Alzheimer's: A Person Already Remembers Far Too Much
Identity Preservation Hypothesis: A Medical Application of Korean Memorization Methods
Prologue
The starting point for thinking about Alzheimer's was a set of ordinary scenes from daily life. Scenes on television and in dramas showing a person with Alzheimer's — an elderly parent who has forgotten her child's name, who no longer recognizes her spouse. Before it was a problem of medical knowledge, it was the weight of watching a person drift away from herself.
What weighed even more was the position of the family watching the patient. The moment a spouse of decades no longer recognizes you. The moment the parent who raised you treats you as a stranger. The family undergoes its own parallel loss alongside the patient. The patient loses memory; the family loses the relationship. The wound that family inflicts on family in this way is a depth of suffering that medical literature does not sufficiently address. Whether there was a direction that could resolve this problem became the starting point of this work.
The heaviness turned into questions without much delay. They were the most ordinary questions a layperson could ask. How does this disease begin. Why do some people break down faster than others. Why is molecular treatment so difficult. And — where, exactly, is the academy treating the self that a patient is losing.
Abstract
The starting observation of this essay is this: a person already remembers far too much. Tens of thousands of faces, hundreds of thousands of words, decades of autobiographical events, countless places and sounds and habits are stored in the brain, and the core information required for a patient to live as herself amounts to less than 0.01% of that vast reservoir. Alzheimer's cognitive intervention must abandon the impossible goal of preserving total memory and instead target only the narrow cluster that constitutes identity, drawn from the memory that already exists. This essay proposes resetting the target of Alzheimer's cognitive intervention from "generalized cognitive ability" to "the specific information circuits that compose the patient's identity." Combining the neuropathological fact that Alzheimer's damage follows an orderly circuit-by-circuit progression with the molecular rule that synapses are asymmetrically preserved according to frequency of use, it predicts that daily multichannel reinforcement of individually defined core identity information will maintain a denser-than-average synaptic density in those circuits, significantly delaying the threshold of functional collapse during disease progression. The hypothesis lies at the intersection of two adjacent academic fields — reminiscence therapy and spaced retrieval training — and shares the same lineage as the algorithmic spaced retrieval work validated in 2024 by Smith et al. with the BST app. What distinguishes it is not the algorithm but what sits on top of the algorithm. When five learning methods accumulated over a century in Korea — the error notebook, consolidation into a single notebook, recitation, multichannel association, and the repetition schedule — are adapted to the cognitive, emotional, and physical conditions of elderly patients and applied to a medical target, a new clinical position opens in the domains of multichannel integration, family reinforcement, and environmental simplification that existing tools have not addressed.
1. Background — The Current Landscape of Alzheimer's Research
Alzheimer's research has operated for three decades around the Amyloid Cascade Hypothesis. Formalized by Hardy and Higgins in 1992, the hypothesis models the accumulation of amyloid-beta (Aβ) in the brain as the primary cause of Alzheimer's, with tau aggregation, neuroinflammation, synaptic loss, and neuronal death as downstream consequences. It rested on strong genetic evidence — familial early-onset mutations in APP, PSEN1, PSEN2, and the high prevalence in Down syndrome — and shaped the drug development direction of the entire pharmaceutical industry.
Over the last twenty years, cracks in the hypothesis have accumulated. Anti-amyloid antibodies such as solanezumab, bapineuzumab, crenezumab, and gantenerumab, and BACE inhibitors such as verubecestat and lanabecestat, failed in succession in Phase 3 trials. Autopsy data increasingly showed that cognitive decline correlates more strongly with the burden of tau neurofibrillary tangles than with amyloid quantity, and the concordance between amyloid burden and cerebral hypometabolism weakened. Lecanemab and donanemab, approved in 2023 and 2024, were the first drugs to meaningfully clear amyloid, yet in eighteen-month follow-up they slowed cognitive decline by only 0.45 points on CDR-SB and 3.25 points on iADRS. The frequency of ARIA (amyloid-related imaging abnormalities — cerebral edema and microhemorrhages) remained a clinical burden. In 2025 the European EMA rejected donanemab and approved lecanemab only for ApoE ε4 non-carriers.
The academy has moved not to abandon the hypothesis but to layer others on top of it. The "Amyloid Cascade Hypothesis 2.0," proposed by Volloch and colleagues, identifies intraneuronal Aβ, not extracellular plaque, as the true culprit. Camps that treat neuroinflammation as the primary driver, infection hypotheses pointing to chronic herpes infection as the trigger, views placing gut-brain axis microbial dysbiosis at the origin, camps identifying cerebrovascular damage and blood-brain barrier breakdown as antecedent events, and the geroscience hypothesis that treats aging itself as the upstream cause — each accumulates its own data and competes for ground. No academic consensus exists, as of 2026, on the true cause.
The limits of this drug-centered paradigm have renewed interest in non-pharmacological cognitive intervention. Yet demonstrating effects has been hard there too. The landmark ACTIVE trial (2002–2017) measured long-term effects of working memory, reasoning, and processing-speed training in about 2,800 older adults, and concluded that gains in the trained domain were robust but transfer to daily function was limited. The academy read this as "the limit of cognitive training" and redirected resources to drug development.
This hypothesis proposes an alternative reading. The failure of transfer from generalized cognitive training in ACTIVE and similar trials can be diagnosed not as "the limit of cognitive training itself" but as "the limit of targeting nonspecific capacity." That is, so long as generalized cognitive ability is the target, effects will remain weak. Redefining the target can change the outcome. Indeed, the five-year follow-up study by Belleville and colleagues (2024) reported that cognitive training in MCI patients significantly slowed cognitive decline at five years and preserved cognitive reserve. The possibility of target redefinition is beginning to open within the academy as well.
2. The Asymmetry of Circuit Damage and Synaptic Plasticity
Alzheimer's neuropathology is not random. Studies of tau neurofibrillary tangle distribution, standardized by the Braak staging system, revealed that damage follows a consistent anatomical order. It begins in the entorhinal cortex and progresses to the hippocampus, lateral temporal lobe, parietal lobe, and frontal lobe; motor cortex, visual cortex, and brainstem are preserved until relatively late. The reason this order is not coincidence lies in the propagation mechanism of tau. Abnormal tau produced in one neuron moves through synapses to neighboring neurons, initiating the same cascade, and damage therefore spreads along the wiring paths of neural circuits.
The decisive fact here is the asymmetry of damage. With identical amyloid and tau burden, not all circuits collapse at the same speed. Autopsy studies accumulate observations that the circuits a patient used most across life — the motor circuit of a lifelong pianist, the native-language circuit of a lifelong speaker — are relatively preserved until the end. The clinical scene in which family members report "she can't do much else but she still plays the piano" is the everyday expression of this asymmetry.
Two molecular rules support the asymmetry. First, Hebb's law: neurons that fire together wire together. When neurons in the same circuit repeatedly activate together, synapses between them strengthen, the synapses themselves grow larger, and new synapses are added nearby. The technical term is LTP (long-term potentiation). Second, the "use it or lose it" rule: frequently used synapses strengthen while unused ones weaken. The strength and number of synapses are bidirectionally regulated by frequency of use.
Both rules continue to operate during Alzheimer's progression. Living neurons retain the capacity to form new synapses throughout life. This capacity is weakened in Alzheimer's patients but not abolished. Chronic inflammation and an amyloid-rich environment suppress new synapse formation, yet formation still occurs when appropriate stimulation is provided. That exercise increases BDNF (brain-derived neurotrophic factor) and stimulates synaptogenesis, and that cognitive training raises synaptic density in targeted circuits, is established in both animal experiments and human research.
Cognitive reserve enters here. The brains of people whose cognitive function is preserved longer despite equal neuropathological burden show higher synaptic density and more bypass routes than average. The epidemiological data are consistent: educational level, cognitively complex occupation, second languages, musical instruments, and social engagement raise cognitive reserve. The 2024 Lancet Commission enumerated fourteen modifiable risk factors and estimated that full management could prevent or delay roughly 45% of dementia worldwide. The molecular basis of that estimate is again synaptic plasticity and circuit preservation.
Combining these facts yields a coherent picture. Alzheimer's progresses circuit by circuit. Circuit damage is asymmetric, not uniform. Frequently used circuits become thicker and collapse later. Synaptic strengthening operates as a lifelong mechanism. With identical damage, the threshold of functional collapse is delayed according to circuit thickness.
2.6 The Scale of Memory and the Narrowing of the Target
One more fact must be raised when designing the target of Alzheimer's cognitive intervention. A person already remembers far too much.
The long-term memory of an ordinary adult stores tens of thousands of faces encountered across life, hundreds of thousands of words, decades of autobiographical events, and countless places and sounds and habits. Academic estimates place the information capacity of human long-term memory at roughly one petabyte — comparable to the storage space of a modern high-performance server. Even in moments when we feel our memory weakening in daily life, the brain in fact maintains an enormous information reservoir.
Of this vast memory, the core information required for a patient to live as herself does not exceed a few hundred items. Her own name, the names and faces of family, her residence, the medications she takes daily, dozens of core relationships, dozens of daily routines. This narrow cluster is less than 0.01% of total memory.
Alzheimer's is not a disease that erases this total memory indiscriminately. It progresses circuit by circuit in a consistent order. A significant portion of what a patient loses is, in fact, information she does not need to live as herself. Lunch menus from twenty years ago, thousands of faces glimpsed in passing, occupational minutiae from decades past, the details of sentences in books she read. These can fade without the self being lost. The problem arises at the moment what is being lost reaches the narrow cluster that composes identity.
This scale asymmetry determines the target of the hypothesis. Any approach attempting to preserve the patient's total memory is aimed at an impossible goal. So long as the brain's total information volume is on the petabyte scale, no cognitive intervention can succeed at total preservation. However, an approach that concentrates resources on the less-than-0.01% identity core cluster becomes an achievable goal. It transforms into a question of which narrow part of already-existing vast memory should be protected.
The logic of this selection meshes directly with the asymmetry of synaptic plasticity. The rule that frequently used circuits become thicker does not apply to all circuits simultaneously. Resources are finite, and so are the circuits that can be strengthened. Thus the question of "what to reinforce" becomes more important than "how much can be reinforced." A strategy of concentrated reinforcement of the identity core cluster produces far greater preservation effects with the same resources than a strategy of uniformly weak reinforcement of total memory.
This perspective connects directly to the Korean learning tradition of consolidation into a single notebook. Just as students compress dozens of study volumes into the one notebook they actually need, the narrow identity cluster is extracted from the patient's vast memory for concentrated reinforcement. This connection is developed in Chapter 5.
3. Statement of the Hypothesis and Mechanism of Action
From the preceding facts, the following hypothesis is derived.
"When individually defined core identity information is reinforced daily through multichannel stimulation, the synaptic density of the circuits carrying that information is maintained at levels higher than average, significantly delaying the threshold of functional collapse during Alzheimer's progression."
The hypothesis has three core components: the definition of identity information, the mechanism of multichannel repeated stimulation, and the measurability of delayed threshold collapse.
3.1 Definition of Identity Information
Identity information refers to the minimum cluster of information required for a patient to recognize herself as herself. The patient's own name and age. The names, faces, and relationships of family members. Her lifelong neighborhood and current residence. Frequented places (hair salon, pharmacy, market) and their frequencies. Medications taken daily and their times. Close friends and acquaintances. Thirty to fifty key events of her life. Her values and preferences. This cluster is not a grand autobiography but a living form of personal identification. Its focus is on the functioning of daily identity rather than on the meaning of autobiographical events.
The decisive difference between identity information and generalized cognitive ability is "patient-specific non-substitutability." Working memory scores are measured on the same scale across all patients, but identity information differs completely from patient to patient. What matters to Kim Sun-ja (her eldest son Park Jeong-min, Miseon Hair Salon, metformin) is not shared with what matters to Park Cheol-su (his second daughter Park Ji-young, the neighborhood pharmacy, Norvasc). Patient-specific customization is intrinsic to the target.
3.2 Mechanism of Multichannel Repeated Stimulation
Multichannel repeated stimulation means exposing the same information repeatedly through different sensory channels (visual, auditory, textual), at different times, in different formats. It applies directly the spaced repetition algorithms that have been studied in the learning field for over a hundred years. Intervals begin short and grow longer as successful recall accumulates. At the same time, the same information is distributed across channels — photographs, audio, text, direct voice recordings from family members. This is what the academic literature calls multimodal reinforcement.
The mechanism of action proceeds in two stages. The first stage is synaptic strengthening. When the circuit carrying identity information is activated daily, LTP occurs repeatedly at its synapses, postsynaptic receptor numbers rise, the synapses themselves grow larger, and in living neurons new synapses form. The second stage is distributed storage of the circuit. When the same information is exposed through different channels, it is stored across multiple brain regions. Visual information lives along the occipital-temporal pathway, auditory information along the lateral temporal lobe, family voices alongside emotional circuits (the amygdala). Even when one region is damaged, a copy of the same information exists elsewhere and recall remains possible. This is a structural preservation effect analogous to the RAID (redundant storage) principle of computer hardware.
3.3 Measurement of Delayed Threshold Collapse
Delayed threshold collapse translates into measurable outcome variables. Domain-specific cognitive tests — accuracy of identity information recall, family identification, self-location awareness, self-schedule awareness — measured at six-month or one-year intervals quantify the preservation of reinforced circuits. General cognitive tests (MMSE, CDR-SB) may decline at similar rates in reinforcement and control groups, and this dissociation between the general battery and the identity-domain battery forms the core prediction of the hypothesis.
4. Position Relative to Adjacent Academic Fields and BST/Braintrust
The hypothesis sits at the intersection of two adjacent academic fields: reminiscence therapy and spaced retrieval training.
Reminiscence therapy, proposed by Robert Butler in the 1960s under the concept of "life review," has since taken its place as one axis of non-pharmacological intervention for Alzheimer's patients. Clinical results have accumulated showing that repeatedly stimulating autobiographical events (childhood, marriage, child-rearing, occupational experience) with photographs and conversation improves autobiographical recall and stabilizes affect. A 2021 study in Frontiers in Psychology reported that repeated reminiscence workshops clearly improved autobiographical memory recall even in late-stage Alzheimer's patients, and simultaneously documented the observation that the patient's sense of identity is relatively preserved into late stages.
Spaced retrieval training is a technique used in clinical settings since the 1980s. A series of studies by Camp and colleagues demonstrated that repeatedly recalling the same information across progressively longer intervals enables Alzheimer's patients to learn new information (family names, medication regimens, simple procedures). American speech-language pathologists use it as a standard tool.
The most direct clinical advance at the intersection of these two fields is the BST (Blank Slate Technologies) app study published in JMIR Formative Research in 2024 by Smith and colleagues. The study implemented a machine-learning-based spaced retrieval algorithm as a mobile app, had 20 MCI patients use it autonomously for four weeks, and reported meaningful effects in both learning new factual information and relearning faded older information (celebrity names). The same company released "Braintrust: Memory Companion" in July 2024 on the Apple App Store, narrowing the target of the same algorithm: patients and family members together enter identity and memory information, and the AI algorithm delivers a personalized review schedule with daily exposure.
This hypothesis shares the same lineage as the BST/Braintrust current and adds four points of differentiation on top of it.
First, multichannel integration. BST is text-Q&A-centric. This hypothesis explicitly combines photographs, audio, text, and direct family voice, and measures the separable contribution of each channel in clinical trials. Visual, auditory, and emotional circuits in different brain regions each hold a copy of the same information, forming a RAID-like distributed storage.
Second, active integration of family. BST trials emphasized minimal caregiver involvement. Braintrust brings family into information input but positions them as "input assistants." This hypothesis actively positions family as "updaters" (monthly information refresh) and "reinforcers" (family members record information in their own voices, which the patient hears). Identity information reinforced by family voices is stored alongside emotional circuits, allowing emotional recognition to persist even after cognitive circuits weaken.
Third, the element of environmental deprivation. The effect of reinforcement grows as unrelated stimulation decreases. BST addresses only reinforcement, not environment. This hypothesis pairs the reinforcement of core identity information with an environmental simplification that reduces unrelated stimulation (unnecessary notifications, noise, identity-irrelevant information). Raising the signal-to-noise ratio amplifies the reinforcement effect.
Fourth, the theoretical frame. BST places itself under "memory retention enhancement." This hypothesis targets "delay of the threshold of Alzheimer's progression itself, through thickening of the identity circuit." The difference of target determines clinical design and outcome measurement. The dissociation between general cognitive tests and identity-domain tests — which BST does not measure — becomes the core prediction of this hypothesis.
On top of these four differentiators, one decisive asset is added: the integrated protocol formed by adapting a learning methodology accumulated in Korea over a century to the conditions of old age.
5. Korean Memorization Methods: Adaptation to Old Age and Medical Application
The algorithm matured in America. But what is placed on top of the algorithm for daily operation developed differently across societies. Korea is the society in which the operational know-how of learning methods has been most intricately accumulated. Yet that know-how was developed for students in their twenties, and applied unchanged to Alzheimer's patients (average age 75+) it collides with cognitive, emotional, and physical conditions. Each method must be adapted to elderly patients. The tension of exam culture, the failure feedback, the speed-centered repetition, the information overload, the fatigue-inducing schedule — all are subject to adjustment. Below, the principle of each of the five methods is presented together with its adaptation for elderly patients.
5.1 The Error Notebook
Principle — A standard tool of Korean learning, in which a student collects only the problems she got wrong and studies them repeatedly. Two axioms: concentrate time on weak points; repeat the same problem until the accuracy rate stabilizes.
Adaptation for elderly patients — Exam culture treats "getting it wrong" as a learning stimulus, but for a 75-year-old patient the repeated experience of "I was wrong" induces frustration, lowered self-esteem, and worsening depression, which reduces compliance. Therefore the failure feedback is removed and only re-exposure remains. When the patient cannot recall her eldest son's name, instead of an "incorrect" display the photograph is shown again in a neutral tone of "let's look once more," and the recall failure is logged only to the family dashboard, not visible to the patient. From the patient's perspective there are no "problems she got wrong" — only "information she sees often." At the same time, channel intensity on failure-prone information is automatically raised: family voice added, photo enlarged, exposure time extended. BST shortens the interval on weak information at the algorithm level but does not attempt the Korean-style deliberate increase in channel intensity.
5.2 Consolidation into a Single Notebook
Principle — Compressing the essentials of many source materials into a single notebook. A hallmark of Korean exam-preparation culture, in which students condense multiple reference books, lecture notes, and mock-exam explanations into one notebook and repeat only that. The principle is to deliberately reduce the total volume of learning material and thereby maximize frequency of repetition.
Adaptation for elderly patients — A student can handle 30–50 items at once, but a patient in mild cognitive impairment cannot sustain more than about 15 at a time. The "layer" of the consolidated notebook is therefore divided into three. Layer 1 (the 10 core items — own name, spouse's name, children's names and faces, current residence, daily medications) is maintained from the time of diagnosis to the end. Layer 2 (the 15 extended items — close friends, frequented places, major life events) is added and rotated seasonally according to patient status. Layer 3 (contextual information) is offered only in early stages and dropped as progression advances. Exposure is limited to one item per screen, in large type (minimum 28pt), with short sentences, and low visual complexity. BST's Braintrust imposes no limit on the amount of information a user inputs. The layered structure of Korean consolidation actively adjusts information volume itself as patient status changes.
5.3 Recitation
Principle — A traditional Korean learning method of reading or memorizing information aloud. The seodang's chanting of the Thousand Character Classic, the school's multiplication tables, the recitation of religious scripture — all share the same principle: the deliberate addition of an auditory channel to visual learning, creating multichannel reinforcement.
Adaptation for elderly patients — A student's rapid recitation places little burden on vocal muscles and respiration, but a 75-year-old patient has diminished lung capacity, hearing, and vocal strength. Rapid recitation induces fatigue, coughing, and dizziness. Therefore recitation speed is slowed to less than half of ordinary conversation, and sentence length is limited to four or five morphemes: "my name is Kim Sun-ja," "my eldest son is Park Jeong-min." Total recitation time per day does not exceed two to three minutes. Seated in a comfortable posture, without becoming breathless. Patients who struggle to vocalize may follow along silently or with only a nod, and these are logged as valid learning. A form in which family recites first and the patient follows ("mother is Kim Sun-ja" first, "I am Kim Sun-ja" following) combines auditory, social, and emotional stimulation, raising efficacy without respiratory burden. BST does not address recitation at all.
5.4 Multichannel Association
Principle — Korean learning has developed mind maps, visual association, story construction, and song-lyric memorization. Words are not memorized alone but bound by visuals, stories, music, and relationships. The principle is to store information not as isolated nodes but as a network of multiple connections.
Adaptation for elderly patients — The complex mind maps used by students impose a heavy visual-cognitive load. For a 75-year-old patient, a "family genealogy" format — generationally familiar — is natural. At the center, a photograph of the patient herself; around her, photographs of spouse, children, grandchildren; under each photograph, only name and relationship in large type. No more than seven to nine people per screen. That photographs of grandchildren and relationships with daughters- and sons-in-law carry the strongest emotional stimulus for this generation is consistently confirmed across intergenerational research. This emotional stimulus is used as the central axis. Connection information follows a simple three-element structure: "eldest son Park Jeong-min — Busan — weekly Sunday call." BST's "name-face associations" remain at single-pair connections. The Korean multichannel association, adapted for old age, forms a family-genealogy network while minimizing visual load.
5.5 The Repetition Schedule
Principle — The Korean student's day operates on a repetition schedule. Morning self-study, school classes, private academy, lecture videos, evening self-study, late-night review. The same information is exposed at different times, in different places, in different formats within one day. The principle is that repeating the same information across varied contexts weakens context-dependent recall and strengthens free recall — the "distributed practice effect" established in learning psychology for over a century.
Adaptation for elderly patients — The packed schedule of a student rotating through academies and lectures induces fatigue and resistance in elderly patients. Therefore no new routine is created; the schedule is laid on top of existing daily routines. Two minutes of exposure during the morning medication routine, two minutes before lunch, two minutes after the evening family phone call. Six to eight minutes total, distributed across the day. The same information is exposed at different time slots, in different places, through different channels (morning in photographs, lunch in audio, evening in family conversation), converting context-dependent recall into free recall. The design's advantage is that the patient does not experience it as "study time." Learning is not a separate activity apart from daily life but a natural extension of existing routines. BST recommends one short session per day. The Korean repetition schedule, adapted for old age, operates as multiple short distributed exposures laid on top of existing routines.
5.6 Common Principles of the Old-Age Integrated Design
The common principles applied across the old-age adaptations of all five methods are as follows.
Emotional safety — The tension and evaluation of exam culture are removed, and the warmth of family is integrated as a channel. The patient must never feel she is being tested.
Consideration of physical limits — Large type (28pt and above), clear and slow speech, high contrast, low-frequency compensation (for hearing loss). The design must be usable by patients with hearing aids or reading glasses.
Dignity — Expressions like "you were wrong," "try again," "why can't you remember" are banned outright. All elements that could induce shame in the patient are removed. Companion expressions like "let's look together" and "let's look once more" replace them.
Relational design — Learning is framed not as a solitary activity but as a relational experience carried out with family. Family voices, family photographs, and family conversations become both the background and the material of learning.
Avoidance of overload — Total exposure under ten minutes per day, each session under three minutes, one piece of information per screen. Design so that sessions end before the patient feels fatigue.
When these common principles carry the old-age adaptations of the five methods, the accumulated know-how of Korean learning culture becomes an integrated protocol for the medical application of preserving the identity circuits of Alzheimer's patients.
6. Testable Predictions and Axes of Clinical Design
If the hypothesis is correct, the following outcomes should be observed.
First, the identity-information-reinforcement group should show significantly slower decline on identity-domain cognitive tests than controls. Measurement instruments consist of domain-specific tests: family-name recall, residence recognition, daily-routine recall, self-location awareness. Decline on general cognitive tests (MMSE, CDR-SB) is predicted to appear at similar rates in reinforcement and control groups, and this pattern becomes positive evidence for asymmetric preservation of the targeted circuits.
Second, the reinforcement effect should show a dose-response relationship with stimulation frequency and channel number. The preservation effect of a group receiving three distributed exposures per day should exceed that of a group receiving one; multichannel reinforcement (photographs, audio, family voice) should exceed single-channel reinforcement (text only).
Third, effect size should distribute across stages of intervention onset. The largest effects are predicted for asymptomatic at-risk individuals and MCI; clear effects for mild Alzheimer's; partial effects for moderate Alzheimer's (more in emotional stability and real-time self-identification than in cognitive scores); little effect for severe Alzheimer's.
Fourth, differential effects by ApoE ε4 status should be observed. ε4 carriers may show smaller effects under the same reinforcement because ε4 weakens synaptic plasticity itself. If preservation effects are clear even in ε4 carriers, the range of application of the hypothesis widens.
Fifth, neuroimaging changes in the reinforced circuits should be measurable. fMRI in the reinforcement group should show better-preserved volume and connectivity in regions activated during identity-information recall compared to controls; DTI (diffusion tensor imaging) should show better preservation of white-matter integrity in those circuits.
The axes of clinical research that can unfold the hypothesis are as follows.
"First axis, standardization of target definition." Standardization of a protocol for extracting 30–50 identity-information items per patient, applying the Korean principle of layered consolidation (10 core + 15 extended). A structured interview tool for joint use by family and patient, an algorithm for prioritization, and a standard for the cycle of information updates form the first area of work.
"Second axis, Korean-style adaptation of the reinforcement algorithm." On top of a baseline SM-2 or FSRS algorithm, a variant that integrates the Korean error-notebook principle (deliberate increase of channel intensity on weak-point information, removal of failure feedback) is compared in effect to the standard algorithm.
"Third axis, separable contribution of multichannel stimulation." The effects of each channel — photograph, audio, text, family voice, patient recitation — are measured separately, and the optimal channel combination is derived. The emotional-circuit activation effect of family-voice reinforcement requires separate measurement.
"Fourth axis, effect of the distributed exposure schedule." The preservation effect of a Korean-style repetition schedule (three or more distributed exposures layered onto existing routines) is compared to a single daily exposure. Patient compliance and fatigue are tracked jointly.
"Fifth axis, multiplicative effect with pharmacological therapy." In patients receiving lecanemab or donanemab, the question of whether the additional application of a Korean-style integrated protocol produces multiplicative effects on cognitive preservation is measured. The potential synergy of molecular-level intervention with circuit-level intervention carries the most significant clinical implication.
These five axes can proceed independently, and as their results accumulate, the hypothesis takes a form ready to enter as an item in a Korean-type Alzheimer's cognitive intervention guideline.
Closing — Background of the Work
The person who wrote this essay is not a medical specialist. He is the representative of a small company called Wonbrand, whose work is to look at social problems from the position of an ordinary citizen and try to build something. One line has guided that work: hongik ingan (弘益人間), to benefit humanity broadly.
This work is the eighth essay in a series of seven prior works conducted with the same method. The prior works are as follows.
1. "Twenty-Five Intuitions on Cancer" — a meta-analysis showing that 23 of 25 free intuitions about cancer aligned with active academic research areas.
2. "Four Questions About Aging That No One Has Properly Asked Yet" — identification of four empty positions in aging biology.
3. "How Anesthesia Turns Off Consciousness" — reflections on the molecular and circuit mechanisms of loss of consciousness.
4. "WWW Theory — How Humans Should Survive in the Age of AI" — a hypothesis on the three axes of human-AI coexistence.
5. "The Aesthetics of Subtraction — Seventeen Proposals for LLM Development" — an outsider perspective on the design of language models.
6. "ADHD Must Box" — a methodological proposal for ADHD intervention.
7. "Depression: The Invisible Wound" — reflections on the molecular and social layers of depression.
This "Alzheimer's Identity Preservation Hypothesis" is the eighth work. Meta-analysis across the works reveals a pattern: the free thought of a layperson frequently aligns with or points to positions less addressed in expert research across medicine, neuroscience, AI, and society. This alignment and these empty positions are not coincidence but cumulative evidence for the hypothesis that, when a problem is approached seriously, there are answers one naturally arrives at.
References
Hardy, J., & Higgins, G. A. (1992). Alzheimer's disease: the amyloid cascade hypothesis. Science, 256(5054), 184-185.
Nelson, P. T., et al. (2012). Correlation of Alzheimer disease neuropathologic changes with cognitive status: a review of the literature. Journal of Neuropathology & Experimental Neurology, 71(5), 362-381.
Altmann, A., et al. (2015). Regional brain hypometabolism is unrelated to regional amyloid plaque burden. Brain, 138(11), 3734-3746.
van Dyck, C. H., et al. (2023). Lecanemab in early Alzheimer's disease. New England Journal of Medicine, 388(1), 9-21.
Sims, J. R., et al. (2023). Donanemab in early symptomatic Alzheimer disease: the TRAILBLAZER-ALZ 2 randomized clinical trial. JAMA, 330(6), 512-527.
Volloch, V., & Rits-Volloch, S. (2023). The Amyloid Cascade Hypothesis 2.0: generalization of the concept. Journal of Alzheimer's Disease Reports, 7(1), 21-35.
Braak, H., & Braak, E. (1991). Neuropathological stageing of Alzheimer-related changes. Acta Neuropathologica, 82(4), 239-259.
Stern, Y. (2012). Cognitive reserve in ageing and Alzheimer's disease. Lancet Neurology, 11(11), 1006-1012.
Livingston, G., et al. (2024). Dementia prevention, intervention, and care: 2024 report of the Lancet standing Commission. Lancet, 404(10452), 572-628.
Ball, K., et al. (2002). Effects of cognitive training interventions with older adults: a randomized controlled trial (ACTIVE). JAMA, 288(18), 2271-2281.
Belleville, S., et al. (2024). Five-year effects of cognitive training in individuals with mild cognitive impairment. Alzheimer's & Dementia: Diagnosis, Assessment & Disease Monitoring, 16(3), e12626.
Butler, R. N. (1963). The life review: an interpretation of reminiscence in the aged. Psychiatry, 26(1), 65-76.
Cuc, A. V., et al. (2021). Boosting autobiographical memory and the sense of identity of Alzheimer patients through repeated reminiscence workshops. Frontiers in Psychology, 12, 636028.
Camp, C. J. (2006). Spaced retrieval: a model for dissemination of a cognitive intervention for persons with dementia. Geriatric Mental Health Care, 1(1), 12-19.
Bourgeois, M. S. (1993). Effects of memory aids on the dyadic conversations of individuals with dementia. Journal of Applied Behavior Analysis, 26(1), 77-87.
Smith, A. M., et al. (2024). Algorithmic Spaced Retrieval Enhances Long-Term Memory in Alzheimer Disease: Case-Control Pilot Study. JMIR Formative Research, 8, e51943.
Bliss, T. V. P., & Lømo, T. (1973). Long-lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit. Journal of Physiology, 232(2), 331-356.
Hebb, D. O. (1949). The Organization of Behavior. Wiley.
Ebbinghaus, H. (1885). Über das Gedächtnis. Duncker & Humblot.
Tonegawa, S., et al. (2015). Memory engram storage and retrieval. Current Opinion in Neurobiology, 35, 101-109.
An Seungwon / Wonbrand / https://wonbrand.co.kr