inputs at different levels of the hierarchy. New properties emerge when a velocity controller is
placed just above a position controller, i.e., arranged hierarchically. This arrangement is similar to a common engineering design called cascade control, with the
higher controller called the inner or master loop and the lower controller the outer
or slave loop.
Suppose we have a position controller with three positions: 1, 2, and 3. A reference signal of 3 is a command to sense the position 3. If the reference signal is ﬁxed
at 3, this position controller will maintain position 3, producing variable outputs as
needed to resist the effect of environmental disturbance. To the casual observer,
however, there is no movement whatsoever. If we plot position (a string of 3 s) over
time, we would see a ﬂat line. If we change the reference to 2—this command tells
the position controller to bring the reading of its position transducer to the new reference value. But to control how quickly the value changes from 3 to 2, velocity is
needed. In the cascade organization, the velocity loop is placed just above the position loop, so that the error in the velocity controller becomes the reference for the
position controller (Fig. 20.7). For this to be possible, the error of the velocity
controller must be converted to the rate of change in the position reference. That is
to say, integration is needed.
The neural implementation of movement velocity control is the sensorimotor
cortico-BG network. As reviewed earlier, striatal activity is related to velocity
whereas nigral activity is related to position. The present hypothesis is that striatal
output reﬂects the velocity error signal, which is integrated by the SNr. Thus the
velocity error is converted to a rate of change for the position reference, regardless
of the start or end position.
Integrators are common in negative feedback systems. When “integral gain” is
used in the output function, the output is proportional to the time integral of the
error signal. Output can be produced even when the current error is zero, maintaining a steady reference command to downstream systems. With no leak or discharging, the amount of output at any given time reﬂects total error accumulated
in the past. In principle, the steady state error of this system can be zero and the
loop gain inﬁnite.
Fig. 20.7 Relationship between velocity control and position control. (a) An illustration of the
hierarchical relationship between a velocity (transition) controller and a position controller. The
velocity error is integrated to yield position reference. The rate of change in the BG output therefore reﬂects movement velocity, while the ﬁring rate itself represents position. (b) Illustration of
charging and discharging the integrator. The lever holding task (Fig. 20.2) is used as an example.
On the left are the velocity errors from the striatum that are assumed to enter the integrator, and on
the right is the position reference from the SNr
When integral gain is used in the transition controller, the latest position achieved
is maintained. The BG output nuclei, e.g., the SNr, are characterized by high tonic
ﬁring rate, which can be explained by membrane properties of the GABAergic projection neurons (Zhou and Lee 2011). But functionally the signiﬁcance of such high
20 The Basal Ganglia and Hierarchical Control in Voluntary Behavior
ﬁring rates has never been clear. According to the present model, this is because SNr
output represents a position vector. To maintain position, the SNr output neurons
continue to ﬁre at the same rates (Fig. 20.2). When the ﬁring rates change, a new
position is achieved. The dynamic range of the neural signaling reﬂects the range of
egocentric position coordinates, and maximum movement amplitude.
Direct and Indirect Pathways
An important feature of the BG is the parallel direct (striatonigral) and indirect
(striatopallidal) pathways (Gerfen 1992; Parent and Hazrati 1995). Although this
anatomical feature remains controversial, there is general consensus that these two
pathways have opponent effects on the SNr output neurons, via direct inhibition and
One possibility is that the direct/indirect organization implements a phase splitter. In a phase splitter, a common signal enters the circuit, and two opposite signals
are generated. One increases as the other decreases, creating antiphase signals.
Because the direct and indirect neurons share largely overlapping inputs, they are
expected to be activated simultaneously by cortical and thalamic input, as supported
by recent results (Cui et al. 2013; Isomura et al. 2013; Tecuapetla et al. 2014). There
must be concurrent activation in order to generate opponent effects on the BG output, i.e., the inhibitory effect of the direct pathway and the disinhibitory effect of the
indirect pathway on the SNr output can create antiphase signals. Because the nigral
neurons inﬂuenced by these two pathways typically show tonically high ﬁring rates,
both increases and decreases from the baseline are possible.
Opponent or antiphase signals are needed for the control of opponent downstream controllers. The reciprocal inhibition organization in the spinal cord, for
example, acts as a phase splitter of the length error signal (McDougall 1903;
Sherrington 1906). For example, to lift a dumbbell (reduce the joint angle at the
elbow) one can contract the biceps while relaxing the triceps. On the other hand,
opponent BG outputs are required to command distinct position controllers that
move parts of the body in different directions along a particular axis of motion. The
antagonism is not between antagonistic muscles, but between directions of
The question is whether these two pathways act on the same SNr output neurons.
If they reach distinct SNr populations, then the two populations of SNr neurons with
antiphase signals can be explained by the opponent inputs (inhibition and disinhibition) from the direct and indirect pathways. Another possibility is that the opponent
SNr neurons are mutually inhibitory via collaterals from their axons. This arrangement can also produce antiphase signals, without relying on a phase splitter upstream
of the SNr. These possibilities are not mutually exclusive. In addition, the “hyperdirect pathway” can also increase the output of the SNr, via glutamatergic inputs from
the subthalamic nucleus (Nambu et al. 2002), and the net effect is the opposite of
that of the direct pathway.
According to the traditional model, direct pathway activation promotes movement, whereas indirect pathway activation inhibits movement (Freeze et al. 2013;
Hikosaka et al. 2000; Kravitz et al. 2010). This interpretation, however, is inadequate in view of the relationship between neural activity and kinematics discussed
earlier. Let us consider the direct pathway ﬁrst. Striatonigral neurons are not identical. Different classes, belonging to distinct velocity controllers, are involved in
movements in different directions. For antagonistic directions (e.g., up and down)
one population must increase ﬁring as the antagonistic population stops ﬁring.
Therefore it is simplistic to conclude that the direct pathway neurons uniformly
promote movement. Some of them promote movement in some direction, but if the
antagonistic units are activated, then the movements will be stopped.
The role of the striatopallidal neurons is more difﬁcult to understand. One important clue comes from observations on turning behavior after unilateral striatal
manipulations. Increasing striatal output on one side produces contraversive turning, largely due to activation of the direct pathway, because unilateral optogenetic
direct pathway stimulation mimics the effect of nonselective unilateral striatal stimulation. By contrast, indirect pathway stimulation produces the opposite effect of
ipsiversive turning, though this effect is usually weaker (Kravitz et al. 2010;
Tecuapetla et al. 2014). This observation suggests that the effect of striatopallidal
activation is indeed the opposite of striatonigral activation, but the opposition is
between two directions of movement.
As described above, a given velocity controller is responsible for motion in one
direction only and contains an integrator in its output function. The accumulation of
signals in this integrator will therefore produce motion in one direction. To get
motion in the opposite direction, the integrator must be “discharged,” and another
antagonistic velocity controller must be activated. The discharging of the integrator
requires an error signal with the opposite sign (e.g., a leak in the bucket), which is
exactly what a phase splitter can provide. Thus, one possibility is that the direct
pathway initiates the action and the indirect pathway serves as a brake by introducing a leak in the integrator. The amount of leak can be independently controlled. The
more leak there is, the more damped the system will be. More “ballistic” actions
will involve less leak. The presence of highly plastic striatonigral axonal collaterals
that target the external globus pallidus suggests a mechanism for adjusting the
damping of the system (Cazorla et al. 2014).
There are only a few neutral postures that are speciﬁed by innate reference settings. These neutral positions often involve body symmetry, whereas movement
requires a transient break in symmetry. Most movements are transient and cyclical,
involving a return to the resting neutral position. The return back to the original
position also requires a discharging of the integrator with the opposite error. Both
the indirect and hyperdirect pathways are capable of introducing the opposite error
(or leak) to the integrator.
In the sensorimotor striatum, a given striatal module generates a set of signals
that will move some body part in a speciﬁc direction with a speciﬁc speed. Such a
module could include a set of striatonigral units and striatopallidal units that are also
activated by the same reference signal for leftward motion. As pointed out above,
20 The Basal Ganglia and Hierarchical Control in Voluntary Behavior
this module is found in the right striatum. In this module, the ﬁring rates of both
striatonigral and striatopallidal neurons are both correlated with leftward velocity,
yet their ultimate effects on the BG output, and on behavior, are different. The direct
pathway moves the body to the left, accumulating signals in the leftward integrator,
whereas the indirect pathway discharges the integrator and brings this movement to
a stop. It was proposed long ago that these pathways can scale movement amplitude
and speed (DeLong 1990), but the original proposal confounded two very different
variables and failed to take into account the key property of direction speciﬁcity.
Why should unilateral stimulation of the indirect pathway produce ipsiversive
movement? According to the present model, to turn left, the leftward module must
be turned on, and the rightward module suppressed. There is therefore antagonism
between these modules. At the same time, within the same module there is antagonism between direct and indirect pathway neurons, because the striatonigral neurons accelerates contraversive movement whereas the striatopallidal neurons
decelerates it, i.e., introduces acceleration in the ipsiversive direction. The deceleration, however, is usually dependent on the same reference command for contraversive motion, used to damp the movement as needed, and to return to the original
position. Normally the striatopallidal unit in this module cannot be activated in isolation, but selective optogenetic stimulation can reveal the acceleration in the opposite direction.
Role of Dopamine
Midbrain DA neurons project to most striatal regions. There are two major DA
pathways: the nigrostriatal pathway targets the dorsal striatum and the mesolimbic
pathway targets the ventral striatum and the prefrontal cortex. DA has the same
computational function in both pathways: it is hypothesized to adjust the gain of
striatal neurons. But the impact on behavior will differ depending on the neural
circuit affected and the types of variables being controlled. We will start by considering the nigrostriatal pathway, by far the dominant DA pathway.
DA modulates synaptic transmission. “Modulation” here does not simply mean
“change,” as used loosely in the physiological literature, but a multiplicative operation in the engineering sense. Nigrostriatal DA alters the responsiveness of dorsal
striatal neurons (Cepeda et al. 1993; Gerfen and Surmeier 2011; Hjelmstad 2004;
Yin and Lovinger 2006). It has opposite effects on striatonigral pathway neurons,
which express D1-type receptors, and striatopallidal neurons, which express D2
receptors (Gerfen and Surmeier 2011; Zhou and Lee 2011). By activating D1-like
receptors, DA increases the responsiveness of striatonigral neurons to glutamatergic
input, as well as its release of GABA at the axon terminal. By contrast, in the striatopallidal pathway, DA reduces the responsiveness of striatopallidal neurons to glutamatergic input, and reduces GABA release.
The known properties of BG synapses are in accord with the present model.
Striatonigral synapses are facilitating. Each additional presynaptic spike in a train
produces a greater response postsynaptically. This pattern suggests nigral integration of the inhibitory input. On the other hand, pallidonigral synapses are depressing, suggesting integration of the excitatory (disinhibitory) input is possible
(Connelly et al. 2010; Zhou and Lee 2011). The nigrotectal synapse appears to be
neither facilitating nor depressing (Kaneda et al. 2008), so there does not appear to
be a neural integrator in the tectum. This arrangement avoids having two integrators
in the overall control loop, which can create undesirable oscillations.
SNr GABA neurons directly inhibit DA neurons in the SNc (Tepper and Lee
2007). Tonic ﬁring of SNc DA neurons at a low rate is mediated by calcium entry
through voltage-gated calcium channels, while burst ﬁring requires glutamatergic
inputs and the activation of NMDA glutamate receptors. In addition, reduced GABA
release can also generate burst ﬁring, so that disinhibition can produce burst ﬁring
in DA neurons (Kang and Kitai 1993; Paladini et al. 1999; Tepper et al. 1995). The
DA neurons are therefore in a position to take the derivative of the GABA outputs
from the BG. This is supported by the ﬁnding that, whereas most DA neurons are
correlated with velocity and acceleration, GABA neurons are correlated with instantaneous position coordinates (Barter et al. 2015b). Due to GABAergic inhibition of
DA neurons, the derivative of the GABA output is subtracted from the output of the
DA neurons. This organization suggests a mechanism for adaptive gain control, in
which the gain can vary according to the movement velocity. Anatomical studies
have shown that nigral output can disinhibit SNc DA neurons that projection back
to the striatum, thus forming a striatonigrostriatal loop (Haber et al. 2000). It could
allow the rate of change in one controlled transition to adjust the gain for the same
controller as well as a different controller.
Dopamine Depletion and Symptoms of Parkinson’s
The hypothesis that DA serves as a gain in the velocity controller sheds light on
common symptoms in movement disorders. PD is associated with degeneration of
the nigrostriatal DA pathway. A major consequence of DA depletion is bradykinesia
or slowness in movement. Indeed, 6-OHDA, a toxin that kills DA neurons, dosedependently slows down movement (Yin 2014a). Recent work also showed that DA
depletion resulted in bradykinesia and abolished striatal representation of velocity
(Panigrahi et al. 2015).
An important property of control system is that even a large reduction in gain
will not necessarily result in system failure. To estimate the value of the controlled
variable at steady state, we can use the equation:
p = r * g / ( g + 1))
where p is the input variable controlled (e.g., movement velocity), g is the loop gain,
and r is the reference. Assuming a loop gain of 100, then the controlled input