A woman in her late seventies can no longer remember what she ate for breakfast an hour ago. She has already asked her daughter the same question three times this morning. Yet when her daughter turns on a record from the 1960s, she sings every verse without missing a word, and she can still describe, in vivid detail, her wedding day from fifty years earlier.
This pattern shows up so often in dementia that families tend to accept it as simply how the disease works. But it raises a genuinely strange question. If dementia is a disease of the brain, why would it destroy the ability to remember a conversation from five minutes ago while leaving decades-old memories, language, and long-practiced skills largely untouched?
The answer has less to do with memory as a single faculty and more to do with where the disease begins. Alzheimer’s disease, the most common cause of dementia, does not spread through the brain randomly. It tends to start in a small, specific region responsible for turning new experiences into lasting memories — and only reaches everywhere else much later.
A Puzzle Hidden in Plain Sight
Memory is not stored in one place, and it is not created through one process. Neuroscientists generally separate memory into distinct systems: the ability to form new memories, the ability to retrieve old ones, procedural memory for skills like riding a bicycle, and semantic memory for general facts about the world.
Early Alzheimer’s disease disrupts these systems unevenly. The capacity to form new memories usually fails first and fails hardest. Long-term memories, particularly ones formed decades earlier, often survive well into the disease. Skills that have become automatic through years of repetition, such as playing a familiar instrument, can remain intact even when a person struggles to recall what year it is.
This unevenness is the real clue. It suggests that Alzheimer’s disease is not attacking “memory” as a whole. It is attacking the specific brain structure responsible for producing new memories in the first place.
The Brain’s Filing Room
To understand why, it helps to know how a memory is normally created. When something happens to you, information about the event arrives from several different sensory and cognitive systems scattered across the brain. On its own, this information is fragmented and temporary.
A structure deep in the brain called the hippocampus, along with a neighboring region called the entorhinal cortex, acts as a kind of coordinating hub. Together, these structures bind fragments of an experience into a single, retrievable memory and gradually help transfer a stable version of it out to the wider cortex for long-term storage.
This means the hippocampus and entorhinal cortex are essential for creating new memories, but they are not the only place where old memories eventually live. Once a memory has been consolidated and distributed across the cortex, it can, in many cases, be retrieved even if the hippocampus itself is later damaged. New encoding and old retrieval depend on different, only partially overlapping neural machinery.
That distinction explains a great deal about the early experience of Alzheimer’s disease. Damage that lands squarely on the hippocampus and entorhinal cortex would be expected to disrupt the formation of new memories far more severely than the retrieval of memories formed years before the disease began. This is close to what researchers actually observe.
Where the Damage Begins
Alzheimer’s disease is defined by two hallmark forms of brain pathology: sticky clumps of a protein fragment called beta-amyloid that build up between neurons, and twisted fibers of a different protein called tau that accumulate inside neurons. Both processes disrupt normal cell function and are associated with the eventual death of neurons and the loss of connections between them.
What matters for the question of memory is not simply that this damage occurs, but where it tends to occur first. Post-mortem studies of Alzheimer’s brains, most influentially a staging system developed by the German neuropathologists Heiko and Eva Braak in the early 1990s, found that tau pathology tends to appear in a fairly consistent sequence. It typically emerges first in the entorhinal cortex and hippocampus, spreads next to nearby structures involved in emotion and association, and only reaches the broader outer layers of the cortex — the regions responsible for language, reasoning, and complex planning — in the later stages of the disease.
This staged progression offers a structural explanation for the clinical pattern families observe. The disease begins its work almost exactly where new memories are made, well before it reaches the regions responsible for vocabulary, long-practiced skills, or the storage sites of decades-old memories. The order in which symptoms typically appear largely tracks the order in which the underlying pathology spreads.
Why Encoding Breaks Before Retrieval
There is also a more basic reason new memory formation is especially vulnerable: it is a more demanding, more fragile process than simply pulling up something already stored.
Creating a new memory requires the hippocampus to actively bind together many strands of incoming information in real time, a task that depends on healthy, rapidly firing neural circuits and functioning synaptic connections. Retrieving an old memory, once it has been fully consolidated elsewhere in the cortex, places fewer demands on that same fragile machinery.
Research on Alzheimer’s pathology has consistently found that the loss of synapses — the junctions where neurons communicate — correlates with cognitive decline more closely than the raw quantity of amyloid plaques does. This supports the idea that it is the breakdown of active neural communication, not merely the presence of abnormal proteins, that produces the clinical symptoms families notice. A demanding, synapse-intensive process like new-memory formation would be expected to falter earlier than a comparatively less demanding one like retrieving a well-worn memory.
What Popular Understanding Gets Wrong
It is worth being precise here, because “dementia” and “Alzheimer’s disease” are not interchangeable, even though they are often used that way in everyday conversation. Alzheimer’s disease is the most common cause of dementia, accounting for a majority of cases, but it is not the only one — and memory loss is not always the first sign.
Frontotemporal dementia, a less common but still significant cause of dementia, typically begins in the frontal and temporal lobes rather than the hippocampus. Its earliest symptoms often involve changes in personality, judgment, or social behavior rather than memory. A person in the early stages may become uncharacteristically impulsive or emotionally blunted while still forming new memories relatively normally.
Dementia with Lewy bodies frequently announces itself through visual hallucinations, fluctuating attention, and movement changes resembling Parkinson’s disease, with memory problems sometimes arriving later. Vascular dementia, caused by reduced blood flow to the brain, can produce a much more variable pattern depending on which blood vessels and brain regions are affected.
The popular idea that “dementia always starts with forgetting” is really a description of Alzheimer’s disease specifically, generalized to a much broader and more varied category of conditions. Even within Alzheimer’s disease, some early-onset and atypical forms can present first with language difficulty or visual-spatial problems rather than classic memory loss.
The Limits of What Scientists Know
Despite decades of research, important uncertainties remain. The amyloid hypothesis, which holds that amyloid buildup is the primary trigger of the disease process, has dominated research funding and drug development for years, but it has faced significant challenges. Some people accumulate substantial amyloid plaques without developing dementia symptoms, while others show significant cognitive decline with comparatively modest plaque levels. Many researchers increasingly view tau pathology, and its close relationship with actual neuron and synapse loss, as more directly tied to symptom severity than amyloid alone.
It also remains unclear why some people appear resistant to the disease’s effects even when scans reveal substantial underlying pathology. Researchers have proposed a concept sometimes called cognitive reserve — the idea that education, complex occupations, and sustained mental engagement may help some brains tolerate a greater degree of damage before symptoms become apparent. The evidence for this idea is suggestive rather than conclusive, and the underlying biological mechanism is still debated.
Finally, the precise sequence of cause and effect within the disease — whether amyloid triggers tau pathology, whether both arise from a shared upstream cause, or whether the relationship is more circular — remains an active and unsettled area of research rather than a fully solved question.
Why This Understanding Matters
Knowing that Alzheimer’s disease follows a predictable anatomical path has practical consequences that extend well beyond satisfying curiosity. It has shaped how clinicians think about early diagnosis, since subtle difficulty forming new memories — rather than problems with language or long-term recall — is often the most useful early warning sign to test for.
It has also shaped the direction of biomarker research, including brain imaging techniques and spinal fluid tests designed to detect amyloid and tau changes years before symptoms become obvious. If the disease reliably begins in the entorhinal cortex and hippocampus, then imaging techniques sensitive to changes in those specific structures offer a plausible route toward earlier detection, at a stage when future treatments might have a better chance of preserving function.
Understanding the staged, structural nature of the disease also reframes what families experience emotionally. A person’s inability to remember what happened five minutes ago, paired with their vivid recollection of a wedding from decades before, is not a sign of selective forgetting or a lack of effort. It is a direct reflection of which brain structures have been affected and which have, for the time being, been spared.
The Larger Picture
Dementia does not erase memory the way a hand might wipe a chalkboard clean all at once. It works more like a light dimming in one specific room of the house before gradually reaching the others — beginning in the small, hardworking structure responsible for turning experience into memory, and only later spreading toward the rooms where language, personality, and long-held memories reside.
That order matters. It explains why a person who cannot recall a conversation from an hour ago may still recognize the faces of people who mattered to them decades earlier, and why families so often describe the disease as taking someone away gradually rather than all at once. Recognizing the specific, structural origin of that pattern does more than satisfy scientific curiosity — it shapes how doctors detect the disease early, how researchers search for treatments, and how families make sense of what they are watching happen to someone they love.
The most important insight may be this: memory loss in dementia is not a single failure but the visible edge of a much more specific, traceable process — one that begins in one small structure of the brain and only gradually claims the rest.
Frequently Asked Questions
Is occasional forgetfulness a sign of dementia?
Not usually. Occasionally misplacing keys or forgetting a name is a normal part of aging and everyday cognitive load. Dementia-related memory loss is typically more persistent, progressive, and disruptive to daily functioning, such as repeatedly forgetting recent conversations or becoming lost in familiar places.
Does every type of dementia begin with memory loss?
No. Alzheimer’s disease, the most common cause of dementia, usually begins with new-memory problems. Other forms, including frontotemporal dementia and dementia with Lewy bodies, often begin with changes in personality, behavior, attention, or movement instead.
Can the brain changes behind Alzheimer’s disease be detected before symptoms appear?
Researchers can now detect signs of amyloid and tau accumulation years before clinical symptoms emerge, using brain imaging and other biomarker tests. However, these tools are primarily used in research and specialized clinical settings, and detecting pathology does not mean dementia is certain to follow.
Why do old memories survive longer than new ones in Alzheimer’s disease?
Old memories, once fully consolidated, are stored more broadly across the cortex and depend less on the hippocampus for retrieval. New memories require the hippocampus and entorhinal cortex to actively form them, and these are typically the first structures affected by the disease.
