Mike Mackus

Matthew Benton
Philosophy 220
December 2nd, 2008

Lewis Leaves Us Right Where We Started

David Lewis offers us a contextualist approach to knowledge and knowledge attributions

that is intuitively familiar with the way in which we converse with people everyday. Generally,

Lewis argues that one’s knowledge can shift depending on the context of the conversation and

situation. More specifically, Lewis states we do this by means of properly ignoring possibilities

that need not be ruled out, cannot be ruled out, or both. After outlining Lewis’s theory I will

direct attention to what appear to be holes in this version of contextualism, specifically problems

within the central rules of how one may properly ignore a possibility, as well as focus on the

more general implications of Lewis’s contextualism.

Lewis begins by stating what he views as the actual crisis of the skeptical problem: if

knowledge, by definition, must be infallible then if S knows that p then he must also not believe

that p in all possible scenarios where p does not obtain. Thus if S cannot eliminate the possibility

of ~p in a certain case and admits to this inability then it appears as though S does not actually

know that p. Such a knowledge that is fallible, where knowledge can remain despite

uneliminated possibilities, is contradictory to the aims of epistemology. Lewis then asks if this

leaves us to decide between the almost equally terrible options of fallibilism and skepticism.

Fallibilism, according to Lewis, is the better option but he states that it’s much better to not have

to choose at all. Thus, he questions the act of epistemology- could the examination of knowledge

force all ascriptions of knowledge to go false? Given this possibility we must then look at what

epistemology requires for knowledge. Taking what Lewis refers to as the ‘ancient idea’ of

justification, we see that the justification required in certain cases is not enough in other cases;
that is, the strength of necessary justification varies according to context. Lewis argues, however,

that justification cannot be a good starting point. Imagine the case of the lottery: one might truly

believe that he is going to lose but, as long as we assume this is a fair lottery, the odds against

him, no matter how high, will never count his belief as knowledge; there is no statistical

threshold that is ever passed that makes the person’s justification in claiming to have a losing

ticket become knowledge. Thus, Lewis moves away from justification noting that it cannot be

the culprit that destroys knowledge.

Without resorting to the necessity of a justificatory argument, Lewis sets up his main

principle: it is the case that S knows that p if and only if p holds in every possibility not

eliminated by S’s evidence. This well-worded principle begins the set-up of defense against the

skeptical argument. Lewis claims that the word ‘every’ acts just as any other quantifier would; it

ranges over a specific domain and does not capture ‘every’ in an absolute sense simply because it

is within an epistemological definition. ‘Every’ has a certain domain that it pertains to and the

scope of the quantification shifts depending on the context and on what is relevant. When a

friend comes up to you and exclaims ‘Why didn’t you come to the party? Everyone was there!’,

you cannot reply ‘False! Everyone in the whole world was not there.’ ‘Every’ ranges over a

particular set of people, probably, in this case, all of you and your friend’s mutual acquaintances.

However, this example illustrates that one cannot merely choose what he wishes to be within the

domain of discourse. Thus Lewis argues that the domain is determined in part by what

possibilities are properly being ignored at the given moment.

Lewis marks three prohibitions outlining the possibilities which cannot be properly

ignored under any circumstance. First, there is the rule of actuality: that which actually obtains

cannot be properly ignored. Lewis argues that the actual state of affairs must always be a relevant

alternative. It follows from the rule that only what is true is known. The rule is externalist in the
sense that S is judged by his success of not ignoring the actual state of affairs rather than being

assessed by his attempt at properly ignoring. Second, there is the rule of belief: that which is

believed to obtain and that which ought to be believed to obtain cannot be ignored. By ‘ought to

believe’ Lewis intends to cover cases where a person is going against what the evidence and

arguments seem to logically lead to. He goes on, though, to reformulate this rule in terms of

degree given that belief is often a matter of degree: the degree of belief must be sufficiently-high;

here, Lewis is forced to allow justification to enter his argument, a problem that we will examine

later. Lastly, there is the rule of resemblance: any possibility that saliently resembles a non-

ignorable possibility also cannot be properly ignored. Lewis argues that this captures Gettier-like

cases insofar that the subject has no grounds to rule out the actual state of affairs given that is

strikingly similar to that which he believes.

Lewis goes on to acknowledge three presumptions necessary given his three prohibitions

on properly ignoring. The rule of reliability states that one presumes the cognitive faculties are

function properly; any possibility where they are not can be properly ignored. While this rule

appears very strong one must remember that it can be beaten out by the rule of actuality or the

rule of actuality and the rule of resemblance working together: Lewis gives an example of a

person living in a world where most people hallucinate and he has only narrowly escaped this

possibility, thus the possibility that he is not hallucinating is too close to actuality and cannot be

properly ignored. The rule of method allows for two uses of non-deductive inference: induction,

where one may take a sample as representative of the whole, and abduction, where one may

suppose that the best explanation of his evidence is the right one. Thus we are allowed to

properly ignore possibilities in which these two methods fail us. Finally, the rule of conservatism

states that if those around us generally ignore certain possibilities (and this is expected of us as

well) then we may also properly ignore these possibilities. This rule helps to capture the nature of
most conversational contexts so that which is tacitly agreed upon not to be a possible scenario

will not be brought into the equation. And as a final note, Lewis adds what he believes to be a

triviality but nonetheless a necessary one: the rule of attention simply states that any possibility

that is being attending to can no longer be properly ignored; it is impossible for one to go from

attending to a certain possibility and then change one’s mind and attempt to ignore it. The rule

makes any possibility that receives attention a relevant one.

The question at the beginning, how can one handle the skeptic without resorting to a

fallible definition of knowledge, disappears. All of Lewis’s rules acting together allow us to

properly ignore a skeptical hypothesis: a possibility where I am a brain in a vat is not one that is

close to actuality, it is not one I ought to believe, it does not have a necessary resemblance to

another plausible possibility. With the skeptic out of mind, I can state that I know I have two

hands, I know my hair is brown and many other types of everyday knowledge we seem to have

everyday. The problem only arises when I give attention to such a skeptical hypothesis and

immediately I no longer seem to have any of my everyday knowledge for it is impossible for me

to recognize any skeptical hypothesis and then immediately ignore it. For Lewis, the context

determines what qualifies as knowledge. If the context is a discussion of Descartes in an

epistemology class then I can no longer rule out that I am not being deceived by an evil demon

and thus no longer know if I even have hands. However, if the context is a conversation with my

mother about my basketball game last week, say, one in which I hurt my hand, it appears that I

must know I have a hand for I am transmitting information about the well-being it.

As outlined by Lewis, contextualism seems to leave us asking the question whether or not

there is any firm, solid knowledge that the skeptic cannot take away from us. Lewis does state

that there likely are necessary axiomatic truths that will always hold, but those are not what I

have in mind. Essentially, Lewis’s contextualism turns knowledge into an indexical term: not
only does knowledge change depending on context but the actual meaning of ‘knowledge’

changes. Knowledge, as the title of Lewis’s paper indicates, becomes elusive. There seems to be

no real, concrete things that fit under the category of knowledge. And, moreover, it appears by

outlining such a contextualist approach we give recognition to the skeptic and his hypotheses. In

trying to circumvent the damage the skeptic can do to our knowledge we recognize him and say,

oh, no, that is something we can ignore. Lewis contends this adding the rule of attention and

stating that until the skeptical notions have left one’s mind or the context of conversation they

must be attended to and all knowledge is destroyed; but once the skeptical hypothesis is out of

mind knowledge miraculously returns again, making it the case that, in fact, I now do know I

have hands. Couldn’t it be the case, however, that one is always in a conversation with a skeptic?

All crazy hypotheses are always on the table and this poor sap cannot escape always being in a

conversation with this loony skeptic. Do we then say that this unlucky man has no knowledge;

doesn’t know he has hands; doesn’t know he’s in the conversation with the skeptic? Well,

according to Lewis, it seems so. I feel as though Lewis has done us a great service in outlining

his theory in detail but it is not one that belongs to epistemology. Lewis has given us elaborate

rules of assertion and this is not demeaning because such was his goal. However, rules of

assertion and requirements for knowledge do not necessarily have to be the same thing.

Furthermore, there are immediate difficulties in Lewis’s rules for properly ignoring. First,

if Lewis allows justification to slip into his rule of belief then he must admit that his theory is

forced to evade the difficulties of traditional accounts of justified-true-belief as knowledge; that

is, we still risk the possibility of falling into an epistemic regress. In the case of contextualism,

though, the regress might look different: if the justification required for holding a belief is a

matter of context then are the rules to be applied in order to determine if a given justification is

sufficiently high? If so, the rule of belief will continue to fall back on itself every time leaving us
in a regress. Even more difficult for Lewis’s contextualism to handle is the difficulties one could

suggest for the rule of actuality: it could be the case that the skeptical hypothesis is the actual

state of affairs; thus, such a possibility cannot be properly ignored. If there is no way to know

that a skeptical hypothesis is not the actual state of affairs it seems unfair for Lewis to claim that

such a possibility can be properly ignored. This problem may be somewhat extinguished by the

presumptive rule of conservatism insofar that one may properly ignore that which others are

expecting him to ignore, which in most conversational contexts would include skeptical

hypotheses. But this highlights the key that there is an equivalence between knowledge and rules

of assertion for the people whom one might be in a conversation with might not be properly

ignoring. Then it seems as though it is proper to ignore that which others are improperly

ignoring. Nevertheless, Lewis’s rule of actuality can be dismissed by the skeptic- we do not

know what the actual state of affairs is! If we are in the matrix… well, then, that possibility

cannot be ruled out and we are back where we started: in a petty fight with skeptic trying to

prove some sort of everyday knowledge exists.