A Counterexample to Evidentialism?
I understand evidentialism to be the view according to which a rational agent should believe what is supported by her total evidence (at least, this is the kind defended by Conee and Feldman[1]). More precisely:
- Evidentialism: Doxastic attitude D is justified for S at t iff D is supported by S's total evidence at t.
Evidentialism is a tidy view. It says that epistemic justification is a matter of having the right sort of evidence. I like it. Here's a possible counterexample (you didn't see that coming, did you!?).
- Evidence Boxes: You are wondering whether P is true. An epistemic authority wheels a cart of ten boxes in front of you. She tells you that inside each box is a piece of evidence for P. In fact, all available evidence about P at the time is contained in the ten boxes. You're told that the evidence isn't difficult to evaluate. In a matter of seconds you'll be able to tell how the evidence in each box bears on P, meaning that it won't take long at all to get through each of the ten boxes. Since you have nothing better to do, and you'd really like to figure out whether P is the case or not, you commit yourself to looking through each of the ten boxes. Moreover, since it won't take long to evaluate the contents in the boxes, and it's all in front of you, you decide to suspend your judgment about P until you've looked at a few of the boxes. You're not going to believe or disbelieve P right away, even if the evidence in the first box justifies one of those attitudes. Once a good number of the boxes are opened, you'll make a judgment about P. For example, you might decide to believe P if, say, the majority of the boxes support P really well, even though you still have a couple boxes left to open. After all, you reason, if there were a box that had extremely strong evidence against P, you probably would have opened it by now. Whatever turns out to be the case, this is your policy: suspend judgment about P until a good number of the ten boxes have been opened.
Intuitively, your policy is rationally permitted. It's okay (in the epistemic sense of "okay") to suspend belief until you've opened a few of the boxes. Here's what I'm claiming:
- Joel's Intuition (JI): in the case of Evidence Boxes, it is rationally permissible to suspend judgment about P until you've opened a few of the evidence boxes.
JI seems incompatible with evidentialism. Why? Imagine you open the first box and get good evidence for P. At that moment, your total evidence supports belief in P. Evidentialism says that you are justified in believing P. But it's not merely that you are permitted to believe that P; rather, according to Conee-and-Feldman-style evidentialism, you are epistemically obligated to believe that P. Failure to believe that P after getting the evidence in the first box is irrational. But doesn't that seem wrong? Reread the above case. Doesn't it seem like you're rationally permitted to have a doxastic attitude (suspension of belief) that isn't supported by your evidence? That is, it seems rational to suspend judgment about P until you've looked at a few more of the boxes. Of course, eventually you should believe what is supported by your evidence, but for the first few bits of evidence, it seems permissible to suspend judgment.
As is to be expected, there are possible responses.
Response1: for the vast majority of your beliefs, you could always get more evidence. There is always easy information available which bears on those beliefs and which you could easily get. Moreover, the proportion of evidence you possess for most of your contingent beliefs is rarely greater than one half of all the evidence their is for those beliefs. Nonetheless, at least some of your beliefs about which this is true are justified. So you don't need to wait until you've looked at most of the evidence before you're justified in believing.
Me: my claim is not that you should exhaust all possible evidence available to you before believing or disbelieving. If that were the case, we would have very few justified beliefs. But we do have many justified beliefs for which we could always get more information. I grant that. For example, suppose there were 10,000 evidence boxes relevant to P and I only opened 100 that were randomly selected, the majority of which seem to support P. Intuitively, I am justified in revising my credence to match what is supported by the evidence. It seems irrelevant that I have only looked at a small proportion of the total available boxes. I don't think you need to have the majority of the evidence for P before you can make an epistemic judgment about P. So I don't think the false rule that you have to examine most of the evidence for P before you're rational in believing/disbelieving P explains my intuition. The My intuition is merely that, in special circumstances (like the one in Evidence Boxes) it's okay to suspend your judgment until you get a few more pieces of evidence. So I'm not saying you ought to suspend your judgment until you've looked at a good bit of the evidence. I'm just saying that, in cases like the one above, it's permissible to suspend your judgment about P. Maybe it would be helpful to identify what the salient features of the above case are.
Response2: That means you're a permissivist (and if I know you, Joel, I know you're not very sympathetic to permissivism).
Me: No, the case need not support permissivism. It might be argued that suspension of belief is uniquely rational for the agent, given her epistemic state. For example, responsiblist evidentialists might argue that the agent has failed the alleged responsibility condition on justification, since a responsible agent would wait to form a judgment about P until she's looked at more evidence (when it's highly convenient for her to do so and when she's barely looked looked through much of this easy-access evidence). Therefore, to believe or doubt P would be irrational. She should suspend judgment until she's looked at a few more boxes (personally, I don't take this route). Moreover, even if one thinks the case supports permissivism about rationality/justification (the agent could rationally suspend judgment or believe P), it doesn't seem to support permissivism about evidential support (and that sort of permissivism seems pretty off to me).
Response3: Maybe what's going on here is this. Think of epistemic rationality as goal-oriented. A doxastic attitude or practice is rational if it is a good means for accomplishing an epistemic goal. Relative to some epistemic goal of yours, suspending judgment about P is rational, even though your evidence supports P after the first box is opened. Perhaps your goal is to examine a good deal of the evidence before making a judgment, or maybe its to minimize the risk of being wrong when you know you're going to get more evidence easily and quickly from which to make a less risky judgment. Relative to either of these goals, your decision to suspend judgment is rational. Evidentialism, on the other hand, is a good policy for accomplishing an entirely different epistemic goal (it's hard to say exactly what this goal is). Hence, evidentialism and the norm captured by JI are different kinds of norms because they are procedures for accomplishing different epistemic goals. If so, why should it be a problem for evidentialism that it is not a good policy for accomplishing either of the above two goals? After all, it was never proposed as a standard for accomplishing those goals. Again, it is a means for accomplishing a different goal. Hence, one can be an evidentialist while still recognizing that suspension of belief in Evidence Boxes is rationally permissible relative to a different epistemic goal. Different epistemic responses are permissible because different epistemic goals can come up in the same epistemic situation. In sum, you have various epistemic goals and, sometimes, the norms determined by those goals make conflicting prescriptions/prohibitions. So what? Lastly, if you're unwilling to concede that epistemic rationality is reducible to the goal-oriented rationality described above, that's okay. Notice that the norm described by JI is a diachronic norm. It tells you how to rationally behave over time. Evidentialism is a synchronic norm. It tells you what you ought to do (or are permitted to do) at a given time. Why should it be so problematic that different kinds of norms conflict?
Me: I'm tired of writing. Plus, this blog is already getting too long. To be continued..
Response: Okay, fine. By the way, your blog is the best.
Me: I know. Thank you. [2]
Response1: for the vast majority of your beliefs, you could always get more evidence. There is always easy information available which bears on those beliefs and which you could easily get. Moreover, the proportion of evidence you possess for most of your contingent beliefs is rarely greater than one half of all the evidence their is for those beliefs. Nonetheless, at least some of your beliefs about which this is true are justified. So you don't need to wait until you've looked at most of the evidence before you're justified in believing.
Me: my claim is not that you should exhaust all possible evidence available to you before believing or disbelieving. If that were the case, we would have very few justified beliefs. But we do have many justified beliefs for which we could always get more information. I grant that. For example, suppose there were 10,000 evidence boxes relevant to P and I only opened 100 that were randomly selected, the majority of which seem to support P. Intuitively, I am justified in revising my credence to match what is supported by the evidence. It seems irrelevant that I have only looked at a small proportion of the total available boxes. I don't think you need to have the majority of the evidence for P before you can make an epistemic judgment about P. So I don't think the false rule that you have to examine most of the evidence for P before you're rational in believing/disbelieving P explains my intuition. The My intuition is merely that, in special circumstances (like the one in Evidence Boxes) it's okay to suspend your judgment until you get a few more pieces of evidence. So I'm not saying you ought to suspend your judgment until you've looked at a good bit of the evidence. I'm just saying that, in cases like the one above, it's permissible to suspend your judgment about P. Maybe it would be helpful to identify what the salient features of the above case are.
Response2: That means you're a permissivist (and if I know you, Joel, I know you're not very sympathetic to permissivism).
Me: No, the case need not support permissivism. It might be argued that suspension of belief is uniquely rational for the agent, given her epistemic state. For example, responsiblist evidentialists might argue that the agent has failed the alleged responsibility condition on justification, since a responsible agent would wait to form a judgment about P until she's looked at more evidence (when it's highly convenient for her to do so and when she's barely looked looked through much of this easy-access evidence). Therefore, to believe or doubt P would be irrational. She should suspend judgment until she's looked at a few more boxes (personally, I don't take this route). Moreover, even if one thinks the case supports permissivism about rationality/justification (the agent could rationally suspend judgment or believe P), it doesn't seem to support permissivism about evidential support (and that sort of permissivism seems pretty off to me).
Response3: Maybe what's going on here is this. Think of epistemic rationality as goal-oriented. A doxastic attitude or practice is rational if it is a good means for accomplishing an epistemic goal. Relative to some epistemic goal of yours, suspending judgment about P is rational, even though your evidence supports P after the first box is opened. Perhaps your goal is to examine a good deal of the evidence before making a judgment, or maybe its to minimize the risk of being wrong when you know you're going to get more evidence easily and quickly from which to make a less risky judgment. Relative to either of these goals, your decision to suspend judgment is rational. Evidentialism, on the other hand, is a good policy for accomplishing an entirely different epistemic goal (it's hard to say exactly what this goal is). Hence, evidentialism and the norm captured by JI are different kinds of norms because they are procedures for accomplishing different epistemic goals. If so, why should it be a problem for evidentialism that it is not a good policy for accomplishing either of the above two goals? After all, it was never proposed as a standard for accomplishing those goals. Again, it is a means for accomplishing a different goal. Hence, one can be an evidentialist while still recognizing that suspension of belief in Evidence Boxes is rationally permissible relative to a different epistemic goal. Different epistemic responses are permissible because different epistemic goals can come up in the same epistemic situation. In sum, you have various epistemic goals and, sometimes, the norms determined by those goals make conflicting prescriptions/prohibitions. So what? Lastly, if you're unwilling to concede that epistemic rationality is reducible to the goal-oriented rationality described above, that's okay. Notice that the norm described by JI is a diachronic norm. It tells you how to rationally behave over time. Evidentialism is a synchronic norm. It tells you what you ought to do (or are permitted to do) at a given time. Why should it be so problematic that different kinds of norms conflict?
Me: I'm tired of writing. Plus, this blog is already getting too long. To be continued..
Response: Okay, fine. By the way, your blog is the best.
Me: I know. Thank you. [2]
____________________________________________________________
Footnotes:[1] Conee and Feldman, "Evidentialism" (1985).
[2] H/T to Nate Lauffer, Bill Vincent, Mike Willenborg, and Tim McGrew for helpful comments related to the content of this blog.
Comments
Post a Comment