1. Introduction
A number of competing theories as well as debates have developed around epistemic justification. One of the most representative of these is the internalism versus externalism debate on justification. The basic claim of internalism is that a belief can be justified by the internal state of the holder of the belief. Alvin Goldman (Goldman, 1967) argues that internalism cannot solve the problem of stored beliefs, and that adherence to internalism on the basis of failure to solve the problem of stored beliefs inevitably leads to skepticism. Frederick Schmitt (Schmitt, 1992) argues that the evidence for internalism is of a “concurrent” nature, in that our beliefs about p are based on the belief Q. Alvin Goldman argues that knowledge is caused by external facts. He proposes a causal theory to repair the evidential theory of knowledge. Goldman argues that our perceptions are triggered by corresponding facts. In a similar vein, Robert Rozick’s (Nozick, 1981) Tracing Truth Theory, was presented in The Interpretation of Philosophy. Similarly, Robert Rozick’s truth-tracing theory in Philosophical Explanations argues that belief justification relies solely on what is objectively true, and that the justification of beliefs that are not objectively false is valid as long as we believe them to be true. David Armsrong (Armstrong, 1973) also argues against an internalism view of falsification. In Belief, Truth and Knowledge, David Armsrong also rejects the internalism view of justification. He argues that true beliefs arise from a trustworthy formation process, a view also supported by Frank Ramssey, Peter Unger and Fred Dretske. The difference, however, is that for Armstrong “proof” is not in itself a necessary condition for knowledge.
But the notion of justification itself, which is the focus of the internalism-externalism debate, has not been sufficiently discussed. I believe that clarifying the concept of justification and defining its content is a good way to resolve the debate. In turn, it is also a prerequisite for resolving skepticism about justification.
2. Knowledge Is More than Ture Belief
By contrasting Ayer’s and Gettier’s definition of knowledge, this section will demonstrate how these two definitions vary in essence from Plato’s explanation of knowledge, and what this difference means. In Plato’s Dialogues, Theaetetus, Socrates discusses the nature of “knowledge” with Theaetetus, implying that true opinion is not yet knowledge, and that the discussion between Socrates and Theaetetus identifies knowledge as “true belief + X” but does not specify what “X” is, and thus does not define what knowledge is. Plato clearly does not say for certain what knowledge is, but everyone says that “the traditional definition of knowledge since Plato”2 is justified true belief (JTB). Actually, Gettier himself came up with this definition (using JTB directly) when he questioned the JTB understanding. In fact, this definition (using JTB explicitly) originally came from Gettier himself, who challenged JTB’s theory. and here, let us circumventing the claim that “since Plato philosopher regards knowledge as JTB” and considering only Gettier’s own understanding of knowledge, for the sake of the following discussion, I will place the theoretical model of JTB here along with Gettier’s first counterexample and refer to Gettier’s definition as D1,
D1: S knows that p, if and only if
1) S believes that p;
2) p is true;
3) S’ s belief about p is justified.
The first counterexample contains two propositions,
p1 the person with ten coins in his or her pocket will get the job.
e1 Jones will get the job, and Jones has ten coins in his pocket.
Assume Jones and Smith both attend an interview to compete for a job offer. For some unknown reason, the manager tells Smith that Jones will get the job, and Smith happens to see ten coins in Jones’ pocket, so Smith has good reason to believe proposition e1, and in turn, according to the second principle mentioned by Gettier himself, the principle of closure, Smith can effectively introduce proposition p1. Smith has a clear Justification for believing the true belief p1. Smith, according to D1, is aware of p1. Imagine, however, that the person who receives the job is Smith, and Smith is unaware that he has 10 coins in his pocket. Smith’s knowledge claims still correspond with D1, since (1) p1 is true; (2) Smith believes p1; and (3) Smith’s belief p1 is justified (by e1), but we do not accept Smith knows that p1 (Gettier, 1963).
It is easy to see that in Gettier’s case, both p1 and e1 take the form of propositions, and while we do not reject that the knowledge under discussion here is still propositional knowledge, that is, we can accept that p1 is the content of knowledge and do not accept that e1 can only be a proposition, but Gettier himself and the series of Justification discussions centered on Gettier’s problem implicitly acknowledge that the meaning of justification is to have a good reason to believe3. We know that “reason” is usually used to describe a relationship between propositions, therefore when we said that p is justified by e, we imply that proposition p may be deduced from proposition e. Because Gettier mentions in his paper that the D1 model is based on A.J. Ayer’s (Ayer, 1956), our next question to examine is whether Ayer’s definition of knowledge asked that “Justification” must be a proposition? We will refer to Ayer’s definition of knowledge as D2, then:
D2: a cognitive subject S knows a knowledge proposition p, when and only when
1) p is true;
2) S is sure that p;
3) S has the right to be sure about the truth of p.
A comparison of D1 and D2 reveals that, first, both Gettier and Ayer recognize that p as a proposition of knowledge must be true; second, Gettier equivalence “to sure p” to “believe p” is understandable, after all, when a cognitive subject says “I am sure that…,” the cognitive subject is also reasonably capable of claiming that “I believe”4. Finally, in Gettier’s Justification, “have the right to be sure about the truth” is equated with “justification,” which, as previously stated, in the instance of Gettier, “justification” refers to a certain relation between propositions. The problem is that Ayer himself does not make this condition, since Ayer uses two sets of criteria to determine whether a cognitive subject knows something. The first is that the cognitive subject must givesome proofs for his or her knowledge claim, that is, when he or she asked as “How do you know?” The cognitive subject must be able to demonstrate that his or her knowledge claim is supported by adequate evidence (i.e., have the right to be sure the truth), and the other set is that the subject can always provide true beliefs in a given domain, even if he cannot provide definite evidence when asked how do you know, and Ayer considers both types of knowledge claims to be acceptable. Let us now attempt to summarize Ayer’s concept of what knowledge really is, which I will refer to as D2’, and then:
D2’: S knows p, if and only if
1) p is true;
2) S believes p;
3) S has good reasons or adequate causes to believes p.
According to D2’, we call “having good reasons to believe” as JR and “having adequatecauses to believe” as JC. Comparing D1, we can see that, for Ayer, the X-factor of knowledge (in terms of X itself) has both connotations meaning JC as well as JR, whereas Gettier’s X-factor contains JR only, so that from D2’ Gettier cannot extract the conclusion that X is equivalent to JR, so Ayer himself does not limit X to expressing propositional relations.
Although Gettier’s generalization in D1 limits the meaning of X, X is still necessity. The name of X is not important, and in order to avoid creating a bias, we will not identify X here for the time being, but we will quickly explain that X is actually necessary for knowledge. First, one of the simplest and most straightforward proofs is that true beliefs cannot be sufficient to knowledge, and while Ayer host a loose norm on knowledge, admitting that certain cognitive subjects who cannot answer “how do you know”, but are consistently able to provide true beliefs possess knowledge in certain circumstances, true beliefs that are acquired accidentally through guesswork or other casual reasons or causes are not knowledge. For example, A and B both claim to know the current lottery winning numbers, but A knows the winning numbers for an unexplained reason, and this claim is not reproducible for A. Although B also knows the winning numbers for an unknown reason, B is able to say the winning numbers of the lottery every time, and intuitively, we can admit that B has knowledge about the lottery, but A does not; secondly, if one insists that X only connotes JR, then from the intuition of knowledge generation, there is even more reason to admit that JR is not a necessary condition for knowledge, that is, X is not a necessary condition for knowledge. Because Reliabilism holds that knowledge does not require JR as long as true beliefs are produced by a reliable progress, and this true belief is knowledge, the definition of Reliabilism has two advantages: the first is that it is consistent with our intuitions about knowledge and the process of knowing, and the second is that this requirement expresses an alternative to Ayer’s original qualification of knowledge, so if Reliabilism rejects X by rejecting JR, but actually, reliabilism just rejects JR (though I think that Reliabilism does not really reject JR) and does not reject JC, i.e., Reliabilism does not reject X. Even if we cannot claim that once we reject JR, we can prove that X is a necessary condition for knowledge by denying the antecedent in order to deny the consequent, it is clear that if we admit that X is a necessary condition for knowledge, then X itself, as a necessary condition for knowledge, must mean more than JR, and indeed, as briefly demonstrated above, X is necessary for knowledge, so X must mean more than JR.
Although the objective of this work is not to provide a new model of XTB, in order for my thoughts to be understood, I have to demonstrate that the meaning of X as a philosophical notion is not the same as the meaning of X being used in the process of being used. That is, I suggest that X has both JR and JC meanings, but whether we require JR, JC, or both JR and JC to offer support for knowledge propositions in a particular knowledge proposition is influenced by the circumstances. We will only consider the former problems in this paper, i.e. X should include both JR and JC, one or the other.
3. AS and Epistemic Justification
In this section I will demonstrate that AS is indeed an extreme form of skepticism about JR theory, and that AS concludes that JR is impossible. The reason why AS is only about JR theory is that the AS argument presumed two premises, namely: that all beliefs can only be justified by other beliefs; and that justification cannot be wrong. And these two presumptions serve to qualify JR, a kind of Gettierian conception of justification, so that to avoid the accusations of AS requires a new understanding of justification. The so-called Agrippa’s Skepticism (Williams, 2001) is that the cognitive subject encounters Agrippan’s trilemma in providing justification to beliefs as the content of knowledge, all three of which are “Justification” dilemmas; if beliefs can only be Justified by other beliefs, then the only three possible forms of Justification for all beliefs are: 1) infinite regress; 2) Beg the questions; 3) Dogmatism, and each of these cannot really provide justification for beliefs, while the skeptic therefore argues that justification is impossible.
For example, a detective whose belief B1: “Zhang San is a thief” needs to provide evidence for his belief, if his evidence is B2: Li Si saw Zhang San come out of the place where the theft occurred. Then B2 is the belief the detective uses to Justify B1, but B2 itself needs to be Justified, so that B3: Li Si would never lie; B4 Li Si is an honest person, etc…
First, the infinitely regress Justification approach is to justify B1 by B2, and justify B2 by B3… The reason why infinite regress is not considered a reasonable way of justification is that the purpose of justify B2 is to provide evidence for the truth of B1, and if B2 is true, then B1 gets Justification, but B2 does not get Justification in this case, so this way of infinite regress does not fulfil the purpose of providing Justification, and so this Justification strategy is a failure one.
Second, the strategy of beg the question fails as well, since when we pursue the how do you know B4, the strategy is that B4 is Justified by B3. The technique of the strategy is ineffective because it justifies B3 with B3, which is essentially tautological and does not give any support. It provides no reasons.
Finally, the dogmatismjustification method is inappropriate because when we ask for justification of B3; the dogmatism will tell us that this is and that Li Si never lies. This does not seem to be rationally acceptable.
The above Agrippa’s skepticism seems to be a major issue; however, it is based on two assumptions: 1) all beliefs can be justified only by other beliefs; 2) justification cannot be false. First, let us look at what these two presuppositions are assuming. (a) As stated previously, limiting all beliefs to being justified only by other beliefs means restricting the knowledge proposition p and e, which provides evidence for the knowledge proposition, to a proposition-proposition relationship, i.e., must also be a proposition, and true transmission between e and p can only be achieved by inference, which is an intrinsic requirement of what I call JR. At the same time, none of these three tactics (infinite regress, beg the questions, and dogmatism) are acceptable since we assume a strong, infallible necessity for knowledge and justification, and this infallibility requirement a kind of certainty. If someone asks A, “how do you know B1?” the infinite regress justification approach is that given knowledge proposition B2, if they continue to inquire, “how do you know B2?” the endless regressor will increase B3, and so on. The reason why endless regressor is not a good strategy in AS’s opinion is precisely because neither B2, B3, or B4 can be determined to be true in themselves, but must be justified by other beliefs, and thus can never provide a complete justification of a belief, so no matter how much justification is provided by infinite regress theory, it cannot satisfy the second condition presupposed by AS -infallibility (or certainty). Similarly, beg the questions arguments and dogmatism are both considered by AS to be failedjustification strategies because they do not satisfy the certainty requirement. But the fact that justification of our beliefs about the external world cannot achieve the same strict certainty as mathematics and logic does not mean that there is no knowledge apart from mathematics and logic. So, Agrippa’s skepticism can be avoided as long as we accept that a) Non-belief justification resources can be used to justify beliefs; b) Justification of beliefs about the external world does not require absolute strict precision (infallibility), i.e., if one does not recognize the Gettier theory of Justification (JR), one would not face the accusation of AS.
4. A Justification Theory that Is Consistent with Cause and Reason
Other than JR, what would a theory of justification, as the third element of knowledge, “X,” entail? As previously stated, JR is insufficient for X. Before delving into the new idea of justification, we first look at how the second component of the new theory of justification, Cause, is represented in epistemology. This approach, sometimes known as the causality scheme, was presented by A. Goldman, also known as the causality scheme. The causality scheme seeks to eliminate the traditional idea of Justification, and Goldman himself states that “the type of condition that I would want to be able to add to the classic analysis is this causality.” Let us refer to Goldman’s notion of knowledge as D3, and then:
D3: S knows p when and only when
1) p is true;
2) S believes p;
3) S’s belief p is caused by a fact P.
Goldman contends that D3 resolves the two counterexamples proposed by Gettier himself, beginning with the first, in which Smith does not know p1 (the person with ten coins in his pocket will get the job) because, while p1 is true and S believes p1, there is no relevant fact P that causes Smith to have the belief p1 (Goldman, 1967).
We will not explore whether this version of causality theory concerning the notion of knowledge really answers the Gettier problem since the Gettier problem is not addressed in this study. Goldman sees its own scheme as a form of knowledge and does not require a Justification position. However, in my view, this is not fundamentally different from the previous JTB theory, because JTB was originally a “true belief + X” model following the Platonic principle, and it is clear that D3 still follows the “true belief + X” model, so both JR’s and Goldman’s proposed causality are merely footnotes to Plato’s model of knowledge, and neither of these footnotes can be a sufficient condition for knowledge in terms of theoretical deductions from knowledge theory.
Returning to Ayer’s knowledge analysis, when we make the assertion “We are not just expressing a true belief when we say, ‘I know that’ but a criteria for determining if we really know is whether we can answer the question, “how do you know?” According to Ayer’s analysis, we encounter two scenarios. The first is a cognitive subject that can directly answer “how do you know?” and we assume that the cognitive subject has knowledge, and the second is a cognitive subject that cannot directly answer “how do you know?” but can always give relevant “true beliefs” in a certain domain and we assume that the cognitive subject also has knowledge. Formally, the second situation described by Ayer does not fit into Plato’s model of knowledge of “true beliefs + X”, but in fact it does. If we accept that “justification” should contain both “reason” and “cause” resources (but which resource to use in a specific knowledge proposition depends on the situation), Ayer’s second case is not difficult to explain, as the cognitive subject who can always give relevant information “in a particular domain is not entirely the cognitive subject who is presumably always capable of providing appropriate “true beliefs” in a given domain is not wholly incapable of answering the question “how do you know?” as long as the response is accepted in addition to the propositional form. Assume Lao Wang has rheumatism and constantly has back discomfort before it rains; in reality, based on his physical state, Lao Wang can frequently anticipate p2 “It will rain tomorrow”. What supplies X for p2 in this case is a fact, i.e., the fact gives birth to Lao Wang’s belief regarding p2.
In summary, a good theory of X should include resources at both the reasons and causes aspects, and I think that “justification” (not JR) always meets this condition, since justification originally meant “accepted reason or explanation.” From the literal meaning, justification demands simply acceptability, not infallibility, and an explanation is a description of the current situation, which does not confine all explanations to propositional forms. Such, in a literal sense, justification fully meets a Platonic X criterion, so that, although the word “justification” had hitherto been regarded as merely JR, its real meaning is expected to include both JR and JC. Both in terms of the original meaning of “justification” and the theory of knowledge’s necessity of X-factors, a theory of justification that incorporates both JR and JC are a superior theory of justification than JR.
NOTES
1Hereinafter referred to as “Justification”.
2The theory of knowledge does not refer to an Aristotelian “species plus genus difference” definition of knowledge, but rather to an extension of the sufficient and necessary conditions. For the purposes of this discussion, the analysis of sufficient conditions for knowledge will be referred to as the “definition of knowledge”.
3In Gettier (1963), Gettier himself makes it clear that one of the principles used in his two cases is the principle of deductive closure of knowledge, and although there was some debate about whether to accept the principle of closure, basically the relationship between Justification and Justification is a propositional one as long as the principle of closure is accepted.
4Whether “belief” and “truth” are necessary conditions for knowledge involves a determination of the content of knowledge, which is not discussed in this paper.