On Easy Knowledge, some problems

Cohen 2002 asserts that the following principle is at least pre-theoretically intuitive:

KR A potential knowledge source K can yield knowledge for S, only if S knows K is reliable.

He notes that some, some evidentialists in particular, will want to deny KR. This entails, for example, that one can know that X is red, on the basis of its looking red, without knowing that X’s looking red is a reliable indication of X’s being red. Thus we can have basic knowledge that X is red, knowledge that is basic in the sense that we do not need to support this belief with knowledge of the reliability of its source. Cohen contends that denying KR commits one—via the entailment just mentioned—to a view which makes knowledge implausibly easy to obtain.

Furthermore, from such basic knowledge we can easily gain knowledge of the reliability of the very faculties in question. Even if basic knowledge were that easy to obtain, surely knowledge of the reliability of the sources of basic knowledge is not. This problem for friends of basic knowledge is called the Problem of Easy Knowledge. There are two versions of the problem, one from combining basic knowledge with closure and one from bootstrapping.

Closure Version
Here is an example of too-easy knowledge via closure. My daughter asks me what color the bike is. I glance over, it looks red, and I reply “red” expressing my sincere belief. Then she says “But how do you know it’s not a white bike with red lights shining on it?” Without looking back, I reply, “because it’s a red bike and red bikes can’t also be white.” Surely this is too easy a way to come to know the bike is not white with red lights shining on it and thus that this perception is veridical.

Bootstraping version
But the case for the reliability of one’s basic sources of justification via bootstrapping is only slightly more complex: record several instances of your knowing that something is j when it’s j-looking, note that you’ve made few or no mistakes, and conclude that you are in fact very reliable in such matters (where j ranges over appropriate predicates). Since theories which allow for basic knowledge underwrite such implausibly easy vindications of reliability there must be something wrong with such theories, Cohen concludes.

Cohen’s suggestion: two-tired knoweldge
Cohen suggests that we take seriously Sosa’s two-tiered view of knowledge: animal knowledge—which is a kind of tracking “not too easily would we be deceived”—and reflective knowledge—which requires a consideration of our epistemic position. Since animal knowledge is merely a form of tracking it needn’t be closed under known entailment. This blocks easy knowledge via closure. Also, bootstrapping is blocked by putting restrictions on the way animal knowledge can combine with self-knowledge (Cohen admits the particular restrictions will seem ad hoc).


Markie
Markie 2005 offers an alternative explanation of the unsatisfyingness of my argument to my daughter that the bike is not white and illuminated by red lights: it begs the question against her skeptical doubts. That is, in the assurance I offer her, I employ as a premise the very proposition she is calling into question. This response does seem to relieve some of the pressure ;-) created by Cohen’s illustrative examples, but the abstract problem remains. It leaves it much easier than is intuitive to gain knowledge of the reliability of basic sources of knowledge. Besides, as Cohen points out the problem can arise in cases where the skeptical dialectic is not present.

Markie also objects that the bootstrapping version of the problem is impossible since in the process of learning to distinguish between, say, red and white phenomenal experiences one acquires evidence that one’s faculties are reliable, such as testimonial evidence. However, it seems conceivable that a person could be created with phenomenal concepts already in place, so even if no this-worldy case of bootstrapping could occur—because no one could ever be in the position of lacking any evidence for the reliability of sense perception before forming perceptual beliefs—the case is a possible one and thus spells trouble for analyses of knowledge that allow for basic knowledge in Cohen’s sense.

When considering holistic support, Cohen very briefly mentions the possibility of an appeal to simplicity. The problem is that when he considers the criterion of simplicity he assumes it is merely a pragmatic feature of an explanation. Following Quine he gives simplicity no truly epistemic role and assumes that we prefer simpler theories simply because they are easier to wield. Any of the infinitely many other empirically equivalent hypothesis would do if we were not so resource bounded. Take away what Russell calls our “medical limitations” and there is no reason to prefer simpler theories.

There are, however, many plausible non-pragmatic vindications of simplicity, however. Some are very technical, like the information-theoretic vindication which purports to prove that simpler statements are intrinsically more probable. Some are not very technical like the suggestion that it is simply an apriori principle of rationality that it is more rational to believe simpler theories. Some are historical such as the dilemma that either simplicity is a rational criterion of theory choice or all of our best science is not really rational simply in virtue of the fact that such theories have in fact historically been preferred for their simplicity.

If any vindication of the epistemic role of simplicity is successful, then there are more options on the table than Cohen has stated. In particular, there is a way of stating how one can be justified—and if all goes well know—that skeptical or skeptic-like alternative hypotheses are false and even that one’s basic cognitive faculties are reliable: commons sense hypotheses are simpler than their skeptical alternatives. Cohen does not treat this response to the Easy Knowledge problem, but I think it is the best line of reply. For one kind development of it see Vogel 1990.