In a recent paper in Mind Dean
Pettit argues that knowing what a term means is not necessary for understanding
the term. He has three examples designed to show this, all cases where we
allegedly have understanding without knowledge of meaning. I don’t find any of
the examples particularly convincing, though I think they are very interesting
examples of a rather different philosophical point.

Here’s
the first example. Travelling in Germany a certain native English speaker who
knows a smattering of German, Mr. Nogot, sees an unfamiliar word,
“Krankenschwester” and asks a local what it means. The local says, “It means nurse.”
This is true, and hr hereby comes to believe it, and hence interprets any
occurrence of “Krankenschwester” in what appears to be a German sentence as
meaning nurse. Pettit suggests that this is enough to understand the
German word “Krankenschwester”, assuming Nogot already possessed the concept
NURSE, and I tend to agree. But he thinks this is not enough to guarantee that
he knows that “Krankenschwester” means nurse. Imagine that the
local always says, “It means nurse” when asked a question in English. In
that case his belief that “Krankenschwester” means nurse is only
accidentally true, and hence does not amount to knowledge. As Pettit notes,
this seems to be a Gettier example, a justified true belief that is not
knowledge. Pettit claims that whatever is lacking from beliefs in Gettier cases
that makes them fail to be knowledge is lacking in this case also.

At
this point Pettit’s argument is no stronger than the argument that Gettier
cases are not cases of knowledge. Quite understandably, Pettit doesn’t think
his argument is vulnerable at just this point, since most everyone thinks that
Gettier cases are not cases of knowledge. But not exactly everyone! I’ve argued
elsewhere that the argument that Gettier
cases are not cases of knowledge is not exactly watertight. Indeed many
theorists accept an ‘argument’ here that is little more than an appeal to raw
intuition. Which is not to say there is no other argument against the claim
that knowledge is justified true belief, but let’s just focus on the appeals to
intuition, which is what it seems Pettit relies upon. There are, at least,
three reasons why we might suspect that such an appeal to intuition won’t do
enough to carry the argument as far as he wants it to go.

Reason One: Not Everyone Shares the
Intuitions

In a fascinating recent paper,
Jonathan Weinburg, Shaun Nichols and Stephen Stich show that epistemic
intuitions differ radically across cultural groups. In particular, many East
Asians, and people from the Indian Subcontinent, tend to think that Gettier
cases are cases of knowledge. In the studies they conducted, about 55%
of East Asians, and 60% of Sub-Continentals, said that a Gettier case was a
case of knowledge, as compared to about 25% of Westerners. The sample sizes are
rather small, but the differences are statistically significant. Given that the
intuition that Gettier cases are not cases of knowledge is culturally bound,
why should we rest much weight on it? One could try appealing to the great
achievements of Western culture, but since this is serious philosophy and not
Fox News such ‘arguments’ won’t get very far. One could try to argue that the
Western intuitions are embedded in a better overall package. The problem is
that this is simply false. Other experiments Weinburg et al run show
that most Western respondents, especially college-educated Western respondents,
tend to hold absurd sceptical positions. For instance, 70% of the Western
respondents said that a person looking at a zebra could not know that it
was a zebra unless he was in a position to distinguish it from a well-disguised
mule! On the other hand, 50% of the Sub-Continental respondents said that
knowledge was possible in this case. I’d think this is pretty good evidence
that the Sub-Continental intuitions are better guides to knowledge than
the intuitions of us Westies, and they think that Gettier cases are
cases of knowledge.

Reason Two: We All Make Mistakes
Sometimes

Ignore all that, and pretend that somehow
the relevant intuitions do support the idea that Gettier cases are not cases of
knowledge. Why should we think that this implies that Gettier cases are not
cases of knowledge? After all, we all make mistakes, and perhaps sometimes we
all make them together. For instance, there’s lots of empirical evidence that
most people have horribly confused intuitions about whether particular
decisions under uncertainty are rational. Here we say that some decisions are
not rational even if most everyone has the intuition that they are. Why do we
say this? Because accepting that the decisions in question are rational would
commit us to denying some principles that we want to hold onto, such as the transitivity
of preferability, or, more contentiously, the sure-thing principle. Could we
find similar reasons to abandon the intuition that Gettier cases are not cases
of knowledge? Well, the principle that understanding requires knowledge seems
just as secure to me as the sure-thing principle! As an argument for the theory that understanding requires knowledge this
is pretty weak, but since that theory is really the null hypothesis, it seems
that we should be allowed to appeal to that status in judging arguments. (This
line of thought is similar to the line I ran in the counterexamples paper.)

Reason
Three: The Intuitions are not Constant

Grant that neither of the above arguments works, so as long as well-educated
Westerners intuit that Gettier cases are not cases of knowledge, that’s enough
evidence to say that they aren’t knowledge. That still won’t get Pettit the
result he needs, because it’s not ever so clear that intuition really does support
this. At least in this particular case, this seems to be the most important of
the three reasons.

When
philosophers say that intuitively x
does not know that p, what they
usually mean is that when asking themselves, “Does x know that p?” they find
it more intuitive to say, “No.” But there are other questions, with potentially
inconsistent answers, that may be more important in terms of judging what our
intuitions really are. Consider the following three intuition pumps.

The
mixed room: Thirteen people are in the ballroom.
Six of them are native German speakers. Six of them are monolingual English
speakers. And the thirteenth is Mr. Nogot. How many people in the room know
what “Krankenschwester” means in German? I think the intuitively plausible answer
here is seven, not six.

The
homogenous room: As in the mixed room, except the
six monolingual English speakers leave. Is it now true that everyone in the
room knows what “Krankenschwester” means in German? Again, it seems very
plausible to say yes here.

The
bet: Herr Sieger and Herr Verlierer are placing
bets on all sorts of things while they wile the day away. (They have decided to
keep all conversations in English, just to help us foreigners!) As they see Mr.
Nogot approach, Seiger says to Verlierer, “I bet that he knows what
‘Krankenschwester’ means in German.” and Verlierer accepts the bet. After some
simple research, they discover all the facts about Nogot as described above.
Who do you think would win the bet? I think that it’s very intuitive that
Sieger wins, or at least should if the games are being fairly played.

We have three methods for testing
intuitions that all point towards Nogot knowing
that “Krankenschwester” means nurse.
But why think that these methods have more evidential force than the simple
method of asking ourselves whether Nogot knows that “Krankenschwester” means nurse? Because in other cases where we
agree on what the answers should be, after reflection, methods like the three
listed here get the answer right and
the simple method gets the answer wrong.
Let’s take a case familiar from Grice. Grice attributes to Hart the view that
in order for it to be true that Nocare drove home carefully, it must not only
be the case that Nocare should have receptive to possible dangers and disposed
to avoid them, but that his method of doing so must be reasonable. So if Nocare stopped at every driveway to check whether
a dog was about to run out, which he did, that would not be careful in Hart’s language, because it
is unreasonable.

Grice
accepts that it would be odd to say that Nocare was careful in this case.
Indeed, it may even be odd to think this, or even intuit it. But that doesn’t
mean that intuition unreservedly says that Nocare was not careful, for when we
apply the above methods we get the correct result that, intuitively, he was
careful. (Very careful, as it turns out.) Imagine Sieger had bet that Nocare
would drive home carefully. Intuitively, he’d win again. Or imagine the
ballroom contains six paradigmatically careful drivers, and six hoons, and
Nocare. Then it contains seven people
who drove home carefully, not six.

I
want to mention one other case here which has some deeper philosophical
importance. (Deeper only because it touches a point in philosophical logic!) Imagine
that a bar serves no whiskey on a particular day, and the whiskey it had on the
shelf was not poisoned. Then it is not exactly intuitively obvious that every
whiskey it sold that day was poisonous. This might be thought to lead some
support to an Aristotelian view of the universal quantifier, where it is only
true that all Fs are Gs if there are in fact Fs. But a version of the homogenous room
test shows that this is the wrong conclusion to draw. Imagine further that
every whiskey the bar ever sold in its existence was poisonous, although as
noted it did not sell whiskey ever day. In that case it seems intuitively true
that every day the bar was open, every whiskey it sold was poisonous. The
proper conclusion to draw is that the intuitive oddness of the belief that
every whiskey it sold on the quiet day was poisonous is driven by some factor
other than the outright falsity of that belief. It’s a hard question just what
the extra factor is. I’m assured by some very trustworthy sources that the
traditional Gricean explanations here are not very plausible. (As if they are
elsewhere.)

So
let’s summarise the dialectical situation. For this example to prove what
Pettit wants, he needs it to be given that Nogot does not know that
“Krankenschwester” means nurse,
because Pettit’s relation to the proposition that “Krankenschwester” means nurse is as a subject’s relation to a
Gettier belief. But the only evidence that this means he doesn’t know the proposition is that we have a
strong intuition that in this case he doesn’t know it. And since (a) the
intuition is not widely shared, including by groups with more reliable
intuitions than ours, (b) the case bears some resemblances to cases where we
all agree intuition goes wrong, and (c) we only seem to have the intuition if
the intuition probe is designed in a certain way, and that way of designing
intuition probes leads to less reliable intuitions than other probes, it seems
like this isn’t much of an argument after all!

Coming
later (perhaps): what to say about the other examples in the paper.

PS: I wouldn’t have thought of
most of the things to say in this note if not for conversations with various
Syracusans, or now more frequently ex-Syracusans, over the years, so I
don’t want to take too much credit for it
all. Indeed, if all goes to plan much of the above material will find its way
into a paper I’m writing on Pettit’s article with another ex-Syracusan, Adam
Sennet. So some of this should count as co-written. But I don’t think the
standards for claiming work on one’s own weblog that is read by about 10 people
a week should be too high, so I don’t feel too bad about putting it under my
name here!

In a recent paper in Mind Dean
Pettit argues that knowing what a term means is not necessary for understanding
the term. He has three examples designed to show this, all cases where we
allegedly have understanding without knowledge of meaning. I don’t find any of
the examples particularly convincing, though I think they are very interesting
examples of a rather different philosophical point.

Here’s
the first example. Travelling in Germany a certain native English speaker who
knows a smattering of German, Mr. Nogot, sees an unfamiliar word,
“Krankenschwester” and asks a local what it means. The local says, “It means nurse.”
This is true, and hr hereby comes to believe it, and hence interprets any
occurrence of “Krankenschwester” in what appears to be a German sentence as
meaning nurse. Pettit suggests that this is enough to understand the
German word “Krankenschwester”, assuming Nogot already possessed the concept
NURSE, and I tend to agree. But he thinks this is not enough to guarantee that
he knows that “Krankenschwester” means nurse. Imagine that the
local always says, “It means nurse” when asked a question in English. In
that case his belief that “Krankenschwester” means nurse is only
accidentally true, and hence does not amount to knowledge. As Pettit notes,
this seems to be a Gettier example, a justified true belief that is not
knowledge. Pettit claims that whatever is lacking from beliefs in Gettier cases
that makes them fail to be knowledge is lacking in this case also.

At
this point Pettit’s argument is no stronger than the argument that Gettier
cases are not cases of knowledge. Quite understandably, Pettit doesn’t think
his argument is vulnerable at just this point, since most everyone thinks that
Gettier cases are not cases of knowledge. But not exactly everyone! I’ve argued
elsewhere that the argument that Gettier
cases are not cases of knowledge is not exactly watertight. Indeed many
theorists accept an ‘argument’ here that is little more than an appeal to raw
intuition. Which is not to say there is no other argument against the claim
that knowledge is justified true belief, but let’s just focus on the appeals to
intuition, which is what it seems Pettit relies upon. There are, at least,
three reasons why we might suspect that such an appeal to intuition won’t do
enough to carry the argument as far as he wants it to go.

Reason One: Not Everyone Shares the
Intuitions

In a fascinating recent paper,
Jonathan Weinburg, Shaun Nichols and Stephen Stich show that epistemic
intuitions differ radically across cultural groups. In particular, many East
Asians, and people from the Indian Subcontinent, tend to think that Gettier
cases are cases of knowledge. In the studies they conducted, about 55%
of East Asians, and 60% of Sub-Continentals, said that a Gettier case was a
case of knowledge, as compared to about 25% of Westerners. The sample sizes are
rather small, but the differences are statistically significant. Given that the
intuition that Gettier cases are not cases of knowledge is culturally bound,
why should we rest much weight on it? One could try appealing to the great
achievements of Western culture, but since this is serious philosophy and not
Fox News such ‘arguments’ won’t get very far. One could try to argue that the
Western intuitions are embedded in a better overall package. The problem is
that this is simply false. Other experiments Weinburg et al run show
that most Western respondents, especially college-educated Western respondents,
tend to hold absurd sceptical positions. For instance, 70% of the Western
respondents said that a person looking at a zebra could not know that it
was a zebra unless he was in a position to distinguish it from a well-disguised
mule! On the other hand, 50% of the Sub-Continental respondents said that
knowledge was possible in this case. I’d think this is pretty good evidence
that the Sub-Continental intuitions are better guides to knowledge than
the intuitions of us Westies, and they think that Gettier cases are
cases of knowledge.

Reason Two: We All Make Mistakes
Sometimes

Ignore all that, and pretend that somehow
the relevant intuitions do support the idea that Gettier cases are not cases of
knowledge. Why should we think that this implies that Gettier cases are not
cases of knowledge? After all, we all make mistakes, and perhaps sometimes we
all make them together. For instance, there’s lots of empirical evidence that
most people have horribly confused intuitions about whether particular
decisions under uncertainty are rational. Here we say that some decisions are
not rational even if most everyone has the intuition that they are. Why do we
say this? Because accepting that the decisions in question are rational would
commit us to denying some principles that we want to hold onto, such as the transitivity
of preferability, or, more contentiously, the sure-thing principle. Could we
find similar reasons to abandon the intuition that Gettier cases are not cases
of knowledge? Well, the principle that understanding requires knowledge seems
just as secure to me as the sure-thing principle! As an argument for the theory that understanding requires knowledge this
is pretty weak, but since that theory is really the null hypothesis, it seems
that we should be allowed to appeal to that status in judging arguments. (This
line of thought is similar to the line I ran in the counterexamples paper.)

Reason
Three: The Intuitions are not Constant

Grant that neither of the above arguments works, so as long as well-educated
Westerners intuit that Gettier cases are not cases of knowledge, that’s enough
evidence to say that they aren’t knowledge. That still won’t get Pettit the
result he needs, because it’s not ever so clear that intuition really does support
this. At least in this particular case, this seems to be the most important of
the three reasons.

When
philosophers say that intuitively x
does not know that p, what they
usually mean is that when asking themselves, “Does x know that p?” they find
it more intuitive to say, “No.” But there are other questions, with potentially
inconsistent answers, that may be more important in terms of judging what our
intuitions really are. Consider the following three intuition pumps.

The
mixed room: Thirteen people are in the ballroom.
Six of them are native German speakers. Six of them are monolingual English
speakers. And the thirteenth is Mr. Nogot. How many people in the room know
what “Krankenschwester” means in German? I think the intuitively plausible answer
here is seven, not six.

The
homogenous room: As in the mixed room, except the
six monolingual English speakers leave. Is it now true that everyone in the
room knows what “Krankenschwester” means in German? Again, it seems very
plausible to say yes here.

The
bet: Herr Sieger and Herr Verlierer are placing
bets on all sorts of things while they wile the day away. (They have decided to
keep all conversations in English, just to help us foreigners!) As they see Mr.
Nogot approach, Seiger says to Verlierer, “I bet that he knows what
‘Krankenschwester’ means in German.” and Verlierer accepts the bet. After some
simple research, they discover all the facts about Nogot as described above.
Who do you think would win the bet? I think that it’s very intuitive that
Sieger wins, or at least should if the games are being fairly played.

We have three methods for testing
intuitions that all point towards Nogot knowing
that “Krankenschwester” means nurse.
But why think that these methods have more evidential force than the simple
method of asking ourselves whether Nogot knows that “Krankenschwester” means nurse? Because in other cases where we
agree on what the answers should be, after reflection, methods like the three
listed here get the answer right and
the simple method gets the answer wrong.
Let’s take a case familiar from Grice. Grice attributes to Hart the view that
in order for it to be true that Nocare drove home carefully, it must not only
be the case that Nocare should have receptive to possible dangers and disposed
to avoid them, but that his method of doing so must be reasonable. So if Nocare stopped at every driveway to check whether
a dog was about to run out, which he did, that would not be careful in Hart’s language, because it
is unreasonable.

Grice
accepts that it would be odd to say that Nocare was careful in this case.
Indeed, it may even be odd to think this, or even intuit it. But that doesn’t
mean that intuition unreservedly says that Nocare was not careful, for when we
apply the above methods we get the correct result that, intuitively, he was
careful. (Very careful, as it turns out.) Imagine Sieger had bet that Nocare
would drive home carefully. Intuitively, he’d win again. Or imagine the
ballroom contains six paradigmatically careful drivers, and six hoons, and
Nocare. Then it contains seven people
who drove home carefully, not six.

I
want to mention one other case here which has some deeper philosophical
importance. (Deeper only because it touches a point in philosophical logic!) Imagine
that a bar serves no whiskey on a particular day, and the whiskey it had on the
shelf was not poisoned. Then it is not exactly intuitively obvious that every
whiskey it sold that day was poisonous. This might be thought to lead some
support to an Aristotelian view of the universal quantifier, where it is only
true that all Fs are Gs if there are in fact Fs. But a version of the homogenous room
test shows that this is the wrong conclusion to draw. Imagine further that
every whiskey the bar ever sold in its existence was poisonous, although as
noted it did not sell whiskey ever day. In that case it seems intuitively true
that every day the bar was open, every whiskey it sold was poisonous. The
proper conclusion to draw is that the intuitive oddness of the belief that
every whiskey it sold on the quiet day was poisonous is driven by some factor
other than the outright falsity of that belief. It’s a hard question just what
the extra factor is. I’m assured by some very trustworthy sources that the
traditional Gricean explanations here are not very plausible. (As if they are
elsewhere.)

So
let’s summarise the dialectical situation. For this example to prove what
Pettit wants, he needs it to be given that Nogot does not know that
“Krankenschwester” means nurse,
because Pettit’s relation to the proposition that “Krankenschwester” means nurse is as a subject’s relation to a
Gettier belief. But the only evidence that this means he doesn’t know the proposition is that we have a
strong intuition that in this case he doesn’t know it. And since (a) the
intuition is not widely shared, including by groups with more reliable
intuitions than ours, (b) the case bears some resemblances to cases where we
all agree intuition goes wrong, and (c) we only seem to have the intuition if
the intuition probe is designed in a certain way, and that way of designing
intuition probes leads to less reliable intuitions than other probes, it seems
like this isn’t much of an argument after all!

Coming
later (perhaps): what to say about the other examples in the paper.

PS: I wouldn’t have thought of
most of the things to say in this note if not for conversations with various
Syracusans, or now more frequently ex-Syracusans, over the years, so I
don’t want to take too much credit for it
all. Indeed, if all goes to plan much of the above material will find its way
into a paper I’m writing on Pettit’s article with another ex-Syracusan, Adam
Sennet. So some of this should count as co-written. But I don’t think the
standards for claiming work on one’s own weblog that is read by about 10 people
a week should be too high, so I don’t feel too bad about putting it under my
name here!