1 Introduction
Imagine that you own a line of fancy transatlantic cruise ships, and business is good. One day, one of your captains alerts you to an apparent problem with one of your ships. Concerned, you have a full inspection done, which reveals that the ship is in very bad shape; in fact thereâs an even chance that it wonât survive another voyage. You therefore put it in dry dock to be repaired and congratulate yourself on your conscientiousness.
Now imagine a slightly different case. You are still the owner of a line of fancy transatlantic cruise ships; one of your captains alerts you to an apparent problem with one of your ships, and to appease your captain, you have a full inspection done. The inspection reveals that the ship is in very bad shape; in fact thereâs an even chance that it wonât survive another trip. Butâcurse your greedy soul!âyou dread the prospect of paying for the repairs and having to refund the tickets that have already been sold. So you proceed to convince yourself that the ship will most likely survive another trip, doctor the inspection results, and send the ship on its final voyage. Out on the deep and treacherous waters, it goes down, taking every unfortunate soul with it.
These cases, modelled on the shipowner case from W. K. Cliffordâs (1999/1897) well-known essay, âThe Ethics of Belief,â sharply contrast in two obvious ways. First, they differ in how responsibly you behave: in the first case, you behave responsibly in deciding to repair the ship, whereas in the second, you behave irresponsibly (indeed, you are criminally negligible). The second contrast is with respect to the belief you form in response to the evidence about the state of the ship that is provided to you in the form of the shipâs inspection. In the first case, you form the belief that is supported by this evidence. In so believing, it seems that your intellectual conduct is perfectly rational. In the second case, however, you ignore this evidence and convince yourself that the ship will likely survive another voyage. It seems that the belief in the second case is irrational; you are conducting your intellectual affairs incorrectly. Even if the ship had managed to survive the voyage, this would have been just lucky for you; your belief that it would survive the voyage was in an important sense irresponsible and irrational.
Reflecting on cases like these, we are apt to conclude that there is a sense in which beliefs count as rational and irrational, a sense that we can call âepistemic.â In the good cases, where people hold beliefs on the basis of good evidence, it seems that theyâre doing something right. In the bad cases, when people ignore good evidence, or they hold beliefs on the basis of no evidence at all (but rather, say, as a result of wishful thinking), it seems like theyâre doing something wrong, something that is under their control, which they could do better. Cases like the two shipowner cases with which we began seem to indicate that there are good reasons for belief which consist of evidence or considerations which bear on the truth value of a proposition. We can call this class of reasons consisting of evidence âepistemic reasonsâ and note that beliefs held on the basis of good epistemic reasons are epistemically rational. And in morally important situations, itâs morally important to have epistemically rational beliefs.
Considering cases like these, and defining a kind of reason and a corresponding kind of rationality as the kind that are displayed in the good cases and absent in the bad cases is a good starting point for our epistemic theorising. But it is only a starting point. Once we acknowledge that there are epistemic reasons and epistemic rationality, and we note that epistemic reasons seem to be evidential in character, a number of questions immediately present themselves. Can we explain why it is that evidence constitutes a kind of reason for belief? Is there anything interesting to say about the relation between epistemic reasons and other kinds of reasons? Do epistemic reasons always have normative force? Where does their normative force derive from? Is it ever morally permissible to act on the basis of epistemically irrational beliefs? And so on.
The aim of this book is to address two of those fundamental questions: can we explain why epistemic reasons consist of evidence? And, given that epistemic reasons consist of evidence, what is it that explains the normative force of epistemic reasons? In this book I will set out what I think is a plausible combination of views about evidence, reasons, beliefs, and normativity in attempting to answer these questions.
1.1. Elements of the View
1.1.1. Accessibilist Internalist Evidentialism
One of the central elements in this view is internalist evidentialism about epistemic rationality. Briefly, evidentialism is the view that beliefs are epistemically rational if and only if they are supported by adequate evidence.1 Internalism is the view that whatever bears on the epistemic status of a subjectâs beliefs (i.e., on their status as epistemically rational or justified) must be internal to the subjectâs cognitive perspective.
Evidentialism is usually taken as an internalist view, although not always.2 Internalist evidentialism is the view that the only considerations that bear on the epistemic status of a subjectâs beliefs are evidential, and the evidence a subject possesses is entirely determined by factors internal to the subjectâs cognitive perspective. As it is sometimes said, for internalist evidentialists, a subjectâs evidence is provided by his or her non-factive3 mental states. Externalist evidentialism, by contrast, is the view that the only considerations that bear on the epistemic status of a subjectâs beliefs are evidential, and the evidence that a subject possesses (or ought to possess) can be in whole or in part determined by factors external to the subjectâs cognitive perspective.
A standard way to illustrate the difference between internalism and externalism is by appealing to New Evil Demon kinds of cases (ânew,â because this is a variation on Descartesâ classic evil demon scenario.) We can proceed in three stages:
Constructing a New Evil Demon scenario
- Stage 1: Think about yourself here in the real world with all your thoughts and fears and experiences and so on.
- Stage 2: Now imagine a different possible world created by an evil demon who delights in fooling his creations. Just about everything his victims believe is false. They believe that they have hands, live in houses, and so on, but (poor folks) they donât even have bodies! Itâs all an illusion. Still, they donât realise any of that; they are, as far as they can tell, in a world exactly as normal as our own.
- Stage 3: Imagine that you have a non-factive mental duplicate in this other possible world, who is a victim of this evil demon. She has all of the very same beliefs and experiences, thoughts, fears, habits of mind, and so on, as you have. Her cognitive life is internally identical to your own.
We can ask: is there a difference between how well justified your beliefs are and how well justified your hapless mental duplicateâs beliefs are?
Internalists think that your mental duplicate is doing just as well as you are, as far as the epistemic rationality or justification of her beliefs is concerned. Externalists, by contrast, will typically say that her beliefs are not in fact justified; she just isnât in a position to know that theyâre not justified. There are various kinds of externalist views in the literature, but for now, all we need to note is that an externalist evidentialist would likely hold that the subject in the New Evil Demon scenario doesnât possess the same evidence as you do in the real world; she only thinks she possesses all of that evidence.
It seems to me that the demon-deceived subject is unlucky but not therefore unjustified in her beliefs. So the sort of view I am interested in exploring in this book is committed to internalism. It will hold that all factors bearing on the epistemic status of a subject Sâs beliefs are internal to her cognitive perspective.
But what does it even mean to say that something is internal to Sâs cognitive perspective? There are two slightly different ways to spell that out in the literature: accessibilism and mentalism. Mentalism is the view that anything that bears on the justificatory status of Sâs beliefs (henceforth, âjustifiersâ) must be part of Sâs mind. Beliefs, wishes, phenomenal statesâthese would all count as justifiers for the mentalist internalist. (The wish that some proposition p were true isnât generally a justifier for the belief that p, but having the wish that p were true can be a justifier for other things, such as a meta-belief to the effect that one wishes that p were true.) Facts recorded in an encyclopaedia on Sâs shelf, by contrast, are not justifiers for S unless S has also learned those facts.
Accessibilism is the view that justifiers must be accessible from Sâs cognitive perspective. Typically this is taken to mean that justifiers must be available to S just by reflection. Accessibilist internalists agree with mentalists in holding that the facts recorded in Sâs encyclopaedia are not justifiers for S because those facts are not accessible to S just by reflection. Of course those facts are accessible, in some sense: S could open the book and start reading. But this is to do more than reflect on the reasons one possesses: it is to engage in some (fairly basic) empirical inquiry and to acquire new reasons.
It is possible to be an externalist accessibilist (as in Gibbons 2006, 2013), by holding that justifiers must be accessible to S to be genuine justifiers for S but allowing that what is accessible through certain kinds of simple empirical inquiry is genuinely accessible and bears on the epistemic status of oneâs beliefs. But typically accessibilism is taken in an internalist sense. We will come back to Gibbonsâs argument for access externalism in Chapter 3.
Accessibilist internalism is stronger than mentalism when it comes to what count as justifiers: it excludes some things that mentalism might allow because there are (plausibly) elements of our minds that are not accessible to us. For example, think of memories of events which we are utterly unable to recall, just by reflecting on them, but which we are able to recall after someone else prompts us with an important hint. Accessibilists would not, whereas mentalists might, count such memories as part of our evidence before we receive the hint.
The view developed here is committed to accessibilist internalist evidentialism. Henceforth, for brevity, I will use the unqualified âevidentialismâ to refer to this specific form of evidentialism. This view has been challenged in recent decades, but in my view it is certainly plausible enough to merit further development.
The main point of this first chapter will be to clarify and illustrate the plausibility of the sort of evidentialism Iâm taking up. I do not, however, undertake to defend evidentialism from all of the challenges that have been raised against it; as long as I can indicate why I think that the view is plausible and interesting enough to merit further exploration, I will count myself happy.4 Towards the end of the chapter, I also argue that rationality and justification are equivalent, and I provide a brief preview of the chapters to come.
1.1.2. Guidance and Doxastic Voluntarism
So one central element of the view set out in this book is evidentialism. A second important element of the view is a guidance constraint on normative reasons: for R to count as a normative reason for S to Ď, S must be such that she could take Ď into account in deliberating about whether she ought to Ď. Versions of this sort of guidance constraint on normative reasons have been appealed to in defence of evidentialism, but I will argue that although a guidance constraint on genuine normative reasons is plausible, it does not yield an argument for evidentialism. That is because we are in fact able to take non-evidential considerations into account in doxastic deliberation. I present a case of doxastic deliberation meant to illustrate that possibility in Chapter 3, and I give a recipe for constructing further such cases. Then in Chapter 4 I go on to argue for a version of strong doxastic voluntarism. In my view the state that a subject S is typically in when S believes that p involves both active and passive elements. The active element is under Sâs direct voluntary control, and it is the active element that is properly subject to epistemic evaluation. If this sort of voluntarism is correct, it follows that deontological terms of appraisal can appropriately apply to doxastic states. It also follows that we are able to take non-evidential reasons into account in our doxastic deliberationsâwe are not forced to be guided only by the reasons which we take to constitute good evidence for the truth of a proposition in deliberating about whether to believe it.
1.1.3. Instrumentalism About Epistemic Normativity
Another key element of the view set out here is a kind of instrumentalism about epistemic normativity. The idea is that we should care about having epistemically rational beliefs because having epistemically rational beliefs is, in general, in the actual world and in most close possible worlds, an appropriate means to take for achieving true beliefs and avoiding false beliefs, which is a goal that we all have with respect to a good many beliefs, at least about domains that are important to us.
But this version of instrumentalism about epistemic normativity is combined here with a rejection of instrumentalism about the nature of epistemic reasons and rationality. Of course if the normativity of epistemic rationality is instrumental, in the service of getting true beliefs that we want to acquire and avoiding false beliefs we want to avoid, it is perhaps natural to think of epistemic rationality as just a kind of instrumental rationality: beliefs are epistemically rational, the idea goes, when and to the extent that holding them promotes the achievement of our truth-related goals. But, as we will see, this view faces some serious objections. For one thing, it does not seem to be able to get the extensions of epistemically rational and irrational beliefs right. For another, it generates a vicious regress when it is fully spelled out. But recognising these problems for instrumentalism about the nat...