A Brain Circuit Hidden Behind the Eyes Could Be the Switch for Dementia

Scientists tracking pupil movements during brain scans have discovered that a tiny cluster of neurons behind your eyes starts degenerating decades before Alzheimer’s symptoms appear.

The locus coeruleus, a brainstem region smaller than a grain of rice, shows tau protein accumulation years ahead of memory centers like the hippocampus, making it the first brain structure to develop Alzheimer’s pathology.

Research published in early 2025 using simultaneous eye-tracking and functional MRI reveals that spontaneous pupil dilation and constriction directly reflect locus coeruleus activity.

The connectivity between this minuscule region and memory networks decreases with age, correlating with tau biomarkers in cerebrospinal fluid and declining performance on memory tests.

The locus coeruleus produces norepinephrine—a neurotransmitter regulating attention, arousal, and memory formation—and projects to virtually every part of the brain.

When these neurons die, the entire brain loses its chemical coordinator, setting off cascading dysfunction across multiple networks.

In individuals over 57 years old, functional connectivity within the pupil diameter network significantly decreases compared to younger adults.

The correlation between total tau levels and reduced connectivity in frontal brain regions suggests this degeneration happens silently for years, detectable through eye measurements long before cognitive tests show impairment.

Your Eyes Reveal Brain Deterioration You Can’t Feel

Researchers analyzed 76 cognitively normal adults, tracking spontaneous pupil changes during resting brain scans.

They identified 20 distinct brain clusters functionally connected to pupil dynamics, including the locus coeruleus buried deep in the brainstem where direct imaging remains technically challenging.

Pupil diameter positively correlated with the salience network—brain regions processing important stimuli and switching attention between internal thoughts and external demands.

The central executive network, controlling goal-directed behavior and working memory, showed negative correlation with pupil size.

This inverse relationship makes biological sense. When your brain focuses inward during rest, the locus coeruleus reduces activity, pupils constrict slightly, and executive networks deactivate.

During alert external attention, locus coeruleus neurons fire rapidly, pupils dilate, and executive regions engage.

But here’s what nobody expected to find: the rate of pupil size change, not just the diameter itself, reveals different neural dynamics entirely.

Analysis of pupil dilation speed identified a separate network of brain regions, and crucially, the locus coeruleus didn’t appear in this network at all.

The Assumption That Pupil Changes Always Mean One Thing

We’ve long assumed pupil dilation serves as a direct readout of locus coeruleus activity—bigger pupils equal more norepinephrine release. Pharmaceutical studies, attention research, and arousal experiments all rely on this premise.

That assumption crumbles under scrutiny.

The first derivative of pupil diameter—how fast it changes rather than its absolute size—activated visual cortex, sensorimotor regions, and frontal areas, but not the locus coeruleus itself.

Eleven brain clusters showed significantly reduced correlation with pupil change rate in older versus younger adults, particularly in medial temporal regions near the fusiform gyrus critical for face recognition.

This dissociation reveals that tonic arousal (steady pupil size) reflects locus coeruleus baseline activity, while phasic changes (rapid fluctuations) involve multiple other brainstem structures—the dorsal raphe nucleus, pretectal olivary nucleus, superior colliculus, and hypothalamus all contribute to moment-to-moment pupil dynamics.

The hypothalamus particularly matters. Orexin neurons there can drive pupil dilation independently, though effects often transmit through the locus coeruleus.

When older brains show disconnection between pupil rate changes and temporal lobe regions, it suggests these auxiliary control systems begin failing alongside the primary locus coeruleus degeneration.

Why This Tiny Region Matters More Than Larger Memory Centers

The locus coeruleus contains only about 30,000 neurons per side of the brain—microscopic compared to billions elsewhere. Yet those neurons project everywhere, releasing norepinephrine that modulates how other brain regions process information.

Think of it as the brain’s volume knob rather than the content itself.

Turn down locus coeruleus activity and every cognitive process—memory encoding, attention, emotional regulation, decision-making—operates at diminished capacity regardless of whether downstream structures remain intact.

Tau pathology appears in the locus coeruleus during Braak Stage 1 of Alzheimer’s progression, preceding both amyloid deposits and tau accumulation in memory centers by potentially decades.

Hyperphosphorylated tau collects in these norepinephrine-producing neurons years before cognitive symptoms emerge, positioning locus coeruleus degeneration as potentially the earliest detectable change in Alzheimer’s disease.

Animal studies demonstrate this causality directly. Experimentally damaging locus coeruleus neurons in rats accelerates tau pathology throughout the forebrain and exacerbates cognitive deficits beyond what occurs from pathology alone.

The norepinephrine these neurons provide doesn’t just support function—it actively protects other brain regions from degeneration.

The Salience Network Connection

Brain regions comprising the salience network—anterior cingulate cortex, insula, supramarginal gyrus—showed the strongest positive correlation with pupil diameter in the 2025 study.

The salience network acts as a switching station, determining whether you focus internally on memories and plans or externally on immediate surroundings.

This network identifies what’s important moment to moment and reallocates attention accordingly.

A loud noise, sudden movement, or emotionally significant memory can trigger salience network activation, which then suppresses the default mode network (internal focus) and engages the central executive network (external processing).

The locus coeruleus provides critical modulation for this switching function. Norepinephrine release sharpens the signal-to-noise ratio across cortical processing, making important stimuli stand out while background noise fades.

Without adequate locus coeruleus activity, the salience network can’t efficiently redirect attention, leaving people stuck in whatever cognitive mode they occupied.

Reduced connectivity between locus coeruleus and salience network explains early Alzheimer’s symptoms that precede obvious memory loss—decreased attentional control, difficulty switching between tasks, mental inflexibility, and reduced awareness of social cues. These subtle changes often get dismissed as normal aging when they actually reflect locus coeruleus dysfunction disrupting the salience network’s coordination role.

Memory, Executive Function, and Visuospatial Abilities All Depend on This Circuit

Eleven brain clusters showed significant correlations between pupil diameter connectivity and neuropsychological test performance. Every correlation involved memory (Free and Cued Selective Reminding Test), executive function (Trail Making Test, Letter Fluency), or visuospatial ability (Rey Complex Figure Test, Judgment of Line Orientation).

The anterior supramarginal gyrus and superior frontal gyrus connectivity specifically correlated with free recall performance—retrieving information without cues. These regions need adequate norepinephrine signaling to maintain working memory capacity and access stored information efficiently.

The proposed mechanism works through network regulation. The locus coeruleus activates in response to salient information flagged by the salience network. This activation then modulates the frontoparietal network (central executive system), which normally suppresses the default mode network during external tasks requiring focused attention.

When locus coeruleus function declines, the frontoparietal network fails to adequately suppress internally-directed cognition. Mind-wandering intrudes during tasks requiring sustained attention. Working memory capacity drops because the neural systems maintaining goal-relevant information don’t receive sufficient norepinephrine support.

Critically, none of these correlations remained significant when researchers analyzed local brain activity alone rather than functional connectivity. The associations depended specifically on network-level communication mediated by the locus coeruleus, not simply on activity within individual brain regions.

Total Tau in Spinal Fluid Tells the Story

Among cerebrospinal fluid biomarkers, total tau showed the clearest correlation with altered pupil diameter connectivity, particularly in frontal pole regions. Two clusters demonstrated significant positive associations between connectivity strength and total tau levels.

Total tau reflects ongoing neuronal damage rather than the burden of tau tangles accumulated in brain tissue. When neurons die or experience injury, tau proteins leak into surrounding fluid and eventually reach cerebrospinal fluid where they can be measured through lumbar puncture.

The positive correlation suggests a compensatory response—brains experiencing early neuronal damage may initially increase connectivity in certain regions attempting to maintain function. This compensation eventually fails as degeneration progresses, but during preclinical stages, heightened connectivity might temporarily preserve cognitive performance despite accumulating pathology.

Intriguingly, phosphorylated tau showed no correlation with locus coeruleus connectivity in this study, despite being considered a more specific marker of Alzheimer’s pathology. Recent research challenges the assumption that cerebrospinal fluid phospho-tau directly reflects tau tangle burden, showing it correlates more strongly with amyloid PET imaging than tau PET.

The disconnection between connectivity changes and phospho-tau levels raises questions about what cerebrospinal fluid biomarkers actually measure. They may better reflect acute injury responses than chronic accumulated pathology, potentially explaining why total tau (reflecting any neuronal damage) predicts connectivity changes while phospho-tau (more specific to Alzheimer’s) does not.

White Matter Hyperintensities and Vascular Contributions

The association between total tau and pupil diameter network connectivity could reflect vascular cognitive impairment rather than pure Alzheimer’s pathology. White matter hyperintensities—bright spots on MRI indicating small vessel disease—correlate with elevated cerebrospinal fluid total tau in elderly populations.

Vascular cognitive dysfunction produces aberrant pupillary responses potentially more pronounced than those seen in Alzheimer’s disease. The locus coeruleus has extensive vascular connections and high metabolic demand, making it vulnerable to small vessel disease that reduces blood flow.

When tiny blood vessels supplying the locus coeruleus deteriorate, neurons in this region experience chronic energy deficits even before overt cell death. The resulting dysfunction disrupts norepinephrine signaling throughout the brain, potentially triggering downstream pathology through multiple mechanisms—reduced clearance of toxic proteins, impaired immune responses, diminished synaptic plasticity.

The interplay between vascular damage and Alzheimer’s pathology complicates efforts to identify primary causes. Does locus coeruleus vascular damage enable tau accumulation, or does tau pathology impair local blood flow? Both probably occur, creating self-reinforcing cycles where each form of damage exacerbates the other.

Why Direct Imaging Remains So Difficult

The locus coeruleus measures roughly 15 millimeters long, 2 millimeters wide, and 2 millimeters deep—smaller than a grain of rice on each side of the brainstem. Its location adjacent to other small brainstem structures and its irregular shape make precise identification on standard MRI scans technically challenging.

Neuromelanin-sensitive MRI sequences can visualize the locus coeruleus because norepinephrine-producing neurons accumulate neuromelanin pigment with age. Lower locus coeruleus contrast on these specialized scans associates with greater tau deposition, faster cognitive decline, and Alzheimer’s disease progression.

But even neuromelanin imaging can’t resolve individual locus coeruleus neurons or distinguish subtle changes in specific subregions. The spatial resolution of functional MRI—typically 1.5 to 3 millimeters—means each voxel contains thousands of neurons from multiple structures.

Using atlas-based approaches to define locus coeruleus location proves unreliable because individual variation exceeds the nucleus’s size. A standardized brain template might place the locus coeruleus in slightly different positions relative to someone’s actual anatomy, causing signal extraction from neighboring structures instead.

This technical limitation drove researchers toward pupillometry as an indirect but more reliable surrogate. Pupil diameter changes reflect locus coeruleus activity through established neural pathways, providing functional information impossible to obtain through direct imaging at current resolutions.

Animal Studies Prove Causality

Mouse and rat experiments demonstrate the causal relationship between locus coeruleus activity and pupil dynamics that human research can only correlate.

Optogenetic activation—using light to trigger specific neurons expressing light-sensitive proteins—causes immediate pupil dilation when applied to locus coeruleus cells.

Conversely, chemogenetic silencing of locus coeruleus neurons prevents normal pupil responses to arousing stimuli. The effects are rapid, reversible, and dose-dependent, confirming direct control rather than indirect association.

Measurement of norepinephrine turnover in awake mice shows tight correlation between locus coeruleus neural firing, norepinephrine release, and pupil diameter changes. When locus coeruleus activity increases, norepinephrine levels rise throughout the brain within seconds, and pupils dilate proportionally.

Studies in rhesus macaques produce similar results using microelectrode recordings from locus coeruleus neurons during behavioral tasks. Spiking activity in this small brainstem nucleus predicts pupil size changes, attention shifts, and decision-making biases with remarkable precision.

The animal data establish that pupil diameter genuinely reflects locus coeruleus function rather than merely correlating through some third factor.

This validation justifies using pupillometry as a window into locus coeruleus health in human populations where direct neural recording isn’t feasible.

The Hypothalamus Wild Card

Orexin neurons in the lateral hypothalamus also drive pupil dilation, complicating the interpretation of pupillometry as purely reflecting locus coeruleus activity. These hypothalamic neurons regulate sleep-wake cycles, feeding behavior, and arousal states.

Animal studies show that orexin effects on pupil size require the locus coeruleus—blocking locus coeruleus activity prevents orexin-induced dilation. This suggests hypothalamic control operates through locus coeruleus activation rather than independently.

However, other hypothalamic circuits might bypass the locus coeruleus entirely. The hypothalamus projects throughout the brain, and its dysfunction in aging and neurodegeneration could produce pupillary changes unrelated to locus coeruleus pathology.

The pupil diameter network identified in the 2025 study included hypothalamic regions correlating with spontaneous pupil fluctuations. Whether this reflects hypothalamus-to-locus coeruleus-to-pupil pathways or direct hypothalamus-to-pupil connections remains unclear without more targeted interventions impossible in human subjects.

Disentangling these contributions matters for using pupillometry as a biomarker. If pupil changes mainly reflect locus coeruleus specifically, they offer targeted information about Alzheimer’s earliest pathology. If multiple systems contribute equally, interpretation becomes muddier and correlations with disease progression weaken.

Cognitive Differences Between Young and Old

Healthy older participants in the study scored significantly worse than younger adults on multiple cognitive assessments despite all being classified as cognitively normal. The Montreal Cognitive Assessment, free recall measures, figure copy tasks, and Trail Making Test B all showed age-related declines.

These differences persisted even after excluding anyone with mild cognitive impairment or dementia. Normal aging brings measurable cognitive changes distinct from disease, though the boundary between healthy aging and preclinical dementia remains fuzzy.

The reduced functional connectivity in pupil diameter networks among older adults parallels these performance differences. Whether connectivity decline causes cognitive changes, results from them, or both simultaneously remains uncertain without longitudinal data tracking the same individuals over time.

Cross-sectional comparisons risk confounding age effects with cohort differences. Older adults grew up in different environments, received different educations, and experienced different health exposures than younger participants. Some observed differences might reflect these lifetime factors rather than pure aging.

Sex Differences Complicate the Picture

The study included more female than male participants, particularly in the older group, likely because women live longer and volunteer for research more readily. This sex imbalance potentially skews findings toward female-typical patterns of aging and neurodegeneration.

Women show different Alzheimer’s risk factors, pathology distribution, and symptom progression compared to men. The locus coeruleus may degenerate differently between sexes, though research specifically examining sex differences in locus coeruleus pathology remains limited.

Hormonal changes during menopause affect norepinephrine systems and might alter locus coeruleus function independent of tau accumulation. Estrogen modulates norepinephrine signaling, and its decline could contribute to the connectivity changes observed in older women.

Future studies need balanced sex representation and analyses examining whether pupil-based biomarkers perform equivalently in men and women. Sex-specific thresholds or interpretation frameworks might improve biomarker utility across populations.

Medications Nobody Tracked

The study didn’t collect comprehensive medication information for participants, creating a significant blind spot. Many common medications alter norepinephrine signaling, locus coeruleus activity, or arousal states—all factors directly affecting pupil dynamics.

Beta-blockers prescribed for hypertension or cardiac conditions reduce norepinephrine effects throughout the body including the brain.

Stimulant medications increase norepinephrine availability. Antidepressants modulate norepinephrine reuptake. Antihistamines cause drowsiness affecting arousal states.

Roughly half the study participants were young adults less likely to take such medications regularly, but older adults frequently use multiple drugs affecting relevant systems.

Unmeasured medication effects could explain some observed age differences in pupil networks independently of neurodegenerative changes.

Controlling for medication use in future research will strengthen confidence that pupil-based measures reflect underlying brain pathology rather than pharmaceutical influences. Alternatively, studying how medications interact with pupil responses might reveal additional information about system integrity.

Dark Eyes Confound Machine Learning

Despite using deep learning algorithms for pupil segmentation, the analysis pipeline struggled with dark-colored eyes where the boundary between pupil and iris becomes difficult to distinguish. This technical limitation forced exclusion of many participants, reducing statistical power.

The quality control criteria—requiring less than 20% failed frames and at least 75% of frames with high confidence scores—balanced data reliability against sample size.

Stricter thresholds would ensure better data but exclude more people, while looser criteria increase noise.

Participants excluded due to poor eye-tracking likely weren’t random. Darker eye colors distribute unevenly across ethnic groups, potentially introducing selection bias.

The final sample may underrepresent populations most at risk for certain forms of dementia if their eye characteristics made them harder to track.

Improving pupil segmentation algorithms to handle diverse eye colors and lighting conditions would democratize this technology and eliminate systematic exclusion of particular groups. Current limitations prevent equitable application as a potential screening tool.

The Path Forward for Early Detection

Pupillometry combined with functional MRI offers a non-invasive window into locus coeruleus function that direct imaging can’t currently provide.

As a potential early biomarker for Alzheimer’s risk, pupil-based measures have several advantages—they’re cheap, quick, non-invasive, and don’t require radiation exposure or specialized equipment beyond a camera.

But translating research findings into clinical tools requires validation across larger populations with longitudinal follow-up.

Do pupil network changes actually predict who develops dementia years later? How much individual variation exists in healthy populations? What thresholds distinguish normal aging from preclinical disease?

Combining pupil measurements with other biomarkers—cerebrospinal fluid proteins, plasma biomarkers, genetic risk scores, cognitive testing—might improve prediction accuracy beyond any single measure.

A multimodal approach acknowledges that Alzheimer’s involves multiple systems failing simultaneously rather than one simple cause.

The locus coeruleus represents just one piece of a complex puzzle, though potentially the earliest piece to show damage.

Understanding its role doesn’t diminish the importance of amyloid plaques, tau tangles, inflammation, vascular dysfunction, or metabolic changes—it adds another layer revealing how diverse pathologies interconnect.

Why Seven Tesla Scanners Might Change Everything

Ultra-high field MRI at 7 Tesla offers spatial resolution fine enough to potentially visualize locus coeruleus subregions and track subtle changes over time. Early studies demonstrate feasibility of direct locus coeruleus imaging without relying on pupil surrogates.

Higher magnetic field strength improves signal-to-noise ratio, allowing smaller voxels that can distinguish the locus coeruleus from adjacent structures.

Specialized pulse sequences designed for brainstem imaging exploit neuromelanin contrast and optimize for the specific anatomy.

However, 7T scanners remain rare, expensive, and available primarily at research institutions. The practical utility of any biomarker depends partly on accessibility—a measurement technique limited to a dozen specialized centers can’t screen populations or guide treatment decisions for millions of patients.

Balancing technical sophistication against practical deployment will determine which approaches actually reduce dementia burden. The most scientifically impressive method means nothing if it never reaches patients who could benefit.

Closing the Window Between Detection and Treatment

Identifying locus coeruleus dysfunction decades before cognitive symptoms creates both opportunity and frustration. Opportunity because interventions might prevent downstream damage if applied early enough. Frustration because we lack proven treatments targeting locus coeruleus preservation.

Current Alzheimer’s drugs don’t specifically protect the locus coeruleus or boost norepinephrine signaling in ways that slow disease progression. Some older medications enhance norepinephrine availability but haven’t demonstrated disease-modifying effects in clinical trials.

The field needs therapies matched to earlier biomarkers. Finding brain changes 20 years before dementia onset accomplishes little without interventions applicable during that window. Otherwise we simply give people more time to worry about an outcome they can’t prevent.

Research into locus coeruleus-targeted treatments lags behind amyloid and tau therapies partly because the locus coeruleus received less attention until recently. The surge of interest following recognition of its early vulnerability may drive development of protective strategies—drugs supporting neuromelanin synthesis, compounds reducing oxidative stress in norepinephrine neurons, or lifestyle interventions optimizing arousal regulation.

The small structure hidden behind your eyes might hold keys to preventing dementia if we can develop tools to support it through decades of aging before damage becomes irreversible. Pupillometry offers one window into monitoring success, closing the loop between detection and intervention that could finally translate neuroscience insights into meaningful prevention.