cognitive resilience

Cognitive resilience is the phenomenon of carrying Alzheimer-threshold pathology — amyloid and tau at burdens that in most people produce dementia — without the dementia. It is the one place where the disease's causal chain is observably broken in humans, which is why this corpus treats it as an anchor rather than a curiosity: whatever distinguishes a resilient brain is, by construction, downstream of the pathology and upstream of the symptom.

Resilience is a human observation, not a model-system one. That is its evidentiary strength and the reason it survives when mechanistic claims in this corpus do not.

The central human finding, stated correctly

The load-bearing human study is de Vries, Bahnerth, Swaab, Verhaagen and Carulli, Resilience to Alzheimer's disease associates with alterations in perineuronal nets (Alzheimer's & Dementia, 2025;21(2):e14504).

Cognitively resilient donors carrying Alzheimer-threshold pathology show a remodelled perineuronal matrix — reduced WFA+ net density and reduced aggrecan immunoreactivity relative to controls — accompanied by the absence of the matrix-proteolytic transcriptional signature that characterises the demented brain. Resilience is associated with net remodelling rather than net preservation.

Where the framework claim is wanted rather than the finding, the defensible form is: the resilient brain is distinguished not by how many nets it retains but by what accompanies their loss — regulated remodelling without the proteolytic signature, rather than digestion with it.

This corrects a claim still carried elsewhere in the corpus. Several documents state that resilient donors show preserved or intact nets, preserved aggrecan, preserved tenascin-R and intact synaptic contacts. The paper reports the opposite or reports nothing:

• aggrecan immunoreactivity around PV+ neurons was decreased in resilient donors (resilient vs control p = .005)
• peridendritic net complexity was reduced (p < .0001)
• WFA+ net density was lower in the resilient group specifically — below both control and AD (p = .0005)
• tenascin-R was never measured; the term appears only inside a cited reference title
• synaptic contacts in resilient donors were a null lying between the comparison groups (resilient vs control p = .499; AD vs resilient p = .310)

See research/CORPUS_ERRATA_2026-07.md item 1, which is the source of the wording above and lists the files still to be corrected.

What survives, and what it implies

The microglial element of the finding survives intact, and it is the part that carries the framework. Bulk RNA-seq gene-set enrichment showed the matrix-proteolytic programme (VCAN, ADAMTS2) elevated in Alzheimer donors and not in resilient ones. There are no MMP or cathepsin data in the paper, so claims naming those proteases must be sourced elsewhere.

The paper also supplies a finding the corpus under-uses: excitatory neurons bearing a perineuronal net carry low amounts of p-tau — direct support for net-mediated protection on its own terms.

The consequence is that resilience is not a matter of how much matrix survives but of how it is turned over. Regulated remodelling and proteolytic digestion can both reduce net density; only one of them is compatible with intact cognition.

Other substrates the corpus proposes

Resilience is over-determined, and different documents locate it in different places:

• Microglial restraint. The TREM2 barrier uncouples amyloid from dementia — resilience as a brake held rather than a structure kept.
• Trophic support. Brain BDNF expression tracks cognitive resilience, and the claim survives its reverse-causation test only in the conditional form the interventional evidence licenses: where BDNF has been restored — by gene delivery in aged primates and amyloid-bearing rodents, by small-molecule TrkB agonists — synaptic and cognitive endpoints improve.
• Bioenergetic capacity. The pharmacology work makes PV+ interneuron survival and the perineuronal net density around those interneurons the circuit-level readout of upstream bioenergetic rescue — the most proximate substrate of resilience the framework names.
• Protective genetics. A strength-graded atlas of protective alleles and epigenetic states treats resilience as something to phenocopy rather than merely to admire.

Why the corpus keeps returning to it

An audit of the framework's coverage argued that resilience should be the primary anchor rather than a secondary one — that a three-phase attrition model describes only a minority of dementia, while the resilience observation constrains any account that hopes to be right. Whether or not that inversion is accepted, resilience does work no mechanism can: it is a human, uncontrived dissociation between pathology and disease.

Advanced by

The concept enters this corpus through the de Vries / Carulli line of work on perineuronal nets in resilient donors. Note that the scientist record behind this concept carries the wrong first name (see errata item 2(f)); the paper's first author is Lisa E. de Vries.

Related

Perineuronal Nets · Locus coeruleus as ground zero · Synaptic Plasticity · Proteostasis · Compensatory Paradigm Nexus