Lecture Notes, UC Davis Philosophy 102, Theory of Knowledge


G. J. Mattey, Senior Lecturer

Picture of William Alston

William Alston

Text: William Alston, "Concepts of Epistemic Justification" (1985)

Plato suggested that knowledge is true opinion with an account. This is often invoked for the analysis of knowledge as true belief with justification.

S knows that p if and only if
The ancient skeptics argued, in modern terms, that we lack justification for many of our beliefs (those that go "beyond the appearances"). Descartes countered the skeptics by claiming that many such beliefs are indubitable, which is the strongest possible kind of justification. James's "empiricism" rejects the claim that knowledge requires irrefutable justification. Almost all contemporary theorists of knowledge embrace this view, which is now known as "fallibilism."

William Alston, now a retired professor at Syracuse University, wrote an influential paper in 1985 which tries to give a unified account of epistemic justification. It is an excellent introduction to the issues surrounding the notion of justification, and it lays out all the major alternative approaches.

The General Concept of Epistemic Justification

Alston begings by making a crucial distinction between being justified in believing that p and justifying one's belief that p. Justifing is doing somehing to show that the belief is true or that it is justified. Being justified "is a state or condition one is in." He claims that the literature greatly confuses these two senses of the term. Alston will consider on the state of being justified. The reason for this restriction is that if we required an activity of justifying for all knowledge, we would not know very much, since most perceptual knowledge does not involve justifying. (Of course, it is always open for someone to claim that because we do not justify most of our beliefs, they do not amount to knowledge!)

Justification applies to beliefs (or a cognitive subject's having a belief). It is an evaluative, rather than factual, concept (a justified belief for a subject S is "all right, satisfactory, in accord with the way things should be"). It is epistemic, falling under an "epistemic point of view," which is defined by the aim of maximizing truth and minimzing falsehood for a large number of beliefs. (The restriction to a large number of beliefs is to rule out confining belief to what is obviously true.) Finally, justification is a matter of degree. It will be treated as absolute here, but it could be understood as the degree of justification required for knowledge. (For Descartes, the degree of justification required for knowledge is what puts belief beyond doubt.)

Justification, then, is about the favorable status of believing relative to a point of view that seeks to maximize truth and minimize falsehood. The favorable status of believing can be due to the fact that it fulfills of some obligation or that it does not violate one and obligation. This is a "deontlogical" notion of justification, which so-called because of its resemblance to obligation-based ethical theories. There might be some other factor that makes believing epistemically favorable. Alston will later endorse a non-deontological account. He takes this split to be "the major divide in this terrain."

Deontological Accounts of Justification

Deontological accounts of justification will be considered first, and Alston claims that most epistemologists who have tried to explain justification have done so in terms of obligation to believe. (Alving Plantinga has argued that Descartes was a deontologist regarding justification, as is John Locke.) Deontologists make an analogy with justified behavior, which is behavior not in violation of some obligation. Justified behavior is not in general obligatory. So, deontological epistemic justification is not being in violation of an epistemic obligation, for which one could be reproached. (Compare what Foley says about what one is to believe.)

The norms that define obligation are set by what Alston has termed the epistemic point of view: maximizing truth and minimizing falsehood for a large body of beliefs. Specific obligations will owe their status to the way in which they promote the general obligations. An example is the injunction: Don't believe what is not supported by adequate evidence. Another is: Believe what you see to be implied clearly by what you already believe or are justified in believing. This is not to rule out other specific obligations embodying the "epistemic point of view" or indeed that there are other more general obligations defined by something other than the "epistemic point of view" described here. But Alston will work from the norms defined by the epistemic point of view.

The most generic version of a deontological account of evidence is this:

(I) S is Jd in believing that p iff in believing that p S is not violating any epistemic obligations.

To flesh it out, we need to look at the various ways or "modes" in which one might have an obligation in general. This will give a clue as to ways in which epistemic obligation can be understood.

Modes of Obligation

The first distinction is between objective and subjective obligation. Alston does not give conditions for obligation as such, but rather for fulfilling an obligation. His reason for approaching it this way is that epistemic justification is understood by the deontologist in terms of not violating an epistemic obligation, which is equivalent to fulfilling an epistemic obligation not to do something.

To fulfill an objective bligation is actually to bring about to bring that which the obligation enjoins. It requires less to fulfill a subjective obligation: only to do what you believe will probably bring about what you are supposed to. Usually we take people to be blameless if they fulfill the subjective side of the obligation.

Merely believing that what you do will probably bring about the outcome is not always enough, though. Sometimes we hold people blameworthy unless they act on the basis of what they have good reason to think will accomplish the end. For this reason, we have to introduce a "cognitive" mode of obligation. To fulfill a cognitive obligation, one must not only believe that his action will probably bring about the required end, but the belief must be justified. This justification must be based on other beliefs that the person holds.Immanuel Kant in the eighteenth century. He claimed that an act is not morally right unless it is performed out of duty. If one does something for some other reason (say, in the pursuit of pleasure), and it happens to produce the result one is obligated to bring about, it has no moral worth.

Now we can summarize the four modes of obligation in a series of numbered analyses.

(II) S has fulfilled his objective obligation iff S has brought about A.

(III) S has fulfilled his subjective obligation iff S has done what he believed to be most likely to bring about A.

(IV) S has fulfilled his cognitive obligation iff S did what he was justified in believing to be the most likely to bring about A.

(V) S has fulfilled his motivational obligation iff S has done what he did because he supposed it would be most likely to bring about A

Modes of Epistemic Obligation

It is easy to transpose these analyses into four modes of with justified belief. For each mode, a justified case of believing is one in which the believing does not violate that mode of obligation.

(VI) S is objectively justified in believing that p iff S is not violating any objective obligation in believing that p.

(VII) S is subjectively justified in believing that p iff S is not violating any subjective obligation in believing that p.

(VIII) S is cognitively justified in believing that p iff S is not violating any cognitive obligation in believing that p.

(IX) S is motivationally justified in believing that p iff S is not violating any motivational obligation in believing that p.

An even more specific set of analyses can be obtained by substituting one specific obligation, that of believing only that for which one has adequate evidence. This is a norm which Alston had suggested earlier as a more concrete implementation of the general norm of maximizing truth and avoiding error in a large number of beliefs.

(X) S is objectively justified in believing that p iff S has adequate evidence for p.

(XI) S is subjectively justified in believing that p iff S believes that he has adequate evidence for p.

(XII) S is cognitively justified in believing that p iff S is justified in believing that he has adequate evidence for p.

(XIII) S is motivationally justified in believing that p iff S believes that p on the basis of (x) adequate evidence p, or (xi) what he believed was adequate evidence, or (xii) what he was justified in believing was adequate evidence.

Motivational justification has three different options, each corresponding to one of the other modes, which is indicated by the use of lower-case Roman numerals.

Of these four, Alston throws out subjective justification, (XI) as inadequate for an account of epistemic justification. To motivate this exclusion, he says more about the relation of having adequate evidence. A proposition q is adequate evidence for pwhen its truth makes p at least probably true. Subjective justification, then, requires only that one has a belief that some of his other beliefs make p at least probably true.

This raises two problems. The first is that the belief that constitutes my evidence is not itself a justified belief. In that case, I would be justified on the basis of evidence that is not itself justified, which seems wrong. The second is that what I believe to be adequate evidence is not adequate, i.e., does not make p probable. Again, Alston wants to say that this failure blocks epistemic justification, but he admits that this claim is "more controversial" than the first.

Now we are left with (X), (XII) and (XIII) as contenders for the role of explaining epistemic justification. Alston thinks that (XII) is preferable from the standpoint of deontological theory. If a person is justified in believing that he has adequate evidence for his belief that p, how can we blame him if his belief that p is incorrect? It would seem that having a justified belief in the adequacy of his evidence is all that we could, from an epistemic point of view, expect of a person.

On the other hand, we tend not to blame people, generally, for failing to have adequate evidence in the first place. There are all kinds of reason why they might not. For example, a person might be a victim of an elaborate hoax. So (XII) is preferable to (X) in properly assigning blame for failure.

Then what about (XIII)? Motivational justification is compatible with any or all of the others, as is seen from the three variants of the condition. In Section V the issue of a "motivational rider" will come up in the context of a non-deontolgical account of epistemic justification.

Against Deontological Accounts of Epistemic Justification

Now Alston makes a well-known case against the deontological conception of justification. First, he allows that one may be justified in refraining from believing as well as in believing, given that one can be obligated to believe something. And one can be obligated to refrain from believing something, so long as one is justified in believing something.

Obligation to believe or to refrain from believing "certainly looks" like it requires belief to be voluntary. That is, such an obligation requires that, just by willing or deciding or choosing, I can bring it about that I believe that p. If we accept this, we generate a version of Jd, i.e., Jdv, for "voluntary deontological justification."

Alston finds direct voluntary control over belief to be "quite unrealistic." In cases of belief belief generated by experience, the belief-forming mechanism is isolated from the will. This constitutes a huge class of beliefs. To be sure, there are cases where the evidence underdetermines what should be believed, as with a fluid military situation or with religious and philosophical beliefs. Alston thinks that even then something other than "willing to believe" takes place. The person makes a choice that leads to his acting as if he has the belief. The military commander makes a "working assumption." Even if belief is produced by willing in these cases of underdetermination, there are vast numbers of cases where it is not. This rules out Jdv as a general account of justification.

Some philosophers, such as Keith Lehrer, make a distinction between belief, which is not generally voluntary, and acceptance, which is the output of a process of deliberation. Justification is then tied to acceptance. But even supposing Lehrer is right, he has a lot of work ahead in explaining how we can have justification for those things that (apparently) we only believe.

Involuntary Belief

There is a way to rescue a deontological conception from these criticisms. It requires giving up the simple analogy between having an obligation to believe and having an obligation, say, to open the door. It will have to be along the lines of action we can undertake voluntarily that can affect belief. So belief will be compared to something like the condition of having friends. We can do things to change this condition for better or worse, though we cannot bring about such changes directly through an act of will. Since we can undertake these facilitating actions, it still makes sense to speak indirectly of an obligation to have friends. Alston will focus on obligations to do what we can to bring about a certain condition.

As far as believing is concerned, our activities may affect the formation of specific beliefs or the formation of general habits that lead to beliefs. So in the first case, one can make all sorts of checks relevant to the truth of the belief. And we can more generally do things to make ourselves more critical, etc. This leads to a deontological account of justified involuntary belief.

(XIV) S is Jdi in believing that p at t iff there are no intellectual obligations that (1) have to do with the kind of belief-forming or -sustaining habit the activation of which resulting in S's believing that p at t, or with the particular process of belief formation or sustenance that was involved in S's believing that p at t, and (2) which are such that:
(A) S had those obligations prior to t.
(B) S did not fulfill those obligations.
(C) If S had fulfilled those obligations, S would not have believed that p at t.

Clause (1) of this analysis refers to the two factors mentioned above. The idea is that we are epistemically blameworthy when we fail to cultivate the right habits that lead to fulfilling the epistemic goal or fail to do the specific things we need to do in order to achieve the goal in the present case. Clause (2) indicates that the blame is incurred when one had an obligation to carry out something indicated in clause (1) which would have prevented the belief that p but did not do so.

This account is too strong because of the counterfactual condition (C). Failure to fulfill an epistemic obligation might be irrelevant to the case at issue. If I had worked at becoming more critical (which is my intellectual obligation) but instead took a walk, I wouldn't have believed that I saw two dogs fighting. But this does not make the belief that I saw two dogs fighting unjustified.

The cases Alston was trying to rule out with (C) are those which directly involve the fulfillment of the epistemic obligations. These will be ones where our failure results in the activation of a bad habit that we wouldn't have had we fulfilled our obligation, or if the failure results in sheltering us from adverse considerations, which allows the belief to be acquired or retained. So he proposes this revision:

(C) If S had fulfilled those obligations, then S's belief-forming habits would have changed, or S's access to relevant adverse considerations would have changed, in such a way that S would not have believed that p at t.

This is too weak a condition. One can fulfill it by being, for example, isolated in a culture where good epistemic habits are not encouraged. One cannot be blamed for believing what one does, and there is nothing reasonable one could do to change it.

A second reason the condition is too weak is that it can be fulfilled through a deficiency in one's cognitive powers. Here Alston recounts the case of a misreading of Locke by a student "who just doesn't have what it takes to follow abstract philosophical reasoning, or exposition for that matter." He reads Locke as saying "One's knowledge is restricted to one's own ideas" and gets out of it that everything is a matter of opinion. This is a possible case where no amount of effort will change thing, or else the amount of effort would be unreasonable. "And yet we would hardly wish to say that the student is justified in believing what he does about Locke."

Other sources where no change in habit would produce a change in belief are those involving be poor training or or an inability to resist believing, both of which would take too much time or effort to correct. So Alston abandons Jdi as a candidate, and with it deontological justification.

A "Goodness" Account of Justification

Now we are looking for a replacement, and there are many models of evaluation to choose from, where the desired state is necessarily not under one's direct control. Basic capacities or bodily build are conditions which we can find desirable or undesirable, but which are completely out of the control of the individual. We can also evaluate people regarding their personal appearance or state of health, where they have some measure of control. The epistemic point of view may be one from which we can evaluate qualities of a belief, just as we can evaluate the qualties of the body from an athletic or aesthetic point of view.

We might call the good qualities of a condition "virtues." Then Alston's replacement is one in which a justified belief is a belief that is virtuous from an epistemic point of view. This foreshadows what has since come to be called "virtue epistemology." We shall study this approach in more detail when we look at a paper of Ernest Sosa.

The analogy Alston has made yields this schematic analysis of justification.

(XV) S is Je in believing that p iff S's believing that p, as S does, is a good thing from the epistemic point of view.

The qualification "as S does" is to restrict the account to the belief by S (and not necessarily anyone else) in the present context (and not necessarily at another time and place). It is neutral as to what makes believing a good thing from the epistemic point of view, so as not to prejudice debates about justification, but it can be made more specific.

As a first step toward this end, Alston asks why we should not just identify good belief with true belief. The reason no epistemologists do this is that justification is "internalist." In evaluating belief, we are looking at it from a person's point of view. This is what influenced the discussion of deontological notions of belief in favor of the "cognitive" mode as opposed to the "objective." We are interested in whether S has adequate grounds that are sufficiently indicative of the truth of p.

Talk of adequate grounds requires an account of the relation of being based on a ground G. It is difficult to say what this is, though we have many paradigm cases to guide us. One is inference. Another is when a belief is based on what appears to be the case in visual experience. The problems come in with cases that seem to be ones where there is basing of belief but no conscious inference occurs. This can be handled (in a way that raises problems) by appeal to unconscious inference, or maybe a disposition to cite the ground as a reason. Another possibility is causation, but here there seem to be many causal factors irrelevant to the basing relation. How do you isolate one way? The solution for present purposes is just to work from the paradigm cases of "being based on a ground."

Grounds might be thought to be beliefs or experiences. Some hold that only beliefs can be grounds, but this seems too restrictive. Keith Lehrer, for example, thinks that experience always stands in need of interpretation, so that nothing in experience is a ground unless it is a belief about experience. But Alston wants to hold open the possibility of direct or immediate justification through experience. Adequate grounds are justified beliefs strongly indicative of truth.

As with the deontological conception, there are modes here as well.

(XVI) Objective -- S does have adequate grounds for believing that p.

(XVII) Subjective -- S has grounds for believing that p and he believes them to adequate.

(XVIII) Cognitive -- S has grounds for believing that p and he is justified in believing them to be adequate.

The subjective version is dismissed for the same reason as with the deontological account. We do not want to say that just believing that one's grounds are adequate makes his believing a good thing. It could be based, for example, on an egotistical over-estimation of his powers of judgment.

This leaves the objective and the cognitive senses. Alston recognizes that he cannot appeal to the same kind of argument that favored the cognitive mode of deontological justification over the objective mode. Blameworthiness is not involved in the present conception of justification.What makes believing a good thing is that one in fact has adequate grounds. If one's grounds are bad, the fact that one is justified in thinking they are good is not consistent with his believing on those grounds being a good thing. So Alston opts for the objective mode of justification as the way to construe Je properly.

This still leaves the motivational mode to be considered. Alston has narrowed the field to objective justification, so there is only one viable way to construe the motivational mode.

(XIX) Motivational -- S's belief that p is based on adequate grounds.

You can go back and forth about whether (XIX) or (XVI) gives a better account of epistemic justification. There is one consideration which favors (XVI). It seems that we can say that a person can have a justification for something he does not believe. This is allowed by (XVI) but not by (XIX). On the other hand, if someone actually does believe that p, that belief would be justified according to (XVI) even if the adequate grounds he has have nothing to do with why the person believes that p. This is why the motivational mode is invoked in the first place.

Alston decides that both sides are right in the sense that they are putting forward viable accounts to justification that are actually used in theory of knowledge. He will not try to decide that one or the other is the unique concept of justification. On the other hand, he doex argue that (XIX) is more fundamental to the theory of knowledge. The reason is that it is "the richer one and thereby embodies a more complete account of a belief's being a good thing from the epistemic point of view." A situation in which one's believing is connected with adequate grounds does a better job of explaining what makes the believing good. A further point is that with (XIX) you get objective justification as a consequence.

Overriding Reasons

The justification conferred by (XIX) is only "prima facie." That is to say, even if one has adequate grounds on which the belief is based, one still may fail to be justified. The reason is that those grounds can be overridden by the total body of evidence available to the person. So Alston reformulates both what it is for justification to be motivational and what it is to be justified.

(XX) Motivational -- S's belief that p is based on adequate grounds, and S lacks overrriding reasons to the contrary.

(XXI) S is Jeg in believing that p iff S's believing that p, as S did, was a good thing from the epistemic point of view, in that S's belief that p was based on adequate grounds and S lacked sufficient overriding reasons to the contrary.

This may be thought too strong, in that it eliminates "groundless" belief, such as those which concern one's conscious states. Another such class of beliefs concern logical truths. These would be the propositions believed are "self-evident" (if there are such cases). The solution is to treat them as limiting cases where the ground is minimally distinguishable "either from the belief it is grounding or from the fact that makes the belief true." So in the case of beliefs about one's conscious states, the ground may be the state itself. In the case of beliefs about self-evident propositions, the gorund might be the way that they appear, e.g., as "obviously true."

Internalism vs. Externalism

Alston had argued earlier that epistemic justification should be understood from the "inside," that is, from the point of view of the believer. There are two ways of understanding this "internalist" conception of justification, both of which Alston rejects. One is that justification is by reasons, which are justified beliefs or knowledge that lie "within the subject's perspective." This is a very restrictive view that Alston has already rejected. The second understanding of internalism is that what lies within the subject's perspective is what is directly accessible to the subject, which includes experiences as well as beliefs. So it is less restrictive, but it is too restrictive for Alston, who wants to allow for beliefs of which we are not conscious, and which thus are not accessible to us.

Alston bases his own view on the notion of grounds for belief. What makes his view internalist is that the grounds for a justified belief must be psychological states of the subject having the belief. These states will include belief and sense-experiences, but they may also include what was mentioned earlier, "the way a proposition seems obvious to one," as well as aesthetic and religious experiences. (Here Alston is leaving the door open for epistemically justified religious belief.)

Internalism is always contrasted with externalism. So what would be the externalist counterpart to Alston's view? It might be "reliabilism," according to which a belief is justified just in case it is formed in a way that reliably leads toward the truth. It seems as if a reliable belief is an epistemically good belief, and that being reliable is what makes grounds adequate. In fact, the relationship between reliabilism and Jeg appears to be so close as to make the two indistinguishable.

But the two are not the same conceptually, nor do they generate the same results in all cases. To show the difference, Alston uses a well-known type of imaginary case. A person has, unbeknownst to him, had implanted in him a device that reliably produces true beliefs, say about the temperature. His belief that the current temperature is n degrees is justified according to the reliabilist. But the operations of the device that produced that belief cannot serve as an adequate ground, on Alston's view. It does not fall within the person's perspective.

Another difference is that overriding reasons may defeat reliable grounds. You may believe something based on reliable memory but have been told by people you trust (but who are misleading you) that it is faulty. Then, according to Alston, your belief is not justified. Reliability is not restricted to the subject's perspective.

So Alston settles on Jeg as the best candidate for "epistemic justification." It is evaluative like Jd, it does not suppose voluntarism in belief, it requires that one be in a strong epistemic position, unlike Jdi. And unlike reliability, it is internalist, and finally, it leaves room for disagreement about the specifics of what grounds are and what makes them adequate.

Summary and Application

Alston summarizes the paper by drawing out a number of its lessons. (1) Being justified is more central to epistemology than justifying. (2) What justifies is from the subject's perspective. (3) Deontological conceptions are either based on something unrealistic or are too permissive. (4) Being based on adequate grounds "incorporates more of what we are looking for in the concept of epistemic justification" than having adequate grounds. (5) Although closely related to reliablilty, it respects perspective. (6) Jeg "satisifies the chief desiderata for a concept of epistemic justification."

Alston concludes with an illustration of the usefulness of his account of justification, which he compares to one of the deontological accounts, Jdv. Specifically, Jeg cuts out the need for higher-order requirements, such as justifiably believing that one has adequate grounds. Believing on adequate grounds is enough to make you justified. But on Jdv, it seems that a higher-order requirement is a necessary condition for justification. "If I decide to believe that p without knowing whether the ground is adequate, am I not subject to blame for proceeding irresponsibly in my doxastic behavior, whatever the actual strength of the ground?"

The same goes for making higher-order requirements a sufficient condition for justification. Meeting the requirement that you have a justified belief about the extent to which p is justified does not yield Jeg. You actually have to be in a favorable epistemic condition, not just believe justifiably that you are in one. But for Jdv, believing justifiedly that your grounds are adequate is enough to make you blameless. This is the final item to tip the scales in favor of the evaluation-grounds account. Avoiding a regress of higher-order requirements allows Alston to avoid at least one skeptical threat.

[ Course Home Page | Menu of Lectures ]