Lecture Notes, UC Davis Philosophy 102, Theory of Knowledge

Epistemic Obligation

G. J. Mattey, Senior Lecturer

Richard Feldman

Text: Richard Feldman, "Epistemic Obligations" (1988)

James described "two ways of looking at our duty in matters of opinion." One duty is to know the truth; the other is to avoid error. This implies that the formation of opinion is voluntary. Opinion seems to be the kind of thing we have voluntary control over and hence suitable as the subject of duty. If we try to extend this duty to belief in general, we run up against the problem noted by Alston, that belief often is involuntary.

One way of dealing with this situation is simply to declare that our duty is confined to the formation of opinion in those cases where we are free to decide what to opine. Another approach, that of Alston, is to claim that matters of "duty" are peripheral to the central concern of theory of knowledge, which is understanding epistemic justification. Richard Feldman takes a third tack. He wants to uphold a notion of epistemic obligation (which might also be called duty) while holding that in many cases we are not capable to carrying our obligation out.

Types of Obligations

Feldman distinguishes between three types of obligations: practical, epistemic, and moral. In Section I, he discusses the relations between the three, coming to the conclusion that they are distinct from one another.

The distinction between practical and epistemic obligations is motivated by a scenario, in which a student, Jones, faces an oral examination from a teacher who passes 10% of students overall, but 20% of those who believe they will pass. Jones is is a normal student and is aware of the passing rates, though not of any reason why he would be different from anyone else in his performance.

Feldman asks whether Jones would be blameworthy if he were to believe that he would pass the examination. From one point of view, the answer would be that he is, since the preponderance of evidence points to his failing, whether or not he believes that he will pass. But from another point of view, the answer is that he cannot be blamed for doing something that would appear to double his chances of success.

The two points of view here are, respectively, the epistemic and the practical. Jones would be epistemically blameworthy but practically blameless for passing. The contrast between the two can be understood in terms of his personal stake in the outcome of the belief. He has an interest in passing, which is why we are inclined to say that he is practically blameless. But with respect merely to the truth of his belief, he is disinterested. There is no outcome-based excuse, so to speak, in his defiance of the evidence.

A necessary condition for epistemic obligations, then is that they "arise from a purely disinterested and impartial perspective." This is not a sufficient condition, though, at least from the standpoint of some standard moral theories. Whether one is morally obligated to do or refrain from doing something is not based on whether the action serves a person's own interests. (The moral philosophy of Kant, for example, is of this sort.)

So suppose that there are moral obligations to have certain beliefs, e.g. that one's family-members are trustworthy. Such obligations could conflict with epistemic obligations, as in the case of a family-member who has always behaved in an untrustworthy manner.

William K. Clifford, as we saw in the lecture on James's "The Will to Believe," had proposed that it is always wrong to believe on the basis of insufficient evidence. If he means that such belief is morally wrong, then he disagrees with Feldman's point that sometimes it is morally permissible to believe what is not epistemically permissible to believe. The reason Feldman separates the two is that "the moral consequences of an epistemically obligatory belief might be very bad." He does not further explain here what he means.

But this is not the thrust of the discussion. Instead, it is meant to draw attention to a problem for the very concept of epistemic obligation. This is the problem to which Alston called attention. There is an argument to the effect that one is not free (at least for the most part) to form beliefs at will. If having an obligation requires that one have the ability to do this, then there is no epistemic obligation.

The Voluntarism Argument

The thesis of doxastic voluntarism is that "having a belief is something a person does voluntarily and can control." The denial of doxastic voluntarism (what might be called "doxastic involuntarism") is taken by critics such as Alston and Alvin Plantinga to imply that there are no epistemic obligations. Feldman reconstructs a simple version of an argument to this end, which he calles "The Voluntarism Argument."

  1. Doxastic voluntarism is false.
  2. If doxastic voluntarism is false, then no one has epistemic obligations.
  3. Therefore, no one has epistemic obligations.
In Section II, Feldman will argue against the conclusion of the argument by denying premise 2.

On first reading, the argument seems persuasive. To motivate premise 1, Feldman gives two examples. The first is that of someone who forms a belief which is contrary to what he wishes were the case: his favorite politician is dishonest. The second is a perceptual belief he forms when stepping outside in Rochester, New York, on a typical winter's day: it is cold and grey out here. This belief "just comes over me."

Premise 2 is motivated in a less direct way, by appeal to "a long tradition that makes only voluntary or controllable actions the subject of obligation." This historical point is far from conclusive, and Feldman himself will show a way around it shortly.

Denial of Premise 1

According to Feldman, Descartes might have rejected premise 1. We have seen in the Meditations that his method was to withhold judgment voluntarily from what he did not find to be certain. Although he made much of the ability to avoid believing, he seems to have thought that having a belief is something that occurs "spontaneously" when one is presented with the evidence. It may be, then, that the voluntarism argument holds only for half of the thesis of epistemic voluntarism: we have no control over forming a belief. If we could have control over avoiding belief, then it would seem that we could have epistemic obligations not to believe.

A different kind of defense of doxastic voluntarism is due to John Heil and was recounted in the Alston article. It is to make a distinction between direct and indirect control over one's beliefs. It is acknowledged that one cannot have direct control over beliefs--cannot decide for any belief that he will have it. But one can do things that promote or discourage the having of beliefs. "I can, for example, study logic and probability theory with the aim of reducing the number of mistaken inferences I draw." (Note that this is a form of control over what one does not believe.)

In response, it could be argued that indirect control is not sufficient for epistemic obligation. It is the wrong kind of voluntary control: one needs direct control in order to have epistemic obligations. Feldman goes along with this argument on the grounds that we make judgments of epistemic obligation with respect to the formation of specific beliefs, not the carrying out of background activities that promote or discourage belief. "Our intuitive judgments do not limit epistemic obligations to cases in which we have the sort of indirect control Heil describes." So we can get a new version of the argument.

  1. "Direct" doxastic voluntarism is false.
  2. If "direct" doxastic voluntarism is false, then no one has "direct" epistemic obligations.
  3. Therefore, no one has "direct" epistemic obligations.
This argument is "equally damaging as the first one because we do make attributions of something like "direct" epistemic obligations. (Though it must be said that it is unclear exactly what "direct" and "indirect" mean in this context.)

For example, although it seems as if nothing Feldman could do voluntarily would keep him from believing that he sees some tables and chairs now, it still seems that it is something he ought to believe. And there are beliefs that some people cannot bring about voluntarily, such as ones pertaining to their personal qualities ("I am a success"), which nonetheless they ought to believe. The second Voluntarism Argument rules these cases out. So how is the apparent truth of epistemic involuntarism to be reconciled with the inclination we have to ascribe obligations in matters outside an individual's control?

Denial of Premise 2

One could object to the original argument that premise 2 is false because there are something like "indirect" epistemic obligations. But this does not help with the threat posed by the modified argument. Feldman thinks "that this response simply evades the point at issue here," which is whether we have an obligation to have or to refrain from having certain beliefs.

As noted in the last lecture, Keith Lehrer has suggested that we draw a distinction between belief, which is involuntary, and acceptance, which is voluntary. In that case, epistemic obligations would be restricted to cases of acceptance or refusal to accept. Feldman thinks there is a better way to reject premise 2. But he also wants to say positive things about epistemic obligations. So he suggests that the reader who is not happy with his alternative response to premise 2 can take what he will later say about obligations to apply only to accepting propositions.

Feldman's own solution is to distinghish between those types of obligation which require voluntariness and those which do not. It seems that moral obligation does require that the act one is obliged to perform or refrain from performing be voluntary. Practical obligation may have this requirement as well. On the other hand, legal and financial obligations seen not to require that the relevant acts be voluntary. If epistemic obligation is like them, then premise 2 is false.

An example of a legal obligation is an academic obligation. A student is required to complete all the work in a course. Perhaps in the case of some student, he cannot do so. This, Feldman contends, does not remove the requirement. The financial case he gives involves a mortgage he has undertaken. If he loses his job, he cannot make his payments, but this failure does not remove the obligation.

Why are these latter two kinds of cases like the epistemic case? All Feldman says about the matter is that we just do assign epistemic obligations regarding beliefs that we cannot help, such as perceptual beliefs. "It is natural to say things like 'He should (or shouldn't) believe that" even in cases in which there is no freedom or control." The point of the analogies, then, is simply to show that it is not in general wrong to do so.

Conditional Obligations

It may be objected that the cases of legal and financial obligations are not appropriate here because they are "conditional" rather than "simple" obligations. So in the financial case, the obligation is this: "given that I want to keep the house, I ought to pay the mortgage." If you detach the condition, leaving "I ought to pay the mortgage," the result is a false statement. There is no obligation apart from the condition itself. So these do not seem to be cases of unconditional obligations that cannot be fulfilled.

Feldman's response is to allow that one could understand epistemic obligations as conditional as well. We might say, "given that I want to attain epistemic excellence, I ought to believe p." The fact remains that the action cannot be undertaken in some cases while the conditional obligation remains. So whether legal and financial obligations are conditional or unconditional, in some cases they cannot be fulfilled, and we can make an analogy with epistemic obligations either way.

The Received View of Epistemic Obligations

The account of epistemic obligations that is "most widely accepted" goes back at least to William James. Recall from two lectures ago that James proposed two "commandments" for believing: to know the truth and to avoid error. He thought that there are choices to be made in how to carry out both, and that excessive caution (error-avoidance) is a bad choice because it precludes beliefs that are of great value.

That the two "commandments" are independent of each other is illustrated by the fact that one could fulfill one while utterly failing to fulfill the other. Believe everything, and you are sure to fall into error. Believe nothing, and you will never attain a truth. We get "epistemic excellence," Feldman contends, through "a suitable mix of the two goals." (James, again, thought that the mix should be a rich one, favoring belief as against unbelief.)

Roderick Chisholm and Keith Lehrer have in more recent times explicitly adopted a Jamesian account of epistemic obligation and built it into their theories of knowledge. Other writers have defended the view and described it as being a familiar one or the standard one.

Difficulties with the Received View

One problem with the Jamesian (as opposed to James's) view is that those who follow it re-interpret "commandments" as "requirements," "obligations," "ends" and as an imperative. "While all these ideas may be fairly similar, there are differences worthy of our attention, and it is not entirely obvious just how the Jamesian line can be put into [a] plausible account of epistemic obligations."

A first attempt to do so runs as follows:

(1) S is epistemically obligated to believe p if p is true and S is epistemically obligated to disbelieve p if p is false.
This amounts to saying that we are obligated to believe all that is true and to disbelieve all that is false. A first problem is that the formulation leaves no room for suspension of judgment, which one surely ought to do in certain cases, as with the results of a coin flip, and "many other cases in which I have no good information about the truth value of the proposition in question."

A second problem is that we epistemically ought to believe what is false when all the evidence, which was carefully collected and weighted, points to its being true. Even if the evidence is misleading, if the person cannot not know that, "the mere falsity of his belief does not demonstrate that he failed to do what he epistemically should have done." Ancient school children who believed the earth is flat were doing what they should have done, epistemically.

To obtain a substitute for (1), Feldman resorts to an analogy. Common advice in the trading of stocks is to buy low and sell high. He formulates this as:

(2) S ought to buy stock x if the price of x will go up and S ought to sell x if the price of x will go down.
(This is really not a good formulation of the advice. If a stock is going up, this does not mean that it is "low," and vice-versa.) Still, (2) is false, because as advice it should be based on what the evidence indicates the stock should do. (Clearly this is meant to be an analogy with (1).)

To get at the truth "in the vicinity of (2)," we can try:

(3) One's goals as an investor are to buy stocks that will go up and to sell stocks that will go down.
This claim is plausible, but it is not trivially true. Children might need to be taught it, and people could disagree, if they have other goals such as socially-conscious investing. More specific advice about investing would concern how to go about satisfying the goal, such as buying stock in expanding industries rather than in industries that are in decline. Generally, this specific advice is about means to an end which is specified as the goal.

Feldman thinks that James's "commandments" should be seen as about goals, not results. They are, as Foley put it, about ends that we should strive for as epistemic agents. Nothing is said about the means to achieve these ends. So we have:

(4) One's goals as a rational believer are to believe things that are true and to avoid believing things that are false.
As with the stock-trading advice, this is not trivial. People have enunciated other ends for belief, such as to make us feel inspired or to enrich our lives. (Note that this was the actual goal that James espoused. Believing certain things that are true has practical benefits.) While (4) has the ring of truth, it says nothing about epistemic obligations, however. The connection is made at the end of the paper.

Chisholm's Account of Epistemic Obligations

The version of the Jamesian view adopted by Chisholm seems to fit with (4) and to avoid the objections to (1). Feldman re-formulates it as follows:

(5) For any proposition p and person S, if S considers p then S is epistemically obligated to try his best to bring it about that S believes p if and only if p is true.
This principle espouses two goals, believing what is true and not believing what is false, and in this way it is compatible with James's "commandments."

A problem in evaluating this proposal is that it says nothing about the relative weights that should be assigned to trying to believe what is true and trying to avoid believing what is false. James himself seemed to advocate in favor of boldness over caution, but Feldman will not pursue this issue because he thinks (5) has deeper problems.

First Objection to Chisholm

The first objection is due to Plantinga, who holds that Chisholm leaves out reference to the importance of some truths over others. (This objection is very much in the spirit of James.) If we are to be "good epistemic agents," we must focus on the right truths. Feldman responds that while it is all well and good to be a "good epistemic agent," this is not relevant to epistemic obligation (as Chisholm was trying to capture), though it may be relevant to some other kind of obligation

Specifically, what Chisholm is after is to give guidelines for what one ought to do in the present situation, confronted with the option to believe a specific proposition p that I am now considering. This issue is irrelevant to any long-term consequences to my belief system in adopting it. It only concerns the truth or falsehood of p, and not what it takes to be, in a more general sense, a "good epistemic agent."

Another broad consideration that is irrelevant has to do with the pursuit of further evidence. I might be a better epistemic agent if I were to search for more evidence, but this does not affect what I should do at this time. This latter issue is, for Feldman, "the central epistemic question, and questions about epistemic agency are quite clearly irrelevant to it."

Even if it is true that in a given case one should seek more evidence, this does not imply that the person should withhold judgment at the time. Heil suggests that there may be pragmatic reasons that dictate that one should seek more evidence even when one has enough evidence to believe in a given case. So, a well-tested drug should be believed safe even though more research is warranted. Such cases show, Feldman contends, that the question what to believe now and the question of whether more evidence is needed are "independent."

This example (and others to come) illustrate how Plantinga's concerns about the "good epistemic agent" are relevant to practical obligations. Thus whether the drug-tester ought to seek more evidence involves calculations about the effects of the drug: if it has severe side-effects, it ought to be tested further before being released, but if it provides treatement for a disease that was previously untreatable, then it should be brought to market. "There seems to be no purely epistemic considerations that decide this matter."

Now on to some further cases that separate the issues of what to believe now and whether to gather more evidence. A response to the claim that I should seek more evidence might be that I have better things to do. This is a practical issue. "There's never any purely epistemological considerations that decide these practical questions." Feldman concludes that the first objection is mistaken.

Second Objection to Chisholm

Feldman thinks that a second objection is more successful. Recall that the obligation is to try one's best to bring about the goal of believing all and only what is true, which is the epistemic goal. One reading of "try one's best" would have it that one has done so when one has actually succeeded in believing a truth or not believing a falsehood. That is, "I've tried in the way that in fact makes me believe all and only truths. What could possibly be a better way to try to achieve a goal than to do what in fact makes one achieve that goal?"

But on this reading of "try one's best," there is a difficulty. One might have to believe what contraverts his evidence. Feldman claims that Chisholm must agree that in some cases, trying one's best leads to falsehood.

A Revision of Chisholm's Account of Epistemic Obligations

Chisholm's idea seems to be that to try one's best is to use the best method at our disposal. This is formulated as:

(6) For any proposition p and person S and time t, if S considers p at t, then S is epistemically obligated to believe p at t if and only if, if S were to follow the best available method for trying to believe truths and avoid believing falsehoods at t, then S would believe p at t.
The methods here are supposed to exclude long-term strategies such as acquiring more evidence. They are simply what are at one's disposal at the time.

We need to exclude "unnatural" or "gerrymandered" methods from this analysis. We want to restrict the methods to ones which draw on the person's available evidence. The example Feldman gives here is not entirely clear. It involves trying to believe everything that is on a list of truths that has been drawn up earlier.

Feldman's Proposal

Feldman's own proposal may be closely related to (6) and may even be what Chisholm himself had in mind. It avoids the problems of (6), however. Not surprisingly, the method Feldman highlights in his modification of (6) is that of believing on the basis of what is supported or justified by his evidence, and not believing what is not so supported.

(7) For person S and proposition p and time t, S epistemically ought to believe p if and only if p is supported by the evidence S has at t.
There may be factors that block believing what is supported by evidence. In that case, one would have an unfulfillable obligation. Doxastic voluntarism is not required for the analysis. (Note the similarity here to Alston's preferred account of epistemic justification, (XXI).)

This seems consonant with Chisholm's later writings, where he substitutes for the reference to truth a reference to reasonableness. Given that what is reasonable to believe is what is supported by evidence, he comes close to endorsing something like (7). Feldman concludes by saying that our epistemic goal is to get at the truth, as embodied in (4), but our epistemic obligations are about how to get to the goal. One is epistemically obligated to believe in accordance with evidence, as this is the best means.

[ Course Home Page | Menu of Lectures ]