PUTTING THE HORSE BEFORE THE CART: A PRAGMATIST ANALYSIS OF KNOWLEDGE
ABSTRACT: The de! nition of knowledge as justi! ed true belief is the best we presently have. However, the canonical tripartite analysis of knowledge does not do justice to it due to a Platonic conception of a priori truth that puts the cart before the horse. Within a pragmatic approach, I argue that by doing away with a priori truth, namely by submitting truth to justi! cation, and by accordingly altering the canonical analysis of knowledge, this is a fruitful de! nition. So fruitful indeed that it renders the Gettier counterexamples vacuous, allowing positive work in epistemology and related disciplines.
KEYWORDS: Justi! ed true belief. Epistemic justi! cation. A priori truth. Justi! cation ad!veritatem. Non-contradiction. Pragmatism. Gettier counterexamples
Since Plato that a promising conception of knowledge is that of justi !ed!true!belief.[2] I say it is promising, because it incorporates the necessary, as well as the suf! cient conditions of knowledge. That its promise has not yet been ful! lled is not due to the concept itself, but to a few, but serious, " aws in its tripartite analysis. Schematically, this analysis, in its now canonical form, is as follows:
(i) !p!is true.
(ii) S believes that p.
(iii) !S is justi! ed in believing that p.
S knows p.
The ! rst problem with this analysis is the adherence to the Rylean distinction between knowledge-that and knowledge-how,3 and the restriction to the ! rst kind, i.e. to knowledge of propositions, namely of propositions ! tting into the structure S believes that ..., where ... can be ! lled in with a wellformed proposition p. Although this is to some extent the cause of the two major " aws discussed in this paper, I will not address it here; suf! ce it to say that I see the Rylean distinction as analytically arti! cial, depicting an unrealistically compartmented picture of human cognition.4 The two major " aws that shall take my attention have a common root: a!priori!truth. This is expressed in the independence of condition (i) from conditions (ii) and (iii) above. The ! rst independence is a mistake, because it implies that propositions are a priori true in an absolute sense; the second, a consequence of the ! rst, is the separation of justi! cation and truth. These two " aws are actually Platonic aspects5 that, once removed or remediated, allow a positive de! nition of knowledge as justi! ed true belief, rendering the Gettier counterexamples vacuous. These " aws are what I call putting the cart before the horse, and what follows is my elaboration on how to bring the horse and the cart to their appropriate positions.
Let us then start with the ! rst " aw: ! rstly, is there actually anything like a priori truth? Tarski showed that the answer is af! rmative, but he also showed transcend this world in order to attain its truth. It is this transcending the physical world that Plato elsewhere, but, interestingly enough, not in the Theaetetus, saw as the required logos, or justi! cation.
3 Cf. RYLE, G. The!Concept!of!Mind. Chicago: Chicago University Press, 1949.
4 For some remarks on this, see AUGUSTO, L. M. Unconscious knowledge: A survey. Advances!in! Cognitive!Psychology!6, 116-141, 2010. For an elaboration on this rejection, see, for instance, STANLEY, J. ; WILLIAMSON, T. Knowing How. Journal!of!Philosophy 98:8, 411-444, 2001.
5 See note 2.
that the semantic conception of truth[3] is literally of no interest to epistemology at large, contradicting the ambivalent fascinating effect it understandably has upon epistemologists of a Platonic vein. If epistemology is concerned with the conditions of knowledge of a subject interacting with the world of which s/he is a part, then a solely semantic approach to truth will simply not do.
Tarski called his de! nition of truth semantic precisely because it applies exclusively to sentences within a speci! ed language: let p be any declarative sentence[4] of a language L; then the truth predicate for p is not true simpliciter, but truein-L. This, L, is a language whose structure has been exactly speci! ed: broadly,[5] it is a semantically non-closed language consisting of an object- and a metalanguage, the latter containing the former as a part (or the former allowing of being translated into the latter); the main role of the metalanguage (L1) is that of allowing the construction of a!name!for every sentence of the object-language (L0)!so that truth, in!the!metalanguage, can be de! ned as what he termed Convention!T:
(T) X!is true <in!L1> iff!p!<in L0>
where X is the name of the sentence p (e.g.: snow is white is the name of the sentence snow!is!white). Graphically,
|
Snow is white is true iff L0 |
L1
The terms true and iff belong exclusively to the metalanguage, this being forcibly richer than the object-language if it is, among other important issues,[6] to de! ne truth as a purely deductive matter. Indeed: given that L0!need not be a natural language, in which speakers form sentences about reality, the sentence snow!is!white!has no relation whatsoever with the world at large. While semantic!notions such as designation, satisfaction, and de! nition express relations between expressions and the objects referred to, the notion of truth is purely logical; it expresses a property of sentences, but because these involve semantic notions, as seen, truth is fruitfully treated as a concept of semantics. What are the gains of this? Material adequacy and formal correctness,[7] and not a bit of anything else; in other words, no truth as we search for it in the world:
In fact, the semantic de! nition of truth implies nothing regarding the conditions under which a sentence like (1):
(1) snow!is!white
(2) can be asserted. It implies only that, whenever we assert or reject this sentence, we must be ready to assert or reject the correlated sentence (2):
(3) The!sentence!snow!is!white!is!true.
And the coup de grâce follows:
Thus, we may accept the semantic conception of truth without giving up any epistemological attitude we may have had; we may remain naïve realists, critical realists or idealists, empiricists or metaphysicianswhatever we were before. The semantic conception is completely neutral toward all these issues.[8]
Saying this less brutally, the semantic conception of truth has no bearing on reality as we perceive it; thus, Convention T sentences in the Tarskian conception yield no knowledge other than of L0.
Despite Tarskis reluctance,[9] sentences like (1) can undoubtedly be valuated within a Convention T perspective of truth in a natural language like English or Portuguese, in which cases the sentence snow!is!white is true-in-English but nottrue-in-Portuguese, and the sentence a!neve!é!branca is true-in-Portuguese, whereas it is not-true-in-English. It is easy to see that we say nothing whatsoever about snow/neve in the world, but merely about the adequacy and correctness of the sentences (1) and (2)!in English and in Portuguese taken!qua!speci !ed!languages.[10]!But our subject, interacting with the world, needs a language that speaks!reality,[11]!i.e. a language in which his/her main concern is to express (de! ne, designate, symbolize, mimic, etc) reality of which, as a speaker, s/he is an unalienable part; as such, S! cannot take a single step without forming beliefs about the world, but S!does not, to that end, go about collecting ps that are a priori true; outside a speci! ed language, sentences are not true or false independently of their being believed, for the simple reason that there might be numberless sentences about!the!world!that may come out true (or false) once believed.The sentences <stating the facts> Arfhs live in Orion, 7~y|*2, or Quarks dont like anti-quarks might all be true for aught we know; they might well be part of the vast number of facts of the universe,[12] but this is wholly irrelevant until someone, say S, believes that Arfhs live in Orion, that 7~y|*2, and that quarks dont like anti-quarks. And the same is valid even for those propositions seen as a priori, or analytically true; for instance, and against platonism,! 2 + 3 = 5 is a true proposition once it is believed by a subject who is in possession of a certain mathematical system; it is within this system that is in itself a set of beliefs that this proposition is true and a priori so, a priori meaning simply that the system of beliefs makes it that it is true in it.
Summing up: sentences are a priori true (here a priori in the Platonic sense of intellectual entities existing per se) within what Tarski called speci! ed languages; when it comes to the world at large, sentences can no longer be true, let alone a priori true, outside the beliefs that contain or express them. In this scenario, and taking it that necessarily some subject S holds the belief p!that snow is white/a neve é branca, this belief can only be valuated as true if in the actual world where S!lives there is such a thing as snow/neve, which s/he calls snow/neve, and if it is white/branca. In this world, namely on planet earth, snow/neve actually is white/branca,[13] and so anyone believing that p can be said to hold a true belief.
Note that so far the following alterations have been made in the tripartite analysis of knowledge:
(i) S believes that snow is white.
(ii) The belief that snow is white is true.
[...]
But beliefs are psychological phenomena, mental attitudes regarding the world. This poses a major problem concerning the question of truth that does not exist in the semantic approach. In fact, in the latter all it takes to assert that X is true is to be able to assert p, X being the name of p, an arbitrary sentence, and one can assert p if it is either an axiom, or a theorem of the formal language in question. Truth is thus a logical relation between two sentences in which the sentence X is true is equivalent to the sentence p; this is what Tarski calls equivalence of the form!(T).[14] Nothing in the world gives us such assurance regarding the truth of our beliefs, unless!we!are!actually!justi !ed!in!holding!them!as!true. Let us now turn our attention to the second " aw in the canonical analysis, the separation of condition (i) from condition (iii).
It is very well to state that Ss belief that snow is white is true when referring to the world, but how can S be assured of the truth of her/his beliefs about the world? Seemingly, the ! rst step for S!to truth-valuate conclusively these beliefs that s/he begins by holding <as true> is to go out in the world and check, i.e. empirically test (see, taste, touch, etc), and verify, given the conditions of the world, and perhaps with assistance from other already valuated beliefs, that they are indeed true beliefs. However, the world (which includes S) is such that S!will not always, only rarely, or even not at all be able to prove that his/her beliefs are true; this is to say that there is no veri! cation method to compare ones beliefs about the world with the world itself.
Nevertheless, the world justi! es, or not, Ss beliefs: if they are justi! ed, then they are true.
Before clarifying this statement in all due detail, let us see how this is a better view than that which makes truth a priori, and independent from justi! cation. Given that I simply cannot state the truth or falsity of the proposition Arfhs live in Orion, I shall opt for it being true; because truth precedes justi! cation, this could well be so, without my bothering to verify whether it is the case or not; for the sake of the argument, let us agree that it is so:
(i) Arfhs live in Orion is true.
(ii) John believes that Arfhs live in Orion.
(iii) John is justi! ed in believing that it is so.
John knows that Arfhs live in Orion.
How is John justi! ed in his belief that Arfhs live in Orion? The best justi! cation seems to be that it is so because the proposition Arfhs live in Orion is true. But this is the Achilles heel of this analysis in that, in!it, justi! cation may have nothing, or very little to do with truth itself, as the Gettier counterexamples are believed to show: easily, one can believe in a true proposition and, alas, not be justi! ed in doing so. If Johns justi! cation is considered invalid (he read about Arfhs living in Orion in a science ! ction book; he dreamt it; a shaman told him so; he was under the effect of hallucinogenic drugs when he formed this belief; etc), then, amazingly, the fact that the belief that Arfhs live in Orion is true does not yield knowledge. We then have the paradox that Arfhs actually live in Orion, John truly believes that they do, but he does not know that.
Let us leave this unfamiliar example of the Arfhs for a better known one; Henry and the barns illusion[15] will do perfectly: Driving along a country road and, seeing façades of barns (only he does not know that they are mere façades), Henry truly believes that there are barns in that region; seeing a particular façade of a barn that actually is a barn, he believes that he sees a barn, that it is a barn that he is seeing, just another among all the others he has been seeing in the last few miles. Yet, according to the canonical analysis, he does not know that it is a barn, because then he just got epistemically lucky: he did not have a good or valid justi! cation. But it is a barn, and he believes it is a barn.
Henrys belief, though true, has no positive!epistemic!status, as Chisholm would put it.[16] Now, one can legitimately ask if there is actually any instance in which ones beliefs have positive epistemic status; after all, it appears to be the case that one is justi! ed in holding beliefs that are a priori true by mere chance, given that ones perceptive and reasoning apparatus is simply not designed for such an accomplishment (after all, contrarily to Platos belief,[17] it is very likely that we were not made acquainted with a priori truths in the huperouranos, aka world of ideas). If it were, then, given a normally functioning cognitive apparatus, we would always, or most of the time, be justi! ed in holding beliefs simply because they are true. But neither our perceptive nor our reasoning faculties are designed for a priori truth: for instance, we still see a stick partially submerged in a more or less transparent liquid as broken even when we are aware of physical reasons explaining that the stick is not actually broken, and by the sole use of reason, without recourse to empirical veri! cation, we very likely would never have an explanation that appears to be good for this phenomenon. In other words: both internalism and externalism are wrong theories of justi! cation, or of knowledge simpliciter, if they stick to the faulty tripartite analysis of knowledge. And a super! cial survey of the literature will show that they do.
The fact is that this notion of justi! cation, in the face of a priori truth, falls prey to all demons and tricks. The internalists, so blaringly Platonic that they require that their subjects be at any time aware (or capable of becoming aware) of the internal justi! cations for their beliefs, see their core assumption thrown to the bin by such playful characters as the Cartesian demons: how can we ever be sure that they are not tricking us into believing that false propositions are true? If you are a Christian, God may enter the scene and save you from a failure in justifying your belief that 2 + 3 = 5 is a true proposition, just because he is too good to want to trick anyone (otherwise, as Descartes claimed, he would not be God), but if God has no part to play in your philosophy, then you are not to be rescued, and your internal justi! cations are prey to all kinds of predators. The externalists should not rejoice, however; their notion of justi! cation is just as fragile when faced with a priori truth. Let us visit one contemporarily prominent faction of externalism, reliabilism, the view according to which one, say S, is justi! ed in holding the beliefs that are produced in S by his epistemic faculties working properly, as Plantinga phrases it.[18] Let us take the same true proposition, 2 + 3 = 5. When dreaming, one apparently has beliefs, beliefs that are produced by ones epistemic faculties (I infer that this means faculties capable of conducing or contributing to knowledge): dreaming is a well-documented phenomenon that appears to be essential for unknown reasons; it is most puzzling precisely because it seems to trick us into all sorts of wrong beliefs, but there is abundant evidence that epistemic faculties are often involved (for instance, there have been many reports of solutions of mathematical problems being found in dreams). Because it is a phenomenon brought about by ones epistemic faculties working just ! ne, then one is justi! ed in holding the beliefs one holds when dreaming. Nevertheless, the following scenario looks rather unlikely:
(i) 2 + 3 = 5 is a true proposition.
(ii)!! S believes that 2 + 3 = 5 is true.
(iii) !S is justi! ed in this belief because s/he got it in a dream.
!!!!!!! ! S!knows that 2 + 3 = 5.
The problem, both for internalism and externalism, is the separation of truth from justi! cation. Simply, this notion of justi! cation, which I propose to call justi !cation!post!veritatem, does not do justice to the basically serviceable de! nition of knowledge as justi! ed true belief. What follows is my attempt to replace this notion of justi! cation with one that is inseparable from that of truth itself, thus eliminating the cause of the apparent success of the Gettier counterexamples.
Going back to Henry and the barn scenario, let us now change the order of the conditions in the canonical tripartite analysis, inserting, in italics, new elements conducing to the elimination of the Platonic aspects above:
(i) Henry believes that the building whose façade he sees is a barn.
(ii) He is justi! ed in believing that it is a barn (because!it!actually!is!a!barn).
(iii) The belief that the building whose façade he sees is a barn is, thus, true. John knows that the building whose façade he sees is a barn.
It is Henrys being justi! ed in believing that he is looking at a barn that makes the proposition/belief this is a barn true. Let us call this kind of justi! cation, to oppose it to the one above, justi !cation!ad!veritatem. The basically far-fetched thought experiments that make Henry fail to be justi! ed lose their science ! ction appeal, and we are back to the serious domain of rigorous analysis.
What is the pay-off of this notion of justi !cation!ad!veritatem? Actually, there are at least three things to be gained: ! rstly, we get rid of the embarrassing need to prove true propositions regarding the world at large, and I say embarrassing, because the notions of provability and truth simply do not coincide, not even in mathematics, which means that we always end up with more true propositions than we can prove;[19] secondly, and particularly when actual propositions are concerned, we actually show that we care for their truth value, contrarily to the canonical analysis that contemplates no such problem (precisely because it considers propositions that are a priori true): in our case, in the very analysis of knowledge a proof of the truth of beliefs is required. Nevertheless, and thirdly, we do not fall into endless debates around truth that, though intricate enough, have brought no progress to philosophy; truth as coherence, correspondence, or even identity, to name but the most querulous stands with regard to truth, cease to interfere with progress in epistemological matters.
This said, let us then see how this concept of justi !cation!ad!veritatem ful! ls the promise of the concept of knowledge as true belief. First of all, it is interesting to remark that I partially borrowed it from a logical conception meant to be of use in an epistemic!logic!with!justi !cation,[20] the major difference being that while its conceptors aim at using it together with the modal conception of truth, I see it as capable of, if not altogether replacing truth, at least submitting it to the required justi! cation in the concept of knowledge at issue. This concept of justi! cation can be summarized in the following three main points:
(1) every belief[21] is amenable, in principle, to justi! cation;
(2) any justi! cation of a belief is compatible with any other justi! cation of the same belief;
(3) justi! cation of a belief entails its truth, and, thus, it implies knowledge.
Elaborating:
(1) Our beliefs concern the world, and they cannot go beyond it. Thus, the world justi! es, or not, our beliefs; the world at large is the theatre of the justi! cation of our beliefs. If S holds the belief that snow is white, then S is either justi! ed in holding this belief because in the world snow is actually white, or S is not justi! ed because snow is of some other colour. Subsystems of the world (e.g.: mathematics, the self, etc) are still part of the world, and thus there is no justi! cation outside the world.
Fundamental for issues in the philosophy of science, (2) simply states that a justi! cation for a belief is compatible with all other justi! cations for the same belief. For instance, 2 + 3 = 5 is a justi! ed belief given the entire edi! ce of mathematics, but it is also justi! ed by the world at large in that if to two oranges one adds three more, then one has ! ve oranges. This at the same time emphasizes the fact that the world is the theatre of the justi! cation of our beliefs, as well as the fact that in order to be justi! ed our overlapping systems of beliefs have to match: if relativistic physics actually is justi! ed, and so is classical physics, then the justi! cations for these systems of beliefs (theories, if you will) have to be compatible. But this also means that no subsystem of beliefs exhausts the explanation and/or the description of the world. As a matter of fact, compatibility of justi! cations can be such as to be simply identity, i.e. two apparently diverse justi! cations for the same belief might actually be one and the same justi! cation (e.g.: 2 + 3 = 5 appears to be differently justi! ed in case the belief is purely mathematical or if one is referring to oranges, but, are there really two different justi! cations, or only one?).
Assumption (3) yields that any justi! ed belief should be true, and that it should therefore yield knowledge. Clearly, more needs still to be said about both justi! cation and truth, and it would perhaps be advisable to begin by carefully distinguishing the new epistemological!notion of truth in play here from the old metaphysical one: while this tries to establish the connection between the mind and the world (correspondence and identity theories of truth) or the internal coherence of the body of beliefs about the world (coherentism theory of truth), the former simply reposes on ones beliefs being justi! ed by the world. We saw above that every belief is amenable to justi! cation in that it necessarily is about the world and must be justi! ed by it; however, we are not endowed with the faculty of being capable of verifying that our beliefs actually correspond, or are identical to the world, or that they cohere in such a way as to make them true. What we can nevertheless claim is that we have strong reasons to consider true a belief that is justi! ed in that it does!not!contradict!the!world.
But we need to re! ne this concept of non-contradiction as the very heart of epistemic!justi !cation: although it is inspired in the mathematical concept, we have already put aside proof as an unfruitful requirement with respect to justi! cation; having distanced ourselves from any restricted notion of non-contradiction, we can simply state that a belief does not contradict the world when it!works.[22] It is wholly irrelevant whether beliefs work because they correspond to the facts in the world, or because they cohere with our web of beliefs, or even because they are identical to the very facts of the world themselves. They might, or they might not. What is of import is that our beliefs do not contradict the world in that they actually promote fruitful relations with and in it. For instance, farmers plant vegetables at speci! ed times of the year they believe they should be planted at because this belief works, i.e. they actually succeed in securing crops by acting on this belief. Therefore, this belief does not contradict the world, namely that part of it that has to do with vegetables. Do they for this know!that those are the correct times to plant the different vegetables? In that they are justi! ed in believing this, their beliefs are true, and they know something about vegetables and their cultivation.
But given that knowledge depends on justi! cation, and given that this is the non-contradiction of the world, knowledge can be revisable, just because justi! cation is: in other words, the external world is in constant change. Given certain climatic changes, the beliefs above may have to be revised once they are veri! ed to contradict the world. There will be a period of trial and error until one gets the right beliefs again, and this is so because new beliefs are undergoing the process of justi! cation.[23] This scenario is in nothing different from that of science: natural philosophers ceased to believe in the phlogiston when the belief that there was such an element in the world appeared to be unjusti! ed because the caloric theory seemed to work better, and this was dethroned by contemporary thermodynamics because this, as a belief, works even better. Here works better is simply does not appear to contradict the world, or does not contradict the world to such an obvious extent. The fact that phlogiston was thought to be an element in combustible bodies justi! ed the belief that, once burned up, they would have lost mass; the Russian 18th-century scientist Mikhail Lomonosov showed that this belief was not justi! ed in that it contradicted the world (it was not the case that the mass of the burnt metals he experimented with decreased).
There is then no room for wild scenarios where S is not justi! ed in believing in propositions that are true and thus fails to have knowledge. I am, among others, referring to the already mentioned Gettier counterexamples, to which I am now, ! nally, ready to pay due attention.
E. Gettiers counterexamples, published in 1963,[24] sparked abundant work in epistemology aimed at rescuing the canonical analysis of knowledge from what has become known as gettierization, the lack of coordination between conditions (i) and (iii); most of this work concentrated on condition (iii), to a great extent neglecting the other two conditions. No one that I know of realized that while one accepts a priori truth in an absolute sense, there will always be cases in which subjects fail to have knowledge in spite of holding true <propositions in their> beliefs. This means the collapse of the tripartite analysis of knowledge, and the dismissal of the de! nition of knowledge as justi! ed true belief, which, as I claim, is unfounded.
At ! rst sight, Gettiers counterexamples work by showing that one can hold a true belief and yet fail to have knowledge because of failure in justi! cation, but actually they work because the notion of a priori truth in an absolute sense allows him to play at will with radically different notions of truth and justi! cation. I next explain this assertion. Let us see the so-called counterexamples:
Gettier!Case!I.
In the ! rst counterexample, Smith, an applicant for a job, holds a true belief that, alas, he inferred!from a false one, which, as Gettier rightly sees it, makes him unjusti! ed in holding this belief in!a!logical!context. In detail, Smith believes that another applicant for the same job, Jones, will get the job, because he has strong evidence for the following conjunctive proposition:
(1) Jones is the man who will get the job, and Jones has ten coins in his pocket. Naturally, Smith infers from (1) that
(2) The man who will get the job has ten coins in his pocket.
It so happens that, unbeknownst to himself, he, too, has ten coins in his pocket, and he, Smith, and not Jones, actually gets the job. This means, according to Gettier, that while his belief (2) is!true, he is!not justi! ed in holding it, and thus ends up without knowledge. But this works precisely because Gettier mixes a merely logical notion of justi! cation (if, of course, there is such a notion in logic) as correct inference with a non-logical, factual notion of justi! cation, i.e. Smith is factually!justi! ed in believing (2) because it so happens that in the real world the man who gets the job has ten coins in his pocket, while he is not logically justi! ed in believing it because he inferred it from (1), which is a false proposition. Moreover, Gettier appeals to two notions of truth that we can see as also factual and logical! truth: (2) is factually true, while (1) is logically!not-true, or false (the conjunction of a false P!and a true Q!is a false proposition). This is a mess, for, regarding proposition (2), the situation is as follows:
(i) The man who will get the job has ten coins in his pocket is factually!true.
(ii) Smith believes that the man who will get the job has ten coins in his pocket.
|
(iii) Smith is factually justi! ed in his belief. |
(iii) Smith is logically not justi! ed in his belief. |
|
Smith knows that the man who... |
Smith does not know that the man who... |
Smith does not know that the man who ...
Gettier!Case!II.
Again, Gettier appeals to logical as well as to factual notions of justi! cation and truth. Now, Smith appears to be justi! ed (he is said to have strong evidence) in believing that
(3) Jones owns a Ford.
For the sake of the argument, we are asked to accept that, having no idea where his other friend Brown is, Smith forms the following beliefs:
(4) Either Jones owns a Ford, or Brown is in Boston.
(5) Either Jones owns a Ford, or Brown is in Barcelona.
(6) Either Jones owns a Ford, or Brown is in Brest-Litovsk.
Smith is said to realize that (4) to (6) are entailed by (3) as P % (P! !Q).[25] It so happens that Jones does not own a Ford but Brown actually is in Barcelona, or, formally, ~P % (P! !Q), which, just like the above, is a true conditional, in case both P and Q are not false. Thus, (5) is logically!true, and logically nothing hinders Smith from making this entailment, but Gettier claims that he is factually! not justi! ed in holding this belief because (3) does not correspond to the facts: Jones does not own a Ford. This means that while in counterexample I logical justi! cation was the strongest, in counterexample II factual justi! cation takes the lead. This puzzling state of affairs is only possible, as said above, because if conditions (i) and (iii) are independent, then one can play with them, namely by varying the criteria for justi! cation and truth at will, as the following additions (in italics) to his original text (cf. op. cit.!p. 121) show that Gettier does:
First, in that sense of justi! ed in which Ss being justi! ed in believing P is a necessary condition of Ss knowing that P, it is possible for a person to be factually justi! ed in believing a proposition that is in fact logically false. Secondly, for any proposition P, if S is factually!justi! ed in believing P, and P entails Q, and S deduces Q from P and accepts Q as a result of this deduction, then S is logically! justi! ed in believing Q.
In the view defended in this text, however, there is nothing in Case I that makes Smith fail to have knowledge: Smith believes (2) and this is a true belief because he is justi! ed in holding it. Regarding proposition (1), it is a logical construct (a complex proposition) that rarely, if ever, corresponds to real beliefs; that is to say that Smith would almost certainly separate (1) into two independent propositions, or beliefs. Does Smith know, in Case II, that Brown is in Barcelona? If he actually! believes it, and not merely guesses it, then, again, he has knowledge in that his belief is true because justi! ed. But what is to be done of the disjunctive belief contemplated by Gettier? As in the case of proposition (1), nothing; that is, nothing outside a logical context, for, again, it seems far-fetched to pretend that Smith is incapable of unmaking the disjunction (if, of course, he would ever make it, to begin with).
The Gettier counterexamples are so only within an analysis of knowledge that puts the cart before the horse, i.e. makes truth precede justi! cation. With these cases, he only revealed what the " aw of this Platonic analysis is: one can have unjusti! ed true beliefs. He showed thus that this analysis is irredeemable,[26] but he actually gave no alternative to that unfortunate state of affairs, succeeding in keeping generations of epistemic Platonists occupied with trying to rescue their sine qua non condition (iii). Unless he himself was a Platonist in this sense, one fails to see why he should have kept them toiling in vain. [27]
I wish to thank the Portuguese Foundation for Science and Technology (FCT), as well as the European Social Fund, for the postdoctoral fellowship that made this work, among others, possible.
RESUMO: a de! nição de conhecimento como crença verdadeira justi! cada é a melhor que possuímos actualmente. Contudo, a análise tripartida do conhecimento que podemos dizer canónica não é a mais apropriada para a sua defesa, devido a uma concepção platónica de verdade a!priori que põe a carroça à frente dos bois. Dentro de uma abordagem pragmática, defendo que esta de! nição é de facto frutuosa, se (1) eliminarmos a verdade a!priori, nomeadamente pela subordinação da verdade à justi! cação, e (2) procedermos às alterações consequentes nesta análise canónica. Com efeito, esta de! nição passa a ser tão frutuosa que torna irrelevantes os contraexemplos de Gettier, permitindo, assim, um trabalho positivo em ! loso! a do conhecimento e nas disciplinas com ela relacionadas.
PALAVRAS-CHAVE: Crença verdadeira justi! cada. Justi! cação epistémica. Verdade a! priori. Justi! cação ad!veritatem. Não-contradição. Pragmatismo. Contraexemplos de Gettier.
ARTEMOV, S.; NOGINA, E. On Epistemic Logic with Justi! cation. In: PROCEEDINGS OF THE 10th CONFERENCE ON TARK, 2005, p. 279-294.
AUGUSTO, L. M. Do unconscious beliefs yield knowledge? Revista!Filosó !ca!de!Coimbra 35, 161-184, 2009.
AUGUSTO, L. M. Unconscious knowledge: A survey. Advances!in!Cognitive!Psycology, 6, 116-141, 2010.
CHISHOLM, R. Theory!of!Knowledge. Englewood Cliffs, NJ: Prentice Hall, 1966.
DEWEY, J. Valuation and Experimental Knowledge. The!Philosophical!Review, 31:4, p. 325351, 1922.
FLORIDI, L. On the Logical Unsolvability of the Gettier Problem. Synthese 142, p. 61-79, 2004.
GETTIER, E. L. Is Justi! ed True Belief Knowledge? Analysis 23, p. 121-3, 1963.
GOLDMAN, A. Discrimination and Perceptual Knowledge. The!Journal!of!Philosophy 73:20, p. 771-791, 1976.
HAACK, S. The Pragmatist Theory of Truth. British!Journal!for!the!Philosophy!of!Science 27, p. 231-249, 1976.
JAMES, W. Humanism and Truth. Mind 13:52, p. 457-475, 1904.
JAMES, W. Pragmatisms Conception of Truth. The!Journal!of!Philosophy,!Psychology,!and! Scienti !c!Methods IV:6, p. 141-155, 1907.
PEIRCE, C. S. Some Consequences of Four Incapacities. Journal!of!Speculative!Philosophy 2, p. 140-157, 1868.
PLANTINGA, A. Positive Epistemic Status and Proper Function. In: TOMBERLIN, J. E. (Ed.). Philosophical!Perspectives!2, Atascadero, CA: Ridgeview Publishing Company, 1988, p. 1-50.
PLATO, Phaidôn. Trad. P. Vicaire. Paris: Les Belles Lettres, 1983.
PLATO, Politeias (IV-VII). Trad. E. Chambry. Paris: Les Belles Lettres, 1970.
PLATO, Theaitêtos. Trad. A. Diès. Paris: Les Belles Lettres, 1963.
RYLE, G. The!Concept!of!Mind. Chicago: Chicago University Press, 1949.
STANLEY, J.; WILLIAMSON, T. Knowing How. Journal!of!Philosophy 98:8, p. 411-444, 2001.
TARSKI, A. The Semantic Conception of Truth, and the Foundations of Semantics. Philosophy!and!Phenomenological!Research 4:3, p. 341-376, 1944.
Recebido em: 17 de maio de 2011
Aprovado em: 26 de junho de 2011
[1] Instituto de Filoso! a, Faculdade de Letras da Universidade do Porto, Via Panorâmica, s/n, 4150-564 Porto, Portugal. E-mail: luis.ml.augusto@gmail.com. Luís M. Augusto completed a Ph.D. in the history of philosophy at the Université Paris IV Sorbonne (Paris, France) in 2006. He is a postdoctoral researcher at the Institute of Philosophy of the University of Porto and a postdoctoral fellow of the Foundation for Science and Technology of Portugal (FCT). His research focuses on human cognition, with emphases on unconscious cognition and abnormal mental processes (especially psychosis). In this context, he is also interested in the use of some philosophical concepts, such as knowledge and belief, in cognitive science and other scienti! c disciplines.
[2] As a matter of fact, in the Theaetetus, Plato does not argue in favour of what appears to be the ! nal of a short list of candidates for knowledge, to wit, what we today call justi! ed true belief (he called it true belief with logos), because in it he implicitly emphasized what in other places he vehemently defended: the claim that knowledge is solely of that which wholly transcends the physical world (cf. the analogy of the divided line in Rep. VI, 20, 510d 21, 511e). This is the idea, or, in other words, the essence of each and every thing. To speak of the essence of a thing, because it is that which each thing ! nds itself being (cf. Phaedo 65d-e; all translations are mine), is the same as to speak of its truth, and to know a thing is thus only possible when one has captured its essence. The central thesis of the Theaetetus, a thesis lost in the dialectic that aims at reaching an aporia, is that there is knowledge when the essence and the truth [of a process of reasoning on the impressions][ ] can be attained (Theaetetus 186d), which, of course, is never. The rationale of the Theaetetus is purely negative: no knowledge can come from sense data, because truth is prior to the material world. Thus, epistemic Platonism, grounded on a metaphysical Platonism that ! rmly establishes a priori the essence, or truth of the objects of the physical world, is the ambition that the human psuchê can
[3] Namely as expounded in TARSKI, A., The Semantic Conception of Truth, and the Foundations of Semantics, Philosophy!and!Phenomenological!Research 4:3, 341-376, 1944. I do not directly discuss the 1933 paper in which Tarski originally, in Polish, elaborated on this semantic concept of truth; the 1944 paper not only clari! es this concept, but it also addresses criticisms from philosophical quarters, thus making it more appropriate for my purposes. A note on a priori truth is called for: There are, indeed, other more widely accepted examples of a priori truth (e.g.: all bachelors are male, circles are round, 2 + 3 = 5, etc), but they necessarily refer to the world, cause of their disputed a priori character. Contrarily, Tarskis semantic concept of truth is wholly a priori in that Convention T sentences refer to nothing at all in the world. As I see it, this, and this alone, is a priori truth.
[4] Tarski uses sentence instead of proposition, a term that he ! nds too inexact; by sentence, he means a declarative sentence (cf. op!cit., p. 342), and throughout this text I use the term proposition with this meaning, too.
[5] For the details of this speci! cation, or formalization, see ibid., p. 346.
[6] E.g., the avoidance of paradoxes and antinomies such as the antinomy of the liar, which seemed to worry him the most.
[7] These are explained by the medieval logic terminology of the supposition. As brie" y as possible, the sentence snow!is!white, or the sentence constituted by three words, the ! rst of which consists of the 19th, 14th, 15th, and 23rd letters, the second of the 9th and 19th letters, and the third of the 23rd, 8th, 9th, 20th, and 5th letters of the English alphabet (snow is white occurs here in a suppositio!materialis) is!true!iff!snow!is!white (snow is white occurs here in a suppositio!formalis). Let any arbitrary sentence be replaced by the letter p and let the name of this sentence be represented by the letter X; let us call this equivalence!of!the!form!(T). Then, a sentence is materially!true if it can be asserted as an equivalence of the form (T), and any such assertion is formally correct; cf. op. cit., p. 343-4.
[8] Ibid., p. 361-2.
[9] Cf. ibid., p. 347.
[10] By proceeding as suggested by Tarski, ibid., p. 347, i.e. by replacing a portion of the natural languages English and Portuguese by languages whose structure is exactly speci! ed.
[11] I am here taking liberties with the transitiveness of the verb to speak in the belief that speaking a language is a means of speaking reality in that language is precisely the attempt to create a one-to-one mapping between words/sounds and reality. To be more precise, when speaking a word, one speaks its reference.
[12] I am using universe, world, external world, and reality as synonyms.
[13] Let us just accept this without further ado for the sake of the argument.
[14] See note 10 above.
[15] Cf. GOLDMAN, A. Discrimination and Perceptual Knowledge. The!Journal!of!Philosophy 73:20, 771-791, 1976.
[16] Cf. CHISHOLM, R. Theory!of!Knowledge. Englewood Cliffs, NJ: Prentice Hall, 1966. It is perhaps important to note that, in Chisholms eyes, the case is that Henry did not ful! l his epistemic!duty of warranting that his justi! cation is waterproof. This is so because Platonism does not end at an ontological transcendence of truth: that Truth is, because it must necessarily be the case that such a thing be in the universe, and that it is ! xed before this world where living organisms strive for knowledge as a means to survival, makes of knowledge, in the case of rational beings, not merely a good possession, but an obligatory one. Platos analogy of the divided line is, again, the most direct source for this perspective, in that whereas the sun is the means by which the physical things are made visible, it is through the attainment of the <Idea of> Good alone that the purely intellectual forms yield knowledge in human reason. According to this view, it is the moral duty of a rational being to attain knowledge, which s/he can do only by attaining truthif not truth itself, then true thingslet them be true propositions for lack of better truths.
[17] Cf. Phaedrus 245c-250a; ibid. 253c-e.
[18] PLANTINGA, A. Positive Epistemic Status and Proper Function. In: TOMBERLIN, J. E. (Ed.).
Philosophical!Perspectives!2, Atascadero, CA: Ridgeview Publishing Company, 1988, p. 46.
[19] TARSKI, A., op. cit., p. 372, n. 17.
[20] ARTEMOV, S.; NOGINA, E. On Epistemic Logic with Justi! cation. In: !PROCEEDINGS OF THE 10th CONFERENCE ON TARK, 2005, p. 279-294.
[21] I am here getting rid once and for all of the distinction between belief and proposition paraded by the canonical analysis of knowledge: p!and the belief that p!are one and the same thing. Take, for instance, the proposition Henry VIII married six times; taking it that beliefs are verbal in essence, in normal circumstances one will express the belief that Henry VIII married six times simply as Henry VIII married six times. Expressions such as I believe that, I think that, etc, are used when stressing ones beliefs, not when normally holding them. By this, I am not excluding p from epistemological discourse; it is a highly useful symbol, namely for formal manipulation.
[22] This is my main point of contact with pragmatism. I am not here saying what pragmatism is; I am solely arguing for my analysis of knowledge, which is avowedly inspired by pragmatism to the point of being a pragmatist theory. However, for those less familiar with pragmatism, some bibliographical indications are in order. As for the seminal texts, the following are, in my view, indispensable readings: PEIRCE, C. S. Some Consequences of Four Incapacities. Journal!of!Speculative!Philosophy 2, 140157, 1868; JAMES, W. Humanism and Truth. Mind 13:52, 457-475, 1904; JAMES, W. Pragmatisms
Conception of Truth. The!Journal!of!Philosophy,!Psychology,!and!Scienti !c!Methods IV:6, 141-155, 1907; DEWEY, J. Valuation and Experimental Knowledge. The!Philosophical!Review, 31:4, 325-351, 1922. For a synthesis of pragmatism, begin with HAACK, S. The Pragmatist Theory of Truth. British!Journal!for! the!Philosophy!of!Science 27, 231-249, 1976.
[23] Note that it will not do to argue that the beliefs have to change because the truth conditions of the world (p!is true as in condition (i) of the wrong analysis of knowledge) themselves changed: the world is not in itself true or false, truth being a predicate of beliefs, and not of the ingredients of those beliefs. Suppose that, faced with the fact that their beliefs seem no longer to be justi! ed, farmers simply become too perplexed to be able to form new beliefs; then, there will be no question of truth at all. This was perhaps an unnecessary reminder, but it is a reminder nevertheless.
[24] GETTIER, E. L. Is Justi! ed True Belief Knowledge? Analysis 23, 121-3, 1963.
[25] I am here making entailment correspond to material implication, as I believe Gettier wanted Cases I and II to be different; if in Case II all we have is, as in Case I, an invalid entailment or deductive inference from a false premise to a true conclusion (i.e., and tentatively, "!# $), then I do not see the point of providing two cases. The absence of formalism in Gettiers paper allows multiple readings; this is another serious problem in it.
[26] For an ardent con! rmation, see FLORIDI, L. On the Logical Unsolvability of the Gettier Problem. Synthese 142, 61-79, 2004.
[27] In AUGUSTO, L. M. Do unconscious beliefs yield knowledge. Revista!Filosó !ca!de!Coimbra 35, 161184, 2009, I address many of the points discussed in this paper from the viewpoint of unconscious knowledge.