By G. J. Mattey, Senior Lecturer, UC Davis Philosophy Department
Externalist accounts of knowledge are serious rivals to the internalist type of account advanced by Lehrer. In this chapter, he discusses a wide range of externalist theories, subjecting them to criticism both for their specific features and for difficulties he attributes to any externalist theory. In the course of the discussion, he also considers "naturalism" in epistemology.
What is Externalism? Lehrer describes the difference between the externalism and internalism by considering what is required to "convert" true belief to knowledge. As he stated in the previous chapter, knowledge requires that true belief must be appropriately related to the world. There must be a "truth connection" between what is believed and what the belief is about. Externalists think that making the right truth connection is sufficient to make true belief knowledge.
Making the truth connection does not, in general, require any higher-level information about how the connection is made. "The central tenet of externalism is that some relationship to the external world accounting for the truth of our belief suffices to convert true belief to knowledge without our having any idea of that relationship" (p. 177). So by contrast, an internalist theory would be one that requires what Sosa calls a "perspective" on one's belief and its relation to the truth. This requires a "conception of how we are related to a fact that yields knowledge" (p. 177). The classic case of an internalist is Descartes, who thought that he had to prove that his faculties are truth-conducive before he would claim knowledge.
Externalist theories of knowledge are rivals to the kinds of foundationalism described in Chapters 3 (infalliblist foundationalism) and Chapter 4 (fallibilist foundationalism) and to the kinds of coherentism described in Chapters 5 (explanatory coherentism) and Chapter 6 (personal coherentism). In all of the above theories there is a common element: that knowledge requires justification, and that justification is based on what a person believes or accepts.
In John Pollock's terminology, all of these theories are "doxastic" theories of justification. As Lehrer put it in Chapter 7, "There is no exit in evaluation from the circle of what we accept. Acceptance is the fuel for the engine of justification" (p. 124). The internalist requirement for a perspective flows directly from the fact that we need a conception of our relation to the world to evaluate the way that we might make the truth connection.
This is not the only way to understand the relation between internalism and externalism. It is often held that our own psychological states are the only evidence we have to support our beliefs about the external world. One might say that "there is no exit from the way in which things appear to us." Taken in this way, internalist theories might include those which are not doxastic and do not require a perspective. Pollock's and Chisholm's theories, for example, are foundationalist theories which take as basic the ways we are appeared to, rather than beliefs, and thus are non-doxastic.
Externalism and Types of Knowledge
Lehrer's preference for internalism over externalism stems directly from his account of the kind of knowledge he is analyzing. He wants to understand "knowledge that the information one possesses is correct," as opposed to "mere possession of information" (p. 6). In Chapter 2, Lehrer had conceded that mere possession of information, even if the result of a lucky guess, can be called knowledge (p. 36). But it is not knowledge that the information is correct.
Externalists understand knowledge as requiring more than possession of correct information. However they understand the "truth connection," it is systematic and not accidental. An appropriate use of one's mental faculties, for example, could supply the connection. In the case discussed in Chapter 2, a person who says, on the basis of memory, that Queen Elizabeth died in 1603 makes the truth connection in a way that looks to many people like it yields knowledge. We might call this sense of knowledge "appropriate possession of information."
Appropriate possession of information can be had by small children, certainly. It seems likely that non-human animals can possess information appropriately so as to yield this kind of knowledge. And we are often tempted to allow that machines such as computers can possess information in this way as well. (David Armstrong used a thermometer reading as a model for knowledge in an early externalist piece!) In each case, though, the truth connection can be made without any reflection. Possession of appropriate information comes about more or less automatically, without reference to a perspective.
One way to accommodate both internalism and externalism is simply to admit that there are two types of knowledge. Following Ernest Sosa, we can say that the kind of knowledge Lehrer is analyzing is "reflective knowledge" and the kind that externalists try to explain is "animal knowledge." There would be no dispute between internalists and externalists if both sides understood that they were interested in different things. The debate between the two camps rages on because both want to lay claim to unqualified "knowledge." I believe that this debate is sterile, because there is no such thing.
A simple way of making the distinction between the two is to respect Lehrer's separation of belief and acceptance. Belief, as he describes it, is the output of a more-or-less automatic system based on "habit, instinct and need" (p. 124). It is not evaluative, but it could function efficiently to yield true beliefs in a wide variety of circumstances. As Lehrer notes, "often" the output of the belief and acceptance systems "coincide" (p. 124). It seems plausible to assume that there is animal knowledge when what is believed coincides with what would be accepted. Reflective knowledge, however, exists only as a result of a process of evaluation. "Some of our beliefs accord with our purposes of obtaining truth and avoiding error, and those we positively evaluate and accept for these other purposes" (p. 40).
Varieties of Externalism
So far, I have written vaguely of an "appropriate truth connection." There are several forms of externalism that flesh this notion out. I classify them slightly differently than does Lehrer.
The basic division is between mainstream externalist accounts of knowledge which reject justification altogether and an account by Goldman who interprets justification itself externally. Within the basic division, some accounts take the crucial element for a belief in making the truth connection to be the history of the way in which the belief is formed. Other accounts are more formal, in that they are concerned only with the way in which the belief itself is related to the truth.
Knowledge-Externalism Most externalist theories reject a justification condition of knowledge and replace it with some sort of external condition. One reason is that they share Lehrer's view that justification is an evaluative notion and think that the appropriate truth connection is not evaluative. We will refer to non-justification theories of knowledge as k-externalist theories. They typically provide a revised three-part analysis of knowledge:
S knows that p if and only if
Belief-Formation and the Truth Connection
One way of understanding the proper relation between a belief and the world is to focus on how the belief came about in the first place. We might think of knowledge as requiring an appropriate response to the conditions in our environment. We make this response by forming beliefs about that environment. And we have knowledge when we form those belief in a way that reflects the way the environment really is.
The simplest such theory is that found in Alvin Goldman's 1967 article, "A Causal Theory of Knowing." Here is Goldman's analysis.
S knows that p if and only if the fact that p is causally connected in an "appropriate" way with S's believing p.Note that although the analysis is in the form of a single condition, the requirements that S believe that P and that P be true are built into it, just as for Lehrer knowledge is reducible to true acceptance that is undefeated (pp. 169-173).
The causal theory is appealing especially as an account of perceptual knowledge because in perception we enter into a causal relation with the object perceived. Since this causal connection is what provides the information about perceived objects, it would seem that our knowledge of the objects is a product of that causal connection. Moreover, the causal theory can handle standard Gettier cases. (In fact, this is what moved Goldman to propose the analysis in the first place.) Take the sheep example of Chisholm given in Chapter 1. S does not know that S sees a sheep because the belief that he sees a sheep is caused by the presence of the dog in the foreground, not the sheep in the background.
Lehrer points out well-known problems with the causal analysis. The most compelling one is that it seems to be limited to knowledge of objects to which we can stand in a straightforward causal relation. These are the ordinary objects of perception. But much of the knowledge we appear to have is not so restricted. Abstract objects such as numbers seem not to cause us to have the beliefs about them that we do (though some rationalists would question this claim). General truths and truths about the future are subject to the same problem. If theoretical objects, such as sub-atomic particles, do stand in causal relations to us, those relations are tenuous and not of the kind that would be easily adapted to the analysis. For these reasons and others, the causal theory has few if any adherents at present.
The second type of belief-formation theory we shall discuss is reliabilism. (Reliabilist theories may also be j-externalist, as will be described below.) In 1929, F. P. Ramsey proposed this analysis of knowledge:
S knows that p if and only if
Most reliabilist theories do not require absolute reliability, though they might. This would make it more congenial for Lehrer than externalist theories requiring an infallible truth connection. Given that it stops short of infallibility, reliabilism leaves us saddled with the problem of specifying the degree of reliability required for knowledge.
Another problem faced by reliabilist theories is known as the "generality objection." Reliabilists tend to describe the processes leading to true belief in fairly vague terms. Ramsey mentioned memory, for example. But not all our memory is reliable to the same degree. Some people have excellent short-term memory but poor long-term memory. Others can remember certain kinds of subjects well but are very poor at recalling others. It is essential for the reliabilist to be able to specify precisely the process which yields knowledge.
But here is where the problem arises. If the specification is too general (e.g. "perception"), then it will probably turn out that the process is not reliable. Pollock points out the human visual perception is not reliable in the universe at large, since it only works correctly in the light of a yellow sun, and yellow suns are relatively uncommon. (Contemporary Theories of Knowledge, second edition, p. 116). But if we try to overcome this problem by specifying that reliability need hold only in a certain domain (as with visual peception in standard lighting conditions), we run the risk of giving an ad hoc account of the circumstances. That is, we end up stating the circumstances in which a process is reliable as being just those in which we think a person has knowledge. Then we would not come to any greater understanding of what knowledge is because the concept of reliability would be entirely dependent on our intuitions about when we know.
The Truth Connection by Design
Alvin Plantinga has recently promoted a different kind of appeal to belief-forming processes. Going beyond the mere reliability of causal processes that produce a belief, he points to the "proper functioning" of the cognitive faculty that produces the belief. The difference here is that Plantinga has a "teleological" conception of the operations of our cognitive faculties. For the ordinary reliabilist, the purpose of one's cognitive faculties such as perception and memory is irrelevant to knowledge. What matters is how reliably they produce true beliefs. For Plantinga, on the other hand, what is important is that these faculties have roles which they may or may not carry out properly.
Plantinga's main argument against ordinary reliabilism is that the process could function defectively in a way that still produces a high ratio of true beliefs. So the process "must be non-pathological; we might say that the process in question must be one that can be found in cognizers whose cognitive equipment is working properly; it must be the sort of process that can occur in someone whose faculties are functioning aright" (Warrant: The Current Debate, p. 208). The type of cases he appeals to are ones involving defects such as brain lesions which happen to cause a person to have reliably true beliefs. In such cases, there is an element of accident in the truth of the beliefs, despite their reliability.
The teleological aspect of "proper functioning" comes into full view when Plantinga tells us that it is the "design-plan" of the faculty that determines its function. He argues that evolution could not have produced the required plan and concludes that our faculties are designed by God. As Lehrer notes, this is a very controversial claim, but it is not relevant to the fundamental issue of the correctness of externalism, since "proper functioning" does not by itself require a perspective on one's own epistemic situation.
Formal Externalist Accounts of Knowledge
Some externalist accounts of knowledge do not make reference to the way in which a belief is formed. Instead, they emphasize the relation of the belief to some general condition that connects the belief to the truth. The condition may be a law of nature or it may be a condition that tracks what would be believed in various circumstances.
The Nomological Account of Knowledge
The nomological theory is closely related to the causal theory. Armstrong in his book Belief, Truth and Knowledge attempted to explain non-inferential knowledge, such as perceptual knowledge, using the model of a thermometer. A good thermometer is one which gives accurate readings in a range of temperatures. It is no accident that the thermometer is successful; rather, there are natural laws which connect the reading with the temperature itself. Just as thermometer readings can be a successful indicator of the temperature, our non-inferential beliefs can be successful indications of facts about the world, so long as their occurrence is lawfully connected with the facts themselves.
My suggestion is that there must be a law-like connection between the state of affairs [S's believing that p] and the state of affairs that makes 'p' true such that, given [S's believing that p], it must be the case that p.
It does not matter, on Armstrong's account, how the law-like connection is established. It may well be that a reliable causal process is the basis of the connection, but it need not be. Moreover, a law-like connection is a very strong relation. Reliablism seems more appealing because it is based on a high frequency of true belief rather than a lawful relation between truth and belief.
The Counterfactual Account of Knowledge
We might say that Armstrong's nomological account of basic knowledge is more general than the causal account, since it does not require that what is known be the cause of the belief, but only that they be lawfully connected. An even more general version of k-externalism is the counterfactual theory. This theory was originally proposed by Dretske in his 1969 book Seeing and Knowing. Robert Nozick came upon the same notion some time later and developed it in some detail in his 1981 Philosophical Explanations.
The idea here is that in order for a person to know, that person's true belief makes the truth connection by "tracking the truth" through other possible but non-actual states of affairs. To see how this works, consider the example of Lehrer's secretary discussed in Chapter 1. Suppose she has an identical twin whom Lehrer cannot distinguish from the secretary herself. We could say that given that Lehrer had never seen the twin, he would believe that the twin was his secretary were he to see her. So his belief that his secretary is in the office does not track the truth through situations in which her twin is present.
Nozick's counterfactual account of knowledge, in its simplest form, runs as follows:
S knows that p if and only if
The tracking theory has come under heavy criticism, particularly by Saul Kripke in an unpublished manuscript from the mid-1980s. Kripke's main complaint is that the formal mechanism of tracking the truth does not coincide with what we intuitively take to be appropriate possession of information. This objection is discussed in Chapter 9, but it is appropriate to look at it here.
There is an intuition appealed to by Goldman that if someone is unable to discriminate perceptually between an object of belief and very similar objects, that person lacks knowledge. Suppose Henry is driving in the Wisconsin ("Dairy Capital of America") countryside and points out a barn to his son. Ordinarily we would say Henry knows it is a barn because he can tell the difference between barns and other things.
Suppose further that there are many fake barns in the area, mere shells which to the ordinary observer look very much like the real thing. Since Henry cannot tell the difference, he does not know that he sees the barn he really sees. If he were to see a fake barn, he would believe that he sees a barn, so his belief does not track the truth. Kripke notes that there might be unusual conditions there such that no fake barn could be red. In that case, there is no possibility of confusing red barns with nearby fake red barns. In that case, Henry knows that he sees a red barn, but he does not know that he sees a barn.
The mismatch creating the problem is that Henry's evidence, based on his experience with barns, supports the view that he sees a barn as well as the view that he sees a red barn. But what makes it true that he sees a red barn, the condition that precludes fake red barns, is not something about which he has any evidence. This problem seems decisive against the tracking theory as proposed by Nozick, and the theory itself lay dormant until Keith DeRose recently incorporated some elements of it into a "contextualist" theory of knowledge (see Chapter 9).
Relevant Alternatives
The counterfactual account of knowledge seems vulnerable to an objection that resembles the generality objection to reliabilism and applies to reliabilism as well. Suppose Lehrer sees his secretary in her office, but the secretary has an identical twin who lives in Australia, and who has vowed never to return to the United States. In general the twin is extremely unlikely ever to fool Lehrer into believing that he sees her sister. Lehrer's belief that his secretary is in her office does not track the truth because of the existence of this sister. That is, if the twin were in the secretary's office instead of the secretary, Lehrer would nonetheless believe that the secretary was there.
But it seems that this should not prevent us from attributing knowledge to Lehrer when he sees his real secretary at her desk. In general, in almost every case there will be some way in which we might fail to track the truth with our beliefs due to an unusual possibility. At the extreme is Descartes's demon possibility, according to which we are unable to distinguish between experience of the physical world and demon-induced "experience" of nothing.
There is a standard answer to this kind of objection: the demon possibility is deemed to be an alternative that is not relevant to whether one knows. So a belief need not track the truth across all possibilites, but only those which are relatively close to the actual case. If Lehrer's secretary's sister is in the same building as Lehrer, his failure to track the truth would be relevant to whether he knows; if she is in Australia and determined to stay there, it is not. A problem with this "relevant alternatives" proposal, however, is to specify exactly where to draw the line. What if the twin is in the same town? The same state? etc. A related problem is how to spell out a mechanism for describing the difference between relevant and irrelevant alternatives. Nonetheless, externalism remains an attractive alternative to skepticism. (Further considerations on externalism and skepticism in Chapter 8 will be discussed in the notes to Chapter 9).
Naturalized Epistemology
Many externalist theories are naturalist in character. A naturalist in epistemology views knowledge as the outcome of processes that are a part of nature. They typically fasten on the physiological process of sense-perception, which is properly studied by natural science. The next step is to view perceptual belief as a natural product of sense-perception and to note that belief is properly studied by psychology. The connection between true belief and the truth can then be seen as a natural connection, to be understood scientifically. Quine went so far as to announce that epistemology is simply a branch of psychology ("Epistemology Naturalized," in Ontological Relativity).
Naturalizers tend to view human beings as a highly-evolved animal. Thus they emphasize the continuity between the human being and the rest of nature. This is in sharp contrast to philosophers like Descartes who claim that the human mind is not a natural object. A second important feature of naturalism is that it operates purely at a descriptive level. It tries describe, in terms of laws of nature, how the truth connection is made. But it steers clear of evaluation of what one ought to believe or is permitted to believe.
At least one philosopher, Richard Fumerton, goes so far as to claim that the very nature of externalism lies in the attempt to understand knowledge in a way that contains only naturalistic components, as does Armstrong.
I argue that the best way to think of the common thread that runs through paradigm externalist accounts of knowledge and justification is their reliance on the reducibility of epistemic concepts to nomological concepts. . . . Armstrong, of course, explicitly invokes causal concepts in explicating basic (foundational) beliefs that operate as that thermometer registering the temperature. (Metaepistemology and Skepticism, p. 66)Whether this is the best way to understand externalism is open to debate, but there can be no doubt that the relation between externalism and naturalism is a very cozy one.
Jutification-Externalism
Let us now turn to justification-externalism. This brand of externalism is due to Alvin Goldman, who formulated it after he had proposed the causal theory. Goldman first abandoned the causal theory in favor of reliabilism and later identified the use of reliable processes with justification. Here is a simple (and far-from-final) version of this thesis, taken from "What is Justified Belief?" (1979)
If S's believing p at t results from a reliable cognitive belief-forming process (or set of processes), then S's belief in p at t is justified.According to Goldman, the problem with internalist theories of justification is that it is possible to meet their conditions of justification (being indubitable, self-evident, self-presenting, incorrigible) in a way that relies on a faulty belief-forming process. The faultiness of the belief-forming process will incline us, intuitively, to regard the belief as unjustified. The solution is to place "causal requirements" on principles which make a belief justified.
Lehrer disagrees with Goldman's claim that intuitively we regard a faulty belief-forming process as blocking justification. He and Stewart Cohen propose a case in which (they contend) intuitively we would agree that a person is justified despite the fact that his belief-forming process is "invincibly" unreliable. This is the old Cartesian evil demon hypothesis, this time put to a new use (hence the name "new evil demon case"). The demon-victim is utterly unrelible in the perceptual judgements he makes about the world, due to the machinations of the demon. Yet Lehrer and Cohen maintain that the victim's beliefs about his environment are justified when he has done the best he can under the circumstances.
How could there be such a conflict of intuitions? It is most likely that there is an ambiguity in the word 'justified.' This is brought out in Lehrer's distinction between personal justification and ultra justification. He would agree that the demon-victim is not ultra justified in his perceptual beliefs but is nonetheless personally justified.
It is interesting to note that Goldman himself finally acknowledged that his reliabilist notion of justification does not capture all our intuitions about justification. In a 1988 article "Strong and Weak Justification," Goldman wrote
There are two distinct ideas or conceptions of epistemic justification. On one conception, a justified belief is (roughly) a well-formed belief, a belief formed (or sustained) by proper, suitable, or adequate methods, procedures, or processes. On another conception, a justified belief is a faultless, blameless, or non-culpablebelief. (Philosophical Perspectives, 2, Epistemology, 1988 , pp. 52-53)Given that there are two notions of justification in play, Lehrer needs to show that the "weak" version is required for knowledge. This is not established by the demon case, since we would all agree that the victim lacks knowledge. Indeed, one critic of internalism, Plantinga, has argued that the internalist notion of a justified belief as blameless has been an impediment to the development of an acceptable theory of knowledge. He calls this a "deontological" notion of justification which he traces to Descartes and John Locke in the seventeenth century. See his book Warrant: The Current Debate, Chapter 1.
I think that internalism, as described by Lehrer, does not depend on a deontological notion of justification. To be sure, the notion that one can only evaluate one's belief from an internal perspective invites the view that knowledge requires being epistemically blameless. But note that internalism as such only requires that justification include evaluation from an epistemic perspective. It may be that all that is required for knowledge (even in the internalist's sense of knowing that the information one possesses is correct) is that the process of evaluation of one's connection to the truth be a reliable one. This possibility is discussed further below.
Trustworthiness and Reliablity
Lehrer does not reject all aspects of externalism. In particular, he thinks that reliability is essential to knowledge. To motivate this condition on knowledge, he describes a variation on the evil demon case, in which an absent-minded demon neglects for a moment to disrupt the victim's belief-forming process. Lehrer would say that despite having true beliefs which are justified (in the same way they were while he was being deceived), the victim does not have knowledge.
Recall that on his view, it is essential to justification that a person place trust in his acceptances. He must regard himself as "trustworthy." This is not quite the same as regarding himself as reliable, despite initial appearances. Trustworthiness is a "deontological" notion: it is an honorific signifying that one is worthy of epistemic praise. Of the victim of the demon, Lehrer comments,
She may be worthy of her own trust because she is as trustworthy in what she accepts as the circumstances allow. She is circumspect and seeks to detect every error in a way that makes her worthy of her trust before accepting what she does. There is no fault in her; the defect is external to her and lies in the circumstances that invincibly deceive her. (p. 192).So accepting one's trustworthiness is really tantamount to accepting that one has done one's best: "proceeding in ways that are worthy of her trust" (p. 192). Reliability, on the other hand, "is a matter of a high frequency of success in obtaining truth in the way she proceeds" (p. 192).
For reliability to be built into the account of knowledge, there must be a connection between one's trustworthiness, which must be accepted for one to be justified, and reliability. Lehrer postulates that this connection itself, as embodied in the following principle, must be accepted (or at least preferred over its negation) for one to be justified.
(TR) If I am trustworthy in what I accept, then I am reliable in obtaining truth and avoiding error in what I accept. (p. 194)Given that (TR) holds up in the ultra system, the person's commitment to her methods reflected in principle (T) yields the truth that she is reliable in what she accepts. This is the externalist element in Lehrer's theory of knowledge.
Externalist Conditions Not Sufficient for Knowledge
Lehrer makes no further concessions to externalism. "The error of externalism is to fail to notice that the subject of knowledge must accept and be correct in accepting that the internal conditions of trustworthiness are connected with the externalist relations" (p. 195). This requires an epistemic perspective, which only internalism demands.
Lehrer goes on to state that this failure is a motivation for externalism.
Externalism is motivated by doubt about whether what we accept can supply the truth connection. The reason for the doubt is the assumption that it is psychologically unrealistic to suppose that beliefs about our beliefs are necessary for knowledge. (p. 202)The solution is to claim that it is in fact realistic to accept that we are trustworthy and reliable in addition to other things we accept. "Is it unrealistic to suppose that people accept that they are trustworthy" (p. 202). As I understand the reply, Lehrer thinks that the acceptance of our trustworthiness and reliability distills our "beliefs about our beliefs" to one or two sweeping ones. So our minds need not be littered with all manner of beliefs about beliefs.
Lehrer basis his positive case for the need for a perspective in knowledge on what he takes to be counter-examples to externalism. We have already seen one, the case of the diamond-sorter who is able reliably to distinguish real diamonds from fakes but does not recognize that she has this ability. Another example plays against Armstrong's thermometer model of knowledge. Mr. Truetemp has had a "computemp" surgically implanted in his brain, though he is unaware of its existence. This device accurately detects the ambient temperature and causes Truetemp to form a belief about what the temperature is at any given time. He does not investigate his newly-acquired compulsion to note the temperature wherever he is.
Lehrer thinks that Truetemp does not know what the temperature is, despite his reliably-formed true beliefs. The key to this intuition is that Lehrer is trying to account for knowledge in the sense of knowing that one's information is correct, which involves an assessment of the reliability of the source of information. Truetemp has not made such an evaluation, so he lacks knowledge in the requisite sense.
Externalists tend not to think of knowledge in these terms. This can be seen most readily in the fact that they tend to focus on perceptual knowledge, which does not straightforwardly require any kind of assessment, but seems automatic. Thus they think that an appropriate connection to the truth does not require any epistemic perspective. If they are inclined to grant that Mr. Truetemp knows the temperature, it is because they construe knowledge in this sense. He has "animal knowledge," though of a peculiarly bionic sort. And then there is no argument between them and Lehrer.
Externalist Conditions Not Necessary for Knowledge
The diamond-sorter and Truetemp examples are supposed to show that the externalists' conditions for knowledge are not sufficient conditions. Another example is supposed to show that they are not necessary either. More specifically, Lehrer attacks that view that the history of the formation of a belief must be good one in order for it to yield knowledge. It is reasons, not causes, which are necessary to knowledge. To think otherwise is to confuse reasons and causes and to commit "the causal fallacy."
The word 'reason' is undeniably ambiguous. If we ask what is the reason for someone's belief, the answer may describe how the evidence supports the belief or it may describe the process whereby one comes to have the belief. But it is doubtful that epistemologists have been guilty of any fallacy of equivocation. For example, in Alvin Goldman's causal account of knowledge, while it is ventured that "inference is a causal process," all that is actually claimed is that, "if a chain of inferences is 'added' to a causal chain, then the entire chain is causal" ("A Causal Theory of Knowing," Journal of Philosophy 64, 1967, widely reprinted). There is no confusion here.
A related ambiguity can be found in the term 'reasoning.' It may indicate a logical argument that, if valid, supports the truth of a belief. Or it may mean a process of drawing a conclusion from premises. Lehrer states that it is not the process but the validity that is relevant to knowledge.
We reason from our evaluation system to defend what we accept, and this has a causal explanation. Nonetheless, the validity of the reasoning is not to be explained causally but in terms of a relation between the premises and what is concluded. (p. 201)The externalist will certainly not deny that validity cannot be explained causally.
What externalists may claim is that when a belief is produced appropriately, the process by which it is produced just is the relation of evidential support. This seems to be reflected in Goldman's notion of "strong justification" mentioned above. In that case, Lehrer's task is to show that the relation of evidential support is really distinct from whatever causal process produces a belief. "Having the reasons we do may justify the belief, however, even though they have no causal influence on the belief at all" (p. 196). To show this, he builds an elaborate example in which, intuitively, a person has evidential support for a belief which is independent of the way the belief was produced. If we are willing to grant knowledge to the person, it would only be on the basis of the evidential support, so knowledge does not depend on the causal process producing the belief.
The example goes like this. A racist doctor, Raco, came to believe, on the basis of his racism, that members of a certain race are susceptible to a certain disease. Later, having become a doctor, he came upon evidence which justified his belief. Is the bad pedigree of his belief, its unsavory origin, enough to deny Raco knowledge when he is in the possession of the evidence? Most of us would think not.
There is a stock externalist response to this example. As was seen the reliabilist analysis of justification given by Goldman, a belief need not be caused in a reliable way, but only sustained reliably. Raco's belief can then be seen as being buttressed by the new evidence he acquired in medical school. The evidence sustains the belief in the sense that if Raco were to cease to be a racist, he would retain the belief. To this Lehrer responds that the case could be altered so that the belief would not be sustained by the new evidence. For example, Raco's ceasing to be a racist might cause his world-view to collapse, to the point that he would lose all his beliefs. Other suggestions can similarly be overcome by varying the example. It will always be possible that something will interfere with the evidence's sustaining the belief.
As Lehrer describes the case, Raco is justified in accepting that members of a certain race are susceptible to a certain disease, and this is what gives him knowledge. "The belief is the result of prejudice, not reason, but his present acceptance of it is confirmed by reason that provides the justification for accepting the belief. Prejudice gives Mr. Raco conviction, but reason gives him justification for acceptance" (p. 196). It is not clear what exactly Lehrer means here by "accepting the belief." I think the best way of understanding it is as accepting the content of the belief upon evaluation of that content. This is because Lehrer thinks of belief and acceptance as outputs of different systems.
Acceptance and belief may come into conflict and . . . the system of acceptance and the system of belief may pull apart in extreme cases. We find in the case of Raco a person whose belief system and acceptance system are more radically separate than they are for the rest of us. (p. 200)If you understand acceptance as an attitude toward a belief, or even a special kind of belief, I do not think this will affect the case I shall now make on behalf of the externalist. (The general issue was first raised in my 1996 course by my teaching assistant, Richard Schubert.)
An externalist could point out that acceptance the product of a process of evaluation. He could concede that acceptance is required for knowledge but claim that being the outcome of the evaluation process is all there is to being supported by evidence. That is, the externalist position about reasons and belief is simply transposed to apply to reasons and acceptance. Applied to the case of Raco, this means that his acceptance that members of a certain race are susceptible to a certain disease is sustained by the evidence he has for the truth of this. Lehrer may be right in asserting that "Why he believes what he does may be explained by almost anything" (p. 198). But why he accepts what he does can be explained by only one thing: the process of evaluation with the goal of accepting all and only what is true.
Lehrer appears to have an answer to this type of move by the externalist. He states,
Of course, acceptance might be causally influenced by the evaluation system of the person, that is by the acceptances, preferences, and reasonings of the person that justify the acceptance. Once again, however, this does not seem necessary. (p. 200)This seems puzzling. If one's acceptance is not causally influenced by the evaluation system, then how could it be the outcome of an evaluation using the evidence? I think Lehrer does not recognize the significance of the fact that acceptance is the outcome of a process. In explaining his assertion that acceptance is not necessarily causally influenced by the evaluation system, Lehrer points to the role of the evaluation system in justification.
If we study the underlying causal processes, we may find that the evaluation system is a causal factor in the explanation of responses of the subject to critical objections. But what is essential is that the person understands how to meet and neutralize the objections to what he accepts. (pp. 200-201)The kind of externalist I am describing might well dispense altogether with the scheme of justification to which Lehrer appeals. He will simply say that if a person's acceptance of what is true is based on a reliable process of justification, the person has knowledge. The validity of the arguments one uses in reasoning to what one accepts can then be understood as a major factor in the reliability of that reasoning.
It must be pointed out that Lehrer's understanding of what externalism makes "acceptance-externalism" seem impossible. If he is right that the evaluation involved in acceptance requires that one have a view of one's own epistemic position, then knowledge analyzed in terms of acceptance is necessarily internalist. But I have pointed out above that one need not construe externalism this way. Any view that having a a good reason is construed as being the result of a reliable process of rational evaluation should be called externalist.