A Mechanistic Reappraisal of Cognitive Dysfunction Following SARS-CoV-2 Infection

Author: John Murphy, CEO, COVID-19 Long-haul Foundation

Abstract

“Brain fog” in Long COVID has often been attributed to persistent neuroinflammation driven by sustained immune activation within the central nervous system. However, accumulating multi-modal evidence suggests this interpretation may be incomplete. While inflammatory signaling is frequently detectable systemically, direct evidence of chronic, compartmentalized brain inflammation in many patients is inconsistent, temporally variable, and often dissociated from cognitive symptom severity.

This review advances the hypothesis that Long-COVID cognitive dysfunction is more plausibly explained by non-inflammatory or post-inflammatory mechanisms, including: (1) cerebral microvascular and blood–brain barrier (BBB) dysfunction, (2) mitochondrial and bioenergetic failure, (3) astrocytic and oligodendroglial dysregulation, (4) persistent peripheral immune signaling without CNS immune activation, and (5) large-scale functional network disintegration affecting attention and executive control systems.

We integrate neuroimaging, neuropsychological, molecular, and systems neuroscience evidence to argue that “brain fog” represents a multi-system neuro-metabolic disconnection syndrome, rather than an ongoing inflammatory encephalitis-like state in most patients.

1. Introduction: The Neuroinflammation Hypothesis and Its Limits

Early conceptual models of Long COVID brain fog were strongly influenced by acute COVID-19 neuropathology, in which systemic inflammation, cytokine elevation, endothelial injury, and microglial activation were clearly documented. This led to a widely adopted assumption:

Persistent cognitive symptoms = persistent neuroinflammation

However, as longitudinal data have matured, this linear model has become increasingly difficult to sustain as a universal explanation.

Recent reviews and imaging studies show a more complex pattern:

Some patients show biomarker evidence of inflammation (e.g., IL-6, GFAP, S100B)
Many do not show sustained CSF or PET evidence of neuroinflammation
Cognitive impairment can persist even when systemic inflammatory markers normalize
Structural brain changes are subtle, regionally selective, or absent in many cohorts

Thus, the central paradox emerges:

How can profound cognitive dysfunction persist without consistent evidence of ongoing brain inflammation?

This discrepancy motivates a shift toward alternative mechanistic frameworks.

2. Neuroimaging Evidence: Weak and Inconsistent Signals of Persistent Inflammation

2.1 PET and MRI findings

Across cohorts, imaging findings in Long COVID brain fog include:

Mild reductions in glucose metabolism (FDG-PET)
Subtle regional perfusion abnormalities
White matter microstructural changes
Occasional limbic or brainstem signal abnormalities

However:

Findings are heterogeneous
Often non-replicated across cohorts
Frequently do not correlate tightly with symptom severity
Do not consistently show patterns typical of active encephalitis

A major imaging synthesis concluded that cognitive dysfunction is more consistent with network-level dysfunction and vascular dysregulation than inflammatory destruction .

2.2 Absence of a consistent inflammatory signature

In classic neuroinflammatory disease (e.g., autoimmune encephalitis), one expects:

CSF pleocytosis
Elevated neuroinflammatory biomarkers
Progressive structural injury
Clear lesion localization

In Long COVID:

CSF inflammation is often absent or minimal
Structural injury is subtle or absent
Symptoms fluctuate rather than progressively worsen

This inconsistency weakens the hypothesis of ongoing active CNS inflammation as the dominant mechanism.

3. Blood–Brain Barrier Dysfunction: A Competing Primary Driver

One of the most replicated findings is blood–brain barrier (BBB) disruption.

A landmark study demonstrated:

Elevated BBB permeability markers (e.g., S100B)
Evidence of endothelial activation
Leakage of plasma-derived proteins into CNS compartments
Association with cognitive symptoms

3.1 Mechanistic implications

BBB dysfunction alone can produce:

Altered neurotransmitter homeostasis
Microglial priming without full activation
Impaired neurovascular coupling
Reduced cerebral energy delivery efficiency

Importantly:

BBB disruption does not require sustained inflammation within the brain parenchyma.

Thus, BBB pathology can mimic inflammatory syndromes without true neuroinflammation.

4. Mitochondrial and Bioenergetic Failure Hypothesis

A growing body of evidence implicates cellular energy dysfunction as a core mechanism.

4.1 Key findings

Reduced cerebral metabolic efficiency
Altered lactate dynamics in brain regions
Impaired oxidative phosphorylation signaling
Mitochondrial regulatory gene perturbation

4.2 Functional consequences

The brain is an energetically expensive organ; small deficits in ATP availability can produce:

Slowed processing speed
Attention instability
Executive dysfunction
Working memory impairment

This aligns strongly with clinical “brain fog” phenomenology:

Patients describe “normal intelligence with impaired cognitive throughput.”

This pattern is more consistent with metabolic throttling than inflammatory destruction.

5. Neurovascular and Cerebromicrovascular Dysfunction

Increasing evidence supports Long COVID as a microvascular disease of the brain.

Key mechanisms include:

Endothelial inflammation and dysfunction
Capillary rarefaction
Microthrombotic injury
Impaired neurovascular coupling

These mechanisms can produce:

Regional hypoperfusion
Intermittent hypoxia at tissue level
Cognitive variability (“good days and bad days”)

Importantly, vascular dysfunction can exist:

with or without ongoing inflammation

and may persist long after immune activation resolves.

6. Glial and Astrocytic Dysfunction: The “Silent Regulator” Model

Astrocytes regulate:

glutamate clearance
synaptic energy supply
potassium buffering
neurovascular signaling

Emerging models suggest:

Astrocytic metabolic reprogramming after infection
Persistent synaptic inefficiency
Dysregulated glutamate cycling

These changes can produce cognitive impairment without inflammation.

Notably, astrocytic dysfunction can also explain:

FDG-PET hypometabolism
Cognitive slowing
Fatigue overlap syndromes (ME/CFS-like phenotypes)

7. Network-Level Brain Dysfunction: A Systems Neuroscience Perspective

A key emerging model reframes brain fog as:

A disorder of functional connectivity rather than structural inflammation.

Recent neurophysiological findings identify:

Altered salience network processing
Frontal-parietal inefficiency
Right insula dysfunction in perceptual control tasks

These findings suggest:

Cognitive symptoms arise from faulty network signaling
Not necessarily from inflammatory tissue injury

This aligns with:

EEG abnormalities
task-based inefficiency
preserved baseline structural MRI

8. Peripheral Immune Activation Without Central Inflammation

A key reconciliation model is:

Ongoing immune signaling outside the brain can produce brain dysfunction without CNS inflammation.

Mechanisms include:

cytokine signaling across BBB
vagal afferent modulation
endothelial receptor activation
microglial priming (non-activated state)

This produces:

“sickness behavior” physiology
fatigue and cognitive slowing
fluctuating symptom patterns

Importantly:

This is a systemic neuroimmune interaction syndrome, not necessarily a brain-inflamed state.

9. Synthesis: Why the Pure Neuroinflammation Model Is Incomplete

The totality of evidence suggests:

Neuroinflammation may be:

present in subsets
transient in many
secondary rather than primary in chronic cases

But brain fog is more consistently associated with:

vascular dysfunction
mitochondrial impairment
astrocytic dysregulation
network-level inefficiency

10. Central Hypothesis

Revised Model of Long-COVID Brain Fog

Long-COVID cognitive dysfunction is primarily a neuro-metabolic disconnection syndrome driven by vascular, mitochondrial, and glial dysfunction, with inflammation acting as an initiating or amplifying factor rather than a persistent central driver in most patients.

Part II — Pathobiology, Cognitive Phenotypes, and Converging Mechanistic Evidence

Author: John Murphy, CEO, COVID-19 Long-haul Foundation

11. Introduction to Part II: Moving Beyond a Single-Cause Model

The hypothesis advanced in Part I—that persistent neuroinflammation is not the dominant driver of Long-COVID brain fog in most patients—requires deeper mechanistic grounding.

This section expands into:

Cellular and subcellular pathology
Cognitive phenotype stratification
Neuropsychological signatures
Comparative disease models (ME/CFS, vascular cognitive impairment, post-sepsis syndrome)
Why anti-inflammatory interventions often show partial or inconsistent benefit

The emerging picture is not of a singular inflammatory brain disease, but of a distributed systems failure affecting energy, perfusion, and signaling efficiency across brain networks.

12. Cellular Pathology: What the Brain Actually Looks Like in Long COVID

12.1 Lack of uniform inflammatory histopathology

Autopsy and biopsy data (limited but growing) show:

Microglial activation in some cases
Endothelial injury more consistently than parenchymal inflammation
Sparse lymphocytic infiltration compared with classical encephalitis
No consistent pattern of widespread neuroinflammatory destruction

This is crucial:

Classical neuroinflammation produces cellular infiltration and structural injury. Long COVID often does not.

12.2 Endothelium as primary target

Across multiple studies, endothelial dysfunction appears more consistent than neuronal inflammation.

Observed features include:

Endothelial swelling and activation
Reduced nitric oxide signaling
Increased vascular adhesion molecules
Microvascular rarefaction

These changes lead to:

impaired perfusion regulation
oxygen diffusion inefficiency
intermittent hypoxic microenvironments

Importantly:

Endothelial pathology can drive brain dysfunction independently of inflammation within neural tissue.

12.3 Microclot and fibrin-amyloid pathology (supportive but debated)

Some studies describe:

persistent fibrin microclots resistant to fibrinolysis
platelet hyperactivation
abnormal clot morphology

While still controversial in extent and generalizability, if present, these processes could:

reduce capillary flow
create patchy cerebral hypoperfusion
amplify fatigue and cognitive slowing

However:

Even if microclot pathology is present, it does not require ongoing brain inflammation.

13. Cognitive Phenotypes: Brain Fog Is Not a Single Entity

A major limitation in earlier literature is treating “brain fog” as a uniform condition. In reality, cognitive profiles cluster into distinct phenotypes.

13.1 Phenotype A: Processing Speed Dominant Impairment

Features:

slowed reading comprehension
delayed mental arithmetic
“thinking through molasses” sensation
preserved memory storage but impaired retrieval speed

Mechanistic alignment:

mitochondrial dysfunction
reduced cerebral glucose utilization
white matter conduction inefficiency

13.2 Phenotype B: Attention Instability / Executive Fragmentation

Features:

inability to sustain attention
task switching failure
mental fatigue after brief cognitive load

Mechanistic alignment:

frontal-parietal network inefficiency
salience network dysregulation
neurovascular uncoupling

13.3 Phenotype C: Memory Encoding Disruption

Features:

difficulty forming new memories
“blank encoding” episodes
word retrieval failures

Mechanistic alignment:

hippocampal metabolic stress
cholinergic signaling disruption
astrocyte dysfunction

13.4 Phenotype D: Fluctuating Cognitive Collapse (ME/CFS-like)

Features:

delayed post-exertional cognitive crash
symptom variability hour-to-hour or day-to-day
disproportionate fatigue relative to activity

Mechanistic alignment:

systemic energy metabolism failure
autonomic dysregulation
immune–metabolic feedback loops

Key Insight

If a single inflammatory lesion were responsible, cognitive deficits would be more uniform and progressive. Instead, Long-COVID brain fog is phenotypically heterogeneous, supporting a multi-mechanism model.

14. Neuropsychological Testing Patterns

Across published cohorts, consistent findings include:

14.1 Most affected domains

processing speed (most sensitive marker)
sustained attention
working memory
executive planning efficiency

14.2 Relatively preserved domains

crystallized intelligence
vocabulary
long-term semantic memory
basic visuospatial function

14.3 Interpretation

This pattern strongly suggests:

a functional inefficiency syndrome, not a degenerative or inflammatory cortical destruction process.

Inflammatory brain diseases typically show:

broader cognitive domain collapse
progressive decline
structural correlates on imaging

Long COVID instead shows:

selective cognitive bottlenecks
variability
partial reversibility in many cases

15. Why Anti-Inflammatory Treatments Often Show Limited Efficacy

If persistent neuroinflammation were the dominant driver, therapies such as:

corticosteroids
NSAIDs
IL-6 or JAK inhibitors
microglial suppressants

would be expected to show strong, consistent benefit.

Yet clinical reality shows:

mixed or modest response rates
transient improvement in subsets
lack of durable cognitive restoration in most trials

15.1 Interpretation

This pattern suggests:

inflammation may be upstream, not sustaining
or inflammation is peripheral/systemic rather than central
or inflammation is secondary to metabolic and vascular dysfunction

15.2 Analogy

Inflammation may be:

the “spark” rather than the “engine” of chronic brain fog.

16. ME/CFS as a Neuro-Comparative Model

Long COVID brain fog shares significant overlap with ME/CFS:

Feature	ME/CFS	Long COVID
Post-exertional fatigue	core	common
Cognitive slowing	core	core
Normal structural MRI	typical	typical
Neuroinflammation	inconsistent	inconsistent
Metabolic dysfunction	strong evidence	strong evidence

Key implication

Two clinically distinct post-viral syndromes converge on energy failure physiology, not persistent brain inflammation.

17. Neurovascular Coupling Failure: A Core Integrative Mechanism

Neurovascular coupling is the mechanism by which:

neuronal activity triggers rapid local blood flow increase.

In Long COVID evidence suggests:

impaired vasodilatory response
endothelial stiffness
nitric oxide dysregulation

17.1 Consequence

Even if neurons are intact:

they may be under-fueled during demand
leading to transient cognitive collapse under load

17.2 Why this mimics “brain fog”

baseline cognition may appear normal
but breakdown occurs under complexity
symptoms fluctuate with exertion

This is highly characteristic of patient reports.

18. Astrocyte–Neuron Metabolic Coupling Failure

Astrocytes normally:

buffer glutamate
supply lactate to neurons
regulate synaptic efficiency

Post-viral dysfunction may lead to:

inefficient lactate shuttle
excitatory–inhibitory imbalance
synaptic fatigue accumulation

Resulting cognitive signature

mental exhaustion after minimal cognitive load
“buffer overflow” sensation
slow recovery after thinking effort

19. The “Silent Hypometabolism” Model

FDG-PET studies in Long COVID often show:

reduced regional glucose uptake
especially in frontal and limbic areas

However:

not all hypometabolism corresponds to inflammation
similar patterns occur in:
- sleep deprivation
- chronic fatigue states
- deconditioning syndromes

Interpretation

Hypometabolism is a final common pathway, not proof of inflammatory causation.

20. Integrated Mechanistic Model (Revised Framework)

Long-COVID brain fog is best conceptualized as a multi-layer cascade:

Stage 1: Trigger

viral infection / immune activation

Stage 2: Systemic injury

endothelial dysfunction
autonomic disruption
mitochondrial stress

Stage 3: Neurovascular and metabolic decoupling

impaired perfusion regulation
reduced energy delivery

Stage 4: Network-level dysfunction

salience + executive network inefficiency

Stage 5: Clinical syndrome

fluctuating cognitive impairment (“brain fog”)

Core conclusion

Persistent brain fog does not require persistent brain inflammation. A more parsimonious explanation is a sustained neurovascular–metabolic inefficiency state with secondary immune signaling.

21. Introduction to Part III: From Mechanism to Clinical Reality

Parts I–II established a central thesis:

Long-COVID brain fog is unlikely to be driven primarily by ongoing neuroinflammation in most patients.

This section moves from theory to clinical consequence, focusing on why treatments behave inconsistently and how emerging biomarker patterns support a heterogeneous, multi-system disorder rather than a uniform inflammatory encephalopathy.

We will address:

Why treatment response is inconsistent across patients
Biomarker stratification and emerging diagnostic signals
Therapeutic failure modes of anti-inflammatory approaches
Why metabolic, vascular, and autonomic therapies show greater promise in subsets
A proposed clinical classification system

22. The Core Clinical Puzzle: Treatment Non-Uniformity

If Long-COVID brain fog were primarily inflammatory, one would expect:

predictable response to immunomodulation
consistent biomarker elevation in responders
dose-dependent improvement with anti-inflammatory therapy

However, observed reality:

Patients respond heterogeneously, often unpredictably, and sometimes paradoxically.

Examples:

some improve transiently with steroids
others worsen cognitively
antivirals rarely reverse chronic brain fog
SSRIs improve fatigue in some but not cognition broadly
stimulants provide partial symptomatic relief but no disease modification

Interpretation

This pattern strongly suggests:

brain fog is not a single-pathway inflammatory disease, but a multi-node physiological disruption syndrome.

23. Biomarker Landscape: What Actually Moves in Long COVID

23.1 Inflammatory markers

Common findings:

IL-6: elevated in subsets, often transient
TNF-α: inconsistent elevation
CRP: often normal in chronic phase
CSF cytokines: frequently normal or minimally elevated

Key insight

systemic inflammation is present in some patients but is neither universal nor sustained in most chronic cases.

23.2 Endothelial and vascular biomarkers

More consistent signals:

von Willebrand factor elevation
soluble ICAM-1 and VCAM-1 increases
endothelial progenitor cell dysregulation
altered nitric oxide metabolites

These correlate more closely with:

cognitive slowing
fatigue severity
orthostatic intolerance

23.3 Metabolic biomarkers

Emerging findings include:

altered lactate/pyruvate ratios
impaired oxidative phosphorylation signatures
mitochondrial stress markers (e.g., mtDNA fragments)
reduced cerebral glucose uptake on imaging

23.4 Autonomic biomarkers

reduced heart rate variability (HRV)
sympathetic overactivation patterns
baroreflex sensitivity impairment

These correlate strongly with:

cognitive fatigue
attention instability
post-exertional worsening

Summary

The most reproducible biomarker signals are vascular, metabolic, and autonomic—not inflammatory.

24. Why Anti-Inflammatory Therapies Often Fail (Mechanistic Analysis)

24.1 Timing mismatch hypothesis

Inflammation may occur:

early in infection
during acute immune activation

But brain fog often persists after:

immune normalization
cytokine decline

Thus:

treating inflammation may target an upstream event no longer driving symptoms.

24.2 Compartmental mismatch

Systemic inflammation ≠ CNS inflammation

Key point:

blood cytokines can normalize
brain metabolic dysfunction may persist

Thus:

systemic immunosuppression does not necessarily correct CNS energy failure.

24.3 Downstream irreversibility window

Some changes may become self-sustaining:

endothelial remodeling
mitochondrial downregulation
synaptic recalibration

These can persist independently of inflammation.

24.4 Heterogeneity problem

Some patients do respond to anti-inflammatory strategies, implying:

multiple subtypes exist
inflammation is a subset driver, not universal cause

25. Therapeutic Response Mapping (Clinical Pattern Analysis)

Below is a synthesis of observed response clusters.

25.1 Anti-inflammatory agents

Examples:

corticosteroids
NSAIDs
cytokine inhibitors

Response pattern:

short-term improvement in some
no sustained cognitive restoration
inconsistent responders

Interpretation:

partial benefit suggests inflammation is modulatory, not primary driver.

25.2 Antiviral therapies

Examples:

nirmatrelvir/ritonavir
acyclovir-class analogs (off-label use cases)

Response pattern:

minimal effect on established brain fog
occasional fatigue improvement

Interpretation:

persistent infection is not the dominant mechanism in most chronic cases.

25.3 Neurostimulants

Examples:

methylphenidate
modafinil

Response pattern:

improved wakefulness and task initiation
no restoration of cognitive efficiency
tolerance development in some

Interpretation:

compensatory activation, not disease modification.

25.4 Autonomic-targeted therapies

Examples:

beta blockers
ivabradine
volume expansion strategies

Response pattern:

improvement in cognitive endurance in orthostatic phenotypes

Interpretation:

supports neurovascular/autonomic contribution.

25.5 Metabolic interventions

Examples:

ketogenic support
mitochondrial cofactors (CoQ10, NAD+ precursors)
GLP-1 agonist class signals (emerging interest)

Response pattern:

modest but sometimes sustained cognitive improvement
particularly in fatigue-dominant phenotypes

Interpretation:

supports metabolic limitation model.

26. Proposed Clinical Stratification Model

To resolve heterogeneity, Long-COVID brain fog can be categorized into four overlapping subtypes:

Subtype I: Neurovascular Dysregulation Phenotype

Core features:

orthostatic symptoms
cognitive variability with posture/exertion
headache and perfusion sensitivity

Mechanism:

endothelial dysfunction
impaired cerebral blood flow regulation

Subtype II: Metabolic Hypofunction Phenotype

Core features:

severe fatigue
slowed cognition
post-exertional cognitive crash

Mechanism:

mitochondrial inefficiency
reduced ATP availability

Subtype III: Network Connectivity Dysfunction Phenotype

Core features:

attention fragmentation
executive dysfunction
task-switching impairment

Mechanism:

salience and frontal-parietal network inefficiency

Subtype IV: Immune-Modulated Neurobehavioral Phenotype

Core features:

fluctuating inflammatory symptoms
systemic malaise
episodic worsening

Mechanism:

peripheral immune signaling
microglial priming without sustained CNS inflammation

27. Why This Model Explains “Brain Fog” Better Than Neuroinflammation Alone

The neuroinflammation hypothesis struggles with:

heterogeneity
lack of consistent biomarkers
weak imaging correlation
poor treatment predictability

In contrast, the integrated model explains:

Feature	Neuroinflammation	Neurovascular–Metabolic Model
Symptom variability	Poor	Strong
Imaging heterogeneity	Poor	Strong
Treatment inconsistency	Poor	Strong
Post-exertional worsening	Weak	Strong
Normal CSF findings	Inconsistent	Expected

28. Central Reframing

Brain fog is better understood as a systems-level failure of cerebral energy delivery and network coordination, with immune activity acting as a trigger or amplifier rather than a sustained central driver in most cases.

29. Therapeutic Implications

This model shifts treatment strategy from:

Old paradigm:

suppress inflammation in the brain

New paradigm:

restore physiological efficiency across vascular, metabolic, and network systems

29.1 Potential intervention classes

A. Vascular restoration

endothelial support strategies
nitric oxide modulation
microcirculation enhancement

B. Metabolic reconditioning

mitochondrial substrate support
controlled aerobic re-training
ketone-based metabolic support

C. Autonomic stabilization

HRV training
graded positional adaptation
baroreflex reconditioning

D. Network rehabilitation

cognitive pacing
neuroplasticity-based training
sensory load management

30. Key Conceptual Shift

The goal is not to suppress a “hot brain,” but to re-optimize a low-efficiency brain network under physiological constraint.

References

Nature Neuroscience (BBB dysfunction in Long COVID cognitive impairment)
Molecular mechanisms of cognitive dysfunction in Long COVID review
Cerebromicrovascular dysfunction in Long COVID
Mitochondrial dysfunction in Long COVID
Neurological imaging and mechanisms review
EEG/insula dysfunction evidence study
Neuropsychological profiling in Long COVID cohorts (processing speed deficits)
Cerebral perfusion abnormalities in post-acute COVID syndrome
Endothelial dysfunction and vascular signaling in SARS-CoV-2 sequelae
FDG-PET hypometabolism patterns in post-viral syndromes
ME/CFS neuroimaging and metabolic dysfunction literature
Neurovascular coupling impairment models in systemic illness
Astrocyte metabolic dysfunction in post-viral syndromes
Cognitive phenotype stratification in Long COVID populations
Endothelial biomarkers in post-acute viral syndromes
Autonomic dysfunction and cognitive impairment correlation studies
Mitochondrial dysfunction and fatigue syndromes literature
Neurostimulant response variability in Long COVID cohorts
HRV and cognitive performance correlation studies
Cerebral perfusion regulation impairment in post-viral illness
Systemic inflammatory marker variability in chronic COVID syndrome
Neurovascular coupling dysfunction models