1. INTRODUCTION
We can draw a distinction between two types of Gettier cases; those in which one's belief that p is caused, at least in part, by the fact that p, and those where one's belief is causally unconnected to the fact that p. The former type causes trouble for a popular thesis linking reasons and knowledge:
reasons-knowledge thesis: The fact that p can be S's reason for ϕ-ing iff
S knows that p
The above formulation of the thesis is taken from Hyman (Reference Hyman1999, Reference Hyman2006). The thesis is central to his claim that knowledge is best understood as a kind of ability: to know that p is to have the ability to ϕ for the reason that p. The thesis is also endorsed by Williamson (Reference Williamson2000), Hornsby (Reference Hornsby and Leist2007), and Littlejohn (forthcoming). Neta (Reference Neta2009) and Hawthorne and Stanley (Reference Hawthorne and Stanley2008) express sympathy to it, and Unger (Reference Unger1975) appeals to it to argue that nobody has a reason for doing, thinking, or feeling anything, since, he argues, nobody knows anything. I'll first sketch Hyman's motivation for the thesis, and then explain the distinction between ‘causal’ and ‘non-causal’ Gettier cases (as I'll call them). I'll then argue that causal Gettier cases undermine the thesis.Footnote 1 My interest here is not so much in the theory of reasons, as it is in a question about knowledge. Recent epistemology has, to a large extent, turned away from the project of attempting a conceptual analysis of knowledge, and towards investigation of the normative and metaphysical roles knowledge might uniquely play. To this end it has been argued that knowledge is, variously, the ‘norm’ of belief (Sutton Reference Sutton2007), assertion (Williamson Reference Williamson2000), and practical reasoning (Hawthorne and Stanley Reference Hawthorne and Stanley2008). Williamson (Reference Williamson2000) has also argued that one's evidence consists of all and only those propositions one knows to be true (‘E = K'). The purpose of this article is to argue that, whatever other roles knowledge might play, uniquely enabling one to ϕ for the reason that p is not amongst them. Contra Hyman, knowing that p is not best understood as the ability to act for the reason that p. Given this focus, my goals are circumscribed. I hope only to show that knowledge does not uniquely play this role. I do not propose a rival epistemology of acting-for-reasons. Indeed, I see no prima facie reason to expect that there exists a true biconditional linking the ability to ϕ for the reason that p and any unique epistemic state.
2. MOTIVATING THE REASONS-KNOWLEDGE THESIS
In order to motivate the necessity direction of the reasons-knowledge thesis, Hyman (Reference Hyman1999, Reference Hyman2006) asks us to consider an example of a justified true belief which falls short of knowledge. Henry's belief is Gettiered:
Henry is watching the television on a June afternoon. It is Wimbledon men's finals day, and the television shows McEnroe beating Connors; the score is two sets to love and match point to McEnroe in the third. McEnroe wins the point. Henry believes justifiably that:
1. I have just seen McEnroe win this year's Wimbledon final
and reasonably infers that:
2. McEnroe is this year's Wimbledon champion
Actually, however, the cameras at Wimbledon have ceased to function, and the television is showing a recording of last year's match. But while it does so McEnroe is in the process of repeating last year's slaughter. So Henry's belief (2) is true, and surely he is justified in believing (2). But we would hardly allow that Henry knows (2). (Reference Hyman1999: 447)
Hyman then asks us to suppose that Henry recalls that his brother placed a £100 bet on McEnroe winning, and so infers that his brother has won £100. Is the fact that McEnroe is this year's champion Henry's reason for believing that his brother has won £100? Clearly not, thinks Hyman. His reason, Hyman claims, is only that he believes that McEnroe is this year's champion. Similarly, the fact that McEnroe won cannot be Henry's reason for being happy, angry, smug, or for congratulating his brother. Hyman takes cases like this to show that p can be one's reason for ϕ-ing only if one knows that p. Whenever one bears an epistemic relation to p weaker than knowledge – justified true belief, unjustified true belief, justified false belief, unjustified false belief – one's reason for ϕ-ing is not p itself, but something else. Similar considerations motivate the other authors who subscribe to the reasons-knowledge thesis (at least those who try to motivate it, rather than just endorse it – I'm thinking in particular of Unger (Reference Unger1975) and Hornsby (Reference Hornsby and Leist2007)).Footnote 2 Hyman goes on to argue that, given the reasons-knowledge thesis, knowledge is best conceived as a kind of ability: the ability to ϕ for the reason that p.
Although it hasn't been remarked upon before, the point extends to other cases of non-knowledge-constituting justified true beliefs that don't have the kind of double-luck structure found in Gettier cases.Footnote 3 Consider the following:
Sarah holds a ticket in a 1 million ticket lottery where only one ticket will win, and every ticket has an equal chance of winning. The draw has been made, and Sarah's ticket did not win, but Sarah has not seen the results yet. On the basis of the probabilities involved, Sarah believes that her ticket is a loser.
Many epistemologists have found it natural to say that Sarah's belief that her ticket is a loser is justified when based on the probabilities involved, but not knowledge (Kyburg Reference Kyburg and Swain1970; Klein Reference Klein1985; Foley Reference Foley1993; Lewis Reference Lewis1996; Hawthorne Reference Hawthorne2004; Pritchard Reference Pritchard and Greco2007).Footnote 4 Insofar as it is, Sarah has a justified true belief that her ticket is a loser, but does not know this. Now suppose that upon forming her belief that her ticket is a loser, Sarah becomes disappointed. Is the fact that her ticket is a loser Sarah's reason for being disappointed? Surely it is not.
3. TWO TYPES OF GETTIER CASE
Hyman's McEnroe case is a Gettier case where the subject's belief that p is entirely causally unrelated to the fact that p. But not all Gettier cases have this feature. There are cases of Gettiered justified true belief where the fact that p is among the causes of the subject's belief. Goldman's (Reference Goldman1976) Fake Barns case is of this kind:
Henry is driving through the countryside looking at objects off the road. He sees what looks to him exactly like a barn. In fact it is a barn, but unbeknownst to Henry, he is in ‘fake barns’ country – an area with only a few real barns and many barn facades designed to look exactly like real barns to passing motorists. Luckily for Henry, he happens to look at one of the only real barns in the area. Henry's belief is justified and true, but not an item of knowledge.
What's the difference between the two cases? In Fake Barns, the fact that there is a barn is (along with various facts about Henry) one of the causes of Henry's belief that there is a barn. It figures in a causal explanation of why Henry believes what he does. By contrast, in Hyman's McEnroe case, the fact that McEnroe is this year's champion is not amongst the causes of Henry's belief that he is. It has no place in a causal explanation of why Henry believes what he does. Many Gettier cases have this causal-disconnect feature. Consider Chisholm's (Reference Chisholm1966) sheep in the field case. You are standing outside a field looking in. You see what looks to you exactly like a sheep, but is in fact a dog cleverly disguised as a sheep. You form the belief ‘there is a sheep in the field’. Actually there is a sheep in the field. But it is hidden from view to you behind a hill. Your belief is both justified and true, but not knowledge. Like the McEnroe case, but unlike Fake Barns, the fact that there is a sheep in the field is entirely causally unrelated to your belief that there is a sheep in the field. Gettier cases come in two varieties: causal and non-causal.Footnote 5
4. COUNTEREXAMPLES TO THE REASONS-KNOWLEDGE THESIS
I agree with Hyman that in his case the fact McEnroe is this year's champion can't be Henry's reason for ϕ-ing. But I reject the reasons-knowledge thesis because I think it is counterexampled by causal Gettier cases. In order to see how such cases cause problems for the knowledge-reasons thesis, we can adapt the McEnroe case as follows:
Henry is watching the television on a June afternoon. It is Wimbledon men's finals day, and the television shows McEnroe beating Connors; the score is two sets to love and match point to McEnroe in the third. McEnroe wins the point. Henry believes justifiably that:
1. I have just seen McEnroe win this year's Wimbledon final
and reasonably infers that:
2. McEnroe is this year's Wimbledon champion
Henry really is watching this year's final. However, due to a network error, all other TV's in the country, including the TV's in Henry's kitchen and bedroom, and the other TV in his lounge just next to the one he is watching, are showing the final from two years ago, in which Connors staged an incredible last-ditch comeback to win the match. Only the TV that Henry is watching has been spared from the error. Henry justifiably and truly believes that McEnroe is this year's Wimbledon champion, but he doesn't know this, because he could have very easily falsely believed that Connors is.Footnote 6
Here Henry's belief that McEnroe is this year's champion is caused, in part, by the fact that McEnroe is this year's champion. As before, suppose that Henry recalls that his brother placed a £100 bet on McEnroe winning, and so infers that his brother has won £100. Is the fact that McEnroe is this year's champion Henry's reason for believing that his brother has won £100? Unlike in the previous case, it seems obvious to me that it is. Similarly, were Henry to be happy, angry, or smug, or congratulate his brother, it seems clear to me that the fact that McEnroe is this year's champion would be Henry's reason for being happy, angry, or smug, or for congratulating his brother. If this is right, then, since Henry does not know that McEnroe is this year's champion, we have a counterexample to the reasons-knowledge thesis.Footnote 7 To be sure, this is a straightforward appeal to intuition. But the same is true of Hyman's argument for the reasons-knowledge thesis. And since intuition appears not to unequivocally support the thesis, we should be reluctant to accept it.
The intuitive verdict extends, I think, to many other causal Gettier cases. Consider, for example, the following variation on the Fake Barns case:
Henry is out hiking. He's lost, and the weather is turning nasty. The situation is getting serious. He sees what he believes to be a hiker's hut in the distance, and feels relieved. In fact, unbeknownst to Henry, he is in fake hiker's-hut county – an area where there are only a handful of real huts, and many hut-facades designed to look exactly like real huts to passing hikers. Henry justifiably and truly believes that the structure in the distance is a hut, but he does not know this.
Is the fact that there is a hut in the distance Henry's reason for being relieved? Intuitively, I submit, it is. If that's right, then the reasons-knowledge thesis is false.Footnote 8
5. SUMMING UP
I've argued that certain kinds of Gettier cases – those where the fact that p causes the subject's belief that p – give us cause to be reluctant in accepting the reasons-knowledge thesis. This is perhaps not surprising, since locutions of the form ‘his reason for ϕ-ing was p’ can usually be replaced salva veritate with ‘he ϕ-ed because p’. Such considerations suggest that the subject's lack of knowledge is epiphenomena in the kinds of cases that have been appealed to motivate the reasons-knowledge thesis. What was really driving our intuitions was the lack of a causal connection between the subject's belief that p and the fact that p. Does this mean that we can truthfully replace the reasons-knowledge thesis with a different biconditional linking an epistemic state with the ability to ϕ for the reason that p? Perhaps one with a causal condition? Maybe. Maybe not. There is no reason to, it seems to me, simply assume that such a true biconditional exists. But the matter ought to be investigated further. That project is not taken up here, however, for even without a rival account we have good grounds on which to reject the claim that knowledge plays the role of uniquely enabling one to ϕ for the reason that p. We should not conceive of knowledge as an ability of this kind.Footnote 9
