Skepticism: Can We Know Anything?

9 minute read

Published: August 21, 2026

You believe you are reading this page.

But how do you know?

Maybe you are:

dreaming,
hallucinating,
inside a simulation.

Skepticism asks whether our apparent knowledge can survive such possibilities.

Skepticism

Skepticism is not simply:

doubting everything.

Philosophical skepticism asks whether our beliefs satisfy the standards required for knowledge.

Local Skepticism

You can be skeptical about one domain.

Examples:

astrology,
memory testimony,
one experiment.

This is ordinary critical thinking.

Global Skepticism

Global skepticism is much stronger.

It asks whether we can know:

anything about the external world.

This is the classical challenge.

Ancient Skepticism

Ancient Greek skeptics explored suspension of judgment.

Pyrrhonian skepticism emphasized:

epoché,

withholding assent when arguments balance.

The aim was partly tranquility.

Academic Skepticism

Other ancient skeptics argued that knowledge may be impossible.

Different traditions used skepticism differently.

Descartes

René Descartes used radical doubt methodologically.

He wanted to find a foundation that could survive every skeptical challenge.

Methodological Doubt

Descartes did not doubt merely for entertainment.

He asked:

What cannot possibly be doubted?

Everything uncertain would be temporarily set aside.

Sensory Error

Our senses sometimes deceive us.

Objects look smaller at a distance.

Straight sticks appear bent in water.

Therefore sensory belief is not automatically certain.

Dream Argument

Sometimes dreams feel realistic.

If I can dream that I am sitting by a fire, how can I prove right now that I am awake?

This threatens ordinary perceptual knowledge.

Dreaming vs Waking

We usually distinguish waking from dreaming using:

coherence,
memory,
stable interaction.

But the skeptic can imagine a perfectly convincing dream.

The challenge targets logical possibility.

Evil Demon

Descartes imagines a powerful deceiver manipulating all experience.

Even mathematical reasoning might be corrupted.

This pushes doubt to its extreme.

Cogito

Descartes reaches:

I think, therefore I am.

Even if deceived, there must be thinking occurring.

The act of doubt seems to establish at least existence of thought.

What Does the Cogito Prove?

It may establish:

there is thinking now.

Whether it establishes a substantial enduring self is more controversial.

The skeptical foundation is thinner than ordinary identity.

Brain in a Vat

A modern version imagines:

a brain maintained in a vat,

connected to a computer generating all sensory input.

Your experiences would seem normal.

Could you know you are not such a brain?

Simulation Scenario

A simulation hypothesis creates a similar structure.

If all observations are generated artificially, perceptual evidence may not distinguish:

base reality

from:

simulation.

Closure Principle

A skeptical argument often uses epistemic closure:

If you know P,

and know that P implies Q,

then you should know Q.

If you know:

I have hands,

then you should know:

I am not a handless brain in a vat.

Skeptical Modus Tollens

The skeptic says:

If I know I have hands, I know I am not a brain in a vat.
I do not know I am not a brain in a vat.
Therefore I do not know I have hands.

This is powerful because each step feels plausible.

Moore’s Response

G. E. Moore famously reversed the argument.

He held up his hands and effectively argued:

Here is one hand.
Here is another.
Therefore an external world exists.

Moorean Shift

Moore’s strategy says:

ordinary knowledge may be more certain than the skeptical premise.

If a philosophical argument implies:

I do not know I have hands,

perhaps the argument is less credible than common-sense knowledge.

Is That Question-Begging?

Critics say Moore simply assumes what skepticism questions.

Defenders say rational argument always begins from some comparatively secure commitments.

The dispute is about epistemic priority.

Contextualism

Contextualism says the standards for “knows” vary with conversational context.

In ordinary life:

“I know the car is outside”

can be true.

When skeptical possibilities become salient, standards rise.

Why Contextualism Helps

Ordinary knowledge survives in low-standard contexts.

Skepticism reveals that we may not meet extraordinary standards.

Both intuitions are preserved.

Invariantism

Invariantists reject shifting standards.

Knowledge means the same thing across contexts.

They seek another response to skepticism.

Relevant Alternatives

Another idea says knowledge requires ruling out only relevant alternatives.

You need not eliminate every bizarre logical possibility.

For ordinary knowledge, brain-in-a-vat scenarios may be irrelevant.

What Makes an Alternative Relevant?

This is difficult.

Relevant alternatives may depend on:

context,
evidence,
practical stakes.

The theory needs principled boundaries.

Safety

Safety theories say:

If you know P, then in nearby possible worlds where you form the belief similarly, P remains true.

Brain-in-a-vat worlds may be too distant.

Ordinary knowledge can survive.

Sensitivity

Sensitivity asks:

If P were false, would you still believe P?

For skeptical hypotheses, this can generate trouble.

If you were a brain in a vat, you would still believe:

I am not one.

Nozick’s Counterintuitive Result

Tracking theories may say:

you know you have hands

but do not know:

you are not a brain in a vat.

This violates closure.

Some philosophers accept that cost.

Externalism

An externalist can say:

knowledge depends on reliable connection to reality.

If you are in fact in the normal world and perception is reliable, you can know.

You need not prove reliability from inside.

Skeptic’s Reply

But how do you know your process is reliable?

The externalist answer is:

knowledge need not require knowing that you know.

This rejects a strong reflective requirement.

Internalism and the Regress

If every justified belief requires another justified reason, we get a regress.

How does justification end?

This is the regress problem.

Foundationalism

Foundationalists say some beliefs are basic.

They are justified without inference from other beliefs.

Perceptual beliefs are common candidates.

Coherentism

Coherentists say justification comes from fit within a mutually supporting system of beliefs.

No single belief needs absolute foundational status.

Infinitism

Infinitism allows an endless chain of reasons.

This is a less common but serious position.

The regress need not terminate.

Agrippa’s Trilemma

Skeptical traditions highlight three unsatisfying options:

infinite regress,
circular support,
arbitrary stopping point.

This is sometimes called Agrippa’s trilemma.

Münchhausen Trilemma

A modern version makes the same point.

Every justification seems to end in:

regress,
circle,
dogma.

This challenges ultimate foundations.

Fallibilist Response

Perhaps justification does not require ultimate certainty.

Beliefs can be justified by:

reliable evidence,
mutual support

without infallible foundations.

This is the dominant practical response.

Scientific Knowledge Under Skepticism

Science cannot prove:

we are not in a perfect simulation.

But it can construct extremely reliable models of experienced regularities.

This may be enough for empirical knowledge.

Instrumental Success

Scientific theories:

predict,
control,
unify.

Even a skeptic must explain why these methods work so consistently within experience.

Skepticism and Action

Complete practical skepticism is hard to live.

You still avoid:

cliffs,
fire.

Behavior reveals deep reliance on ordinary beliefs.

Hume’s Response

David Hume argued radical skepticism is psychologically unstable.

Nature forces us back into ordinary belief.

We cannot sustain universal doubt continuously.

Skepticism as Tool

Skepticism is valuable when used diagnostically.

Ask:

What supports this belief?
Could I be mistaken?
What alternative explanation exists?

This improves reasoning.

Skepticism as Paralysis

If every possibility of error prevents belief, no action is possible.

Epistemology must work for finite agents.

Absolute doubt is often too demanding.

Radical Skepticism and Probability

Even if certainty is impossible, probabilities can differ dramatically.

Maybe:

brain in vat = logically possible.

That does not make it equally plausible with:

ordinary external world.

Possibility is not probability.

Inference to Best Explanation

The external world explains:

stable perception,
intersubjective agreement,
successful prediction.

A skeptical scenario can reproduce these too, but often at greater complexity.

Ockham’s razor favors ordinary realism.

Underdetermination

Still, if two hypotheses generate exactly the same possible observations, evidence cannot distinguish them.

This is genuine underdetermination.

Some metaphysical questions may remain empirically undecidable.

Pragmatic Vindication

Even if ultimate proof is unavailable, trusting perception works.

It enables:

survival,
science,
coordination.

Pragmatic success does not prove metaphysics, but it supports practice.

Skepticism and AI

A machine also operates under partial observability.

It cannot directly know hidden world state.

It maintains beliefs and updates from evidence.

Epistemic uncertainty is not uniquely human.

POMDP Analogy

In a partially observable environment, an agent maintains a probability distribution over states.

It acts despite uncertainty.

This is a useful model of finite epistemic agency.

Knowing Without Certainty

The lesson is not:

nothing can be known.

It is:

knowledge need not require ruling out every conceivable skeptical scenario.

Fallible knowledge can still be robust.

Intellectual Humility

Skepticism teaches humility.

Our beliefs are:

revisable,
perspective-bound.

But humility should not collapse into:

all claims are equal.

Evidence still matters.

The Philosophical Lesson

Skepticism exposes a gap between:

certainty

and:

ordinary knowledge.

We may be unable to prove that deception is impossible.

But this does not force us to abandon well-supported belief.

The practical task is to distinguish:

reasonable doubt