Skepticism and its limits

Scenario

Ask someone, seriously, whether they can prove right now that they are not dreaming, or that some far more capable deceiver is not feeding them a false stream of experience in place of the real world. Most people, pressed honestly, admit they cannot. No observation settles it, because any observation offered as proof would itself be part of the dream or the deception, and so cannot be trusted to rule it out. Then the same person gets up, eats breakfast, crosses the street when the light changes, and takes a doctor's diagnosis at face value. They live exactly as though the doubt were false, while granting they have no way to show that it is. That gap, between a doubt nobody can defeat and a life nobody actually lives as doubtful, is what this entry is about. The question worth asking is what a person committed to the doubt would actually have to give up in order to hold it consistently.

Definition

A skeptical argument, in the sense meant here, argues that a belief lacks the justification needed to count as knowledge. It works by demanding a reason for the belief, then a reason for that reason, and so on. Call this chain the regress of justification: for any reason offered, it is fair to ask why that reason should be trusted, and a further reason is owed. If the chain never terminates in something that needs no further defense, then no belief is ever more than justified relative to assumptions taken on faith rather than fully justified in its own right. A skeptic aims this regress at a target and asks what could stop it there.

Global skepticism aims the regress at everything at once. The dreaming argument and the deceiving-demon argument both ask for a justification that would survive even if all of experience turned out false, and neither can be met, because any evidence a person points to is itself experience, which is exactly what is in question. Local skepticism aims the same regress at one narrower class of belief while leaving the rest alone. The problem of induction is the clearest example: it asks what justifies moving from observed cases to unobserved ones, without doubting the senses or the existence of an external world at all. That entry works the induction argument in full; here it marks the other end of a scale, since a single regress structure can be run at the scope of everything or at the scope of one inference pattern, and the two scopes call for different responses.

The self-undermining point is what gives the global version its real limit. Whatever standard is strict enough to defeat ordinary knowledge defeats the skeptic's own reasoning by the same measure. The move from "you cannot rule out the demon" to "you do not know you are sitting in this room" is itself an inference, and it rests on a principle, roughly that knowing something requires ruling out every alternative that would make it false, which the skeptic must be claiming to know. If the demon hypothesis is strong enough that nothing at all can be known while it stands open, that principle is among the casualties, and the argument removes the ground it needed to stand on in order to be stated. Applied partway, as a challenge to overconfidence, the doubt does real work. Applied all the way, consistently, it takes down its own premises along with everything else.

One ancient school worked through this consequence rather than skip past it. The Pyrrhonists, confronted with the same regress, did not conclude that nothing is known, a claim that would itself fall to their own argument the moment it was asserted. Their move instead was to suspend judgment: for any claim, they held, an argument of roughly equal strength can be built for its denial, so the position that survives its own scrutiny is neither to affirm the claim nor to deny it, only to withhold assent and keep looking. Burnyeat (1980) treats this suspension as the mechanism by which Pyrrhonism tries to escape the self-undermining trap that catches a flatter assertion like "nothing can be known." Whether the escape fully works is disputed; what it shows plainly is that even the oldest form of the doubt was built with the collapse in mind, rather than in ignorance of it.

Common mistakes

Treating a skeptical argument as something that must be refuted, point by point, before it can be set aside. A doubt that resists direct rebuttal is not thereby a doubt a person is required to live by; the question this entry asks instead is what applying it consistently would cost, which is a different and answerable question.

Collapsing "I might be wrong about this" into "nothing at all is known." The first is an ordinary, local admission that a specific belief could be mistaken, and it is compatible with knowing a great many other things. The global skeptic's conclusion is a claim about every belief at once, and confusing the two makes a reasonable modesty look like a radical thesis it never had to be.

Ignoring that the skeptic's own argument leans on the very kind of reasoning it doubts. A regress built from ordinary inference cannot, when pushed to its full strength, exempt the argument that built it; a reader who spots this is checking the argument's own coherence, not merely disliking its conclusion.

Confusing global and local skepticism because both use the word "doubt." Doubting one inference pattern, as in the problem of induction, leaves the rest of a person's beliefs standing; doubting the reliability of experience as such does not. Naming which scope a given skeptical move occupies is most of the work of assessing it.

Test yourself

Pick one skeptical argument: the dreaming argument, the deceiving demon, or the brain-in-a-vat scenario. Work out exactly what it would cost to apply consistently, all the way, rather than halfway. Check whether its own reasoning survives being turned back on itself, the way the self-undermining point above describes. Then name the specific commitment the argument's own user would have to give up to hold the doubt honestly, a shared standard of evidence, the reliability of memory, or the very principle that generates the doubt in the first place, rather than gesture at the cost in general terms. Decide, and state your reason: is the honest response to accept that cost and suspend judgment, in the Pyrrhonian manner, or to treat the doubt as one nobody can actually live and set it aside on those grounds?

Success looks like two things written down, not just felt. First, the exact commitment the argument would have to sacrifice if pressed all the way, named specifically rather than described vaguely as "a lot" or "everything." Second, a stated verdict on whether the doubt is one a person could actually hold while acting in the world, together with the reason behind that verdict, not the verdict on its own.