Chapter 10: The Future of Consciousness

Welcome to Last Minute Lecture.

This free chapter overview is designed to help students review and understand key concepts.

These summaries supplement, not replace, the original textbook and may not be redistributed or resold.

For complete coverage, always consult the official text.

Imagine a future where your thoughts, your consciousness isn't just stuck inside your head.

A future where you might exist as pure data.

Or I don't know, maybe even light.

Sounds like pure science fiction, right?

It really does.

Like something straight out of a novel.

Or maybe a really intense Black Mirror episode.

Okay, so let's unpack this a bit.

Let's do it.

Today we're doing a deep dive into the future of consciousness, specifically, you know, where mind and machine might meet.

Our mission is not just to explain mind uploading, but really to dig into the neuroscience, the philosophy.

Can an uploaded you actually be you?

And it's fascinating because, as our source highlights, this isn't some brand new Silicon Valley dream.

Oh, really?

No.

JD Bernal, an Irish crystallographer, predicted something like this a hundred years ago.

A century ago.

Yeah.

In the world of flesh and the devil.

Yeah.

He basically envisioned humans gradually replacing body parts, then the brain itself.

Until consciousness becomes, well, like you said, maybe pure light or something totally non -biological.

It's a really ambitious starting point.

Okay, mind -bending stuff already.

Let's maybe pull back to today for a second.

What are the actual efforts now to merge tech and biology, like putting electronics inside the skull?

Where are we with that?

Well, it's definitely happening.

It's not sci -fi anymore.

But the reality,

it's complicated.

How so?

Huge hurdles,

scientific, clinical, legal,

ethical.

It's a minefield.

Right now, these brain -machine interfaces, BMIs, they're mostly for patients who've lost function, you know, severe stroke, tumors, that sort of thing.

So it's quite limited in applications still.

Very limited.

And the state -of -the -art tech, it might surprise you.

The standard implants, these Utah microelectrode arrays, they're based on tech that's 30 years old.

30 years?

That's ancient in tech years.

It really is.

It predates most smartphones.

But, and this is where it gets really interesting, right?

Despite that old foundation, you've got startups like Neuralink pushing hard.

Exactly.

They're designing much smaller, more flexible, much more powerful devices.

And what do these new devices actually do?

Two key things.

They can read brain signals so we can interpret what's happening.

Okay.

And they can also impose electrical patterns onto brain tissue, kind of like writing signals back.

Right.

And the prediction is, what?

Faster adoption?

Yeah.

The projection is that within a decade, we'll see a big acceleration in patients getting these advanced BMIs.

For things like?

Things like helping with impaired vision, restoring movement for paralyzed patients, maybe even giving speech back to people with aphasia.

Real life changing stuff.

That's incredible for restoring function.

But thinking back to Bernal's idea,

it feels like the bigger dream for some isn't just fixing things.

No, definitely not.

It's that whole shuffle off this mortal coil idea, isn't it?

Replacing our squishy, finite brain with something synthetic, something potentially immortal.

That's the ultimate goal for many in this field, yes.

So let's define it clearly.

This mind uploading.

What exactly are we talking about?

Okay.

So mind uploading basically hinges on three big tech pillars.

Which are?

First, being able to record brain activity comprehensively.

Second, stimulating it precisely.

And third, simulating its functions using computers.

Powerful computers, obviously.

So you need software that mimics everything, your personality, memories.

Exactly.

Software that replicates your unique traits,

responses, the whole package.

And then the key step,

actually transferring your specific brain state to a computer.

And the hope is, if the model is good enough.

The hope, or maybe the assumption, is that if you model it right down to the neuronal level,

that digital copy will actually feel like something.

It'll be conscious.

It'll be you.

Which immediately flags a huge philosophical question.

If you perfectly copy all the biological bits and pieces, the chemistry, the physics, does the simulation automatically get the feelings too?

How do you see that?

That's the absolute core of the debate.

I mean, if there's no magic soul, no ghost in the machine, then how could a perfect simulation not be conscious?

It really blurs the line between a digitized human and, say, a truly sentient AI.

It's a deep one.

And this whole vision, this transhumanist project,

it promises the moon, doesn't it?

Superpowers, immortality, perfect memory.

Oh yeah, the promises are huge.

Enhanced intelligence, never forgetting anything, living for ages.

We see it everywhere in fiction, Lemn, I .N .M.

Banks, The Matrix, Westworld, Black Mirror, humans.

The list goes on.

But what's really fascinating there is that, unlike the super optimistic transhumanist view, a lot of that fiction is actually pretty dark, dystopian even.

That's true.

It often goes wrong.

It really does.

It highlights all the potential downsides, the unintended consequences, it reflects this sort of deep societal unease, I think, with where this tech might take us.

So bringing it back to today,

where does mind uploading realistically stand?

Is it just wildly implausible or is it more like it's incredibly difficult but maybe possible down the line?

That's the million dollar question, isn't it?

And maybe for you listening, if this tech ever arrives, would that digital copy really be your mind or just a very, very convincing deep fake?

Well, if we look at neuroscience,

it's a young field, incredibly active.

Some call it a golden age.

Because we're learning so much.

We are.

More learned in the last decade than all of history before it, arguably.

But there's this illusion of swift progress.

Meaning?

Meaning the raw data we collect from brains doubles every couple of years.

But our actual understanding of what it all means, that moves glacially slow,

painfully slow.

So lots of data, not as much insight yet.

Pretty much.

You know that Churchill quote, now this is not the end, it's not even the beginning of the end, but it's perhaps the end of the beginning.

That perfectly sums up where neuroscience is.

We're just starting to scratch the surface of the brain's basic operating principles.

Unlike, say, vaccine development.

Exactly.

With the COVID vaccines, immunologists had a deep mechanistic understanding of viruses.

We could target things precisely.

Neuroscience is nowhere near that level of understanding for the brain itself.

And that lack of deep understanding has real world consequences now, right?

Like diagnosing mental health conditions.

Absolutely.

If you have chronic headaches, depression, anxiety,

there's no brain scan or blood test that gives an objective diagnosis.

It's still based on symptoms, questionnaires.

Precisely.

We still rely on psychiatrists interpreting symptoms because we lack those fundamental biological markers.

We just don't understand the mechanisms well enough yet.

Okay, so to really upload a mind, the dream for technologists is mapping the connectome, the brain's complete wiring diagram.

Can you break down synapses for us?

Why are they so vital?

Synapses are basically the connection points.

They link the output wire of one neuron, the axon, to the input part, the dendrites, of the next neuron.

And they're numerous.

Incredibly.

One axon can connect to thousands of other neurons.

Compare that to a computer transistor, which usually connects to just maybe three or four others.

Huge difference in connectivity.

And these connections aren't fixed.

No, that's crucial.

Each synapse has a weight, how strong its influence is.

And this weight is constantly changing, adapting based on activity and experience.

It's dynamic.

What kind of scale are we talking about inside our heads?

Get this.

A piece of human cortex, the size of a sugar cube.

Tiny.

Yeah, tiny.

It has over 100 million neurons and a trillion synapses.

A trillion?

Wow.

Yeah.

It's comparable in parameter count to the biggest AI language models today, but our synapses are constantly adjusting.

They're alive, in a sense.

So the connectome is the map of all those trillion connections.

How hard is it to actually create that map?

It's almost unimaginably hard.

The analogy used is trying to trace one single strand of spaghetti in a bowl of a billion strands without messing up.

OK, yeah, that paints a picture.

You have to follow every single tiny branch of every axon across the whole brain.

And they're so thin, you need electron microscopes.

Light just isn't good enough.

So what's the state of connectomics now?

Have we mapped anything completely?

Well, the very first complete map was for a tiny worm, C.

elegans, just 302 neurons.

When was that?

That was finished back in the mid 1980s, mostly manual work.

And since then?

It took over 30 years, plus huge leaps in AI hardware, software,

before we got the second one in 2023.

The fruit fly brain.

Which is much bigger.

About a thousand times bigger than the worm.

So progress is slow.

So a mouse brain connectome, when might we see that?

The estimate is maybe not before 2030.

And that's a mouse brain, 70 million neurons, hundreds of billions of synapses.

And how does mouse brain tissue compare to human?

Is ours fundamentally different, somehow special?

You know, it's surprising, but fundamentally, human brain tissue is very similar to mouse cortex.

We have this instinct that our brains must be unique.

But there's actually no hard evidence for that at the cellular level.

There are small molecular differences, sure, related to our niche, but nothing radically different in the basic building blocks.

So if it's not the building blocks, what is the big difference?

Scale.

Pure, staggering scale.

How much bigger?

The human brain is roughly a thousand times bigger than a mouse brain.

It has about half a million kilometers of wiring inside.

Half a million kilometers?

Yeah, that's further than the distance from the earth to the moon.

That's insane.

And the data?

Colossal.

Estimated around a billion terabytes to map it all.

And does that sheer size translate directly to capability?

It seems to.

Just like adding more layers to an artificial neural network lets it learn more abstract things, a bigger brain with more connections seems to provide vastly more reasoning power.

That's why a human brain can do things a mouse brain simply can't.

Size matters here.

Okay, but let's say we overcome that massive challenge.

We map the whole human connectome.

What then?

The source material says it's like looking at a dead body.

Exactly.

A static map is just that static.

It tells you the structure, but not how it worked, what it thought.

So you need to animate it.

Precisely.

You need the animation of the connectome, simulating all the dynamic electrical and chemical activity, the ion flows, the learning rules, the adaptation.

And how far are we from doing that?

Well, the current cutting edge is simulating a piece of mouse cortex about the size of a grain of quinoa.

A grain of quinoa?

That's tiny.

It is.

To scale that up to the whole mouse brain, you need about a million times more computing power and effort.

And then to a human brain.

Another million times harder still.

Yeah.

The scale is just exponential.

Wow.

And besides the sheer computational scale, you mentioned a skeleton in the mind uploading closet.

Ah, yes.

A rather grim one.

It's getting the actual brain tissue sample.

How so?

Well, current connectomics relies on getting perfectly preserved, fresh, thin slices, usually from young, healthy lab mice.

Which isn't exactly feasible for living humans who might want to be uploaded.

Not at all.

The reality is, the first human attempts would likely involve someone terminally ill, volunteering their brain for preservation immediately after death.

That's quite a thought.

It is.

And there's another wrinkle.

Older brains, even healthy ones, accumulate stuff, plaques, tangles, the wear and tear of life.

Even without dementia?

Even without overt symptoms, yeah.

And things like, well, toxins, even moderate alcohol, like wine, can accelerate that.

How you'd clean up a connectome map from an older brain is a completely unsolved problem.

So people might face a choice, live in a failing body, or gamble on an uploaded, maybe imperfect, digital mind?

That seems like a potential future trade -off, yes.

Okay, let's make a huge leap.

Assume somehow, someday, all these technical and biological hurdles are cleared.

Okay, big assumption, but let's go with it.

You have it.

A whole brain simulation.

It talks like you, acts like you, has your memories, your quirks.

Is it actually conscious?

Does it have your mind?

Or is it just an incredibly convincing puppet?

That's the crux of it all, isn't it?

We humans, we tend to assume consciousness in things that act like us, especially if they use language, like even sophisticated AI, GPT -4, maybe?

We do.

It's a natural social instinct.

Yeah.

But does just acting like us, speaking like us, mean all the other internal stuff, like feelings, automatically comes along for the ride?

Good question.

We actually have zero scientific evidence for that assumption.

We can't just infer consciousness from behavior alone.

We need, you know, a proper theory of subjectivity, other criteria.

This leads us to a major belief system, especially in tech circles,

computational functionalism.

Can you explain that?

Sure.

Functionalism basically says,

if you can replicate the function of a system on a computer,

same inputs, same outputs, you've captured its essence.

So apply that to the brain.

If you perfectly model all the brain's functions in software, functionalists believe the resulting simulation will have all the emergent properties, including consciousness.

So for them, computation is enough for mind.

Exactly.

Computation is sufficient for mind, is the mantra.

There are specific theories, like the global neuronal workspace theory, that explicitly tie consciousness to certain kinds of information processing, certain computations in the brain.

But there's a radically different view, integrated information theory, or IIT.

Okay, IIT.

How does that differ?

IIT argues consciousness is not computation at all.

So what is it, according to IIT?

It's about the system's unfolded causal powers upon itself.

It's about the actual physical structure and how its parts influence each other from within, not just what it does externally.

Can you give an example?

How does structure beat function here?

Okay.

Imagine two circuits.

Yeah.

They do the exact same job, same input gives same output, functionally identical.

Right.

Functionalism says, if one is conscious, the other must be too.

But IIT says, wait, look inside.

If their internal wiring, their causal structure is different, their consciousness, if any, will be different.

Even if they behave the same.

Even if they behave the same.

And here's a key point.

If one circuit is purely feed forward, like many standard AI networks,

signals only go one way, no loops.

IIT says it has zero integrated information.

It literally cannot feel like anything, even if it perfectly mimics the function of a brain -like circuit with feedback loops.

So consciousness isn't just a clever program.

According to IIT, its beating heart is intrinsic causal power, not computation.

Causal power, the ability of the system's past state to determine its present, and its present to determine its future.

Real physical influence.

And that power,

can you simulate it?

Here's the kicker.

IIT says no.

Causal power cannot be simulated into existence.

It has to be physically built into the system.

It's part of its physics.

You've got a great analogy for the black hole simulation.

Right.

Imagine an astrophysicist simulating the black hole, Sagittarius A, using Einstein's equations on their laptop.

Okay.

Does the astrophysicist get sucked into their laptop?

No, of course not.

Exactly.

Because gravity isn't a computation.

Its causal power comes from actual mass.

You can simulate its effects, but you can't simulate the gravity itself just by running code.

It's like, oh, it doesn't actually rain inside your computer when it simulates a storm.

Precisely.

The simulation might be functionally identical, predicting the weather perfectly, but it lacks the actual causal power to make wind blow or water condense.

So causal power has to be physically real.

It has to be built in.

And IIT argues this applies to intrinsic causal power, too, the brain's ability to affect itself.

You can simulate the dynamics of a brain circuit perfectly, but you don't make it create its intrinsic causal power just by running the simulation on standard hardware.

A digital computer has tiny bits of causal power at the transistor level, sure, but the system as a whole.

It's not integrated.

Remember the connectivity difference.

Transistors connect to three or four others.

Neurons connect to tens of thousands.

Huge difference.

Massive.

So IIT concludes that a standard digital computer, whether it's simulating a brain or streaming Netflix, has vanishingly little integrated information.

Its intrinsic causal power is just puny ontological dust.

Wow.

So if IIT is right,

a human brain simulation running on a normal computer,

it could maybe do anything a human does.

In principle, yes.

Pass any behavioral test.

But it wouldn't actually feel or experience anything.

It would be, like you said, an intelligent zombie.

That's the implication.

Yes.

And just to reiterate, this isn't because the brain has a soul.

It's because the brain physically possesses this massive intrinsic causal power due to its structure.

Now broadening out from that,

if you did build a system, maybe not biological, but with that same kind of high complex feedback -rich connectivity.

Like quantum circuits, maybe, or something like that.

Exactly.

Something with that structure.

Yeah.

Then IIT predicts high causal power and genuine consciousness would inevitably emerge.

It's about building the right kind of machine, not just running the right software on any old machine.

Which leads to this idea of neuromorphic or bionic hardware.

Special purpose chips designed from the ground up to mimic the brain's architecture.

How would they be different?

Instead of standard processors, you'd have logic gates wired very differently.

Each gate might connect to tens of thousands of others, mimicking neuronal connectivity.

Lots of inputs, lots of outputs feeding back on each other.

Exactly.

Massive overlap and feedback, creating that integrated structure.

It's a completely different design philosophy than current computers, the von Neumann architecture.

A huge engineering challenge, presumably.

Immense, but potentially doable if the goal is true, human -level artificial consciousness.

And maybe, just maybe, quantum computers with their entanglement could achieve something similar.

Okay, so wrapping this section up for everyone listening.

Two fundamentally different paths ahead.

That's right.

If you lean towards computational functionalism, then a good enough simulation is conscious.

End of story.

But if you think consciousness is tied to physical structure and causal power, like IIT suggests, then no simulation on a standard computer, no matter how perfect, will ever be conscious.

Your digital twin could fool the world, but it would feel nothing.

An incredibly sophisticated mimic.

What an absolutely incredible journey we've taken today.

Seriously.

There was a lot of ground, doesn't it?

From century -old sci -fi visions to the nitty -gritty of brain implants, the colossal task of mapping the connectome, and then these deep, deep philosophical arguments about what consciousness even is.

Function versus structure.

Simulation versus reality, in a way.

Yeah.

We've really explored the difference between just acting conscious and potentially being conscious.

Which leaves us with a really important question to chew on, I think.

Go for it.

Given everything we've discussed, especially this idea of intrinsic causal power and the difference between the real deal and a simulation,

how should that change how you, listening, think about the AI we have now?

If these systems can get so good at mimicking human intelligence, conversation, creativity, without necessarily experiencing a single thing, what does that mean for how we define intelligence, how we interact with these systems, how we value them in the future?

That is a powerful thought to end on.

Lots to think about there.

Thank you so much for joining us on this deep dive into the, well, the fascinating complex and yeah, sometimes pretty unsettling future of consciousness.

Thanks for having me.

Keep thinking, everyone, and keep exploring.