Chapter 11: Cognitive Systems and Representing-in-the-World

Welcome to Last Minute Lecture.

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

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

For complete coverage, always consult the official text.

Welcome to The Deep Dive, where we extract the essential, game -changing insights for the latest cutting -edge research.

Today, we're wading into a really fascinating, and let's be honest, an often intense debate.

It's about the current identity crisis in cognitive science.

That's absolutely right.

The field is, you know, actively searching for its proper paradigm, and we've taken a deep dive into a critical commentary that really assesses where we stand right now.

In this paper, it zeroes in on the challenges from what's called the

Right, the idea that cognition is embodied, embedded, inactive, and extended.

So our mission today is pretty clear.

We're summarizing this comprehensive critique that looks at four specific proposals that are floating around, things from predictive coding to, you know, the extended mind thesis.

And we're asking the core question, really.

Are these new models leading us toward a unified, stable science?

Or are some of the more radical claims, actually,

undermining the very foundations of what we call cognitive science?

We'll be tracking their arguments about representation, dynamics, and I think maybe the most controversial part,

where the mind actually ends.

To really feel the weight of this debate, you have to look at the history.

I mean, cognitive science is a young field.

Right.

And like any young field, it's still grappling with these foundational questions.

It kind of reminds me of Kant's famous claim about metaphysics.

Oh, that it hadn't found the secure path of a science yet.

Exactly.

We're still looking for that secure path for studying the mind.

Okay.

So let's unpack the path it did follow.

Let's start with a classical approach.

This was the first paradigm.

It defined everything for decades.

What were its two sort of central pillars?

Okay.

So first, and this is the big one, was representation.

The basic idea was that cognitive states, so your beliefs, your desires, your perceptions, they have to stand for something.

They reflect parts of the world.

Their internal structures, little symbols in your head.

Pretty much typically housed in the brain, abstracting information about the world outside.

And second, and this is crucial, was computation.

Cognition was seen as the formal, rule -governed manipulation of those internal representations.

The brain was the symbolic processor,

a computer.

The mind is software running on the brain's hardware.

That was the dominant metaphor.

It was.

But, you know, that view started showing some cracks, which led to the second paradigm.

And that was the connectionist approach.

Right.

It challenged that idea of neat symbolic manipulation.

It suggested a whole different kind of architecture things, like neural networks, distributed processing.

So it changed how the brain computed.

Exactly.

But it often kept the idea of representations.

They just existed in a different format, you know, as distributed weights and nodes in a network, not as clean little symbols.

And now we are smack in the middle of the third paradigm, the 4U movement.

And this is where the challenge becomes, well, almost existential.

Because it's questioning not just how the mind works, but what it is, and maybe more importantly, where it is.

Exactly.

And right away within the 4E camp, you see this massive critical split.

On one side, you have radical inactivism.

So people like Hutto, Mayen, Camero.

Yes.

They take the 4Es very seriously, but they use them to fundamentally reject the idea of representations.

For everything.

Well, for basic mental stuff, so simple action, perception.

They argue you can explain all of that just by the dynamic coupling between an organism and its environment.

You don't need any internal mental representations at all.

They're just throwing out the whole playbook.

They really are.

And then you have the middle ground,

the, let's call it the more politically palatable option for the traditionalists.

That's moderate inactivism.

Yes.

Figures like Andy Clark or Michael Wheeler.

They try to integrate the core truths of 4E, because you can't deny the importance of the body and the environment.

But they do it while retaining the notion of mental representation.

So for them, it's an enrichment, an expansion of the classical view, not a total replacement.

Precisely.

What's so interesting is that the 4Es themselves aren't all equally controversial.

I mean, nobody seriously denies that cognition is embodied, right?

It needs a brain and a body.

Of course not.

But the two claims where you see the real fireworks are, one, radical anti -representationalism, and two, the idea that cognition is truly extended, that your cognitive system can literally include parts of the world.

That's where things get really wild.

And the rise of these ideas, plus the hugely influential predictive processing framework, has just fundamentally blurred lines that used to be, well, they used to be axioms in classical cognitive science.

Okay, let's track those blurred distinctions because they really set the stage for the four papers we're looking at.

First, the line between perception and cognition.

That's getting blurry.

Totally.

In the classical view, perception was just sensory input.

Cognition was what happened after you perceived.

But in predictive processing, which we'll get into, sensing and inferring are basically the same thing.

So you don't perceive and then think about it?

No.

The act of perceiving is an act of thinking.

It's a generative inference about the world.

Okay, second blurred line between intentional action and non -intentional behavior.

This is a big one for radical inactivism.

They look at a simple coordinated movement, like a baby reaching for a toy,

and they want to explain it as purely dynamic coupling, just a mechanical interaction.

Which removes the need to talk about an internal intention, a desire or a goal.

Exactly.

The claim is if it's a basic action, it's just behavior.

It doesn't need all that complex mental machinery behind it.

Third, and this one feels like it's underneath almost the entire debate,

the line between causation and constitution.

Oh, this is absolutely vital.

This is the heart and soul of the extended mind debate.

Right.

Does the environment just cause a cognitive outcome?

Like, does my notepad cause me to remember something?

Or does the notepad constitute, does it form a part of the cognitive process of remembering itself?

And if you don't distinguish those, you fall into what's called the coupling constitution fallacy.

Which we will definitely have to come back to.

It's a huge point.

And finally, the line between the cognitive system and the environment itself, which the extended mind thesis just wants to erase completely.

Precisely.

So, to guide you through this critical commentary, we're going to evaluate four specific proposals from Howie, Froze, Mennery, and Lam and Kemero by organizing our discussion around the four key tensions that come out of these blurred lines.

So we've got predictive processing versus Fourier, the role of phenomenology, the viability of non -representational dynamics, and finally, the defense of where you draw the system boundaries.

That's the plan.

That sets the table perfectly.

Let's jump right into that first tension, predictive processing, or PP, and 4E cognition.

And we're looking specifically at Howie's version, which is very rigorous and very brain -bound.

It's called prediction error minimization, or PEM.

And you're right, this framework is just dominating neuroscience right now.

Why is it so influential?

Because it offers this unifying principle for basically all brain function, from your most basic reflexes right up to conscious experience.

The central idea is incredibly powerful.

The brain is a prediction machine.

It's not passive.

It's not just taking in information.

Not at all.

It's constantly, actively generating expectations about the world.

Okay, so how does this prediction machine actually work?

What are the nuts and bolts?

So the brain maintains an internal statistical structure.

They call it a generative model of how the world works, what causes its sensory inputs.

And using this model, it's constantly sending out top -down signals, basically hypotheses,

about what sensory input it expects to get next.

And then it compares that prediction to what actually comes in from the senses.

Right.

And if there's a mismatch, the sensory input is unexpected, that generates what they call surprise.

That's the prediction error.

That's the prediction error.

And the brain's single overarching goal in everything it does is to minimize this error.

How does it do that?

Two ways.

It can update its internal model, its hypothesis, to better match the world.

We call that perception or learning.

Or it can change the world itself to match its prediction.

And we call that action.

So perception, action, even attention.

It's all just different strategies for reducing surprise, for making the world less uncertain.

Can you give us a simple everyday example of this?

Absolutely.

Think about walking through your own house late at night in the dark.

You have a generative model of your house.

You expect the floor to creak in a certain spot.

You expect the light switch to be on the left wall.

If you reach for that switch and your hand hits bare wall, it's not where you expected it to be.

You instantly feel that jolt of prediction error.

Right.

That little moment of confusion.

And your brain has to do something.

It can either, one, update its model, that's perception.

Oh great, I remember we moved the furniture.

Or two, it changes its action.

Okay, I'll move my hand a few inches to the right to find the switch I expected to be there.

Every single second, your brain is doing this, trying to make reality match its best guess.

Which brings us to the big clash with 4E.

Howie is adamant that because of this generative model,

PEM is fundamentally representational and inferential.

Exactly.

The brain is relying on an internal model representing the outside world to make all these predictions.

So perception isn't direct.

It's not raw sensory input shaping our experience.

Not according to Howie.

The sensory input is just, as he puts it, feedback to the queries issued by the brain.

The generative model is the star of the show.

So he argues PP is only compatible with moderate 4E approaches.

The ones that keep representation.

And the radical and activist would say this is just old wine and new bottles.

Repackage classical science.

They would.

But the implications of Howie's view are, well, they're truly mind -bending.

The critical note quotes him saying that what we experience is not the world itself.

But the predictions of the currently best hypothesis about the world.

Which leads to that famous slogan from Chris Frith.

Our perception of the world is a fantasy that coincides with reality.

Wow.

I mean, think about that for a second.

You aren't experiencing the world directly.

You are experiencing your brain's best, most probable guess about the world.

A guess that's just being constantly updated by error signals.

Brain is effectively trapped inside the skull running a simulation.

That's the view.

And it flies directly in the face of the radical 4E claim that perception is direct.

Right.

This is the radical opposition.

They're often inspired by Gibson's ecological approach.

And they say these internal models are completely unnecessary.

They do.

They rely on the idea of affordances.

Can you explain that?

An affordance is a possibility for action.

You see a chair and it affords sitting.

A handle affords grasping.

Their claim is that we perceive these affordances directly without any intermediate inference or representation.

The explanatory work is all done by the dynamic coupling between the organism and its environment.

So my brain doesn't need to build a representation of sitability for the chair.

The sitability is just an objective feature of the world relative to my body and my system just picks up on it.

That's the core of the anti -representationalist challenge.

But the critique here really highlights how PP responds to this.

How he argues that perception has to be indirect and inferential because the brain needs to be flexible.

And the generative model is what provides that flexibility.

Yes.

As Daniel Dennett put it, the generative model is a method for generating affordances galore.

It's an inferential representational analysis of what's possible.

So why is that better?

This is the flexibility challenge that the conservative views are always pushing.

Well, radical 4E tends to rely on fixed, quick and dirty processing in the organism environment coupling.

And that works great for simple reflexes or basic navigation.

But human behavior is incredibly complex and flexible.

We can switch goals on a dime.

We can consciously override our habits.

Let's use that all -you -can -eat buffet analogy from the paper to really hammer this home.

Perfect.

So imagine you walk into the buffet.

Your coupled system, your eyes, your nose, your stomach.

It's immediately set up to perceive all the gustatory affordances.

Eat this, grab that, taste this.

Right.

That's the quick dynamic coupling.

But you can pause.

You can stand there, look at the dazim table, and start contemplating your doctor's recent warnings about your cholesterol.

That pause, that high -level deliberation about future consequences, that requires a representational generative model.

A model of your diet plans, your long -term health, your caloric goals.

That model is what we call representation hungry.

And that's what allows you to overcome the immediate coupled affordance of just grabbing the cake.

Exactly.

The radical view might explain why you automatically reach for the doughnut, but it really struggles to explain why you might hesitate, put it down, and pick up a carrot instead.

That deliberate choice, that overriding of a default behavior, seems to demand a higher -level, goal -directed representation of some future state.

So it's fundamentally a goal -directed representational mechanism, which is why Howie's camp is seen as moderate and conservative.

That's right.

For him, action is active inference.

It's not just coupling.

It's a goal -directed process.

That's a really compelling defense of internal representation.

So let's shift gears completely now and move to our second major tension.

Phenomenology and cognitive mechanisms.

If Howie's PP tries to win by being a better science of the brain,

Froese takes a totally different approach.

He wants to use the experience of the mind to dictate how the science should be done.

And he focuses specifically on social understanding.

Froese starts from what he calls the phenomenological starting point.

He asks, how do we experience social interaction?

So not what the brain is doing, but what it feels like.

Exactly.

And he focuses on what he calls genuine intersubjectivity.

He argues that when you're interacting with someone, that other person isn't just an object you observe.

They play a constitutive role in the social phenomenon itself.

So for dancing or having a great conversation, it feels like a single shared unified thing.

And this leads him to his big demand,

a demand for isomorphism.

Isomorphism.

Okay, what does that mean in this context?

It means demanding a strict structural similarity.

Froese insists that because the phenomenal experience, what it feels like is intersubjective and unified, then the sub -personal mechanisms.

Well, the neural stuff, the bodily stuff, the environmental stuff.

All of that must equally reflect this structure of mutual incorporation.

The mechanism has to look like the experience.

That's a fascinating and very strong constraint.

He's saying the scientific model has to be shaped like the subjective experience.

He is.

But the critical note in the paper rejects this pretty strongly.

Why?

Why shouldn't our science be constrained by how things appear to us?

Well, this really gets to the heart of what science is supposed to do.

The critic argues there's really no historical or philosophical reason to expect this kind of strict structural match between the first -person view and the underlying mechanism.

Science often reveals things that are totally counterintuitive.

Totally.

And the mechanism is often structurally very different from our subjective experience of it.

Can you give us a concrete example of that?

Let's take the unity of conscious experience.

Okay, so right now you are experiencing consciousness as a rich, unified, single stream of thought and sensation.

Right, it all feels like one thing happening to one me.

Exactly.

Now, if Froese's isomorphism demand were right, neuroscientists would be forced to go looking for a single physical place in the brain.

What philosophers sometimes mockingly call a Cartesian theater, where all the information comes together into one unified hub.

And what do they actually find?

They find distributed integrative processes.

Things like 40 hertz oscillations networks of neurons firing in coordinated patterns across huge stretches of the brain.

The structural feature of the mechanism is this distributed oscillation, which in no way matches the unified structure of the experience itself.

And nobody rejects the science of 40 hertz oscillations just because it doesn't feel like an oscillation.

Of course not.

The phenomenal structure just doesn't constrain the mechanical structure in that way.

That makes a lot of sense.

The second counter example they use is self -awareness.

Right.

We all experience ourselves as a continuous, stable, enduring subject.

D 'Amazio calls it a haven of stability, and this gives rise to what the paper calls Cartesian gravity.

The powerful, seductive intuition that the self must be a single, separate thing inside us.

Yes.

And if we demanded isomorphism, we'd have to go looking for the physical equivalent of a soul or some singular self -module in the brain.

But the reality that science is uncovering might be much messier.

Maybe it's just a bunch of parallel processes that only give rise to the illusion of a single self.

Exactly.

As Dennett predicted, the true theory of the mind is often going to be counter -intuitive at first.

So the commentary concludes that the features of our conscious experience just don't put these kinds of structural constraints on our scientific theories.

To demand isomorphism is to risk holding science hostage to introspection.

Okay, let's go back to the social context and Froese's claim of mutual incorporation.

He says that when two agents interact, they become parts of a dynamic senseware motor system.

The critique suggests this is often based on the coupling constitution fallacy.

It is.

The problem is that leaped from recognizing a very strong causal dependency to claiming a constitutional one.

So just because two things are tightly linked causally doesn't mean they've become one single thing.

Right.

If your expression causes a reaction in me, which in turn causes a change in you, that's a tight causal loop, and Froese even describes it that way sometimes.

But a causal dependency, no matter how tight, doesn't mean we've literally fused into one system.

If my friend tells a joke that makes me laugh, their joke caused my laughter.

But the joke isn't a constitutive part of the physiological process of my laughing.

Precisely.

And what's more, these emotional exchanges rarely result in perfect incorporation anyway.

If you express profound sadness, I usually feel sympathy or empathy, not identical sadness.

Our experiences remain distinct.

Exactly.

We are interacting as separate subjects, and traditional accounts can handle those complex causal dynamics just fine without having to make these big constitutional claims.

Froese also levels a solipsism charge against traditional theories, saying they can't ever really guarantee the other person exists because they're all about what's happening in one person's head.

The critic finds this a very strange charge, especially against someone like Dennett, whose whole theory begins by assuming the other agent exists.

But what's more critical is that Froese defines the second -person perspective as

two first -person perspectives temporarily becoming integrated.

But wait, if it's the integration of two first -person views, then the first -person perspective is still the primary thing.

It is.

So if solipsism was a problem for the classical accounts, his view doesn't solve it, it just renames the starting point.

It's kind of a self -defeating charge.

This all leads to what the critique says is the most damaging flaw for radical inactivism in the social domain,

the precondition problem.

Can you break this down?

Sure.

So radical inactivism insists that genuine social interaction is this coordinated coupling that takes on a life of its own.

But for that required coordination to even begin, the agents have to first achieve some minimal cognitive state on their own.

Okay.

Let's use the shy student thought experiment from the paper to make this concrete.

Perfect.

So imagine a shy student at a bar.

He spots someone across the room he might want to talk to.

Now, to initiate the kind of coupling or coordinated interaction that Froze is talking about, that student has to first make an initial

individual cognitive assessment.

He has to think about it first.

Yes.

He has to presuppose internally that the person across the room is a potential candidate for interaction.

More than that, he has to assume she's a second -order intentional system.

Meaning she's capable of having beliefs about his beliefs, like she might believe he's about to approach her.

Absolutely.

That initial moment of social cognition, that little bit of minimal mind reading has to happen before any coupled interaction can be established.

So if radical inactivism rejects all forms of representation,

how does it explain that initial individual cognitive step that gets the whole thing started?

It can't really.

The theory seems to presuppose the very representational machinery it wants to get rid of.

It feels like a massive hole in the argument.

It's a huge problem.

And the traditional view, including PP, handles this much better because it acknowledges that interaction is just a complex causal loop between two individual representational systems.

That strong rejection of phenomenological constraints brings us to our final and maybe the most philosophical tension of all,

cognitive systems.

Where does cognition end?

This is the ultimate question of the extended mind debate.

It is.

The classical view just drew the line at the skull and skin.

And the 4E movement says that's arbitrary.

It ignores how tools, technology, and our environment change how we think.

So we have to figure out where the boundary should really be.

Let's start with the most extreme version, the dynamical systems approach from Lam and Chemerow.

How do they define the cognitive system?

Well, they reject fixed boundaries completely.

They argue the system includes all and only those elements that depend, to a specified extent, upon one another.

But here's the catch.

The system is determined relative to a given phenomenon of interest.

So the boundary changes depending on what a researcher is studying.

Exactly.

If you're studying someone reaching for a coffee mug, the system might be the arm, the eye, the mug.

If you're studying them solving a math problem with a calculator, the system now includes the calculator.

That leads to what they call a messy picture of the boundaries.

Openly, yes.

But if the boundaries depend on the aims and interests of cognitive researchers, that starts to sound dangerously subjective.

It sounds just as arbitrary as the skull and skin view they're criticizing, maybe even more so.

That's precisely the critique.

The traditional view might be arbitrary, but at least it's a fixed target for research.

This dynamically shifting boundary risks being even more arbitrary because it's variable.

It depends on an external researcher's goals, not on any internal logic of the system itself.

And this creates a huge practical problem for anyone trying to study development over time.

This is Rupert's worry.

Yes, or the diachronic identity problem.

Diachronic identity is just a fancy way of saying the problem of making sure you're studying the same thing over time.

If the system's boundaries are constantly shifting, sometimes it includes a notebook, sometimes a colleague, sometimes nothing, how can a developmental psychologist study how the same child acquires language or how the same adult's memory declines with age?

You need a stable, persistent core system to track change.

You absolutely do.

The organism, the animal, provides that core identity.

The radical dynamical systems approach just doesn't.

So if Lamm and Kimmer's approach is to go bigger and risk arbitrariness, Froese's autopoietic approach is to go smaller and draw a biological boundary.

Right.

Following the tradition of Maturana and Varela, Froese suggests we should identify cognitive systems with fundamental features of life itself,

specifically metabolism.

And this leads to the very strong claim, where there is life, there is mind.

Yes, which means single -celled organisms, bacteria, they all possess some basic mental capacities.

So wait, is the bacteria in my yogurt cognitive?

If you follow the logic all the way, yeah, they have a little bit of mind.

But the critique here questions the scope and the utility of this view.

If the cognitive and the biological are the same thing, if a single cell is cognitive, then cognitive science just collapses into general biology.

The word cognition loses its special meaning.

It stops being useful for explaining complex human and animal behavior.

Exactly.

And there's a fascinating ironic tension here because this autopoietic view is often motivated by something called the free energy principle,

or FEP.

The idea that all living things must minimize free energy to maintain their identity.

Right.

But the irony is that Howey's predictive processing framework, which Froese opposes, also uses FEP.

But Howey assigns that task of error minimization specifically to the brain.

So if the principle is supposed to be universal, applying to bacteria and plants just to maintain their structure, why should minimizing free energy automatically count as cognition?

That's the million -dollar question.

If FEP explains both why a bacteria maintains its cell wall and why a human brain predicts sensory input, then FEP is probably too broad to be a useful marker for the cognitive.

It's a principle of life, not a unique signature of the mind.

Precisely.

We need a lot more work to clarify that scope issue.

Okay, let's look at the moderate conservative version of extension, which is Mennery's cognitive integration.

This seems to be the one most compatible with the classical view.

It is.

Mennery is motivated by explaining how we use external things, like mathematical notations or a laptop, to do things we couldn't otherwise do.

It makes us smarter.

And he says this happens through niche construction, where we use our basic sensor motor skills to incorporate these external tools.

What's crucial here is that it presupposes an existing bounded cognitive system, the human, that starts the process of using the tool.

Yes.

It's a conservative enrichment of the classical view.

It's an internal system skillfully extending its reach.

The system's boundaries aren't fundamentally dissolving.

Now, the whole argument for the extended mind hangs on the claim that drawing the boundary at the skull is arbitrary.

The paper offers a powerful defense of that traditional boundary.

The asymmetry argument.

Right.

This argument rests on the fundamental asymmetry between the brain's role and the environment's role.

The environment and the body are crucial enabling factors.

They cause cognition to happen.

But the brain is the necessary and unique central controller.

How do we know this?

How do we prove this asymmetry?

The critique points to psychopathology.

Specifically, research on brain lesions, like the work of DeMassio.

Think about it.

If I lose my arm, I can still think, I can plan, I can calculate.

My ability to interact is reduced, but my core cognitive functions are there.

But if I suffer a specific lesion to my prefrontal cortex?

Your ability to plan, calculate, and reflect might just disappear entirely, even if your body and all your external tools are perfectly intact.

So the brain's role is asymmetrical.

Without a functioning brain, no complex cognition gets off the ground.

It's the non -substitutable central hub.

And that's also the best defense against the cognitive bloat objection.

If we start mistaking any causal factor for a constitutive part of the system, where do we stop?

The sun is causally necessary for my perception, but we can't seriously include the sun in my cognitive system.

You'd have a system so vast and ill -defined, you couldn't possibly study it.

Right.

And the extended car analogy from the paper really drives this home, distinguishing causation from constitution.

Let's hear it.

A car absolutely depends on gas, oil, and roads.

They are necessary causal factors for it to work.

We could philosophically decide to define the extended car as the metal plus the gas plus the road.

But nobody does that.

Of course not.

For practical, legal, and economic reasons, it is profoundly useful and sensible to draw a clear line demarcating the car, the self -contained machine, from its enabling environment.

We study the car's engine as a system with its own identity, even though we know it relies on external things like gas stations.

Exactly.

And the critique argues the organism, the animal, the human, is the most sensible persistent cognitive system to study.

Laptops, sunlight, friends.

They're often essential causal contributors, but that doesn't make them constitutional parts of the cognitive system itself.

That brings us to the conclusion of this really deep critical evaluation.

We've unpacked these four proposals and tested them against these huge tensions of representation, phenomenology, and system boundaries.

And the author's assessment is pretty clear.

It's decisively tilted toward integration, not revolution.

The compatible or conservative views, so Howey's predictive coding and Mennery's cognitive integration, they're seen as highly fruitful.

Because they embrace the insights of Fourier like embodiment, but they do it while holding onto the core classical mechanisms of representation, and it defines system identity.

Conversely, the radical or problematic views, Froese's anti -representational and activism, and Lamb and Jimmerow's dynamical systems, they face what look like significant, maybe even insurmountable conceptual hurdles.

So Froese's social and activism is crippled by that precondition problem.

Right.

It can't explain the individual cognitive act needed to start an interaction.

And it rests on that philosophically dubious demand for phenomenological isomorphism.

And Lamb and Jimmerow's approach fails the diachronic identity problem.

It makes the system boundaries so arbitrary and variable that you can't really do science on them over time.

That's the verdict.

So what does this all mean for the future of the field?

I mean, can radical anti -representationalism ever really account for complex things like language or planning?

If it struggles to explain a student looking across a bar.

The critique concludes that it's highly unlikely.

It probably can't scale up to account for these representation -hungry phenomena.

So while the Fourier movement was incredibly successful in highlighting the importance of the body and the environment, the core truths of it can be fully integrated into a more conservative representational account.

That seems to be the takeaway.

What is radical about the movement will likely be abandoned as the field matures.

The future paradigm looks like one of conservative integration, not radical revolution.

A truly critical analysis of this ongoing search for the mind's blueprint.

This really provides a clear shortcut to understanding the complex intellectual hurdles facing cognitive science today.

And this raises one final provocative thought for you to mull over.

If we accept the insights of the free energy principle, which applies to all life, and we see the power in the autopoietic view,

the core empirical challenge that remains is figuring out where we draw the line in the animal kingdom.

At what level of complexity does a non -representational system stop being just alive and start being genuinely cognitive?

This boundary is absolutely essential if cognitive science is going to keep a distinct, useful subject matter that's separate from general biology.

A question that defines not just the boundaries of the mind, but the boundaries of the very science that studies it.

Thank you for joining us for this deep dive.

We hope you feel well informed and ready to tackle the debate.

Until next time.