There is broad agreement that knowledge is more than just true belief. What, though, must be added to true belief to get knowledge? According to traditional epistemology, two more ingredients are necessary: justification understood as having good reasons, and beliefs that are non accidentally true.

Having justified belief is not a way of possessing truth. This is because, unlike true belief, justification does not entail truth: a belief can be justified even though it is false. This does not mean that there aren’t some forms of justification that do entail truth – justification by mathematical proof might be one. Such forms of justification are called “infallible” justification. But most forms of justification are fallible: they don’t guarantee truth; e.g. justification by induction, or by the evidence of our senses. So, in general, having a justified belief does not entail that the justified belief is true.

Monism about epistemic values defends the view that truth is the primary epistemic value. It would appear that we want justified beliefs because we want true beliefs rather than we want true beliefs because we want justified beliefs. This would allows us to view the truth goal as the ultimate and primary epistemic goal. There aren't any intrinsic value accruing to justified belief that would be independent from the value of non-accidentally true belief.

Pluralism about epistemic values defends the view that truth is not the primary epistemic value. In addition to truth, there are many other epistemic values: for example, understanding, justification, and making sense of something. Among these, truth is not primary, but just one goal among others. Just truth, as primary candidate is unduly reductive. Epistemology is more than truth as a goal: it is the study of successful cognition. Thus the pluralist rejects the account of truth as the primary epistemic goal, touching upon such topics as the beliefs that are non accidentally true, and reliabilism. A central concept is that of the reliability of a process or method of belief formation.

Is epistemology about justified beliefs for their own sake? Truth can be regarded as the primary epistemic goal?

  • I like to think that the development of the process is as important as the product, the fallible truth. Commented Mar 4, 2013 at 18:13
  • In mathematics, this is also known by the 'formalism vs platonism' debate. You might want to check up the literature related to it. Commented Jun 13, 2018 at 14:59

5 Answers 5


Foundations don't require truth. They do require correlation to involve communication and a foundation for belief, but this is much weaker.

Systems that look to truth are very common throughout philosophy and have to face a very fundamental issue: where does foundational truth come from? There are many attempts at answers to this. Most forms of idealism dating from Plato on have had to find a way to relate ideal truth (as a metaphysical concept) with epistemic truth and what we can communicate about. Kant had his synthetic a priori as his point for the growth of truth into the epistemic world. Heidegger posited related transcendental relations of truth in being. Popper has tried to turn the relationship around by focusing on falsifiability and hypothesis, but still confronts the fundamental issue of where the meaning of concepts is meant to come from in this approach.

It's my belief these are all missing a much more fundamental relationship, and by attempting to look at bivalent ideas of truth as foundational are completely led askew. When you look at all the examples that Kant tried to show were consequences of synthetic a priori truth (things like logic, geometry, even at one point stating the inverse square law) - all have been invalidated by modern science (we live in a universe who's logic is not classical, it is an orthomodular logic based on the projection of Hilbert operators and noncommutative, our universe does not obey classical geometry, and the inverse square law is only a good approximation...). Similarly, it is well known in Computer Science that you don't build a semantics from a negative logic.

What you can do, though, is something that has been discussed by people like Quine, Putnam, Wittgenstein, and a number of phenomenologists: you can build meaning from correlation. This is now something that has rigorous mathematical foundations today, which actually refute a number of the initial philosophical objections to this approach. When you point to a scene with a rabbit and utter "rabbit" and another person is able to pattern-recognize a similar object, that correlation is able to begin a process of training meaning into words. If there is any correlation in shared experience, repeated utterance is sufficient to begin to build shared meaning. This is not sufficient to build absolute truth at any point, but that's not necessary when the correlation makes likelihood high enough for information transfer. Information theory codifies this in the concept of channel communications and symbol error. Effectively, repeat usage of symbols trains the relationship between syntax and semantics to a certain likelihood (never certainty).

The fact that language appears to exist is a good indication that shared experience is likely happening fairly frequently. In this way, you don't need foundational truth to build a theory of epistemic values. Epistemic values come from likely semantics of our experience, interpreted possibly in various terms like "experienced" or "exists" indicating internal states or their mapping to metaphysical ontologies.

  • Is justification internal or external? I am an externalist like you. It is a very good question, but it is a different question. There are two modes of epistemic evaluation: objective and subjective. The former is uncontroversially externalist. The latter is concerned with epistemic appropriateness, or responsibility. According to internalists, that is precisely the kind of evaluation that is subject to the internality constraint. Externalists challenge the assumption that knowledge requires the possession of good reasons. +1 for your reasoning on foundational truth. Commented Mar 5, 2013 at 12:24
  • 2
    @ex0dus:A minor point - Kants understanding of geometry hasn't been invalidated by modern science. That there are other consistent geometries mathematically is interesting but besides the point. That physical spacetime is non-euclidean in-the-large is also besides the point for us who live in-the-local. As a matter of fact a manifold is precisely defined as being locally flat, that is locally euclidean. Flat-Earthers would have been correct had they insisted that the Earth was locally-flat. But this is also besides the point for Kant. Commented Mar 5, 2013 at 12:43
  • What we apprehend immediately to our intuition is euclidean space and this is in us. If we then conceptually (that is scientifically), even locally determine spacetime is otherwise doesn't change the fact that our immediate apprehension says otherwise. Commented Mar 5, 2013 at 12:47
  • +1 - I think what you say is true. Er, wait. I find a correlation betw...no...there has been information transfer...uh. Well. In any case, I agree. (But with the additional point that the next step is to recover the word "true" as it is a very good approximation to various properties of one's correlational models, and it's a dreadful chore to communicate without such a shorthand.)
    – Rex Kerr
    Commented Mar 5, 2013 at 16:55
  • "you can build meaning from correlation. This is now something that has rigorous mathematical foundations today, which actually refute a number of the initial philosophical objections to this approach" - which are these approaches? How do they relate to modal Bayesian logic?
    – Nikolaj-K
    Commented Jun 6, 2013 at 9:38

My own view is that epistemology centres on rational belief, not on truth or the related concept of knowledge. We can certainly achieve rational belief. Whether we can achieve truth-as-knowledge is a separate and more dubious matter.

Very roughly indeed, a belief (Β) is rational for a person (Ν) at a time (t) if and only if Β is more probable than not-B on the basis of the evidence available to Ν at t. (Gary Iseminger, 'Successful Argument and Rational Belief', Philosophy & Rhetoric, Vol. 7, No. 1 (Winter, 1974), pp. 47-57 : 48.)

If we look back on the history of virtually any discipline, particularly the natural sciences but other subjects as well, we find only beliefs that once were believed to be true - to amount to knowledge - but in the event were at best rational beliefs. Historically, knowledge claims turn out to be false; truth claims are invalidated. But rational belief remains undented.

It was rational for Euclid, given the evidence available to him, to believe that his geometry described space. It does describe one kind of space but not a space with negative curvature. He was wrong to think of space simpliciter as 'flat' and homogeneous. He thought he knew what space was like but he didn't. But his belief was rational.

Geocentrism was a rational belief in the West in the early middle ages. Muslim scholars had established considerably earlier a number of doubts about geocentrism at least in the form in which it had been formulated by Ptolemy but their work was not known to the major and most influential Western scholars. Geocentrism, once seen as a view known to be true, is now regarded as false. But on the evidence available in the West, it was a rational belief. It is still a rational belief, relative to the evidence then available; it is not rational given the evidence available now.

My point is essentially that we can attain rational belief; we want more, we want knowledge, but we don't actually need more than rational belief and are in any case unlikely to get more. In epistemology I put truth second in the queue behind rationality.



Epistemology is about asking the question "How?"

How do you know?

How did you find the answer? (Not, "what is the answer?")

An apt analogy would be with Formal logic where the form or validity of an argument is studied independently of 'truth values', or the truth of the conclusion.

Truth is usually thought of as a "What".

What is the conlusion?

What does really Exist?

If Truth is the goal of any philosophical branch, it would be Ontology.


I don't think that it is. Etymologically philosopher is friend of wisdom, not of alethia-truth. Friendship (in its true sense) is reciprocal, so wisdom comforts the philosopher as refuge and retreat, as the Stoics viewed it.

Truth is an ideal and (infinitely) distant. Truth in its own place may be a whole, but from our own finitude truth is many. There is not one goal but many and each as distant from each other as we from it.

  • I am a pluralist about epistemic values too. But I don't think that “truth is an ideal and (infinitely) distant”. Many past theories were not approximately true or truthlike. Ptolemy's geocentric theory was rejected in the Copernican revolution, not retained in the form “approximately Ptolemy”. Indeed, the progressive steps from Ptolemy to Copernicus or from Newton to Einstein are not only matters of improved precision but involve changes in theoretical postulates and laws. Commented Mar 5, 2013 at 15:28
  • There is no theory-independent way to reconstruct phrases like "really there", each theory has its own ontology. Convergence to the truth scientific progress seems to be impossible, if ontologies change with theories observations and ontologies are relative to theories. Science is progressive only on values other than the truth, such as simplicity, predictive accuracy, comprehensiveness, and requirements for consistency. Commented Mar 5, 2013 at 15:29
  • @Bevilaqua: Newtons theory had a particle move in uniform motion in a straightline in flat space; Einsteins theory has it moving uniformly in a straightline in a curved space. In this perspective Einsteins theory is no more than a tweak. I'm not using 'really there' or 'truth' in a theoretic-pinned-down way; perhaps poetically, perhaps religiously but definitely ambiguously. Commented Mar 5, 2013 at 16:11
  • @Bevilaqua: One can only speculate as to why Ptolemys Geocentric world was taken up rather than the Heliocentric World, proposed by Aristarchus. Perhaps a future generation may question why an incredibly Baroque Theory as String Theory was taken up by a whole generation of physicists when perhaps a theory with the right idea but in the wrong location or not looked at the right way now by the community was available. (This shouldn't be taken as me knocking String Theory - I'm not). Commented Mar 5, 2013 at 16:21

Truth is not available in a concrete formula. ‘Justified beliefs’ allows us to follow working models that are true only until they are not true. The best we can get to as human beings is “what works – for now” One of the reasons for that, ironically, is how we seek the ‘truth’. The Cartesian and Galilean methodology for research, which we base most educational systems on, has as its main questions, “What do you know, and how would you prove it is valid?” This became Western view of epistemology.

The Greeks held that epistemological truth is available through dialogue, which would make it transitory. Relating it back to my first sentence, that for humans, truth is not concrete formula.

Cartesian doubt allows us to deconstruct, take apart, and use doubt to reconstruct, what we know. When it works, we’ve succeeded. When it doesn’t, we keep looking for what works. That is not the same as finding the ‘truth’. We humans are temporal, impermanent and operate as using our sentient frailties as we do with our mental acumen. The ‘truth’ we seek is usually evidence for what we wish to believe. Time is not a friend of the ‘truth’ as it gets poked, prodded, fictionalized and retold told fit within the present moment, again and again. When the idea that holds that truth together is no longer solid, the clarity it was founded on waivers. It is only the ‘truth’ as long as the results we base it on remain solid and valid.

What we have instead is rote and processional data. “Columbus discovered North America in 1492.” = rote data. “Control + Alt + Delete keys will bypass normal computer functions to immediately apply administrative functions.” = processional data. And they would seem closer to qualifying as concrete truths than most justified beliefs, but they are no where near as interesting or engaging.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .