If something like the modified version of the Traditional Analysis of Knowledge proposed in Chapter 3 is correct, then justification is a crucial necessary condition for knowledge. Furthermore, justification is an interesting and puzzling concept in its own right. It will be the focus of this chapter and the one that follows. The present chapter will cover a traditional, and still widely accepted, account of justification. The next chapter will introduce some rather different, and more recent, accounts of justification (and of knowledge).
To help focus clearly on the central questions, it will be best to use an example in which two people believe the same thing, but one is justified in that belief and the other is not.
Example 4.1: Thievery
Someone has broken into Art's house and stolen a valuable painting. Officer Careful investigates the case and comes up with conclusive evidence that Filcher committed the crime. Careful finds the painting in Filcher's possession, finds Filcher's fingerprints at the scene of the crime, and so on. Careful comes to believe:
1. Filcher stole the painting.
Meanwhile, Hasty also hears about the theft. Hasty happens to live next door to Filcher and has had some unpleasant dealings with him. Hasty dislikes Filcher intensely and blames him for many bad things that happen. Hasty has some vague idea that Filcher works in the art business but has no specific knowledge about what he does. With nothing more to go on, Hasty also believes (1).
The Standard View holds that in Example 4.1 Careful is fully justified in believing (1) but Hasty is not. If you need to add more to the story to convince your-self of those evaluations, you can make those additions. However, the example should be fairly persuasive as it is.
The goal of the present chapter is to spell out in a systematic and useful way what distinguishes Careful's belief from Hasty's and, more generally, to identify the general features that distinguish justified beliefs from unjustified beliefs. There are lots of differences between Careful's belief and Hasty's belief that are irrelevant to this project. For example, Hasty's belief is about his neighbor, but Careful's belief is not about his (Careful's) neighbor. That is true, but it is not what makes one justified and the other not. Neighborly beliefs can be justified, and it is not even remotely plausible that this accounts for the difference in justification. In general, nothing about the subject matter of the belief alone is likely to be of value in answering the question, because people can have justified beliefs, as well as unjustified beliefs, about nearly any topic. What, then, is the difference?1
In thinking about this question it will be useful to keep in mind the following idea. Whether a belief is justified or unjustified, its epistemic status is an evaluative fact about the belief. Reflection on this suggests that epistemic status must depend on other, nonepistemic facts. It might be easier to understand the idea by first considering an analogy. Suppose a professor returns a set of graded papers to the students in her class. She says of one paper that it is excellent and gives it a very high grade. She says another one is a poor paper and she gives it a low grade. The teacher thus ascribes certain evaluative properties to these papers. These are properties concerning how good the papers are. (Although it is not crucial for the discussion that follows, assume that there is an objective truth about the quality of each paper.) The quality of the paper is dependent upon other features of the paper. For example, having misspelled words detracts from the quality of the paper, as does having ungrammatical sentences. Perhaps being clearly written adds to its quality. There are various other factors that enter into the evaluation. These factors involve the descriptive properties of the papers. The key idea to understand is that if there is an evaluative difference in the papers, then there must be a descriptive difference. In other words, if there is no descriptive difference, then there is no evaluative difference either. The following principle captures the idea:
Necessarily, if two papers have all the same descriptive properties, then they have the same evaluative properties.
This is sometimes described as a supervenience thesis—the evaluative properties of the papers supervene on, or depend upon, their descriptive properties.
The plausibility of the supervenience thesis about the two papers can be appreciated by considering the plight of a student who gets a low grade. Suppose such a student asked the teacher what made his paper inferior to the paper of a classmate who got a higher grade. There would surely be something wrong with a teacher who replied to this student, “There is no descriptive difference between the two papers. They are exactly alike in all descriptive ways. It's just that, unfortunately, your paper is not as good as that one.” This student can properly complain that if the paper is not as good, there must be something about the two papers that brings about this evaluative difference.
Something similar is true in epistemology. Being justified or unjustified is an evaluative epistemic property of a belief. Facts about the causes of a belief, whether it is true, whether other people also believe the same thing are nonevaluative facts about the belief. In addition, facts about what experiences a person is having, what other things the person believes, and so on are all nonepistemic facts. The evaluative epistemic facts depend upon these other facts. Thus, if one belief is justified and another is not, there must be some nonevaluative difference between the two beliefs that accounts for this evaluative difference. This idea can be summed up in the following epistemic supervenience principle:
Necessarily, if two beliefs have the same nonepistemic properties, then they have the same epistemic properties. (If two beliefs are exactly alike nonepistemically, then either both are justified or both are not justified, or they are justified to the same degree.)
Defenders of all the theories of justification that we will consider in this chapter and the next agree with this thesis. The difference between the various theories concerns which properties determine epistemic status, or which descriptive facts make an epistemic difference.
I. EVIDENTIALISM Our question about Example 4.1 concerned what made Careful justified in believing (1) but Hasty unjustified in believing that proposition. It may seem that the answer to our question is rather simple: Careful has good reasons, or evidence, for believing (1) whereas Hasty does not. It is the possession of evidence that is the mark of a justified belief. We will call this the evidentialist theory of justification, or evidentialism.
While evidentialism may be correct, as stated so far it is not a well-developed theory. Philosophers who agree that justification is a matter of having good reasons differ markedly over just what having good reasons amounts to. There is, then, more to be done to develop a satisfactory account of justification. The questions will become clearer as we examine the idea more carefully.
A. Epistemic Evaluations In a famous essay, "The Ethics of Belief," published in 1877, William K. Clifford describes the following example:
2Example 4.2: The Negligent Shipowner
A negligent shipowner decides, without doing any careful checking, that his ship is seaworthy. The ship sets sail, and then sinks. Many lives are lost, largely because this shipowner believed that his ship was seaworthy without bothering to get it checked out.
Clifford draws a harsh conclusion about this shipowner. And building upon this example, and some others, he formulates a general conclusion well worth examination. That conclusion is Clifford's thesis, (C):
C. It is wrong always, everywhere, and for anyone, to believe anything upon insufficient evidence.3 There are obvious questions to ask about this, most notably, "What counts as insufficient evidence?" We can bypass that question for now, assuming only the following: If a person has more and better evidence for the conclusion that proposition p is false than for the conclusion that proposition p is true, then that person has insufficient evidence for believing that p is true. Maybe Clifford thinks that to have sufficient evidence requires even more, something like very strong evidence. But we can bring out an issue concerning (C) by using this weaker condition. In discussing and defending (C), Clifford writes:
It is not only the leader of men, statesman, philosopher, or poet, that owes this bounden duty to mankind. Every rustic who delivers in the village ale-house his slow, infrequent sentences, may help to kill or keep alive the fatal superstitions which clog his race. Every hard-worked wife of an artisan may transmit to her children beliefs which shall knit society together, or rend it in pieces. No simplicity of mind, no obscurity of station, can escape the universal duty of questioning all that we believe.4 His idea is that by believing on insufficient evidence one helps to keep alive “fatal superstitions” and that failing to follow one's evidence will tear society apart (“rend it in pieces”). Although Clifford's claims may seem a bit extreme, perhaps there is some merit to his thesis.
Some critics may object to Clifford's thesis on the grounds that modest amounts of evidence, especially in cases in which a decision must be made quickly, can make belief acceptable. Here is an example designed to illustrate the point.
Example 4.3: Chest Pains
You are about to go away on vacation. Shortly before you are scheduled to leave, you feel some slight chest pains. You know that such pains are typically associated with indigestion, but they can be signs of heart problems. Worried that there might be a serious problem, you call your doctor.
This is a sensible action. Yet the evidence you have is fairly weak. You do not have sufficient evidence to believe that you have a serious medical problem. Hence, it might be concluded Clifford's thesis is wrong. Sometimes, a little evidence is good enough.
Clifford has a good reply to this objection. (C) is not a thesis about when it is wrong to act. It is a thesis about when it is wrong to have a belief. So if this ex-ample makes any trouble for (C), the example must be one in which having a belief is not wrong, even though one does not have sufficient evidence for it. If the situation is as just described, it would be wrong to believe that you have heart trouble (if the symptoms described are the only reasons you have for thinking this). You are pointlessly going well beyond your evidence if you believe that. But you do have good enough evidence to believe a different proposition, namely that there is a possibility that you have heart problems. Furthermore, this belief provides good reason to take precautionary action. There is nothing wrong with this belief or with acting on its basis. So distinguishing a belief from a related action, and distinguishing the proposition that there is a chance you have a heart problem from the proposition that you do have a heart problem, provides all that is needed to escape this objection.
There are, however, some other objections to Clifford's thesis that are more effective.
Example 4.4: The Optimistic Batter
A major league baseball player is coming to bat in a crucial situation. This player is a good hitter: He gets a hit about one-third of the times he comes to bat. Still, more often than not, he fails to get a hit. Like many other major league players, he is supremely confident: Each time he comes to bat he believes that he will get a hit. This sort of confidence, we may assume, is helpful. Players do better when they are confident (believe that they will succeed), and they do worse when they lack confidence.
The details of Example 4.4 suggest that it is not wrong for the batter to believe that he will get a hit. In fact, it seems far better for him to believe this. Yet he does not have “sufficient evidence” for the proposition that he will get a hit.
Example 4.5: Recovery
A person has a serious illness from which few people recover. But this person is not willing to give in to her illness. She is sure that she will be one of the lucky ones. And confidence helps: Those who are optimistic tend to do a little better, even though, unfortunately, most of them do not recover either.
Clifford’s thesis says that it is wrong for this patient to believe that she will re-cover. And that judgment seems to be entirely too harsh. Imagine criticizing the hopeful patient, claiming that it is wrong to be optimistic. If optimism helps, it is hard to find fault with her for being optimistic.
These examples seem to show that there are cases in which it is not wrong to believe something, even though one does not have good evidence for it. Still, Clifford maybe right to think that each case of believing on insufficient evidence has one bad feature: It does run the risk of encouraging bad habits of thought. However, (C) depends upon the idea that this factor always outweighs other considerations. The two examples just given are designed to show otherwise. Some-times the good of believing on insufficient evidence outweighs the potential harm.
You may find that you are of two minds about these cases. On the one hand, past performance suggests that the batter in Example 4.4 will not get a hit. This seems to indicate that there is something wrong with the belief that he will get a hit this time. On the other hand, the fact that believing that he will get a hit tends to improve his performance suggests that it is not wrong for him to believe that he will get a hit. After all, this belief aids his performance, just as concentrating, holding the bat properly, and, perhaps, scratching and spitting do. Similar considerations apply to Example 4.5. The statistics about recovery from the disease suggest that there is something wrong with the belief that the patient will recover. The belief "flies in the face of the facts." Yet this is her best chance to recover. How can we condemn a person for trying?
One good way to resolve these apparent conflicts is to say that there are two (or more) different notions of wrongness under consideration here. One notion concerns morality (or prudence or self-interest). The other is more intellectual or epistemological. A plausible thing to say is that in these examples the beliefs are morally right but epistemically wrong. We need not get into any detailed discussion of morality here. It will be enough to say that, typically, behavior is immoral when it has bad effects on others (or oneself) and there is no offsetting benefit. Believing on insufficient evidence can have the bad effects Clifford notes. However, in Examples 4.4 and 4.5 there are clear offsetting gains. Clifford's thesis is completely general. He says “everywhere and always” it is wrong to believe on insufficient evidence. If Clifford's thesis is about morality, as it seems to be, then it is incorrect. It just is not immoral to have the optimistic and beneficial beliefs in these circumstances. Thus, it is likely that Clifford went too far in asserting (C) in full generality. Sometimes it is not morally wrong to believe on insufficient evidence.
However, thinking about these examples and Clifford's thesis may help us focus on the more central epistemological issues. Suppose a person interested only in getting at the truth were in the position of the people in our examples or were forming a belief about those people based on exactly the evidence they have. Such a person would set aside self-interested concerns such as winning the game or recovering from the disease. (You might think about a person who is going to place bets on the outcomes and is interested only in winning the bets.) That person would be interested only in what in fact is true. What would that person believe in that situation? It is clear that such a disinterested believer would not believe that the batter will get a hit or that the patient will recover. You might put this point by saying that in these situations, with the evidence as described, there would be something wrong with believing these things. But this is not a moral matter. It is a matter of rationality or reasonableness. In other words, it is epistemically wrong to believe these things in the situations described.
The key idea to get from this is that we can evaluate beliefs in two ways. We can evaluate them morally5—are they beneficial? Do they cause any significant harm? In the two examples, the beliefs are beneficial (when held by the batter or the patient). They therefore get a favorable moral evaluation. We can also evaluate beliefs epistemically. In the view about epistemology under discussion here, this is determined by whether they go against the evidence. If Clifford had said that it is epistemically wrong to believe on insufficient evidence, he would have asserted a view that many philosophers take to be correct. But his claim about morality is mistaken.
Clifford's discussion helps us focus on the notion of something being epistemically wrong. It is this evaluation that the justification condition of the TAK is about. An epistemically justified belief is a belief that is favorably evaluated from an epistemological point of view, no matter what its moral or prudential status.
B. Formulating Evidentialism The central idea of evidentialism can be stated in the following evidentialist principle about justification:
EJ. Believing p is justified for S at t iff S's evidence at t supports p.
A version of (EJ) that covers other attitudes is possible as well. It says that the justified attitude—belief, disbelief, or suspension of judgment—is the one that fits the evidence. A fully fleshed-out evidentialist theory will say something about what a person's evidence at any given time consists of and what it is for that evidence to support a particular belief.
In general, evidentialists will say that the evidence a person has at a given time consists of all the information the person has to go on at that time. This will include the memories the person has and the other justified beliefs he or she has. When evidentialists speak of someone "having evidence," they do not mean the same thing someone discussing a legal matter might mean by that phrase. Suppose a certain document is a crucial item in a case. You have that thing among your possessions, but you do not know about it. In the legal sense of "having evidence," you may have the relevant evidence. But in the sense intended here, it, and facts about it, are not part of your evidence. The evidence you have consists of the information you have available, in some hard-to-specify sense, for your use. The key idea, then, is that the evidence a person has consists of the data the person has to go on in forming beliefs, not of the items the person physically possesses.
For it to be true that a person's evidence supports a proposition, it must be that the person's total evidence, on balance, supports that proposition. It is possible to have some evidence that supports a proposition and some evidence that supports the denial of that proposition. If these two bodies of evidence are equally weighty, and the person has no other relevant evidence, then the person's total evidence is neutral and suspending judgment about the proposition is the justified attitude. If one portion of the evidence is stronger than the other, then the corresponding attitude is the justified one. In all cases, it is the total evidence that determines which attitude is the justified one. Call this the total evidence condition.
There is a distinction, not so far mentioned, that is important to evidentialism. An analogy from ethics will make the distinction clear. A person can do the ethically right thing for the wrong reasons. For example, suppose a wealthy person is asked to give some money to a charity and agrees to transfer the funds electronically. The charity gives her the account number so that she can make the transfer. Armed with this information, the person decides to take money from the charity rather than to give money to the charity. However, by mistake she pushes the wrong button and does transfer money to the charity. She does the right thing, but she does it by mistake. Her action is right, but it is not "well intentioned" or "well motivated." She is to be condemned for her character and motivations, even though she did the right thing.
There is an epistemological analogue of this example. Suppose that you have good reasons to believe something and you do believe it. However, you believe it not on the basis of those good reasons but because of an astrological prediction or as a result of some logical blunder. You believe the right thing for the wrong reasons. In such cases, believing that proposition really does fit your evidence, and so according to (EJ) believing it is the justified attitude. But it is an epistemically "bad" belief. You are not doing so well, epistemically speaking, in holding that belief.
These examples show that there are two related ideas of justification that we should distinguish. One is properly expressed in (EJ). It is the epistemic analogue of the action that is in fact good, i.e., the best one to do, given the situation. There are several different ways we will express this idea:
S is justified in believing p.
Believing p is justified for S.
S has a justification for believing p.
None of these imply that S actually does believe p. They just imply that S has what is needed to make believing p epistemically appropriate.
The second sort of justification is the epistemological analogue of the idea of doing the right thing for the right reasons. This is the idea of a "well-formed" or a "well-founded" belief. We will typically express this idea by saying such things as
S's belief in p is justified
S's belief in p is well founded.
S justifiably believes p.
Sentences of these forms do imply that S believes p and that S does so for the right reasons. Here is a more precise account of this concept:
BJ. S's belief that p at time t is justified (well founded) iff (i) believing p is justified for S at t; (ii) S believes p on the basis of evidence that supports p.6 Clause (ii) of (BJ) is intended to capture the idea of believing on the basis of the right reasons. Call this the basing condition. A generalized version of (BJ), applying to disbelief and suspension of judgment, could also be developed.
Evidentialism affirms of both (EJ) and (BJ). It holds that the justified attitude toward a proposition for a person at any time is the attitude that fits the total evidence the person has at that time. And an actual belief (or other attitude) is a justified (well-founded) one provided it fits the person's evidence and the belief is held on the basis of evidence that really does support it.