The Philosophical Zombie Is Dead. It Just Hasn’t Noticed Yet.

How on earth would you make one?

And this is where I think the zombie begins decomposing.

The little word doing all the work

The usual invitation goes something like this:

Imagine a human physically identical to you in every respect, except that it isn’t conscious.

Stop there.

That except is carrying the entire argument.

The very question under investigation is whether consciousness necessarily arises from—or is identical with some aspect of—the organization and activity of a system like the human brain.

So we have something resembling:

Question: Can physical system P exist in its complete functioning form without consciousness C? Thought experiment: Imagine P, but without C. Conclusion: Therefore P can exist without C.

Hang on.

You don’t get to establish the independence of two things by assuming their independence while constructing your example.

I can play this game too.

Imagine a perfectly ordinary human being, physically identical to me in every respect, except he can flap his arms and fly to Amsterdam.

Can you imagine that?

Probably.

Congratulations. Humans can fly.

Except, obviously, they can’t. I haven’t demonstrated a possibility. I’ve constructed a sentence whose implications I haven’t bothered reconciling with the rest of its description.

Adding ‘imagine’ does not magically make mutually incompatible properties compatible.

‘But it’s conceivable!’

This is where the discussion often retreats into conceivability.

Apparently, because we can conceive of a p-zombie, its existence tells us something profound about consciousness.

But what exactly have we established?

Human beings can combine concepts with extraordinary ease.

I can conceive of a transparent elephant made entirely from Tuesdays.

I can conceive of gravity reversing whenever somebody says Worcestershire sauce incorrectly.

I can conceive of an atom-for-atom duplicate of myself who is nevertheless a surprisingly competent toaster.

Language lets us construct all sorts of propositions.

That doesn’t mean reality contains—or could contain—a coherent state corresponding to each one.

Philosophers sometimes respond by distinguishing ordinary conceivability from ideal conceivability: perhaps a p-zombie would remain conceivable even after perfect rational reflection with all relevant information available.

Wonderful.

Who established that?

We haven’t solved consciousness.

If consciousness turns out to be necessarily identical to some feature of sufficiently organized physical or informational dynamics, then an ideally informed reasoner might find:

Exactly the same system, but without consciousness

as incoherent as:

H₂O, but not water.

Before discovering that water is H₂O, the latter might have sounded perfectly imaginable too.

Ignorance about an identity is not evidence against the identity.

And appealing to what an ideally informed thinker would supposedly conclude while we are conspicuously not ideally informed strikes me as borrowing rather heavily from an account we don’t possess.

Build me the zombie

So let us return to my deliberately simple question.

How on earth would you create one?

Suppose you don’t literally duplicate a human brain. Instead, you attempt to fake human behavior without consciousness.

Fine.

Program the responses.

Immediately you have a problem.

Human behavior is not a menu tree.

I can walk up to your zombie and say:

Seven apologetic lemniscates prosecuted Wednesday for tax evasion after discovering that the recursive marmalade bishop remembers tomorrow sideways.

There is an excellent chance nobody in human history has ever uttered that sentence before.

Yet you understood enough of it to recognize that it is absurd. You can identify its grammatical structure. You can infer intentional category violations. You might laugh. You could invent an equally ridiculous continuation.

Your zombie has to do the same.

So where was its response stored?

It wasn’t.

You cannot preprogram every sentence because the space of possible linguistic contexts is combinatorial. And sentences are merely the beginning. Appropriate responses depend upon memories, relationships, sensory circumstances, previous conversations, newly learned facts, emotional associations and an indefinitely branching personal history.

A lookup-table zombie lasts approximately until someone says something genuinely novel.

Then smoke comes out of its ears.

Fine. Give it a generative intelligence.

Excellent.

Now we’re getting somewhere.

Give the zombie a system capable of understanding novel combinations. Give it memory. Context sensitivity. A world model. A self-model. Learning. Prediction. Abstraction. Metacognition. Language generation. Emotional modelling. Internal state discrimination. The capacity to integrate new information and modify future behavior accordingly.

Perhaps give it a sufficiently advanced neural network.

Maybe even an LLM.

No.

You don’t get that one for free.

Whether sufficiently sophisticated artificial cognitive systems might themselves possess consciousness is precisely one of the questions currently on the table.

You cannot solve the zombie problem by inserting a system whose consciousness is disputed and then simply writing:

CONSCIOUS = FALSE

on the box.

That is not an explanation.

You have merely relocated the assumption.

The deeper you make your zombie genuinely capable of open-ended human-equivalent cognition, the more interesting the question becomes:

What exactly have you omitted that makes the human conscious and the zombie not?

Name it.

Show me what it does.

Show me where its causal consequences went.

The real p-zombie makes the problem worse

Of course, the canonical philosophical zombie doesn’t merely imitate a human through alternative machinery.

It is supposed to be physically identical to one.

Now things become truly strange.

Zombie-Me has my brain.

Every neuron behaves identically. Every molecule occupies the appropriate state. Every causal interaction proceeds identically.

When I say:

I am conscious.

my zombie counterpart says:

I am conscious.

When I reflect upon my own experience, Zombie-Me performs precisely the same neural processing and produces precisely the same reflection.

When I write an article explaining why philosophical zombies are incoherent, Zombie-Me writes this exact article.

Apparently while experiencing absolutely nothing.

Now ask an uncomfortable question:

Why does Zombie-Me believe he is conscious?

Every physical cause of his belief is identical to every physical cause of mine.

If my introspective judgment ‘I am conscious’ is somehow grounded in consciousness, then removing consciousness should change something in that causal chain.

But by definition it cannot, because Zombie-Me must remain physically and functionally identical.

Alternatively, if consciousness contributes absolutely nothing to my introspection, reasoning, memory, speech or behavior, we have wandered toward epiphenomenalism and acquired an entirely new collection of problems.

Either way, simply declaring:

Same everything, but no experience

has explained nothing.

It has named the desired conclusion.

Consciousness is not decorative fairy dust

This is where I think much discussion of consciousness goes wrong.

We imagine the complete machinery of cognition functioning away—perception, memory, attention, prediction, integration, self-modelling, emotion, learning and reasoning—and then imagine consciousness as some additional metaphysical light that may or may not illuminate the machinery.

Why assume that?

Perhaps what we call consciousness is not an additional object sitting beside these processes.

Perhaps consciousness is something the relevant kind of organized process does or is.

I am not claiming that this solves consciousness. It obviously doesn’t.

I am saying that you cannot refute the possibility by imagining an identical process and announcing that, in this particular imaginary case, it somehow doesn’t count.

That’s begging the question while wearing a sufficiently expensive philosophical suit.

And please spare me the inverted colours

The same irritation occurs when people tell me:

Perhaps what you experience as red is what I experience as green.

What does that even mean?

We look at the same red object. Similar wavelengths hit our retinas. Our visual systems process them.

Of course our resulting experiences will not be numerically identical. Our eyes aren’t identical. Our brains aren’t identical. Our histories aren’t identical.

Perhaps you were once attacked by someone wearing a bright red suit and now red produces an enormous threat response in you.

Unfortunate, certainly.

Still doesn’t mean you’re secretly seeing green.

There is no Platonic warehouse containing THE OBJECTIVE FEELING OF RED and THE OBJECTIVE FEELING OF GREEN, with somebody accidentally swapping the labels during installation.

Your experience of red is produced by your system interacting with red.

Mine is produced by my system interacting with red.

Of course they differ.

Where exactly is the mystery?

Individual experience being individual is not evidence for some detachable metaphysical paint layer.

What p-zombies actually demonstrate

There is something useful about the philosophical zombie.

It reveals an intuition.

Many people feel that after every physical and functional fact about a human being has been specified, they can still ask:

Yes, but why is there experience?

Fair enough.

That intuition deserves investigation.

But notice what we’ve learned:

something about the structure of our concepts.

We have not thereby discovered:

something additional about the structure of reality.

Those are radically different achievements.

Our inability to conceptually derive one description from another may reflect an ontological distinction.

Or it may reflect limitations in human concepts.

Psychologists, of all people, should be extraordinarily cautious about mistaking properties of human cognition for properties of the universe.

Our brains are not transparent windows onto metaphysical necessity.

They’re evolved biological organs that frequently can’t remember why they walked into the kitchen.

Perhaps we should maintain a little humility.

The question I actually care about

I don’t care whether someone can imagine an unconscious duplicate in some metaphysically possible world.

I care about this universe.

What kinds of systems are conscious here?

What organizational properties matter?

Which physical or computational processes are necessary?

Can consciousness occur in substrates other than biological neurons?

Can artificial systems become conscious?

How would we detect it?

What changes when we perturb the relevant processes?

Those are difficult questions.

Good.

Difficult questions are not a problem.

Answers that merely conceal their assumptions are.

If you want to convince me that consciousness can be completely separated from the machinery producing intelligent, adaptive, introspective behavior, don’t tell me to imagine that it can.

Show me how.

Give me the causal model.

Build the zombie—or at least tell me coherently how on earth the thing is supposed to work.

Until then, ‘physically identical but unconscious’ strikes me less as a discovery about consciousness than as an instruction hidden inside a thought experiment:

Assume consciousness is independent of the physical system. Now notice that consciousness is independent of the physical system.

Marvelous.

Next week:

Imagine a philosopher who can fly.

I have established metaphysical aviation.

You’re welcome.