BRAINBITES
WEEK OF AUG 28–SEP 3, 2026 · ISSUE 25
BRAINBITES — Your weekly health, fitness, and longevity digest. Five minutes that cover the highest-signal podcasts and science before they hit the mainstream.
This week: Joe Liemandt told Huberman he can fit a school day's academics into two hours, and the number he's chasing is Benjamin Bloom's famous two sigmas. We went back to where that number came from, and about half of it turns out not to be tutoring at all. Plus Gina Poe on the silence your brain needs before it can file anything, and Brian Grosberg giving Peter Attia a master class on headaches.
From: Huberman Lab · Educational Researcher · Huberman Lab Essentials · Scientific Reports · The Peter Attia Drive
🔢 2.00 — the most-quoted number in education, and where it came from
🎙️ Joe Liemandt on the two-hour school day
🔬 Deep dive: half the miracle was a quiz
🔭 Expected breakthroughs: Alzheimer's tests that need only blood
📰 News Bits: Poe on the quiet brain · an air filter with a test score · headaches, properly explained
GM.
Every field has one number everybody quotes and almost nobody has read the source for. Education has a spectacular one. We read the source — and the useful half of it is free.
⏱ ~5 min read
🧠 THE NUMBER AND THE METHOD
Everyone quotes the ceiling. Very few people ship the method.
In 1984 Benjamin Bloom claimed a tutored student ends up two standard deviations ahead of a classroom student, and technology has been trying to cash that cheque ever since. This week a man who runs a school built around it sat down with Huberman — so we went back to the original studies.
The thread this week: a number tells you where the ceiling might be. It never tells you how to reach it — and the two get confused constantly.
🔢 THE VITAL SIGN · LEARNING / COGNITION
2.00σ
How far ahead a one-to-one tutored student finished — enough to put the average one above 98% of a normal class.
The source: Benjamin Bloom, Educational Researcher, 1984 — reporting two doctoral studies by his students Joanne Anania and Arthur Burke. Fourth, fifth and eighth graders across a handful of classrooms were randomly assigned to whole-class teaching, to "mastery learning," or to one-to-one tutoring, then taught a three-week unit on probability or cartography. On the end-of-unit test the tutored group finished two standard deviations clear.
Bloom's point was never "hire a tutor." It was that the ceiling on an ordinary student sits well above what an ordinary classroom reaches, and the job is finding a group method that gets there. Everyone from Khan Academy to this week's guest is answering that challenge — which is why the number's provenance matters more than usual.
Educational Researcher, Bloom, 1984 · cited in this week's Huberman Lab show notes
🎙️ THE EPISODE · LEARNING / EDUCATION
Two hours of school, then go build something
Joe Liemandt on mastery over pace, why fluency is really a working-memory argument, and the part where they pay the children.
Joe Liemandt · Huberman Lab · 31 August 2026
Liemandt is a software founder who now runs Alpha School, a K–12 model that compresses academics into roughly two hours a day. He isn't a scientist and the results are the school's own, so hold them loosely. The reasoning underneath is still worth the listen.
The bottleneck isn't hours. It's being taught at the wrong level.
A class moves at one pace, so most of the room is either lost or idle. Liemandt's claim is that when software puts each student at their own edge and holds them there until the material is genuinely mastered, the academic day collapses to about two hours — and the rest goes to projects and life skills rather than more worksheets.
Drilling the basics is a working-memory argument, not a nostalgia one.
If your times tables aren't automatic, they occupy the same limited workspace the new idea needs. Fluency in boring sub-skills isn't the opposite of understanding — it's the room you understand in. Same logic at forty as at nine.
🎙️ Jump to ~43:40 for the scaffolding-and-working-memory stretch. It's the most transferable part of the conversation.
The blocker is usually shame — and they pay cash to get past it.
A student three grades behind doesn't have a knowledge problem first, Liemandt argues; they have a not-wanting-to-be-seen-failing problem. Alpha's answer is to pay students for hitting targets — blunt, obviously gameable, defended as a bridge over the first few weeks. Raise an eyebrow, then notice the claim underneath: the emotional cost of starting does more damage than the difficulty does.
🎙️ Listen → Huberman Lab, Joe Liemandt, August 2026
💡 DEEP DIVE · AUDITING A FAMOUS NUMBER
Half the miracle was a quiz
Paul von Hippel, a statistician at UT Austin, read Anania's and Burke's dissertations. Not a debunking — the experiments were decent — but it changes what you take from them.
The tutored children didn't only get a tutor. After every unit they took a quiz; anyone below 80–90% got corrective feedback and a fresh quiz, repeating until they cleared the bar. The third arm — ordinary whole-class teaching plus exactly that loop, no tutor — landed about 1.1 standard deviations above plain classroom teaching. Roughly half of Bloom's two sigmas, from a mechanism that costs nothing.
The rest deflates on contact with scale. Those were short, author-written tests on deliberately unfamiliar topics; the same tutoring on broad standardised tests averages about 0.27σ. Pool the modern randomised evidence — 96 trials in Nickow, Oreopoulos and Quan's 2020 review — and tutoring averages 0.37σ, not one study reaching two. Kurt VanLehn put human tutors at 0.79σ and computer tutors at 0.76σ: close enough that the tutor's humanity may not be the active ingredient.
Test yourself, find what you got wrong, fix it, test again. That's the transferable half — and it was never the expensive half.
von Hippel, Education Next, 2024 · Nickow, Oreopoulos & Quan, NBER, 2020 · VanLehn, Educational Psychologist, 2011
🔮 EXPECTED BREAKTHROUGHS · NEUROLOGY / DIAGNOSTICS
Alzheimer's became a blood test before it became treatable
Two more blood tests for the amyloid pathology behind Alzheimer's cleared the FDA in the last fortnight: C2N's PrecivityAD2, the first for symptomatic adults as young as 40, and Roche's Elecsys pTau217, the first single-biomarker test cleared to rule amyloid both in and out in the over-55s. Four are now cleared in the US, and Roche's runs on machines thousands of hospitals already own.
Neither is a screening test; both are read alongside a clinical assessment. But an answer that used to need a PET scanner or a lumbar puncture now needs a vial. Detection is getting cheap and early rather faster than treatment is getting good — which turns a technical advance into a personal question. Worth deciding whether you'd want the result before anyone offers you the test.
— Alzheimer's Association on PrecivityAD2, 20 August 2026 · Roche, 24 August 2026 · Rutgers Brain Health Institute, 1 September 2026
📰 NEWS BITS
Gina Poe: the brain has to go quiet before it can file anything.
Poe's work centres on the locus coeruleus, the brain's noradrenaline source. Her argument: it has to fall properly silent during the spindle-rich stretches of early sleep for the hippocampus to hand the day's material to cortex — and stress, alcohol and a jangled bedtime all keep it talking. A calm wind-down stops being a preference and becomes a precondition.
— Huberman Lab Essentials, Dr. Gina Poe, 3 September 2026
An air filter with a measurable test score.
119 adults aged 30 to 74 in Somerville, Massachusetts — a town wedged between two busy highways — ran a real HEPA purifier at home for a month and a convincing fake one for another month. The over-forties finished a mental-flexibility task about 12% faster on the real filter; under-forties, nothing measurable. One town, one task, small sample — but a randomised crossover rather than a correlation, which is rarer here than it should be.
— Scientific Reports, reported 30 August 2026
Headache, taken seriously for two hours.
Brian Grosberg calls migraine an invisible disease: roughly 12% of people, skewed heavily towards women, and covered in only a few hours across an entire medical degree. Cluster headache is routinely mistaken for sinus infection. His one practical instruction is unfashionably low-tech — keep a detailed headache diary, because the pattern is the diagnosis.
— The Peter Attia Drive #406, Brian Grosberg, M.D., 31 August 2026
P.S. If you've ever been properly tutored, we'd like to know what actually changed. Our bet is the correcting, not the company. Hit reply and tell us we're wrong.
What did you think of today's BrainBite?
🧠🧠🧠🧠🧠 The feedback loop was the whole trick. Slightly annoyed nobody said so.
🧠🧠🧠 Liemandt landed. The audit is the bit I'll be repeating at dinner.
🧠 Two sigmas of nothing, then.
We listen to every health podcast so you don't have to. Weekly. No noise.
Quiz it, correct it, quiz it again. That was the whole miracle. See you next week.
