§1. Varieties of Scepticism

Some preliminary clarifications and distinctions need to be made if we are to approach the topic of scepticism.

(A) I shall take ‘Scepticism’ to be the negative view that we lack knowledge or justified belief of some kind or other. I shall define ‘dogmatism’ to be the positive view that we have knowledge or justified belief of some kind. Using the word ‘dogmatism’ in this way is at a slight remove from the ordinary meaning, which implies something like strongly believing things without adequate grounds. But there is no convenient word for the common sense view that scepticism denies. In my preferred terms, the dogmatist asserts the instantiation of positive epistemic properties, while the skeptic asserts the instantiation of negative epistemic properties. The negative/positive distinction is crucial in epistemology. Scepticism says that we do not know and do not have justified in beliefs. The same holds of other epistemic properties such as wisdom or intelligence. They have opposites or contraries: foolishness or stupidity. A skeptic about intelligence says that no one is intelligent and everyone is stupid. (This is different from being skeptical about the notion of intelligence.) This kind of scepticism contrasts with what is called ‘Pyrrhonian’ scepticism. Pyrrhonian skeptics do not assert that we lack knowledge or justified belief, but say that we should withhold or suspend belief. My focus will be on scepticism as I have defined it. Pyrrhonian scepticism, in its various guises, needs separate treatment.

(B) Dogmatism and scepticism come in plain and in modal versions. That is, dogmatists may say that we actually instantiate positive epistemic properties, such as having knowledge or justified belief. Or they might say that we can have knowledge or justified belief. Similarly there are two types of scepticism, according to which we lack, or necessarily lack, knowledge or justified belief. In other words, the skeptic either says that knowledge or justified belief are non-actual or that they are impossible.

(C) Such claims might be restricted to certain kinds of bearer of epistemic properties, or they might be general. For example, perhaps true wisdom is the property of God alone, as Socrates thought. Or perhaps women know things that men cannot know. Or perhaps everyone is stupid except me. (This is a popular view!)

(D) Such claims might be global, about all knowledge or justified belief about any subject matter, or just about some restricted sub-set of subject matters. For example, we might assert or deny knowledge of the external world, or other minds, or mathematics, or theology. Or we might divide subject matters in a different way. Perhaps we have or lack knowledge or justified belief with certain characteristics. For example, we might assert or deny that we have a priori knowledge, or that we have empirical knowledge.

(E) Dogmatism or scepticism may apply to all orders of knowledge or just to certain orders. Some dogmatism or scepticism is only first-order. It says that we know, or do not know, for example. But some claims are meta-dogmatism or meta-scepticism. This is one kind of restriction on the subject-matter of knowledge or justified belief. One second-order claim might be that we know that we know, while another might be that we do not know that we know. Formalizing this: KK, ~KK. Other possibilities are K~K and ~K~K. This yields four possibilities. Suppose we add justification. This yields, KK, KJ, JK, JJ and so on, 16 possibilities in all, including negatives. Suppose we add modalities: for examples, we can know that we know, or we can know that we can know, or that we know that we can know, and so on. So must we add four extra possibilities for each of the 16 above? No: eight more, if we consider the possibilities and their negations. For KK, we have, possiblyKK, KpossiblyK, possiblyKpossiblyK, ~possiblyKK, K~possiblyK, ~possiblyKpossiblyK, possiblyK~possiblyK, ~possK~possiblyK. So there are 8 times 16 = 128 modalized meta-epistemic doctrines, for two epistemic properties, to add to the 16 non-modalized meta-epistemic doctrines, making 144. Furthermore, why should there be just two epistemic properties? We might also hold combinations of such views, such as ~possiblyKK & ~possiblyK~K.

My point? There are a lot of views out there! They differ in plausibility, and some arguments support some views but not others.

§2. Skeptical Views and Arguments

There are skeptical views and skeptical arguments. These should be distinguished. Corresponding to any of the many possible epistemic views, there are a number of possible arguments for that view. And one argument may support, or be put forward in support of more than one epistemic view.

Now why would someone assert some epistemic view? Let us start with simple first-order epistemic views. Why might someone assert that we know something or that we do not know something or that we can or cannot know things – positive or negative epistemic claims? Why be dogmatic or skeptical, in plain or modalized forms? For example, why think that we know or can know when there is a hand before us?

The answer, surely, is that we have a view about that in virtue of which we know, or can know, or about that in virtue of which we do not know or cannot know. This ‘in virtue of’ signifies a dependence relation, which is a metaphysical relation, distinct from necessitation – a relation that links epistemic and non-epistemic property instantiations.Footnote ¹ That is, we have views about the maker of knowledge, or the maker of the possibility of knowledge, or the maker of the absence of knowledge, or the maker of the impossibility of knowledge. For example, one common dogmatic view is that I know that there is a hand before me because I see it. With other epistemic notions, such as justification, understanding, wisdom, intelligence, or their opposites, this pattern is replicated.

A dogmatic or skeptical argument would be the offering of reasons for asserting a positive or negative epistemic view. It would be an argument for the conclusion that something is a maker for a positive or negative epistemic property instantiation. But that argument would be a second-order one that says that something justifies us in asserting a dependence relation between positive or negative epistemic properties and nonepistemic properties.

In the case of skeptical arguments, we can observe that it seems that skeptical arguments depend on dogmatic views. Skeptics seem to have dogmatic views about what the lack of knowledge or justification depend on. Such skeptical arguments presuppose epistemic/non-epistemic dependence relations and because of that they assume dogmatism. Scepticism presupposes dogmatism. My complaint is not just that scepticism is self-refuting, since it assumes dependence relations, it is also that the burden of proof is clarified in a way that disadvantages the skeptic.

Let us look at some skeptical arguments. There are three prominent forms of skeptical arguments: challenge, comparison, and possibility arguments. As we shall see, these arguments all assume epistemic dependence relations. In challenge arguments, the skeptic responds to a knowledge or justification claim made on certain grounds by asking: ‘Why does that make for knowledge or give me a reason to believe something?’ In comparison arguments, the skeptic compares two cases that seem epistemically equivalent and infer scepticism. In possibility arguments, the skeptic appeals to the possibility of error, and infers scepticism.

§3. Challenge Scepticism

Given a knowledge or justification claim, which is made on a certain ground, the challenge skeptic replies: ‘But why does that make for knowledge or justification?’ This form of argument is of limited force. For the skeptical challenge can be met by the dogmatic assertion not only that the epistemic/non-epistemic dependence relation in question obtains but also that it is that it does not obtain in virtue of anything else. If something is a epistemic property-maker, that may just be a basic dependence fact. It may be a fact that some X makes for knowledge or justification and there need be nothing that makes X make for knowledge or justification. Of course, some dependence relations are explained by others. But some dependence relations are not explained by others. Those are the unexplained explainers. So the skeptic's question may be rejected as begging the question against the idea that the dependence relation is basic.

For example, suppose someone claims that a person knows something in part due to a causal relation between his belief and the fact it represents. The challenge skeptic then asks: but why does such a causal relation make for knowledge? The dogmatic answer is: it just does – that making relation is an unexplained explainer. The dogmatist might concede that in some sense it could be that God wills that a causal relation makes for knowledge, by analogy with the divine command theory of ethical norms.Footnote ² But the making relation between causation and knowledge may also be brute. So the challenge argument achieves little.

The challenge skeptic might reply by denying that X makes for knowledge. But then there is the meta-question about this denial. It looks like the dogmatic assertion that X is not a knowledge-maker. Such skeptics then faces skeptical challenge questions of their own there. The denial that X makes for knowledge is just as dogmatic as the assertion that X makes for knowledge. Thus the skeptic is a dogmatist. Scepticism looks like an impossible doctrine because at some point a positive or negative epistemic judgment is asserted dogmatically. There is no escaping dogmatism!

Suppose that the skeptic replies instead by asking: ‘Ah but how do you know that X (for instance, a causal relation) partly makes for knowledge?’ But this is to ask a meta-knowledge question. It is a different question from asking why X makes for knowledge. We must be scrupulously careful to avoid what William Alston called ‘level-confusions’, which are endemic in discussions of scepticism.Footnote ³ Claiming a justification for saying that we know as if that were the same as claiming knowledge is a classic level confusion in Alston's sense. These are utterly different issues. One is a meta-epistemic issue, the other a first-order epistemic issue.

Moreover, suppose it is true that we do not know that the causal relation makes for knowledge. It does not follow that we do not know. For from ¬KKp, ¬Kp does not follow. And Kp does not imply KKp (for obvious explosive reasons). So there are no first-order skeptical consequences from second-order scepticism. It also looks as if the skeptic is asserting K¬Kp, which would be a dogmatic assertion.

Does this argument only show that there is a problem with global scepticism but not about local forms of scepticism? Perhaps one cannot be a skeptic about everything since scepticism always makes some assumptions; but perhaps that does not impede many powerful kinds of scepticism about local ranges of subject matters or certain kinds of knowledge. For example, one might know one's own mental states but nothing beyond them, or we might have a priori knowledge of the concept of knowledge from which we can deduce that we have no empirical knowledge of anything. This is incorrect. Even local scepticisms, for example about the external world or about empirical knowledge, make assumptions, local assumptions, about the epistemic making relation. And that is what generates problems with skeptical arguments, even locally.

So far I have defending dogmatism. But there is another corollary: one must be a certain kind of skeptic. Where n is the order of knowledge, we know that there is some n such that ¬K_np. So there is no escaping scepticism! But that is an acceptable kind of scepticism. Perhaps I do not KKKKKKKp. We can live with that. Thus, strangely, both scepticism and dogmatism are true. Scepticism is our fate just as much as dogmatism!

§4. Comparison Scepticism

The second form of argument – comparison scepticism – asserts various intuitive epistemic equivalences and argues from there. The argument is: here is one case, an ordinary one where common-sense says that we have knowledge or justified belief; here is another case, an unusual case, where we have a strong intuition that we do not have knowledge or justified belief. But they seem to be epistemically equivalent. If we know or are justified in one case, then we also know or are justified in the other, and if we do not know or are not justified in one case, then we also do not know or are not justified in the other. But it is undeniable that we do not know in the unusual case. So common sense is wrong that we know or have justified belief in the ordinary case.

For example, perhaps I was previously deceived looking at things from a distance or when dreaming. I might compare those cases with my current situation and conclude that both have the same epistemic properties. This is a comparison of two actual cases. Or I might compare my actual situation with a possible case: if I were a brain in a vat then I would have the same epistemic properties that I do in an ordinary actual situation. So I do not know or my beliefs are not justified in the actual situation.

However, both these comparison forms of argument assume the dogmatic view that certain features are not epistemic difference-makers between the two cases. This form of argument in epistemology fares like common moral/political arguments over animal rights, and gender, race etc, where it is asserted that some feature does not make a moral difference. This is just as dogmatic as asserting that the feature does make a moral difference. Although in many cases we might agree with the dogmatic view, the argument is dialectically ineffective if that is all that is said, since the claim that such and such is not a difference-maker may have no more credibility or justification than the claim that it is. I do not say that species, gender or race do make a moral difference, only that the argument had better be better than a dogmatic assertion that they do not, coupled with an convenient assumption of a favorable onus of proof. Similarly, the skeptic dogmatically asserts that something is not an epistemic difference maker.

The differences between justification and knowledge are worth exploring in this context. Consider two cases where we have similar experiences, where one case is an ordinary case and the other case is one in which some skeptical hypothesis obtains. The situations seem to be similar with respect to those factors that are relevant to determining justification or rational belief; in both cases our beliefs are justified, and we believe rationally despite the falsity of beliefs in one case but not the other. Thus the comparison case scenario has no skeptical consequences for justification. What about knowledge? We can can say that there is a significant difference between the two cases, and it is a difference in what knowledge depends on. The difference is not a difference in experiences but in some other fact. In virtue of that fact, we know in one case but not in the other. It is difficult to see any problem here. This is a dogmatic assertion, it is true. But then all the skeptic does is dogmatically to deny it, and then we have a stand-off. And it is the skeptic who should really be trying to avoid being dogmatic.

Once again, dogmatism is our fate. Scepticism of this kind is implausible and indeed incoherent, because one can only be skeptical at one level by being dogmatic at the level above.

§5. Possibility Arguments

Consider this familiar argument. (1) Knowledge of the external world depends on sense experiences. (2) There is a possibility of error in sense-experience. Therefore (3) we do not know about the external world. One wonders how this argument can ever have been thought to be so strong. Thousands of articles engage with it. But why? Yes, there is the possibility that there is a sense experience with content p together with p; and there is also a possibility of a sense experience with content p together with ~p. But so what? It is difficult to see why the existence of these possibilities has skeptical consequences. The skeptical thought would have to rest on the dogmatic assumption that the existence of one or both of these possibilities is a maker of ~K.

This is already dogmatic, as we have seen. But it is worth fully appreciating how implausible this dogmatic skeptical view is. The question is: why should either a possibility or a conjunction of possibilities be the makers of the actuality of ~K? And why should the lack of a possibility or conjunction of two possibilities be a maker of the actuality of K?

It seems that there are considerations that count against possibilities counting. If we know or don't know, this is a matter of actuality: we actually know or don't know. So what can possibilities have to do with it? Compare truth and goodness. The makers of both these properties are not mere possibilities! They are actual facts. Surely knowledge too.

Suppose for example that a dogmatist says that actual knowledge holds in virtue of the existence of actual causal relations between beliefs and facts. Then it would not be relevant that it is possible that someone believes p or has an experience as of p when ~p. If the causal relation actually obtains then that makes for knowledge, and the possibility is irrelevant.

Quite generally, let us ask: what have possibilities to do with actuality? Suppose I am a human being. Might I have been a chimpanzee, or a tortoise? How would these possibilities be relevant to the question of whether I am actually a human being? This gets things the wrong way round. The modal claims beg the question about actuality. If I am a human being, then I could not have been a chimpanzee or a tortoise; actually being human explains the impossibilities. Actuality claims of this sort are prior to the possibility claims. If so, we cannot argue from claims about what is possible to claims about what we know or do not know (actually). Take a different kind of case. My wall is actually blue. But it might have been yellow. So what? How does that bear on its being blue? The possibility of being yellow lacks any connection with actuality. Many cases in epistemology are like this. There are certain skeptical possibilities. But it is completely unclear what they have to do with the actual instantiation of epistemic properties. It is not question-begging to assert skeptical possibilities (as in the human/tortoise case). Instead the possibilities are irrelevant to how things actually are. Yes, if I were a dreaming, a victim of the evil demon, or a brain in a vat, then I would not know what I actually do. But so what? If my wall were yellow it would not be blue. But it is blue. Why does the existence of such possibilities have implications for whether I actually know or not?

It is difficult to see how mere possibilities could be relevant to actual knowledge or ignorance. But then we might wonder why they have played such an important role in recent epistemology. This seems mysterious. One diagnostic suggestion might be that knowledge is the product of an ability or capacity, which is an actual fact about a person. Where we have knowledge or justified belief, we have the ability to get it right and not to get it wrong. (Abilities or capacities are not merely dispositions.) If I know London, then if I am asked directions then it is no accident that I get it right. Now it is evidence that one has the ability to get it right that we generally believe that p when p and we generally do not believe that p when ~p. And it is evidence of lacking the ability to get it right that this is not generally the case. There need not hold universally. Good jugglers may occasionally drop a ball even though they have the ability to juggle, and bad golfers can by fluke hit a hole in one, although they lack the ability to hit a hole in one. But even in these cases, the actual presence or absence of ability explains the generalities. Similarly, actual knowledge explains our generally being right and generally not being wrong. If so, the ability suggestion does not really explain the emphasis on possibilities.

In the case of justification it might be conceded that possibilities themselves are not relevant to the actual instantiation of the property of epistemic justification or to the failure actually to instantiate the property of epistemic justification. Instead, it might be said that possibilities are relevant in that the mental act of ruling out possibilities is a source of justification. The idea would be that ruling out the possibility of not-p is a justification-maker for believing that p. There is a worrying circularity in this idea, since we must surely justifiably rule out possibilities. More importantly, the mental fact of ruling out possibilities in an actual fact about a person. The maker of justification in this case is not a possibility but an actual belief about a possibility arrived at by an actual mental act.

I am, thus, empty-handed; I have no convincing explanation of what seems bizarre – that possibilities have been taken to be so important in epistemology, almost defining the genre for two generations. But, then, perhaps the turn to dependence in epistemology explains why epistemology, modally conceived, often resembled a wild goose-chase. A thing, property or fact is one thing, and necessary or sufficient conditions for it are another. If you want to show that people do or do not know something, or are or are not justified in believing something, you had better appeal to some actual fact about them. Claiming possibilities gets you nowhere.

§6. Coda

I confess that I no longer feel the pull of scepticism in the way that I did for much of my philosophical life. Once we prioritize epistemic dependence and cast scepticism in those terms, rather than in modal terms, it seems obvious that skeptical arguments are no good. The crucial point is that skeptical arguments necessarily make dogmatic assumptions about what makes for knowledge or justification, or about what makes for lack of knowledge or justification. A dogmatist asserts an epistemic dependence relation at some level or other – that something makes for knowledge or justification. But the skeptical assertion that something makes for the lack of knowledge or the lack of justification is then asserted dogmatically by the skeptic, and asserting that is as dogmatic as a non-skeptical assertion. Even if it is then said that both positive and negative epistemic views are not justified then we have a negative meta-epistemic dogmatic assertion. Wherever we draw the line, epistemic making relations are assumed. Both skeptical assertions and skeptical arguments assume such dependence claims. Dogmatism is our fate, and any attempt to fight dogmatism, like a fly struggling in a spider's web, only reinforces that fate.