When we discuss the criteria for knowledge, we typically consider ourselves as neutral observers assessing the status of knowledge claims. It is implicit in this assumption that the conditions for knowledge is independent from the context of the discussion, or the standpoint of the attributer of knowledge. Contextualism rejects this claim and argues that context determines the truth conditions for knowledge ascription. This essay will discuss what contextualism is, a thought experiment than motivates contextualism, contextualism’s treatment of scepticism, and arguments against contextualism.

We have many words in our natural language that are context-dependent: “flat”, “high”, “rich”, etc. These words have different semantical meanings under different circumstances. It is worth considering, then, if knowledge is context-dependent as well. If knowledge is context–dependent, then I can correctly be said to know a statement P in some contexts, but not others. Contextualism argues for exactly this conclusion. Contextualists believe that the meaning of the statement “Edison knows P” is dependent on the context in which the claim is made, which leads to the truth conditions of knowledge claims being context-dependent as well. More specifically, since knowledge is typically conceptualized as justified true belief with some qualification, and truth and belief seem to be invariant regardless of context, contextualist claim that ascriptions of justification are context-sensitive. Just as I might be said to be tall in some contexts but not others, my belief might also be said to be justified in some contexts but not others. Contextualism, however, does not pretend to know exactly how standards for justification are determined by context. To give a rough picture, Cohen[1] suggested that the standards are determined by “some function of speaker intentions, listener expectations, presuppositions, salient relations, etc”. The lack of a complete theory, however, should not discourage readers from subscribing to contextualism, as one cannot give a complete theory of how standards of tallness are determined either; one nevertheless find the notion of highness clearly context-dependent.

Consider the following thought experiment: Mary and John are at an airport and want to confirm if they have a stop-over since they have an important meeting coming up. They overhear another person Smith check his itinerary and say that “I know! My itinerary says that there is a stop-over!” Mary and John agree that Smith does not know that there is a stop-over because they have experienced itineraries sometimes being inaccurate and they decide they’d better check with the travel agent directly. Notice two claims that we are inclined to make here. We are inclined to agree with Smith that he knows that there is a stop-over. This seems to be because we normally consider itinerary good-enough evidence for a knowledge claim about flight information. However, Mary and John also seem to be justified in their claim about Smith’s knowledge, as there is a possibility that the itinerary is wrong, which is especially non-negligible given that they have an important meeting that they can’t miss.

Under invariantism, either Mary and John are too strict, or Smith is too forgiving, only one of the knowledge claims can be right. However, contextualists claim that this goes against our intuitions. Suppose Mary and John are too strict in their knowledge ascription, it nevertheless seems that they are prudent to double-check the itinerary. But then they are required to say that “Smith knows that there will be a stop-over, but we better check if he is right or not.” If we were to say that Smith is too forgiving, it would seem to imply that lots of our daily knowledge claims are in fact false, which seems too high a price to pay. The contextualist can say that Mary, John, and Smith are right about their knowledge claims. Because Mary and John’s intentions and suppositions were different when they made their knowledge ascription, their standards for knowledge are higher than those of Smith. As Mary and John’s claim expresses a different meaning than Smith, the two claims are not contradictory to one another.

We have so far motivated contextualism through a thought experiment, but, so far, the theory seems limited to the semantics of knowledge claims. A more epistemically important motivation for contextualism is from its treatment of scepticism. The skeptic argument can be formulated as follows:

1. If I know that I have a hand, then I am in a position to know that scepticism is false

2. I don’t know that scepticism is false

Therefore, it is false that

3. I know that I have a hand

Cohen argues that all three claims seem to be intuitive. They are, however, incompatible with each other. Yet, simply rejecting one of the claims above seems arbitrary, as all three claims have an intuitive appeal. Since we don’t want to concede to scepticism, Cohen argues that a satisfactory solution needs to explain away the skeptic premise’s appeal while not conceding to scepticism. Contextualism seems to be able to do just this. When the context of a discussion concerns whether I have a hand or not, the most salient pieces of information is typically whether I have the corresponding sensory experience that supports my claim about hands. However, when the context is shifted to a discussion about scepticism, the possibility of skeptic scenarios become more salient. Contextualists can claim that what is expressed by “I know I have hands” is, as such, different depending if I am in an everyday discussion about hands, or in a philosophical discussion regarding scepticism. Under this framework, both the second and the third claims can be true under different contexts. The reason why the three statements seem to be inconsistent is because we fail to notice the shift in context. In skeptic discussions, the standards for knowledge are higher than when we are in daily discussions.

Let’s evaluate contextualism’s treatment of scepticism. As Cohen admitted, the above example only applies to what he calls “high-standards skepticism”. It concedes to the skeptic that our evidence is not good enough to warrant knowledge claims in skeptic contexts, but it nevertheless retains the intuitive claim that we have sufficient evidence that we have hands, at least in everyday contexts. It is with global scepticism that contextualism has more trouble with. Global scepticism casts doubt on the reliability of exactly the sensory experience we base our claims about our hands on. Since global scepticism is immune to rejection from any empirical evidence, it seems that the contextualist is at a dead end, as we are unable to provide any sort of evidence against global scepticism no matter what the relevant context is. Cohen argues that justification is impossible for global scepticism only when defeating global scepticism requires evidence. A belief can, however, be rational and therefore justified to a certain point without the need for evidence according to Cohen. I find this claim dubious. It is certainly not helped by the fact that Cohen gave little arguments for this claim.

Why might it be rational for me to believe something even when one has no evidence for it? One possibility is a priori deduction, but this is clearly not the case with scepticism. Another answer might be that it is intuitive that global scepticism is false. Given the absence of contradicting evidence, we might say that a supporting intuition is sufficient for a belief to be rational. After all, lots of our fundamental axioms in logic are taken to be true primarily due to our strong intuitions for them. However, there is a disanalogy between logical axioms and scepticism. It seems inconceivable for logical axioms to be false, but it seems possible for scepticism to be true. In addition, intuition seems to be too relaxed a standard for rationality. Critically, it seems to permit contingent a priori knowledge, as Cohen himself noted. Cohen responded to this problem by proposing that only when we use our empirical evidence in conjunction with the non-evidential rationality is our justification sufficient for knowledge, at least in everyday contexts. However, this proposal doesn’t seem to completely remove the problem. While non-evidential rationality is no longer sufficient to establish contingent a priori knowledge, the a priori intuition still appears to contribute to the justification of one’s epistemic status about a statement that is contingent. The possibility of this still appears highly questionable.

Another answer is based on pragmaticism. In at least some situations, scepticism also holds very little relevance to our practical reasoning. In situations where what is relevant to us is our conscious experience, whether scepticism is true or not would not influence our decision. It then seems that it is rational to live as if I believe that scepticism is false at least in some contexts. The objection is that this seems to confound a belief being pragmatically rational with a belief actually being rational. In some scenarios, believing in certain unwarranted claims might be pragmatically beneficial, these beliefs do not therefore become rational. Consequently, this account of non-evidential rationality also seems unsatisfactory.

Apart for issues regarding contextualism’s treatment of scepticism, is there a more general argument against contextualism? Hawthorne argues that contextualism entails a disconnection between facts about knowledge and facts about assertion and practical reasoning. Suppose I have a lottery ticket and I need the money from winning the lottery to take a trip to Africa. It seems inappropriate to assert that I know I won’t win the lottery ticket, because it is a possibility you will win the lottery. But according to contextualists, someone else can correct say that I know I will not win the lottery at least in some contexts, because the probability of winning the lottery is incredibly low. A belief that I correctly know to have such a high probability of being true is usually considered justified in everyday contexts. Therefore, contextualists seem committed to admitting that it is sometimes inappropriate to assert something that I know, because my epistemic position is not strong enough. On the flip side, if there you have too high a standard for knowledge, such as in the contexts of discussions of scepticism. Then you are committed to saying that I can assert something that I don’t know and yet not be blameworthy. The same applies to practical reasoning. It seems that contextualists must claim that it is sometimes appropriate to base our decisions on premises that we don’t know, and it is sometimes inappropriate to base our decisions on premises that we do know about. More generally, the problem, according to Hawthorne, seems to be that the normative standards for what can be asserted and what can be the basis of practical reasoning are invariant to context, and yet knowledge is context dependent. Contextualists are then forced to disassociate standards for assertion and practical reasoning with standards of knowledge, even when we find the association between them very intuitive.

How can contextualists respond to this objection? I believe the correct move is to point out that the association between assertion, practical reasoning and knowledge is a spurious relationship. There seems to be a certain requirement for the strength of a person’s epistemic position for this person to appropriately assert something or base their practical reasoning on something. This requirement is usually quite close to the requirement for knowledge. Yet, it is not quite the same thing. The difference between standards for practical reasoning and knowledge is perhaps the most obvious. Sometimes, it is normatively permissible if not required to make practical decisions on the basis of information one does not know. Examples are abundant, such as when I decide to go a restaurant because I am “pretty sure” I heard from a friend that it is pretty good, or when a pro-poker player makes a decision about how much to bet on his hand based on his readings of his opponents that he is not quite certain about. Yet these cases of practical reasoning don’t seem to be questionable or inappropriate.

Furthermore, it seems that a contextualist error theory can account for the counter-intuitiveness of the dissociation between standards for knowledge and standards for assertion and practical reasoning. We associate knowledge and assertion, for example, because we typically assess the knowledge claim of an individual in the context of the assertion. My claim about the existence of my hands is appropriate when what is salient is my sensory experience of the existence of my hands. It sounds weird when someone states that I don’t know the existence of my hands and yet the assertion is appropriate because we don’t typically consider my knowledge claims in a separate context as my assertion.

In conclusion, we considered what contextualism is, one of Cohen’s argument for contextualism, contextualism’s treatment of skepticism, and finally one of Hawthorne’s arguments against contextualism. We argued that Cohen’s notion of non-evidential rationality seems to be problematic, and we also questioned the assumptions behind Hawthorne’s argument from assertions and practical reasoning.


Cohen, S. (1999). Contextualism, skepticism, and the structure of reasons. Noûs, 33(s13), 57–89.

Hawthorne, J. (2004). Knowledge and lotteries. Oxford University Press.

[1] Cohen, S. (1999). Contextualism, skepticism, and the structure of reasons. Noûs, 33(s13), 57–89.

This blog contains a collection of satires, notes, and essays on philosophy, economics, etc. I’m a master’s student in Philosophy at Oxford.