«Ascribing Knowledge in Context: Some Objections to the Contextualist’s Solution to Skepticism MICHAEL HANNON T history of skepticism is extensive ...»
Aporia vol. 17 no. 1—2007
Ascribing Knowledge in Context:
Some Objections to the Contextualist’s
Solution to Skepticism
T history of skepticism is extensive and complex. The issue has
changed shape numerous times, thus making it difficult to combat a
general skeptical problem. Contemporarily, the dilemma is structured in
the form of the skeptical hypothesis (SH), and it is this formulation that is the focus of this paper.
The core element of the skeptical hypothesis is the possibility of a delusion (D) that is irreconcilable with some ordinary empirical proposition (O) that one allegedly knows. The delusion may be that “I am dreaming” (Descartes), “I am a brain in a vat” (Putnam), or “zebras are actually cleverly painted mules” (Dretske). Equally, the empirical propositions that I claim ordinarily to know might include “I am standing,” “I have hands,” and “zebras are not actually cleverly painted mules.”
The skeptical hypothesis (SH) can be expressed in three parts:
(1) It is not the case that I know that I am not deluded.
(2) If I do not know that I am not deluded, then I do not know some ordinary empirical proposition.
Michael Hannon recently finished his undergraduate studies at York University where he majored in criminology and philosophy. He was admitted to the University of London (King’s College) for philosophy but has decided to defer his acceptance for one year in order to study the history of Western philosophy. For his graduate studies, he plans on researching both moral education and moral intuition through the works of Wittgenstein.
56 MICHAEL HANNON (3) Therefore, it is not the case that I know some ordinary empirical proposition.
However, a significant fact is missing from this formulation, namely:
(4) I know some ordinary empirical proposition (e.g., that I have hands).
Individually, both SH and (4) seem plausible. On the one hand, however unlikely it may seem to suppose that I am deluded, it also seems true that I do not know that I am not deluded; indeed, how could I know such a thing? On the other hand, if I propose that I have hands, or if I propose any other ordinary empirical assertion, it seems persuasively true.
Nevertheless when taken together, these claims result in a logical contradiction; that is, it is not the case that I know O—given (1) and (2)—and yet I know O. Something has to give.
There are three possible ways to reject SH. One can deny the first premise, maintaining instead that one does in fact know that one is not deluded. This position was famously—perhaps infamously—defended by G.E. Moore in his “Proof of an External World.” One can also deny the second premise by arguing that people can know both O and D simultaneously. Fred Dretske and Robert Nozick have taken this position by denying the validity of the epistemic principle of deductive closure. Finally, one can allow both (1) and (3) in contexts where SH has been raised, while nevertheless allowing for the denial of (3) in ordinary conversational contexts.
Keith DeRose, Stewart Cohen, and Ludwig Wittgenstein are the principal defenders of this position.
Contextualism holds that skepticism is insoluble and that once SH has been raised, any attempt to refute the skeptic on her own ground is destined to fail. For this reason, the contextualist is willing to concede both the first and second premises to the skeptic since considering the premises implies that SH is necessarily in play. However, the contextualist also maintains that skepticism does not necessarily conflict with our ordinary claims of possessing knowledge. It is wrongly assumed that in order to refute the skeptic one must show that the skeptical possibility (D) does not obtain.
The contextualist would argue that ascriptions of knowledge are contextsensitive, and that the truth-values of utterances involving the word “know” (and its cognates) depend on standards that are contextually
ASCRIBING KNOWLEDGE IN CONTEXT 57determined. Skeptical arguments only succeed because they exploit the conversational context in which words that have epistemic significance are uttered. However, in ordinary conversational contexts where there is no possibility of skeptical error, it is perfectly appropriate to ascribe knowledge to subjects who utter an O. Therefore, we can in ordinary contexts be said to know O.
In this paper, I intend to criticize the contextualist solution to the problem of skepticism. The contextualist would have us believe that skepticism does not necessarily conflict with our claims of ordinary knowledge possession and that knowledge is possible given what the skeptic says. This is because ascriptions of knowledge are ostensibly context sensitive, and their truth-values depend on contextually determined standards. However, although contextualism provides very persuasive arguments against skepticism, I will demonstrate that this position is vulnerable to a number of objections. I maintain that if contextualism is to be considered a viable theory, these objections must be resolved.
This paper is divided into three sections. In the first, I examine two notable alternatives to the skeptical hypothesis. Specifically, I outline both the Moorean denial of the first premise and the Dretskean denial of the second premise. I demonstrate that both positions are flawed and thus ineffective in disarming the skeptic. In the second part, I critically analyze the anti-skeptical position offered by the contextualists. This analysis consists of providing a general outline of contextualism, as well as its relevance to the problem of skepticism. In the third part, I levy some objections to the contextualist treatment of skepticism.
Moorean Denial of the First Premise
Moore has reacted to SH by arguing that, despite the initial plausibility of (1), he is significantly more certain of (4) (“Here is a hand,” says Moore, with a characteristic wave). However plausible the premises of SH may be, it is more reasonable to maintain that we do in fact know many things; thus, SH lacks the impetus to topple our knowledge of many ordinary empirical facts. When reflecting on Descartes’ dreaming hypothesis, Moore willingly concedes the second premise, stating, “I agree with the part of the argument which asserts that if I do not know that I am not dreaming, it follows that I do not know that I am not standing up.” Moore then proclaims, 58 MICHAEL HANNON however, that he does know that he is standing up (4), and concludes that the skeptical alternative must therefore be false given (2). Since the paradox arises from maintaining both SH and (4), Moore is content with rejecting SH in favor of (4) because it has persuasive, intuitive plausibility.
However, this solution is not very satisfying for several reasons. In On Certainty, Wittgenstein suggests that Moore’s argument fails because his claim to know O invites the question of how he came to know O, thus dragging him back into the skeptical debate. By failing to account for how he came to know O, Moore also fails to demonstrate how D is false. Thus, Wittgenstein remarks that “Moore’s view really comes down to this: the concept ‘know’ is analogous to the concept ‘believe’” (5). Moore resorts to mere picking and choosing without a sufficient grounding for his decision.
Simply to argue that it is adequate to rely on one’s intuition of O rather than D fails to recognize that D itself has considerable intuitive pull. In seeking a solution to skepticism we should seek to explain how we fell into this trap in the first place. Since each premise seems intuitively plausible when taken individually, Moore indeed seems to resort to arbitrary preference. For these and other reasons, many philosophers have rejected the Moorean anti-skeptical response.
Dretskean Denial of the Second Premise
In “Epistemic Operators,” Dretske denies the second premise of SH (131–44). As stated, (2) holds that if it is not the case that one knows that one is not deluded, then it follows that one does not know some ordinary empirical claim. For instance, if I do not know that zebras are not actually cleverly painted mules, then I do not know that those animals are zebras. This premise relies on the epistemic principle of deductive
closure (DC). In logic, this principle can be formulated as follows:
(1) A knows that P
As regards SH, DC holds that if I do not know that not-D, then it follows that I do not know O. The principle of deductive closure thus illustrates that knowledge is closed under logical implication.
Dretske feels obligated to concede the first premise to the skeptics.
Unlike Moore, Dretske maintains that one does not know that one is not deluded. However, he does not admit that not knowing that not-D necessarily entails not knowing O. This is because he rejects DC, maintaining instead that DC is only semi-penetrating, and thus does not necessarily hold in every instance. Dretske believes that by giving up DC we can defeat the skeptic because SH hinges on deductive entailment.
The denial of the closure principle is an unpopular anti-skeptical position in contemporary debates. This position has been rejected by Stine, DeRose, Cohen, and many others, each of whom suggests that knowledge should remain closed under logical implication. The principal reason for sustaining DC is that by denying closure Dretske licenses an abominable conjunction—meaning it is possible to know both that one sees a zebra (O) while simultaneously maintaining that one does not know that the zebra is not actually a cleverly painted mule (D). Although this conclusion is clearly counterintuitive, it nonetheless follows if one rejects DC. Thus, most anti-skeptics advocate retaining DC or (2), refuting SH by some other method.
General Outline of Contextualism
Two weeks ago my sister went to the doctor for a routine checkup.
Our family physician, Dr. Shan, was measuring my sister’s height and remarked that she was “quite tall.” Yesterday, my sister decided to become a model, so she contacted a local modeling agency to set up an appointment. When she arrived, the recruiter took one look at my sister and said, “Too short, next!” Something funny is going on here. My sister did not shrink from the time of her doctor’s appointment to the time of her appointment with the modeling recruiter (she remained five feet seven inches), nor was she wearing different shoes or using any other means of changing her height.
Yet, she was nevertheless tall when at the doctor’s office and short while at the modeling agency. How can this discrepancy be explained?
60 MICHAEL HANNON Contextualism maintains that the inconsistency in the above scenario is apparent and does not amount to a genuine contradiction. This example can be explained by examining the environment in which the words “tall” and “short” were uttered. By paying attention to the role of context, one can come to understand how the meaning of a word may change when used in different circumstances. In the case above, the standards of height are relative to the context in which the claim was uttered, and the truth-value of a proposition is shaped by a particular context. The standards in play with regard to height at the doctor’s office were different from the standards in play at the modeling agency, where the relative standards of height were more restrictive.
So how does contextualism relate to the skeptical hypothesis? As mentioned above, SH seems sound. However, SH directly contradicts that we do allegedly have ordinary knowledge of many things (4), such as having hands or knowing that zebras are not cleverly painted mules. In order to resolve this contradiction, the contextualist maintains that our knowledge of O can have different truth-values in different contexts, since different contexts call for different standards. For instance, in the case of my sister’s height, the doctor (S) claimed to know that my sister was tall (O), while the modeling agent (S) asserted that she was not tall (not -O). The propositions “S knows O” and “S does not know O” were thus shown to not logically contradict. Both propositions are correct because of the relative truthvalue of knowledge ascriptions, which are shown to vary cross-contextually.
In a similar vein, contextualism holds that I do not logically contradict myself when asserting, “I do not know that not-D,” and also assert, “I do know that O,” so long as the context in which these statements are uttered prescribes different standards of knowledge. Such standards are more restrictive in the case of the former, while more liberal standards are in effect in the latter. Thus, the context of attribution allows for both claims to be true when uttered in the appropriate context.
Let us briefly consider another example in order to clarify how standards of knowledge can change in different contexts. Imagine that I have a roommate named Smith. In all the years I have known Smith, he has owned the same two pairs of shoes: his running shoes and his dress shoes.
One day I arrive home at 4:30 PM with a good friend of mine, Jones. Jones and I enter the house and proceed to the kitchen, where we converse for several minutes. After a bit of discussion, Jones asks, “Is Smith home?”
ASCRIBING KNOWLEDGE IN CONTEXT 61Keep in mind that neither Jones nor I have seen Smith since our recent arrival.
Nevertheless, I peer down the hallway and see both pairs of Smith’s shoes on the shoe rack, and thus conclude, “Yes, Smith is home.” In this case, there are relatively relaxed standards of knowledge in effect. However, consider a second example. Imagine that everything in the above scenario remains true. But once Jones and I finish our conversation we decide to go out for dinner. After dinner, Jones and I part company, and I soon return home. Upon my return, I find that Smith is being arrested under suspicion of murder. A long-time enemy of Smith’s had been killed that day, and Smith is the prime suspect. Coincidently, the coroner had concluded that Smith’s enemy was murdered at 4:30 PM.
At Smith’s trial, the prosecutor puts me on the stand and asks, “Did you know that Smith was home at the time of the murder?” I respond, “Yes.