Introduction
The roots of skepticism are almost as deep as the roots of philosophy itself. The word “skepticism” is derived from the ancient Greek word “skeptikos” (σκεπτικός), an adjective meaning “inquiring” or “doubting”. Today, the word has come to mean a sort of extreme or corrosive doubt that denies the existence of the subject matter with which the skeptic is concerned.
If I’m a skeptic about UFOs, for example, I doubt the existence of UFOs and question any evidence that someone proposes in favor of their existence. Used in this way — “a skeptic about X” — skepticism is local or confined to a particular subject matter.
In contrast, there is also the possibility of being a global skeptic. A global skeptic does not just doubt whether a particular subject area makes sense – like the study of UFOs, say, or astrology — but doubts the existence of knowledge or even evidence more generally. The most common form of global skepticism is skepticism about knowledge of—or evidence about—the external world, the world beyond our senses. However, it is also possible to be a skeptic about other minds, or morality. In each case, the global skeptic is doubting that there can be any good evidence capable of supporting knowledge about the external world, or other minds, or moral judgments.
As we’ll see, philosophers have tended to see global skepticism as a challenge to be defeated. However, even if global skeptical arguments overplay the doubting, questioning impulse that characterizes skepticism, there is still a place for the sorts of challenges to dogmatism that skepticism represents. However, the lessons of skepticism must be applied carefully.
For example, the rise of fake news and demagoguery might suggest that the skeptical message of “Question Authority” is one that should be applied without reservation. However, the widespread public inability to understand scientific topics like global warming, genetically modified organisms, or nuclear power—not to mention the shape of the earth—suggests that it’s not blind skepticism that is called for in today’s networked society, but rather appropriate skepticism.
To explore skepticism more deeply, we’ll begin with a brief look at ancient skepticism before discussing a new form of skepticism and the challenges it presents. After considering some contemporary answers to the challenges posed by what is called “Cartesian skepticism”, we’ll look at a way of thinking of the role of skepticism that will help to appreciate when skeptical stances may be of value.
Ancient skepticism
Suppose I tell you that it’s going to be sunny this afternoon. You’re trying to decide how to spend your day so you double-check with me: how do I know what the weather is going to be like? What evidence do I have for my claim? At this point, I might respond by saying, “Well, I looked outside, and the weather is nice and sunny now.” Alternatively, I might say, “I just checked the weather app on my phone; it’s forecasting sun for the afternoon.”
You might follow up with a counter-claim: how the weather looks now isn’t a guarantee of how it will look later, or that different weather apps are more trustworthy than others. And perhaps I’ll respond with further information, or perhaps, at some point, I might not be able or willing to give you any more evidence and say that that’s all the evidence I need to support my original claim about the weather. Most of us engage in conversations like this every day and think nothing of it. But this is the basis of a process for searching for truth that was prized by the ancients. It is a process that, when understood correctly, we moderns should prize as well.
The ancient Greek Mediterranean and the Hindu and Buddhist schools of the Indian Subcontinent were two centers of culture that put a high premium on debate and argument. It’s little wonder, then, that thinkers in both cultural contexts spent a lot of time thinking about features of the general structure of argument.
For example, around the first century C.E., the philosopher Agrippa, a disciple of the Pyrrhonean school, observed general features of the structure of the way we give evidence that lead to the recognition of five “modes” or ways that evidence can be presented both for and against a claim and how a debate about that claim can run its course. Sextus Empiricus, in the Outlines of Pyrrhonism, gives this account:
According to the mode deriving from dispute, we find that undecidable dissension about the matter proposed has come about both in ordinary life and among philosophers. Because of this we are not able to choose or to rule out anything, and we end up with suspension of judgement. In the mode deriving from infinite regress, we say that what is brought forward as a source of conviction for the matter proposed itself needs another such source, which itself needs another, and so ad infinitum, so that we have no point from which to begin to establish anything, and suspension of judgement follows. In the mode deriving from relativity, as we said above, the existing object appears to be such-and-such relative to the subject judging and to the things observed together with it, but we suspend judgement on what it is like in its nature. We have the mode from hypothesis when the Dogmatists, being thrown back ad infinitum, begin from something which they do not establish but claim to assume simply and without proof in virtue of a concession. The reciprocal mode occurs when what ought to be confirmatory of the object under investigation needs to be made convincing by the object under investigation; then, being unable to take either in order to establish the other, we suspend judgement about both. (Sextus Empiricus, Outlines of Skepticism, p. 41)
In other words, there are five ways that a request for additional evidence can ultimately end: (1) Dissent, or suspension of judgment; (2) Infinite regress, where the giving of evidence goes on forever; (3) Relativism, where each person arrives at something that is “his (or her) truth”; (4) Dogmatism, where the giving of evidence stops at an assumption that isn’t questioned, but also isn’t supported by further evidence; and (5) Circular reasoning, where the evidence loops back on itself.
If we’re focusing on debates that result in accepting or rejecting a claim (for example, “it’s going to be sunny this afternoon”), we can leave out the first option, Dissent, and the third option, Relativism, neither of which result in a situation in which the giving of evidence leads to agreement that the initial claim is true (or false). We’re left with the other three options that have come to be known as the Agrippan Trilemma (tri meaning “three” and lemma meaning “to take”: three options to consider taking). When attempting to support a claim, ultimately we arrive at a situation in which we are supporting our argument on the basis of:
- Infinite regress,
- Dogmatism, or
- Circular reasoning.
When searching for truth, none of these options is appealing. According to Agrippa, it seems that the very nature of giving reasons means that we are doomed to base our arguments ultimately on very shaky foundations! This is the crux of classical skepticism: the general structure of argument makes it impossible to present an argument that is immune to criticism.
Modern skepticism: Descartes and contemporary skepticism
Whereas ancient skepticism motivated doubt on the basis of general features of argumentation and the giving of evidence, French philosopher Rene Descartes (1596-1650) derived his skeptical doubt from a different source. Descartes believed that the mind was the seat of thought and radically separate from physical matter. For Descartes, then, the only evidence about which I can be immediately certain is evidence involving my current mental states: what I am thinking, feeling, and experiencing right now, in the moment in which I am thinking, feeling, and experiencing it.
Contemporary versions of Descartes’s skeptical argument generally involve a consideration of the plight of the brain-in-a-vat. Imagine that an evil scientist has kidnapped you, removed your brain, keeping it alive in a vat of nutrient solution, and hooked your brain to a sophisticated computer. The computer feeds your brain information so that you believe you are still a person, with a body, moving around in the world — all of this despite the fact that you are now a disembodied brain, floating in a vat of nutrient solution in a lab somewhere! This may remind you of the popular movie The Matrix. It’s the same idea.
Here’s an illustration. Let’s consider some claim about the external world, say the claim that I have hands. We can see that the skepticism motivated by the brain-in-a-vat scenario is global skepticism by considering the following argument:
- If I have good evidence that I have hands, then I also have good evidence that I am not a (handless) brain-in-a-vat. [Assume]
- All of the experiences that I now have are indistinguishable from experiences that I would have if I were a brain-in-a-vat. [Assume]
- If all of the experiences that I now have are indistinguishable from experiences that I would have if I were a brain-in-a-vat, then I do not have good evidence that I am not a brain-in-a-vat. [Assume]
- Thus, I do not have good evidence that I am not a brain-in-a-vat. [From 2 & 3]
- Thus, I do not have good evidence that I have hands. [From 1 & 4]
Since the claim that I have hands was just a representative claim and could instead have been any other claim about the world outside of my immediate experience, the skepticism prompted by the brain-in-a-vat scenario is very far-reaching indeed!
Contemporary answers to global skepticism
The brain-in-a-vat argument can help us consider the various responses to skepticism that have been proposed in contemporary discussions. The argument contains three assumptions—premises 1, 2, and 3 above—and the various responses each involve rejections of one or more of those assumptions. While there are contemporary responses that focus on each of the premises, we’ll concentrate on premises 1 and 3, since it’s those two that will help us appreciate where those of us who are unable to take comfort in Cartesian certainty might still find value in skepticism. Let’s take them in order.
I can have evidence that I have hands without having evidence that I’m not a brain-in-a-vat
One of the most radical solutions to the skeptical argument has been suggested, in different forms, by philosophers such as Fred Dretske and Robert Nozick. According to them, the problem with the skeptical argument is the first premise, and the principle of deductive closure that underlies it.[1]
What is deductive closure? Suppose you have good evidence for the claim that p. Suppose furthermore that you have good evidence for the claim that, if p, then q. The principle of deductive closure allows you to conclude that you therefore have good evidence for the claim that q.
Applied to the brain-in-a-vat argument, it’s deductive closure that provides support for the first premise. I have good evidence that, if I have hands, then I’m not handless—including that I’m not a handless brain-in-a-vat. So I have good evidence that, if I have hands, I’m not a (handless) brain-in-a-vat.
Dretske and Nozick argue, however, that deductive closure only holds within a certain context of evidence. If I’m considering evidence about whether I have hands, then that evidence will include what I see while I’m typing on the computer, say, or what I feel when my fingers touch the keyboard, or what I hear when my fingertips hit the keys. All of which, in circumstances when I’m trying to think about normal, everyday physical objects in my vicinity, counts as excellent evidence that I have hands.
But there’s a problem. Dretske and Nozick accept premise two: I would have all of the same experiential evidence about the normal, everyday physical objects in my vicinity if I were a brain-in-a-vat that I do right now. For them, as soon as realize this I must accept that the consideration of the brain-in-a-vat scenario means that none of the evidence about normal, everyday physical objects in my vicinity counts in the same way.
What Nozick and Dretske argue is that deductive closure no longer holds because the standards for the evaluation of evidence have changed between the normal scenario in which I’m considering whether I have hands—the scenario in which hands are everyday, physical objects interacting with other everyday, physical objects—and the extraordinary scenario in which I’m considering whether I am a brain-in-a-vat—a scenario in which experiential evidence no longer has the same evidentiary weight.
A more everyday analogy might motivate the rejection of deductive closure. When I arrived at work this morning, I parked my car on a parking deck some distance from my office—far enough away that I cannot now see 
my car, or hear my car alarm, etc. I have good reason for thinking that my car is where I parked it, on the top floor of the parking deck.
Unbeknownst to me, the parking deck might have been built with a structural flaw, so that while I was at work this morning, the parking deck collapsed, along with my car. (Luckily, nobody was injured when the deck collapsed.) Of course, if I have good reason for thinking that my car is where I parked it, on the top floor of the parking deck, then I also have good reason for thinking that the parking deck has not collapsed, along with my car. But I’m no civil engineer and I’ve not done an evaluation of the structural soundness of the parking structure at my work! I have no good reason for thinking that the parking deck has not collapsed—I have no evidence about the structural soundness of the parking structure at all.
Back from the brink: Contextualism and deductive closure
I said that the denial of deductive closure was a radical solution, and it is. Without deductive closure, then there is no guarantee, even if we have a deductively valid argument, and even if we have strong evidence for each of the premises, that we are evidentially supported in drawing the conclusion of that argument. And this would be a very radical claim indeed!
One fairly recent attempt to draw on the lesson of Dretske and Nozick without having to give up on deductive closure is contextualism. According to the contextualist solution to the skeptical argument, Dretske and Nozick were right to draw our attention to the extraordinary nature of skeptical scenarios. Their mistake, according to the contextualist, is to interpret the consequences of the extraordinariness of those scenarios as affecting deductive closure. Rather, the contextualists suggest that we take the consequences of the extraordinariness of skeptical scenarios to affect the functioning of the skeptical argument as a whole.
Here is how they suggest that we do this. Consider the first premise again:
- If I have good evidence that I have hands, then I also have good evidence that I am not a (handless) brain-in-a-vat.
This premise can be the basis of an anti-skeptical argument, as follows: AS1. If I have good evidence that I have hands, then I also have good evidence that I am not a (handless) brain-in-a-vat. AS2. I have good evidence that I have hands. AS3. Thus, I have good evidence that I am not a (handless) brain-in-a-vat.
Or it can be the basis of a pro-skeptical argument, as follows: PS1. If I have good evidence that I have hands, then I also have good evidence that I am not a (handless) brain-in-a-vat. PS2. I don’t have good evidence that I am not a (handless) brain-in-a-vat. PS3. Thus, I don’t have good evidence that I have hands.
Now, the contextualist says that in the ordinary, everyday context in which you are sitting at your computer, watching your fingers dance across the keyboard, feeling the keys under your fingertips and hearing the clicking as you type, your evidence is such that the argument from AS1 – AS3 is true of you. It is important to note, though, that in those contexts you are not even considering the possibility that you might be a brain-in-a-vat!
As soon as you even consider the possibility that you might be a brain-in-a-vat, the contextualist suggests, you now change the context of evaluation of your evidence. And, according to the new context of evaluation, the argument from PS1 – PS3 is now true of you. So, as soon as you consider a skeptical scenario, the skeptical argument wins. To paraphrase a line from the eighties movie War Games, the only way to win against the brain-in-a-vat skeptic is not to play.
In this way, the contextualist solution recalls David Hume’s observations about the pull of philosophical, skeptical worries, and how those worries seem to have no force when engaged in everyday activities:
[everyday activity] cures me of this philosophical melancholy and delirium [of skepticism], either by relaxing this bent of mind, or by some avocation and lively impression of my senses, which obliterates all these chimeras. I dine, I play a game of back-gammon, and I am merry with my friends; and when after three or four hours’ amusement, I wou’d return to these speculations, they appear so cold, and strain’d, and ridiculous, that I cannot find it in my heart to enter into them any farther.
Not all evidence is experiential evidence
So here’s where we are. We have one solution–Nozick and Dretske’s–that requires us to deny deductive closure. This led us to a second solution, the contextualist one. The problem with this one, however, is that it is not a rejection of skepticism, but an abject surrender to skepticism. As soon as the skeptic enters into debate with us, it’s the skeptic who wins.
There is a further way of dealing with the skeptical argument that actually rejects skepticism. This way is to block the brain-in-a-vat argument by rejecting premise three and denying that the only evidence I can have is the sort of experiential evidence that is available to the brain-in-a-vat. For example, one of the contemporary theories of knowledge is externalism, according to which at least some of the evidence that supports our beliefs is external to our minds. There are different forms of externalism, but they all share the idea that people could differ with respect to the quality of the evidence that they have for their beliefs, in ways that are unknown to them.
Here’s one way that this could work that is relevant to the brain-in-a-vat case. Consider people who suffer from Capgras Delusion — the delusional belief that familiar, significant others, like family, friends, and loved ones, have been replaced by cleverly disguised robots or aliens. A standard explanation for the cause of Capgras Delusion is that, though the facial recognition system of those who suffer from the delusion is undamaged, their emotional system is damaged so that they don’t have the expected emotional feedback when they see the familiar faces of their loved ones. It is this that causes them to believe that the faces that they see are imposters.
An externalist would attempt to solve this by claiming that those suffering from the delusion have all of the same evidence as a person not suffering from the delusion—the visual evidence of seeing the familiar face—despite the fact that they are not reliably forming the appropriate beliefs on the basis of that evidence; namely, that the familiar face that they see is the face of their loved one. What the externalist could say in such a case is that the experiential evidence alone is not sufficient to account for what constitutes good evidence: your face recognition system also has to be reliably integrated with your emotional system in order to form the appropriate beliefs on the basis of that experiential evidence.
Whether your face recognition system is reliably integrated in the appropriate way, however, is not something of which you can be aware. Those who suffer from Capgras Delusion are not aware that anything is wrong with them. Rather, they think there is something wrong with their loved ones: they have all been replaced by robots or aliens.
Note that the externalist provides us with a solution to ancient skepticism by establishing that there can be forms either of infinite regress or of dogmatism that are in fact virtuous because they depend on processes that reliably form true beliefs on the basis of evidence despite the fact that the parties to the dispute are unaware of the reliability of those processes.
For example, suppose that my spouse, knowing that I use whatever weather app is on the first screen that I see when I unlock my phone, has researched which weather app is most reliable, and, without telling me, has put that one on the first screen that I see when I unlock my phone. When I tell you that it’s going to be sunny this afternoon, and I do so on the basis of the weather app on my phone, I am in fact relaying reliable information to you–indeed, information that my spouse has curated on the basis of its reliability–although I am unaware of any of those facts myself. I’m giving you reliable and trustworthy information about the weather but I don’t know that I am.
In a case like this, if I am dogmatic in relying on the weather app on my phone, that strikes me as a different sort of case than that of someone who dogmatically relies on unreliable evidence. It’s different not because of the psychology of the dogmatists themselves, but because of the quality of the evidence that they’re relying on.
Tying the strands together
Although externalism provides a genuine rejection of skepticism, it doesn’t do a very good job of explaining why skepticism exerts any pull on our imagination at all. But, as Hume’s description of skepticism as “delirium” suggests, skepticism does exert such a pull—at least at times. This is where the lesson from contextualism comes in. What the contextualist gets right is that the skeptic attempts to raise the standards of evidence so high that no normal response will be satisfactory. What contextualism gets wrong, however, is in abandoning the field to the skeptic.
On the contextualist’s account, the skeptic doesn’t just have the last word in every debate, but deserves to have the last word in every debate. This, however, doesn’t seem right either. Although the contextualist explains why the skeptic’s challenge seems compelling, she doesn’t explain why arguing with the skeptic is so frustrating. If the skeptic deservedly wins every debate, then presumably we shouldn’t be frustrated when she does.
The problem with the contextualist is that she treats the raising of standards of evidence by the skeptic as a legitimate move, and one that the skeptic’s interlocutor has to accept. In other words, the contextualist treats skepticism as a serious contribution to discussion–even if it’s to a different discussion than the one that we might have thought we were having.
What we need, however, is a way to determine what forms of skepticism are appropriate and in what contexts. It’s at this point that we can apply what we’ve learned from the discussion of Nozick and Dretske. My evidence that my car is parked on the top floor of the parking structure at my work is based on the reliability of my memory and not on me being a civil engineer with information about the structural integrity of the parking structure. What we should take from this is not that everyone who parks in parking structures should study civil engineering and gather data on the structural integrity of the places they park. Rather, it’s that even our everyday knowledge depends on the world around us being reliable in ways that we normally don’t even consider.
Furthermore, this suggests that the value of skepticism doesn’t always consist in each person’s being skeptical all of the time. The result of that would be paralysis, not more knowledge.
Rather, we should seek to try to ensure that each, more circumscribed community of knowledge is one in which skepticism has a place. I can trust the parking structure because if the civil engineers who serve as inspectors in my town notice structural deficiencies in the parking garage, they’ll have the power to shut it down until they’re confident it’s been fixed.
Here’s a great way to think about which types of considerations it pays to be skeptical about. If you’re faced with a claim, ask yourself how much money you would bet on the truth of that claim. In effect, I’m already doing that when I park my car on the top floor of the parking structure at work: I’m betting at least the replacement value of the car that the parking structure is of good structural integrity. Then, to see when skepticism is warranted, ask yourself what sort of information it would take for you to change your bet. For example, if I learned that code inspection in my town was very shoddy and the inspectors are poorly trained, would I still be willing to bet the value of my car in parking there, or would I choose to park somewhere else.
It can also help to see how people who actually have money on the line with respect to a certain claim behave–not how they talk, but what they actually do. To take one example, I don’t know what sorts of opinions Exxon investors express about the reality of global climate change, but I do know that they are pushing Exxon to do more to plan for the effects of climate change.
The more precise you can be in formulating your claims, and the more honest you can be with yourself about the actual cash value you would wager on outcomes, the better this strategy can be. For more concrete tips, a great resource is the book Superforecasting by University of Pennsylvania professor Philip E. Tetlock and Dan Gardner.
[1] For discussion, see https://plato.stanford.edu/entries/closure-epistemic/
About the Author
Joseph H. Shieber is Associate Professor of Philosophy at Lafayette College. Prior to teaching at Lafayette, he taught at Brown University, the Massachusetts Institute of Technology, and Connecticut College. He is the author of numerous articles on epistemology, the philosophy of language, and the history of philosophy, and of the book Testimony: A Philosophical Introduction.
By the Author
