Are the Conditions of Belief, Truth and Justification Necessary and Sufficient for Knowledge?

World Knowledge MedalThis is an essay I wrote back in 2008. I evaluate the ‘standard analysis’ of the conditions for knowledge in the light of Gettier’s counter-examples. I cover three possible solutions: 1) redefining the conditions of knowledge (I follow especially Nozick’s conditional theory as advocated by Dancy, 1985), 2) explaining away Gettier’s refutation (particularly following Fogelin, as elaborated in Williams, 2001), and 3) radical scepticism. These positions each have merits in their own ways, but it is the conditional theory that is most plausible and as a result the elaboration of the standard analysis found in the question above should be rejected.

The standard analysis of knowledge is so called because it was the most popular from at least the time of Plato [1] until the mid 20th century. It is also called the tripartite account because it is composed of three conditions, purported to each be necessary, and to collectively be sufficient to ascribe a belief state as one of knowledge, namely: belief, truth and justification.[2] It can plainly be seen that these conditions (or an equivalent in the case of justification) are necessary because they each exclude a particular fatal deficiency in the construction of knowledge, respectively: ignorance, error and opinion.[3] In the tradition inherited from Plato there is a strong qualitative distinction between knowledge and mere opinion. We do not know something if we believe it for no reason and it turns out to be true. Such ‘right opinion’ has equal utility to knowledge, but it is characterised as ephemeral because it is necessary for it to be ‘tied down’ so we can be certain it is right.[4] Some contemporary theorists, including Fogelin, however, do not think that knowledge needs to be so stable.[5]

In an article published in 1963, Gettier put forward possible scenarios in which people have justified true beliefs that do not meet our intuitions for knowledge, thus refuting the tripartite account.[6] One of his counter-examples runs as follows: two other people are in my office and I am justified on the basis of much evidence to believe that the first owns a Ford car; unbeknown to me they no longer do. I am equally ignorant of the fact that the second person owns one. I believe truly and justifiably that someone in my office owns a Ford car, but I do not know someone does, and therefore justified true belief cannot be sufficient for knowledge.[7] This provokes a response of scrutinising Gettier’s argument or altering the analysis, perhaps adding new conditions. There is also the possibility of scepticism, for whilst we have a capacity for justifying our beliefs, it may be we are not able to meet the extra conditions that turn justified true belief into knowledge.[8]

First I will consider constructive responses which aim to alter the analysis to escape the ‘Gettier problem.’ Two important attempts prior to the conditional theory were that of indefeasibility and the causal theory. The indefeasibility approach holds that the Gettier problems arise because there exist truths which would have destroyed the agent’s justification had they been aware of them. Thus, we add the criterion that the justification for the belief is indefeasible (not liable to be destroyed by further truths).[9] This, however, is too strong a requirement because it requires us to know a great many other truths. Moreover, there are often truths which, taken in isolation, are enough to defeat our justification in cases of true belief, so it has the consequence that our range of knowledge would be dramatically reduced. In attempting to solve this problem the theory collapses into the alternative criterion that our justification would not be defeated by the sum of all truth, which effectively amounts to its refutation because we cannot be required to know all truth before we can have justified belief.[10]

Goldman’s causal theory comprises the criterion ‘s knows that p if and only if the fact that p is causally connected in an appropriate way with s’s believing that p.’ [11] This is initially very plausible because it excludes true beliefs which are justified by coincidence, but it does suffer from major difficulties. First, facts are true propositions which reflect the world rather than affect it; events alone, not facts can stand in causal relations. Furthermore, establishing universal truths is a much bigger problem; there does not appear to be any causal link between the fact that all men are mortal and my belief that all men are mortal, for example. Instead it is known by examples from which the generalisation is inferred. Ultimately, while we may be able to form a view of facts in which they cause knowledge, this theory is not viable due to its failure to account for universal truths.[12]

The conditional theory of Nozick has a strategy of confirmation similar to the casual theory; it is general enough to encompass a casual relationship, but subtle enough to avoid the problems associated with the causal relations of facts. The theory replaces the third (justification) criterion with: ‘If p were not true, s would not believe that p’. This leaves the problem of how to prove that something isn’t knowledge if, in a slightly different circumstance it remains true but a no longer believes it (if it is possible for p to be true without a believing it, s’s reason for belief isn’t sufficiently dependent on its truth). For this reason the forth criterion is added: ‘If it were the case that p (in changed circumstances), it would be the case that s would still believe that p.’ These criteria are subjunctive conditionals, they test the reliability of the justification against counterfactual conditions.[13] For a belief to be knowledge it must therefore be particularly sensitive to the truth of the proposition believed; when it is true, s believes it; when it is false, s does not believe it. Nozick calls this property ‘truth-tracking’.[14] The objections to this theory are similar to that of the indefeasibility approach- that it is too abstract and demanding. But it is a very plausible solution and its high standards provide us with knowledge that remains stable.

I turn now to a major competitor to the conditional theory, Fogelin’s conception of unstable knowledge. Fogelin does not take the Gettier problems as having refuted the tripartite account, but to have shown its justification criterion is too vague. The criterion can be interpreted in two ways: personal entitlement to believe a proposition and adequate grounds for the proposition itself being justified. The former way, ‘epistemic responsibility,’ we ascribe when a standard of behaviour is met, that belief is responsibly formed and held. The latter way is more objective and outcome-orientated, we ascribe it when epistemic procedure reliably establishes a truth. Epistemic responsibility is fundamental because it is subjective and more practical than the ‘god’s eye view’ demanded by adequate grounding.[15]

A feature of the Gettier examples is that they require us to imagine having extra evidence not accessible to the agent, meaning that we are aware not only of their epistemic responsibility but of the grounding of their belief also, whereas they only have sufficient amounts of one of these to warrant the label ‘justified’.[16] For Fogelin this is the key to the problem raised by Gettier, and he proposes new criteria for the analysis accordingly, such that both types are necessary: “s knows that p, if and only if s justifiably came to believe that p on grounds that establish the truth of p.” [17 ]

Superficially, the condition of ‘grounds that establish the truth of’ may sound too much like the unworkable demands of indefeasibility, but there is an important difference owing to Fogelin’s conception of knowledge. He argues that to claim the right to be sure and to admit fallibility is not inherently contradictory. Focusing on the epistemic responsibility element of justification rather than on adequate grounds inclines us towards fallibilism. If we accept fallibilism then we are liable to accept, as does Fogelin, the ‘instability of knowledge’ (a further departure from the conception of knowledge in Plato).[18]

In the common interpretation, only the stable, final ‘truth of the matter’ can be associated with knowledge, but this is not coherent with our actual practice of knowledge where being able to perceive the end truth is not workable, as indefeasibility has taught us. Rather, knowledge varies with the addition of new information, thus, he argues, this conception of instable knowledge is more plausible than the other views.[19]

Having encountered the contrasting views of Nozick and Fogelin it can be seen that our choice of theory depends upon the strength of criteria for knowledge we hold; in contrast to both these views sceptics argue that in the absence of a wholly sound analysis of empirical knowledge we must fall back on deductively-reached knowledge because this is all that we can have certainty of. The identification of knowledge with certainty is intuitive today as it was when held by Plato, but it conflicts with another strong intuition, that we can claim ‘a right to be sure’ about things in the world (Ayer’s expression).[20] This has led philosophers since Plato to move to a fallibilist conception of knowledge, in which the possibility of error admits, but the criteria for knowledge are stringent.

In Fogelin’s analysis we have looser criteria for knowledge which are closer to the tripartite account than that of Nozick. Fogelin’s theory may cohere with our mental experience of inductive justification being contextually sensitive, that is, varying with the addition of new information, but what is important is coherence with our experience of knowing. As Plato emphasised, a belief must be stable in order for us to have confidence in its truth. But because right belief is just as useful, we should not hesitate to call right belief knowledge when we can be sufficiently sure of its truth. Unlike the conditional theory, the instability revision of the tripartite account does not provide this sufficiency because it is possible to misplace our justification, while a belief that tracks the truth will always result in knowledge.[21]

Overall, the tripartite account is flawed because it is too ambitious, which is probably result of its historical link with infallibilists such as Plato. Whilst scepticism is a viable alternative, it is neither as intuitively plausible nor as readily utilisable as a knowledge based on the conditional theory. The proxy knowledge it yields must be distanced from knowledge in the traditional sense because it is fallible, and consequently it is not fitting to call it true knowledge. But it is stronger than the fragile ‘knowledge’ of Fogelin, which amounts to little more than belief because the ‘knowledge’ it gives does not assure lasting stability. Because the conditional theory is much more demanding it gives us the right to be sure, but in line with human fallibility, admits the possibility of error. In conclusion, the conditional theory, although imperfect, is the most plausible response available to the Gettier problems. This is such that the tripartite account remains necessary at a general level, but we must reject it insofar as it lacks the more specific form of justification -the truth tracking criteria- because it too is necessary and completes the joint sufficiency of the criteria. 


1 See Plato, Theaetetus, 201c-210d & Meno, 98a: “For true opinions, as long as they remain, are a fine thing and all they do is good, but they are not willing to remain long, and they escape from a man’s mind, and are not worth much until one ties them down by [giving] an account of the reason why. … That is why knowledge is prized higher than correct opinion, and knowledge differs from correct opinion in being tied down.”

2 Dancy, 1985, p. 23

3 Williams, 2001, p. 16

4 Plato, Meno, 97a – 98c

5 Williams, 2001, p. 52

6 Dancy, 1985, p. 25. See also Gettier (1963).

7 Nozick, 1981, p. 173

8 Williams, 1999, p. 142

9 See Swain, 1974.

10 Dancy, 1985, pp. 29-31

11 See Goldman, 1967.

12 Dancy, 1885, pp. 33-4

13 Ibid, pp. 37-8

14 Nozick, 1981, p. 178

15 Williams, 2001, pp. 21-5

16 Ibid, pp. 51-2

17 Fogelin, 1994, p. 94

18 Williams, 2001, pp. 54-5

19 Ibid, p. 55

20 Ibid, p. 17

21 And here by ‘truth’ I mean that which currently appears true by reliable inductive methods.


Dancy, Jonathan, (1985), Introduction To Contemporary Epistemology, Oxford, Blackwell Publishers Ltd.

Fogelin, Robert, (1994), Pyrrhonian Reflection On Knowledge and Justification, Oxford, Oxford University Press

Gettier, Edmund, (1963), ‘Is Justified True Belief Knowledge?’, Analysis, 23, pp. 121-3

Goldman, Alvin, (1967), ‘A Causal Theory of Knowing’, Journal of Philosophy, 64, pp. 355-72

Nozick, Robert, (1981), Philosophical Explanations, New York, Oxford University Press

Plato, Meno, In Five Dialogues, Grube, G. M. A. (trans.), (1981), Indianapolis, Hackett Publishing Company Inc., pp. 85-6

Plato, Theaetetus, Jowett, Benjamin, (trans.) (1871)

Swain, M., (1974), ‘Epistemic Defeasibility’, American Philosophical Quarterly, 11, pp. 15-25

Williams, Michael, (1999), ‘Fogelin’s Neo-Pyrrhonism’, International Journal of Philosophical Studies, 7, (2), pp. 141–158

Williams, Michael, (2001), Problems of Knowledge, Oxford, Oxford University Press

On the Definition of ‘Knowledge’


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s