Phantom Limbs: Why the Brain Keeps a Map of What Is Gone

Introduction

In 2016 a group of researchers put people into an MRI scanner and asked them to move the fingers of a hand that had been amputated years earlier [1].

They could not move it. Nobody can move a hand that is not attached. But they could try, and trying is a brain event, and brain events show up on a scanner.

What came back was not noise. It was a map. Index finger here, middle finger there, ring finger next to it, each one in its own small patch of cortex, laid out in the same order as the fingers of the hand you are reading this with. The pattern was specific enough to tell the fingers apart. In some cases the limb had been gone for decades.

The hand was still in there. Filed, indexed, and apparently waiting.

That result is a problem, because it contradicts what most people are told about phantom limbs. The standard explanation, the one on hospital websites and in medical school slides, says the opposite. It says that when a limb is amputated, its patch of brain goes silent and the neighbours move in. The face invades the hand's territory. The map is overwritten. And the overwriting, the story goes, is what produces the pain.

For about twenty years that was the answer. It is a good answer. It is elegant, it explains a bizarre clinical observation, and it comes with a treatment attached. It is also now the subject of one of the longest running arguments in clinical neuroscience, and if you read the pages that rank highest for this topic you would not know the argument existed.

This article is about that argument. Not just about what a phantom limb is, though we will get to that, but about how a field convinced itself of something, spent two decades building on it, and then had to look again. Along the way there is a woman born without forearms who nonetheless felt them. There is a Civil War surgeon who named the phenomenon in 1872. There is a trial where both groups got better and nobody could say why. And there is the awkward fact that the two scientists who disagree most sharply about the cause eventually sat down and wrote a review together.

What a Phantom Limb Actually Is

Start with the vocabulary, because three different things get blurred together constantly and the blurring makes the research impossible to read.

The first is a phantom sensation, meaning the felt presence of a body part that is no longer there. It is not necessarily unpleasant. People report position, temperature, weight, itch, the sense that a hand is clenched or that toes are wiggling. Many describe it as simply having the limb, in the ordinary way you currently have your left foot without thinking about it.

Then there is phantom limb pain, which is pain that seems to come from the missing part. Burning, cramping, shooting, crushing. Sometimes it feels like the limb is locked in the position it was in at the moment of injury.

The third is residual limb pain, often called stump pain, and it happens in the tissue that remains. It has different causes from the other two, and treating it as the same thing is a genuine and frequent error [2].

The blurring matters for a practical reason. When you see a claim that some very high percentage of amputees experience phantom limb pain, check which of the three the number refers to. Most phantom experiences are sensations rather than pain. The largest modern picture comes from a nationwide survey of 3374 unilateral limb amputees, which is a sample size that makes the older single-clinic studies look like pilot work [3].

There is a second reason to keep the three straight. If you conflate sensation with pain, you end up believing that nearly everyone who loses a limb is in constant agony. That is not what the research says, and it is not what most people with amputations report.

Two Danish studies from the 1980s still anchor everything known about the time course. The first followed amputees through their first six months [4]. The second followed them for two years and found something that turned out to matter enormously: pain in the limb before the amputation predicted phantom pain after it [5].

Hold onto that. It comes back later.

The short version of the time course is that phantom sensations usually begin almost immediately, often within days. Phantom pain tends to be worst early and to decline. It does not usually disappear. A phenomenon that fades but persists is a strange thing to explain, and it has constrained every theory since [6].

The Surgeon Who Named It

The phenomenon is older than the name. Ambroise Paré, a French military surgeon working in the sixteenth century, described patients complaining of pain in limbs he had removed himself. Battlefield medicine produced a great many amputees and a great many observations.

The name came later. Silas Weir Mitchell was an American physician who spent the Civil War at a Philadelphia hospital that specialised in nerve injuries. He saw enough amputees to notice the pattern, and in 1872 he published a book on nerve injuries in which he called it a phantom limb [7].

Mitchell's phrase was better than he knew. He did not mean it as a ghost story. He meant something present but not solid, felt but not touchable. That is precisely what the modern imaging work keeps finding.

For most of the century that followed, phantom limbs were a curiosity. They were catalogued, described, occasionally dismissed as psychological. What changed was not a discovery about phantoms. It was a discovery about maps.

The Map Inside Your Head

Your brain contains a map of your body surface. Not a metaphorical map. An actual spatial arrangement, running in a strip across the top of the brain, in which adjacent patches of cortex handle adjacent patches of skin.

This was worked out the hard way. Neurosurgeons operating on conscious patients stimulated points on the exposed cortex and asked what the patient felt, and recorded evoked responses directly from the surface [8]. The result was a body laid out along the strip, distorted in proportion to sensitivity rather than size. Lips and hands get enormous territory. The back of your thigh gets almost none.

The map is not drawn to scale, and it is not drawn in the order you would expect. The hand sits next to the face. That single piece of anatomy is about to do a lot of work.

Here is the question that produced everything that follows. If a hand is removed, the patch of cortex that used to receive signals from it is still there. It has not been damaged. It has simply stopped receiving mail. What does a piece of brain do when its input stops?

Three answers have been offered. All three are still in play. The rest of this article is the story of how each one was argued for, and what happened when the evidence arrived.

Touch the Cheek, Feel the Thumb

The first serious clue was an observation so odd that it sounds made up.

In the early 1990s, V.S. Ramachandran and colleagues took a cotton swab to the faces of people who had lost an arm. When they touched particular spots on the cheek, the patient reported feeling the touch in the missing hand. Not instead of the cheek. As well as it. Touch here, feel the thumb. Touch a few centimetres away, feel the index finger [9].

There was a whole hand mapped out on the side of the face. It was consistent, it was reproducible, and in some patients a second map appeared on the upper arm above the amputation line. Drops of warm water running down the cheek were felt running down the phantom [10].

Try to imagine what that is like from the inside. Someone brushes your cheek and you feel your thumb. Not a memory of your thumb. Your thumb, right now, being touched. The patients were not confused about where the swab was. They felt both.

Now look back at the body map. The face sits next to the hand. If the hand's territory went quiet and facial inputs spread into it, then touching the face would activate cells that the brain has always interpreted as hand cells. The brain would report a hand.

That is a beautiful piece of reasoning, and it is worth pausing on why it was so persuasive. It took a bizarre subjective report and explained it with known anatomy. No new mechanism required. Just a map, a gap, and the neighbours moving in.

Ramachandran laid the whole picture out in a lecture published in 1998 that is still one of the most readable things written on the subject [11]. The idea had a name by then: cortical reorganization. And it had just acquired its most important piece of evidence.

1995: The Correlation That Became a Textbook

You now have a theory that explains a symptom using anatomy you can point at. That is the strongest position a young idea can be in. What it still lacked was a number.

In 1995, Herta Flor and colleagues published a paper in Nature that changed how the whole field talked about phantom pain [12].

They used magnetic imaging to measure how far the mouth representation had shifted into former hand territory in upper-limb amputees. Then they compared that shift against how much phantom pain each person reported.

The correlation was 0.93.

In human neuroscience, that is an extraordinarily strong number. Correlations that tight are rare enough that people remember them. And the direction was exactly what the theory predicted: the more the map appeared reorganized, the worse the pain.

This is the moment the idea became doctrine. It acquired a name that carried a judgement inside it, maladaptive plasticity. The brain's capacity to rewire itself, the same capacity that lets you learn an instrument or recover from a stroke, was here doing harm. The rewiring was the disease.

You can see why it spread. It explains the pain. It explains the referred sensations. It connects phantom limbs to a broader story about how experience shapes cortex, which was the most exciting idea in neuroscience at the time.

Supporting evidence arrived steadily. Reorganization showed up in motor cortex as well as sensory cortex [13]. Blocking the nerves with regional anesthesia changed the pain and the cortical picture together [14]. Imaging during attempted phantom movements showed shifted activation [15]. Transcranial magnetic stimulation mapping showed the same thing from a different angle [16].

Notice the shape of that evidence. Different labs, different machines, different measurements, all pointing the same way. When you see a run like that you are looking at a field converging, and convergence feels a great deal like confirmation. It is not quite the same thing, and the difference only becomes visible later.

By 2006 the theory had a review article in Nature Reviews Neuroscience summarising it. The title asked whether phantom limb pain was a case of maladaptive plasticity, with a question mark [17]. The question mark did not survive the journey into clinical teaching. What reached the handouts was a statement.

The same logic seemed to apply well beyond limbs. Tinnitus, the ringing that follows hearing damage, was linked to reorganization in auditory cortex [18]. Neuropathic pain after spinal cord injury was linked to reorganization in somatosensory cortex [19]. A general principle was taking shape: deprive a sensory area of input, and the resulting reorganization hurts.

This is a good place to say something about how the brain's rewiring capacity is usually described. If you have read about how the brain rebuilds itself after injury, you have met plasticity as the hero of the story. Phantom limb pain was the case that made people ask whether it could also be the villain.

The Theory Comes With a Treatment

A theory that explains a symptom is interesting. A theory that suggests a cure is powerful, and this one did.

If the problem is a mismatch between what the brain expects from the limb and what it receives, then feeding the brain a convincing image of an intact, moving limb should help. In 1996 Ramachandran and Rogers-Ramachandran described exactly that: a box with a mirror down the middle, positioned so the reflection of the intact hand appears where the missing one should be [20].

Move the good hand, watch the reflection, and see the phantom move.

Patients reported unclenching phantom fists that had been painfully closed for years. It is a striking clinical story, and it needed no drugs, no surgery, and almost no money. A mirror.

By 2007 a small controlled trial in the New England Journal of Medicine reported that mirror therapy reduced phantom pain more than covered-mirror or mental-imagery controls [21]. Related approaches followed. Graded motor imagery, a staged programme working from recognising limb images through imagined movement to mirror work, was tested on chronic pain conditions [22]. Sensory discrimination training, where patients learn to identify electrical stimuli on the stump, reduced pain and shifted the cortical map at the same time, which looked like direct confirmation of the mechanism [23].

One quick note before going further, because the naming is genuinely confusing. The mirror box has nothing to do with mirror neurons. Different mechanism, unrelated discovery, similar word. The box works with reflected vision. Mirror neurons are cells that fire both when you act and when you watch someone else act. If a page tells you one explains the other, it is guessing.

So by the mid-2000s the field had a mechanism, a correlation of 0.93, a treatment, and a growing body of supporting imaging. This is usually the point where a story ends.

2013: The Correlation Runs Backwards

Except that it did not end, and the reason is worth stating plainly. A correlation tells you two things move together. It does not tell you which one is doing the moving, or whether a third thing is moving both. Everyone involved knew that. The problem is that once a mechanism sounds right, the caveat stops getting repeated, and by the time it reaches a patient handout it has vanished entirely.

Tamar Makin's group scanned upper-limb amputees and looked at what remained of the hand representation. The prediction from twenty years of theory was straightforward. People with the worst phantom pain should show the most degraded, most invaded hand area.

They found the opposite.

The amputees reporting the most phantom pain had the best preserved structure and function in the former hand area [24]. Not the most overwritten. The most intact.

Read that twice, because the significance is easy to skate over. This is not a failure to replicate. It is not a null result. It is the same relationship with the sign flipped. If pain tracks preservation rather than erasure, then the phrase maladaptive plasticity is describing something backwards.

A follow-up in 2015 went after the reorganization claim directly, reassessing what happens in primary sensorimotor cortex after arm amputation and finding much less large-scale takeover than the standard account requires [25]. Then came the finger-mapping study this article opened with, showing the individual digit layout of a missing hand still legible in cortex years later [1].

The table is a summary, and summaries flatten things. It is worth being precise about how far the challenge goes.

It does not say reorganization never happens. Changes after amputation are real and measurable. What is disputed is the scale of those changes, and whether they cause the pain or merely accompany it. A systematic review of fMRI studies published in 2025 reported a dissociation between motor cortex reorganization and pain intensity, which is the sort of quiet finding that does more damage to a theory than any single dramatic result [26].

It is also not a new complaint. As early as 1998 a paper in Brain asked whether the perceptual correlates of reorganization were stable enough to bear the weight being put on them, under the pointed title "Plasticity of plasticity?" [27]. That paper was not ignored so much as absorbed. The doctrine kept moving.

The Congenital Problem

There is a group of people whose existence has been quietly awkward for the reorganization theory since long before the imaging debate started.

Some people are born without a limb. If a phantom is a map built from a limb's input, and there was never any input, there should be no map and therefore no phantom.

In 1997 Ronald Melzack and colleagues reported on 125 people with missing limbs. Among them, 41 individuals experienced phantoms despite either being born limb-deficient or having lost the limb before the age of six. Fifteen of those were congenital cases. Twenty-six had been amputated in early childhood [28].

Their numbers put it plainly. At least twenty percent of the congenitally limb-deficient participants experienced phantoms, against fifty percent of those amputated before the age of six.

Among those who did report phantoms, the proportions describing them as painful were about twenty percent in the congenital group and forty-two percent among the young amputees. Note that this second pair of figures counts only people who had phantoms in the first place, which is a different denominator from the bars above and worth keeping straight.

A phantom of a limb that never existed cannot be a memory of that limb. So where does it come from?

Melzack's answer, proposed in 1990, was the neuromatrix [29]. His suggestion was that the brain carries a built-in template of a whole body, distributed across several regions, partly genetically specified rather than assembled entirely from experience. On this view the body you feel yourself to have is generated from the inside and then corrected by sensory input, rather than being built up from that input in the first place. Remove the input and the template keeps running.

It is worth slowing down on what that proposal is really claiming, because it inverts the intuition most people start with. You probably assume your sense of having a body is assembled from the signals your body sends. The neuromatrix says the assembly runs the other way. The template comes first and the signals correct it.

If that is right, then a phantom is not a malfunction. It is the template running without correction, which is exactly what it would do.

The most vivid single case is a woman born without forearms and without legs who nonetheless reported phantom limbs, studied with imaging and magnetic stimulation and published in 2000 [30]. That is a single patient. One person. A case like that cannot establish a prevalence or settle a mechanism, and any article treating it as if it could is overreaching. What it can do is rule things out, and a phantom in someone who never had the limb rules out any account that requires the limb's own input to have built the map.

Single cases get treated badly in popular writing. They are either dismissed as anecdote or promoted into proof, and neither is right. What a single case does well is kill a universal claim. If your theory says something can never happen, one clear instance is enough.

A direct comparison of congenital and traumatic amputees followed in 1998, looking at cortical reorganization and phantom phenomena in both groups [31]. Modern prevalence figures for congenital and early-life amputees arrived much more recently [32].

Then, in 2022, a study did something that should have been done years earlier. It compared the facial remapping story in people born without a hand against people who had lost one. The complex facial reorganization showed up following congenital hand absence, and not following acquired hand loss [33].

Sit with the shape of that. The invasion is clearest in the group that mostly does not have severe phantom pain, and much less clear in the group that does. If the invasion caused the pain, it is turning up in the wrong people.

The Third Answer: It Might Not Be the Brain

Both accounts so far are arguments about cortex. There is a third position that says the whole debate is happening at the wrong end of the nervous system.

When a nerve is cut it does not go quiet. The severed ends attempt to regrow and can form a tangled knot called a neuroma, and damaged nerve tissue can start generating signals with no stimulus at all [34]. Those signals travel up to the spinal cord and onward to the brain, which has no way to know they are spurious. It receives traffic on the hand line. It reports a hand.

On this account the cortex is not malfunctioning. It is doing its job correctly with bad input.

The most direct test came in 2014. Researchers applied the local anesthetic lidocaine to the dorsal root ganglion, the cluster of sensory cell bodies just outside the spinal cord, in 31 amputees. Phantom sensation and phantom pain were abolished or strongly reduced [35]. Silence the periphery and the phantom fades.

That exchange between periphery and cortex runs through the thalamus, the relay station almost every sensory signal passes through on its way to cortex, and the deeper mechanics of how a damaged nerve can fire without being stimulated come down to the ordinary business of how neurons communicate going wrong.

The peripheral paper did not go unchallenged. Within months the same journal published a response arguing that the peripheral origin of phantom limb pain was very far from resolved, pointing at evidence the lidocaine result could not account for [36]. The original authors replied [37]. All three papers ran in Pain in 2014, and reading them in order is the single fastest way to understand how this field actually argues.

Read that exchange and you get an unusually clear view of how a disagreement actually gets conducted. Nobody accused anybody of fraud. Nobody claimed the other result was fabricated. The argument was about what the evidence can and cannot support, which is the only kind of argument worth having.

There was already older evidence pointing the same way. Ketamine, which acts centrally, reduced phantom pain, and the paper reporting it described the condition as a central disorder maintained by peripheral input [38]. That phrasing is probably closest to where the evidence now sits. Not central or peripheral. Both, with the argument being about which one you have to interrupt.

Recent molecular work has gone looking at the cellular processes in the nerve and cord that could sustain the signal for years [39].

None of this makes the cortical work wrong. A signal has to arrive somewhere to be felt, and where it arrives is cortex. What the peripheral evidence does is move the question. Instead of asking what the brain did wrong, it asks what the brain was told.

Here is where the pre-amputation pain finding from those 1980s Danish studies comes back. If pain before surgery predicts phantom pain after it, something is being carried across the event. Katz and Melzack called these pain memories in 1990 and argued the nervous system holds a trace of the pain state that outlasts the tissue [40]. If that is right, then blocking pain around the time of amputation should reduce phantom pain later. A 1988 study of preoperative epidural blockade reported exactly that [41], and a later trial of optimised perioperative analgesia reported reductions in phantom pain intensity, prevalence and frequency [42]. The idea that a nervous system can hold a trace of a state long after the tissue is gone should sound familiar to anyone who has read about how trauma rewires memory.

The Limb That Shrinks

One more phenomenon, because it constrains all three theories and almost nobody writes about it properly.

Phantom limbs sometimes shorten. Over months or years the phantom hand can seem to retreat toward the stump, until it feels attached directly to the shoulder with no arm in between. The phantom is still there. The distance is gone. This is called telescoping.

Ask yourself what would have to be true for that to happen. The brain is not losing the limb. It is losing the space the limb occupied. Presence and position turn out to be separable, which is not something you would guess from the inside of your own body.

It sounds like a curiosity. It is actually a constraint, because whatever explanation you favour has to allow the felt geometry of the body to change while the felt presence stays intact.

Telescoping has been badly under-studied, and the newest work on it is very recent. A 2026 study examined telescoping in people with limb loss and its relationship to anxiety, depression and pain measures [43], with a companion qualitative study on how people actually describe the experience of living with it [44]. Neither appears anywhere on the first page of search results for this topic, which tells you something about how slowly the readable web updates.

What the Treatments Actually Show

This section needs a warning at the top. Nothing here is advice, and none of it is a recommendation about anybody's own body or treatment. It is a summary of what trials have reported, and the honest summary is messier than most pages admit.

The row that deserves the most attention is the 2024 one.

An extended reality treatment, in which patients use a system that detects residual muscle signals and renders a moving virtual limb, was tested in a multicentre double-blind randomised controlled trial across nine outpatient clinics in seven countries [45]. Phantom pain fell by 64.5 percent in one arm and 68.2 percent in the other. Thirty-seven participants in the first group and nineteen in the second reached a clinically meaningful reduction.

Both groups got substantially better. That is the finding.

A result like that is genuinely hard to interpret, and it is worth being honest about why rather than picking whichever reading suits. It might mean the active treatment works and the comparator, which was itself an active phantom-imagery task rather than a sugar pill, also works. It might mean much of the benefit comes from attention, structure, contact with clinicians and the expectation of improvement. It might mean both. What it does not do is show that the specific proposed mechanism is doing the work.

There is a habit worth resisting here, and you will see it on almost every page that covers this. When a trial reports improvement in both arms, the improvement gets attributed to the treatment and the comparator quietly disappears from the summary. That is how a mixed result becomes a success story.

Methodological critiques of mirror therapy trials have been around for a while, pointing at small samples, weak blinding, and the sheer difficulty of constructing a credible sham for a treatment where the patient can see what is happening [46]. Pooled analyses of movement representation strategies report benefit while flagging the quality of the underlying evidence [47]. The augmented reality and phantom motor execution work reported striking results in chronic cases that had resisted everything else, in a single group design without a control arm [48]. Virtual reality displaying the missing leg was tested in an explicitly small clinical trial [49]. Mirror therapy has also been trialled before amputation to see whether it prevents the problem rather than treating it [50].

None of that means mirror therapy does not work. It means the confidence with which it is described on most health pages is not matched by the strength of the trials behind it.

Drug evidence is similarly mixed. A crossover trial of gabapentin reported benefit [51]. A randomised controlled trial of amitriptyline did not [52]. Negative trials matter as much as positive ones and get cited a great deal less.

The most interesting recent movement is surgical and technological rather than psychological. Targeted muscle reinnervation, where cut nerves are redirected into nearby muscle so they have somewhere to go instead of forming a painful neuroma, was tested in a randomised controlled trial [53] and has also been done preemptively at the time of amputation [54]. Prosthetic limbs that send sensory signals back into the nervous system improved walking and reduced phantom pain in leg amputees [55]. Closed-loop spinal cord stimulation restored a sense of the foot and reduced phantom pain [56]. A 2026 review in Lancet Neurology surveys where hand prostheses with somatosensory feedback have got to [57].

Notice what those approaches have in common. They do not try to erase the representation of the missing limb. They try to give it something to receive. That is the preserved-representation view turned into engineering, whether or not anyone designed it that way.

Given a field this unsettled, it is unsurprising that a formal consensus exercise was needed just to establish what clinicians should agree on [58], and that current multidisciplinary reviews are careful about how much they claim [59].

Phantoms That Are Not Limbs

Limbs get the attention because they are dramatic and because the cortical map is so well characterised there. The phenomenon is not restricted to them.

Phantom sensations have been documented after mastectomy, with a 1989 study reporting incidence and clinical characteristics and finding, once again, a relationship to pain experienced before the surgery [60]. Persistent pain referred to an extracted tooth has its own clinical literature under the name atypical odontalgia [61].

The pattern generalises. Remove a mapped part of the body and the map does not simply close over the gap.

That should reframe how you read the whole topic. Phantom limbs are not a weird exception that happens to arms and legs. They are the most visible instance of something the nervous system does whenever a represented part stops reporting in.

This is where the phantom limb story connects to something broader about how the brain handles missing input. A deprived visual area does not sit idle either, and what happens in the visual cortex when its input is lost has been studied along similar lines. The comparison is worth making carefully rather than casually, because the two cases differ in ways that matter, but the underlying question is the same: what does a piece of specialised brain do when the thing it specialises in stops arriving?

One clarification worth making, since it comes up constantly. None of this involves growing new tissue. Whatever is happening after amputation is existing circuitry changing how it behaves, which is a different process from the birth of new neurons. Readers conflate the two routinely and the distinction changes what kinds of change are plausible and how fast they can happen.

Where This Leaves Us

Thirteen years after the 2013 result, the argument has not been settled. It has done something better and much rarer.

In 2020, Tamar Makin and Herta Flor published a review together [62]. The principal proponent of preserved representation and the author of the 0.93 correlation, in one paper, working through what brain reorganisation after amputation actually means for phantom limb pain and where the two positions can and cannot be reconciled.

That is not how disputes usually end in public. It is worth noticing, because the version of science most people encounter is a sequence of confident announcements, and this is what the actual process looks like.

Makin followed with a 2021 piece arguing the field needed to think outside the mirror box, questioning whether treatments built on the reorganization story were aiming at the right target [63]. A 2022 review tried to make sense of phantom limb pain across all the competing accounts [64]. A 2024 paper worked on connecting the basic science of cortical reorganization after limb loss to what actually helps patients recover, which is the gap that matters most [65].

So what can you take away with any confidence?

Phantom limbs are common, normal, and neurological. They are not imagined and they are not a psychiatric symptom. The idea that phantom pain is somehow in the mind is both wrong and harmful, and it persisted for a long time because nobody could point to a mechanism.

Something in the brain continues to represent the missing limb, sometimes for decades. Both sides of the argument now agree on that much. Kikkert's finger maps and Makin's preserved hand area are not disputed observations. What they mean is what is disputed.

Reorganization is real but smaller and more complicated than the textbook version, and whether it causes pain is genuinely open. Maladaptive plasticity is contested, not refuted. There are published methodological criticisms of the preserved-representation analyses too, and anyone telling you one camp has simply won is selling a cleaner story than exists.

The periphery matters. A cut nerve that fires on its own can drive the whole experience, and silencing it can quiet the phantom, which no purely cortical account predicts.

No treatment reliably works for everyone, and the evidence base is thinner than the confidence of most summaries suggests.

And underneath all of it is the finding in the title. The brain keeps the map. Not as a memory, not as a ghost, but as a working representation that continues to run whether or not the limb is there to answer it. Living with amputation is difficult in ways that go well beyond the neurological question, involving body image, adjustment and the ordinary weight of a permanent change [66] [67], and prevalence work continues to refine who is most affected [68].

The strangest part of the whole story is not that the map survives. It is that we assumed it would not.

The Argument in Order

A last observation about that list. The gap between 1995 and 2013 is eighteen years. That is how long a persuasive correlation held a field, and there is nothing shameful in it. The result was real, the reasoning was sound, and the tools to check it properly did not exist yet. What matters is what happened next, which is that people checked, published the awkward answer, and then wrote it up together.

The mechanism of how sensory representations form and persist without ongoing input runs through everything above. If that question interests you, the way the brain briefly holds an impression after the input stops, explored in the quarter second you miss, is the same question on a very much shorter timescale.