Can We Upload a Brain?

Have you seen Transcendence?

It’s a 2014 sci-fi film about artificial intelligence. It’s alright. Well, it’s pretty average. But it does have some cool explosions and commentary on the future of artificial intelligence. It’s the kind of movie you can put on for a passive, empty-headed Sunday afternoon.

Basically, Johnny Depp is an artificial intelligence researcher named Dr. Will Caster (handsome first name, that) who, after being shot by anti-tech terrorists, has his conciousness ‘saved’ by his wife, who uploads his mind into a computer.

Predictably, bad things follow.

All this aside though, it does prompt some interesting questions:

Could you actually upload a brain into a computer?
How big is a human brain?

I’m no scientist. But I have really strong Google-Fu. And this is what I found out.


Could you actually upload a brain into a computer?

Astoundingly yes! This has actually already happened.

Sort of.

It’s not ‘uploading’ per se — like sending some photos on your iPhone to the Cloud. Instead, the idea is to map a brain completely and then recreate that in software.

The Open Worm Project did this with the round worm Caenorhabditis elegans — and you really ought to take a moment and say that awesome name out loud.

wormbook.org

Caenorhabditis elegans has only about 302 neurons which is positively tiny compared to the number of neurons estimated to be in a human brain — but hold that thought.

C. elegans is our Johnny Depp worm.

The Open Worm Project mapped all the connections between these neurons to create a full connectome. Connectome is an awesome word:

So, essentially a map of all the connected neuron roads and highways within a brain.

This round worm connectome was then emulated in software and put in a LegoBot — just to see what would happen. Because science.

This is where it gets both exciting and kind of creepy.

So, you’ve mapped an organism’s brain and emulated that in software. But, you haven’t given this software any commands or processes to run — you just put it in robot. The Johnny-worm’s nose neurons were rewired to a sonar sensor on the robot and its creepy-crawling neurons to motors on each side of the LegoBot.

This is what happened:

That’s the Open Worm LegoBot moving around, encountering obstacles and looking for other routes — all by itself. There was no programming other than the freshly-squeezed-from-nature connectome. LegoBot is just doing what Johnny-worm would do.

Granted, it’s not emulating everything C. elegans would do at this stage, they haven’t gotten that far yet —

But the implications of this are astounding.

The worm’s behaviour is dictated by it’s connectome — by how its wired up. A really complicated, really tiny electric Rube Goldberg machine. A machine that, once mapped, we can recreate with software.

Not only does this hold the potential for immense insight into how the brain actually works and where things go wrong, such as Alzheimers — but some pretty sci-fi sounding concepts.

But hold up a second.

C. elegans is currently the most studied and understood organism in the world (of those that have brains, anyway). It’s also a pretty simple creature — it’s comprised of only around 1,000 cells.

The human brain is something far, far more complex.


How big is a human brain?

It’s kind of funny to me that this was the harder question to answer, because, well, nobody really knows for sure.

I mean, in physical terms, brains are around 1.3–1.4 kilograms and between 1130 and 1260 cubic centimetres. But that’s just our hardware. Or, jellyware, I suppose.

What about our software? How big is that?

Well… how big is our brain storage capacity? How much stuff can we store in a brain?

Estimates seem to vary from 1 terabyte to 2.5 petabytes.

1 terabyte is 1000 gigabytes. In terms I can conceptualise, that’s 31-and-a-quarter 32gb iPhones. That doesn’t seem like that much for a whole human brain.

2.5 petabytes is 2,500,000 gigabytes. That’s 78,125 iPhones at 32gb a piece. Somewhat more flattering.

A side note: If you were to text the contents of your 1 terabyte brain from your phone to someone else’s phone, this would require about 71,428,571,428.6 text messages. And at 20c per text message this would cost two $2,857,142,857 and fourteen cents. I have no idea why you would do that, but I just thought it was interesting.

But we just don’t know for sure. In terms of neurons, while we certainly do have a bucket load more neurons than C. elegans we also are not one hundred percent sure on that figure either.

For a long time, the accepted number was around 100 billion neurons. Recently however, that number was put into question and estimates were more like 86 billion neurons.

We know so little about our own brains. I think that’s just fascinatingly bizarre.

We haven’t had an upgrade in jellyware since our ancestors hunted antelope on the savannah. Today, we’re using that same brain to build machines to transmit invisible information through the air, land fragile machinery on other planets and copy an organism’s brain into an autonomous robot.

But we still don’t quite know how we’re able to do all that.

Our brains are really, really complex. But they are just jellyware. With enough resources its foreseeable you could create a full human connectome — in fact there are projects under way to do just this, such as the aptly named Human Connectome Project.

So once we have that detailed map of our own skull-clad Rube Goldberg machines and provided we have the requisite storage capacity, what’s to stop us from putting that data into a complex LegoBot and seeing what happens?

Ethics, mostly. But that’s a whole other question.


Still, how many decades might it be before insurance companies offer high level packages that include Connectome Protection?

You might have regular medical check ups that include a full connectome scan — a sort of Save File for your mind. In the event of a catastrophic car crash you can simply download a copy of your connectome into a new, artificial body. See: science-fiction novels/films.

That’s pretty out there — and riddled with ethical and philosophical issues: would this connectome clone be you? What makes you, you? To be fair, Transcendence touched on this.

But the fact is — it’s looking like all of that is not impossible.

Weird.


I have no affiliation with Open Worm, but I think their work is really fascinating. If you’d like to help them, you can donate here.

And if you want to dissect a virtual C. elegans you can do that here.


Probably the best thinker around on these sort of questions is Ray Kurzweil who poses a host of concepts that will either make you feel very scared or very excited. Or both.