What Is Belief?

You believe:

the Earth orbits the Sun.

You believe:

your keys are on the table.

You may believe:

a friend will keep a promise.

What exactly is a belief?

A belief is not merely a sentence spoken aloud.

It is a state that guides how an agent interprets and acts in the world.

Belief as Attitude

Philosophers often describe belief as a propositional attitude.

A person takes a certain attitude toward a proposition.

Example:

Proposition:

it will rain tomorrow.

Possible attitudes:

• believe it,
• doubt it,
• deny it.

Proposition

A proposition is roughly:

the content that can be true or false.

Different sentences can express the same proposition.

“Snow is white.”

and a translation in another language may express the same content.

Belief and Truth

Belief is psychological.

Truth is not.

You can believe something false.

This is one of the most important distinctions in epistemology.

Belief as Representation

A cognitive system may represent the world as being a certain way.

To believe:

the door is locked

is to maintain a model in which:

door = locked.

This representation can guide action.

Belief and Action

If you believe the door is locked, you may:

• look for a key,
• choose another exit.

Beliefs influence behavior through their content.

Dispositional Belief

Not every belief is consciously active.

You believe:

Paris is in France

even when you are not thinking about Paris.

This is a dispositional belief.

Occurrent Belief

An occurrent belief is actively present in thought.

Example:

you suddenly think:

“The meeting is at three.”

The distinction matters because minds contain more beliefs than are consciously accessible at once.

Explicit and Implicit Belief

Some beliefs are explicit:

you can state them directly.

Others are implicit:

they influence behavior without clear verbal endorsement.

Implicit Bias

A person may sincerely endorse equality while displaying automatic biased associations.

Does that association count as a belief?

The answer depends on theory.

Belief may come in layers.

Belief vs Acceptance

Sometimes we accept a claim for practical purposes without fully believing it.

A scientist may say:

“Assume the ideal gas law.”

This is acceptance within a model.

Not necessarily belief about literal reality.

Belief vs Guess

A guess is a weak commitment.

You may guess:

the coin will land heads.

But you may not genuinely believe it.

Confidence matters.

Belief and Confidence

Beliefs are often treated as binary:

believe / do not believe.

But real cognition is graded.

You may be:

• 55% confident,
• 95% confident.

This suggests probabilistic belief.

Credence

A credence is a degree of belief.

Example:

[ P(\text{rain tomorrow}) = 0.7 ]

Bayesian epistemology often models beliefs this way.

Binary vs Graded Belief

Do ordinary beliefs reduce to thresholds on credence?

Perhaps:

believe P if confidence > 0.8.

But thresholds create problems.

The relationship is not straightforward.

Lottery Paradox

Suppose a lottery has one million tickets.

For each ticket, you believe:

this ticket will lose.

Yet you also believe:

one ticket will win.

Your highly probable beliefs cannot all be jointly true.

This is the lottery paradox.

Belief Consistency

Ideally, beliefs should be mutually consistent.

Humans often hold contradictions.

Belief systems are not perfectly logical databases.

Cognitive Dissonance

When beliefs or actions conflict, people may experience cognitive dissonance.

They sometimes revise beliefs to reduce tension.

Belief change is psychologically motivated, not purely logical.

Confirmation Bias

People tend to seek and interpret evidence in ways that support existing beliefs.

This makes beliefs self-reinforcing.

We will examine this more deeply later.

Belief Perseverance

A belief can survive even after its original evidence is discredited.

This is belief perseverance.

Once integrated into a worldview, beliefs can become resistant to correction.

Motivated Reasoning

Desire can shape belief formation.

People may accept evidence more easily when it supports:

• identity,
• goals,
• group membership.

Belief is not always neutral inference.

Wishful Thinking

You want:

the test result to be good.

You begin to believe:

it probably is.

This is wishful thinking.

Desire contaminates evidence assessment.

Belief and Desire

Belief represents:

how the world is.

Desire represents:

how the world should be.

This is a classic distinction.

Direction of Fit

Beliefs have a mind-to-world direction of fit.

If belief conflicts with reality:

change belief.

Desires have a world-to-mind direction of fit.

If reality conflicts with desire:

try to change reality.

Intentions

Intentions are different again.

They commit the agent to action.

Belief:

the door is closed.

Desire:

I want it open.

Intention:

I will open it.

Belief and Evidence

A rational belief should respond to evidence.

If evidence changes, belief should change.

This is an epistemic norm.

Humans often violate it.

Evidence Is Not Self-Interpreting

The same evidence can support different conclusions depending on:

• background assumptions,
• models.

Belief formation always occurs within context.

Background Beliefs

To interpret one claim, we rely on many others.

Example:

You believe the thermometer is reliable.

That background belief affects whether you trust the reading.

Web of Belief

W. V. O. Quine and others emphasized that beliefs form a network.

Evidence usually confronts not one isolated statement but a larger web of belief.

Revision can occur at different points in the web.

Central and Peripheral Beliefs

Some beliefs are easy to revise.

Others are deeply connected to identity or theory.

Central beliefs are more resistant to change.

Belief Revision

Formal belief-revision theory studies how rational agents should update inconsistent belief sets.

One influential framework is AGM theory.

AGM Intuition

When new information arrives, an agent may need to:

• expand beliefs,
• contract beliefs,
• revise beliefs.

The goal is minimal coherent change.

Belief and Memory

Many beliefs depend on memory.

But memory is reconstructive.

A remembered event can be:

• distorted,
• incomplete.

Belief inherits memory’s imperfections.

Testimony

Much human belief comes from other people.

You believe facts about:

• distant countries,
• history,
• science

without direct observation.

Testimony is essential.

Epistemic Dependence

No individual verifies everything.

Knowledge is social.

Belief systems depend on:

• experts,
• institutions,
• records.

This makes trust epistemically central.

Authority

Authority can be a legitimate source when the authority is:

• competent,
• reliable.

But authority can also be abused.

Belief should not become blind obedience.

Belief and Identity

Some beliefs express group membership.

They function socially as signals.

This can make factual correction difficult.

Changing belief may feel like changing identity.

Ideological Beliefs

Beliefs can organize whole worldviews.

They may determine:

• what counts as evidence,
• whom to trust.

Epistemology therefore intersects sociology.

Delusion

A delusion is a strongly held belief resistant to counterevidence, often associated with psychiatric conditions.

Clinical definitions are complex.

The key point is that confidence alone does not guarantee truth.

Hallucination vs Delusion

A hallucination is typically a perceptual experience without corresponding external stimulus.

A delusion is a belief.

Perception and belief can dissociate.

Seeing Is Not Believing

You can experience an illusion while believing:

“I know these lines are equal.”

Belief can override perceptual appearance.

Belief and Imagination

You can imagine:

a dragon

without believing dragons exist.

Representation does not imply belief.

This is crucial for understanding fiction.

Fiction

While reading a novel, you may emotionally respond to characters you know are fictional.

Imagination can guide emotion without literal belief.

Pretense

Children can pretend:

this box is a spaceship.

They represent one world while believing another.

Cognition can maintain multiple models.

Belief in Machines

Can machines have beliefs?

A simple system can store:

door_locked = true.

Is that a belief?

Functionalists may say yes in a minimal sense.

Others require richer semantic commitment.

Database vs Belief

A database contains information.

But it does not necessarily:

• act,
• update autonomously,
• care about truth.

Belief may require embedding representation inside agency.

Agent Belief

In AI, agents often maintain belief states about uncertain environments.

A belief state can be a probability distribution over possible world states.

This is operationally precise.

Bayesian Belief State

Example:

[ P(s_1)=0.6,\quad P(s_2)=0.4 ]

The agent acts based on this uncertainty.

Belief becomes mathematically actionable.

False Belief in Machines

A robot can maintain an incorrect map.

Then it behaves according to a false representation.

This resembles false belief functionally.

Belief and Understanding

A system can store a proposition without understanding it deeply.

Belief and understanding are distinct.

Humans also hold many shallow beliefs.

Belief and Commitment

Belief involves some degree of commitment to truth.

If challenged, the agent should defend or revise it.

This distinguishes belief from temporary token storage.

Contradictory Beliefs

Humans can hold:

P

and effectively:

not-P

in different contexts.

This suggests beliefs may be fragmented rather than globally consistent.

Contextual Activation

Different situations activate different representations.

A person may reason differently at:

work, home.

Belief may be context-sensitive.

Fragmentation

Some philosophers propose fragmented belief models.

The mind contains partially isolated belief clusters.

Contradictions can persist because they are rarely activated together.

Rational Ideal vs Human Reality

Logic gives ideals:

• consistency,
• evidence responsiveness.

Human cognition is bounded.

We operate under:

• limited memory,
• limited time.

Belief rationality is therefore approximate.

Bounded Rationality

Herbert Simon emphasized bounded rationality.

Agents satisfice rather than optimize perfectly.

Belief formation occurs under cognitive constraints.

Belief as Prediction

One modern view says beliefs are parts of predictive models.

The brain predicts:

• sensory input,
• consequences.

Belief becomes probabilistic world-modeling.

Predictive Processing

Under predictive-processing frameworks, perception itself involves hypotheses about causes.

The boundary between:

perception

and:

belief

becomes less sharp.

Belief and Action Loop

Beliefs guide actions.

Actions generate new evidence.

Evidence updates beliefs.

This forms a loop:

belief → action → observation → belief.

Epistemic Agency

A good agent does not merely hold beliefs.

It manages them.

It asks:

• what evidence do I have?
• how confident should I be?
• what would change my mind?

This is epistemic agency.

Belief Is Cheap

Anyone can believe anything.

The hard question is:

When does belief become justified?

And when does justified belief become knowledge?

The Beginning of Part XVIII

Part XVIII examines:

• belief,
• knowledge,
• skepticism,
• bias,
• uncertainty.

The central question is:

How should a finite mind decide what to believe?

The Philosophical Lesson

Belief is a commitment to a representation of the world as true.

It can be:

• explicit,
• implicit,
• graded,
• false.

Belief guides action.

But belief alone is not enough for knowledge.

The Next Question

You can sincerely believe something and still be wrong.

So what must be added before belief becomes knowledge?

First we must make one distinction absolutely clear:

Belief Is Not Knowledge.