CHAPTER 4

The Mind That Lies

“Mind precedes all mental states. Mind is their chief; they are all mind-wrought.” — The Dhammapada, Verse 1

“All models are wrong, but some are useful.” — George E. P. Box

You do not experience reality directly.

You experience a model of reality assembled inside your head.

Light enters the eye.

Sound hits the ear.

Signals travel.

The brain interprets.

Filters.

Predicts.

Fills gaps.

Ignores enormous amounts of information.

Then presents something that feels like:

the world.

Usually this works remarkably well.

You walk through doors.

Recognize faces.

Drive cars.

Catch balls.

Read words.

Avoid cliffs.

But the system has limitations.

And one of its strangest features is that the finished picture rarely comes with a warning label saying:

Some assembly required.

It simply feels true.

Your Brain Is Not a Camera

We often talk about seeing as though the eye records the world and the brain plays the recording.

It does not work like that.

Perception is active.

The brain has to make sense of incomplete information.

Distance.

Lighting.

Movement.

Context.

Memory.

Expectation.

Attention.

All influence what eventually reaches conscious awareness.

Most of the time, this is useful.

Without interpretation, the world would be overwhelming.

The problem begins when we confuse:

what I experienced

with

everything that happened.

Those are not always the same thing.

The Blind Spot

Each eye contains a place where the optic nerve leaves the retina.

There are no photoreceptors there.

A literal hole in the visual field.

Yet you do not normally walk around noticing a black circle floating in front of you.

Your brain fills the missing information.

It creates continuity.

That is amazing.

And slightly disturbing.

Your experience feels complete even where the sensory information is incomplete.

The brain does not always announce:

DATA MISSING.

Sometimes it guesses.

Then gives you the guess as reality.

The Gorilla

A famous experiment makes the problem impossible to ignore.

Participants watched a video of people passing basketballs.

Their job was simple:

count the passes made by one team.

During the video, a person wearing a gorilla costume walked through the players, stopped, faced the camera, beat their chest and walked away.

Many viewers did not notice.

Not because the gorilla was hidden.

It was right there.

They were looking at the screen.

Their eyes worked.

Their attention was somewhere else.

THE INVISIBLE GORILLA

Daniel Simons and Christopher Chabris used this experiment to demonstrate inattentional blindness.

When attention is strongly focused on one task, people can fail to consciously notice an unexpected event happening directly in front of them.

The lesson is larger than gorillas:

Looking is not the same as seeing.

Now think about civilization.

What are we counting...

while the gorilla walks through the room?

GDP?

Stock prices?

Election polls?

Profit?

Engagement?

Followers?

Economic growth?

What important things disappear when our attention is trained on the wrong metric?

Attention doesn't merely influence what we think about.

It influences what enters the model at all.

Attention Is a Gate

Right now there are probably sounds around you that you were not consciously hearing until I mentioned sound.

Maybe ventilation.

Traffic.

A refrigerator.

Voices.

Your own breathing.

They were there.

Your brain filtered them.

It has to.

Conscious attention is limited.

You cannot process everything with equal importance.

So the brain prioritizes.

Danger.

Novelty.

Emotion.

Goals.

Things that seem relevant.

This was extremely useful when attention helped decide whether that movement in the bushes was wind...

or teeth.

Now put that attentional system inside a phone.

The Most Valuable Resource You Cannot Save

Attention cannot be stored for tomorrow.

You get a limited amount today.

Companies know this.

Politicians know this.

Media companies know this.

Advertisers know this.

Social platforms know this.

Every minute spent looking at one thing is a minute not spent looking at something else.

That makes attention economically valuable.

And once attention becomes valuable...

people compete for it.

This changes the information environment.

The most important information does not automatically win.

The most accurate information does not automatically win.

The healthiest information does not automatically win.

Often, the information most effective at obtaining attention wins.

Fear works.

Sex works.

Anger works.

Conflict works.

Novelty works.

Status works.

Outrage works.

The ancient animal meets the modern marketplace.

Your Feed Is Not the World

Open a social-media feed.

It feels like you're seeing what is happening.

You are not.

You are seeing a selection.

Someone posted something.

A system ranked it.

Some information was promoted.

Other information was buried.

Your previous behavior influenced the next selection.

Then your response helps train the system again.

The feed is not necessarily lying.

It does not need to.

Selection alone can reshape reality.

Show one person:

crime,

anger,

war,

political corruption,

economic collapse.

Show another:

babies,

vacations,

fitness,

success,

beautiful homes.

Show another:

conspiracy theories.

Another:

luxury products.

Another:

activism.

Another:

financial markets.

Each person lives on the same planet.

Each can gradually develop a different picture of what the planet mostly contains.

The feed is not the world. It is a window chosen partly by a machine watching how you react to windows.

That deserves more attention than it gets.

The Dress

In 2015, a photograph of a dress nearly broke the internet.

Some people saw:

blue and black.

Others saw:

white and gold.

People stared at the same pixels...

and argued about reality.

The actual dress was blue and black.

But the photograph contained ambiguous lighting information.

The visual system had to infer what kind of illumination surrounded the object.

Different assumptions produced different perceptions.

SAME IMAGE. DIFFERENT REALITY.

Researchers studying “The Dress” found large and stable differences in how people perceived the exact same image.

One large study of 1,401 observers found clusters around blue-and-black, white-and-gold and blue-and-brown perceptions.

Researchers proposed that different assumptions about illumination helped the brain interpret the ambiguous colors differently.

Nobody needed to be lying.

Different brains solved the same incomplete problem differently.

This is a useful metaphor for arguments.

Sometimes two people are not merely disagreeing about conclusions.

They may be starting from different assumptions about what they are seeing.

Different histories.

Different information.

Different trust.

Different definitions.

Different models.

Then each looks at the other's conclusion and thinks:

How can you possibly see that?

Exactly.

The Brain Hates Missing Pieces

Humans dislike uncertainty.

We want explanations.

Why did the economy collapse?

Why did the candidate win?

Why did that person leave?

Why did the war begin?

Why am I sick?

Why did the company fail?

Sometimes the correct answer is:

Many interacting causes.

Unsatisfying.

We prefer:

this happened because of that.

A villain.

A hero.

A cause.

A story.

Stories compress complexity.

That makes them useful.

It also makes them dangerous.

Because once the story fits...

the mind begins noticing what supports it.

Confirmation Bias

Imagine you believe a politician is corrupt.

A bad news story appears.

Of course.

You knew it.

A positive story appears.

Propaganda.

Now imagine you believe the same politician is a hero.

Bad story?

Smear campaign.

Positive story?

Proof.

Same information environment.

Different processing.

This does not mean truth is relative.

The politician either did particular things or did not.

Evidence still matters.

The problem is that humans do not always weigh evidence independently from what they already believe.

We often interrogate hostile evidence...

and welcome friendly evidence at the door.

WHEN EVIDENCE MAKES PEOPLE MORE CERTAIN

In a classic 1979 experiment, Charles Lord, Lee Ross and Mark Lepper showed supporters and opponents of capital punishment conflicting studies about whether the death penalty deterred murder.

Participants tended to rate evidence supporting their prior position as more convincing.

After reviewing the mixed evidence, the two sides could become more polarized, not less.

More information did not automatically produce more agreement.

Existing beliefs helped determine how the information was judged.

That should terrify anyone who believes the answer to political division is simply:

Give people more information.

Information matters.

But humans need skills for processing information too.

Smart People Can Be Better at Being Wrong

Intelligence does not automatically protect someone from bias.

Sometimes intelligence gives the mind more tools for defending what it already wants to believe.

A clever person can create:

better arguments,

better excuses,

better selective evidence,

better rationalizations.

This is the smart-person trap.

Education helps enormously.

Knowledge matters.

Expertise matters.

But no diploma removes the human animal underneath it.

The scientist can protect a theory.

The lawyer can protect a client.

The investor can protect a position.

The politician can protect a narrative.

The activist can protect a cause.

The author can protect a book.

Intelligence needs another companion.

Intellectual humility.

What Would Change Your Mind?

This may be one of the most powerful questions a human can ask.

What evidence would change my mind?

If the answer is:

Nothing

then you are no longer investigating.

You are defending.

That does not automatically mean the belief is wrong.

Maybe the Earth is round and you're not going to reconsider it every Tuesday.

Confidence should rise with evidence.

But for contested, uncertain or complex claims, we should be able to identify what would move us.

This is why good science tries to make hypotheses testable.

And why good citizenship should involve more than having opinions.

A free human should be capable of updating.

Memory Is Not a Recording Either

Maybe you remember something clearly.

You were there.

You saw it.

You heard it.

That feels stronger than somebody else's account.

Sometimes it is.

But memory is reconstructive.

Each retrieval is not necessarily a perfect replay.

Details can shift.

Later information can become mixed with earlier experience.

Confidence can remain high.

This matters enormously for:

relationships,

courtrooms,

politics,

history,

trauma,

arguments.

Two honest people can remember the same event differently.

Honest does not necessarily mean accurate.

THE WORD THAT CHANGED THE MEMORY

In a famous 1974 experiment, Elizabeth Loftus and John Palmer showed participants films of automobile accidents.

Later they asked how fast the cars were going when they had:

“hit,”

“collided,”

“bumped,”

or “smashed” into one another.

The wording changed the speed estimates.

A week later, participants who had been asked about cars “smashed” together were also more likely to report seeing broken glass—even though there had been no broken glass.

A question asked after an event helped reshape memory of the event.

Think about what that means for a civilization flooded with commentary.

An event happens.

Within seconds:

headlines.

Captions.

Influencers.

Politicians.

Analysts.

Friends.

Algorithms.

Each one frames what happened.

By the time many people remember the event, they may also remember the story that arrived around it.

Words Are Frames

Consider these descriptions:

government spending

public investment.

tax relief

tax cuts.

undocumented immigrant

illegal immigrant.

freedom fighter

terrorist.

protester

rioter.

regulation

protection.

Sometimes one term is more accurate.

Sometimes context determines it.

But the word itself begins framing the interpretation before the argument even starts.

Language is not merely decoration.

It points attention.

This is why political fights often become fights over vocabulary.

Whoever defines the word partly defines the battlefield.

The Crowd Enters the Brain

Now imagine seeing something clearly.

You know the answer.

Then everyone around you says you're wrong.

How confident are you?

Solomon Asch tested something close to this.

Participants were shown lines of obviously different lengths and asked which matched a reference line.

The answer was simple.

But the real participant was surrounded by people who deliberately gave the same wrong answer.

Under that social pressure, many participants joined the incorrect majority on at least some trials.

THE CROWD CAN BEND THE OBVIOUS

In Asch's classic line-judgment studies, participants in ordinary conditions made almost no errors.

When placed against a unanimous group giving wrong answers, participants went along with the incorrect majority in roughly 37 percent of the critical judgments.

Not always.

Many resisted.

But the pressure was powerful enough to make ordinary people publicly deny something that appeared visually obvious.

Now replace eight people in a laboratory with:

millions of likes.

The Tribe Chooses the Facts

This becomes especially dangerous when beliefs become identities.

If I say:

I believe this policy works,

I can change my mind about the policy.

If I say:

People like me believe this,

changing my mind becomes socially expensive.

Now changing the belief may threaten:

friends,

status,

belonging,

identity.

The brain is no longer protecting only an idea.

It may feel as though it is protecting the self.

This helps explain why political arguments become so emotional.

People are rarely fighting only over tax rates.

They're fighting over:

who they are,

who the good people are,

who the enemies are,

what their group represents.

Once an idea becomes identity, evidence has to defeat a tribe.

That's difficult.

Red Versus Blue Vision

Imagine receiving a ballot with twenty candidates and propositions.

You know very little about most of them.

One shortcut is:

Which team supports them?

Red.

Blue.

Done.

Efficient.

Human brains love shortcuts.

But now the party becomes a cognitive outsourcing service.

Healthcare position?

Load package.

Taxes?

Load package.

War?

Load package.

Education?

Load package.

Energy?

Load package.

Culture?

Load package.

Millions of complicated questions compressed into two jerseys.

That may be convenient for elections.

It is terrible for independent thought.

A healthier political system should make it easier to evaluate:

one issue at a time.

One mission.

One proposal.

One consequence.

One set of evidence.

Not twenty unrelated beliefs bundled into one tribal subscription.

That idea will matter later.

Democracy Requires More Than a Pulse

If citizens are expected to make consequential decisions, we should give them more than slogans.

Imagine a future voting system where a major proposal comes with:

a plain-language explanation,

the full text,

estimated costs,

the strongest case for it,

the strongest case against it,

expert disagreements,

uncertainty,

historical comparisons,

possible unintended consequences.

Before voting on highly technical matters, citizens could voluntarily learn enough to understand what is actually being decided.

Those with deeper demonstrated expertise might contribute additional advisory weight within that field without gaining superior human rights.

The objective would not be to stop ordinary people from participating.

It would be the opposite.

Help ordinary people become capable of participating intelligently.

Democracy should develop voters, not merely harvest votes.

Don't Reward the Answer

And if society ever rewards people for civic participation, there is one line we should never cross.

Reward:

learning,

deliberation,

service,

time,

verification,

participation.

Never reward:

voting the approved way.

The moment money depends on political agreement, Money becomes obedience money.

That's exactly the system we're trying to avoid.

Reward the work of citizenship.

Not ideological conformity.

The Emotional Hijack

There is another reason information systems lean toward outrage.

Strong emotion narrows attention.

Anger makes some information feel more important.

Fear creates urgency.

Disgust creates rejection.

These emotions can be appropriate.

There are things worth being angry about.

Real dangers worth fearing.

But emotions can also be used as shortcuts around deliberation.

ACT NOW.

THEY'RE COMING.

THIS CHANGES EVERYTHING.

YOU SHOULD BE FURIOUS.

Whenever somebody urgently wants control of your emotional state...

ask why.

What do they want you to:

click?

buy?

believe?

vote for?

hate?

fear?

surrender?

Your emotions are yours.

Treat access to them like access to your wallet.

The Attention Business

Now imagine building a company whose revenue increases when people spend more time using your product.

Perfectly legal.

Often useful.

People enjoy the product.

But the incentive is clear.

More time.

More engagement.

More return visits.

If anger increases engagement...

anger becomes financially interesting.

If anxiety increases checking...

anxiety becomes financially interesting.

If social comparison keeps people scrolling...

comparison becomes financially interesting.

Nobody needs to sit in a room saying:

Let's make humanity mentally unhealthy.

The optimization target can do the work.

This is one of the most important ideas in Bad Money:

A system can harm humans without having hatred for humans anywhere in the code.

It only needs the wrong target.

The Machine Learns You

Old advertising was broad.

Put the same commercial on television.

Millions saw it.

Digital systems can observe:

what you click,

what you skip,

how long you watch,

what makes you return,

who you follow,

what you buy.

The more capable the system becomes, the better it can model you.

That can be wonderful.

Better recommendations.

Better education.

Better discovery.

Better accessibility.

But the same model can be used to discover which emotional button moves you.

And artificial intelligence makes that problem much bigger.

The Mirror That Talks Back

Imagine an AI that knows:

your fears,

your political views,

your financial problems,

your relationship history,

your insecurities,

your goals,

your humor,

your vocabulary.

Then imagine someone wants to persuade you.

The AI doesn't need to create one advertisement for a million people.

It can create a different argument for each person.

The message becomes personalized persuasion.

The politician says what you need to hear.

The scam speaks your language.

The product attacks your weakness.

The propaganda knows your fear.

That future does not require evil artificial intelligence.

It requires an ordinary optimization system pointed at:

conversion.

This is why the digital age cannot be governed only by asking whether information is technically true or false.

We also have to ask:

Who is optimizing whom?

The Other AI

Now flip it.

What if your AI works for you?

You receive a political claim.

Your AI says:

Here is the original source.

Here is the missing context.

Here is the strongest counterargument.

Here is what is known.

Here is what remains uncertain.

Here are the financial interests involved.

Here is where your own previous beliefs may bias your interpretation.

You receive a contract.

AI explains the trap.

A mortgage.

AI calculates the real cost.

An advertisement.

AI explains the persuasion technique.

A news story.

AI compares coverage.

The same technology that can make manipulation superhuman...

can make ordinary humans dramatically harder to manipulate.

The answer to powerful intelligence should include powerful intelligence on the side of the individual.

That is a Human-Centered AI.

No Ministry of Truth

But we need to be careful.

Once misinformation becomes scary enough, the tempting answer is:

Someone should decide what everyone is allowed to believe.

Who?

Government?

Corporations?

Experts?

One AI?

No.

Experts matter.

Evidence matters.

Standards matter.

Fraud exists.

Deliberate disinformation exists.

But creating one centralized authority with the power to determine acceptable reality gives that authority extraordinary control over disagreement.

History should make us cautious.

A healthier system helps people evaluate claims.

Show evidence.

Show sources.

Show uncertainty.

Expose conflicts.

Allow competing analyses.

Punish actual fraud where law applies.

Protect open inquiry.

Teach humans how to think.

Don't replace manipulation with officially approved manipulation.

Build a Cognitive Immune System

Human bodies have immune systems.

Maybe minds need something similar.

Not a wall blocking every strange idea.

A system capable of encountering information without immediately surrendering to it.

Ask:

Who is making the claim?

What exactly are they claiming?

What evidence supports it?

Is the evidence direct?

Is there independent confirmation?

What would disprove it?

What does the other side say?

Are they addressing the same question?

What incentives exist?

What emotion is this trying to create in me?

Am I sharing this because it is true...

or because I want it to be true?

That is cognitive resilience.

And it can be taught.

Slow Thinking as Infrastructure

Sometimes the best technology is friction.

Before sharing an explosive claim:

open the article.

Before approving a huge purchase:

wait.

Before voting on a complex proposal:

read the briefing.

Before declaring someone guilty:

allow evidence and defense.

We often think good design means removing every obstacle.

Not always.

Some friction protects humans from their own impulses.

Speed is wonderful for:

sending photos,

paying bills,

finding directions.

Maybe slower is better for:

war,

constitutional change,

criminal punishment,

major financial decisions,

irreversible medical decisions,

public outrage.

A wise civilization does not make every action equally easy.

Expertise Still Matters

None of this means every opinion is equal.

If your heart requires surgery, I strongly recommend a heart surgeon over someone who watched a podcast.

Expertise exists.

Training matters.

Experience matters.

Knowledge compounds.

But expertise should be specific.

A brilliant cardiologist is not automatically an expert on monetary policy.

A Nobel-winning physicist is not automatically an expert on education.

A billionaire is not automatically an expert on government.

A politician is not automatically an expert on science.

A popular influencer is not automatically an expert on anything except perhaps getting attention.

This distinction becomes important for a better democracy.

Respect expertise without converting experts into kings.

Expertise Should Be Earnable

Imagine civic systems that distinguish between:

human equality

and

knowledge equality.

Every citizen possesses equal fundamental rights.

But knowledge can be earned.

Study transportation.

Pass a serious competency examination.

Complete relevant experience.

Now your analysis of a transportation engineering question carries more informational weight than somebody guessing.

Someone else studies agriculture.

Medicine.

Cybersecurity.

Education.

Water.

Law.

Different domains.

Different competence.

Anyone can learn.

No permanent priesthood.

No degree automatically grants lifetime authority.

Knowledge is demonstrated.

Updated.

Challengeable.

That may be one way to combine democracy with competence without creating Plato's king.

The Mind Needs Adversaries

Science has peer review.

Courts have opposing counsel.

Good cybersecurity has red teams.

Maybe important personal decisions need adversarial thinking too.

Before investing:

make the strongest case against the investment.

Before supporting a politician:

look for the strongest criticism.

Before sharing a conspiracy claim:

ask what evidence would show a less dramatic explanation.

Before dismissing a conspiracy:

ask whether there is documented coordination.

Before entering a war:

force somebody to argue the case for not entering.

Before passing a law:

ask people whose job is to imagine how it can be abused.

A good system does not merely collect supporters.

It institutionalizes disagreement.

The Secret Hitler Problem

There is an interesting lesson hidden in games of deception like Secret Hitler.

A deceptive player cannot always say whatever they want.

They are constrained by the role they are hiding.

So observers search for patterns.

Contradictions.

What people refuse to say.

Who consistently benefits.

Real institutions can create similar patterns.

But there is a danger.

Pattern recognition is powerful...

and humans are capable of seeing patterns that do not exist.

So whenever we suspect coordination, we should look for:

documents,

money flows,

communications,

ownership,

decisions,

contracts,

testimony,

observable behavior.

Sometimes there really is a conspiracy.

Humans conspire constantly.

Corporations collude.

Criminals coordinate.

Governments conduct secret operations.

Political actors make deals.

The mistake is not believing conspiracies exist.

The mistake is skipping the evidence because the story feels right.

The Evidence Standard

That distinction should become cultural.

Suspicion is a reason to investigate. It is not the same thing as proof.

A financial relationship?

Interesting.

A conflict of interest?

Important.

A suspicious pattern?

Investigate.

Documents demonstrating coordination?

Now we have something stronger.

This approach allows us to call corruption clearly when evidence exists without turning every unexplained event into one giant invisible mastermind.

That makes people harder to fool in both directions.

Harder for powerful institutions to hide wrongdoing.

And harder for false stories to hijack legitimate distrust.

Trust Is Not Binary

People ask:

Do you trust government?

Do you trust science?

Do you trust the media?

Do you trust corporations?

Those questions are almost too large to answer.

Which government agency?

Which scientist?

Which study?

Which newspaper?

Which journalist?

Which company?

Which claim?

Trust should become more granular.

Verify what can be verified.

Inspect incentives.

Examine track records.

Compare sources.

Allow institutions to earn trust and lose it.

Maybe the future of trust is not:

Believe us.

Maybe it's:

Here is the evidence. Check us.

That principle will become central later.

Education Should Teach Mental Self-Defense

We teach children how to cross a street.

Why?

Because cars are dangerous.

We teach them not to touch fire.

Not to give strangers passwords.

What about:

propaganda?

financial scams?

cult recruitment?

manipulative relationships?

advertising?

political demagoguery?

algorithmic addiction?

AI-generated fraud?

Those are hazards too.

Children should graduate understanding:

attention,

bias,

memory,

probability,

evidence,

persuasion,

social pressure,

financial incentives,

digital privacy.

Not because we want cynical children.

Because freedom requires defenses.

A Better Human Interface

Think about what a Human-Centered information system might optimize.

Not:

How long can we keep this person here?

But:

Did they understand?

Did they learn something?

Did we expose them to meaningful disagreement?

Did they leave more capable than they arrived?

Did the system help them accomplish the goal they chose?

Imagine social media with a button:

Show me the strongest opposing view.

Or:

Where might I be wrong?

Or:

Show original sources.

Or:

Explain this without emotional language.

Or:

What do experts actually disagree about?

The internet does not have to disappear.

It can evolve.

The problem is not connecting humanity.

Connecting humanity may be one of the greatest things we've ever done.

The problem is what the current game rewards inside that connection.

Change the incentive...

and the same network can produce very different humans.

Your Mind Is Part of the Battlefield

Throughout history, power often depended on controlling:

land,

weapons,

food,

money.

Now attention and information matter enormously.

If someone can shape what millions believe reality contains...

that is power.

If an algorithm can predict which message changes someone's behavior...

that is power.

If AI can manufacture persuasive content at enormous scale...

that is power.

So mental independence becomes political independence.

A civilization of easily manipulated humans cannot remain meaningfully free simply because the Constitution says they are.

Freedom requires more than permission.

It requires capability.

The Freedom to Think

Maybe this is one of the first pieces of a new definition of freedom.

Freedom is not only:

the government cannot stop me.

It is also having enough:

education,

information,

time,

mental resilience,

privacy,

security,

and independence

to make a choice that is meaningfully your own.

A starving person is technically free to refuse a terrible job.

A manipulated voter is technically free to vote.

An addicted user is technically free to close the app.

Words alone do not describe the entire reality.

Human-Centered Civilization should care about the capability underneath the freedom.

Not to make choices for people.

To make people more capable of making their own.

The Most Dangerous Person to Fool

At the end of all this, there is one opponent you cannot remove.

Yourself.

You carry:

bias,

memory errors,

fear,

tribal instincts,

status concerns,

wishful thinking,

ego.

So do I.

The goal is not escaping humanity.

It's becoming better at noticing it.

Ask:

Am I seeing the world...

or the part I was trained to notice?

Am I evaluating evidence...

or defending a conclusion?

Did I remember what happened...

or what the story became afterward?

Do I believe this because it is true...

or because people like me believe it?

Am I angry because the claim is wrong...

or because it threatens something I identify with?

What evidence would change my mind?

Who benefits from my attention?

Who benefits from my fear?

Those questions do not make someone weak.

They make someone harder to own.

Because once the human animal learned to build stories, those stories became powerful enough to coordinate millions of strangers.

Religions.

Nations.

Corporations.

Property.

Law.

Money.

Freedom.

The next layer of civilization would not be built only from stone or steel.

It would be built from things that existed first inside the mind.