In Wittgenstein's work, criteria are contrasted with symptoms, where symptoms are taken to be outward manifestations of something while criteria point to the thing itself. For instance, noticing the barometer drop may be taken as a symptom of rain, while
seeing and feeling water drops fall from the clouds is a criterion for rain. Criteria differ from symptoms in that they are definitional: while we may note that the barometer drops when it rains, we don't define rain as, "the weather when the barometer dr
ops," but rather as, "water drops falling from clouds." If the barometer drops, there is room for dispute as to whether it will indeed rain, but not if water drops fall from clouds.
Contrary to Wittgenstein's earlier interpreters, criteria are important not for telling us what something is as much telling us what sort of thing something is. Wittgenstein does not use criteria to distinguish someone who is in pain from someone who is p
retending to be in pain. Rather, Wittgenstein uses criteria to determine that pain—real or feigned—is the thing in question here. If someone is writhing on the floor and moaning, that person could plausibly be faking the pain. But there is no
question that pain is in question here. If someone says, "he's faking it," we know that "it" is "his pain," and not "his love for Wagner." The distinction between symptoms and criteria is not as sharp as it would be with Wittgenstein's earlier interpreter
Criteria are grammatical tools, not factual tools: they do not help us settle matters of fact such as "is he in pain?" but help clarify grammatical matters. For instance, a criterion for something's being an order is that it is obeyed. Obviously, not all
orders are obeyed, but a person giving an order will at least hope to be obeyed. When we say a criterion for something's being an order is that it is obeyed, we are not saying that all orders are always obeyed, but that orders are the sorts of things that
one is normally expected to obey. If all orders were always disobeyed, the word "order" would no longer make sense.
This example of giving orders shows the extent to which the meanings of our words are dependent upon our practices. Someone arguing against Wittgenstein might say that an order is an order; we all know what it means, and whether it is obeyed or not is a p
ractical issue that has nothing to do with the meaning of the word. But Wittgenstein asks, what if orders were never obeyed: would that have no effect on the meaning of the word? An order is only an order because of the social practices they are embedded
in. We generally forget the importance of criteria like "an order is something that is obeyed" because they are always right before our eyes. Wittgenstein's odd examples, like orders that are never obeyed, or asking whether a stove feels pain, remind us
that the words we use and the questions we ask only have a sense because their use is cemented in place by our forms of life.
In discussing first-person ascriptions of pain or color-sensations, Wittgenstein is not trying to deny anything that is obviously true. He is not denying that I can feel pain, or that I can express this pain by saying, "I am in pain." He is challenging m
y temptation to talk about "my pains" or "the pains I have" in order to contrast them with other pains, and arrive at some conclusion about the nature of pain by means of this contrast. But, Wittgenstein points out, what is there to contrast? I can say
"my shoes" while pointing at a particular pair in a row of shoes, but there is no row of different pains that I can point toward, singling out the one that is mine. And if I could somehow survey a number of different pains, only one of which was mine, wha
t criteria could I use to distinguish my pain from others' pain? There are no criteria regarding my own pains: there are no questions of knowledge, doubt, investigation, and so on, which I can raise regarding my own pains. Thus, Wittgenstein is not oppos
ing the idea that first-person pain ascriptions exist, but he is opposing the idea that these ascriptions can then be treated as objects of knowledge.