The oldest interface for memory

The method of loci — placing things you want to remember at points along a route you already know — is more than two thousand years old, and it is still what competitive memorisers use. Position turns out to be a far better handle than sequence.

It is also the reason a canvas beats a list: you remember where you put something long after you have forgotten what you called it.

The technique’s founding story turns on the same gap between where and what. Cicero tells it in De Oratore: the poet Simonides of Ceos was dining at the house of a nobleman, Scopas, in Thessaly, when he was called outside, and while he was gone the roof of the hall fell in. The dead could not be recognised. Simonides could still name every one of them, because he remembered where each had been sitting.

Places, images and order

Whether or not it happened, Cicero’s lesson is that order is what makes memory clear. Someone who wants to train it should choose places, form images of the things to be remembered, and store the images in the places. The order of the places then keeps the order of the things, and places and images work like a wax tablet and the letters written on it. Loci is simply Latin for places.

The earliest surviving account of the method is older still. The anonymous Rhetorica ad Herennium, from the 80s BC, builds an artificial memory from ‘backgrounds’ and ‘images’. The backgrounds should be conspicuous and kept in a fixed order, so you can start at any one and recite forwards or backwards; the images should be striking — unusually beautiful or ugly, or comic — because ordinary things slip away while the remarkable stay. Stripped of its Latin, the method comes down to three moves.

• Choose a route you know well — your flat, the walk to work — and fix distinct places along it: the front door, the coat hook, the kitchen table.
• Turn each item into a vivid image and leave it at one place. The odder or more active the image, the better it holds.
• To recall, walk the route in your head and look at what you left at each stop. The route supplies the order; the images supply the content.

Strategy, not special brains

Competitive memorisers can look like a different species. Eleanor Maguire and colleagues studied people known for memory feats in competitions such as the World Memory Championships, using neuropsychological tests and brain imaging (Maguire et al., 2003). Their performance was not explained by exceptional intelligence or by structural differences in the brain. What set them apart was a spatial learning strategy, and while they memorised they engaged regions, such as the hippocampus, that are important for memory and for spatial memory in particular.

If the difference is strategy, it should be learnable. Martin Dresler and colleagues compared 23 of the world’s leading memory athletes with matched controls, then trained newcomers in the method of loci: forty half-hour sessions over six weeks (Dresler et al., 2017). The trained group more than doubled the number of words they could recall from a list of 72, and four months later, without further practice, they still recalled substantially more than before. The changes in their brain networks also resembled the pattern that distinguished the athletes from their controls. The advantage lives in the technique, not in rare hardware.

Why position works

The brain, for its part, takes location seriously. In 1971 John O’Keefe found cells in the hippocampus of rats that fired when the animal was in a particular place, and with Lynn Nadel he later argued that the hippocampus holds a cognitive map of the environment (O’Keefe & Nadel, 1978). In 2005 May-Britt and Edvard Moser described grid cells, which supply something like a coordinate system, and the three shared the 2014 Nobel Prize in Physiology or Medicine. There is evidence that similar cells exist in humans.

It would be overreach to say this explains the method of loci, or that arranging notes on a screen recruits place cells. The safer claim is narrower: people carry a capable, constantly exercised memory for places, and a technique that borrows it borrows something robust.

There are also plainer reasons. A known route already exists, so new items hang on something stable instead of scaffolding and content being built at once. Each place is a retrieval cue: standing, in your mind, at the coat hook prompts the question of what you left there, which is far easier to answer than ‘what was the fourth item?’

And places have neighbours. Items at adjacent stops are seen together, and walking on from one stop leads you to the next. On a memorised route, though, which item sits beside which is an accident of the list — a limit worth keeping in mind.

What it does not do

The method is a technique for recalling arbitrary items in order — the points of a speech, a sequence of cards, a list of words. It does not by itself produce understanding. A student revising for a history exam can place the causes of a war along the route to school and recite them without a slip, and still be unable to say how one led to another. The route holds the items; it says nothing about how they relate.

It also takes work. The Rhetorica ad Herennium is blunt that the theory is almost worthless without sustained effort, and the newcomers in the 2017 study trained for six weeks. The images have to be invented, and the routes chosen and learned, before any of it pays off.

So the lesson for everyday work is not ‘build a memory palace’. Few people need to recite forty items in order. What transfers is the principle underneath: give material a stable place, and let the place carry some of the remembering.

Stable places for digital work

A paper desk gives things places without being asked: the pile on the left is the unpaid invoices, the folder by the lamp the draft that needs a reply. Many digital tools quietly undo that. A list sorted by date modified reorders itself whenever you touch something, and search ranks by a relevance you cannot see. Neither gives an item a place it keeps, so the only handle left is its name.

There is evidence that spatial arrangement helps with documents too. In a 1998 Microsoft Research study, George Robertson and colleagues let people place web pages anywhere on a sloping virtual surface they called the Data Mountain. The layout drew on spatial memory, and it showed reliable advantages over the browser’s ordinary favourites list for managing pages people wanted to keep (Robertson et al., 1998).

Consider someone planning a kitchen renovation. Over a month they collect quotes, tile photos, a PDF of building regulations and a note about moving a water pipe. In a folder these become a column of filenames; on a canvas they can have regions — costs along the top, materials on the left, permits in one corner, the plumbing note beside the floor plan it affects. A week later, ‘that thing about the pipe’ is found by going to the floor plan, not by guessing what the file was called.

A few habits strengthen the effect. Regions work best with landmarks — a heading, an image, a distinctive block — visible from a distance, much as the Rhetorica ad Herennium asked for conspicuous backgrounds. Related things belong near each other, and here a canvas improves on the mental route: neighbours are chosen, not accidental, and moving a block is a small decision about what it is, a form of thinking by arranging.

A PlansFlow project applies the same idea: an infinite canvas where notes, PDFs, images, links and web search answers sit as blocks, and stay where you put them until you move them. Connections drawn between blocks can play the part of the route — the order in which you would walk someone, or yourself, back through the material.

Revisiting works better by travelling through the space than by searching it. An engineer back from two weeks away can pan from the open question to the test results to the half-written recommendation and pick up the state of the work; search hits, shown one at a time and out of context, do not give that back. The same logic argues against anything that tidies or re-sorts your material for you. It trades your spatial memory for someone else’s order.

In practice

1. 01

   Give each piece of material a fixed home, and move it only when its meaning changes, never just to tidy up.

2. 02

   Divide a project into a few regions, each with a visible landmark such as a heading or an image, so you find the area before you look for the item.

3. 03

   Put things that inform each other side by side; if two items must be read together, their positions should say so.

4. 04

   Return to a project by moving through it region by region before you reach for search.

5. 05

   Turn off automatic sorting where you can, and be wary of views that reorder items by date or relevance each time you open them.