Knowledge is always true


The analysis of the concept of knowledge is one of the main tasks of epistemology. This reflects the current discussion: After the enormous flood of essays that Edmund L. Gettier's question "Is Justified True Belief Knowledge?" Had triggered in 1963 ebbed noticeably in the 1980s, the last decade has seen a wealth of interesting new approaches and further developments of known positions.

1. Basics

The search for a correct analysis of the concept of knowledge is primarily motivated by the skeptical challenge. As is well known, there are two basic skeptical arguments: the Agrippa trilemma and the argument from the skeptical hypotheses. The Agrippa Trilemma is based on the assumption that only he who knows that p, who has good reasons and thus a justification for this belief. Since the reasons must also be known, if you try to justify a belief, you end up in an infinite regress or a circle, or you break off the justification dogmatically. In all three cases, the original conviction must be considered unjustified and therefore does not represent knowledge. The argument from the skeptical hypotheses is based on the assumption that we have skeptical hypotheses - for example the hypothesis that an evil demon deceives us in all things (René Descartes), that we have brains in the tank (Hilary Putnam), that the world could only have arisen 5 minutes ago with all the evidence of the past (Bertrand Russell) etc. - we cannot clear it up. But if we can't, then we have no knowledge (at least in certain areas). Both arguments worry us because they fundamentally question our knowledge and thus our position in the world. We just don't want (or can't) believe that we can't know anything (or next to nothing).

The Agrippa trilemma is obviously based on a certain conception of the concept of knowledge: If and only if justification (in the manner described) is part of the nature of knowledge, one gets into the trilemma. The argument therefore shows, if it is successful, that we cannot know anything. The argument from the skeptical hypotheses, like the Agrippa Trilemma, is ultimately based on the assumption that knowledge requires reasons: namely, the skeptical hypotheses cause us to look at our beliefs (in a certain area) at once (Barry Stroud). Since, as it seems, we cannot necessarily give reasons for all of our beliefs (from a certain area) - namely because no corresponding beliefs remain as soon as we want to justify them all at once - we cannot know (in certain areas) to have. (The argument, however, allows a second reading: According to this, it is only intended to show that we de facto (in certain areas) have no knowledge because the skeptical hypotheses (allegedly) give us reasons for doubting certain beliefs. In this second reading the argument is much weaker; I will not consider them further in the following.) Since both arguments are thus based on a certain conception of the concept of knowledge, dealing with the skeptical challenge requires an analysis of this concept.

According to the standard analysis of knowledge already ascribed to Plato, knowledge is nothing more than true, justified conviction. The fact that one cannot know what is not the case is (almost) indisputable. The fact that one cannot know what one does not even believe is more controversial, especially because of certain examples that Colin Radford brought into the discussion. Most questions, however, are raised by the condition of justification. At first glance, it is necessary because a belief that is only accidentally true cannot count as knowledge and justification is the obvious way to rule out chance. (The question of to what extent epistemic happiness is compatible with knowledge is discussed in detail today by Duncun Pritchard, among others.)

The well-known examples by Edmund L. Gettier are intended to show that the three conditions of the standard analysis are not sufficient. Consider one of them: Smith has good reason to believe that Jones will get a certain job and that he has ten coins in his pocket. Smith concludes that whoever gets the job has ten coins in his pocket. In fact, he himself, Smith, will get the job, and coincidentally he also has ten coins in his pocket (which he does not know, however), so that whoever is going to get the job actually has ten coins in his pocket. Smith thus has a true, justified belief, and yet we would not ascribe any knowledge to him.

Above all, this consideration has provoked two fundamentally different reactions: Internalists assume that the justification condition of the standard analysis is actually necessary. What turns a true conviction into knowledge must therefore, at least partially, be accessible “internally” to the knowledge subject. In order to escape the Gettier examples, an additional, fourth condition is then required. Externalists, on the other hand, are of the opinion that justification (understood internally) is not necessary for knowledge. What distinguishes knowledge from true conviction must therefore be described in a new third condition. Let's take a closer look at the two approaches.

2. Internalistic theories of knowledge

In the example above, Smith is justified insofar as he can be described as rational, or in so far as he fulfills his epistemic duties and lives up to his epistemic responsibilities. It is justified because its evidence speaks for the aforementioned belief. In the opinion of internalists (especially evidentialists) this is also necessary for knowledge. Nevertheless, Smith's justification from the outside is flawed insofar as the reasons he can give for his true beliefs are ultimately not good reasons: After all, his justification is based on a false belief, the belief that Jones will get the job. An obvious fourth condition is therefore the requirement that justification must not be based on false beliefs (Gilbert Harman). However, the following example made known by Alvin Goldman shows that this condition does not solve the Gettier problem: During a trip to the country, Henry sees a barn from the window of his car and is convinced that he has seen a barn. However, there are numerous barn dummies in the area (for whatever reason) that Henry would have thought were real barns from the car. So it was just a coincidence that he had just looked out the window of the only real barn in the area and therefore came to a true, justified conviction. Accordingly, we would not ascribe any knowledge to him. Nor can one say that Henry's justification is based on a false belief: his justification is in a sense not based on reasons at all. Rather, Henry lives up to his epistemic responsibility simply by the way he acquires his conviction.

Examples of this kind lead to the development of more complex fourth conditions in what are known as the theories of incontestability, as proposed by Keith Lehrer, Marshall Swain and John L. Pollock, among others. For example, a suggestion by Peter D. Klein is as follows: Person S only knows that p if S's justification cannot be challenged with additional information. Henry's justification, for example, could be challenged by pointing out that numerous barn dummies can be found in the area. That's why he has no knowledge. However, as Keith Lehrer's Grabit example shows, this condition seems to be too strong: Suppose Jones sees Tom Grabit tucking a book under his coat in the library and leaving the room with it. Jones is justified in believing that Tom stole the book, which in fact he did. But let's continue to assume that Tom's father, who is in the mental hospital, often speaks of Tom's kleptomaniac twin brother who doesn't even exist. (Tom's thieving tendencies drove the father into this delusion.) Jones's justification in these circumstances could well be challenged by the selective information that Tom's father claims there is a twin brother. Nonetheless, according to teachers, we ascribe knowledge to Jones. So the condition of Klein is not a necessary condition for knowledge.

The fourth condition that the internalist needs should look more like this: Person S only knows that p if the justification of S cannot be challenged by relevant additional information. The information from Tom's sick father's allegation is not relevant but misleading; the information from the dummy barns is not misleading but relevant. The crucial question is then: What distinguishes relevant information from irrelevant information? All theories of incontestability are confronted with this problem, which I call the relevance problem. It has not yet been resolved to everyone's satisfaction.

At first glance, the internalist plays into the hands of the skeptic. Both the Agrippa trilemma and the argument from the skeptical hypotheses begin with the (internalistically understood) condition of justification, which is accepted here as a necessary condition for knowledge. Accordingly, one can only escape the skeptical consequence if one can show that breaking off justification is not necessarily dogmatic (fundamentalism), that a circle is not necessarily "diabolical" (coherentism) or that recourse is only the result of a wrong understanding of justification is (contextualism). All three strategies have been extensively researched.

3. Externalistic theories of knowledge

According to externalistic theories of knowledge, even the (internalistically understood) condition of justification is a mistake: in fact, knowledge and true conviction differ through factors that do not necessarily have to be accessible to the knowledge subject. A first suggestion for an alternative to the justification condition is as follows: The person S only knows that p if this true belief was caused by the fact that p. However, such a causal theory of knowledge, as worked out in particular by Alvin Goldman, Fred I. Dretske and David M. Armstrong, also fails with Goldman's barn example: Henry essentially believes that he drove past a barn because he drove by drove past a barn. Yet we do not ascribe any knowledge to him. Reliabilistic knowledge theories overcome this weak point in causal analysis. Accordingly, it is important for knowledge that the corresponding true conviction has come about in a reliable way. Henry's conviction did not come about reliably because he cannot distinguish between barns and dummy barns (Goldman's principle of discrimination). For this reason we do not ascribe any knowledge to him. Reliabilism is a generalization of causal theory: often correct causation is a reliable method, but not always. Reliabilistic theories of knowledge were represented in various variants, especially by Alvin Goldman, Fred I. Dretske and Robert Nozick. The concept of the relevant alternative often plays a decisive role: You have acquired your conviction in a reliable way if you are able to dispel all relevant alternatives.

The central problem with which the reliableist is faced is the so-called generality problem (Earl B. Conee, Richard Feldman). It consists in the fact that the reliabilistic knowledge analysis presupposes that one distinguishes between reliable and non-reliable methods. To do this, however, it must first be clear which method is actually used to acquire a belief. For example, did Henry use the (unreliable) look-out-of-the-window-near-barn-dummies? Or is it the (generally reliable) method of looking out of the window? Or did he even use the (even in a special case reliable) method of looking-out-of-the-window-at-the-only-real-barn-near-barn-dummy? As long as the reliableist cannot say which is the relevant method, his analysis remains vague. We are dealing with a variant of the relevance problem described above! This becomes particularly clear if one looks at the form of reliability, in which reference is made directly to relevant alternatives: Here it would again be important to specify a general criterion for what makes an alternative relevant.

In relation to the skeptical challenge, the externalist is initially better off. Since he denies the condition of justification, the Agrippa Trilemma finds no starting point at all. Nevertheless, the skeptical hypotheses remain dangerous, as they show that we cannot differentiate between certain alternatives and thus cannot reliably come to our convictions. In order to get around this skeptical argument, externalists often dispute the principle of unity, that is (to put it simply) the assumption that a subject who knows that p, and who knows that q follows from p, also knows that q. How successful this antiskeptical strategy is cannot be further discussed here.

The decision between externalism and internalism is not an easy one. Laurence BonJour has suggested a number of examples to show that externalism is untenable. These examples are based on the fact that we do not want to ascribe knowledge to persons who arrive at true convictions in an irrational way, even if the method they use is de facto reliable. However, these examples are countered by the intuition that small children and animals can acquire knowledge, even if there is hardly any question of justification, epistemic obligations, etc.

4. Naturalized epistemology

Proponents of the naturalized epistemology initiated by Willard V. O. Quine mostly start from an externalistic analysis of the concept of knowledge. In principle, this approach is shaped by the conviction that it is impossible to refute the skeptic. Epistemology should therefore turn to another task, namely the investigation of the way in which we (if one disregards the skeptical challenge) actually come to true beliefs about the world. According to the naturalist, this research should be carried out in direct cooperation with the empirical sciences, in particular with psychology, ethology, biology and the neurosciences. These sciences provide us with crucial information about how we arrive at beliefs about the world through perception and reflection and under what circumstances the corresponding mechanisms are reliable.

Certain results of empirical psychology then show, for example, that we tend to systematic errors in special situations (for example with regard to the assessment of probabilities), ethology shows fundamental similarities and differences between animals and humans with regard to cognitive abilities, and the theory of evolution sheds light the development of our cognitive faculties. As a reliableist, one can say that all of these researches make mechanisms of knowledge understandable and assessable with regard to their reliability and thus also clarify the nature of our knowledge. In addition, the concepts of truth, conviction and reliability are less problematic from the naturalist's point of view than the concept of justification. The latter has a normative component prima facie, the former lacking prima facie. Critics of the naturalistic position, such as Jaegwon Kim, argue that even the concept of conviction is tied to the concept of rationality and is therefore also normative. Naturalists, such as Hilary Kornblith, on the other hand, try to show that many prima facie normative terms are actually unproblematic for the naturalist.

The previous considerations tacitly assume that the nature of knowledge can be clarified by clarifying the concept of knowledge and that this can be done by analyzing the sentence “S knows that p”. Naturalists are critical of this view. Hilary Kornblith argues that epistemology should examine not the concept of knowledge, but knowledge itself. However, he does not want to forego exploring the nature of knowledge. According to Kornblith, however, the means of empirical science should be used. It is true that one has to start from our everyday understanding of the word “knowledge”. This understanding then has to be deepened or revised through scientific research.Just as physicists did not discover the nature of gold through conceptual analysis (although they too initially examined samples that we would typically refer to as "gold"), so too should the epistemologist discover the nature of knowledge not through conceptual analysis, but through empirical research. Saul Kripke's idea of ​​the possibility of recognizing necessary connections a posteriori forms the basis for these considerations.

5. Virtue epistemology

Another approach, which is currently being intensively discussed, also starts from externalist theories of knowledge: virtue epistemology. Any innate or acquired habit that reliably leads to the acquisition of true convictions is considered an epistemic virtue. Our perception is just as important here as the striving for coherence (Ernest Sosa). Knowledge is then to be analyzed as true belief acquired through the exercise of an epistemic virtue. Various philosophers have replaced this image of an epistemic virtue with a more internalist view. Just as the right attitude belongs to virtue in ethics, the epistemically virtuous person must also live up to his epistemic responsibility (James Montmarquet). This includes, among other things, the willingness to appreciate (foreign) arguments as well as the openness (not excessive to gullibility) to new knowledge. The most important approaches to virtue ethics (Linda T. Zagzebski, John Greco) understand the concept of epistemic virtue both internalistically and externalistically: According to the Aristotelian model, virtue not only includes a motivational component, but also the component of reliability. The one who acts for (internalistically seen) wrong reasons can therefore no more be considered virtuous than the one who is (as a rule) unsuccessful in his actions. Accordingly, the concept of knowledge is not analyzed in a purely externalist manner.

What all the theories of virtue epistemology have in common is the idea that epistemic concepts must ultimately be analyzed by referring to properties of the knowledge subject. The concept of epistemic virtue is therefore to be seen as fundamental in these approaches. On this basis one then tries to make progress both with regard to the gettier problem and with regard to the skeptical challenge. Whether virtue epistemology really brought new aspects to light is a matter of dispute. In any case, the interesting connection between epistemology and ethics becomes particularly clear in these approaches.

6. Contextualist theories of knowledge

The theories discussed so far assume that there is a “fact of knowledge”, that is, the question of whether a person knows something or not has to be answered regardless of the context of the person asking the question. In contrast, the basic thesis of contextualistic theories of knowledge is that the truth value of an ascription of knowledge depends on the context of the ascription. Just as the truth value of the sentence “The Lüneburg Heath is flat” depends on whether it is uttered in the hiking club or when meeting high-speed test drivers, the truth value of the sentence “S knows that p” should also depend on the context of the utterance. The standards that someone has to meet in order to be considered a knower are therefore not always the same.

The question of which aspects of the utterance context exactly determine the truth conditions of “S knows that p” is answered differently by different contextualists. According to David Lewis, for example, the question of which alternatives the knowledge writer has become aware of plays a decisive role: For a person S to know that p, these alternatives must be excluded by the evidence of S (among other things). On the other hand, many of the alternatives, to which the knowledge writer did not become aware, can be ignored.

Contextualist theories of knowledge have been intensely researched in recent years. Different variants are represented by David Lewis, Stewart Cohen, Michael Williams and Keith DeRose, among others. The main motivation for these approaches is again the skeptical challenge. If the truth value of a knowledge ascription depends on the context of the ascription, a new way of dealing with the skeptic opens up: You can then say that the skeptic is right when he (in his context) claims that we have next to nothing knowledge. However, the man on the street is also right (in his context) when he claims to know a lot. The attribution contexts are different (for example, because in one case attention is directed to alternatives that may be ignored in the other). The skeptic would not be refuted, but “domesticated”. (Michael Williams goes a lot further here and tries to show that the skeptical arguments are based on "unnatural" theoretical assumptions.)

Whether the skeptical challenge can be averted satisfactorily in this way is controversial. Equally controversial is the basic thesis of the contextualist that the attribution of knowledge is actually context-dependent in the required way. Prima facie they are not. Sentences such as "Helga says something true when she says" S knows that p ", but S doesn't know that p" initially sound paradoxical, but should be able to be true if the contextualist is right (because Helgas Context can be different from the context of the person uttering the said sentence). Nikola Kompa speaks here of the "unpleasant consequence" of contextualism. These and a whole series of other linguistic observations, but also the connection between knowledge, assertion, epistemic possibility and practical consideration, make contextualism difficult to create (John Hawthorne, Jason Stanley).

7. Beyond contextualism

Based on these observations, there are various attempts to save the insights and strengths of contextualism without adopting its weaknesses. A new approach, represented by John Hawthorne and Jason Stanley, among others, shifts the context dependency from the writer to the knowledge subject: Whether or not knowledge can be ascribed to him should depend on his situation. That sounds all too familiar at first, as it is the starting point of all non-contextualistic (i.e. invariantistic) theories of knowledge. The novelty of subject-sensitive invariantism is that here non-epistemic factors are seen as relevant for the truth value of the attribution of knowledge. Let us consider the following two cases (Keith DeRose): (1) It is Saturday. Helga would like to deposit money in the bank. She believes the bank is open because she also deposited money on Saturday two weeks ago. There is of course the possibility that the bank has changed its opening hours, but for Helga it does not depend much on whether she pays in the money on Saturday or not until Monday. (2) As in the first case, Helga would like to deposit money on Saturday. This time, however, a lot depends on the fact that the money is actually in the account on Saturday, otherwise a check will burst, which is associated with a lot of trouble. But Helga doesn't know that. If we are inclined to say that Helga can rightly ascribe knowledge in the first, but not in the second, then this is probably because Helga's practical situation has changed. Your epistemic situation, on the other hand, has remained the same.

In my opinion, one of the main problems with this approach is that the intuitions on which it is based are extremely ambiguous. With reference to the examples cited, the following applies: If one concentrates on the possibility of error, one considers Helga's claim to knowledge to be inadequate, even if little for her depends on the truth of her convictions. If, on the other hand, you concentrate on the fact that the bank has actually not changed the opening times - you have to assume that in the example, because otherwise you are dealing with a false conviction, which is then of course no knowledge - you are inclined to do so To recognize Helga's claim to knowledge even if more depends on it for her. The extensive use of ambiguous examples, however, is a weak point of various approaches.

An alternative to this is to examine the interests that guide us in using the word “knowledge”. Following Oswald Hanfling, I myself differentiate between two fundamentally different interests, each of which requires its own analysis of knowledge: In one case, we are concerned with recognizing good informants - this function of the concept of knowledge was examined above all by Edward Craig - in the other case it is possible it is a matter of identifying potential recipients of information. Accordingly, one must distinguish between two types of knowledge: a perspective (i.e. contextualistic) and an objective one (i.e. one that is independent of the context of the assignee). In my opinion, this distinction makes it possible to adopt the strengths of the contextualist approach without failing because of the weaknesses. This not only results in a plausible starting point for dealing with the skeptic, but in particular a possibility of solving the relevance problem.

One way out of the difficulties that the analysis of the concept of knowledge raises is to dispense with such an analysis: Timothy Williamson argues for the thesis that the concept of knowledge should be viewed as fundamental (and not analyzed using the concept of belief). Knowledge is about the “most general active mental state”. Williamson applies a view that semantic externalists have developed in relation to the concept of belief to the concept of knowledge. This gives him a new starting point for dealing with epistemological questions, including clarifying the concept of evidence and the relationship between knowledge and assertion, as well as dealing with the skeptic. The thesis that knowledge is a mental state is subject to various objections, in particular the objection that we have special access to our mental states that we do not have to our knowledge. Williamson denies that. His approach is discussed intensively in current epistemology.

Overall, today's debate about the concept of knowledge is shaped by refined variants and further developments of established approaches on the one hand, but also by real new beginnings on the other. There is an increasing intertwining of epistemological questions with topics of the philosophy of language (especially in the context of contextualism) and practical philosophy (especially in virtue epistemology and subject-sensitive invariantism).


Gerhard Ernst is a private lecturer at the Philosophy Department of the Ludwig Maximilians University in Munich. He has published on the topic:
- Introduction to epistemology. 168 pp., Kt., € 14.90, 2007, Wissenschaftliche Buchgesellschaft, Darmstadt (including detailed references).
- The problem of knowledge. 232 pp., Kt., € 31.50, 2002, Mentis, Paderborn.


Overview literature:

Baumann, Peter: Epistemology: Textbook Philosophy. 300 pp., Kt., € 19.95, 2002, Metzler, Stuttgart.
Blaauw, Martijn / Pritchard, Duncun (eds.): Epistemology A-Z. 208 p., Ln. £ 40.—, kt. £ 10.—, 2006, Edinburgh University Press, Edinburgh.
Fogelin, Robert J .: Pyrrhonian Reflections on Knowledge and Justification. 256 pp., Ln., £ 49.85, 1994, Oxford University Press, Oxford.
Steup, Matthias; Sosa, Ernest (ed.): Contemporary Debates in Epistemology. 360 S., Ln., £ 21.—, 2004, Blackwell, Oxford.

Current monographs:

Craig, Edward: Knowledge and the State of Nature. An Essay in Conceptual Synthesis. 182 S., Ln., 1999, Oxford University Press, Oxford.

Greco, John: Putting Skeptics in Their Place. 280 S., Ln., 2000, Cambridge University Press, Cambridge.

Grundmann, Thomas: On the trail of the truth. A defense of epistemological externalism. 403 pp., Kt., € 50.—, 2003, Mentis, Paderborn.
Hawthorne, John: Knowledge and Lotteries. 214 pp., Kt., £ 18.—, 2004, Oxford University Press, Oxford.

Kompa, Nikola: Knowledge and Context. A contextualist theory of knowledge. 166 pp., Kt., € 37.80, 2001, Mentis, Paderborn.

Kornblith, Hilary: Knowledge and Its Place in Nature. Ln. £ 35.—, 2002, kt. £ 16.—, 2004, Oxford University Press, Oxford.

Teacher, Keith: Theory of Knowledge. 224 pp., Kt., € 24.95, 2000, Westview Press, Boulder / CO.

Pritchard, Duncun: Epistemic Luck. 304 p., Ln., £ 40.—, 2005, Oxford University Press, Oxford.

Sosa, Ernest / BonJour, Laurence: Epistemic Justification. Internalism vs. Externalism, Foundations vs. Virtues. 248 p., Ln. £ 65.—, kt.,. € 21.—, 2003, Blackwell, Oxford.

Stanley, Jason: Knowledge and Practical Interest. 208 S., Ln £ 27.—, 2005, kt. £ 23—, October 2007, Oxford University Press, Oxford.

Williams, Michael: Problems of Knowledge. A Critical Introduction to Epistemology. 288 pp., Kt., £ 20.—, 2001, Oxford University Press, Oxford.

Williamson, Timothy: Knowledge and Its Limits. 352 pp., Ln., £ 45.—, 2000, Oxford University Press, Oxford.

Zagzebski, Linda T .: Virtues of the Mind: An Inquiry into the Nature of Virtue and the Ethical Foundations of Knowledge. 383 S., Ln. £ 60.—, kt. £ 23—, 1996, Cambridge University Press, Cambridge.