How the Epithalamus Regulates Sleep, Melatonin, and Mood

The epithalamus is a small segment of the brain perched on top of the thalamus, and despite its size it houses structures with outsized influence on sleep, mood, pain, and motivation. It consists mainly of two components: the pineal gland, which produces melatonin, and the habenular nuclei, a pair of tiny relay stations that funnel signals from the forebrain down into the brain’s major mood-regulating chemical systems. A few accessory structures round out the region, including the stria medullaris (the fiber highway that feeds information into the habenula) and the posterior commissure (a bundle of fibers crossing the midline just behind the pineal gland). Researchers have increasingly focused on the epithalamus over the past two decades, and the findings connect it to an unexpectedly broad range of functions, from nicotine addiction to treatment-resistant depression.

What Makes Up the Epithalamus

The epithalamus develops from the most dorsal (uppermost) part of a region called prosomere 2 in the embryonic brain, though the genetic programs that guide its formation are still being worked out.¹ In the mature brain, the pineal gland sits at the back end of this zone, tucked between the two cerebral hemispheres. The habenular nuclei sit just in front of the pineal stalk, split into a medial habenula and a lateral habenula on each side. These two divisions look similar under a basic microscope but have sharply different chemical profiles and connect to different downstream circuits.

Feeding into the habenula is the stria medullaris, a discrete white-matter tract that carries signals from forebrain regions involved in decision-making and emotion.² The stria medullaris terminates at the triangular eminence of the pineal stalk, where the habenula resides.³ From the habenula, outputs travel via the fasciculus retroflexus down into the midbrain, where they influence the production of serotonin and dopamine. This arrangement makes the epithalamus a critical switchboard: information about goals, expectations, and threats flows in from the forebrain, and the habenula translates it into chemical signals that adjust mood, motivation, and arousal.

The Pineal Gland and Melatonin

The pineal gland is the epithalamus’s best-known resident, largely because it produces melatonin, the hormone that helps time your sleep-wake cycle. The gland does not generate its own rhythm independently. Instead, it receives instructions from the body’s master clock in the hypothalamus, which tracks light exposure via the eyes. When darkness falls, sympathetic nerve endings in the pineal release norepinephrine, which activates receptors on the gland’s cells. Those receptors trigger an enzyme that converts serotonin into the melatonin precursor N-acetylserotonin, which is then converted into melatonin itself.⁴ Melatonin secretion peaks at night and drops during the day, giving the body an internal darkness signal used to coordinate everything from sleep onset to seasonal reproductive timing in other mammals.

One quirk of the pineal gland is that it sits outside the blood-brain barrier. Most brain tissue is shielded from circulating blood proteins by tightly sealed capillaries, but the pineal’s capillaries are fenestrated, meaning they have small pores that allow proteins to pass through freely.⁵ This makes the pineal one of the brain’s circumventricular organs, structures that act as portals between the central nervous system and the rest of the body.⁶ The open blood supply presumably helps melatonin get released quickly into the bloodstream and may also allow the pineal to sense circulating signals that other brain regions cannot detect directly.

Pineal Calcification and Aging

If you have ever seen a CT scan of an older adult’s brain, you may have noticed a small bright dot near the center. That is calcified pineal tissue, sometimes called “brain sand.” Pineal calcification is extremely common, increasing with age, and for decades was considered just an incidental finding. More recent work, however, suggests that the process is not entirely passive. One hypothesis is that pineal calcification resembles bone formation in some respects, involving mesenchymal stem cells and possibly melatonin itself in an active biological program.⁷ A 2025 preprint went further, reporting that the calcification process redeploys molecular pathways normally used for skeletal bone formation, including Hedgehog and Notch signaling, suggesting it is a regulated, lineage-specific program rather than simple mineral buildup.⁸

Whether heavy calcification meaningfully reduces melatonin output in older adults is still debated. Melatonin levels decline with age regardless, and separating the effect of calcification from the general aging of the gland is difficult. But the finding that calcification is actively regulated rather than degenerative opens up the question of whether it could eventually be slowed or prevented, something that would matter for sleep quality in aging populations.

The Habenula and Mood

While the pineal gland handles circadian chemistry, the habenular nuclei handle something arguably more complex: evaluating whether the world is going well or badly, and adjusting your brain’s chemical environment accordingly. The lateral habenula in particular has become one of the most studied structures in depression research. It sits at the intersection of circuits governing reward, aversion, and stress, receiving inputs from cortical and limbic forebrain regions through the stria medullaris and projecting outputs to brainstem nuclei that produce serotonin and dopamine.⁹

When something disappointing or aversive happens, the lateral habenula fires more intensely, suppressing dopamine and serotonin release. This is a normal, healthy response; it is how the brain updates its expectations. But in depression, the lateral habenula appears to become hyperactive, firing in bursts even in the absence of negative events. Research in mice showed that in animals subjected to chronic stress, the proportion of burst-firing neurons in the lateral habenula was significantly higher than in unstressed mice. A single injection of ketamine suppressed this burst firing within minutes, and the suppression persisted for up to 24 hours, paralleling the drug’s rapid antidepressant effects.¹⁰ This finding helped explain why ketamine works so quickly compared to traditional antidepressants, which take weeks to alter mood: it directly quiets the habenular activity pattern that may be driving the depressive state.

Deep Brain Stimulation for Treatment-Resistant Depression

The habenula’s role in depression has led to a natural therapeutic question: could directly stimulating or inhibiting it help patients who do not respond to standard treatments? In rats exposed to chronic stress, deep brain stimulation of the lateral habenula reduced depression-like behaviors and reversed the region’s hyperactivity.¹¹ The idea of targeting the habenula specifically was proposed as a formal hypothesis over 15 years ago, based on the structure’s known anatomy and early imaging evidence.¹²

In a recent clinical trial, six patients with treatment-resistant depression received bilateral electrode implants in the habenula and were given high-frequency stimulation. Compared to baseline, their depression scores dropped substantially: roughly 62% at one month, 64% at three months, and 66% at six months. Imaging showed that the stimulation modulated activity not just in the habenula but also in the medial orbitofrontal cortex, the raphe nuclei, and the substantia nigra, all regions involved in mood and reward processing.¹³ Six patients is a very small sample, and larger trials are needed, but the results are striking enough that the habenula is now firmly on the map as a potential deep brain stimulation target for depression that resists medication.

Nicotine Addiction and the Medial Habenula

The medial habenula, the lateral habenula’s smaller neighbor, has a different specialty: it is densely packed with nicotinic acetylcholine receptors, the same receptors that nicotine hijacks to produce its rewarding and addictive effects. The medial habenula projects to the interpeduncular nucleus in the midbrain, forming a circuit now recognized as critical for nicotine dependence. In mice, the nicotinic receptor subunits expressed in this pathway are necessary for the appearance of nicotine withdrawal symptoms, and blocking nicotinic activity in the medial habenula alone is enough to trigger withdrawal in nicotine-dependent animals.¹⁴

The medial habenula’s cholinergic neurons fire spontaneously at a steady pace, driven by pacemaker channels. Acute exposure to nicotine speeds up this firing. After chronic nicotine use and withdrawal, reexposure to nicotine doubles the pacemaking frequency of these neurons, which may explain the intense relief smokers feel when they relapse.¹⁵ Genetic variants in the nicotinic receptor gene cluster CHRNA5-CHRNA3-CHRNB4, which is heavily expressed in this circuit, have been linked to heavier smoking and higher relapse risk in humans, giving the medial habenula a clear place in the biology of tobacco addiction.

More recent work has added complexity. The excitability of medial habenula neurons is modulated differently depending on nicotine dose, the presence of menthol (a common cigarette additive), and the sex of the animal. In male mice exposed to high-dose nicotine with menthol, neurons showed strong correlations between excitability and drug-seeking behavior, while female mice showed different patterns at the same doses.¹⁶ These sex differences in habenula physiology may eventually help explain why men and women respond differently to smoking cessation treatments, though that link has not been clinched yet.

Pain and Aversion Processing

The lateral habenula does not just handle disappointment and depression. It also appears to play an integrative role in pain, connecting the sensory experience of pain with its emotional fallout, including anxiety and avoidance behavior. Anatomically, this makes sense: the habenula receives input from forebrain structures and sends projections to brainstem pain-modulation centers like the periaqueductal gray and the raphe nuclei.¹⁷

In a mouse model of trigeminal neuralgia (a condition that causes severe facial pain), researchers found that inhibitory GABAergic neurons in the lateral habenula became markedly less active. When these neurons were artificially reactivated using optogenetic or chemogenetic techniques, both pain behaviors and anxiety-like behaviors were reduced. An inhibitory pathway from the lateral habenula’s GABAergic neurons to the posterior paraventricular thalamus primarily alleviated pain-related behaviors, with less effect on anxiety, suggesting that the habenula’s contribution to pain and its contribution to pain-related anxiety may run through partially separable circuits.¹⁸ This kind of dissection matters clinically, because a future treatment that could target pain-specific habenula output without blunting emotional processing more broadly would be far more useful than one that suppresses the whole region.

Sleep and the Lateral Habenula

Beyond its better-known roles in mood and motivation, the lateral habenula also regulates sleep architecture. Functional manipulation of its glutamatergic neurons bidirectionally modulated both non-REM and REM sleep in mice. When researchers recorded activity across sleep-wake cycles, they found that lateral habenula neurons were preferentially active during REM sleep. A subpopulation of REM-active neurons, when selectively stimulated, specifically promoted REM sleep via projections to the ventral tegmental area. In a depression mouse model, these REM-active neurons were potentiated: a larger fraction of them were active, they fired more, and their activity patterns shifted. Inhibiting them not only reduced the excess REM sleep seen in depression but also alleviated depressive-like behaviors.¹⁹

This is a useful piece of the puzzle because disrupted sleep, particularly excess REM sleep early in the night, is one of the hallmark features of major depression. The finding that the same habenular neurons driving mood disturbance also drive sleep disturbance provides a mechanistic link between the two symptoms, and suggests that a treatment targeting habenular hyperactivity could address both simultaneously.

Asymmetry in the Habenula

The habenula has an unusual property among brain structures: in many vertebrates, the left and right habenulae are strikingly asymmetric. In zebrafish, which have been widely used to study this phenomenon, the left and right dorsal habenulae are wired differently. Calcium imaging in young zebrafish showed that most neurons responding to a light stimulus were located on the left side, while most neurons responding to an odor were on the right, with very few neurons responding to both.²⁰ This lateralization appears to let the two habenulae divide labor, processing different sensory modalities in parallel.

Whether this kind of asymmetry holds in humans is less clear. A meta-analysis pooling data from over 50 datasets found an average left-right volume difference of just 0.41 cubic millimeters, which was not statistically significant.²¹ Individual studies have reported left-larger or right-larger findings, but they do not converge on a consistent pattern. The habenula’s small size, roughly 30 to 35 cubic millimeters per side, makes it difficult to measure accurately on standard MRI, and part of the challenge in human neuroimaging has simply been reliably locating the structure amid surrounding tissue.²² The functional asymmetry seen so clearly in fish may be present in subtler form in mammals, but proving it will require better imaging tools or more creative experimental designs.

The Subcommissural Organ and Cerebrospinal Fluid

Tucked near the back of the epithalamus, at the entrance to the cerebral aqueduct, is the subcommissural organ, a small secretory gland that produces glycoproteins. These glycoproteins are released into the cerebrospinal fluid, where they polymerize into a long thread-like structure called Reissner’s fiber that extends through the aqueduct, the fourth ventricle, and down the central canal of the spinal cord.²³ Knocking out this system immunologically in animal models led to hydrocephalus, an accumulation of fluid in the brain, suggesting the fiber plays a role in cerebrospinal fluid dynamics or drainage.

The story has some unexpected twists, though. In zebrafish embryos lacking the protein that forms Reissner’s fiber, direct measurements of cerebrospinal fluid flow through the central canal showed no change in velocity or net flow. Instead, the embryos developed abnormal spinal curvature, pointing to a role for the fiber in body-axis straightening during development rather than in fluid circulation per se.²⁴ The subcommissural organ remains one of the least understood structures in the brain, and these conflicting results across species only add to the mystery.

The Evolutionary “Third Eye”

The pineal gland’s evolutionary backstory is stranger than its current job description suggests. In many lizards and the tuatara, a photoreceptive structure called the parietal eye, or “third eye,” sits on top of the skull with a simplified retina capable of detecting light. This structure develops from the pineal complex and expresses a suite of nonvisual opsins, including pinopsin, parapinopsin, and parietopsin.²⁵ A newly discovered opsin, dubbed “lepidopsin,” was found to exist solely in the genomes of the tuatara and most lizard groups but is absent from all other vertebrates. Lizards that have lost the parietal eye, such as geckos, have also lost most or all of these pineal opsin genes, confirming a tight evolutionary link between the eye structure and its molecular toolkit.

The parietal eye was far more widespread in the deep past. Fossil evidence shows that the parietal foramen, the opening in the skull that housed it, was present across both major amniote lineages during the late Paleozoic and Triassic periods. Its size varied across time, and the trends suggest it may have been tied to thermoregulation. The foramen disappeared completely in lineages that evolved higher metabolisms and possible warm-bloodedness, including the ancestors of mammals and archosaurs, while it grew dramatically larger in some marine reptiles and dicynodonts.²⁶ In at least one case, evolution went even further: a fossil monitor lizard, Saniwa ensidens, had both a pineal eye and a parapineal eye present simultaneously, giving it four eyes total, the only known jawed vertebrate with that arrangement.²⁷

In mammals, the pineal gland has lost its photoreceptive capacity entirely and retreated to a purely endocrine role, buried deep in the brain with no direct light access. But its evolutionary origin as a light-sensing organ explains why it remains so tightly linked to the circadian system and why it still uses serotonin, a neurotransmitter with ancient photoreceptor roots, as the raw material for melatonin synthesis.

Why the Epithalamus Has Been Hard to Study

For most of the history of neuroscience, the epithalamus was largely ignored, and there are practical reasons for that. The habenula is tiny, smaller than a pea in humans, and it sits deep in the brain surrounded by much larger structures. Standard functional MRI has a resolution too coarse to cleanly isolate habenula signals from neighboring tissue, and guidelines for reliably locating and tracing the habenula in neuroimaging were only published within the past decade or so.²⁸ The pineal gland, meanwhile, was long dismissed as a vestigial structure, an impression reinforced by its tendency to calcify with age.

The surge of interest over the past 15 years has been driven largely by animal work showing the habenula’s outsize influence over dopamine and serotonin, two of the most therapeutically important neurotransmitter systems. As higher-resolution imaging and more precise stimulation techniques become available, translating those animal findings to humans is becoming increasingly feasible. The early clinical results with deep brain stimulation of the habenula in treatment-resistant depression are one sign that this translation is already underway.