QuestionsCan we be certain of anything?Agrippa's trilemma: every justification collapses
Can we be certain of anything?We must doubt

Agrippa's trilemma: every justification collapses

The five modes of Agrippa; the Münchhausen trilemma; the regress problem

Sextus Empiricus c. 160 · c. 210Against every claim an equally strong one can be set: we must suspend judgement.
See profile
What it claims

On any disputed question, we must suspend judgement: no claim can be securely justified.

Why

Every proof needs another, and the chain ends only in regress, circle or assumption, none of which proves anything.

The same idea through history
Sextus Empiricus190Sextus Empiricus (c. 160-210) hands down the five modes as those of "the later Sceptics": disagreement, infinite regress, relativity, hypothesis and circularity. Other sources attribute them to a sceptic named Agrippa. The last three modes form the trilemma presented here.
Michel de Montaigne1580Montaigne, a reader of Sextus, reuses the figure in the Apology for Raymond Sebond: to judge, we need a verified instrument, and to verify it another one, "to all infinity".
In short

"Agrippa's trilemma: every justification collapses" is an argument by Sextus Empiricus (Outlines of Pyrrhonism I, 164-177 (c. 190), 190) on the question "Can we be certain of anything?" (answer: we must doubt). Its claim: On any disputed question, we must suspend judgement: no claim can be securely justified. According to AskPhilosophy's logic check (Z3 solver), the argument is valid only if one grants a premise the author leaves unstated (P5). AskPhilosophy presents 4 objections, 2 of them with a documented reply, and the original text behind each premise, checked word for word.

Outlines of Pyrrhonism I, 164-177 (c. 190)
Valid with an implicit premise

You claim something, and someone asks: "How do you know?" You give a reason. Then they ask: "And how do you know that reason?" And so on. The Greek sceptics noticed that this series of questions can end in only three ways, and that none of them is convincing. This is Agrippa's trilemma, one of the most lasting arguments in all of philosophy. Sextus Empiricus does not set it out as a list of premises: the P1 to P5 form below is our reconstruction from his five modes (Outlines of Pyrrhonism I, 164-177).

P1When we give a reason to prove a disputed claim, that reason itself needs a proof, and that proof another, and so on.
P2This chain of proofs can end in only three ways: it goes on without end (the regress), it comes back to its starting point (the circle), or it stops at a claim that is simply assumed without proof (the hypothesis).
P3An endless regress proves nothing, for lack of a starting point; a circle proves nothing, because the proof depends on what it was supposed to prove.
P4A hypothesis proves nothing either: if the one who makes it deserves to be believed, his opponent who makes the opposite hypothesis deserves it just as much.
P5 · implicit premiseWhen a disputed claim can be justified only in these three ways, and none proves it, the only reasonable position is to assert nothing, that is, to suspend judgement.
Therefore, on any disputed question, we must suspend judgement: no claim can be securely justified.

click a premise to reject it

If you accept every premise, logic forces you to accept the conclusion. So the debate is about the premises.
Checked by a logic solver: the premises cannot all be true while the conclusion is false.
Without the implicit premise: Without P5, one could accept P1 to P4 and still go on believing what one believes, out of habit or trust, simply admitting that one cannot prove it. Many philosophers do this: being unable to prove does not force us to suspend.
The argument is valid: P1 says a proof always calls for another, P2 that the chain has three outcomes, P3 and P4 that each fails, and P5 draws the suspension of judgement. The weak point is P2 (are there really only three outcomes?), then P4 (is every hypothesis arbitrary?) and P1 (does a proof always call for another?). Sextus adds a precaution: the sceptic speaks of what is disputed, not of what appears to everyone.
Analogy. Picture a child who answers every explanation with "why?". A parent can keep answering forever, go in circles ("because that's how it is") or say "that's just how it is, full stop". In all three cases the child has received nothing that proves. The sceptic says this is the situation of all our knowledge.

Objections

4

Each objection is a conversation: click to see who replies and where the debate stands.

René Descartes 1637 · attacks premise 4

There is a starting point that is not an arbitrary hypothesis: "I think, therefore I am". When I try to think everything false, I who think it must be something. This truth is so firm that the most extravagant suppositions of the sceptics cannot shake it. The chain therefore stops on a secure foundation, evident by itself.

Empiricus replies
A
Antoine Arnauld 1641 · Arnauld's objection (Fourth Objections)
Arnauld asks how Descartes avoids a circle. Descartes says we are sure of what we conceive clearly because God exists. But we are sure that God exists only because we conceive him clearly. We would therefore already have to be sure that everything clear is true before proving God: we fall back on Sextus's circle.
René Descartes 1641 · Descartes's reply (Replies to the Fourth Objections)
Descartes denies the circle. At the moment itself, we are sure that God exists because we attend to the reasons that prove it. Afterwards, it is enough to remember having conceived a thing clearly to be sure it is true, which would not suffice if we did not know that God exists and cannot deceive.
The debate goes on (our summary) The "Cartesian circle" is one of the most discussed problems about Descartes, and many readers find his reply clever but not always convincing. The cogito itself holds up better: even its critics admit that it is hard to deny. But it gives only a very narrow starting point, and the sceptic can ask whether it is enough to rebuild the rest of our knowledge.
L
Laurence BonJour 1985 · attacks premise 1

Coherentism rejects the picture of a chain. A belief is not justified by a single line of reasons resting on one another, but by its place in a set of beliefs that support one another and form a coherent whole. A narrow circle is vicious, a large coherent system is not: the regress never even starts.

no reply
Empiricus never answered this objection.
The debate goes on (our summary) Critics ask two questions. Is a very coherent novel true for that reason? And where does contact with the world come from, if everything is justified inside the system? Coherentists reply by building observation into the system, which is debated. Our summary rests on the Stanford Encyclopedia of Philosophy's presentation, not on BonJour's own text.
Aristotle 350 BC · attacks premise 1

In the Posterior Analytics (I, 3), Aristotle examines this difficulty in advance. Some say: if all knowledge comes from proof, we fall into the regress, so there is no knowledge. Others accept the circle. Aristotle rejects both: not all knowledge is demonstrated, because first principles are known without proof, and this is what lets demonstration begin.

no reply
Empiricus never answered this objection.
The debate goes on (our summary) Aristotle's reply is the root of all foundationalism. The sceptic is not disarmed: he asks how we know these principles are true, and why another philosopher would not lay down other principles with the same assurance. This is the objection of P4. Disagreement among philosophers about what is self-evident feeds the sceptic.
David Hume 1748 · attacks the conclusion

Nobody can live as a complete sceptic: this is the so-called objection from apraxia (the impossibility of acting). Hume writes that the great destroyer of Pyrrhonism is action and the occupations of common life: these principles triumph in the schools, but vanish like smoke once they meet the realities of life.

Empiricus replies
Sextus Empiricus 190 · Sextus's anticipated reply (c. 190)
Sextus foresaw the objection. The sceptic does not abolish appearances: he sticks to them. He lives by the ordinary rules of life, without dogmatizing, since one cannot remain completely inactive. He doubts what things are in reality, not what they appear to be.
The debate goes on (our summary) Sextus and Hume do not aim at quite the same thing. Hume attacks a total scepticism that would also govern life. Sextus says he suspends judgement only on disputed theoretical questions, and follows appearances for the rest. The debate is about whether this line can be held: can we really suspend judgement on everything disputed, and stay calm?