Category Archives: Confirmation Theory

Section 4 of chapter 7 in The Philosophy of Philosophy aims to identify the judgment skeptic’s mistake. In context, Williamson argued in section 3 that the same line of critique judgment skeptics use against folk theory can be used against elements of judgment skepticism that rely on folk theory. A judgment skeptic holds that we cannot know mountains exist because our evidence is neutral between the ordinary hypothesis and the skeptical hypothesis. Instead, there are only micro-events that humans errantly, though conveniently, classify as mountains. The result, however, is that we cannot possess knowledge or justification about beliefs concerning mountains. When this kind of reasoning is ported over to general skepticism it become clear (according to Williamson) that the reasoning is unsound. With the context of section 3 in mind I return to section 4. Williamson wants to identify the mistake in the judgment skeptic’s reasoning. What makes this line of reasoning bad?

There are two mistakes that Williamson identifies. The first mistake is the use of so-called appearance principles, and the second mistake is committing the consequence fallacy. I will discuss each of these mistakes in turn.

An appearance principle is defined as follows:

[O]ne should be confident that P (on the basis of common sense) only if its appearing (by the standards of common sense) that P is good evidence that P. (2007: 227)

Williamson shows that appearance principles can be used as premises in an argument for general skepticism as well as judgment skepticism. This is a problem because judgment skeptics want to exclude the results of particle physics from skepticism so that they can claim underlying micro-physical events entail the impossibility of mountains. I will provide you with an overview of Williamson’s argument.

Let SS be the judgment skeptic’s scenario in which there are no mountains. In this scenario it falsely appears that there are mountains even though mountains are a metaphysical impossibility. If there really are mountains, then SS must not obtain. For the judgment skeptic: one should be confident that SS does not obtain only if its appearing that SS does not obtain is good evidence that SS does not obtain. However, appearing that SS does not obtain is not good evidence that SS does not obtain, according to the judgment skeptic, so one should have low confidence (in one’s judgment) that SS does not obtain. Now, I turn to a distinction.

Roughly, something is truth-indicative if the appearance of it raises the probability of P. If, on the other hand, appearance (used as a conditional on P) does not raise the probability of P above the probability of P alone, then appearance is falsity-indicative. Appearance principles require one to modulate one’s confidence in P according to how appearance that P provides evidence that P, and only if the appearance of P is truth-indicative should one be highly confident in P.

The use of appearance principles in the reasoning above can also generate general skepticism. Let p be a description of the external world that jives with the judgment skeptic’s understanding of particle physics. Imagine SS* is an evil demon scenario in which p is false but an evil demon makes the truth of p seem to hold. By the same reasoning, the appearance that SS* does not obtain is not evidence that SS* does not obtain (i.e., it is not truth-indicative) because appearances to a subject are systematically deceived by the demon. So, given the appearance principle, one should have low confidence that SS* does not obtain. Because p (the existence of the external world) entails that SS* does not obtain, then one should modulate one’s confidence in p to accord with one’s confidence that SS* does not obtain. The result is that confidence in p should be low even when its appearance raises the probability of p. So, we should be skeptical about the existence of the external world as described by particle physics. Williamson cuts the legs out from under the judgment skeptic’s reasoning. Or, does he?

I’m not satisfied with Williamson’s pattern of pulling the judgment skeptic into general skepticism. Why? The mere possibility of an evil demon scenario precludes the use of appearance principles. In such a scenario appearances are false and, consequently, apperance principles do not hold. Who would reasonably argue that in a Matrix world one should be confident that P only if it appears that P is good evidence that P? By the assumptions of the scenario it appearing that P will not be good evidence that P. So, to argue that appearance principles used in such a scenario result in skepticism about a domain judgment skeptics endorse (particle physics) seems like a ticky-tacky move at best and unwarranted at worst.

The second mistake in judgment skepticism is the consequence fallacy. This fallacy involves criticizing confidence in a theory by focusing on a logical consequence of the theory whose probability is not raised by the evidence. Take the following argument Williamson outlines (2007: 233):

1. Physical events occur that folk geography takes to constitute the presence of mountains in Switzerland.
2. If physical events occur that folk geography takes to constitutes the presence of mountains in Switzerland, then there are mountains in Switzerland.
3. There are mountains in Switzerland.

A person who subscribes to folk geography is likely to endorses the whole argument. However, a judgment skeptic jumps off the boat at premise 2. That is, the evidence may increase the probability of premise 1 but not premise 2. The fallacy comes from arguing that the failure of increased probability in 2, conditional on the evidence, is reason to hold that a high degree of confidence in both 2 and 3 is not warranted. It may be that it is still reasonable to hold a high degree of confidence in 2 and 3 even though evidence raises the probability of 1 but not of 2. The problem comes from, “identifying a logical consequence of the theory (not itself a logical truth) whose probability is not raised by the evidence” (2007:232). It is not the case that evidence raising the probability of a hypothesis makes more probable a logical consequence of that hypothesis. In fact, according to Williamson, when the evidence makes the hypothesis more probable, but not certain, it decreases the probability of the logical consequence of the hypothesis. When evidence makes a hypothesis certain it does not make a logical consequence of that hypothesis more probable. Thus, evidence making more probable premise 1 but not 2 is not a basis from which to argue that one is not entitled to a high degree of confidence in premise 2 and 3.

Williamson’s logical consequence point brings up issues in confirmation theory. His point has prompted me to explore confirmation theory in more detail. Some useful reads in this regard can be found here and here.