The Basal Ganglia

The components

The basal ganglia are not a single organ but a functional family of nuclei, and the membership list has varied over the history of anatomy. The modern list, defined by the circuit rather than by proximity, has five members.

Two features of this list are worth pausing on. First, the substantia nigra is in the midbrain, not the cerebrum, yet it is functionally inseparable from the rest; the basal ganglia are defined by their circuit, not by an anatomical boundary. Second, almost every connection within the system is inhibitory, using the neurotransmitter GABA. A circuit built mostly of inhibition, with one excitatory nucleus in it, is a circuit built to suppress by default and to release selectively. Keep that in mind: it is the key to everything below.

The basic loop

Information enters and leaves the basal ganglia in a loop, and the shape of the loop is the same whichever function is being served.

1. Cortex to striatum

   Excitatory glutamate fibres descend from almost every area of the cortex onto the striatum. This is the input, and it is massive: the striatum hears from the whole cortex.

2. Through the basal ganglia

   The striatum processes this input through the direct and indirect pathways described below, involving the globus pallidus, the subthalamic nucleus, and the substantia nigra.

3. Output to the thalamus

   The GPi and the substantia nigra pars reticulata send inhibitory GABA fibres to the thalamus, specifically the ventral anterior and ventral lateral nuclei. Crucially, these output neurons fire tonically, all the time, so the thalamus is held down by default.

4. Thalamus back to cortex

   The thalamus, when released, excites the motor and premotor cortex, which then issues the command. The loop is closed.

The critical insight is in step three. The default state of the basal ganglia output is on: continuous inhibition of the thalamus, which means continuous suppression of movement. To move, you do not send a "go" signal through the basal ganglia. You remove the brake. Movement is released by disinhibition, and this is why the whole system can be understood as a gate rather than an engine.

The same loop architecture is repeated in parallel channels. A motor loop passes through the putamen and serves movement. An oculomotor loop serves eye movement. An associative loop through the caudate serves cognition and connects with the prefrontal cortex. A limbic loop through the ventral striatum serves motivation and reward. The circuit is one design applied to several domains, and this is why basal ganglia disorders can affect thought and mood as well as movement.

The direct and indirect pathways

Within the loop, the striatum splits its output into two routes with opposite effects. Tracing the sign of each connection is fiddly, but it is worth doing once carefully, because everything about Parkinson's and Huntington's disease follows from it.

Both pathways start in the striatum but they start in different cells. Direct-pathway neurons carry D1 dopamine receptors; indirect-pathway neurons carry D2 receptors. The two receptor types respond to dopamine in opposite ways, and this is the single anatomical fact that makes dopamine so powerful in this system.

A third route, the hyperdirect pathway, runs straight from the cortex to the subthalamic nucleus, bypassing the striatum entirely. Because the STN excites the output nuclei, this route applies a rapid, broad brake. It is thought to be the mechanism by which an action already under way can be aborted at short notice, the neural basis of stopping yourself mid-reach when you realise the cup is hot.

What dopamine does here

The substantia nigra pars compacta sends a dense projection of dopamine fibres to the striatum, called the nigrostriatal pathway. Dopamine here is not a message in the ordinary sense; it is a modulator, changing how the striatal neurons respond to the cortical input they are receiving.

Its effect is opposite on the two pathways, and this is the crux. Acting on D1 receptors, dopamine excites the direct pathway, strengthening the "yes" signal. Acting on D2 receptors, it inhibits the indirect pathway, weakening the "no" signal. Both effects push the same way: dopamine favours the release of movement. Take dopamine away, and both effects reverse. The "yes" weakens, the "no" strengthens, and the whole system tips toward suppression.

Dopamine has a second, slower role here, in learning. When an outcome is better than expected, midbrain dopamine neurons fire a burst; when it is worse than expected, they pause. This signal, a reward prediction error, arrives at exactly the striatal synapses that were active when the action was chosen, and it strengthens or weakens them accordingly. Over many repetitions the striatum learns which actions, in which contexts, tend to pay off. See dopamine and reward for the fuller account.

Movement as selection

The most useful way to think about the basal ganglia is as a selection mechanism operating on a set of competing candidates. At any instant, many cortical circuits are proposing actions. The parallel loops of the basal ganglia receive these proposals in the striatum. Through the direct pathway, the winning proposal has its brake released. Through the indirect and hyperdirect pathways, the losing proposals have their brakes tightened. The result is a single action, cleanly executed, with its competitors suppressed.

This account explains several things at once. It explains why the basal ganglia are not needed to move: patients with basal ganglia damage can still move, they just cannot select well. It explains why their disorders come in two flavours: too little release (hypokinetic, as in Parkinson's) or too little suppression (hyperkinetic, as in Huntington's, chorea, tics, and hemiballismus). And it explains why a system built for selecting movements is also used for selecting thoughts and habits, since the same problem, choosing one from many, arises in every domain.

The basal ganglia work alongside two other systems in motor control, and the division of labour is clean. The cortex plans and commands. The cerebellum, described on the brainstem and cerebellum page, corrects and coordinates, comparing intended movement with actual movement. The basal ganglia select and gate. Damage each and you get three quite different disorders: paralysis, clumsiness, and difficulty starting or stopping.

Habits, skills, and reward

A skill you have practised until it is automatic, typing, driving a familiar route, playing a scale, is not held in the same memory system as the fact that Paris is the capital of France. Declarative memory of that kind depends on the hippocampus. Procedural memory, the memory of how to do things, depends heavily on the basal ganglia, and particularly on the striatum.

The evidence for the separation is unusually clean. Patients with severe hippocampal damage who cannot form any new conscious memories can nevertheless learn new motor skills across days of practice, improving steadily while insisting each session that they have never done the task before. Conversely, patients with striatal degeneration lose the ability to acquire such habits while their conscious memory remains intact. Two memory systems, two anatomies, dissociable by lesion. For the wider map, see types of memory.

The reward function follows from the same machinery. The ventral striatum, including the nucleus accumbens, receives dopamine from the ventral tegmental area, the substantia nigra's medial neighbour, and forms the limbic loop with the prefrontal cortex and the amygdala. This is the circuit that assigns value to outcomes and shapes which behaviours are repeated. It is also the circuit that addictive drugs hijack: nearly every drug of abuse, by one route or another, raises dopamine in the ventral striatum, and in doing so it borrows a learning mechanism that evolved to make food and safety worth pursuing.

Parkinson's disease

Parkinson's disease is, at its core, a disease of one small population of cells: the dopamine-producing neurons of the substantia nigra pars compacta. They degenerate progressively, and as they die the nigrostriatal supply of dopamine to the striatum fails.

Follow the circuit and the symptoms follow. Without dopamine, the D1-driven direct pathway is under-excited, so movements are not released; the D2-suppressed indirect pathway is under-inhibited, so movements are over-suppressed. The output nuclei fire more, the thalamus is held down harder, the motor cortex is less driven. Clinically that appears as bradykinesia, slowness of movement; rigidity, stiffness of the limbs; a characteristic resting tremor; difficulty initiating movement; and a general poverty of spontaneous motion including reduced facial expression and diminished arm swing.

Two features of the disease are worth emphasising. The first is that symptoms appear late: a substantial fraction of the nigral dopamine neurons is already lost by the time the first motor sign is noticed, because the striatum compensates for a considerable time. The disease is well established before it announces itself. The second is that Parkinson's is not purely a movement disorder; the same dopamine loss and the same wider pathology affect mood, sleep, olfaction, and cognition, and non-motor symptoms often precede the tremor by years.

Treatment follows the anatomy. Levodopa, a precursor that the surviving neurons convert to dopamine, replaces what is missing and is strikingly effective, at least at first. Deep brain stimulation, in which electrodes are placed in the subthalamic nucleus or the internal globus pallidus, works by damping the overactive suppression at its source, and the fact that it works is one of the strongest confirmations of the circuit model.

Huntington's disease

Huntington's disease is the mirror image, and its logic is just as clean. It is an inherited condition, caused by an expanded repeat in a single gene, and the neurons it destroys first are the medium spiny neurons of the striatum, with the indirect-pathway population particularly vulnerable early on.

Lose the indirect pathway and you lose the "no". Suppression fails. The output nuclei, no longer driven by the subthalamic route, inhibit the thalamus less, and the thalamus over-drives the cortex. The result is chorea: continuous, involuntary, flowing, dance-like movements that the patient cannot suppress. The disease progresses to involve the direct pathway and the cortex as well, so the movement disorder is eventually joined by cognitive decline and psychiatric symptoms.

The same principle explains hemiballismus, a dramatic disorder in which a stroke damages the subthalamic nucleus on one side and the patient develops violent flinging movements of the opposite arm and leg. Destroy the nucleus that amplifies suppression and suppression collapses. Anatomy predicts the symptom.

Related topics