"Any way you look at it, all the information that a person accumulates in a lifetime is just a drop in a bucket."
Taken from 1995’s Ghost in the Shell, this sentence perfectly sums up the complexity and ambiguity of the human mind and the brain. Despite enormous advances, our ignorance on what lies within our cranium remains vast and sometimes unyielding. Those drops, and their relation to the bucket, are profoundly enigmatic.
We understand how different parts of the brain communicate and often what they specialize in, but what exactly is a “thought” made of? Where, amongst the 86 billion neurons in our brain, do they come from, and where do they go? What, scientifically speaking, is consciousness – is it a single entity or the sum of trillions of permutations? Is it the addition or average of all those firing neurons?
The gulf of knowledge is clear, but science fiction gleefully disregards it. From the aforementioned classic to the Netflix shows Altered Carbon and Black Mirror, uploading our brains or minds to a machine or a computer interface of some kind is simply a piece of cake.
So will we ever be able to do this or is this just a pipe dream? Plenty has been written about the future and what we may be able to do one day, but not much attention gets paid to the hurdles we have yet to overcome. Forget all the techno-babble, philosophy, and transhumanism – how close is this brave new world to our present time?
No matter what you call it – uploading the mind, mind transfer, or whole brain emulation (WBE) as it’s technically referred to on occasion – we absolutely don’t have the ability to do this now or anytime soon. The brain is too complex, and the nature of “mind” and “self” too uncertain, to even know what exactly we’re transferring.
Think about what we’d have to transfer. All our thoughts, recorded feelings, memories, intellect, cognitive abilities, fears, hopes, dreams, and so on. Additionally, those neural connections we currently have would have to remain intact in our new digital palace or we simply wouldn’t be the same person we once were, in a manner of speaking.
Our limited knowledge of the workings of the brain is a titanic obstacle that we’re nowhere near close to overcoming. Not knowing what exactly it is we’re transferring in the first place means that we can’t engage in any WBE without a biomedical revolution taking place.
At the same time, there’s a data storage problem. The numbers are deeply imprecise at this stage, but the human brain is thought to be able to store between 1 terabyte and 2,500 terabytes of data.
Storing all that on a computer would be a colossal undertaking, considering that only the world’s most powerful supercomputers – like Oak Ridge National Laboratory’s terabyte-heavy beast “Titan” – have the capacity to store your memory in that way.
It’s unlikely we’d use a digital architecture for WBE. After all, biology is a far more efficient storage mechanism. A recent study revealed that just a single cubic millimeter – a tiny drop – of synthetically built DNA can store 1 billion gigabytes’ worth of data on it, whether that be encyclopedias or selfies.
Forget storage space, though: How would we physically store the mind on a computer in the first place? Leaving out the still-emerging field of quantum computing, conventional computers operate using bits, which are digital yes/no signals represented by ones and zeroes.
Thoughts and neural impulses aren’t digital entities. They’re more akin to a spectrum of bioelectrochemical pulses, in which information flows freely. They do not communicate at set rates like computers currently do.
Yes, there are chips that mimic our neurons, and they are very cool. Using artificial synapses, they have replaced these 1/0 logic gates of old and instead send highly accurate pulses of information across a crystalline gap whenever needed. This doesn’t mean that they are artificial neurons, though – they’re just imitations.
Yes, we also have neural networks, which refers to an artificial intelligence (AI) that exists on a computer network made up of interconnected nodes. They can learn without any human input at their best, and they’re somewhat better than us at pattern recognition, but again, their design merely mimics our own biological neural networks. They’re not a copy we can paste onto, they’re inferior versions of our brains.
There's more to neurons than just electric pulses, mind you.
"Neurons do not exist in a vacuum: they are surrounded by chemicals that affect them, and release chemicals that affect nearby neurons," Matthijs Dorst, a computational neuroscientist researcher at the Karolinska Institute, told IFLScience.
"Do we need to include all these processes? Do we need to compute every single chemical interaction?"
Whichever way you cut it, the meeting of mind and machine is not about to emerge over the horizon.
I’ve Seen Things You Wouldn’t Believe
It can’t be overemphasized that we don’t quite understand how thoughts, or even emotions, work at this point.
We’ve repeatedly attempted to communicate with the brain, and transfer thoughts into digital responses, in a wide range situations. We can invent prosthetic limbs controlled by thought processes, but using motor (movement-based) neuronal impulses to move a metal arm is very different from transferring a thought into a machine.
Even the US military’s research unit, DARPA, has barely made inroads in this regard: A recent endeavor called RE-NET used 100 brain implants to register distinct neurological activity on a computer, but it was often inaccurate and full of noise.
We can, based on MRI scans of our brains, make predictive models that let a machine know what you’re about to do. In one example, researchers let people watch movie trailers while they identified which segments of their brains’ visual cortices lit up.
After enough time, a computer program learned to reconstruct the visual images they were seeing based on this information. Very cool, but this is still predictive – it’s not directly understanding our thoughts or neuronal impulses.
At this point, we know we can implant false memories into different brains using light. That’s amazing, but if we can’t directly communicate with machines, then there’s no hope in hell when it comes to transferring our mind – whatever that actually is – into a computer system that doesn’t currently exist.
If we could somehow use a low resolution transferal mechanism to upload what we could onto a machine that can't handle our brain's complexity, we'd be a grim shadow of our former selves.
What we need, ideally, is a system that is a near-perfect reflection of the human brain, but that’s going to supersede the biological limits of our biology; one that doesn’t decay and die. Simulations of a brain-like system have emerged, but scaling these up would require mindboggling power sources and a heck of a lot of processing power to be able to generate unique thoughts from a seemingly infinite well of information.
"Even for simple models, the fastest supercomputer needs an hour to compute what the brain does in a few seconds," Dorst points out.
Based on Moore's Law, which posits that computing power doubles every two years, "we may have supercomputers fast enough to simulate as many neurons as the human mind somewhere in twenty to thirty years."
Monkey Needs A Hug
Despite this chasm of uncertainty, it hasn’t stopped some experts planning for a future where virtual immortality is achievable. Some endeavors appear more advanced and rigorous than others, and there are non-profits and international initiatives out there aiming to promote and advance WBE technologies.
In the meantime, researchers and futurists will continue to try to find ways to get computers and brains to converse and exchange information directly. After all, we don’t fully understand how anesthetics works, but that doesn’t stop us using them.
Here’s the thing, though: Comparing our brains to computers, and assuming they’re similar enough to be compatible, might not make that much sense. Throughout history, the brain has been compared to water clocks and mechanical beasts.
Today, the analogy is made with computers.
It’s almost certain that, if we ever achieve WBE, the technology that we use won’t be a computer – or, at least, what we think a computer is. Science may surprise us, but we suspect it’ll take a technological marvel unlike any we currently know of before we can start putting our ghosts in any kind of shell.
Either way, we might have some competition. The predictive capabilities of technology, from targeted ads to those MRI-based models, are fairly astounding. Perhaps we're not as complex as we like to think.
"Maybe the human mind is overly-redundant, and we could get the same behavior with a much simpler model. Relatively simple computers can already beat the best of us at Chess, Go, even Jeopardy," Dorst notes.
"Perhaps they can also one day beat us at consciousness."