The sceptic's move is disarmingly simple. Whatever you claim to know, they ask how you rule out a scenario in which you are wrong but everything seems identical. If you cannot rule it out, they ask what your claim to knowledge was resting on. Most people find they run out of answers surprisingly quickly.
Ancient scepticism had a therapeutic purpose that the modern version lost. The Pyrrhonists observed that for any position, arguments of roughly equal force could be assembled on either side. Their recommendation was epoché — suspension of judgement — and their claim was that this produced ataraxia, tranquillity. The dogmatist, committed to how things really are, is perpetually disturbed by challenges; the sceptic, holding nothing, is not. They continued to live by appearances and local custom, so this was a stance about belief rather than a refusal to act.
Descartes turned scepticism into a method rather than a destination. To find something certain, he doubted everything doubtable. The senses sometimes deceive, so they cannot be trusted absolutely. Dreams can be indistinguishable from waking, so no experience proves you are awake. And a sufficiently powerful deceiver could manipulate even your reasoning about mathematics. From this he extracted the one thing doubt could not touch: that something is doing the doubting. The cogito was designed to defeat scepticism — though many think the rebuilding that followed was less successful than the demolition.
The brain in a vat is the modern demon: a brain sustained in fluid, stimulated to produce precisely the experiences you are having now. Every observation you could make is one the scenario predicts. Since the evidence is identical either way, it appears you have no grounds to rule it out — and if you cannot rule it out, the sceptic asks how you know anything about the external world at all.
ResponsesHow the challenge gets answered.
Best explanation. The ordinary picture explains our experience better than the vat hypothesis — it is simpler, more predictive, and less ad hoc. The sceptic replies that this assumes simplicity tracks truth, which is exactly the sort of assumption under dispute.
Lower the bar. Perhaps knowledge never required certainty. On reliabilist accounts, a belief counts as knowledge if produced by a reliable process — and if your perception is in fact reliable, you know, whether or not you can prove you are not a brain in a vat. This is powerful, though the sceptic notes it makes knowledge depend on facts you cannot verify.
Contextualism. The standards for "know" shift with context. In ordinary conversation you know where your car is parked; in a philosophy seminar where vat scenarios are salient, the standards rise and the same claim fails. On this view the sceptic isn't wrong so much as changing the subject.
Semantic responses. Putnam argued that a lifelong brain in a vat could not even formulate the hypothesis truly, since its words would refer to features of the simulation rather than to real brains and vats — so the scenario cannot be coherently stated from the inside.
None of these commands universal assent, which is the honest situation. Scepticism's importance is not that anyone believes it but that any theory of knowledge which cannot say what is wrong with it has not finished its job.