Person, Action, Object
Give every number from 00 to 99 a person, an action and an object, and six digits fit in one picture. The same trick holds a deck of cards in eighteen stops.
photo (adapted): Orren Jack Turner, Public domain
PAO (Person, Action, Object) is a memory system that gives every two-digit number from 00 to 99, or every playing card, a person doing an action with an object, then builds one picture from three of them: the first one's Person, the second one's Action, the third one's Object. On this site's PAO list, 12 is Tina Turner stomping with a hand microphone, 34 is Marie Curie stirring a vial of radium and 56 is Elijah Wood trudging in a grey elven cloak, so the six digits 12 34 56 become one scene: Tina Turner stirring a grey elven cloak.
The Method of Loci gives you a place to store data, and the Major System gives you a way to encode digits into things that fit on shelves. PAO is the trick that lets you fit three things on one shelf instead of one, and it is the technique behind most modern deck records, including the 12.74-second speed-cards record. Alex Mullen, who switched to a two-card system of 1,352 images in mid-2014, used a one-card PAO system in his first year of training.
PAO is a multiplier: it is what happens when you decide that one locus, one image, one peg word is no longer enough, that the substrate of human memory is generous, and you should pack more in. It is also the technique most likely to make a person doing it look, to a bystander, like they are mildly out of their mind. There is no discreet way to memorise a deck of cards by imagining Albert Einstein scribbling on a chalkboard inside your kitchen sink, so you just have to commit.
The cards-as-a-deck problem
A standard deck has 52 cards. If you wanted to memorise the order of a shuffled deck using only the Method of Loci, you would need 52 loci (one per card) and 52 distinct images, one per card. By the time you'd built that, you'd have a working palace; by the time you'd memorised the cards, you'd have spent forty minutes.
The world records for memorising a shuffled deck are held by competitors who routinely encode a full 52-card deck in under twenty seconds (at the top of the field, comfortably under fifteen). At that pace, the encoder is spending less than a third of a second per card: eyes on the card just long enough to register rank and suit, hands flicking through.
This is not possible with single-image encoding: there aren't enough quarter-seconds in fifteen seconds for fifty-two distinct mental acts. The trick has to be that each "image" carries more than one card, and PAO carries three.
The competitive history
Person-Action-Object encoding for playing cards became the standard technique at the World Memory Championships through the late 1990s and early 2000s. Andi Bell of the United Kingdom, who won the World Memory Championship in 1998, 2002 and 2003, was an early high-profile proponent of the technique; his speed on the cards discipline helped establish PAO as the dominant method for that event. By the mid-2000s, the technique had spread through the memory-sport circuit and was in near-universal use among top competitors.
The origin of the method is older than the competitive circuit, but its systematic form (one canonical Person, one Action, one Object per card, rigidly composed in triples) was formalised and disseminated primarily through that community. There is no single inventor; the technique emerged from an iterative refinement process across multiple competitors.
Today the top competitors have moved to either PAO or a two-card system, which turns each pair of cards into one image. Plain PAO is the smaller build: 52 Person, Action, Object triples, 156 images in all, rather than more than a thousand.
How it works
Each card in the deck (there are 52) gets assigned a stable Person, Action, Object triple. For example:
The Ace of Spades, in the canonical Major Dictionary PAO list, is Albert Einstein (Person) scribbling (Action) on a chalkboard (Object). Each PAO is unique; no other card has Einstein, no other card has scribbling, no other card has the chalkboard. All 52 are listed, suit by suit, in the full 52-card PAO list.
Now, when you draw a card from a shuffled deck, you think Einstein, not "ace of spades", and after a few hundred reps the mapping is automatic.
But the trick is that you don't use the full PAO when only one card is drawn. You compose three cards into a single image, like this:
- Card 1: Take the Person.
- Card 2: Take the Action.
- Card 3: Take the Object.
Three cards become one image: Person from card 1 doing Action from card 2 with the Object from card 3. That image goes on a single locus.
As a worked example, suppose the first three cards are:
- A♠ (Einstein, scribbling, chalkboard)
- 2♥ (Beyoncé, twirling, fan)
- 3♦ (Chaplin, tipping, bowler hat)
The composed image is Einstein twirling a bowler hat: the Person from the first card, the Action from the second, the Object from the third, and nothing from anywhere else. That single image goes on locus 1 of your palace.
The next three cards become locus 2, and three more become locus 3. A deck of 52 cards becomes 17 composed images plus a 52nd-card remainder (which is just the Person of the last card, alone), so you get 52 cards in 18 loci: 17 composed images plus the leftover card's Person.
If you have an 18-locus palace memorised cold, and a 52-card PAO list memorised cold, you can encode a deck in eighteen quick walks-through-rooms, which with practice takes about a minute. The champions get it under fifteen seconds because their PAOs are so overlearned that the Person → Action → Object recovery is sub-conscious; they're seeing Einstein and the chalkboard in the same moment they see the rank-suit, with no intermediate deliberation.
PAO for numbers
The same rule works on digits. Give every two-digit number from 00 to 99 its own Person, Action and Object, and six digits become one picture: the first pair gives the Person, the second pair the Action, the third pair the Object. The first six digits of pi, 31 41 59, become Mother Teresa directing, with a long wand for a baton. A hundred rows is a bigger build than fifty-two cards; the 00-99 PAO list is a ready-made set built on Major System sounds, so every person's name spells its own number. Adopt it whole and swap what does not land.
Choosing your Persons
This is the hard part of PAO: you will be living with these 52 people for years, so pick them with care.
The principle is to make them maximally distinct from each other. If you have one Bruce Lee, don't add a second martial artist, and if you have one Chaplin, don't add a second silent-film clown. The encoding works by recognition; the more your 52 People feel like 52 different silhouettes, the faster the recovery.
Some heuristics:
- Mix sources. Half historical figures, half pop culture. A PAO with only US presidents is hard to disambiguate; one with Einstein, Beyoncé, Sherlock Holmes, Florence Nightingale, and Yoda is easy.
- Pick people whose silhouette is iconic. Hermione Granger has a wand and a bushy hairstyle, and Ron Swanson a moustache and a steak, which makes them easier than two photogenic-but-similar actresses.
- Stay away from family and close friends. You will be imagining them doing strange things 17 times per shuffled deck, which is faintly disturbing in a way that strangers and historical figures are not. (This is genuine advice.)
- Stay away from people you'll forget. That obscure jazz musician you love might not work as a PAO if you can't picture him reliably, so pick the ones whose face you can summon without effort.
The Major Dictionary canonical PAO list is opinionated. We pick people that are visually distinct, culturally durable, and have at least one defining action you can pin them to: Einstein scribbles, Beyoncé twirls, Chaplin tips his hat. The list ships under the 52 cards system, and you can adopt it as a starter set and edit people you'd rather replace.
The canonical PAO 52 cards ships with Major Dictionary as a starter system you can edit, override, and drill against. Adopt it in one click and it joins your dictionary, lighting up the practice surfaces.
Choosing your Actions
Each Person needs a stable Action: what they're always doing in their PAO (the thing you'd describe them with in one verb if you only had one verb).
The principle is that the Action must be visualisable and energetic. Static actions ("standing," "thinking") fail, while dynamic ones ("scribbling," "kissing," "throwing," "splitting") succeed.
Heuristics:
- One action per Person, never overlapping. If Einstein scribbles, no one else scribbles, and if Cleopatra drapes, no one else drapes. The Action has to be unique to the Person, which is what lets a deck of 52 give you 52 distinguishable verbs.
- The Action should fit the Person. Einstein scribbling is easy, Einstein moonwalking harder: the closer the Action is to the Person's archetype, the less effort the recall takes.
- Compose Person + Action without an object first. If Einstein scribbling, Beyoncé twirling and Chaplin tipping feel solid in isolation (if you can hold each one for two seconds with closed eyes), your Person-Action pairing is sturdy; if they don't, swap the Action.
The 52 Actions on Major Dictionary are tied to each card; you can override them in your own dict, and the system surfaces your overrides at the top of the entry detail page. Editorial overrides are first-class.
Choosing your Objects
The Object is the third leg, and it is the most flexible. Pick something the Person interacts with characteristically, ideally an object that the Action acts on.
Einstein's Object is a chalkboard because that's what he's scribbling on, Beyoncé's is a fan because that's what she's twirling, and Chaplin's is a bowler hat because that's what he's tipping. In each case the Object completes the Person's characteristic scene.
Heuristics:
- Make the Object physical, ideally something a Person can hold or wear. Abstract Objects (the universe, justice, the concept of love) don't work, while hand-sized ones (a bowler hat, a fan, a wand) are the easiest to move between scenes. A few Objects in the starter set are too big to hold, like the grand piano, the freight train and the North Star: they make each card's own scene easy to learn but are harder to hand to another card's Person. If one stalls in practice, swap it for something graspable.
- Make the Object distinct. No two cards should share an Object: if Einstein has a chalkboard, no one else has a chalkboard, even a school-themed Object.
- Sometimes the Object is the Person's iconic prop. This is the easy mode: Chaplin's bowler hat, Cleopatra's asp, Gandhi's spinning wheel. These need almost no memorisation because the Person and the Object are already linked in pop culture.
The King of Hearts in the canonical list is Elvis Presley (Person) hip-swivelling (Action) in a rhinestone jumpsuit (Object). All three legs reinforce each other: the King of rock, his signature move, the costume nobody else on earth wears. From ace to ten the set is arranged so the Person's initial matches the rank, which is why the two of hearts is Beyoncé, and the queens and three of the four kings carry the rank in their name, which is why the king of hearts is the King; you are never remembering a mapping, only a picture. The recall pathway is not "K♥ → Elvis"; it is the visual scene itself, locked in once and replayed each time.
Composing three cards
Once you have your 52 PAOs, the composition rule is rigid:
Card 1 contributes Person. Card 2 contributes Action. Card 3
contributes Object.
Apply it in that order every time, without improvising: the rule is the contract. If the first three cards are A♠, 2♥, 3♦, the image is Einstein (P from A♠) twirling (A from 2♥) a bowler hat (O from 3♦).
If the next three cards are J♠, 5♦, 2♣, the image is Person of J♠ (let's say it's Yoda) Action of 5♦ (let's say "shooting") Object of 2♣ (let's say "a sandwich"), giving Yoda shooting a sandwich for locus 2.
You will, in early practice, sometimes accidentally swap A and O, or P and A, and the recall will surface a wrong card. This goes away with reps. The discipline of the composition order is what keeps the encoding consistent.
When you have leftover cards
A deck of 52 divides into 17 groups of 3 with one card remaining. That remainder card is encoded as just its Person, alone, on the 18th locus, with no verb and no object: Einstein, sitting on the kitchen counter. When you reach the 18th locus on recall, you read off the Person's card and stop.
If you're memorising multiple decks (the championship circuit demands several), you walk into the next palace and start over, grouping deck 2 in threes from its own first card.
Common pitfalls
The technique fails most often in three places.
1. Overlapping People
If two of your 52 People look or sound alike, the recall will silently grab the wrong one. Albert Einstein and Mark Twain are both white-haired men in old photographs; if you have both, their PAOs will smear. Pick distinguishable archetypes.
2. Boring Actions
Static actions ("sitting," "thinking," "looking") don't form images that survive composition. Einstein sitting on a sandwich is forgettable where Einstein scribbling on a sandwich is not, so pick verbs with motion in them.
3. Mismatched Objects
The Object is the slot most likely to feel arbitrary, because unlike Person and Action it's often a free choice. Resist the temptation to default to "thing" Objects. A box is a bad Object. A briefcase Sherlock would carry is a better one.
Variations the championship circuit uses
Double-PAO uses six cards per locus, three for each of two PAO images stacked at the same locus. This halves the locus count, doubles the recall density and makes the images even more bizarre, but six-card composition is hard.
PAO + Object-locus. Some memorists drop the Action and use Person-Object (PO), encoding two cards per locus instead of three. PO is faster to learn and slightly faster to recall, but loses the 3:1 compression PAO provides. It's the right choice if you don't need the density and the wrong one if you do.
Sharing images between cards and numbers. Some competitors translate each card into a two-digit number so that one 00-99 PAO list covers the deck as well. Larger systems share too: 992 of the 1,352 images in Alex Mullen's two-card system overlap with his 000-999 numbers system, so only 360 are card-specific. The 52-card list here is not built that way: it keeps its own people, separate from the 00-99 list.
Drilling discipline
The encoding is only as fast as the slowest leg. If 49 of your 52 PAOs are sub-second but three of them require deliberate recollection, every shuffled deck will hit those three slow cards seven or eight times and bog the recall. Drill the entire 52 to uniformity, not to "good enough on average."
The right practice loop is: pull a single card, produce the Person, Action, and Object out loud (or silently, if you must), in order, in under a second. Do this for all 52, in random order, every day for two weeks. A physical deck is the right tool for that half: the discipline is not creativity, it's reproducibility.
The card drill is for the other half. It deals five cards, thirteen or the full fifty-two from a shuffled deck, revealing them one at a time over thirty seconds (a minute for the full deck), and then asks you to re-place them: tap a slot, tap the card that was there, in the order you saw them. It scores the whole run afterwards, card by card, and the ones you got wrong are recorded against their entries, so the set reports its own weak spots instead of waiting for you to notice them.
By week two, the slowest 5 cards in your set will have surfaced themselves through the drill, and those are the ones to reconsider: maybe the Person isn't iconic enough, the Action overlaps something else, or the Object is the wrong scale. Replace the weak PAOs deliberately, then re-drill. By week three, you'll have 52 sub-second cards and the encoding will feel automatic.
The other discipline is composing on the fly, and no drill will hand it to you. The PAO encoding is only useful in groups of three; drilling cards in isolation gets you fluent at the legs but not at the composition. Once your single-card recall is solid, deal three cards face-up off a real deck and produce the composed Person-Action-Object image, such as Einstein twirling a bowler hat. Then deal the next three. This is the actual deployed motion of the technique; if you can't do it in three seconds with the cards in front of you, you can't do it in fifteen seconds with a stopwatch.
When not to use it
PAO is overkill for short sequences: if you need to memorise a 3-card hand, just use the Method of Loci with each card's Person on its own locus. PAO's compression only pays off when you have enough cards to fill multiple loci.
PAO is also fragile to overlap. If you accidentally pick a Person-Action-Object triple where the same iconic image already appears on another card, you will silently lose data. The canonical 52 shipped here are de-duplicated; if you bring your own, de-duplicate carefully.
Where to start
The shortest path is below. The PAO deck journey walks you through the 52-card list in seven stages, from the four suits to a timed full-deck attempt.
- Adopt the Major Dictionary canonical 52-card PAO. Each card carries a Person, an Action, and an Object, illustrated and pinned by the editor. You can override any card with your own PAO; the system surfaces your version at the top of the entry detail.
The canonical PAO 52 cards ships with Major Dictionary as a starter system you can edit, override, and drill against. Adopt it in one click and it joins your dictionary, lighting up the practice surfaces.
- Drill the PAO list against itself until you can produce all three legs of any card in under a second. Then take the run to the card drill at a small count and a generous timer, and raise the count as the scores hold. Two to three weeks of ten-minute daily sessions gets most people to fluency.
- Build an 18-locus palace. (The six stock palaces here have 12 to 16 loci each, enough for 36 to 48 cards while you practise; for a full deck, create your own from a place you know with 18 stops.)
- Encode a shuffled deck. The first time will take five minutes and feel like flailing, the tenth ninety seconds, and the hundredth twenty.
The technique is not a magic trick but a translation pipeline, and the pipeline is the work: you build it once and then you have it.
See also: The Method of Loci for the spatial substrate; The Major System for digit encoding (which composes well with PAO for mixed digit/card content).