- Why Steroids Are NOT Functional – Don’t Trade Your Health for Muscle
- How to Keep Leveling Up INFINITELY – Like Sung Jin-Woo
- The Ideal Physique is Easy for Most Guys When They Learn This – Toji Workout
- How to Train Your FOOT Muscles for Balance, Power, & Injury Prevention
- How to Do Sit Ups CORRECTLY for Ripped, Powerful Abs
- How to Train Your Nervous System Like a NINJA
- Pike Push Ups are Good and You Should Probably Do Them, Maybe
- Supercharge Your Mitochondria for Energy, Endurance, And Longevity
- Calisthenics will change you.
- How to Track and Progress Multiple Goals at the Gym… And Win!
Can Mind Uploading Ever Work?
Warning: May induce existential dread.
Here on The Bioneer, I like to speculate about the future of human performance. Often, that means contemplating the potential impact of transhuman technologies. My views on these are varied and changing, but it’s no doubt fascinating stuff.
These are technologies that promise to enhance our bodies beyond the limits of what we consider human. Examples include prosthetic limbs and augmentations, brain chips, and genetic modifications. Perhaps the most seemingly out-there technology, though, is mind uploading. The notion that we will someday be able to upload our entire consciousness to a computer and exist as an immortal, digital being. This mind uploading is also called “whole brain emulation.” Don’t tell Nintendo.
But is this really possible? How can it be?
The answers may surprise you. And confuse you. Let’s dig into what the science says and what we can infer from that research, and what we already know.
Credit Where Credit is Due!
Before I go further, I want to credit my source. This post is very heavily inspired by a similar essay from author Ramez Naam, included at the back of his fiction book Crux. Crux is a story about brain implants and mind uploading and Naam, being a scientist himself, included this essay to demonstrate just how possible the story could be. He collated some really fascinating research in that text, so I wanted to share it here. But I highly recommend checking out the original book and the two others in the series (Nexus and Apex – Crux being the second in the trilogy).
How Would This Work?
So, how could something like mind uploading possibly work? Most likely, the answer would lie with brain simulation. A computer with sufficient power could, theoretically, simulate a human brain. Just as games designers can create digital cities with night/day cycles, traffic, weather, and a bustling populace with its own schedules and habits… so too could a computer scientist model the human brain and its behavior. With a sufficiently complex physics engine and roughly 86 billion digital neurons, each programmed to behave they would in the real world, we could theoretically create life within a digital environment.
And in fact, this has already happened. Several times, in fact.
IBM’s Blue Brain project has previously run similar simulations (similartions?). One such artificial neural network aimed to simulate half of a mouse brain, running on the Blue Gene supercomputer. This followed a previous simulation that simulated a mouse neocortical column – the smallest functional unit of the neocortex (reference, reference).
In 2013, researchers from Japan and Germany went one step further by simulating 1% of the human brain. This simulation modelled a network of 1.73 billion nerve cells and 10.4 trillion synapses. This was made possible by the K Computer, which was the 4th fastest supercomputer in the world at the time.
Researchers from Japan and Germany went one step further by simulating 1% of the human brain.
Incredibly, you can actually download some open-source models of the human brain, free of charge, from sites like Github. You can also find some of my programming tutorials there…
The world is crazy right now.
We Need More Power!
Before we get too excited though, there are some serious setbacks and limitations to keep in mind.
Firstly: these simulations are only representing fractions of brains. The sheer amount of data processing necessary means that even today’s super computers are woefully ill-equipped to render a full, human brain in real time.
Not only are these small cross-sections of brain, but they also run extremely slowly. One simulation Naam discusses reportedly took 600 seconds to calculate 1 second of real-world brain activity. As the brain sample grows larger, this processing time will also increase, exponentially.
Of course, though, technology is improving all the time. As devices get more powerful, we could start to see entire brains like these emulated in real-time (this is several steps short of mind uploading, as we’ll see in a moment). Inevitably, they will eventually overtake the speed of the flesh-and-blood brain – which could result in a superintelligence! (This might take longer than you would assume, however. We’ll get to that in a moment, too.)
See also: Does Moore’s Law Still Apply Today?
For a digital human brain running at super speed, the subjective experience of being alive would be akin to time slowing down. We could live for decades in the space of hours.
We could live for decades in the space of hours.
Seeing as such an intelligence could self-replicate in the blink of an eye, assimilate more information, and more… this would no doubt bring about the singularity. If it hadn’t already occurred.
See also: Building a Superintelligence: AI vs Exo Cortex
But again, we’re getting ahead of ourselves. Like, big time. There are still quite a few more issues to raise before we’re even close to true mind uploading.
Mind Uploading… Or Body Uploading?
Because, another issue with mind uploading, is that it renders us disembodied. Being a mind without a body to inhabit, may in fact be contradictory.
This would no doubt bring about the singularity.
As I argued in a previous article (The Mind Body Connection – Movement and the Brain), our brains and bodies are inexorably linked. We need a body to interact with our environment, and interacting with the environment is what our brains are for. Embodied cognition demonstrates how we might use our awareness of our own bodies and senses in order to make sense of our own thoughts. Essentially, we understand what people say to us by visualizing, sensing, and simulating the meaning based on our experiences. If someone talks about walking through the woods, you understand that by activating neural pathways as though you were walking through the woods!
See also: Embodied Cognition – You Think With Your Body
This is easier to solve, however. You simply would need to provide the simulated brain with some kind of simulated body and some kind simulated environment. To begin with, this body should be familiar to the brain being uploaded. This could mean you also need to map the rest of the nervous system, muscular system, skeletal system, organs, etc. Mind uploading might not be enough on its own.
This is especially true, when you consider how other mechanisms throughout the body directly impact on personality and cognition. For example, it’s now widely acknowledged that the gut, and particularly the gut bacteria, play a large role in the production of hormones and neurotransmitters.
The Brain is, Uh, Complicated
And this is the other problem: in order to simulate the brain, we need to fully understand how it works.
(The same, of course, also holds true for the rest of the body.)
Simply scanning all the neural connections that make up an individual (referred to as their individual connectome) is not enough to ensure that the simulation will behave like that individual.
In order for this information to be useful, we also need an accurate model to simulate how all those connections actually work. This is the problem: the brain is immensely complex and there is more we are discovering all the time.
Neurons communicate via electrical impulses that travel between synapses (gaps between the neurons). However, this is not simply a binary signal. Many neurons will also release a range of different chemicals called neurotransmitters, for example. These chemicals will then impact how excitable (likely to “fire”) the post-synaptic neurons are going forward. Specific neurotransmitters act on specific neurons at particular receptor sites. It’s likely that we haven’t yet discovered all the neurotransmitters.
These effects can drastically alter our mood and increase the likelihood of us paying attention to certain events, or remembering them. It also affects networks on a local level to alter behavior through “short-term plasticity.”
The Unimaginable Complexity of the Human Brain
Neurons themselves are also capable of computation via “dendritic computation.” There are white “glial cells” that carry out a range of tasks to support neuron repair and function (and possibly even facilitate computation) that we don’t fully understand. Glial cells, by the way, outnumber brain cells. And they aren’t included in any brain scans thus far.
Brain plasticity means that the brain is constantly changing shape and growing new neurons. Our grasp on the rules that dictate plasticity are primitive at best, meaning that any simulation would likely remain “static.” Would such a simulation be able to create new memories? Change its mind? Learn new skills?
Recent research now suggests that “double-stranded breakages” in DNA may be fundamental in supporting rapid plasticity (article). This goes against a lot of assumptions we had about cellular function. It’s likely that countless more paradigm-changing discoveries will be made in the next few years alone.
What it Would Really Take to Emulate a Human Brain
Ideally, in order to create an accurate model of the human brain, we would first need a complete “theory of everything,” so as to properly emulate chemistry, physics, and quantum physics. In short, we’re miles off.
The simulations that have so far been carried out, therefore, are not truly emulating the workings of the brain, but rather estimating them. With sufficient data, we can predict when a neuron would be likely to fire, or not. Such artificial networks are referred to as “spiking neural networks.”
Even if we were able to calculate the brain’s behavior down to single molecules, the amount of computational power would be immense. Prominent transhumanists Anders Sandberg and Nick Bostrum suggest that a model based on stochastic behavior of single molecules wouldn’t be possible, even on a $1 million supercomputer, until the year 2111.
Indeed, many claims surrounding brain uploading have proven to be hyperbolic and overblown. Henry Markram, lead researcher of the Blue Brain Project, stated in 2009 that we would have a “detailed functional artificial human brain” by 2019. Only two years later, the project came under fire for being mismanaged and unrealistic. Makram was forced to step down from the role.
I mean, this guy is meant to be a lead researcher on an extremely expensive project… and I could have told him that was nonsense!
The Case of the Amoebae
In playing devil’s advocate in his think-piece, Naam uses a really great counter-argument: the amoebae.
An amoebae can remember the places it has been. It can remember where it found food. And it can hunt. It is able to do all those things despite being significantly less complicated than a single neuron. We don’t know how it does all that.
So, if a simulation were to be as basic as simply handling inputs and outputs, it would likely be missing a LOT of what it means to be alive; never mind what it means to be human!
Machine Learning and the Brain
A glimmer of hope lies with machine learning. If enough data can be collected regarding the behavior of the brain, then we might be anticipate and model the behaviors of individual neurons without fully understanding them. The brain would remain a black box, but algorithms could run the show on our behalf.
Without a complete understanding of the human brain, we could never build one in a virtual environment. But with enough data and computational power, we could create a damn good impression of one.
Or perhaps machine learning and the singularity will lead to that elusive “theory of everything” and a complete understanding of the human brain. Maybe an AI will crack that code, resulting in true mind uploading. But I wouldn’t hold your breath!
Would it be You?
Assuming everything goes extremely well and we have an accurate simulation of the human brain in a virtual space, would it even be you?
For the answer to be yes, a minimum requirement would be that your precise arrangement of neurons – your precise connectome – be scanned and recreated in virtual reality. We are a LONG way from being able to do this. MRI machines simply don’t have the resolution to show individual neurons.
The only way around this would be to scan the brain destructively. Harvard researcher Kenneth Hayworth has an electron microscope that can be used to scan extremely high-resolution maps of the brain. For this process to work, however, Hayworth’s machine needs to replace the blood with a special plastic to stiffen the surrounding tissue. Following this, slices are made in the brain that measure just 30 nanometers thick (100,000 times thinner than a human hair).
So, yeah, your brain gets turned into plastic and then sliced into millions of tiny pieces.
You would only want to use such a process after you were dead. This might not seem like a big deal, seeing as most people consider mind uploading to be an “endgame” rather than an alternative to living.
But is it really you in there?
We’ve established that it probably wouldn’t be. If the simulation isn’t accurately simulating the behavior of subatomic particles, then the brain model won’t behave precisely as yours would have. It wouldn’t be you: just a spooky copy.
Your brain gets turned into plastic and sliced into millions of tiny pieces.
A Thought Experiment
But for argument’s sake. If it WAS a perfect emulation… now is it you? Would you have subjective continuity?
This is the territory of deep philosophical debate. My gut instinct is to say no. Why?
Because you could theoretically exist at the same time as your upload. If there a way to non-destructively scan the brain, then you could exist both in the flesh-and-blood sense AND the digital sense simultaneously. Would you experience a divided consciousness in this case? I very much doubt it.
If anything, having a digital upload would be more like having a twin.
A digital upload might be more like a twin.
But there is a fascinating counter-argument to all this. That is as follows: if your flesh and blood body ceases to exist and you wake up in a digital environment, how would you know if you were still “you?”
The answer is that you wouldn’t. And if you pull on this thread, you realise that the same thing happens all the time. Your brain is constantly changing shape and your cells are constantly dying. Are you truly the same person as you were 10 years ago? Or do you simply share that person’s memories?
Are you truly the same person that you were one second ago? Or is the entire experience of continuity entirely illusionary?
Are you in fact, simply information? Is that all that matters?
Closing Comments
Answering those last questions is a real test of any materialist stances. I don’t consider myself spiritual in any way (or any the worse off for it). However, even I find it hard to think that simply recreating the “information” that makes up my brain is sufficient to make a new “me” with my subjective experience of consciousness. Perhaps there really is something more.
Either way, it seems highly doubtful that even that will be possible in our lifetimes. Or our children’s lifetimes. Or theirs even. This is singularity stuff, though there is the faintest chance it could happen someday.
But our attempts remain valuable, nonetheless. An imperfect simulation of the human brain could teach us a huge amount about ourselves. It could lead to other, future breakthroughs. It might even be the spark we need for true AI to take off.
At the very least, it forces us to confront some fascinating questions about ourselves.
Facinating article. If only the brain was simpler to understand but then we would be too simple to understand it.