Skepticism

The Strange Story of the Idea of Error Avoidance

Skeptics have claimed that there is no knowledge, that everything we say is doubtful; hence, error is unavoidable.

Socrates presented himself as a skeptic.

The idea that the suspension of judgment is a virtue became the central teaching of Pyrrho and his followers, the Pyrrhonists. There is no written evidence attesting to Pyrrho’s position, but the prevalent view regarding it follows Sextus Empiricus and presents Pyrrhonism as follows. Those who wish to dwell in a peaceful mood (ataraxia) should try to argue against every position that they tend to prefer: their view of any question should be as balanced as possible. As a result, they will refrain from assuming any position, and the outcome will be a permanent state of peace of mind. This is a recommendation for a skeptical way of life, in which one neither denies nor asserts any statement about the world. Like all schools of philosophy that seek peace of mind, Pyrrhonism rests on the refuted idea about what brings about peace of mind; whereas, in truth, we still do not know how peace of mind is best achieved.

Sixteenth-CenturyEuropean scholars discovered the writings of Sex­tus and found his arguments most congenial.

Modern philosophy starts at the beginning of the seventeenth cen­tury with the ideas of Sir Francis Bacon and Rene Descartes, who advocated the dismissal of all past opinions without distinction so as to be able to start fresh, thereby achieving true knowledge. Bacon said that unless one totally suspends all judgment about everything, one is unable to investigate nature without bias. After exercising the total suspension of judgment, one can observe accurately and construct science on solid foundations with ease - even in one’s spare time.

Descartes accepted Bacon’s view of utter skepticism as the starting point of all serious thinking. He practiced this idea of the full suspen­sion of judgment by applying it as methodically as best he knew how. He was no skeptic, but for the sake of reaching the desired certitude, he assumed as extreme a skeptical position as he could imagine. In this process, he evoked the “evil spirit hypothesis.” This is the possible hypothesis that all my experiences are creations of an evil spirit that controls all my beliefs and wishes to mislead me. Hence, the reason for my experiences and beliefs is the evil wish of that evil spirit. Later writ­ers simplified this hypothesis; in its simplest version, it is the solipsist hypothesis: all my experiences are misleading because the only thing that really exists is I (i.e., the first-person singular, whoever it hap­pens to be) and all the rest is but a figment of my vivid imagination.

Now, let there be no mistake: this hypothesis, in all of its variants, is utterly crazy, and no sane person believes in it or even takes it seriously. Psychologists suspect anyone who does take it seriously or who even thinks one should give it serious attention of a mental imbalance. Indeed, a leading twentieth-century psychotherapist, Ronald Laing, said that Descartes’ initial ideas sound suspiciously sick to modern therapists. Thus, all efforts to defend Descartes’ hypothesis result from having lost sight of the point of his discussion: there is no need and no possibility to defend a crazy hypothesis. The point is not to believe or disbelieve this or any other crazy idea but rather to disprove it (and all of its variants, of course) because any proof of our scientific theories is by itself a disproof of these crazy ideas. Therefore, anyone who claims that extant science is a system of proven ideas should be able to disprove with ease all of these crazy ideas - but no one can.¹ And the crazier the idea, the greater is the insult that its irrefutability presents to epistemology. Immanuel Kant considered this irrefutability the scandal of philosophy (he claimed that he overcame it, but he was just bragging).

In the early twentieth century, G. E. Moore and Ludwig Wittgen­stein, the most influential philosophers in England at that time, sug­gested that only philosophy takes this crazy idea seriously so that phi­losophy itself is not serious and even pointless.

Bacon and Descartes, the two great fathers of modernity, were con­cerned with science, and their efforts were directed toward teaching people to avoid error.

Hume is the philosopher who is mostly associated with skepticism in the eighteenth century, although he considered this a great injustice to himself - even an insult^[7] - because, he said, nothing is easier than the indiscriminate throwing of doubt in all directions. His moderate skep­ticism was the recognition of the futility of Pyrrho’s version of skepti­cism. He referred to two types of skeptical views: Pyrrhonism, which he rejected, and moderate skepticism, which he embraced, according to which, despite skeptical arguments, our psychology saves us from suspending all belief.

Incidentally, mostWesterners want science to guarantee tomorrow’s sunrise but, to their great chagrin, it tells us that there is some likeli­hood that the sun will explode instantly. The likelihood is not great, even if we consider not only tomorrow but also, say, the next billion years or so. Oddly, science never gave assurance about the sunrise, even though many great researchers (including Laplace, no less) said that it does. But they always were careful to qualify their statements. Their audiences, however, were more than ready to overlook their qualifications.^[8] Now, science says, the sun loses tremendous amounts of energy every moment; therefore, sooner or later, it will run out of fuel.^[9] And then, even if it goes on rising, it will do us no good. H. G. Wells described this vividly in his science fantasy, The Time Machine.

We digress. Hume said he had away out of skepticism. Unlike most other philosophers, he did not use epistemological arguments to that end. Rather, he claimed that what saves us from the Pyrrhonist trap is the force of psychological processes that make us take for granted what we believe in, regardless of all skeptical deliberations.

To repeat, Hume doubted physics but accepted psychology, which is also quite questionable, of course. Kant made a great effort to avoid psychology. He spoke of the difference between facts and judgments and said that epistemology discusses not facts but rather the validity of judgments. Yet, he failed: he described the human thinking apparatus and ascribed to it some of the best qualities, such as harmony (which he called “the transcendental unity of the apperception”) in assertions that are obviously descriptive - too much so, thus inviting doubt even about the meaning. Thus, the nineteenth-century commentators on his works had great trouble sifting his epistemology from his psychol­ogy. This explains why the Humean view of epistemology as a part of psychology is so very popular. It reappears in history in diverse garbs and, at times, also in all nakedness. For example, in the 1930s, an able historian of science, A. N. Meldrum, and more recently also the famous philosopher and historian of science, Thomas S. Kuhn, both declared that the philosophy of discovery is too faltering and should give way to the psychology of discovery. In all of its variants, this idea is psychologism, the view that we cannot help but believe and behave the way we do and that therefore the explanation of our belief and of our conduct are inherently parts of psychology.

Hume failed to convince his readers that his skepticism was moder­ate. Many of them saw him as a skeptic proper, a Pyrrhonist, which still is the dominant view of him - despite his touching protest - because he strengthened the traditional Pyrrhonist arguments. This is regret­table because he was of the opinion that sense data are certain, so he obviously was not quite as much a Pyrrhonist as legend says he was. But that is another story.

Perhaps as the result of the failure of the French Revolution that was the product of the Enlightenment Movement, irrationalism flourished in the nineteenth century, especially on German soil, and its adher­ents offered some crazy metaphysical systems as expressions of their irrationalism and as a means to undermine science proper. Rational­ists could not bring themselves to admit that the irrationalist critique of rationalism had some value. So, the rise of irrationalism regrettably pushed the excessive defenders of rationalism to the enhancement of the traditional hostility to metaphysics, which got a new name: pos­itivism, meaning faith in reason and science, only to the exclusion of faith in any metaphysics. This made rationalism a kind of politi­cal party, one that allied itself regularly to the radical political parties proper; loyalty to it made discussions of skepticism improper because skepticism undermines science. This is an error: skepticism and pos­itivism often go together. Indeed, the name of David Hume always invokes both together, perhaps because his hostility to metaphysics was brave and systematic. (His positivist ideas about religion appeared in print only posthumously.)

The great breakthrough came around the middle of the nineteenth century, in the same period in which positivism flourished: William WhewelFs theory of scientific verification, which was one of the great­est discoveries of psychology ever made. The greatest perhaps was made by Sir Francis Bacon, who said that our speculations operate as spectacles: we see the world in accord with them. Hence, whether or not what we see agrees with our ideas, it looks to us as if they do. This is now a familiar idea: we refuse to admit cognitive dissonance - that is, we refuse to admit the very possibility of empirical refutations of our ideas. Hence, said Bacon, better have no ideas to begin with because then what Mother Nature shows is the truth. No, said Whewell: with no ideas, there is no vision at all - what does not enter our intellectual horizon is also beyond our visual horizon. Many psychologists agree about that and yet they do not see that this is a rejection of the idea that perception is passive. Rather, they often say that we cannot escape the constraints of the intellectual framework of the traditions to which we were born. This is relativism that today the postmodernists have made very popular. It is obviously false because we do learn and we do see new things: otherwise, science would be impossible. Also, some people do change their minds and at times do say so audibly, which Whewell explained by saying that there is a way to defy the disposition to interpret observations as if they agree with our ideas. We develop a new idea, we conclude from it that a special arrangement should give rise to certain new observations, and we test this conclusion. Because we are the source of the idea, it is unlikely to be true. Therefore, our test normally refutes our ideas, so we try again and again. When a test result is positive, we have verified our theory. This is how science progresses, Whewell concluded.

This terrific theory did not appeal to philosophers because it makes the success of scientific research depend on luck. Researchers, how­ever, liked it very much and for the same reason. So, philosophers forgot Whewell and the idea of verification. Scientists did remember it but, in writing history, they forgot its originator. When Whewelfs ideas were rediscovered, philosophers were less averse to the idea that research needs luck; therefore, the popularity of positivism increased. However, the idea came too late: in the meantime, Einstein caused a quiet revolution in philosophy. It was rightly in the shadow of his greater revolution in physics; however, in the present study, it concerns us more: Einstein destroyed the equation of error with sin that the sci­entific tradition unwittingly shared with religious traditions, especially with the religious traditions of the West.

Under Einstein’s influence, Karl Popper advocated fallibilism, the view according to which any scientific theory may turn out to be false no matter how well it fits our experience to date. Hejoined Einstein in claiming that experience may undermine theories or allow for them but never support them in any way. Popper went further and said that to be scientific, even experience has to be tentative.

Our form of skepticism is even more radical: we hold that every statement is doubtful, that information and theories are never certain, plausible, corroborated, or justified - in the philosophical sense of these terms.

Radical Skepticism

size=1 color=black face="Book Antiqua">Before presenting the arguments for skepticism, we make seven rather technical points, as follows.

First Note

The words skeptic, skepticism, and their like have diverse uses. The etymological sense of the words refers to search (skeptomai = I search); whether or not this was initially true, it is not true today because it invariably refers to doubt, not to search.

Second Note

Our view that no statement is certain, plausible, corroborated, or otherwise justified refers to the epistemological status of statements. There are cases in which these terms refer to some established or agreed-on procedures and, at times, they make ample sense. For exam­ple, laws demanding that certain types of statements should be plau­sible, corroborated, or otherwise justified before courts would admit them as true. Their demand is that contestants should meet some received standards of plausibility as listed in books of laws of evidence. These procedures do not guarantee the certainty or plausibility that philosophers usually seek: there is no guarantee that these received procedures lead to results that are invariably true, and they often are found to be erroneous. When it seems that a certain kind of error appears repeatedly, legislators try to alter the procedures with the aim of reducing error.^[10] They often promise that the reform they advocate is the last and that after its implementation, the rules become perfectly reliable. How many times should this procedure of reform occur, asked Israeli law professor, Benjamin Akzin, a half-century ago, before we realize that the law cannot be infallible?^[11] At times, courts admit false evidence: they still cannot fully guarantee all error avoidance. This is general knowledge: legislators and courts repeatedly try to reduce error by reforming laws of evidence.

Thus, in particular, the cosmetic and the pharmaceutical industries have to test their products. Recently, the U.S. Supreme Court decided that these tests are often too perfunctory; it therefore declared that the tests should be severe, which means that testers should do their best to find fault. It is not easy to demand a manufacturer to seek fault, just as it is difficult to demand that accountants find fault with the data supplied by their employers. But, if a court of law decides that a certain test procedure was not up to the standard and then fines the manufacturer of a defective product for negligence or the accountants for their negligence to report faults, then the scenario looks different. This, too, is not yet a guarantee from any court, and even if there were such a guarantee, it would still be not very efficient because it would not be watertight and not even the best technique is available to avoid error. But, then, no method is perfect, and even if there were a perfect method, we would not know how to identify it; even if we had identified it, we would not know how to implement it with the least amount of distortion.

This is not merely theoretical. Whenever innovation leads to a dis­aster, inquests or courts may face the question: Was it predictable and thus avoidable?^[12] Suppose a test did not eliminate some harmful error. The producers defend their conduct, saying that proper tests failed to disclose the defect. Verdicts then depend on the view of the sincerity, or good faith, of the effort of the producers to find the fault in question and the severity of their tests. The jury may have mere common sense to go by - that is, some unarticulated criteria for sincerity and for severity - but no sooner than it passes the verdict then its procedure will crystallize as articulated criteria, which will be imperfect and invite efforts to improve on these very criteria. This is our conjectural story that may illustrate real-life situations that conform to Popper’s philos­ophy much better than to traditional philosophies - and, of course, it is obviously inherently fallibilist.^[13]

The great achievement of the scientific tradition that philosophy could not recognize before the Einsteinian revolution is this: there is no need to justify conduct unless the law requires it, and then the requirement should be specific and depend on standards that have improved through the ages. The same is true for disagreement. The scientific tradition looked askance at it in preference for error avoidance because, clearly, disagreement rests on error: by definition, among several inconsistent answers to a given question, no more than one can be true; therefore, in any dispute, at most, one party may be in the right. Democracy always went with liberalism and disputes, yet its rationality became clear only when Popper rejected the equation of rationality with proof and allowed into philosophy the commonsense idea that some but not all error is rational; some but not all error is due to preventable neglect or impermissible stupidity. (Those who view science as error-free must find all error in research stupid or negligent or prejudiced: error then must be due to some extra-scientific fault.) Hence, Popper’s fallibilism is a new version of skepticism. In partial agreement with that of Pyrrho, Popper’s skepticism is a denial of the possibility of proof coupled with the new, Einsteinian encouragement to make bold conjectures and put them to a rational test.^[14]

In what sense, then, are everyday, commonsense standards of plau­sibility reasonable? Briefly, they are reasonable as contemporary the­ories of what seems credible. We return to this discussion in Chap­ter 3.

Third Note

The assertion that all statements are doubtful is open to the following two interpretations:

ι. No statement is certain or demonstrable.

2. No statement is plausible, corroborated, justified, and so forth (in the epistemological sense of these terms).

We advocate the second interpretation, which is obviously the stronger of the two.

From the days of ancient Greek philosophy to the very end of the nineteenth century, most philosophers understood skepticism along the lines of the first interpretation - namely, that a statement is doubt­ful as long as it is not certain and there is no certainty. Opposing it, they tried to find what could be known with certainty; however, they could not produce any instance because the skeptics had no trouble dismiss­ing all evidence. (Skepticism, as Hume observed, is very facile.) This made them seek a criterion for certainty; plausibility was not enough of a guarantee for rationality, however. In particular, it is less than reli­gion demands: religious authorities condemn as heretic (if not even downright atheist) the claim that the existence of God is not certain but merely plausible. (This way, Kant made light of the argument from design - the claim that the world’s design proves the existence of a designer - the physico-theological proof of the existence of God so- called, although he considered it very strong because, he admitted, it is not clinching.) Some philosophical discussion about plausibility did take place but seldom. The general view was that, according to skepticism, all statements are uncertain even if most people take them for granted.

This interpretation of skepticism ceased to be effective as the result of the discovery that even the best scientific theories are not proven. Tradition deemed scientific only a theory that already has won the honor of having passed successfully the process of empirical test. But any such proof, any proof by successful tests, is merely inductive. Today, philosophers agree that induction from empirical observations cannot possibly reach certainty. We may declare, for example, that all the many observed ravens are black; we may declare that these observations are true, that the ravens seen as black are indeed black. But then even the prediction that the next raven to appear on our horizon will be black does not follow from these declarations. Hence, this prediction is quite uncertain even if its inductive premises are certain.

The discovery that scientific theories cannot be certain, shocking as it was, chimed with the traditional skeptic view that even predictions of tomorrow’s sunrise are uncertain. Philosophers soon introduced a new theory of plausibility, just to block the skeptic’s claim for obvi­ous victory. This new theory, then, had to justify the familiar faith in tomorrow’s sunrise: although it is uncertain, it is beyond reasonable doubt - it is plausible; it possesses a high degree of plausibility.^[15]

This is the dominant position in contemporary mainstream episte­mology. Mainstream philosophers today admit that certainty obtains only in the fields that do not involve experience: logic and mathemat­ics. Mainstream philosophers resist skepticism nonetheless: some state­ments about our possible empirical experience of the world, such as the prediction that the sun will rise tomorrow, they say, although uncer­tain in principle, are certain in practice; they are plausible, corrobo­rated, or otherwise justified. This is the most that we can expect science to achieve, they add, demanding that more is unreasonable - and the skeptics demand more. Hence, the skeptics are unreasonable (A. J. Ayer).^[16]

Note that some versions of fallibilism are stronger than others; our disposition is to advocate a version as strong as we can make it - with­out becoming unreasonable, of course. To reiterate, fallibilism is the position that any view about the world might be false. It is associ­ated mainly with the philosophies of Peirce and Popper, who stated categorically that no statement about the world is certain. But falli- bilism does not entail the logically stronger position, that by itself no statement is plausible, corroborated, or otherwise justified (in the epistemological sense of these terms). And, indeed, Peirce claims that the process of knowledge-seeking reduces doubt. Popper claims less: he asserts that although corroboration by empirical observations can­not make a theory probable, refutation by empirical observations can render it implausible. Although we observe this to be the usual case, and although it is often reasonable, we do not suppose that it is gen­erally the case or that to be reasonable it should be. In this sense, our position is more radical than Peirce’s fallibilism and even Popper’s.

In what follows, we use the concept of plausibility as a synonym for corroboration and justification. There are, of course, different kinds of justification, but for our purposes we may ignore them; the concept of plausibility represents them all. True, it is always possible to make distinctions and so we, too, can distinguish between the different jus­tificatory concepts; however, it does not serve our present discussion.

Fouith Note

We do not claim that all statements have the same degree of plausibil­ity. Rather, we claim that plausibility (in the epistemological sense of the term) does not apply to statements. We distinguish between the following two positions:

ι. Statements can be plausible, but then all statements have the same degree of plausibility.

2. There are no plausible statements, just as there are no magical spells.

We agree that the first position is absurd. However, here we are pre­senting the second position and, in reference to it, we are suggesting that it is consistent and reasonable. Unlike the Pyrrhonists, we are not ascribing the same degree of plausibility to the prediction that the sun will rise tomorrow and to its opposite. Rather, our suggestion is that regardless of what we feel about these two statements, to say that they are more plausible or less is no more than saying that they are more potent or less than spells because the plausibility (in the epistemolog­ical sense of the term) of a statement is but a form of the potency of a spell.

Many philosophers who advocate the view that science lends plau­sibility to its predictions tacitly admit this: they add a principle of induction as the missing item that helps render tomorrow’s sunrise plausible, given all past sunrises. The trouble is, no one has managed to formulate this principle sufficiently clearly and in a manner that does not immediately raise unanswerable objections to it. (The same holds for magic.) The traditional wording of it is the principle of the simplicity of nature. This or any other wording of the principle should make the future look like the past. But which future? We know that some past events are gone forever, that death is final, that some species are extinct. To know which past events are repeatable and which are not, and under which conditions, is what researchers are trying to find out. It is too much to expect philosophy to wave a magic wand and provide answers to all the tough questions that scientific research has labored so hard for generations to find. Therefore, we suggest that the philosophers who are seeking the principle of induction are doing what is necessary to establish their position, but that this they cannot possibly achieve. Moreover, the reason for which they are trying so hard to find the principle of induction is their wish to justify research and perhaps even to assist it, under the profound conviction that as philosophers, they should not try to replace it. Yet, this is exactly what they are doing, trying to replace research — all quite unwittingly, of course.

Fifth Note

Many discussions in the philosophical literature refer to variants of skepticism that limit it to certain fields. One such example of limited skepticism is the position according to which all scientific theories are doubtful but that there are empirical repeatable observations beyond doubt.^[17] (This is the standard empiricist view, and diverse thinkers have advocated it - from Locke in the seventeenth century to Quine in the twentieth.) This kind of skepticism appears as a generaliza­tion because it refers to all scientific theories, but the generalization itself applies to a limited domain. Not so our position. We assert that, because there is no certainty and no plausibility, the very concepts do not apply to statements by themselves. Nor is there any need for this: we can do without the certainty and without the plausibility that philosophers seek once we see that we can admit that all statements are doubtful and that in admitting this, we lose nothing that is worth keeping. In particular, this does not lead us to the preference of the inaction that the Pyrrhonists and other ancient skeptics advocated.

Hume’s skepticism was almost total but, contrary to total skepticism, he applied it neither to immediate reports of experience nor to beliefs that rest on simple intuitive logical theorems. On this point, our posi­tion differs from Hume’s. As previously mentioned, we assert that all statements are doubtful, including reports of immediate experience as well as mathematical and even logical statements. This sounds strange because logic and mathematics are rich with proofs. We do not wish to deny that. We nonetheless assert that the proofs are neither capable of eliminating doubt nor is there any need for this elimination because the doubt that we advocate can do good and do no harm - except when we misread ideas in line with some traditional errors that here we do our best to eliminate.

Sixth Note

It is not our aim to explode as meaningless terms such as certainty, plausibility, corroboration, and justification, any more than spell; they are meaningful, easily understandable words.

Many philosophers followed Wittgenstein and tried to resolve (not solve) some insoluble classical philosophical problems by claiming that the terms that necessarily enter the wording of these problems are meaningless or that their wordings are necessarily ungrammati­cal (not that Wittgenstein and his followers were expert grammari­ans). We have no wish to take this path. We do not like problems to disappear; we consider sheer magic all play with words that allegedly serve as spells and make them disappear; the magic does not work. People often use sentences in which some terms denote certainty, plausibility, and corroboration or other terms that denote justifica­tion. They appear in discussions of some statements, and we do not see how all such extensive use can be dismissed as meaningless without limiting unreasonably our exchanges of ideas.

Wittgenstein declared that he had demolished all philosophical problems by exposing certain words and sentences as devoid of mean­ing. At first, he referred to meanings in an artificial language of his own making that he deemed ideal (namely, perfect). He was in error, but at least one had to learn his language in order to show this. Later on, he and his disciples applied this idea to natural languages. He then proscribed the literal use of many common words (God forbid) so that it became obvious at once that he was changing the meaning of the word meaning without warning and without explaining this change. So, we do not try to fathom it: his proscription is too arbitrary for our taste whatever he exactly meant by the word meaning, which we pre­fer to use in its ordinary meaning or else in accord with its standard use in formal logic - and we take great care to distinguish these two meanings.

Seventh Note

We do not herein refer to probability. We deny the widespread view that it is incumbent on rational people to choose the most probable alternative (to believe in). We also have no use for the probability of statements. The distinction between plausibility and probability, incidentally, is obvious: probability has a well-defined characteris­tic because it follows the formal calculus of probability; plausibil­ity, whatever exactly it is, does not. What exactly probability means when applied to statements rather than to ensembles of events is unclear and under debate. Consider a generalization such as “all ravens are black.” This statement does not describe an event and, although it is possible to assign probability to such a statement, it is not clear which statement is more probable than another, except for a small set of comparisons imposed by the calculus. The calcu­lus declares that the probability of a conjunction is smaller than that of any of its conjuncts. This is clear when the conjuncts are events: it is less probable to throw two sixes in a row (1/36) than to throw one six (1/6). What it means when the two conjuncts are statements is not clear, but at least we can say that if statements have proba­bilities proper, then the probability of one statement is higher than that of two; therefore, according to Popper, those who want to make only probable statements should say as little as necessary. Such peo­ple, he added, are not scientific researchers because they must be adventurous.

Considering probability as frequency is the most obvious appli­cation of its calculus. No one knows how to measure the relative frequency of a universal statement, much less how to calculate its probability as a frequency. An interpretation of the calculus of prob­ability exists that assigns probabilities to all statements in any given language. It renders all tautologies most probable and all contradic­tions most improbable so that no evidence can support or undermine any of them. This reading is very interesting for some purposes but not for the purpose of identifying the plausibility of statements with their probability. By that reading of the calculus, the probability of a universal statement like “all ravens are black” is minimal because the statement refers to infinitely many possible ravens. Singular state­ments that describe evidence, such as the appearance of black ravens flocking by the drove, do not raise the probability of any universal statement. Of course, if we declare - for example, in the name of the principle of simplicity - that all birds of the same species are the same color, then one instance of a black raven suffices to inform us that all of them are black. But then, all statements involved in this deduction are and will remain doubtful, and most of us will admit that the added premise we just suggested is false (although it holds for some bird species). Still, all is not lost because if instead of speaking of colors we speak of skeletons, then the result is less foolish, at least seemingly so. This shows again that even the simplest scientific deliberations are context-dependent.

The Case of Mathematics and Logic

Traditionally, logic and mathematics were deemed the strongest argu­ment against skepticism: Descartes suggested that even God cannot render 2 + 2 = 4 untrue. In any case, it was generally recognized that no sane person can seriously doubt arithmetic or geometric truths. The truth of Newtonian mechanics, then, had to follow suit: even though the certainties here are of a different sort, the skeptical doubt concerning them is the same and so it can be dismissed wholesale.

Hume denied that. Our theories are causal and rest on experience, yet causality is not given to experience (= it is in no sense datum). Further, the assurance with which geometry is endowed differs from that which accompanies arithmetic because it rests on intuitions that may mislead. All this invites reconsideration because all ascriptions of certitude to perceptions are refuted, and whether we observe causality is an open question. Yet, remarkably, Hume endorsed the received view of logic as unassailable.

Kant revolutionized both logic and mathematics when he offered new arguments to buttress them because he first deprived them of the traditional arguments. He allowed within logic two or three types of formulas (and, presumably, their corollaries): they are the forms style='font-weight:bold'>a = a (a is a) and ab ⊆ a (ab is a), to use modern parlance, as well as all verbal definitions. The latter are true as mere stipulations. He rightly insisted that such definitions add nothing to knowledge (while rightly dismissing definitions as theories in disguise).

The important point is not what Kant allowed but what he excluded from logic to begin with: he declared the truths of arithmetic not a part of logic. This was excessive because he could allow for it by the use of Leibniz’s idea that they are verbal definitions. Leibniz suggested that ι + ι = 2 is not an equation but rather a definition of the word 2, so that spelling out 2 + 2 = 4, we get 1 + 1 + 1 + 1 = 4, which, likewise, is not an equation but a definition of the word 4. (Leibniz’s idea was revived and updated by Alfred Tarski.) It is hard to know whether Kant deemed the truths of arithmetic at that stage more or less doubtful than the axioms of Euclidean geometry or even those of Newtonian mechanics. His concern was to move to the next stage and offer a new justification for them, and his intent was to elaborate on that stage. He therefore lumped them all together and called them by the fancy name of transcendental logic, which he defended at length. His defense is pointless because transcendental logic is not logic at all; he never attempted to explain why he called this cluster of transcendental assertions a logic except to say that they are all beyond doubt. This is obviously flimsy. (Hegel soon followed suit: he declared as logic all the fancy pronouncements that he decreed, thus identifying with logic not only the allegedly certain but also the allegedly true.) Still, Kant’s hardly argued opinion about the status of certain theories led to huge changes: the search for their refutations. A new geometry and a new logic thus emerged. Geometers first omitted some of Euclid’s axioms and then they got more daring and examined all sorts of non-Euclidean alternatives. The concluding point of this process was Einstein’s dissent from Kant that is the consensus today: inasmuch as geometry is about the world, it is not certain, and inasmuch as it is certain, it is not about the world. We dare to go further: even if it is not about the world, we propose, geometry is not certain. No assertion is.

The quest for certainty led Kant’s critics to the study of the founda­tions of mathematics. Georg Cantor (who was contemptuous of him) managed to present arithmetic as a part of abstract set theory. Gottlob Frege developed a new logic to show that set theory is a part of logic; Russell followed suit. Their theory is known as logicism, which today is taught as the last word, even though it was refuted by Russell and more thoroughly by Quine.

Prior to that theory, David Hilbert developed the revolutionary idea that the axioms of a mathematical system are definitions of sorts: they define or characterize that system. Thus, the axioms of Euclidean geometry are true by the convention of seeing them as characterizing a system; they are truths by convention: we call Euclidean any system that abides by Euclid’s axioms.

Thus, mathematical truths are unassailable as parts of logic or by definition. Does this invite further justifications for them? We do not know. Frege and Russell did offer one: they deemed logic a charac­teristic of the ideal language. Today, logicians no longer entertain the view that an ideal language is possible: they deem all languages con­ventional. And so, it seems, Hilbert’s view prevails. Why then endorse the conventions that mathematics offers? Hilbert was more intent on defending his view that the conventions are commendable than on explaining why they are (he found this unnecessary). The question, then, is: Do the truths of logic possess greater force than those of mathematics? If not, are alternative systems of logic possible akin to the alternative possible systems of geometry? And, if yes, do these cre­ate alternative systems of arithmetic? These questions engaged many thinkers in the twentieth century and led to exciting results, some by Hilbert’s former students and some by his critics.

Enter Imre Lakatos. He presented the history of mathematics - only snippets, to be precise - as that of trial and error. He convinced Popper that mathematics is as much a series of conjectures and refutations as science is. Popper still kept logic apart: he, and more so Quine, took the demand to eliminate contradictions as more than a mere convention: if anyone allows for them, said Quine, then one thereby shows oneself to be confused or else using words in a totally new and different way. Popper went further. He developed a version of logic (called natural deduction) that presents it as a set of rules of inference. He showed that different systems of logic simply follow different sets of rules of inference, and he opted for the one that is strongest because it is the best tool for the elimination of errors.

The numerous varieties of reasonable claims for truths and more so of ways to elaborate on these claims, contrary to received impressions, are agreeable to skepticism because, clearly, none is assuredly error- free and none is in need for such assurance: taking some of these truth claims seriously, we suggest, invites their careful examination. We suggest that skepticism encourages taking them seriously; perhaps they are possible at all only after skepticism is admitted.

Let us conclude this outline by opening one such discussion. What exactly is a truth by convention? Who are the people who convene and how do they convene? What are their aims? Are there better conventions to further these? In particular, no one can show that the Socratic search for contradictions and the efforts to eliminate them are the best tools in the search for the truth. The progress of logic and of mathematics in the last two centuries is tremendous, and it did overthrow many ideas that were deemed indisputable truths, culminating with Popper’s overthrow of the equation of rationality with certitude. It is consequently becoming increasingly easier to be skeptical if not also increasingly imperative.^[18]

The Skeptical Arguments

Skeptics have raised two kinds of argument against the claim that cer­titude is attainable. One is general and has already been mentioned: a statement is certain or justified if it is proved, but proof is impossi­ble because it is question-begging - any criterion for the validity of a proof requires a different proof (because self-validation is too easy and always possible and therefore leads one anywhere one likes; hence, it is useless). Ajustification procedure invites ajustification for itself and so on ad infinitum; hence, no justification is final. For example, char­acter witnesses are possibly unreliable and then, for their testimony to be acceptable, they need character witnesses to testify that they are reliable, and so on until we exhaust the population.

The arguments of the second kind are specific: they are counter­examples against each presentation of an allegedly reliable source of knowledge. Lord Acton was a Roman Catholic, yet he found intoler­able the idea that the Pope is infallible, saying that this grants him absolute power and absolute power corrupts absolutely. We sympa­thize but try to hold even this wise adage in doubt. We go further and say that no number of convincing arguments can prove that our skep­ticism is true; rather, they disprove instances for the opposite view, the view that proof is somehow attainable.

As it happens, today - when most philosophers still combat skepti­cism any way they can - almost all of them still view only empiricism as a serious contender for a theory of the attainability of certitude (the rest of them retry Kant’s way). So, nowadays, the presentations of alternatives to skepticism refer mainly to variants of empiricism. This is the view that knowledge comprises what we perceive and what we infer from what we perceive by diverse means, deductive and inductive, and nothing more. These two assumptions are exactly what Hume referred to in his proof of the impossibility of induc­tion. Latter-day efforts to refute him rest on innovative ideas about both perception and inductive inference. (Deductive inferences stand

some success first. It is in this sense that - his general skepticism notwithstanding - Einsteinspokeofaspecificpartofphysics (i.e., thermodynamics) asrelativelycertain. Even that was more certain in general terms (the phenomenological theory) than in detail (the statistical theory) that he managed to have replaced (by the Bose Einstein statistics). apart simply because they are not under debate in this context.) No one knows exactly what perception and inductive inference are, much less which perception is reliable and which inductive inference is acceptable. Still, broadly speaking, we know what together they are supposed or hoped to do: they should lead from empirical observa­tions to generalizations and other theories without allowing error to creep in or, at least, while reducing significantly the rate of scientific error.

Discussions of perceptions should refer to empirical studies, to the field of study known as psychophysics. The pioneers in this field were hoping to illustrate Kant’s idea about observation reports as the outcome of the application of theory to observation. It is doubtful that this can be done, but it was possible to refute some specific simple theories, those that are definitely no longer endorsed in up-to-date science but are still extant in the philosophical literature as a sort of fossil. Specifically, it is easy to refute the specific empiricist naive theory of perception that Locke and Hume espoused and that most empiricist philosophers still piously uphold in a sincere effort to find cases of error-free perceptions. Thus, these efforts assure that they are steeped in error. They proceed regardless of the empirical refutations of their views in search of inductive inferences that rest on what is directly perceived; although what it is that is directly perceived, no one knows - except that it is not what Locke and Hume said it is and that it should be error-free. However, just because of this last characterization of what is directly perceived, we suggest that looking for it is a wild goose chase.

For our part, although we are quite ready to take psychophysics seriously, we do not need it as a source of skeptical arguments. We consider the ancient classical arguments against the reliability of what we perceive - directly or not, as you like it - both interesting and sufficient. Yet, as it happens, the best presentations of these lovely ancient arguments are now in the psychophysical literature. Anyone interested in this can consult a modern textbook on perceptions or, better yet, visit a psychological laboratory.^[19]

So much for the arguments about perception, classical and mod­ern; the next issue for the defenders of empiricism is induction.^[20] Now, whatever exactly induction is, a prolific philosophical discussion about it is taking place in an attempt to dispel the refutations of induction, misnamed as the paradoxes of induction, which seriously undermines the very idea of induction as a means for tapping the source of our knowledge. The endorsement of the refutations of all extant theo­ries of induction (or of all versions of the theory) is permissible even though uncomfortable, as long as the insistence that learning from experience by induction is nevertheless quite possible accompanies it, and that once we show this, we can also vindicate induction. The discomfort just mentioned is due to the inability to explain why no study of induction by inductive means takes place. Such a study may fail to vindicate induction; nevertheless, for inductivists, it is worth­while because it will elicit the theory of induction that they wish to vindicate. In truth, however, this is impossible because real science is fallible and induction is supposed to prevent error or, at least, reduce it significantly. Even if certitude is replaced by some surrogate, that surrogate is supposed to reduce error as well as possible, thereby pro­viding a surrogate infallibility, which likewise cannot be found in the real world. If any induction is clear, it is that like in the past, future philosophers will fail to exemplify valid induction, much less to vin­dicate it. But let us leave this discussion and report the paradoxes or, rather, the most famous of them.

One famous paradox of induction is attributed to Nelson Goodman and it is as follows: Let us arbitrarily choose some date in the future - for example, the year 2100 - and define a color “grue” (i.e., a hybrid of “green” and “blue”) such that an object is grue if and only if it is green until 2100 and blue thereafter. Now, consider the following two statements:

1. All emeralds are green.

2. Allemeraldsaregrue.

By any theory of inductive support, available empirical evidence supports or undermines both statements equally because whatever evidence supports or undermines the one supports or undermines the other for the very same reasons, whatever they may be. This is no surprise because Goodman devised the predicate “grue” artificially and solely to meet that end. And, because “grue” is artificial, we can replace it with any other hybrid term, and so find that the evidence about colors of emeralds in the past does not favor their present color any more than any other. This example is only a sophisticated demon­stration of the idea that Hume proved: there is no valid inference from past events alone to any conclusion about any future events. Those who take Hume’s argument seriously do not need Goodman’s; the latter made waves only because it makes more conspicuous the hopelessness of efforts to get around Hume’s proof. But, for those who do need such conspicuous examples, there are some that are much easier to follow, like those developed earlier by Russell: all events are observable, or all events happen prior to the current date.

Yet, there is a reason for the success of Goodman’s paradox. It is in its being a (redundant) critique of a specific, popular theory of induction - that induction rests on simplicity. Thus, “all emeralds are green” is simpler than “all emeralds are green or blue,” and the simpler competitor pushes the less simple one out of the ring when they compete for the title of the empirically most supported theory. Goodman merely created a new predicate that renders a deplorably complicated theory apparently and respectably simple. Of course, the artificiality of “grue” is deplorable; Goodman himself agreed to that. But he looked in vain for a reason to dismiss “all emeralds are grue.” The artificiality of “grue” is insufficient reason, he said, because many respectable predicates that science uses regularly are also artificial. The arbitrariness of this predicate is objectionable, but the motive for the suggestion of a new term is irrelevant: like clues that detectives can employ, it should lead to better arguments, to ones that can stand up in court. The detectives are still working on it because they are convinced that the suspect - Hume’s proof - is guilty as charged, that induction is alive and well, and that therefore Hume’s proof is somehow invalid. The way they go about it is their effort to refute Goodman’s paradox. No one has yet found any new satisfactory resolution to this paradox. Until they do, we must leave them to their toil.

Another famous refutation of induction is the paradox of confir­mation (so-called) of Carl G. Hempel. His is somewhat earlier and somewhat less general than Goodman’s because it rests on the very popular hypothesis that whatever inductive inference is, it involves the validation of general laws by their observed instances, via corrob­oration. This is known as the theory of induction from instances or confirmation by instantiation. For example, because a black raven is an instance of the general law that “all ravens are black,” any obser­vation of a black raven at any time and in any place is supposed to confirm the law. Likewise, the same theory of instantiation implies that every observation of a non-black non-raven confirms the general law that “everything non-black is a non-raven.” Now, a white shoe is non-black; it is also a non-raven. Hence, by the theory of instantiation, every observation of a white shoe confirms the general law that “every non-black thing is a non-raven.” Further, the two statements in ques­tion are logically equivalent: of necessity, “all ravens are black” is true if - and only if - the statement “every non-black thing is a non-raven” is true. (This is the famous law of classical logic called the law of contra­position: “all S are P” is equivalent to “all non-P are non-S.”) Hence, an observation of a white shoe - that is, an observation of anything at all except for a non-black raven - confirms the theory that “all ravens are black.” Thus, every observation whatsoever either agrees or conflicts with every theory.^[21] This sounds absurd because we naturally judge most observations irrelevant to any given theory; that is, almost any observation is utterly indifferent to the choice between a given theory and its negation. Hence, the theory of induction by instantiation is false. Yet, Hempel’s refutation is not taken any more seriously than Hume’s; therefore, he called it not a refutation but rather a paradox. And this is how it is seen quite generally (which, alas, is indicative of the state of the art).

Hempel tried to cope with his paradox, to resolve it, to show why it only looks strange but is not really. It is clear why he wanted to do away with it: he was convinced that induction by instantiation is right. He offered two excuses for the paradox. One is that in choosing between two competing theories (e.g., “all ravens are black” and “all ravens are blue”), we deem irrelevant any item (e.g., “this is a white shoe”) that supports both equally. The other is that the support that seems to us paradoxical is too small to matter. Now, the first excuse is questionable. Whenever there is a tie, we want to know about it because we may seek a change in the satiation that would break the tie. In the present case, we may not care about this because the contribution of the evidence is too small; that is, the first argument relies on the second. As to the second argument, that the quantity of support that some allegedly relevant data lend a given hypothesis is too small to matter, the idea that support is quantitative, is not one that Hempel studied; he simply pulled it out of the hat hoping that it would make the paradox vanish. It does not, and it raises more paradoxes, as we show when we discuss the quantitative theory of support, typically called inductive logic or the theory of probability of hypotheses.

No one has found any satisfactory argument to make Hempel’s paradox go away and no one noted that Hume’s initial criticism of induction is aimed at induction by instantiation.

Russell, although an inductivist, rejected induction by instantia­tion. He had a nicer refutation than Hempel of the view that a theory is supported in the presence of instances for it and in the absence of instances contrary to it. Take any theory and extend the class of which it speaks in any way that dodges meeting counter-instances (by including, say, all elementary particles) and (because elementary par­ticles are colorless) you have added confirmations for free. Of course, it is strange to add any items to a given class just to have a better confirmation. But how do we legitimate any definition of any class? Modern logic says that all such classes are equally legitimate. Yet, inductivists such as Goodman feel that some classes are more natu­ral than others and if we knew why, we would solve the problem of induction. A whole vast literature is devoted to this matter, the so- called study of natural classification. That study is doomed to failure because what is natural either rests on a theory, and that theory may be false, or else it may rest on intuitions; even some of the best nat­uralists have changed their views. Of course, the way out may be to prove the theory of natural classification by induction. But because the search for the proper principle of induction depends on the search for the right principle of natural classification, this move is dis­qualified.^{[22] [23]}

A third argument against induction refers to ad hoc changes of theories. The Latin word means “for this” and it applies to all changes that fit the need for change without examining the possibility that perhaps a bigger change is needed. This method is repeatedly used in law courts because judges should not legislate, but at times they simply cannot apply a law where its application makes no sense. So, in such cases, they have few options but to make exceptions, and they make them as narrowly as possible, sometimes while pleading with legislatures to reform the law - not ad hoc, but radically. Ad hoc changes of theories work as follows. Suppose some people share the prejudice against Ruritanians and honestly consider all of them stupid. Later on, they bump into a wise Ruritanian, Tom. Consequently, they change their view and assert that all Ruritanians but Tom are stupid. They bump into another wise Ruritanian, Dick. Consequently, they change their view again and make exceptions for both Tom and Dick. Thus, they go on changing their view whenever they find a counter-example to their prejudice. Such changes are ad hoc, and a theory that one alters to encompass corrections ad hoc is an increasingly ad hoc theory. Credence in it diminishes, they say, but the rules of induction allow for it: according to them, nothing is wrong with ad hoc changes. By these rules, all the necessary ad hoc modifications to the theory are welcome: they should not reduce support for it.

A fourth argument refers to the status of simple theories. Consider the presentation of at least three measurements of two parameters A and B on a grid that appear as if they more or less form a straight line. We tend to see a straight line going through the points.

A

This line is not there: the conclusion that the data are on a straight line does not follow from the observed points because there are infinitely many ways to draw a line through them (e.g., an infinite num­ber of wave functions that coincide with these points). The conclusion that the correct line is straight, then, is an extrapolation. All extrapola­tions raise the same difficulty for the empiricists: it is our old acquain­tance, the problem of induction. This problem blocks the way to any explanation of the preference for the linear function over each of the countless other functions that fit the same data. Some empiricists have tried to solve the problem by means of some criterion of simplicity. We may, of course, consider tentatively the hypothesis that linear functions are the simplest, and perhaps also the hypothesis that as such, they are the best explanations. This quickly leads us to two questions: By which criterion of simplicity do we judge the straight line the simplest line? Why is the simplest theory preferred and in what sense? Many empiri­cists have tried to answer these questions but, as far as we know, they despondently admit that none of the answers that they know stands up to the most obvious criticism. They still are optimistically convinced that a simple and obvious answer evades them, although it is simple and obvious that they are looking for a foolproof method which is but a pipe dream. (Also, the need for a foolproof method to sustain one’s optimism hardly seems so optimistic.^[24])

A similar analysis applies to what is known as Occam’s razor, which is the demand to eliminate unnecessary entities. It possibly includes the demand for simplicity and possibly a special case; this, in turn, depends on what we consider unnecessary under which conditions. Consider the following situation. Wanting to know what makes some substances combustible, Stahl stated that combustibles can emit large quantities of phlogiston (phlox is Greek for flame), which they do as they burn. Lavoisier then said that combustibles can absorb large quantities of oxygen, which they do as they burn. For a time, there was evidence against the older theory and none against the new one.

Nevertheless, some people found it difficult to let phlogiston go. They could, of course, consider Lavoisier’s theory true, give up any refuted assertion about phlogiston, but still insist that, while absorbing oxy­gen, combustibles also emit phlogiston. Most of us consider this idea silly; it is name-calling, not an argument. We present this idea as a chal­lenge to explain or disregard the feeling that this idea is stupid. Many meet this challenge by asserting that Occam’s razor should shave off this remnant of an outdated theory because it is quite superfluous. Admitting this, we may ask: Is Occam’s razor a part of the principles of empiricism and induction? Does it even chime with them? Empiri­cists have failed to show that it does, and Popper showed that the two are in conflict: the more unnecessary entities we admit, the more we will be prepared to meet events that do not fit our present ideas about the world. However, ignoring Popper’s critique, we can still ask: Is Occam’s razor not superfluous? If it is, should we not throw it away?

Finally, a fifth argument against induction rests on the difference between prediction and explanation. Consider a theory and the obser­vations that conform thereto (an observation conforms to a theory if it is an instance of a generalization that follows from said theory). Does it matter which comes first? Most advocates of traditional empiricism follow Keynes’s assertion that because credibility, plausibility, or prob­ability is a logical relation between theory and observation, the time sequence in which they appear should make no difference. They are all aware that a theory that yields a prediction that is confirmed seems thereby more convincing than the explanation of already known obser­vations. The majority of advocates of traditional empiricism deemjust such persuasiveness to be an illusion of presentation. But skeptics view the difference as another entirely salient refutation of traditional empiricism. Both explanations of observed phenomena and predic­tion of new results are hypothetical, but all the more so, prediction affords experimental tests of hypotheses, not random unbiased obser­vations toward induction.

These are some of the main arguments that refute traditional empiricism. They do not prove that skepticism is true; rather, they refute the given empiricist answers to skepticism. They do not refute all versions of empiricism, especially not those versions (if any) that are fallibilist. As to the claim that there is a possibility of newer and better arguments against skepticism, it is a general claim and it invites a response that must be general as well. We have already discussed this and are only going around in circles, as we have in philosophy for the last two or three centuries - all because of the refusal to abandon traditional empiricism and perhaps re-embrace it in some fallibilist version or seek such a version if none is available.

The traditional refusal to abandon traditional empiricism is very strong. There is a strong argument in its favor: we do learn from expe­rience. Now, skeptics need not deny the possibility of learning from experience; they deny that progress is due to error avoidance and they deny that it leads to certain or even plausible results. To reiterate, the call for error avoidance is but a call for inaction. Learning from experience is wonderful but, by being a human activity, it is risky; it is not error-free. What is objectionable in traditional empiricism is the demand to limit one’s beliefs to justified ideas. This demand seems to make a lot of sense. But then, justification is either error- free and therefore impossible or prone to error so it is useless for the traditional empiricist ends. Worse than that: empiricists cannot justify their view that skepticism is false; so, by their own standard, they should not oppose it even if they may insist on not endorsing it. But, the suspension of judgment about the suspension of judg­ment is itself a suspension of judgment. This sounds clever, but it is merely a roundabout way to say that skepticism is trivial. It sounds perverse because some skeptics - Pyrrho and his followers - have appended to the trivial thesis the perverse idea that action should not rest on doubtful ideas. The following section discusses this in some detail.

Skepticism in Epistemology

Skepticism was never popular in the West. By the traditionally popular view, skepticism is just a teaser; it comprises an assault on rationality and even on plain common sense, a kind of conjurer’s trick. No mat­ter how clever and convincing conjurers can be, no sane audiences will fall for them, except that conjurers please their audiences when they tease their imagination. Skeptics do not please their opponents, who try hard to find the skeptic’s sleight of hand but have failed to do so for centuries. And any competent and honorable debunker of conjurers might have conceded that much by now! Contrary to the popular view, then, we repeatedly advocate the skeptical view as outlined herein, claiming as straight-forward, rational, and common sensical that there simply is no way to guarantee that any discourse can ever be certain or plausible in the epistemological sense of these concepts.^[25] It is easy to show the difference between the two senses of plausibility - the epistemological and the common: only the former claims infallibility (in some sense or another); common sense rests on commonly shared suppositions that admittedly may be false. We have little to say about common sense, being that it is such a little studied field.^[26]

The view that skepticism in epistemology is absurd and a conjurer’s trick rests on the mistaken idea that it entails the following conclu­sions:

ι. Skepticism is objectionable because the assertion that every statement is doubtful condemns itself as doubtful. In this sense, skepticism is inconsistent.

2.      Skepticism blocks the adoption of any view and any mode of conduct: it prevents its adherents from saying why they believe that the sun will rise tomorrow and why they prefer the elevator over the window for exiting from the top floor of a high-rise building.

3.      Skepticism blocks all reasonable accounts of the progress of sci­ence and of common views about the world (e.g., the superiority of the views of experts over those of the ignorant or the views of the sane over those of the insane).

Now skeptics can easily parry these attacks, even though possibly not the traditional skeptics of the Pyrrhonist persuasion.

The first of these assertions is confusion between truth and cer­titude. Skeptics consider every statement doubtful, not false: some truths are obviously doubtful. Take the different bets in a given horse race. Obviously, all of the horses are uncertain and only one will be the winner. Here is a simpler argument. Consider any conjecture; if it is not true, then its negation is. Yet, both are doubtful. Hence, some true statements are doubtful. Skepticism is too, we contend. The dis­comfort that doubtful truths cause for some people, we suggest, is due to dogmatic education.

The second assertion rests on the false assumption that views are open to choice, that they abide by decisions. Indeed, confusingly enough, the very term decision making may even be used to mean draw­ing conclusions: decision theory rests on the idea that (in principle) a computer fed with the right information (about our psychological composition and circumstances) can decide for us better than we can. But, as Robert Boyle, Charles Saunders Peirce, and George Orwell^[27] asserted and as Benedict Spinoza and David Hume argued at length, beliefs are given. One cannot choose to believe that the sun will not rise tomorrow. And, because skeptics are at liberty to assume (tenta­tively, of course) that beliefs partake in the determination of conduct, they need not assume that the choice of a mode of conduct is arbitrary. We return to the psychological theory of the change of belief in the next chapter.

Contrary to the third assertion, skeptics may resort to common sense. Their critics do not have this right because common sense takes much for granted with no proof. Otherwise, common sense could never prove erroneous as occasionally it has. Thus, skeptics may explain scientific progress as a process of evolution. The superiority of the expert and the sane over the ignorant, the dogmatist, and the insane rests on the ability of the former to criticize the errors of the latter and, at times, to also point out options that they omit or ignore. In particular, at times skeptics criticize and improve the standards of criticism in specialized contexts. Indeed, the great skeptic historian, Armando Momigliano, suggested that all the improvements of standards of scholarship that historians regularly employ today are the product of skeptical historians of the relatively recent past.

Practical Implications

Let us consider cases in which we are asked to justify views or actions. Usually, when a view is considered true, the demand to justify it does not arise. The idea that it is considered true due to its justification is plainly false because too many views are received despite obvious criticism. We are usually not asked to justify errors that have led to successful actions. Mark Twain told the story of a commander who confused right and left so he made his soldiers meet head-on an ambush that was meant to hit them from behind; he won a medal. We are usually asked to justify errors that have led to disaster because we are accountable and must show that the disaster was not due to negligence. For example, commanders consult subordinates and ask for criticism of their plans. If the commanders reject the criticism with disastrous results, they must subsequently justify that rejection, which is not always easy.

Empiricism had a profound influence on our culture and mainly but not all to the good. Most reasonable and responsible people try to avoid error, but when they do err, they regrettably try to hide it because they do not like to admit it because empiricism says that their errors were avoidable. It relieves the tension all around to admit that even though, in most cases, the effort to avoid error is obviously laudable, the hope to completely succeed is plainly harmful. This is so because it puts excessive demands on everyone and thus makes all of us foolish and irresponsible every time we err, even though we all agree that some of us are wiser and more responsible than others. (As previously mentioned, one of the standard anti-skeptical allegations is that skepticism precludes the distinction between the judicious and the foolish, between the sane and the insane; the shoe is on the other foot.)

Few intellectuals find unsatisfactory theories that they had publicly declared satisfactory, especially if they are their own, and fewer are then ready to withdraw them openly and clearly. Even fewer intellec­tuals admit that this is the outcome of some justified criticism of their opponents. Most of the few who have admitted mistakes declared that they changed their mind following their own criticism. The admission of having made a mistake is still considered self-humiliation of a sort, and the admission that this was forced by others seems worse.

Politicians, to mention an obviously worse example, regularly try to steer clear of refutations by presenting views that they hope no one can refute. Instead of presenting any cogent political platform as, for example, in suggesting a detailed plan about how to reduce poverty, they only declare the intention to do so or to apply new social policies - but they do not specify which novelties they intend to introduce, much less how they could help - keeping their claims impossible to refute. This, then, implies a practical suggestion: when politicians present their ideas and plans, concerned citizens should not attempt to criticize them because it is futile; rather, they should insist on demanding some refutable ideas. This way, if their ideas turn out to be false, we may hope to be able to seek new plans for improvement, demand governments to adopt them, and, if need be, elect new candidates to office.

Philosophers habitually apply similar strategies. They reduce possi­bilities of being found in error by the use of methods the sole merit of which is that they prevent risk-taking by excluding the presentation of ideas that may be refuted if they are erroneous. Because such ideas are usually stale trivialities, philosophers use techniques that mask the staleness and triviality of their presentations; usually these tech­niques are verbal: they employ logical or metaphysical terms. The dif­ference between the two leading schools of philosophy today is in their choice between these two options: analysts employ logical terminolo­gies, whereas phenomenologists prefer metaphysical terminologies; the simultaneous employment of the two vocabularies is known by foes as fence-sitting, muddles, or mix-ups and by friends as syntheses, mediations, or displays of broad outlooks.

It is, of course, regrettable that the search for ways to avoid error leads to techniques for masking rather than revealing them. But, we sympathize: the demand for the impossible must lead to pretense, so it is better to dismount those proverbial high horses and reconcile with mere mortal fallibility. As skeptics, we suggest that the best way to do so is to join us in skepticism. We suggest the abolition of all norms that aim at error avoidance and replacement with humbler and more rea­sonable ones. Obviously, there are two extremes to avoid here: (ι) the irresponsible conscious advocacy of erroneous theories and the all- too-common indifference to the truth; and (2) the demand to avoid error and the all-too-common pretense that one is free of error. The aims should be to meet more reasonable targets, to avoid repeating corrected past errors, to raise questions, to seek new and interest­ing solutions, to debate them critically within the limits prescribed by circumstances, and to render these circumstances as conducive as possible to all critical assessments of new solutions. In brief, we rec­ommend the replacement of the pretense of being always in the right with responsible conduct. We elaborate on this in the next chapter.