External world skepticism is usually set up by first presenting two scenarios that are indistinguishable from each other. One is the good ‘real world’ scenario that we believe ourselves to be in. The other is the bad ‘skeptical’ scenario. It could be an evil demon deceiving us, or it could be that we are a brain in a vat hooked up to a detailed computer simulation. Whatever the case might be, the point of the setup is that nothing can count as evidence that we are in one rather than the other scenario, while at the same time there is a massive difference between the scenarios in terms of how good are beliefs are. If we are in the real world, we know a lot of truths. If we are brains in vats, most of our beliefs will be false.

It is important to notice that the mere presentation of these two scenarios is not enough to force us into skepticism. After all, if we are not brains in vats, then it seems we are the lucky people who know a lot. Skepticism gets off the ground only once we have managed to convince ourselves that:

1. We do not know that we are not brains in vats.
2. If we do not know that we are not brains in vats, then we don’t know any of the familiar facts of daily life either.

Many contemporary responses to skepticism attack #2 and the closure principle that is usually used to defend it, but let us focus on the more fundamental #1. Clearly, if we are brains in vats, then we do not know that we are not brains in vats — knowledge has to be true. But what if we are normal people in a real world? What then would make it the case that we do not know that we are in a real world? After all, our belief is true.

But truth is not the only standard to which our thought has to answer before it can count as knowledge. Usually, knowledge is also supposed to be justified. So if our true belief that we are not brains in vats fails to be knowledge, this failure is due to a failure of justification. But how exactly is justification undermined by the fact that there is a skeptical scenario indistinguishable from the scenario we take ourselves to be in? The point of the skeptic cannot be that our belief might be false. Of course it might be false. My belief that there is no elephant in my house might be false. But I’m still justified in believing that there is no elephant in my house, because (1) there are very few elephants around, (2) it would be really hard to get one into my house, (3) there’s no reason to believe that anyone would attempt such a feat, and (4) I have no positive empirical reasons for believing that there is an elephant around. Of course we end up skepticism if we claim that knowledge requires infallibility; but that’s not a very interesting route to skepticism, because it is so easy to simply shrug and refuse to accept the demand for infallibility.

Reasons, then, are almost always compatible with falsity of the belief. The point of the radical sceptical scenarios must be different than pointing out fallibility. The point must be that we do not have any reasons at all for our normal belief; that given indistinguishability, nothing can count as a reason for believing that I’m in the real world rather than in a vat manipulated by brain scientists. It’s not that none of my observations can make it certain that I’m not a brain in vat; it’s that none of my observations can even count as positive evidence in favour of this hypothesis. What the skeptical scenario is meant to accomplish is a total breakdown of reasons, leaving us entirely unable to justify the belief that we are in the epistemically good situation.

It’s not obviously true that the skeptical scenario succeeds at this. But it is notable that quite a few responses to the skeptic are willing to grant this point. For instance, Wittgenstein would say that giving and asking for reasons is a practice with certain presuppositions that are not met in the skeptical case; and so the language game of justification cannot be played there. This is to agree with the skeptic that we cannot give reasons for the claim that we are not brains in vats. (Wittgenstein would of course go on to deny that this impairs our ability to give reasons for more mundane beliefs.)

What is the structure of the skeptical scenario that allows it to generate a full breakdown of reasons? It posits two things: an external standard of knowledge and an internal standard of knowledge. The external standard is sometimes called ‘truth’, but it’s better to call it ‘reality’. For thought to rise to the status of knowledge, it has to answer to reality. The internal standard is justification. For thought to rise to the status of knowledge, it has to be justified; which means that it has to answer to our standards for judgement.

Now if the two standards are to be two standards, they must be different. It must be possible to answer to one standard while failing to answer to the other: a justified belief can be false, and a true belief can be unjustified. And this is true not only for isolated cases. The skeptical scenario is specifically engineered to showcase a global misalignment of the two standards. Since one standard is purely internal and the other is purely external, there would seem to always be room for such a scenario. What things look like from the point of the view of the subject is logically independent from how they really are. And thus a skeptical wedge can always be driven between them; we can always come up with a scenario in which even the most perfect justification falls short of truth.

But if that is always possible, then we have to start wondering what the point of the internal standard is. If it might fail to have any connection with the external standard; if our reasons could be completely divorced from reality, and (on pain of begging the issue) no reasons can be given to think that they are not so divorced; then why think of the internal standard as a standard at all? Why care about reasons if there are no reasons to believe that they are a guide to reality? No answer would seem to be forthcoming. And if no answer is forthcoming, then we have a radical breakdown of reasons. We indeed know nothing; worse, it no longer seems to make sense to think or investigate.

Given this analysis, two replies to skepticism are especially interesting. The first is the externalist reply which claims that thought has only to answer to external standards. The second is what I shall can the transcendental idealist reply, which claims that thought has only to answer to internal standards.

The externalist answer to the skeptic is to say that knowledge does not depend on an internal standard. An example is reliabilism, which holds that knowledge is reliably formed true belief. If I’m not a brain in a vat, but walking around in the real world, then I’m forming my beliefs in a reliable way. And so I know that I have two hands, and presumably also that I’m not a brain in a vat. If I were a brain in a vat, then my beliefs would be both false and formed in an unreliable way, and hence they would doubly fail to be knowledge.

It is indeed hard to see how the skeptical scenarios could get any purchase on the externalist. This doesn’t mean that externalism is successful. I believe it fails to do justice to the nature of thought, for instance by making it very mysterious why we should care about having reasons at all. The externalist theory considers a person’s judgement from the outside; but judgement shows its true nature only from the inside, from the first-person perspective of the judging subject for whom the question of truth, the question of belief, and the question of justification are the same question. (There is a difference between wondering whether S believes that p, whether S‘s belief that p is true, and whether S is justified in believing that p. But there is no difference between wondering whether you believe that p, whether p is true, and whether you have good reasons for believing p. All of this is the same thing: judging whether p.)

The transcendental idealist answer to the skeptic is to say that knowledge does not depend on an external standard. Kant’s claim that thought is spontaneous is the claim that thought is autonomous, that it answers only to the standards that it sets itself. And of course if there is no external standard, then no skeptical scenario can drive a wedge between it and the internal standard.

Now abolishing the external standard may seem a total non-starter. Isn’t that tantamount to saying that truth or reality don’t matter?

It is most certainly not. To drop the idea of an external standard is emphatically not to say that if I believe something, then it is true; nor is it to say that if I am justified in believing something, then it is true. For the internal standards of thought are still standards; they have normative force, and we can fail to live up to them in all kinds of ways. Most importantly, our standards are such that they not only allow for but actively call for revision of our beliefs as new observations are done and new ideas are developed.

But that talk about observations may seem to sit uneasily with the idea that there are only internal standards of thought. Isn’t an observation something that brings us in touch with reality; and is it not clear that here something external comes to us, something to which thought then has to live up?

Yes and no. Yes, it is clear that something external comes to us. Kant tells us that thought is spontaneous; but he also tells us that thoughts without intuitions are empty; and intuition, in finite beings like ourselves, is receptive. For thought to have an object, we need the world to give us something. Without perception, our spontaneous capacity for thought would remain completely useless. So that’s the yes. But also no, because insofar as what is given is truly given, that is, external to the spontaneity of thought, it is not a standard. One could make this point the way Sellars does, which is more or less by pointing out that what is truly given ought to be non-conceptual, and what is non-conceptual cannot be related to thought as a standard of correctness. Kant makes the point in a slightly different way, arguing that if what we encounter is to be an object, it must already be structured by the internal standards of spontaneous thought — it must appear at a determinate location in space and time, be conceivable as a substance with intensive and extensive magnitudes, and stand in relations of causation and interaction with other objects. The Kantian point here is that these structures are both the internal standards of thought and the criteria for reality. To be real is to stand in causal relations to everything else; and causation is a fundamental category of the finite mind, that is, a structural element of all judgement of objects.

Henry Allison explains transcendental idealism as a rejection of the theocentric model of cognition: a rejection of the idea that God’s knowledge acts as an external standard to reality. This is precisely right. We can forget about a God’s point of view: the only standards to which finite thought has to answer are those that are internal to itself; the only criterion of reality that we have to take seriously is our criterion of reality. And this is not the case because we have lowered our standards, but because it makes no sense to apply divine standards to finite cognition. God literally doesn’t think. Whatever God might be doing is utterly incommensurable with our thought.

Okay, but what about the brain in the vat? Kant of course never discusses this idea, but I think what we might say is this: the brain can only be held to its own internal standards, and those define what reality is for the brain. So its reality is the computer simulation, and its beliefs, including the belief that it is not a brain in a vat, are simply true (and knowledge). I believe that this is also what Putnam and Davidson might say about the case, though perhaps on slightly different grounds. What then if we unhook the brain from the simulation and hook it up to a robot body, so that it can find out that it was ‘really’ a brain in a vat? Well, are we sure that the brain would understand this as awakening to the real world? Is it not more like being transported to another world? Might it not even be the case that the brain is no longer the same person? I’m far from certain that much philosophical good comes from contemplating such scenarios, but I am certain that we should not be too convinced that our interpretation of what happened (“you were first in a computer simulation and now you’re in the real world”) is binding on the brain.