You Don't Remember — You Rebuild It

— In 2003 they took 10 competitors from the world memory championships and 10 ordinary people of the same age, put them in a scanner and went looking for what makes those brains different. They found nothing. Not one significant difference in grey matter volume — not at a strict threshold corrected for multiple comparisons, and not at a deliberately loose one with no correction at all. They weakened the criterion on purpose, to give the differences a chance to show up. There were no differences.

— They measured that too. Verbal intelligence — 111 for the champions against 112 for the control group. Matrix reasoning — 12.9 against 12.4. So the controls are a point higher on one measure and half a point lower on the other, which in plain language means "the same." Not one of them is a genius. That's the most inconvenient thing about this work: if they'd found geniuses, everything would be explained and we could all go home.

— They found something, just not where they were looking. What differed wasn't the hardware, it was which areas switched on at the moment of memorising. In the champions you saw the medial parietal cortex, the retrosplenial cortex and the right posterior hippocampus light up — areas that are otherwise busy with spatial orientation. They weren't memorising a list. They were walking through it. 9 out of 10 used the method of loci, and in the control group not a single person used any technique. Not one.

— Then take S.F. An ordinary student, digit span of 7, like everyone. They sat him down to train, and over a bit more than two years, more than 230 hours of laboratory practice, he got to 79 digits. He was a long-distance runner and he heard running times in the digits: this one is nearly a mile record, that one is a hopeless time.

— And now the part that ruins the whole picture. They gave him a string of letters instead of digits — and he recalled 7. Same as day one of training. And it stayed at 7 the entire time he trained.

— The opposite — that is the result, just not the one people expect. He didn't expand his memory, he built one very good translator for one type of material. For letters there was no translator, and the memory was instantly ordinary again. Which means it definitely isn't about capacity. Capacity doesn't vanish when you change alphabet.

— In 2017 they took 51 volunteers with no training at all and split them into 3 groups. One group spent 40 days, half an hour a day, laying words out along imaginary routes; the second spent 40 days grinding short-term memory exercises; the third did nothing. The first group went from 26 words out of 72 to 62. The second gained 11, the third 7. And 4 months later, without a single review, almost all of the first group's gain was still there.

— And rightly so, because I now owe you three cracks in it before you find them yourself. First: in that very scanner study the champions did beat the controls on one measure — digit span, 16.8 against 12, and that difference is significant. And on delayed recall of a story as well.

— From that data you can't see whether the advantage is innate or already a trace of training. Digit span grows beautifully with chunking — with exactly the thing they spent years learning. Which is cause and which is effect, the study doesn't answer. I have to say that out loud, because keeping quiet would be convenient.

— The brain can in fact change with practice. The same researcher had earlier found an enlarged posterior hippocampus in London taxi drivers. In mnemonists she didn't find one — and she herself explains it by saying the taxi driver holds a whole city in his head, while the mnemonist runs the same few routes over and over. So "no differences" doesn't mean "practice changes nothing," it means "this particular practice leaves no anatomical trace." That's a big difference, and it usually gets swallowed.

— Practice explains a lot, but not everything, and that's been counted. A meta-analysis over 88 studies: deliberate practice explains 26% of the variance in performance in games, 21% in music, 18% in sport, 4% in education and less than 1% in professions. It's the largest known factor — and it isn't the only one. There's been a 20-year fight about exactly what sits in the remaining percentages.

— There is. The argument isn't whether it exists, it's where the ceiling is. On the innate share of working memory capacity I won't even give you a number — twin-study estimates spread widely, there's no single figure, and any specific percentage here would be invention dressed as statistics. But notice what neither side argues about: that 26 words become 62 in 40 days. On that we have enough data.

— One person crumples things up and stuffs them in — half of it fits. Another rolls them and lays them out in sections — it all fits, and he still knows where the socks are. The suitcases were measured in a scanner, they're identical down to the centimetre. The difference is that one of them was taught to pack and you weren't. By the way, the man who won the US memory championship in 2006 had come to that championship a year earlier as a journalist, to write a piece about prodigies. He found no prodigies, he found people with a technique, got himself a coach, and a year later memorised a deck of 52 cards in 1 minute 40 seconds.