Memory: How the Brain Holds the Past

Yesterday we found the mechanism by which experience rewires a synapse. Today we ask what that rewiring is for — because if the brain can change its own wiring with use, then memory is simply what that change stores. And here is the strange, beautiful truth the science keeps returning: a memory is not a file the brain saves and plays back. It is a pattern of altered synapses that, when re-activated, re-creates a version of the past. The recollection is rebuilt, not retrieved — and, remarkably, the act of remembering briefly returns the memory to a changeable state, so that each time you recall something, you quietly rewrite it. The past you carry is held not as a recording but as a living change in the wiring of your own brain.

Not one thing, but several

The first thing to understand about memory is that there is no such single thing. What we call "memory" is a family of distinct systems that can be damaged separately. There is working memory — the small, seconds-long scratchpad that holds a phone number while you dial it, limited to only about four items at a time in modern estimates (the famous "seven, plus or minus two" was revised downward once psychologists controlled for rehearsal and chunking). There is declarative or explicit memory — the conscious recall of facts and events — and there is non-declarative or implicit memory, including the procedural skills, like riding a bike or typing, that you know in your hands without being able to say how. The most dramatic proof that these are genuinely separate came not from a theory but from a single patient.

The man who could not make new memories

In 1953, a young man named Henry Molaison — known for decades only as H.M. — underwent surgery to control severe epilepsy. The surgeon removed much of his medial temporal lobe on both sides, including most of the hippocampus. The seizures improved. But so did something no one had anticipated: from that day forward, H.M. could not form new conscious memories. He could hold a conversation, but minutes later have no memory of it. He met his doctors thousands of times, each time as a stranger. His intelligence was intact, his childhood memories mostly preserved, his working memory fine — but the machinery for laying down new declarative memories was gone.^[2] Scoville and Milner's 1957 report of his case is one of the most important papers in the history of the mind.

And then came the twist that split memory into pieces. The neuropsychologist Brenda Milner had H.M. trace a star while watching his hand only in a mirror — an awkward task everyone improves at with practice. Day after day, H.M. got better at it, his motor skill steadily climbing — even though, each day, he had no memory whatsoever of having done it before, and insisted he was trying it for the first time. His hands were learning while his conscious self recorded nothing. That single dissociation proved that declarative memory (which he had lost) and procedural memory (which he kept) are different systems, in different circuits — and that the hippocampus is essential for the former but not the latter. Memory is not one faculty. It is many.

A memory, molecule by molecule

But what is the trace, physically? The answer came from an unlikely creature: a giant sea slug called Aplysia, chosen by Eric Kandel precisely because its nervous system is simple — only about twenty thousand large, identifiable neurons. Studying the slug's gill-withdrawal reflex, Kandel showed that learning is a change in the strength of specific synapses — and, crucially, that short-term and long-term memory are built by physically different means.^[1] A short-term memory, lasting minutes to hours, is a transient chemical modification of proteins already present at the synapse: signaling through cAMP and the enzyme PKA tweaks existing channels and release machinery so the synapse fires more strongly. Nothing new is built. A long-term memory is different in kind: the signals travel all the way into the cell's nucleus, switch on genes through a transcription factor called CREB, build new proteins, and grow new synaptic connections. The rule is startlingly clean — block protein synthesis and you block the formation of long-term memory while leaving short-term memory untouched. For revealing how memory is written into the molecules of the synapse, Kandel shared the 2000 Nobel Prize in Physiology or Medicine (with Arvid Carlsson and Paul Greengard), "for their discoveries concerning signal transduction in the nervous system." As the Nobel Assembly put it, our memory can be said to be "located in the synapses."

1. Step 1 · EncodingThe hippocampus binds an experienceA salient event drives coordinated activity; the hippocampus links the scattered cortical fragments of an episode into one pattern.
2. Step 2 · Short-term strengtheningA quick chemical changeSignaling (cAMP → PKA) modifies proteins already at the synapse, strengthening it for minutes to hours — no new protein needed.^[1]
3. Step 3 · Long-term consolidationNew genes, new synapsesStronger signals reach the nucleus, activate CREB, build new proteins, and grow new connections — a durable structural trace (block protein synthesis and it fails).
4. Step 4 · Systems consolidationHanded off to the cortexOver days to years the memory becomes distributed across neocortex and less dependent on the hippocampus — why H.M.'s old memories survived.
5. Step 5 · Reconsolidation on recallRemembering is re-writingRetrieval returns the memory to a labile state that again needs protein synthesis to re-stabilize — so each recall can subtly update it.^[4]

The hunt for the engram

The physical trace of a memory has a name — the engram, coined around 1904 by Richard Semon — and finding it became one of the great quests of twentieth-century biology. The psychologist Karl Lashley spent some three decades cutting and removing pieces of rat cortex, trying to locate where a given memory was stored. He never found it; wherever he looked, the memory stubbornly survived in what remained, leading him to conclude that the trace is distributed, spread across the brain rather than filed in one place. His 1950 essay reporting this defeat is famous for its rueful humor.

I sometimes feel, in reviewing the evidence on the localization of the memory trace, that the necessary conclusion is that learning just is not possible. — Karl Lashley, "In Search of the Engram," 1950

Learning is, of course, possible; Lashley's joke marks the limit of the tools of his era, not of the brain. What he could not find, modern neuroscience can now grab hold of. In 2012, Susumu Tonegawa's laboratory used optogenetics — the light-gated ion channels we met earlier — to genetically tag the specific neurons that were active while a mouse formed a fear memory, and then, later, to switch just that set of neurons back on with a pulse of light. The mouse behaved as if remembering the fear, on cue.^[3] A follow-up study went further and implanted a false memory, pairing the artificial reactivation of one context with a shock experienced in another. Lashley's ghostly, distributed engram had become something a researcher could tag, trigger, and even forge. It is breathtaking work — and it is, to be exact, mouse neuroscience, requiring genetic engineering and implanted optical fibers, not anything close to a human therapy.

Where the honesty lives

Two honesties close this out, and both matter. The first is scientific humility about memory itself: it is reconstructive, not a recording. Because retrieval reopens a memory to revision, and because the brain fills gaps with plausible inference, memories drift, blend, and can be confidently, vividly false — as the psychologist Elizabeth Loftus demonstrated by implanting entirely fabricated childhood events in a sizable fraction of ordinary volunteers. Your memory is not a camera, and the sooner marketing stops promising that it could be one, the better. Which brings the second honesty: the frontier here — engram editing, and reconsolidation-based approaches to easing traumatic memories — is genuinely exciting and genuinely unfinished, an active research program in animals and early trials, not a menu of available treatments.

And so to the overclaim this Journal exists to name. Memory is built by specific molecular and structural events at particular synapses — CREB switching on genes, proteins being synthesized, spines growing. A generic pill or a "frequency" that promises to boost your memory wholesale has nothing coherent to act on; it points at a target the mechanism does not contain. The most visible example is instructive: in 2017 the U.S. Federal Trade Commission and the New York Attorney General charged the marketers of the memory supplement Prevagen with deceptive claims, alleging that the company's own study had "failed to show that Prevagen works better than a placebo on any measure of cognitive function."^[6] "Photographic memory," in the literal sense, is largely a myth; the memory champions who recite thousands of digits are using trained mnemonic strategies, not a mental camera. Real memory is specific, effortful, and reconstructive — which is exactly why a gadget or supplement promising to boost it is a non-sequitur, not a discovery.

Quick answers

Bioelectric Mechanisms · The resting potential · The channels · The spike · The synapse · The rewiring · The trace · Biofield Hub →

Tomorrow on the Journal

Day 65 — Sleep and the Filing of Memory. The consolidation we traced today does much of its work while you are unconscious. Next: how the sleeping brain replays the day, moves memories from hippocampus to cortex, and why a night's rest is part of the machinery of remembering.