Remembering Names and Faces

Nobody forgets a face. People say they are "bad with names", and they are right, but the phrase hides what is actually going on. You recognise the person across the room instantly. The failure is narrower and more specific: the name did not attach to the face, because a name is an arbitrary sound and there was nothing for it to attach to.

That is a fixable problem, and it is fixed the same way every other mnemonic problem is fixed: you give the arbitrary thing a picture, and you join that picture to something that will be present when you need it. Here, the something is the face. This article is the method, step by step, with the two mistakes that account for most failures and a plan for practising without having to attend a party.

Why names are hard, specifically

Psychologists have a tidy demonstration of this. Show people a photograph of a stranger and tell them the person is a baker, and they remember it. Show a different group the same photograph and tell them the person's name is Baker, and they forget it. Same word, same face. The difference is that "a baker" comes with pictures (flour, an oven, an apron) and connections to things you know, while "Baker" is a label with no content, connected to nothing.

The lesson is not that names are inherently harder than jobs. It is that names arrive without hooks and jobs arrive with them. The method below is simply a way of manufacturing the hooks that a job gets for free.

There is a second, more embarrassing reason names fail: you did not hear it. In the two seconds after an introduction you are managing eye contact, a handshake, what you will say next, and whether your drink is about to spill. The name goes past. You cannot remember what you did not encode, and no technique helps with that. So the method starts before the mnemonic.

The method

Step one: get the name. Actually hear it. If you missed it, ask again immediately ("sorry, was that Marion or Marian?"), which is always fine in the first ten seconds and awkward after that. Say it back once: "Marion, good to meet you." If it is unusual, ask how it is spelled; people with unusual names are used to the question and usually pleased to be asked. This step alone, with no imagery, fixes a surprising share of the problem.

Step two: turn the name into a picture. Names are sounds, so use sound. Some names are already words: Rose, Bill, Hunter, Summer. Others contain words: Marion has a marionette in it; Kowalski has a cow on skis; Patel can be a pat of butter on a telephone; Nguyen sounds, to an English ear, like "when" and can be a clock with a question mark on the face. You are not looking for a clever or respectful picture; you are looking for the first concrete thing the sound suggests, and you are keeping it to yourself.

For common names it is worth having a fixed picture you always use. Every John is a toilet, every Mike is a microphone, every Anne is an ant, every Tom is a tom-tom drum. Build these once and the step becomes instant for most people you meet.

Step three: pick a feature. Look at the face and choose one thing that is distinctive: a high forehead, thick eyebrows, a gap in the teeth, a mole, a sharp jaw, a particular hairline. It should be something you will see again, so not glasses or a hat if you can avoid it, and not clothing. If nothing stands out, pick the most prominent ordinary feature, usually the nose or the eyebrows. It only needs to be what your eye lands on.

Step four: join them. Put the name-picture onto the feature and make it interact. Marion's marionette dangles from her high forehead on strings. Mr Kowalski's cow is skiing down the slope of his long nose. Mike's microphone is wedged in the gap between his front teeth. Make it move, make it a little absurd, and hold it for two or three seconds while you look at the person. This is the same join that powers the link method; the face is just a peg that walks around.

Step five: use it, soon. Say the name once more in the conversation and once when you part. Then, in the first quiet moment afterward (walking to the next room, in the taxi), run through the faces you met and pull the names back off them. This first rehearsal within an hour is worth more than any amount of effort at the moment of meeting.

A worked example

You arrive at a dinner and meet five people in ninety seconds. This is the situation the method was built for, so here it is at speed.

Priya, who has striking dark eyebrows. "Priya" sounds like "prayer": you picture her eyebrows as two hands pressed together in prayer, which then part and rejoin.

Doug, with a broad, round face. Doug is "dug": a shovel is digging a hole in the middle of that round face, turning up earth.

Beatriz, whose hair is pulled tightly back. "Bea-triz": a bee lands on a trellis (a garden lattice) that is growing out of her scraped-back hairline.

Sam, with a neat pointed beard. Sam is always Uncle Sam for you, so a tiny Uncle Sam in the star-spangled hat is swinging from the point of the beard.

Chen, who has a small scar through one eyebrow. "Chen" is a chain: a bicycle chain is threaded through the scar and out the other side, and it rattles.

Ninety seconds, five images, and each one took about as long as the handshake. In the taxi home you go round the table: eyebrows in prayer (Priya), shovel in the face (Doug), bee on the trellis (Beatriz), Uncle Sam on the beard (Sam), chain through the scar (Chen). Do it once more the next morning and most of these will be there in a month.

The two mistakes that account for most failures

You skip the feature. The name-picture without a feature is floating. You remember that there was a marionette somewhere at the party but not on whom. The feature is what ties the image to the person, and it has to be a physical feature you will see again, not a position in the room or what they were drinking.

You do not rehearse within the hour. The image is fragile at first. Names practised once in the first hour and once the next day survive; names encoded perfectly and then never revisited are gone in a week. If you take one thing from this article, take the taxi rehearsal.

Two smaller ones. Pictures that are too polite ("I imagined him near a rose") do not hold; give the rose thorns and stick it somewhere. And do not try to do this for everyone at a conference of two hundred. Choose the ten you will need and do those properly.

The evidence

This is one of the few mnemonic methods that has been tested more or less as it is taught. In a well-known study from the late 1970s, one group of participants was trained in exactly this sequence (find a feature, convert the name to an image, link the two) and another was given no strategy. On a test of matching names to photographs the trained group's recall was around eighty percent higher. Later work found the same pattern with older adults, who tend to report the most frustration with names, and found that the gains held up on delayed tests as long as people kept using the method.

Two honest caveats. The improvement is large but it is not total: trained participants still forget some names, especially with long lists and no rehearsal. And the studies mostly used photographs, where you can stare; real introductions give you two seconds and a handshake, which is why step one (actually hearing the name) matters so much outside the laboratory.

Names that resist

Names you cannot make a picture from. Break them into syllables and picture the syllables: "Oyelaran" becomes oil, a lair, and a run. Or use the person's face to remind you of someone famous with the same name. If neither works, spell it: the letters themselves can be pictured (a capital O as a ring, and so on). This is slow, and for a name you will use often it is worth doing once, properly.

Names in a language you do not speak. Ask for the pronunciation, then treat the sounds as if they were English words. This is exactly what the keyword method does for foreign vocabulary, and it works better than it has any right to. Be aware that your private image is for retrieval only; the respectful thing is to say the name correctly, and the picture helps with that too, because it fixes the sounds.

Two people with the same name. Use the same name-picture and different features. Two Toms are two drums, one balanced on a bald head and one on a large chin. Recall the feature first, the drum second, and the name comes.

Names you learn from a list, not a face. For a class register or a seating plan, where there is no face yet, peg the names to the positions (the peg system is built for this) or chain them with the link method, and attach the faces when you meet them.

Practising without a party

The method has an awkward property: you cannot practise it alone, or so it seems. In fact you can, in three ways.

Photographs. Any set of faces with names attached will do: a company's team page, a cast list, a sports squad. Give yourself thirty seconds per face, build the image, then test yourself an hour later and the next day. Twenty faces a day for a week is more introductions than most people get in a year.

Television. Every film and every news broadcast introduces strangers by name. Do the method on them in real time, at the speed of a real introduction. This is the closest thing to live practice and it costs nothing.

A personal deck. Keep the names you actually need (colleagues, clients, your children's teachers, the neighbours) in a deck and review it. Each card is a name on one side and the feature plus the image on the other. This site lets you build a personal deck and drill it, and the review schedule takes care of spacing the practice so you are not re-learning the same twenty people every week. Faces are not stored; the image you built is a private note to yourself and works as the cue.

Names on screens

More introductions now happen on video calls than at parties, and the method needs two adjustments there.

The name is written on the screen, which solves step one entirely and creates a new failure: because the label is always visible, you never encode the name at all, and the day you meet the person in a corridor you have nothing. Treat the on-screen name as the cue to do steps two to four deliberately, in the first minute of the call, while the label is still there to check against.

Features are harder on a small tile in poor light, so use what a screen gives you instead: the voice (a low voice, a particular accent), the background (the bookshelf, the bicycle on the wall), or a gesture the person makes when they talk. Backgrounds and voices are less permanent than a jawline, but for colleagues you will mostly see on screen they are what you will actually be shown again.

Email and chat names are the easiest case and the most neglected. You have the name, the spelling and unlimited time. Build the image once, the first time you reply to someone new, and the name will be there when you finally meet.

When a name has already gone

The method prevents most losses, but everyone still meets someone whose name has vanished. Three recoveries, in order of dignity.

Run the alphabet. Say the letters silently and pause on each; the first sound of the name often surfaces when you reach its letter. This works far more often than it should, because the initial sound is usually the part that was encoded.

Recover the context. Where did you meet, who introduced you, what were you talking about? Names are stored with their episode, and finding the episode frequently returns the name.

Ask, properly. "I'm so sorry, I've lost your name" said plainly in the first minute of the second meeting costs nothing and is far better than an hour of avoidance. Then do the method on the spot, so that it does not happen a third time.

Keeping names for the long term

For people you will see occasionally over years (a client, a cousin's partner, the parents at a school gate), one good encoding and one rehearsal is not enough, and it does not need to be much more. A review the day after, a week after, and a month after will hold most names for a year. The personal deck mentioned above does this scheduling for you; if you prefer paper, a page at the back of a notebook with the name, the feature and the image, glanced at before you go anywhere you might meet the person, does the same job.

A plan for the first two weeks

Days one to three. Build fixed pictures for the twenty most common first names where you live. Practise on a team page of twenty faces each day; test yourself the following morning.

Days four to seven. Do the method live on every name you meet or hear, including on television. Keep it to steps one to four; do not worry about the rehearsal yet. The goal is to get the image built at conversational speed.

Week two. Add the rehearsal within the hour, every time. Start the personal deck with the names you actually care about. At the end of the week, go somewhere you will meet people and count how many names you have the next morning. Most people who do the fortnight properly find that "bad with names" has quietly stopped being true.

What this is not

It is not a trick for looking clever. If you use someone's name three times in a sentence they will notice and it will be worse than forgetting it. It is not a substitute for interest; the method works best when you are genuinely curious about the person, because curiosity does step one for free. And it does not make names permanent. A name you never use will fade even with a good image, which is fine; the point was never to remember everyone forever, only to remember the people who matter for as long as they matter, and to stop losing names in the first ten seconds.