The Thalamus

Where the thalamus sits

The thalamus occupies the geometric heart of the brain. It is the largest component of the diencephalon, the part of the forebrain that lies between the two cerebral hemispheres and the midbrain. There are two thalami, one on each side, and they sit like a pair of eggs flanking the narrow slit of the third ventricle. In many people the two are joined across the midline by a small bridge of tissue, the interthalamic adhesion, though this is not a functional connection and it is absent in a substantial minority of brains.

The neighbours matter. Below the thalamus lies the hypothalamus, separated from it by a shallow groove. Behind and below lies the midbrain. Laterally, the thalamus is bounded by the internal capsule, the great sheet of fibres carrying traffic between the cortex and everything beneath, and beyond that the basal ganglia. This position, wedged between the sensory pathways rising from below and the cortex spreading above, is not an accident of packing. It is a consequence of what the thalamus does: it is placed exactly where every ascending route must pass.

Wrapped around the outer surface of each thalamus, like a thin shell, is the thalamic reticular nucleus. It is worth flagging now because it is the exception to almost every rule that follows: unlike the other nuclei, it does not project to the cortex at all. It is inhibitory, and it points inward, at the thalamus itself. Its role becomes clear in the section on gating.

The nuclei: a map

The thalamus is not one thing; it is dozens of things in a shared package. Each nucleus receives a defined input and projects to a defined patch of cortex, and it is by their nuclei that thalami are understood. A sheet of white matter shaped like a Y, the internal medullary lamina, divides each thalamus into anterior, medial, and lateral groups, with additional nuclei sitting inside the lamina itself.

Six nuclei do most of the work you need to know about.

Two further groups complete the picture. The ventral lateral and ventral anterior nuclei are the motor relays: they receive input from the cerebellum and the basal ganglia and project to the motor and premotor cortex, which is how those two subcortical systems influence voluntary movement. The anterior nuclei are part of the limbic circuitry, connected via the mammillary bodies and the fornix to the hippocampus and projecting to the cingulate gyrus, with a role in memory. And the intralaminar nuclei, scattered within the internal medullary lamina, receive input from the brainstem reticular formation and project diffusely across the cortex; they are central to arousal and to maintaining consciousness.

Relay: the sensory gateway

The single most repeated fact about the thalamus is true and worth stating precisely. With one exception, all sensory information travelling to the cerebral cortex passes through a thalamic nucleus first. Light striking the retina generates signals that reach the LGN before the visual cortex. Sound reaching the cochlea travels a long route through the brainstem, but its last stop before the cortex is the MGN. Touch on the hand reaches the VPL before the parietal lobe. Taste reaches the VPM.

The exception is smell. Olfactory signals leave the olfactory bulb and travel directly to the olfactory cortex on the medial surface of the temporal lobe, bypassing the thalamus entirely. This is often presented as a curiosity, but it has consequences: olfaction has an unusually direct line to the amygdala and hippocampus, which is one reason smells can evoke memory and emotion with a vividness and immediacy that other senses rarely match.

Relay nuclei are organised so that the map of the sensory surface is preserved. Neighbouring points on the retina project to neighbouring points in the LGN; neighbouring frequencies in the cochlea project to neighbouring points in the MGN; neighbouring patches of skin project to neighbouring points in the VPL. The cortex inherits these maps from the thalamus. The orderly cortical maps that appear so striking in the somatosensory and visual cortex are, in this sense, thalamic maps passed upward.

Gate: why relay is the wrong word

If the thalamus were merely a relay, it would be an odd design. A relay simply passes a signal on, and a system that only needed to pass signals on would not need a large, cellularly complex, energetically expensive structure at the centre of the brain. Two anatomical facts show that something more is going on.

The first is the ratio of connections. A thalamic relay nucleus does not receive most of its input from the sense organ it serves. In the LGN, retinal fibres, the ones actually carrying visual information, account for only a minority of the synapses. The majority arrive from the cortex, from the brainstem, and from the thalamic reticular nucleus. In other words, most of what the visual relay is listening to is not the eye. It is the rest of the brain, telling it how to treat what the eye is saying.

The second is the reticular nucleus. This inhibitory shell of GABA-releasing neurons receives copies of both the ascending signals passing through the thalamus and the descending signals coming back from the cortex, and it sends inhibition into the relay nuclei. It is, in effect, a control surface: it can dampen the flow through one nucleus while leaving another open. This is the physical machinery of a gate.

The consequence is that the thalamus does not treat all incoming signals equally. It weights them, and the weighting depends on the state of the rest of the brain. When you attend to a sound and stop noticing the room around you, part of what has happened is that the thalamic gate has been adjusted. When you fall asleep and stop registering the traffic outside, the gate has largely closed: relay neurons switch into a rhythmic bursting mode in which they no longer faithfully transmit the details of incoming sensation. This is not a metaphor. It is a measurable change in how thalamic neurons fire.

Thalamocortical loops

Every relay nucleus of the thalamus sends fibres up to its target area of cortex, and every one of those cortical areas sends fibres back down. The connection is reciprocal, and the return path is not a token. Corticothalamic fibres outnumber thalamocortical fibres by a large margin, which means the traffic is predominantly downward. Whatever the cortex is doing with these fibres, it is doing a great deal of it.

Circuits of this kind, in which a thalamic nucleus and a cortical area are wired into a loop, are called thalamocortical loops, and they are the fundamental architecture of the forebrain. They provide a mechanism for the cortex to shape its own input: an area that has begun to interpret an ambiguous signal one way can bias the thalamic relay to favour that interpretation. They also provide a substrate for oscillation. A loop with a delay in it will tend to resonate, and thalamocortical loops resonate at characteristic frequencies. The rhythms picked up by EEG, including the alpha rhythm of quiet wakefulness and the sleep spindles of light sleep, are generated largely by thalamocortical circuitry with the reticular nucleus as pacemaker.

Motor loops follow the same principle in the other direction. The basal ganglia and the cerebellum both project into the ventral thalamic nuclei, which project to the motor cortex, which projects back down to both. Movement is therefore not commanded by the cortex alone but decided in loops that pass through the thalamus, which is why the thalamus appears in any honest account of motor control despite not being a motor structure in the ordinary sense.

Arousal, sleep, and consciousness

The intralaminar nuclei sit apart from the specific relays. They receive input from the reticular formation of the brainstem, the diffuse network that governs wakefulness, and they project widely and non-specifically across the cortex. They are, in effect, a broadcast system, and they are the route by which brainstem arousal reaches the cortex.

This gives the thalamus a role in consciousness that goes beyond relaying content. It is not enough for information to reach the cortex; the cortex must be in a state to make use of it. That state is set, in large part, by thalamic activity. In deep sleep, thalamocortical neurons enter a bursting mode that decouples the cortex from the outside world. In anaesthesia, thalamic activity is markedly suppressed, and several lines of evidence point to the thalamus as a critical node in the loss of consciousness under general anaesthetic.

The clinical evidence is stark. Bilateral damage to the intralaminar and medial thalamic nuclei, as can occur when a stroke affects the artery supplying both thalami, can produce a profound and lasting disorder of consciousness, even though the cortex itself is intact. A cortex without its thalamus is not a conscious cortex.

For more on how sleep is produced and what it does, see sleep and the brain.

When the thalamus is damaged

Because the thalamus packs many functionally distinct nuclei into a small volume, the consequences of damage depend acutely on exactly where the damage falls. A lesion of a few millimetres can produce a syndrome; a lesion a few millimetres away can produce a completely different one.

Thalamic stroke. The thalamus is supplied by small perforating branches of the posterior cerebral artery, and blockage of one of these produces a lacunar infarct. If the VPL and VPM are affected, the result is a pure sensory stroke: loss of sensation over the whole opposite side of the body, with motor power preserved, because the motor fibres run elsewhere. If the lesion extends laterally into the internal capsule, weakness is added. If the medial and intralaminar nuclei are involved, the picture is dominated instead by impaired arousal, attention, and memory, and can be mistaken for a psychiatric or dementing illness.

Thalamic pain syndrome. One of the most striking consequences of thalamic damage, sometimes called Dejerine-Roussy syndrome after the neurologists who described it, is a delayed and persistent pain. Weeks or months after a sensory thalamic stroke, the affected side of the body, initially numb, begins to hurt: a burning, aching, poorly localised pain, often accompanied by allodynia, in which an ordinary light touch is experienced as painful. It is difficult to treat and is a paradigm case of central neuropathic pain, pain generated by damage to the pain-processing system itself rather than by any injury at the site where the pain is felt.

Memory. Damage to the anterior and mediodorsal nuclei, and to the mammillothalamic tract that feeds them, impairs the formation of new declarative memories. This is a central feature of Korsakoff's syndrome, seen in severe chronic thiamine deficiency, and it makes an important theoretical point: the memory circuit does not begin and end in the hippocampus. It runs through the thalamus, and it can be broken there.

Related topics