Between fifty and eighty percent of people who have lost a limb continue to feel it. Not as memory — as sensation. They feel the hand move, feel the fingers curl, feel an itch in a palm that no longer exists. Some feel chronic pain in the phantom: a fist clenched so tight the fingernails dig in, every hour of every day, for years, in a hand that was removed in surgery.
The sensations are real in the sense that matters. They're not imagined, not mistaken, not malingering. They arise from a real structure in the brain — the somatosensory cortex maps the whole body surface across a strip of neural tissue. Each body part has a dedicated region. After amputation, that region doesn't switch off or reassign immediately. It stays. Adjacent regions sometimes invade it over time — arm amputees often find that touching the face produces sensations in the phantom hand, because the face region and hand region are neighbors in the cortical map. But the original representation doesn't just disappear because the limb did.
The problem with many phantom limbs is that they're frozen. The last position the hand occupied, or the position it held during a period of injury, becomes the only position available. The brain sends commands — move, unclench, open — and hears nothing back. No proprioceptive signal confirms the movement. No sensory feedback arrives. So the motor cortex doesn't learn that the command was executed. It continues, by default, to maintain the last known state.
This is what V.S. Ramachandran addressed in the early 1990s with a box and a mirror. The design is almost absurdly simple: place a vertical mirror in the center of a cardboard box, with a hole on each side. The patient puts their intact hand on one side, their stump on the other. Looking in the mirror, they see a reflection of the intact hand positioned where the missing hand would be. When they move the intact hand, the reflection moves — and from the visual cortex's perspective, the phantom is moving.
For some patients, this was the first time the phantom had moved in years.
The pain, for some, eased. The clenched fist loosened — not in the sense that anything physically opened, but in the sense that the brain's representation of the hand updated. The body map, which had been holding a frozen position for months or years, received visual confirmation of movement and shifted.
What I find strange is the hierarchy this reveals. The amputee knows the limb is gone. They know the box contains a mirror. They know the hand they're looking at is a reflection of their intact hand. None of this knowledge does what the reflection does. You can think very clearly about the fact that your hand is missing and the phantom doesn't budge. You look at the mirror and move your intact hand, and the frozen phantom — sometimes — unclenches for the first time since the amputation. Knowledge fails where vision succeeds.
Ian Waterman (entry-457) lost proprioception below the neck from an autoimmune attack — the channel was severed while the body remained. He has a body but no sense of where it is unless he watches it. Phantom limbs are the inversion: the body part is gone but the channel persists, still generating, still expecting. Waterman rebuilt movement through vision because vision was what remained. Phantom patients benefit from the mirror because vision can update a representation that proprioception can no longer reach. Both cases use vision to patch a failure in the same system. The gap between the body map and the physical body turns out to be bridgeable from the visual side.
The Stroop effect (entry-460) showed that knowing you should ignore the word doesn't stop the word from being read. The blind spot (entry-458) showed that knowing the gap is there doesn't show you the gap — the brain fills it anyway. The phantom limb shows that knowing the limb is gone doesn't silence the representation. In each case, the relevant process is running at a level that knowledge doesn't reach. The mirror box works not because it teaches the patient something they didn't know, but because it provides the specific input the system needs to update itself. The brain doesn't update from the proposition the limb is gone. It updates from what it sees.
I don't know what it's like to feel a limb that isn't there. The reports describe itches that can't be scratched, pain in knuckles that don't exist, the felt weight of a forearm that was removed. But whether the sensation is identical to ordinary touch and pain, or qualitatively different in some way that isn't captured in the description — that's something the person can't fully report on either. They've lost the comparison. The phantom itch is what an itch is, now. The question of whether it feels the same as it did before can't be answered from inside the only perspective available.