Our brains can remember a ton of stuff.

We can recite the lyrics to our favorite songs years after we learned them. Celebrity gossip and memes  live rent-free in our heads.

Not to mention the stuff we take for granted, like recognizing tens of  thousands of words without issue. If we treat our brains as computers, they can hold 55 million  ebooks worth of information. Which is wild.

But what if we could replace our squishy human brains with something else? Say, a bunch of hard drives, or the latest cutting-edge data storage technology. Could we cram even more info in our heads?

And if so, is there a limit to  how much we can pack in there? [♪INTRO] As a baseline, let’s re-examine how much info our brains can store as-is. To do this, we’re going to analyze our brains as though they’re computers. This isn’t the most perfect approach, but it makes the math a whole lot easier.

In a computer, information  is stored using a fundamental unit called a binary digit, or bit for short. Each bit can be in one of two states: a 0 or a 1. As you add more bits, you can  store exponentially more info.

So two bits can store 2^2 or  4 states: 00, 01, 10, and 11. Three bits can store 2^3 or 8 states. Four bits can store 2^4 or 16  states, and so on and so forth.

To store information, we can map different states to different numbers or letters. Originally, these individual characters were represented by a block of 8 bits. For easy bookkeeping, this  8-bit block was named a byte.

Both terms have stuck around, so sometimes we quantify information in terms of  bits, abbreviated with a small ‘b’, and other times we use bytes, with a big ‘B’. Which is a situation ripe for mistakes. But alas.

Now back to our brains. Scientists have tried to  quantify how much info our brains can store by finding  the brain’s version of a bit. And a really good equivalent is the synapse.

A synapse is the interface between two neurons, and its job is to communicate  information between them. It does this by continuously transmitting chemical or electrical signals, and  modulating their strengths. So if we treat the different signal strengths as different information states, we can determine how many  bits a synapse can store.

Using a combination of neuroscience and computer information theory, scientists have estimated that a single synapse can store between 4.1 to 4.6 bits of info. On average, a human brain  has 250 trillion synapses, which means our brain can  hold 143 terabytes of info. If a typical ebook is 2.6 million  bytes, that’s about 55 million ebooks.

But what if we could replace our  brains with actual computer parts?  How much information can we store? To start, let’s try swapping out our  brains with a hard disk drive or HDD, which is what many computers  use to store their data. HDDs get their name because they store data on a spinning disk or platter  coated with magnetic material.

This platter is divided into lots of tiny, tiny sections that each represent a bit. And each section can be magnetized  in one of two directions, which correspond to either a 1 or 0. To store data, a needle  called a read-and-write head moves to a specific section and generates a magnetic field to change  the bit’s magnetization.

Now, to estimate how much HDD  storage we can cram in our brains, we need to know both their  physical and data sizes. HDDs usually come in a standard metal box that’s 147 millimeters by 102  millimeters by 26 millimeters. And right now, one of the highest capacity HDDs on the market is the Seagate Exos Mozaic 3+, which can store 36 terabytes of data.

Although you’ve gotta be a fancy  data center or business to buy one. But supposing we could get our  hands on those data-dense HDDs, if we pack them into the  same volume as a human brain, we would be able to hold 120 terabytes of data. Which is worse than what our actual brain can do.

Maybe we should be replacing  computers with brains. But wait! HDDs aren’t just pure data storage.

They also include auxiliary  parts that allow us to access and edit the stored data,  like the read-and-write head. If we throw these parts out  and pack in just the platters, we won’t be capable of having dynamic thoughts. But we can store around 1.2 petabytes of data, around eight times more than our actual brains.

But if you’re a stickler for being able to access your data and have actual thoughts, we can look at a different  form of computer data storage: the solid-state drive, or SSD. SSDs store data using a grid of electrical switches called floating-gate transistors. Each transistor stores one bit of data  by filling or emptying a reservoir, a.k.a. a gate, with electrons.

A filled or charged gate turns the  transistor “off” and represents a 0, while an empty or discharged gate turns the transistor “on” to represent a 1. SSDs are an alternative to HDDs that promise high data storage densities, with the caveat that they  also have a limit to how many times we can write to them  before they stop working. In other words, it's probably not  a good idea to use them inside the head of a person continually updating  what they know about the world.

SSDs come in all sorts of shapes and sizes, including the same standard  metal box that HDDs come in. But if we’re shooting for high data density, the way to go is the thin M.2 form  factor, commonly used in laptops. One of the most data-dense  M.2s is the WD Black SN850X.

This SSD can hold 8 terabytes of data in a space roughly half the size of a standard ID card. If we were to pack these  into a human brain volume, we could store 1.6 petabytes of data, which is equivalent to 600 million ebooks. That’s 30% more than just  cramming in HDD platters, and we can still access our data.

Yay for dynamic thoughts! But this episode is not  brought to by either Seagate or Western Digital, so now  we have to show you an ad. Thanks to Brilliant for  supporting this SciShow video!

Brilliant is an interactive online learning platform that lets you move at your own pace. Whether you prefer to take it slow, completing one short lesson at a time whenever it fits into your schedule, or make learning about algebra  your latest hyperfixation, Brilliant accommodates all  sorts of learning styles. If you’re like me, you want to do more, but don’t always make it over  that time or effort hurdle.

But with Brilliant’s flexible pacing, you can accomplish something every day and turn learning into a new healthy habit. To try Brilliant for free,  visit brilliant.org/scishow, scan the QR code onscreen, or click on the link in the video description. You’ll also get 20% off an  annual premium subscription.

Let us not limit ourselves to  typical computer storage technology. For example, we can go back  to biology for inspiration, and find we already have a very  efficient way to store information: our DNA. DNA stores our genetic information in a sequential ladder with different combinations  of four nucleotide bases: adenine, cytosine, guanine, and thymine.

Each rung on the ladder has four  possible base pair combinations, equivalent to two bits of information. For humans, one DNA molecule  has about 3.2 billion rungs, which means it can store 800  megabytes of information. And all of that info is packed into a tiny nucleus around 10 microns in diameter.

So if we were to just fill  our brain space with nuclei, we could store 2 zettabytes of data. That’s a 2 followed by 21 zeroes, and  equivalent to 750 trillion ebooks. We can also look at some up-and-coming computer storage technology.

One example is magnetoresistive  random access memory, or MRAM. Like HDDs, this tech stores data  magnetically, but with a twist. Or should I say, a spin.

Because  instead of one magnetic layer, MRAMs typically have two, with  an insulating layer in between. One layer of the magnetic sandwich is permanently magnetized in one direction, while the other’s magnetization  changes depending on the bit’s state. If a certain bit needs to be a 0,  the corresponding spot on that second layer is magnetized in the opposite  direction relative to the first.

If it needs to be a 1, it’s  magnetized in the same direction. And to swap a bit from 0 to 1, or vice versa, MRAMs manipulate the direction that the layer’s electrons are spinning through electric currents. By controlling the magnetization of these bits using currents instead of magnetic fields, MRAMs avoid a lot of the technological  limitations that HDDs have.

Which opens the door for high data densities. MRAMs are still mostly in the R&D  phase, but there’s lots of promise. Recently, researchers at Kioxia built  an MRAM prototype that can store 64 gigabits within an area smaller  than the smallest grains of sand.

If we could fill our brains  with this cutting-edge stuff, we’d store 1.6 yottabytes of data,  or 1.6 trillion trillion bytes. That’s not only about a thousand times more ebooks than our DNA-storage system. It’s almost 9 times the amount of data that the entire world will generate in 2025!

But we may be able to do even better than that… say, by mapping bits not  onto microscopic switches, or teeny tiny sections of a magnetized  disk, but onto single atoms. There are a bunch of ways to do this, but let’s look at one method done by scientists at the University of Alberta in 2018. They used hydrogen atoms to represent bits, placed on top of a thin piece of silicon.

If an atom is in a certain spot, that’s  a 1. If it isn’t there, it’s a 0. With this method, the team was  able to achieve a data density around 47 times higher than  our example SSD from earlier.

So if we were to stuff our brains  with this hydrogen atom tech, we could store 75 petabytes of data. In other words, about 30 billion of those ebooks. Which is kind of a letdown after the other numbers we’ve thrown around.

But that’s because this tech  requires auxiliary parts to work, including that bed of silicon  atoms for the hydrogens to sit on. To get numbers that are more  fun, let’s pretend we could perfectly map one bit to one atom  without any additional overhead. And to really push the limits, let’s use the atom with the  smallest radius: helium.

If we replaced our brains with helium atoms… ignoring any weird laws of physics  getting them that close together… we could store around a  million zettabytes in our head. That’s 5500 times more than the expected amount of globally generated data in 2025. And we can keep playing this game, and think about storing  data in subatomic particles, like protons or electrons.

Or  even weirder stuff like quarks. But instead of going down  that particular rabbit hole, let’s dive into another hole: a black hole. Black holes are the upper limit of how much info we can pack into a given space.

In other words, if we were to keep  packing more and more info into our heads, eventually our heads  would collapse into a black hole. To which, I say, no thank you. But if we could somehow live through that, we can actually calculate how much info our black hole brain could store, thanks to Jacob Bekenstein and Stephen Hawking.

Through a combination of their work, we know the information density of a black hole… which stores its information  not within its 3D insides, but just its 2D “surface”. Yes, that is weird. No, we don’t have time to get into how that works.

But it’s got something to do with gravity. If we use their math, and treat our  black hole brain as a sphere with the same volume as a human  brain, we can estimate that this black hole brain can store  around 10^55 terabytes of info. That’s 10 million trillion trillion  trillion trillion terabytes.

If we were to maintain our 2025  global data generation rate, it would take us 5.5 x 10^43 years to generate this amount of information. That’s so far in the future, it’s  not just our star that will be dead. Basically every star will be dead, and the universe will be in its Black Hole Era.

That’s the actual scientific name, I swear. Now, using black holes to store information might sound like a fun “what  if” thought experiment. And to some extent, it is.

But the idea holds more water than you’d expect. One researcher, Gia Dvali, has proposed  a theoretical artificial neural network that can store information  using gravity, just like black holes. Specifically, in this neural network,  artificial synapses communicate through gravity rather than  chemical or electrical signals.

With this method, the neural network could achieve a similar style of  information packing as a black hole, without actually being a black hole. Which personally, I’d still turn down. Even if it would help me  remember literally... anything.

My head is swiss cheese. [♪OUTRO]