In brief
How quickly can a person turn unfamiliar information into something they can recall? Memory-sport records provide measurable answers. Their most surprising pattern is that the information rate falls as the task becomes longer: extended mathematically to an 80-year lifetime, the fitted curve amounts to only about 6 MB.
Picture the task
Andrea Muzii memorised an 80-digit number in 9.75 seconds and recalled it correctly. Imagine doing that with this sequence:
58310472 90623158 17492063 80541729 36105894 72043615 98420137 65081724 43906281 17534890
Illustrative digits, not the sequence used in his attempt. Spaces only make the 80 digits easier to read.
First, what is a bit?
A bit is a unit of information: one choice between two equally likely possibilities. Eight bits make a byte. A Wi-Fi connection running at 100 megabits per second transfers 100 million bits each second; a 6 MB phone photo contains about 48 million bits. The memory study finds rates up to 42 bits per second in short elite tasks. These numbers share a unit, so the difference is informative: a photo file carries pixel detail, while a person can turn the important parts of a scene into a few meaningful cues. The mind works with a compact representation rather than a pixel-by-pixel copy.
Why look at memory competitions?
Competitors memorise items such as numbers, images or cards under time limits. Records let us compare how much information highly trained people retain across different tasks and durations. This is evidence about exceptional, practised performance—not a direct measurement of brain activity or the memory speed of an average person.
What do the results suggest?
In short tasks, reading the items takes up a large share of the available seconds; perception and memorisation may overlap. Across tasks lasting from seconds to an hour, the records follow a power law: the longer the memorisation time, the lower the information rate. The fitted rate is about 47.79 × T−0.36 bits per second, where T is the task duration in seconds. The relationship describes the records more closely than a single fixed rate for every duration.
What would that mean over a lifetime?
Multiplying the fitted rate by the task duration gives about 47.79 × T0.64 bits memorised. If that same relationship held through 80 years of uninterrupted memorisation, it would yield roughly 50 million bits, or 6.2 MB—about the size of a single phone photo. The records themselves extend only to one hour, so the lifetime figure extends a measured pattern far beyond the observed tasks. It concerns new information in this memorisation model, not a count of everything a person knows. That small-looking number also makes the role of existing knowledge vivid: a familiar concept, image or place can carry meaning without needing to be learned again from scratch.
What remains to be understood?
The records do not isolate reading, association and retrieval, or show how much of each happens in parallel. It is also unknown whether the same power law continues beyond the measured durations or applies to ordinary learning. Understanding why longer tasks require disproportionately more time is the next question raised by the data.
Why it matters beyond competition
Memory athletes turn abstract material into images, stories and routes through a memory palace. Years of practice make these codes fast, but they build on ordinary human strengths in imagery, narrative and spatial memory. A few cues can evoke a richly structured scene. This is one way to see human intelligence beside computers and language models: digital systems can move enormous quantities of bits, while people use meaning and prior knowledge to make selected information memorable. The record rates describe trained memorisation, not a direct speed contest with a machine. Exploring how to use these techniques for everyday learning guides my courses and independent practice on Unicorn Memo.