Lecture Notes, Lehrer's Theory of Knowledge, Chapter 6, Internal Coherence and Personal Justification

2002 Lecture Notes, Lehrer's Theory of Knowledge, second edition

Chapter 6, Internal Coherence and Personal Justification

By G. J. Mattey, Senior Lecturer, UC Davis Philosophy Department

After having devoted three chapters to criticism of rival theories of justification, Lehrer presents his own analysis in Chapter 6. Although there is much new detail here, Lehrer has in fact been laying the groundwork very extensively in the earlier chapters. I will summarize his general account as a series of positions on the nature of justification, using the terminology of other epistemologists. I shall here adopt a couple of abbreviations for the sake of economy. Acceptance with the goal of accepting all and only what is true will be called "t-acceptance" where the context demands that it be distinguished from other kinds of acceptance. The goals of t-acceptance will be called "t-goals."


The most fundamental tenet of Lehrer's theory of justification is that the only thing that can serve as evidence to support the acceptance that p is other things one accepts. His theory of justification is a doxastic theory. One's "acceptance system," which is just the body of one's acceptances, is an internal source insofar as acceptance is an internal cognitive state of a knowing subject. Moreover, acceptances are said to be the outcome of evaluation, so they are not opaque to the person whose acceptances they are. Indeed, the acceptances which support all justification are the sort of things to which one would appeal when questioned, "How do you know?"

The reason Lehrer adopts a doxastic theory is that he thinks there is no available alternative. So he claims at the beginning of Chapter 6 that "there is no exit from the circle of what one accepts" (p.124). We would like to be able to consult our sense-experience directly, but, "The stimulation of our senses raises the question of what we should accept, however, instead of answering it" (p. 124). The question is "whether what is suggested to us by our senses is true and accurate rather than false and illusory" (p. 124). To determine this, we must appeal to the acceptance system.


What one accepts in the interests of attaining truth and avoiding error is said to be the appropriate support of our knowledge because knowledge itself is correct information that is recognized as such (pp. 5-6). In order to accept anything, a person must evaluate whether accepting it furthers his t-goals. This is a rational or intellectual task, which requires, says Lehrer, a reasonably mature human being to perform.


If one is to determine whether one would be meeting the goals of acceptance in a given case, one must have a view of one's competence to judge rightly in such a case. In other words, one must place one's cognitive abilities in perspective. (This terminology comes from Ernest Sosa.) As Lehrer states it, our acceptance system "includes our conception of the world and our access to it" (p. 124). Having a view about one's own general competence is required to judge one's performance in a particular case. For example, in Chapter 2, Lehrer held that the quiz-show contestant failed to know that Elizabeth died in 1603, though he remembered the information, because of his negative assessment of his competence to judge such matters of history.


Neither assessments about competence nor assessments about performance are basic. One needs to accept that one has performed well in order to be justified in accepting that one is competent to perform well. (This is usually called the appeal to one's "track record.") But to determine one's competence, one must t-accept that one has been performing properly. So assessments of competence and performance must support one another if there is to be any justification at all. This is the fundamental relation of mutual support in Lehrer's coherentism.

Personal Justification

The acceptance system (AS) of a person provides the evidence which may support the acceptance that p for S. It provides this support in the context of an "evaluation system" (ES) of a person. Besides the "core" provided by the AS, the evaluation system includes information about one's attitude toward accepting that p relative to accepting something else, q. It also includes information about S's "reasonings." These two components of the ES will be discussed in turn.

Lehrer postulates that we have a range of epistemic "preferences" to accept p relative to accepting that q. Epistemic preferences are not limited to what one actually accepts, any more than dining preferences are limited to what we actually eat. So a person who has never entertained a skeptical hypothesis such as Descartes's "evil deceiver" may still prefer accepting his ordinary descriptions of the world to a description under which our faculties are defective due to the mechanations of a powerful being. Preference is a "partial ordering" on what may potentially be accepted. In most cases, accepting one thing may be preferred to accepting others, but there may be ties. The point of introducing preferences is to allow comparisons with respect to how reasonable it is to accept piece of information versus accepting another. An argument will be made later to connect preferences with reasonableness.

The final component of the evaluation system is "the system consisting of reasonings from acceptances to further acceptances as conclusions" (p. 127). What Lehrer seems to have in mind is the set of all instances of reasoning that S has made by the time when S's acceptance that p is said to be justified or not. "Reasoning uses and extends acceptance" (p. 127) in the sense that we use it to generate new acceptances from premises we already accept. The relevance of reasoning to justification is that the reasonableness of the premises is passed on to the conclusion. As we will see in Chapter 7, there are cases in which S fails to know that p due to faulty reasoning.

The Lottery Paradox

In Chapter 1, Lehrer stated that to be justified in accepting that p, it must be more than merely reasonable to accept it. One must be able to exclude conflicting possibilities. The motivation for this view lies in a problem with equating justification with a high degree of reasonableness. This problem, known as the "lottery paradox," is couched in terms of probability, but it would apply to a non-probabilistic notion of reasonableness as well.

Suppose we say that to be justified in accepting that p, p must be highly reasonable relative to S's evaluation system. Now consider a fair lottery with one thousand entries. It is highly reasonable to accept, of each lottery ticket, that it is a loser. Suppose it is reasonable enough that we could be said to be justified in accepting of each ticket that it loses. Then, on a plausible principle of conjunction, we are justified in accepting that all tickets are losers, which is paradoxical.

Lehrer's solution is to appeal to relative reasonableness. To be justified, one's acceptance that p must be more reasonable than anything whose truth would diminish the reasonableness of accepting that p. In the lottery case, the information that ticket 1 is a loser diminishes the reasonableness of accepting that ticket 2 is a loser. It is slightly more likely that ticket 2 wins given that ticket 1 loses. But it is no more reasonable to accept that ticket 1 loses than to accept that ticket 1 loses. So, Lehrer concludes, one is not justified in accepting that ticket 1 loses.


To be personally justified in accepting that p, p must cohere with one's evaluation system. Coherence is understood in terms of answering objections to the propriety of accepting that p. An objection is any piece of information q whose acceptance as true would make the acceptance of p less reasonable than its acceptance as false. We shall introduce some symbolization to make the discussion of reasonableness more compact. Where p and q are pieces of information that S might accept, we shall indicate that p is more reasonable to accept than q by this schema.

r(A(p)) > r(A(q)).
The concept of an objection requires a notion of "conditional reasonableness: how reasonable it is to accept something on the condition that one has accepted something else. In symbolic terms, we will use a '/' to indicate a condition on acceptance. So we get this characterization of an objection q to the acceptance of p:
r(A(p)/A(q)) < r(A(p)/A(not-q)).
(Note that we take it here that accepting that not-q is equivalent to accepting that q is false.) In the literature, what Lehrer calls "objections" are called "skeptical hypotheses." We will shortly look at some specific objections in the context of a couple of cases.

To answer an objection, one must be able to draw on the evaluation system to show the objection to be relatively unreasonable. One might think that an objection is answered when it is refuted or shown unlikely to be true, but this is not how Lehrer proceeds. Instead, he says that an objection is answered when it is less reasonable to accept that p itself:

r(A(p)) > r(A(q)).
Descartes thought that he had to refute the hypothesis of an evil deceiver by proving it to be false. Lehrer's standard is much more lenient, though he does think that in fact that hypothesis is made very improbable by the evidence in the case he discusses (pp. 133-4). If one can answer all objections, that person has justification, but we shall see below that answering all objections is not a sufficient condition for justification, as some objections can be "neutralized."

The Justification Game

To illustrate this abstract conception of justification that all objections be answered, Lehrer describes a game to be played between a claimant and a critic. The game is wholly imaginary, and whether one is justified in accepting that p in a given case does not depend on the existence of any game. Instead, the game is a "heuristic" device designed to give the reader a feel for the logical workings of justification.

Where p is the information whose justification for the claimant is in question, the claimant asserts that p. After this initial move, the critic advances an objection to p. The claimant then shows, appealing to his own evaluation system, how the objection is answered. (These appeals are given in parentheses.) Of course, if the objection is not answered, the critic wins and the claimant is not justified. If all objections are answered, then there is justification.

Lehrer develops an example, due to Fred Dretske, of a person at the zoo who accepts that he sees a zebra. Dretske was trying to illustrate that one may be justified in accepting that he sees a zebra though he is unable to rule out objections such as that the animal is a mule painted to look like a zebra. His paradoxical claim was that one does not know that it is not a painted mule while at the same time he knows it is a zebra. (Fred Dretske, "Epistemic Operators," Journal of Philosophy 67, 1970. pp. 1007-1023. Widely reprinted.)

On Lehrer's treatment, one does not have to know that it is not a painted mule, either, though he may know this as well. It must only be more reasonable to accept that one sees a zebra than to accept that one sees a painted mule. The claimant responds to the skeptic's objection in this way: "(Though I have no specific information about the stripes, I have no reason to believe that the Edinburgh Zoo would paint a mule to look like a zebra and identify it as a zebra, or be deceived by somebody else doing so. The Scots are known for their honesty. So my perceptual evidence is trustworthy information that the animal is a zebra.)" (p. 133).

The zebra example Lehrer illustrates how one would respond to the critic's objections in a justification game, but note that the few objections advanced are far from all those that would have to be brought forward by the critic were such a game actually to be played. Moreover, most of the objections Lehrer uses in the example are fairly extreme skeptical hypotheses. Though these have to be answered, they do not play a central role in the justification of every-day acceptances.

Summary: Answering Objections

I will now review the fundamental definitions and principles of Lehrer's analysis of justification, as they have been developed thus far. The information that o is an objection to the acceptance of a proposition pfor a subject S with an evaluation system ES if and only if it is less reasonable for S to accept p given that S accepts o than it is for S to accept p given that S accepts not-o. Objection ois answered by appeal to S's evaluation system if and only if it is more reasonable relative to S's ES to accept p than to accept o. If, based on S's ES, all objections to accepting that p are answered, then S is personally justified in accepting that p.

Varieties of Objections

Lehrer's account of Chisholm's "sheep" example on p.19-20 of the text will be used to give a further illustration of how these notions are supposed to work. In the example, a farmer sees what he takes to be a sheep and "knows a sheep when he sees one." Lehrer suggests that in this circumstance, the farmer can be said to be justified in accepting that he sees a sheep. Call the proposition that the farmer sees a sheep 'p.'

Contradicting Information

The default first objection to the farmer's accepting that p (not-p) is that the farmer does not see the sheep. This is the negation or denial of p. It is relatively uncontroversial that not-p is an objection. If the farmer accepts that not-p, it will not be at all reasonable for him to accept that p, since if he did, his acceptance system would be contradictory. (There are, however, a few dissidents who are willing to accept the rationality of accepting contradictory information, it should be noted.) On the other hand, if one accepts that not-not-p, then given the equivalence of that information to the information that p, one already accepts that p. (Again, there are dissenters, this time regarding the equivalence.) So it is most reasonable to accept p given not-not-p. On just about any account, it is more less reasonable to accept p given that one accepts not-p than it is to accept p given that one accepts not-notp. So the only question that remains is whether it is more reasonable for the farmer to accept that he sees a sheep than it is to accept that he does not see a sheep.

The direct objection to p, not-p, is answered with the evidence which favors p in the first place. The farmer accepts that he sees something nearby that looks like a sheep, and accepts that he can tell a sheep when he sees one. So it is more reasonable for him, given his evaluation system at the time, to accept that he sees a sheep than to accept that he does not see one.

Other pretty clear objections are those which imply not-p. Thus an objection would be that S sees a robot. Since robots are not sheep (i.e., they are not animals at all), the truth of the skeptical hypothesis implies the falsehood of p. And so we can apply the same considerations as in the preceding paragraph to see that the objection that the farmer sees a robot is a legitimate objection, which would have to be answered by showing that it is more reasonable for him to accept that he sees a sheep than to accept that he sees a robot.

Here the answer gets more complicated. Not only must the farmer have evidence that the animal is a sheep, but he must also have evidence that it is not a robot. This will have to involve reference to his competence to distinguish sheep from robots. This will be a theme in the answers to the other objections. Typically, the evidence would be that sheep have certain characteristics that robots lack, perhaps a smoothness of movement, etc.

A related kind of objection makes explicit reference to an error S has made. Thus the information that the farmer has mistaken a sheep for something else would be an objection. If he were to accept that he had committed a perceptual error of this kind, it would be very unreasonable for him to accept that he sees a sheep, whereas accepting that he has not made this error makes it eminently reasonable to accept that he in fact sees a sheep. Once again, this is because the claim that he has made a perceptual error of mistaking a sheep for something else implies that he does not see a sheep.

The objection that an error has been made would be answered by acceptances about the normality of the present situation. The farmer accepts that there is nothing wrong with his vision or other relevant faculties,that the object he takes to be a sheep is within normal identification range, that there are no unusual lighting conditions or other abnormalities in the visual setup.

Objections to Comptence

Other objections are less direct, as an objection that S does not have the competence to make the judgment. This objection could be that the farmer is generally unable to distinguish sheep from other objects that look like sheep. If he were to grant this, then it would be unreasonable to accept that what he takes to be a sheep is a sheep, and hence that he sees a sheep. But notice that the objection does not imply that he has made a mistake. He might be lucky by accident. But the odds of this are low, so accepting that he is incompetent greatly lowers the reasonableness of accepting that he sees a sheep, while accepting that he is competent raises it a good deal. For the objection to be answered, it must be more reasonable for the farmer to accept that he sees a sheep than to accept that he is incompetent.

The objection concerning the farmer's general competence are answered by what the farmer accepts about his abilities. He may, for example accept that he has been raised on a farm around sheep, that he raises sheep, and so forth. He has made thousands of successful sheep identifications under such conditions.

The final kind of objection we will consider here is been used to generate an important class of cases to be discussed in Chapter 7. Sometimes one is in an unusual environment, such that one's normal competencies are no longer operative. For example, the fields surrounding those which the farmer is viewing are filled with many robots which look very much like sheep but are not sheep. Under these conditions, one may be easily fooled, even though in ordinary circumstances one would not. So let an objection be that there are many robots in the vicinity, robots so realistic that the farmer cannot tell them from real sheep. We can say that if the farmer were to accept this, then it would not be reasonable for him to suppose that his ability to discriminate sheep from non-sheep is sharp enough in the present situation, so it would be less reasonable for him to accept that he sees a sheep if he accepts that there are realistic-looking robots in the area.

This final objection, concerning abnormal surroundings where the countryside is filled with sheep-like robots, would be dealt with differently. The farmer accepts that there is no reason to think that the area is filled with ersatz sheep, it has never happened before, there is no apparent motive for anyone to go to the trouble to do this kind of thing, etc.

There are many other objections that would also have to be answered. It is important to note that the definition of an objection and the partial analysis of justification do not require that S have any awareness of what the objections are. S must merely have the kind of evaluation system which is capable of supplying what is needed to answer the objections. The "justification game" described by Lehrer is no real game, but only an dramatization of the way in which competition is beaten within the structure of an evaluation system.

Objections to Perspective

Many objections challenge whether S has the proper perspective from which to judge the truth or falsehood of the information. In the sheep case, an objection might be that the farmer is too far away from the sheep to be able to identify it properly, given his powers of vision and his ability to recognize distinguishing characteristics of sheep. This objection might be answered by reference to the distance of the sheep and how easily its distinguishing characteristics can be judged from that distance.

This kind of objection may seem unremarkable, but it does pose an interesting challenge to Lehrer, as pointed out by some students in my 2002 class. Suppose the farmer is in the farmhouse, and that he has entrusted care of his sheep to his ultra-reliable chief shepherd. Sheep are scheduled to be in that field at that very time every day, and they have not failed to be there at that time for thirty years. An objection to "Sheep are in the field" might then be, "You are not in the field now." It seems, at least intitially, that the it would be less reasonable for the farmer to accept that sheep are in the field given that he is not in the field now then it would be for him to accept it if he is in the field now. This objection cannot be answered, since it is clearly not more reasonable for the farmer to accept that the sheep are in the field now than to accept that he is not now in the field. Does this mean that the farmer is not justified in believing that the sheep are in the field now?

If we require that to be justified, a person must physically perceive the object of his information, then very few of our acceptances will be justified. (John Locke claimed that claims to knowledge extend only as far as present experience, since there is no necessary connection between present experience and what exists in the recent past, other parts of space, etc. (An Essay Concerning Human Understanding, Book IV, Chapter 11, Section 9.) As a fallibilist, Lehrer wants to allow that a person can be justified in matters beyond present experience, yet he must be cautious. He had claimed in Chapter 1 that he is not justified in accepting that his secretary is in her office, when he is not there to see her. "I am unable to exclude the possibility that she is out of the office on an errand, for example, and in that way, my resonableness falls short of justification" (15).

Where is the line to be drawn? Which possibilities do we have to exclude? I suggest first that the only way to salvage justification given Lehrer's scheme is to say that some apparent objections are not really objections. Suppose it is mid-summer in Davis, I just arrived at a resturant after bicycling in 110 degree (Farenheit) heat. The owner takes me into the walk-in freezer, and I cannot see, hear or feel anything outside the freezer. Am I justified in believing that it is not snowing outside? The apparent objection that I am not seeing that there are no snowflakes outdoors could be taken not to be an objection. Going outside and looking would not make the acceptance any more reasonable than just thinking about the situation from inside the freezer. That it is not snowing now is so reasonable to believe that it is no objection that I have no further evidence that it is not.

We can adapt a story recounted by David Hume to reinforce the point. Suppose someone entered the freezer crying, "It is snowing outside!" The fact that he said this might count as an objection, and it might be a bit more reasonable for me to accept that he said this than to accept that it is not snowing outside. Would I have to be able to answer this objection to be justified in accepting that it is not snowing outside? It seems not to be the case. The testimony of the person makes it no less reasonable to accept that it is not snowing outside than it would be to accept this without the testimony. As Hume put it in the context of a seemingly air-tight report of a miracle, "such an evidence carried falsehood on the very face of it" (An Enquiry Concerning Human Understanding, Section X, Part II).

Neutralizing Objections

In general, not all objections can be answered, and nearly all, if not all, cases of justification will have to be able to deal with unanswerable objections. This class of objection works indirectly, through a kind of innuendo. Take the case of the zebra in the Edinburgh Zoo, described in Chapter 6. Lehrer accepts that p, he sees a zebra and shows how various competitors to this proposition are answered given his evaluation system. But now consider:

o*: I sometimes make mistakes about what I see.

This proposition is an objection to accepting that p, in that accepting that o* is true makes it less reasonable to accept that p than accepting that o* is false. If I accept that I never make mistakes about what I see (i.e., that o* is false), I should be more confident about what I accept based on my vision than if I accept that I sometimes make mistakes about what I see.

But o* is not answered in Lehrer's evaluation system (or so we may surmise). He has probably made numerous errors of visual identification, and the degree of reasonableness of o* should be quite high for him, higher even than that for the original p, that he sees a zebra. If o* is not answered, how can it be dealt with? Lehrer's claims that it can be met head-on and "neutralized." That is, if he is in a position to say that he is not now making a mistake about what he sees, the fact that he sometimes makes mistakes is irrelevant.

Lehrer gives a technical account of the conditions under which an objection is neutralized. Suppose we let the neutralizer be this: n: I am not now making a mistake about what I see. The objection o* is neutralized when it is no less reasonable to accept the conjunction o* & n than to accept o* itself. With this definition of neutralization in hand, Lehrer completes his analysis of justification:

not-(r(A(o* & n)) < r(A(o*)))
S is justified in accepting that p in evaluation system ES if and only if all objections to p are answered or neutralized relative to ES.

What is Reasonableness?

The account of neutralization raises the issue of the nature of reasonablness. Accepting both o* and n is riskier than accepting o* by itself, in that the conjunction contains more information than the single statement. In general, the probability of a conjunction whose conjuncts are independent of each other and have probabilities that are less than 1 and greater than 0 is less than the probability of either conjunct. The risk of error is compounded when information is combined. Lehrer states that for justification, "there is a barrier of risk beyond which we should not fall," given the goal of accepting only the truth. In the present case, "the risk of error is negligible" (p. 147), so the minimal requirement of justification is met.

What allows the conjunction to be no less reasonable than one of the other conjuncts, despite its being riskier, is that is more informative. It better promotes the second goal of acceptance, to accept all that is true. "The more informative a statement is, the more it tells us about the world, and the greater our gain in acccepting it if it is true, and the greater the risk of error" (p. 145). In general, informativeness is a factor in the reasonableness of accepting that p. Lehrer also mentions explanatory power, simplicity, pragmatic value, and conservation of what is already accepted as other "spices of truth" that may be factors in reasonableness (p. 148). But he does not define what reasonableness is, a decision he defends in part by appeal to the diversity of factors that have been suggested by philosophers (p. 144).

We might say intuitively that the reasonableness of accepting some information p as true is a function of how well accepting that p promotes our t-goals, given our evaluation system. Lehrer assumes that "we can tell, at least intutively, when it is more reasonable to accept one thing than another on the basis of our own evaluation system" (p. 128). But he also thinks that we should be able to explain what makes our acceptance reasonable. This will be a key issue in the sections that follow.

The Evaluation System as Evidence

Justification is relative to an evaluation system. Specifically, the evaluation system provides a way of answering or neturalizing objections. Lehrer raises a "foundationalist" challenge to this use of the evaluation system. He specifically targets the acceptance system, which is the "core" of the evaluation system, and which contains the information used to formulate answers to and neutralizers of objections. In order to serve as the basis for justification, members of the AS must themselves be justified, according to the objection. "The fundamental reply is that each of the things that I accept may be claims that I am justified in accepting because of other information I accept" (p. 137).

This reply is typical of coherentist responses to foundationalism. If each of what I accept in defense of p is itself justified, there is a proper basis for the justification of p. The foundationalist will not be satisfied with response, however, because ultimately p will be used in the justification of at least some of the acceptances that justifies it. Lehrer does not pursue this "fundamental reply" any further here. Critics of coherentism have held that justification of this kind would be circular. Lehrer has more to say about circularity in another context, but here his attitude seems to be guided by a claim he made early in the chapter, that "there is no exit in evaluation from the circle of what we accept" (p. 124). If we are going to have an account of justification at all, it will have to work this way.

Notice that Lehrer claimed only that what one accepts may be justified by other things that one accepts. This suggests that acceptances can play their role in justification without being justified themselves. Instead, they might only be "reasonable." Whether or not this is sufficient to answer the foundationalist, Lehrer gives a high profile to a "principle of trustworthiness" whose consequence is that accepting that p is reasonable.

The Principle of Trustworthiness

A typical objection to coherentism is that it allows justification within a small evaluation system, since the smaller the system, the fewer the elements that must cohere with one another, which should make coherence easier. In Chapter 5, Lehrer had noted that "beliefs must cohere in some comparatively strong way with other beliefs within a system for such coherence, however it is explicated, to yield justification" (p. 117). This strong kind of coherence is manifest in the requirement that every objection be answered or neutralized. A small evaluation system will not have the resources to answer the many objections that can be raised against any acceptance.

As noted above, a typical objection to the acceptance that p is one that raises the issue of a subject's competence or "trustworthiness" in accepting the kind of information that p encapsulates. In the example of the farmer and the sheep used above, the objection is that the farmer is generally unable to distinguish sheep from other objects that look like sheep. To answer such an objection, the farmer would appeal to his evaluation system to show that it is more reasonable to accept that he sees a sheep than that he is unable to distinguish sheep from sheep-like objects.

The claim that I am trustworthy in any particular matter under any special set of cirucmustances may be justified on the basis of other things that I accept: I accept that I have had success in reaching the truth about similar matters in similar circumstances in the past and that the present circumstances do not differ in any relevant way from past circumstances when I was correct. (p. 138)
We can say that the farmer can appeal to his acceptance that he has a good "track record" in identifying sheep and to his acceptance that the present situation is normal.

A higher-level appeal would be to the trustworthiness of one's faculties. "I may accept that my faculties, perception, memory, reasoning, and so forth are trustworthy guides to truth in circumstances of the sort I find myself in when I accept what I do" (p. 138). This would answer an objection that these faculties are defective in some way. These high-level acceptances would, it seems, have to be themselves supported by appeal to one's track record and circumstances as well.

The highest-level objection, that can always be raised, is directed at the acceptance system in general:

o**: You are not trustworthy in what you accept.

This is an objection that must be answered rather than neutralized, as it is a direct threat to the basis of the justification. My acceptance system must yield the result that it is more reasonable to accept that I am trustworthy in what I accept than to accept that I am not trustworthy in what I accept. (Of course, this will require that we mobilize the evaluation system in its own defense, as a coherence theory allows.)

Lehrer claims that "I must accept . . . that I am worthy of my own trust" (p. 138). That is, if I am to be justified in accepting anything at all, I must accept that "when I accept something, it is a good enough reason for thinking it to be true" (p. 138). It is not clear why this is something that I must accept. To answer objection o**, it must be more reasonable for me to accept that p than to accept that o**. Perhaps I can get this result without actually accepting that I am trustworthy with respect to what I accept if my evaluation system simply makes it unreasonable to accept o**. So we shall look for another role for the acceptance of one's trustworthiness.

This acceptance is unique in a couple of ways. First, it is a "second-order" acceptance, in that it is an acceptance whose content is what one accepts. This has the consequence that it applies to itself, as we shall see. Second, its content includes an evaluative notion, unlike ordinary acceptances. It is in effect a general description of the quality of evaluation that produces the acceptance system. In order for my evaluation system to justify my acceptance of p, it must be an epistemically worthy system of evaluation. For Lehrer, then, this acceptance is the keystone of justification. It is what gives the main support for the other parts of justification, though it is still supported by the others to some extent.

Lehrer calls the "principle of trustworthiness of acceptance," a "special principle of an evaluation system."

T. I am trustworthy (worthy of my own trust) in what I accept with the objective of accepting something just in case it is true.

He claims that if one accepts that T, then "my accepting something will be a reason for me to accept it" (p. 139). If this is the case, then acceptances will automatically have the status of being reasonable, which makes them better candidates for providing justification for other things one accepts. This helps to answer the foundationalist objection. Lehrer enunciates similar principles for the other elements of the evaluation system, preference and reasonableness. The trustworthiness of preference ties rankings of preference to rankings of reasonableness, which in turn are the basis of all justification.

Trustworthiness is a rich notion. It is said to be a "capacity or disposition" of a person to accept all and only what is true (p. 139). This capacity is manifested in the way in which information is evaluated and integrated into one's evaluation system.

"My trustworthiness is not just a matter of what I now accept, but also of how I change what I accept and even of how I change my method of changing in order to correct what I accept and improve in my quest to obtain truth and avoid error." (p. 140)
Trustworthiness is the result of how one learns from experience and from other people. It is a measure of one's ability to respond to criticism.

The Role of Principle T in Justification

The most important role assigned to principle (T) is to create a linkage between the contents of one's evaluation system and reasonableness. Accepting that one is trustworthy with respect accepting that p can generate the conclusion that it is reasonable to accept that p through the following conditional.

TR: If I am trustworthy in accepting that p, then it is reasonable for me to accept that p.

If S accepts that p, and accepts principle (T), then there is a general argument for the reasonableness of accepting that p in with the objective of accepting all and only what is true.

The Trustworthiness Argument

  1. I am trustworthy in what I t-accept. [Principle (T)]
  2. I t-accept that p.
  3. Therefore, I am trustworthy in t-accepting that p.
  4. If I am trustworthy in t-accepting that p, then it is reasonable for me to t-accept that p.
  5. Therefore, it is reasonable for me to t-accept that p.
The trustworthiness argument makes the fact that S accepts that p something that can play a legitimate role in justification. This is not to say that the foundationalist will be silent. He can counter that the premises of the argument consist in things that S accepts, so that the conclusion is reasonable only as far as the premises are. But the argument is supposed to establish that what one accepts is reasonable. Thus, the foundationalist can claim that the trustworthiness argument begs the question by assuming the reasonableness of what one accepts to show that what one accepts is reasonable.

Trustworthiness as an Explanation of Reasonableness

There is a second role assigned by Lehrer to principle (T) and the trustworthiness argument. Trustworthiness explains why it is reasonable to accept what we do. It shows why we can exploit the acceptance system to answer and neutralize objections. Moreover, it explains the basic notion of relative reasonableness that is the basis of justification. The principle of the trustworthiness of preference explains why preferring p to o makes it more reasonable to accept p than to accept o.

Once we have explained the reasonableness of preference, the comparative reasonableness of a target acceptance over objections to it is an immediate result of the preference for accepting one over the other. Trustworthiness explains the reasonableness of preference. The reasonableness of preference explains reasonablesness of comparative evaluation on the basis of the evaluation system. (p. 141)

In the following discussion, we shall focus our attention on the reasonableness of accepting that p, as this is the simplest case, and we should be able to transfer it to the reasonableness of preference (and reasoning) in a straightforward way.

The ability to explain how we are justified is central to Lehrer's whole project of analyzing the concept of knowledge (in the "correct information" sense).

The explanatory role of an analysis is of fundamental importance and must be appealed to in support of an analysis. We shall be concerned with an analysis that will be useful for explaining how people know that the input (the reports and representations) they receive from other people, their own senses, and reason is correct information rather than error and misinformation" (p. 10).
Part of what explains how people know that the information they receive is correct is that it is reasonable for them to believe it is correct, and this in turn is explained by their acceptance of their own trustworthiness.

Explanation begins with a phenomenon to be explained. Here, the phenomenon would be the fact that it is reasonable to accept some piece of information p as true. Since it is hard to demonstrate how reasonableness issues from an evaluation without begging the question, we will take it for granted that the information that p is reasonable for S to accept given his evaluation system. The explanation of the reasonableness of accepting p consists of two elements: the explanatory hypothesis that S is trustworthy in accepting p and the general conditional (TR) establishing the relation between trustworthiness and reasonableness.

We will grant here that the explanatory hypothesis and the link logically imply what they are said to explain. Lehrer argued in Chapter 5 that this kind of logical relation is not sufficient for explanation, however. So we must wonder why it is that he thinks that in this case there is an explanation.

One way to approach the issue is to say that the notion of trustworthiness is better-understood than that of reasonableness. Lehrer describes trustworthiness as a complex capacity of human beings to respond to information in the quest for truth (p. 140). But reasonableness is undefined, partly because it has complexities of its own that seem to be too much to sort out (pp. 144-148). This approach seems initially promising, but it should be noted that in one place, Lehrer virtually defines trustworthiness in terms of reasonableness, which would undermine its explanatory role.

I must accept, moreover, that I am worthy of my own trust, that is, that I am trustworthy as well: that when I accept something, that is a good enough reason for thinking it to be true, so it is reasonable for me to accept it. (p. 138)

We will proceed on the assumption that trustworthiness can be understood independently of the concept of reasonableness, though this will be questioned later.

Does Trustworthiness Explain Reasonableness?

The most obvious way that trustworthiness explains reasonableness is not available to Lehrer. If we understand trustworthiness in terms of a high ratio of success to failure in accepting all and only what is true, it is hard to think of anything better than trustworthiness to explain the reasonableness of acceptance. We accept what we do in pursuit of specific goals, and we are good at attaining those goals. Unfortunately, this avenue is not open:

My trustworthiness in what I accept is not simply a matter of my current rate of success in obtaining truth and avoiding error in what I accept. I may proceed in a manner that is worthy of my trust in what I accept but be deceived through no fault of my own. (139)
So trustworthiness is a subjective notion. In the paragraph just quoted, Lehrer speaks of "great intellectual integrity" in the pursuit of truth as the reason one may be worthy of trust (p. 140). This integrity shows itself by the means one employs in the "dynamic process of evaluation and amalgamation of information" (p. 140).

So perhaps what explains reasonableness of acceptance is one's honest pursuit of the goals of t-acceptance.

TR': If I pursue the goals of t-acceptance with intellectual integrity in t-accepting that p, then it is reasonable for me to accept that p.
I think that we still do not have a real explanation of reasonableness. As in the quotation from Lehrer given above, it seems that having carried out the evaluative process with great integrity is just what it means to be reasonable in what one accepts. An acceptance is reasonable for a person when it is the result of a person's best efforts to accept all and only what is true.

The best approach to giving explanatory power to trustworthiness is to take seriously Lehrer's claim that trustworthiness is a capacity or disposition. Having a disposition to accept what is true and not to accept what is false does not require that one be actually successful in meeting these goals. Unusual circumstances, such as the mechanations of an evil deceiver, might interfere with one's success. And intellectual integrity may be part of such a disposition. So we get a new bridge principle.

TR'': If I am disposed to realize my t-goals when I t-accept that p, then it is reasonable for me to accept that p.
The possession of such a disposition would have real explanatory value. Trustworthiness could be understood as a "naturalistic" property of a person, which accounts for the "normative" property of reasonableness (p. 144). (It would not be a "reduction" of reasonableness to some naturalistic property such as probability, however.) As was mentioned above, we intuitively think of the reasonableness of t-acceptance in terms of the way in which our t-goals are fulfilled. The disposition to realize those goals thus allows us to understand why an acceptance is reasonable.

The Indirect Defense of Principle (T)

The trustworthiness argument depends on acceptance of principle (T) as a premise. Appeal to acceptance of this principle to explain the reasonableness of what one accepts seems to beg the question unless the principle itself is independently reasonable to accept. As Lehrer notes, one could defend principle (T) by appeal to one's track record.

She may, of course, appeal to the character of what she accepts, to the various things she accepts, and reason inductively from premises concerning the trustworthiness of individual acceptances in support of the conclusion that (T). She might reflect on what she has accepted and her fine track record of mostly accepting what was worthy of her trust to accept. This argument would establish that the trustworthiness of her acceptances manifests her disposition to be trustworthy in what she accepts. (p. 142)
This seems to me to be exactly the right approach. A person who accepts that she is trustworthy is accepting something about her disposition, and the standard way in which we establish the existence of dispositions to behave in a certain way is to see how one has behaved.

Note that one could also appeal to more general considerations. For example, Descartes tried to establish his trustworthiness (in some matters) by proving that God exists and made him in such a way that he would not be deceived (in those matters). Or one could appeal to the natural makeup of our cognitive architecture. These are not routes Lehrer takes.

This defense of Principle (T) is characteristic of the reciprocal support typical of coherentism. If we appeal to our success in accepting all and only what is true, we must have information that our acceptances have hit the mark. But this information will consist of other things which we accept. They will get their support, in part, by our acceptance of the success we have attained in accepting what we do in the interests of truth. As Lehrer points out in Chapter 4, our acceptance of particular information supports our acceptance of general information, and vice-versa. The foundationalist will still find this to be question-begging.

The Direct Justification of Principle (T)

Lehrer spends considerable time developing a "more direct argument for the reasonableness of accepting (T)" (p. 142). This is to apply (T) to itself. Since (T) is a piece of information that might be accepted, and it is not restricted in its content, it can indeed apply to itself. Lehrer claims that it is "natural" to suppose that people do apply (T) to itself, though it is not clear to me why he says this. To make the self-application of (T) more plausible, Lehrer invokes Reid's analogy of light illuminating itself as well as other things.

This does not mean that (T) is self-justified, however, as there may be objections to (T) that go unanswered, Lehrer notes. The principle contributes to its own justification, but we also need information about "the sort of circumstances in which we err and those in which we do not" (p. 143). It seems that the contribution (T) makes to its own justification is that it establishes its own reasonableness.

Lehrer notes the application of (T) to itself produces a circular argument.

There is obviously a circularity in the trustworthiness argument when we use the principle (T) to support the conclusion that the other acceptances are reasonable and then use those acceptances and the principle itself to conclude that it is reasonable to accept it. Should we find the circle vicious?(p. 143)
He maintains that if we invoke (T) in response to a skeptical challenge, we have violated the "rules of rhetoric." This seems to me to undermine the role of (T) in establishing its own reasonableness. If a skeptic (or a foundationalist, for that matter) were to challenge the reasonableness of accepting that (T), it would seem that Lehrer's restrictions on violating rules of rhetoric would prevent the use of (T) to establish that it is reasonable.

Lehrer claims the "loop" is not vicious, but instead is virtuous, when we understand that the role of (T) in supporting itself is explanatory, not justificatory. If it did not explain its own reasonableness, principle (T) would be "a kind of unexplained explainer that explains why it is reasonable for us to accept the other things we accept and then falls mysteriously silent when asked why it is reasonable to accept the principle itself" (pp. 143-4).

As we have seen, what does the heavy lifting in explaining why what we t-accept is reasonable is appeal to a disposition to accept all and only what is true. Principle (T) encapsulates the information that our disposition is completely general, so that we are well-disposed to accept that we are well-disposed to accept what we do. One could postulate that we have a higher-order disposition to accept information about the quality our base-level dispositions, but this threatens a regress.

The issue here is whether appeal to our overall truth-conducive disposition explains why that disposition is truth-conducive. Here, the answer must be in the negative. A disposition explains nothing about itself, though it explains a great deal about phenomena to which it is relevant. It is itself explained by something more basic. We can explain why salt disappears when sprinkled in water by saying that it is soluable, i.e., that it is disposed to dissovel in water. In turn, we understand why salt is disposed to dissolve in water by understanding the molecular structures of salt and water and how the two interact. Similarly, if I were to explain my trustworthiness, I would appeal to some structural features of my evaluation system which make it truth-conducive. This is what "naturalistic" accounts of knowledge try to do. (See Chapter 8.)

I do not think that appeal to explanation shows that application of principle (T) to itself produces a virtuous loop. As Lehrer pointed out in Chapter 5, it is difficult to establish whether anything explains itself. It is this feature of explanation that has led philosophers over the ages to postulate the existence of an "uncaused cause." Those moved by the self-illumination metaphor find the notion of a "self-caused cause" (causa sui) more satisfying. But the case of human trustworthiness does not demand any such radical treatment. If we really are disposed to accept what is true and not to accept what is false, this should be explained by showing how it is that our evaluation system does well in handling the information it evaluates.

Previous Lecture | Next Lecture | Lecture Notes Menu

Philosophy 102 Home Page | G. J. Mattey's Home Page