An Invariantist's Treatment of Stake Intuitions
Transcript of An Invariantist's Treatment of Stake Intuitions
Master Thesis Paper
An Invariantist’s Treatment of Stake Intuitions
Copenhagen UniversityPhilosophy Department
Date: March 3, 2014
Author: Jasper F.J. Witjes
Student ID: nhm883
Phone: 0045 52406560
E-mail: [email protected]
Supervisor: Søren Overgaard
Summary:In this paper, I give an invariantist’s treatment of the intuitions concerning knowledge ascriptions in situations with a difference in stakes. The purpose is to show that contextualism does not have an advantage with its treatment of the intuitions provided by DeRose (1992). In my invariantist’s treatment, I focus on the influence of the belief formation for both the subject and the ascriber. The main argument is twofolded. For one thing, the invariantist can explain that it’s more difficult to form the right belief in high stake situations. If a subject forms a false belief, or no belief at all, invariantism can accommodate our intuitions concerning high stake situations. For another, the ascriber may not form the right beliefs about the subject’s knowledge. The invariantist can explain that, in high stake situations, a knowledge ascription may be incorrect, but conversationally appropriate. In this way, the invariantist is able to properly explain our intuitions concerning situations with a difference in stakes.
Number of standard pages: 38
Introduction
In everyday life, we make knowledge ascriptions without thinking about
the meaning of ‘know’ or the conditions for knowledge. We easily follow
our intuitions about what is correct to claim. Keith DeRose (1992) defends
contextualism by showing that it can accommodate these pretheoretic
intuitions.
In this paper I want to defend invariantism by investigating its treatment
of these intuitions. I will argue that a proper treatment of intuitions
doesn’t need to accommodate them. It is more important to provide an
explanation of our intuitions and, if they are incorrect, why this is the case.
I will show how invariantism can explain this. I focus on intuitions
concerning situations with a difference in stakes. My explanation will be
centered around the influence of belief formation on making knowledge
ascriptions. In high stake situations, we make incorrect knowledge
ascriptions, because we’re not able to produce the belief that the subject
knows, with the confidence that is necessary to make this claim. I will
explain why this is the case.
It is not my purpose to show that invariantism is better at explaining the
intuitions than contextualism. I want to defend invariantism, by showing
that it has a proper treatment of the intuitions. This provides us with the
modest argument that contextualism does not have an advantage over
invariantism with its treatment of our pretheoretic intuitions.
First, I will explain what both contextualism and invariantism entail. Then,
I will make clear what our intuitions are concerning knowledge
ascriptions in situations with different stakes. I will argue that an
accommodation of our intuitions is not an advantage in a proper treatment
of intuitions. A theory of knowledge ascription needs to provide a good
explanation of our intuitions. The third chapter provides my invariantist’s
treatment of the intuitions. I rely heavily on the concepts and arguments
from Jennifer Nagel (2008), Kent Bach (2005) and Jessica Brown (2005b).
Chapter 1: Invariantism and Contextualism
I will defend a moderate (that is: non-skeptical), classic version of
invariantism. In this first chapter I will describe what this position in
detail entails and how it relates to the opposed theory of knowledge
ascription: contextualism. First we will need a basic understanding of a
theory of knowledge and what a knowledge ascription entails. I will start
with that, then describe contextualism and finally explore invariantism. At
the end of this chapter, it should be clear what, in particular, classic
moderate invariantism entails and how it relates to contextualism.
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 3
1.1 Propositional knowledge
Before we go into theories of knowledge ascription, we need to have a
basic understanding of a theory of knowledge. In the epistemological
tradition, there is mainly a distinction between three kinds of knowledge:
knowledge-who, knowledge-how and propositional knowledge .1
Knowledge-who concerns our recognition of other people.
Knowledge-how concerns our knowledge of how to do things, like riding a
bike or playing chess. I will leave these kind of knowledge completely
aside in this paper, so when I use the term ‘knowledge’, I mean to refer to
propositional knowledge. Basically, propositional knowledge is
knowledge of facts in the world. For example, someone can know that the
Eiffel Tower is in Paris or that Nelson Mandela died in December 2013.
These facts are described in the proposition that - in its most standard
form - follows after the ‘that’-clause. Propositional knowledge is more
than merely a fact stated in a proposition. We need a particular subject (S)
who knows a proposition (p), so when we talk about propositional
knowledge, it concerns a statement of the kind ‘S knows that p’.
A theory of knowledge needs to explain what the necessary and sufficient
conditions are for S to know that p. Since the ancient Greek philosophers,
for a very long time it has been generally accepted that knowledge should
be understood as a justified true belief (JTB). This theory entails that the
conditions for knowledge are that:
1) p needs to be true,
2) S needs to belief that p and
3) S needs to have a justification for his belief that p.
According to this theory, whenever these conditions are satisfied, it is
correct that S knows that p. However, Edmund Gettier (1963) showed that
these three conditions are not satisfactory for a definition of knowledge,
because they ‘do not constitute a sufficient condition for the truth of the
proposition that S knows that P’ (Gettier 1963, p. 192. Italics in the
original). Gettier supports his claim with two hypothetical cases that
describe a subject who has JTB, but of whom we cannot say that he knows.
Since Gettier’s paper, there have been numerous attempts to solve what
came to be known as ‘the Gettier Problem’ with the purpose of defining
necessary and sufficient conditions for knowledge. The main line of
response was to propose an extra condition in addition to JTB that would
establish necessary and sufficient conditions. This strategy has been
unsuccessful so far, since there has always been a counterexample in
response that wasn’t covered by JTB and the extra condition. Other lines
of response have been to reject JTB as a theory of knowledge altogether.
Jonathan Jenkins Ichikawa and Matthias Steup (2001) mention in this
1 For an extensive introduction in different kinds of knowledge, see Fantl, J. (2012)
‘Knowledge How’ and Ichikawa, J.J. and Steup, M. (2001), ‘The Analysis of Knowledge’.
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 4
respect the causal theory of knowledge that “includes a clause requiring
that the belief that p be caused by the fact that p.” A last resort has been to
conclude that knowledge is unanalyzable. The most important conclusion
that we take from this discussion is that there is no consensus about the
correct theory of knowledge or if JTB should be maintained as a condition
for knowledge. Nevertheless, I will adopt the orthodox view that JTB
provides us with necessary conditions for knowledge, leaving open how
sufficient conditions should be fully defined.
1.2 Knowledge ascription
First and foremost, for a knowledge ascription to be true, the subject
actually needs to know that p. For this to be the case, as I assume, the
subject needs to have JTB. Besides that, an ascriber needs to claim that S
knows that p. Whereas S can know that p without anyone claiming that he
does, an ascriber can only make a knowledge ascription by making the
claim in a knowledge-ascribing sentence. The most common way of
ascribing knowledge is in a conversational context by saying that S knows
that p, but we can also ascribe knowledge by thinking something along the
lines of ‘S knows that p’. It’s not directly the assertion that makes it a
knowledge ascription, but rather the meaning that the ascriber has in
mind with the knowledge-ascribing sentence.
The ascriber can be the same person as the subject. We regularly
self-ascribe knowledge in a conversation, often for practical purposes.
Imagine for example that you want to decide if you should go to the
supermarket, because you’re hungry and you don’t have anything in the
fridge. Your partner questions if the supermarket is open at that time,
since it’s 19:00 on a Saturday. When you claim: “I know that the
supermarket is open. It’s open every day until 21:00”, you use this
self-ascription to convey the message that you can go to the supermarket
without the disappointment of it being closed.
It could also be that you’re not sure if the supermarket is open every day
until 21:00 or perhaps only on weekdays. In that case you might make the
knowledge ascription: “I don’t know if the supermarket is open.”
Knowledge ascriptions include both cases of ascribing and denying
knowledge to a subject, but in the case of a denial, we can also speak of a
knowledge denial to be more specific. For clarity, I will use ‘(making a)
knowledge ascription’ for the general cases where it can turn out to be an
ascription or a denial and ‘ascribing knowledge’ when it concerns a
positive ascription of knowledge or ‘denying knowledge’ when it concerns
a denial of knowledge.
An important element in a theory of knowledge ascription is how the
epistemic standard should be defined. The epistemic standard defines
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 5
whether the evidence is sufficient for the subject to have knowledge. A
low standard implies that relatively little evidence suffices to have
knowledge. A high standard implies that strong evidence is required to
have knowledge. I will focus on two ways of defining the epistemic
standard. One way of defining the epistemic standard is by having it vary
with the context. In that case, in some contexts we apply a low standard
and in others we apply a high standard. Contextualism adopts this
strategy. In the next chapter I will describe what this entails for the view.
The other way is to use a fixed standard that doesn’t vary with the context.
In that case, we need to define how high or low of a standard we want to
apply, regardless of the context in which we use it. Invariantism adopts
this way of defining the epistemic standard. I will discuss this in more
detail when I analyse invariantism. First I will provide a detailed
description of contextualism.
1.3 Contextualism
Keith DeRose is a prominent advocate of contextualism in its most common
understanding, so I will mainly be referring to his work. To describe
contextualism as a theory of knowledge ascription, we should start by
mentioning Peter Unger, who proposed the use of the terms
‘contextualism’ and ‘invariantism’. DeRose adopts Unger’s terminology
when he states the following central claim for contextualism: “It’s a theory
according to which the truth conditions of sentences of the form ‘S knows
that p’ or ‘S does not know that p’ vary in certain ways according to the
context in which they are uttered” (DeRose 1992, p. 914).2
Now, the truth conditions of a knowledge-ascribing sentence depend on
the meaning of that sentence. A common way to explain this is that: “They
[meaningful sentences] are so in virtue of having constituents in the
world, which are brought together in the right way. There are many
complications about the nature of meaning, but at a minimum, this tells us
what the truth conditions associated with a proposition are.” (Glanzberg
2006, online). If we know the meaning of a sentence, a theory of knowledge
ascription can explain how the truth conditions lead to the truth-value
‘true’ or ‘false’.
To understand the meaning of a sentence of the form ‘S knows that p’, we
need to understand its parts and how they are related. In the case of
knowledge-ascribing sentences, the meaning depends on the existence of
both S and the fact stated in p, and on how ‘know’ establishes the relation
between S and the fact stated in p. We can define the existence of S and p,
by their constituents in the world. To define the right way in which S
2 DeRose calls sentences of the form ‘S knows that p’ or ‘S does not know that p’
‘knowledge attributions’, which is the same as ‘knowledge ascriptions’. These terms can
be used interchangingly, although I prefer to use ‘knowledge ascriptions’.
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 6
needs to be related to p to have knowledge, we need to define an
epistemic standard.
According to contextualism, “the standard is determined (somehow) by
the context” (Bach 2005, p. 9). As Bach proceeds: “[...] ‘knows’ does not
express a fixed two-term relation. It expresses either a contextually
variable two-term relation or a fixed three-term relation whose third
term, the operative standard, varies with context. And, as the
contextualist stresses, ‘there is no context independent correct standard’
(Cohen 1999: 59)” (Bach 2005, p. 9). The epistemic standard defines if a
knowledge ascription is true or false: there’s nothing in between. A
justification comes in degrees, but this is not the case for knowledge. To
define if a justification is good enough to provide us with knowledge, we
set an epistemic standard that demarcates the shift between knowledge
or not.
The epistemic standard has an important role in connecting the meaning of
a knowledge ascription to the context in which it is made. What
contextualism therefore entails is that the meaning of ‘S knows that p’ can
vary according to changes in the context. All other things being equal, the
same knowledge-ascribing sentence can be true in one context and false in
another context. This can only be correct, because ‘know’ means
something different in these contexts. In one case it refers to knowledge
defined by a low epistemic standard. In the other case it refers to
knowledge defined by a high epistemic standard. Therefore, when we
describe two cases in which the same subject has the same evidence and
the same true belief, the contextualist can claim that the evidence might
be sufficient for knowledge in one case, but not in the other.
1.3.1 DeRose’s Bank Cases
DeRose constructs two cases at the beginning of his influential article
‘Contextualism and Knowledge Attributions’ (1992) to explain and defend
contextualism. I will introduce his Bank Cases here in a slightly different
form to emphasize the important elements. The point of these two cases is
to show that only a difference in context changes the truth-value of the
knowledge ascription. The description will help us understand which
contextual factors make the difference for the change in knowledge
ascription.
In Bank Case Low, DeRose is driving home with his wife on a Friday
afternoon and they plan to stop at the bank to deposit their paychecks (as
people did in the nineties). They notice that the lines inside are very long
and since, in this case, it’s not especially important to deposit the
paychecks as soon as possible, DeRose suggests to his wife to go home and
deposit them on Saturday morning. His wife says: “Maybe the bank won’t
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 7
be open tomorrow. Lots of banks are closed on Saturdays.” DeRose
replies: “No, I know it will be open. I was just there two weeks ago on
Saturday. It’s open until noon.” (DeRose 1992, p. 913)
In Bank Case High, DeRose and his wife are in the same situation, but there
are two important differences. For one thing, the stakes are high. They
have just written a very large and important check that needs to be paid.
If their paychecks are not deposited into their account before Monday
morning, the important check will bounce, leaving them in a very (as
DeRose also emphasizes) bad situation. If the bank is closed on Saturday,
they cannot deposit their checks in time, because the bank is not open on
Sunday. So one difference with Bank Case Low is that it’s very important
that DeRose is right in claiming that the bank will be open on Saturday.
Another difference is that DeRose’s wife reminds him of these facts and
mentions a relevant alternative: “Banks do change their hours. Do you
know the bank will be open tomorrow?” DeRose remains as confident as he
was before, which I will interpret as that he still has the belief that the
bank will be open, but then replies: “Well, no. I’d better go in and make
sure.” (DeRose 1992, p. 913)
DeRose provides us with the additional information that p is true (the
bank will be open on Saturday) and there is nothing unusual about the
cases that has been left out.
According to DeRose, In Bank Case Low we intuitively would ascribe
knowledge to S, whereas in Bank Case High we intuitively would deny that
S knows. Contextualism can explain that these intuitions are right,3
because ‘S knows that p’ means something different in both cases. In Bank
Case Low, it is correct to say that S knows that the bank will be open on
Saturday, because we would apply a low epistemic standard. His
evidence, that the bank was open when he was just there two weeks ago,
is sufficient for knowledge. In Bank Case High it is false to say that S knows
that the bank will be open on Saturday, because we apply a high epistemic
standard. The same evidence (that he was just there two weeks ago) is not
sufficient in this case.
DeRose adds to say: “Yet I seem to be in no better position to know in Case
A [Bank Case Low] than in Case B [Bank Case High]. It’s quite natural to say
that (3) If I know that the bank will be open on Saturday in Case A, then I
also know that it will be in Case B” (DeRose 1992, p. 914). DeRose adds a
third intuition here, that seems to be incompatible with the first two. This
is not the case though, because in the third intuition he uses ‘know’ two
3 Questions have been raised about these intuitions. In experimental philosophy it has
been claimed that people actually do not have these intuitions (see e.g. DeRose, K.
(2011), ‘Contextualism, contrastivism, and X-Phi surveys’ in Philos Stud 156, p. 81 - 110).
In this paper, we’ll grant the contextualist that we do have these intuitions.
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 8
times with the same low epistemic standard. ‘Know’ means the same here
as in the first intuition, but not the same as in the second intuition.
These three intuitions are not compatible with another claim that looks
quite similar to the third intuition: “(4) If what I say in Case A in claiming to
know that the bank will be open on Saturday is true, then what I say in
Case B in conceding that I don’t know that the bank will be open on
Saturday is false” (DeRose 1992, p. 914). The contextualist can deny that
this fourth claim is correct, because ‘know’ has different meaning from
when it’s being used in Bank Case Low in claiming to know than when it’s
being used in Bank Case High in claiming not to know. The subject is in the
same epistemic position to know in Bank Case Low and Bank Case High, but
the ascriber applies a different epistemic standard when he claims that
the subject knows or not. We will come back to these intuitions in the
second chapter.
We should distinguish between two contextual elements that influence
the epistemic standard. The first element is a difference in stakes. In Bank
Case High the stakes are high. It’s very important for DeRose and his wife
to deposit their paychecks in time. If DeRose is right in claiming that the
bank will be open on Saturday, it’s no problem to go home and deposit the
paychecks on Saturday, but if he’s not, they are in serious trouble. Since it
is so important, we ascribe knowledge according to a high standard. In
Bank Case Low it would not be so problematic if DeRose was wrong in
claiming that he knows, so the stakes are low. We ascribe knowledge in a
low stake situation according to a low epistemic standard. According to
the contextualist, a difference in stakes influences the epistemic standard
we apply in making a knowledge ascription.
The second contextual difference is that his wife mentions the possibility
that the bank changed its hours since the last time DeRose was at the bank
two weeks ago. This implies the alternative to p that the bank will not be
open on Saturday. When DeRose makes his knowledge ascription, he is in
the same epistemic position to know that p as in Bank Case Low; his
evidence doesn’t change and he has a true belief. Even though, he cannot
just disregard the alternative as irrelevant for his knowledge ascription.
DeRose: “Since my wife raised the possibility of the bank changing its
hours in Case B [Bank Case High], I have that possibility in mind when I
utter my sentence.” (DeRose 1992, p. 915). This means that DeRose perhaps
has to be able to rule out the alternative. I say ‘perhaps’, because it may
be that he can’t rule it out with his current evidence and information. But
even if he can’t rule out the alternative to p, his evidence in favour of p
may be good enough to ascribe knowledge that p.
Contextualism entails that the alternative becomes relevant in Bank Case
High, merely because it's mentioned. Relevance is not defined by the
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 9
likelihood of the alternative. In Bank Case Low the alternative is not
mentioned and therefore does not change the epistemic standard. In Bank
Case High, since it’s on the mind of the ascriber, the epistemic standard is
raised, even when it’s a far-fetched alternative.
1.3.2 Ascriber’s context
The contextualist distinguishes between a low context in which there are
low stakes and no mentioned alternatives and a high context in which
there are high stakes and mentioned alternatives (or at least one). The
contextualist refers to the context as the context in which the knowledge
ascription is being made, so this is the ascriber’s context. It is very
important to understand that the contextualist refers to the ascriber’s
context and not to the subject’s context. This may sound confusing at first,
because the subject is in a high stake situation and the alternative is
mentioned to the subject. What makes it even more confusing, is that the
Bank Cases are first-person cases: the subject is self-ascribing knowledge,
so DeRose is subject and ascriber at the same time.
How to understand the relation between the subject and the ascriber
when it comes to the context of a knowledge ascription? We need to make
a distinction between ascriber factors and subject factors in the ascriber’s
context. The ascriber factors determine the epistemic standard that the
ascriber sets for what counts as knowledge, or in DeRose’s words: “They
affect how good an epistemic position the putative knower must be in to count
as knowing” (DeRose 1992, p. 921. Italics in the original). The subject
factors determine whether or not the subject is able to live up to this
standard, or: “They affect how good an epistemic position the putative
knower actually is in” (DeRose 1992, p. 922. Italics in the original).
If we first look at the mentioning of relevant alternatives, we can see that
this is an ascriber factor, because it affects the standard the ascriber sets
for the subject to live up to. If a relevant alternative is mentioned, this
will raise the bar for what the ascriber can count as knowledge. In such a
case, the subject’s evidence needs to be better, to live up to the standard
for a knowledge ascription. We should understand this as follows. Before
the alternative was mentioned, the question was if S knows that the bank
will be open on Saturday, because many banks are closed on Saturdays. S has
a valid reply: “This bank is not closed on Saturdays, because I was there
two weeks ago”. After the alternative is mentioned, the standard has been
changed and the question is if S knows that the bank will be open on
Saturday, because banks do change their hours. S does not have a valid reply
to the addition, because he doesn’t know if this bank changed its hours
since two weeks ago.
The mentioning of relevant alternatives is not a subject factor, because it
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 10
does not affect whether the subject is able to live up to the standard that
is set by the ascriber. Say that DeRose, as the subject, is aware that banks
do change their hours, because his wife mentioned it to him. If the ascriber
doesn’t know that this relevant alternative has been mentioned, the
question still is if S knows that the bank will be open on Saturday, because
many banks are closed on Saturdays. The alternative may be relevant for
the subject, but is not relevant in relation to the knowledge ascription.
The subject will still live up to the standard set by the ascriber, because
the ascriber demands a justification that the subject can live up to.
Remember that it’s not important if the subject is able to self-ascribe
knowledge. Even when a subject doesn’t claim to know, the ascriber can
ascribe knowledge to him. Note also that we are talking about a subject
factor in the ascriber’s context and not in the subject’s context. The
question is if the ascriber is aware of an alternative that has been
mentioned.
Now let’s see if the difference in stakes is an ascriber and/or subject
factor. An important observation should be that, at this point, we are only
talking about the stakes for the subject. We’re not yet discussing the
stakes for the ascriber that may have an affect on the knowledge
ascription. We will come back to this in chapter 3.
According to the contextualist, a difference in stakes is an ascriber factor,
because it affects the standard that the ascriber sets for what counts as
knowledge. When S is in a high stake situation, we apply a more demanding
notion of ‘know’ than when S is in a low stake situation. We do this, because
it’s more important to be correct. The better the evidence, the more
certain we can be that someone knows. In a high stake situation we need to
be absolutely certain, so we only say that someone knows if the evidence
is really good. High stakes raises the epistemic standard.
Finally, the stakes for S are not a subject factor, since it doesn’t say
anything about how good of an epistemic position the subject is in. If S is in
a high stake situation, this doesn’t mean that he is in a better or worse
position to know that p, than when he is in a low stake situation. His
epistemic position remains the same, while the epistemic standard
changes.
An important conclusion then is that both relevant alternatives and stakes
for S are ascriber factors and not subject factors in the ascriber’s context.
The subject factors are those that answer the question if S knows that p,
so it is up to a theory of knowledge to define what the subject factors are.
If we hold on to JTB, then these conditions are (at least) whether the
subject believes that p, whether p is true and whether he has a
justification for p. These are the conditions that define the epistemic
position of the subject.
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 11
1.3.3 Advantages and challenges for contextualism
We’ll conclude this part on contextualism with some advantages and
challenges for the theory. Contextualism is attractive mainly for two
reasons. For one thing, it provides a good response to skepticism. The
skeptic tries to show that we hardly know anything, if anything at all, by
bringing in a skeptical argument when someone claims to know something.
Since there always is a skeptical argument to defy the claim that we know
something, the skeptic claim that we do not know in any case. An
often-used example that has been introduced by G.E. Moore is the
ascription ‘I know that I have hands’. We clearly seem to know that we
have hands, but the common skeptical arguments against it, are that there
could be an evil demon deceiving us or that we may just be a (bodiless)
brain in a vat or that we could be living in the Matrix and a supercomputer
is controlling our minds. If we take these arguments seriously, we need to
accept that we don’t know that we have hands, which is quite absurd. The
contextualist responds that when the skeptic raises his skeptical
argument, he changes the standard for what we mean when we say n we
use ‘knowledge’. In this skeptical context that is created we know indeed
very little or next to nothing, but that doesn’t generalize to all contexts.
Usually we do know a lot, because we can easily meet the regular
standards that apply to ordinary situations. David Lewis described this
way of dealing with skepticism by referring to what he calls a ‘sotto voce
proviso’: “S knows that P iff [if and only if] S’s evidence eliminates every
possibility in which not-P - Psst! - except for those possibilities that we
are properly ignoring” (Lewis 1996, p. 694. Italics in the original). So in
most contexts we are properly ignoring skeptical arguments, simply
because they are are not being raised. Lewis jokingly uses the ‘Psst!’,
because we’re supposed to be silent about possible alternatives. The
moment we mention a sceptical argument, we lose our knowledge of p.
Another - closely related - reason why contextualism is attractive, is that
it supports our intuition that most of our everyday knowledge claims are
actually correct. A theory that would claim that most of our knowledge
ascriptions are wrong, will have a hard time explaining this and gaining
support. A theory of knowledge ascription needs to provide us with
insight in our way of ascribing knowledge. Contextualism does this by
showing that a knowledge ascription depends on context. We have seen in
the Bank Cases that contextualism is compatible with the knowledge
ascription being correct in Bank Case Low (low stakes and no mentioning
of alternatives) and with the knowledge denial being correct in Bank Case
High (high stakes and mentioning of alternatives). This fits particularly
well with our intuitions about knowledge ascriptions.
A common resistance against contextualism is that it’s hard to accept that
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 12
changes in context immediately have an effect on knowledge ascriptions.
So we can imagine a Bank Case where DeRose in the end says: “I knew that
the bank is open on Saturday just before, but now that it became so
important that it will be open, I don’t know anymore”. In this sentence he
first uses ‘knew’ according to a low standard and later ‘know’ according to
a high standard. Can what we mean when we use ‘know’ vary so easily from
one moment to the other? As the critic may reply, a more reasonable
response for DeRose would be to say that he actually didn’t know that p
just before, but that response is not compatible with contextualism.
Another objection against contextualism is that it seems odd from a
subject’s point of view that the ascriber’s context has anything to do with a
claim about what the subject knows. Imagine a case where S knows that p
and self-ascribes knowledge, but another ascriber is aware of a relevant
alternative that has been mentioned to him and therefore claims that S
doesn’t know that p. According to contextualism, both the subject and the
other ascriber are right in their knowledge ascription. The subject will
find it difficult to accept though that someone else is right in denying that
he knows something that he correctly claims to know himself. There is a
tension between two contexts, when they apply to the same subject in the
same situation.
1.4 Invariantism
By now we have a good overview of DeRose’s contextualism. Although
there are some differences in contextualist views, we can take DeRose’s
version as a mainstream standard. Invariantism, on the other hand, is a
view that we will need to flesh out in more detail. In broad lines,
invariantism is the epistemic view that claims that the truth-value of a
knowledge ascription does not vary with the ascriber’s context. Unlike the
contextualist, the invariantist applies a fixed epistemic standard that
cannot change according to the ascriber’s context. This means that
‘knowledge’ and ‘know’ have a fixed meaning.
Stephen Jacobson (2010) explains this by comparing ‘know’ to natural
terms like ‘gold’, ‘water’ or ‘aluminium’. There is a strict meaning for these
terms: they refer to their natural constituent in the world. Nonetheless,
people use these terms more or less loosely across contexts. But this
doesn’t mean that the meaning of these terms also varies across contexts.
For example, we can use the term ‘water’ in more or less precise ways.
When we use murky rainwater to water the plants we can refer to it as
‘water’. When we ask for a glass of water to drink, the rainwater doesn’t
qualify. We are more demanding when it comes to the water we drink; in
many countries even tap water is unacceptable for that purpose. Even
more, when water is being used in a laboratory under extremely tight
tolerances, it needs to be pure H2O. Officially, ‘water’ refers to its
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 13
chemical constitution H2O, but when we speak of ‘water’, we rarely mean
pure H2O.
Usually when we speak of ‘water’, it is clear how tolerant we are with
what we mean. The context defines the use of a term. In a garden, it is clear
that we mean rainwater and at home it is clear that we mean tap water. It’s
even clear that you get served bottled water and not tap water, when you
ask for water in a restaurant. The use of the term is variable across
contexts, but the meaning of the term is fixed. The invariantist claims that
this is the same for ‘knowledge’ or ‘know’. The meaning is fixed by one
epistemic standard that applies to any context. So when we make a
knowledge ascription, the truth-value depends on the subject’s epistemic
position and the epistemic standard. Say that a subject forms a belief
based on his evidence and the belief is true, then the epistemic standard
will define if the evidence suffices for knowledge and if it’s correct to
ascribe knowledge or to deny knowledge to the subject.
Invariantism is divided over two main issues that I will discuss more
extensively in the remainder of this part of the chapter. The first
disagreement is about the definition of the epistemic standard. This result
in the divide between skeptical invariantism and moderate invariantism.
The second disagreement is about how strict the invariantist should be in
claiming that the context has no influence on the truth-value of a
knowledge ascription. This results in the divide between classic
invariantism and subject-sensitive invariantism. I will discuss these
differences, so that we eventually clearly understand what classic
moderate invariantism entails and why I choose to defend this view.
1.4.1 Moderate and skeptical invariantism
One aspect that divides different versions of invariantism is the question
how moderate or skeptical we need to be in ascribing knowledge. This
depends on how the fixed epistemic standard is defined. We can explain
this by using DeRose’s Bank Cases as an example. Remember that the
contextualist says that a knowledge ascription is correct in the low
context and a knowledge denial is correct in the high context, because we
apply a different epistemic standard. The invariantist, on the other hand,
has two options. Either a knowledge ascription is correct in both cases or
a knowledge denial is correct in both cases. Since the subject’s epistemic
position is the same in both cases, it is not possible for the invariantist to
make a shift from ascription to denial or vice versa between the cases. The
fixed epistemic standard defines whether we should ascribe or deny
knowledge to the subject.
Skeptical invariantism entails that we need to apply a high epistemic
standard when we speak of knowledge. In the Bank Cases, according to a
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 14
high epistemic standard, we should deny that the subject knows that the
bank will be open on Saturday. His evidence is not sufficient, because he
can’t rule out the alternative that the bank would be closed on Saturday.
The subject doesn’t know if the bank changed its hours since two weeks
ago. It is not likely that the bank did, but since we’re applying a high
epistemic standard, the alternative is sufficiently relevant. Notice that the
alternative is relevant in both Bank Case Low and High, even though it is
not mentioned in Bank Case Low. According to invariantism, if a relevant
alternative exists, it needs to be taken into account, whether it’s
mentioned or not.
Moderate invariantism entails that we need to apply a low epistemic
standard when we speak of knowledge. According to a low epistemic
standard, the evidence is good enough to ascribe knowledge to the
subject in the Bank Cases. This means that the alternative to p that the
bank could be closed on Saturday, is not relevant for the moderate
invariantist. It is so unlikely that the bank changed its opening hours since
two weeks ago, that we can properly ignore the alternative and claim that
S knows that p. Of course this knowledge ascription is only true, because
the subject also believes that p and p is true. The fact that the alternative
is mentioned in Bank Case High, doesn’t make it relevant. In both cases it
can be properly ignored.
Jennifer Nagel and Kent Bach, who will have a prominent role in the last
chapter, are both moderate invariantists. As Bach admits, their position
leaves open how to define a ‘meetably low’ epistemic standard. He4
writes: “[...] I will assume that much of what passes for knowledge really
is, including perceptual knowledge, such as that you are sitting and that
there are lines of print in front of you, and knowledge about simple facts,
e.g. that California is a state and that chickens lay eggs” (Bach 2005, p. 25).
These are quite obvious cases of knowledge that the skeptical invariantist
also might agree on. The Bank Cases are therefore a better example to
show the difference between skeptical and moderate invariantism. The
moderate invariantist does ascribe knowledge to S in both Bank Cases,
whereas the skeptical invariantist denies that S knows in both Bank Cases.
For the purpose of this paper, I will not be more specific about the
difference in epistemic standard between skeptical and moderate
invariantism. Moderate invariantism just accepts a more loose justification
for knowledge, whereas skeptical invariantism is more strict about the
justification. I will leave open how loose or strict a justification needs to
be according to these views.
I will follow Nagel and Bach in adopting a moderate invariantist’s view.
One reason for this, is that in everyday situations people quite easily claim
4 Bach cites DeRose who defined moderate invariantism as a theory that keeps the
epistemic standard ‘meetably low’ (Bach 2005, p. 25).
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 15
to know things. Everyday knowledge ascriptions are more common than
knowledge ascriptions in situations with special concerns and I believe
that we shouldn’t claim that most of our knowledge ascriptions are
actually false. When we try to explain our use of knowledge ascriptions,
first and foremost, we need to look at the knowledge ascriptions that we
actually make and provide a good explanation for them. My second reason
for adopting moderate invariantism is that it has the burden of explaining
why our intuitions about knowledge ascriptions in situations with high
stakes are wrong. I believe we can explain this particularly well by
looking at the process of belief formation that is involved. We will get to
this in my treatment of the stake intuitions in chapter 3.
1.4.2 Classic and subject-sensitive invariantism
The second distinction we can make is between classic invariantism and
subject-sensitive invariantism (SSI). Classic invariantism is most radical in
that the context has no influence on the truth-value of a knowledge
ascription. SSI accepts that contextual factors do have an influence, but
only the subject’s context and not the ascriber’s context. Classic
invariantism rejects this idea and is therefore most radically opposed to
contextualism. It may be helpful to visualize schematically the difference
between classic invariantism and contextualism :5
Figure 1.1
Figure 1.1 shows that the ascriber makes a knowledge ascription based on
the truth of the proposition, the justification that follows from the
5 SSI will not be displayed in the scheme, because my focus is on contextualism and
classic invariantism. A brief description of SSI will follow shortly.
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 16
evidence, the subject’s believe that p and an epistemic standard. The
contextualist claims that the truth-value of the knowledge ascription
depends on a standard that follows from the context. The classic
invariantist claims that the standard follows from the theory he supports.
He either supports skeptical invariantism and applies a high standard or
moderate invariantism and applies a low standard. Anyway, the context
has no influence on the standard and therefore on the truth-value of the
knowledge ascription. Jessica Brown summarizes classic invariantism as
follows: “[The theory] denies that the truth-value of attributions of
knowledge depends on whether, in either the attributor’s or the subject’s
context, the issue is important and error has been raised” (Brown 2005b,
p. 137).
The last theory we need to review is SSI . I will only mention it briefly. It6
differs from classic invariantism, because it assigns a special role for the
subject’s context. Brown explicitly notes that neither the ascriber’s
context nor the subject’s context matters. We have seen that
contextualism claims that the ascriber’s context does matter. SSI on the
other hand claims that the subject’s context has an influence on the
truth-value of a knowledge ascription. The view accepts that “knowledge
depends on a larger range of factors than is traditionally thought” (Brown
and Gerken 2012, p. 2). So it combines the invariantist’s claim that the
truth-value of knowledge ascriptions does not depend on the ascriber’s
context with the contextualist claim that it does depend on contextual
factors, albeit only the subject’s context.
1.4.3 The Airport Case
The differences in the views have important consequences for the
truth-value of knowledge ascriptions. To understand the consequences
for SSI, contextualism and classic invariantism, I’ll introduce an example
that John Hawthorne (2003) uses to support SSI. He describes what we
will call the ‘Airport Case’, originally introduced by Stewart Cohen (1999).
This case is particularly useful, because - unlike in the Bank Case - we
have several persons who act like the ascriber and the subject in the case.
Mary and John are at L.A. airport to take a flight to New York. They need to
know if the flight has a layover in Chicago, because they have to meet a
very important business contact there, so this is a high stake situation for
them. They overhear someone ask Smith, who is also flying to New York, if
he knows whether the flight stops in Chicago. Smith looks at his flight
itinerary and responds: “Yes I know, the flight stops in Chicago”. This is a
self-ascription of knowledge. Mary and John discuss the situation. Since
the stakes are high for Mary and John, Mary raises the alternative that the
6 SSI has also been referred to as interest-relative invariantism (IRI) or sensitive
moderate invariantism, but we’ll only use the term ‘SSI’.
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 17
itinerary may not be trustworthy: it may contain a misprint or the
schedule could have been changed. Mary and John then agree that Smith
does not know that the plane will stop in Chicago and they decide to check
further. So Smith self-ascribes knowledge that the flight stops in Chicago,
but Mary and John deny that he knows this. (Cohen 1999, p. 58).
We can distinguish three knowledge ascriptions:
1) Smith self-ascribing knowledge that p,
2) Mary and John denying that Smith knows that p and
3) Mary and John implicitly denying that they themselves know that p.
I will for our present purposes assume that all these knowledge
ascriptions intuitively are correct. We can evaluate for the three theories
of knowledge ascription how they judge the truth-value of the knowledge
ascriptions that are being made.
Contextualism is compatible with all knowledge ascriptions and therefore
our intuitions about knowledge ascriptions. It can accommodate the
difference in these three knowledge ascriptions by explaining that the
different contexts in which these ascriptions have been made ask for
different epistemic standards to be applied. Mary and John apply a high
standard for what ‘know’ means, because of the high stakes and the
mentioned alternative in the ascriber’s context. Smith applies a low
standard for what ‘know’ means, since there are no high stakes or
mentioned alternative involved in his context (as an ascriber). According
to contextualism, they are both right in their own contexts.
SSI entails that only the subject’s context can influence the truth-value of
a knowledge ascription, so concerning 1): Smith (as ascriber) rightly
asserts that he (as the subject) knows, since he (as the subject) is in a low
stake context. Concerning 3), Mary and John correctly deny knowledge to
themselves, because they (as subjects) are in a high stake context. Finally,
concerning 2), Mary and John should ascribe knowledge to Smith (as the
subject), who is in a low stake context, but they incorrectly deny that
Smith knows. SSI entails therefore that one of our intuitions is false.
Invariantism claims that context has no influence on the truth of the7
knowledge ascription and that the standard needs to be meetably low, so
Smith’s knowledge ascription is correct and Mary and John should ascribe
knowledge to Smith and themselves, so their denial in both 2) and 3) is
incorrect. So the invariantist claims that in the latter two cases our
intuitions are false.
Our intuitions are quite important when we value a theory of knowledge
ascription. Contextualism is fully compatible with the intuitive knowledge
7 From now on, I will simply use ‘invariantism’ for what fully entails the position ‘classic
moderate invariantism’.
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 18
ascriptions and denials that follow from the Airport Case. SSI is
compatible with two of them and invariantism is compatible with only one.
Contextualism seems to have an advantage in accommodating intuitions, so
in our defence of invariantism we need to show that this is not an
advantage and therefore doesn’t exclude invariantism as a good candidate
for a theory of knowledge ascription. In the next chapter we will have a
closer look at the intuitions that are involved in making knowledge
ascriptions and how to deal with them.
Chapter 2: Intuitions, knowledge ascriptions and stakes
Contextualism can accommodate the knowledge ascriptions that we
intuitively would say are correct to make in the Bank Cases and the
Airport Case. There are other relevant intuitions though and there is more
to say about them. To properly assess the treatment of the intuitions by
the invariantist, we need to specify the relevant intuitions and be clear
about what it means to accommodate and/or explain them. Our focus will
be on the difference in intuitions that arise from situations with low
stakes as compared to high stakes.
DeRose is quite explicit about the intuitions that concern his Bank Cases
and that support his contextualist view. Therefore, I will take these
intuitions as a starting point. If invariantism has a good treatment of these
intuitions, the contextualist treatment can not be considered an advantage
for the theory. In this chapter I will have a closer look at the intuitions
from DeRose’s Bank Cases. Since these intuitions only apply to that
particular case, I will show that they can be generalized to apply to other
cases with changing stakes. First, I will say more about how intuitions are
used in philosophy and how they should be used.
2.1 Using intuitions in philosophy
Spicer (2006) distinguishes three senses in which we can use the concept
‘intuition’. In the first sense, if we have the intuition that p (where p is for
example one of the four intuitions that we are assessing), it means that we
are “disposed to endorse, affirm or assert [p] in that situation; the second
way is by our being disposed to think that [p] would be a natural thing for
a subject to endorse or assert in that situation; a third way would be by8
our thinking that a subject is correct to endorse or assert [p] in that
situation”. (Spicer 2006, p. 373. Italics in the original.) So the latter two
senses entail that we would expect from other people that they share our
intuition. It’s not just something that we merely endorse ourselves (as in
the first sense), but something we assume others naturally (second
sense) or correctly (third sense) endorse as well.
8 Note that with ‘subject’, Spicer refers to any subject, not - as we have been doing in this
paper - in particular to a subject that features in a knowledge ascription.
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 19
Neither DeRose (1992) nor Nagel (2008) are very explicit about their use
of intuitions. They don’t mention literally that they appeal to the reader’s
intuitions, they hardly use the term ‘intuition’ and whereas DeRose at
least explicitly mentions and numbers them, Nagel does not flesh out the
separate intuitions in any detail. Nonetheless, we can get an idea of their
use of intuitions. First and foremost they appeal to intuitions in the first
sense, because they invite the reader to form their own intuitions about
the Bank Case. DeRose seems to be most clear in using intuitions in the
first sense: the reader is disposed to endorse or affirm the claims about
knowledge ascriptions that he makes. With the reader on his side, the
attraction of contextualism is immediately clear. Nagel is not merely
appealing to the reader’s intuition, but also affirms that other people have
these intuitions (at least most contemporary epistemologists). Since she
refers to the claims about the knowledge ascriptions as ‘natural
inclinations’, it seems that she commits to the second sense of ‘intuition’. In
this paper, we will also adopt ‘intuition’ in the second sense. We assume
that people have the proposed intuitions and that it’s a natural thing to
endorse them. We are also aware though, that our intuitions may be false,
so it is not always correct to endorse our intuitions about knowledge
ascriptions.
2.1.1 Accommodation and explanation of intuitions
A theory of knowledge ascription can accommodate intuitions, explain why
we have intuitions and why an intuition is (in)correct. It may seem that
accommodating intuitions is always preferable to not being able to
accommodate intuitions, even when a theory is able to explain why
particular intuitions are wrong. When a theory of knowledge ascription
can accommodate our intuitions, we don’t need to change our pretheoretic
views. The theory will immediately make sense, because it is in
accordance with the intuitions we already developed. A theory that
disregards our intuitions as incorrect demands that we revise our
pretheoretical framework, which is something we are hesitant to do. We
have to be careful though, not to rely on accommodating intuitions too
easily. Our intuitions are not always correct, so a theory cannot be
justified simply because it accommodates our intuitions. As Spicer (2006)
remarks: “It is usually counted as a point in favour of a theory that it
accommodates our pretheoretic intuitions [...]. It is also usually thought
that a theory has a burden to explain just those intuitions that it fails to
accommodate. It is not clear, though, that either common thought is true.
Both thoughts become less plausible in a case where we have reason to
suspect that the method by which our pretheoretic intuitive judgements
are made is not truth-conducive” (Spicer 2006, p. 371). So although it
speaks in favour of a theory if it does accommodate our intuitions, this
only is an advantage if the intuitions follow from a reliable,
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 20
truth-conducive cognitive process.
Nagel (2007) compares intuitions with visual perception. We do not
actively have to use cognitive effort to produce intuitions, in the same way
that visual perception appears to us with a certain immediacy. Intuitions
are established by a subconscious process. We experience things and
learn to see patterns that we can apply to new experiences without
consciously processing them. ”The [intuitive] judgments flow
spontaneously from the situations that engender them, rather than from
any process of explicit reasoning.” (Gopnik and Schwitzgebel 1998, p. 77.)
We cannot explain our intuitive judgements, because we can’t observe the
cognitive process that leads to them. Comparably, we can’t explain why
we see things; we simply see them. But what we see is not always correct.
A perceptual observation can be in conflict with empirical theory.
Likewise, our intuitions can be in conflict with our theory of knowledge
ascription.
Good examples of our perception being incorrect are optical illusions like
the Müller-Lyer illusion and the Ebbinghaus Illusion (see figures 2.1 and
2.2). In the Müller-Lyer illusion, we see the lines as having a different size,
although they are of the same size. In the Ebbinghaus illusion one circle
seems smaller than the other, but both circles are equal in size. Even
though we know that this is the case, we keep perceiving the two identical
objects as of a different size. Nonetheless, in general we still trust our
perception. The point is similar for intuitions. We can use our intuitions as
a guidance, but we can’t rely on them without a proper explanation. Nagel
confirms: “Many epistemologists take epistemic intuitions to yield prima
facie rather than conclusive evidence about the nature of knowledge or
justification, subject to correction in the light of a systematic theory, or
perhaps subject to outright rejection in cases where there is something
problematic about the conditions in which the relevant intuitions have
arisen.” (Nagel 2007, p. 794. Italics in the original.)
Figure 2.1: Müller-Lyer illussion Figure 2.2: Ebbinghaus illusion
A common way of using intuitions in philosophy is as a source for forming a
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 21
working hypothesis. We shape theories by testing our intuitions,
investigating where they come from and see if they prove to be correct.
Theories of knowledge ascription have in this respect a disadvantage in
comparison to theories about perceptual experience. In the latter case we
can easily gather independent evidence of the facts. In the case of the
illusions, we can measure the lines and the circles to find out what is
correct and then propose a theory that explains why our perception is
wrong. In the case of our epistemic intuitions we can ‘measure’ our
intuitions against a theory of knowledge ascription, but the theory
provides no objective evidence. That is, when two theories disagree on
whether an intuition is correct or not, we have no objective evidence of
what theory provides us with the right answer. Bottom line, a theory of
knowledge ascription may accommodate intuitions or not, but anyway has
to explain why we have the intuitions and why they are correct or not.
2.2 The four stake intuitions
It needs to be clear what intuitions we’re working with. DeRose mentions
the following intuitions that concern his Bank Cases:
1) “[W]hen I claim to know that the bank will be open on Saturday in
Case A [Low], I am saying something true.”
2) “I am saying something true in Case B [High] when I concede that I
don’t know that the bank will be open on Saturday.”
3) “If I know that the bank will be open on Saturday in Case A [Low],
then I also know that it will be in Case B [High].”
4) “If what I say in case A [Low] in claiming to know that the bank will be
open on Saturday is true, then what I say in Case B [High] in conceding
that I don’t know that the bank will be open on Saturday is false.”
(DeRose 1992, p. 914)
The description of DeRose’s Bank Cases and the intuitions that follow from
them have some characteristics that makes them very suitable to explain
contextualism, but they can be quite confusing. First of all, as mentioned
before, DeRose features as the subject and the ascriber of knowledge,
which can make it difficult to understand what role he is referring to. He
also introduces both stakes and the mentioning of alternatives as
contextual factors, whereas I want to focus merely on a difference in
stakes. Finally, he emphasizes that the subject in Bank Case High still
believes that p (‘remaining as confident as he was before’), whereas it may
be more realistic to leave open if he still believes that p. Bach (2005)
agrees on the last point by mentioning: “It seems to me that unless he’s
trying to placate his wife, his belief would have to be shaken at least
somewhat” (Bach 2005, p. 26).
Nagel (2008) rewrote the Bank Cases to the following version that suits
the purpose of this paper better, since I focus on invariantism, a
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 22
difference in stakes and on the process of belief formation:
“It’s late on a Friday afternoon, and raining hard. Lo and her next-door neighbour
Hi are thinking about going to the bank, but wondering whether the trip could be
postponed until tomorrow. Both of them can remember a recent Saturday visit to
the bank, but neither of them has any further information relevant to the question
of whether the bank will be open tomorrow. Nothing much is at stake for Lo - her
bank errand could be done any time in the next week - but for Hi the question has
burning practical importance. Hi knows he must deposit his paycheck before
Monday, or he will default on his mortgage and lose his home. Does Lo know that
the bank is open tomorrow? Does Hi?
If you are like most contemporary epistemologists, you find it easier to ascribe
knowledge that the bank will be open on Saturday to Lo than to Hi, despite the
fact that the two subjects are relying on the same information in trying to settle
that question.” (Nagel 2008, p. 279.)
In this case description, the subject has the same evidence as in DeRose’s
Bank Cases and p is also true, but Nagel deliberately leaves out any
information about the subject’s belief. By leaving open if Lo and Hi believe
that p or not, we are able to explore the possibility that they didn’t (yet)
form a belief or a false belief. Concerning the contextual elements, there’s
only a difference in stakes and no mentioning of alternatives. Additionally,
we gain clarity by the distinction between the ascriber and the subject,
because the ascriber is not the same person as the subject, but the reader
of the case.
We assume that Nagel is right (following the intuitions from DeRose’s
Bank Cases) and the reader/ascriber - like most contemporary
epistemologist - indeed finds it easier to ascribe knowledge to Lo than to
Hi. So we can formulate a definition of the intuitions that follow from
Nagel’s Bank Case based on DeRose’s intuitions:
1) An ascription of knowledge to Lo is correct.
2) A denial of knowledge to Hi is correct.
3) If Lo knows that p, then Hi also knows that p.
4) If it’s correct to ascribe knowledge to Lo, then it must be false to
deny knowledge to Hi.
It needs to be noted that the first two intuitions are pretheoretic
intuitions. This means that they are shaped without any exposure to the
framework of a theory of knowledge. Any person can have these intuitions,
because everybody has an idea of what it means to know something. The
last two intuitions are not pretheoretic, because they assume some
understanding of a theory of knowledge. They entail that we understand
that, to our best information, Lo and Hi are in the same epistemic position
to know that p. The average person that is not trained in epistemology will
not have very established intuitions concerning the last two.
Contextualism relies most on the pretheoretic intuitions, but for
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 23
completeness I will assess all four stake intuitions.
2.1.1 Generalizing Nagel’s Bank Case intuitions
Before we start analysing the stake intuitions, I will show that they can be
generalized. This is important, because I want to show that the
invariantist’s treatment does not only apply to the Bank Cases, but in
general to cases with a difference in stakes. I will show that this is the
case, by comparing them to the intuitions that Finn Spicer (2006) mentions
in his article ‘Epistemic Intuitions and Epistemic Contextualism’.
First of all, Spicer says: “Let O be some ordinary proposition about the
external world that is obvious to normal subjects and which typically we
take ourselves to know (such as the proposition that one has hands), and
let H be a proposition detailing a sceptical hypothesis, one that we
typically take ourselves not to know not to obtain (such as the proposition
that one is a handless brain in a vat being stimulated to have handish
experiences)” (Spicer 2006, p. 367. Italics in the original). Then, he
mentions the following claims about knowledge ascriptions we intuitively
would make:
1) S knows that O
2) S does not know that not-H
And later he adds the following two additional more general intuitions,
where p is any proposition:
3) “If A and B [two subjects] possess the same evidence for p, then A
knows that p if and only if B knows that p. [...]
4) If A [one ascriber] says ‘S knows that p’ and B [another ascriber] says ‘S
does not know that p’, then A contradicts B” (Spicer 2006, p. 372. Italics
removed).
The first intuition concerns a low stake situation for the subject: O is a
proposition that we typically take ourselves to know. Typically, we are in
situations with low stakes. This intuition therefore means that we would
ascribe knowledge to a low stake subject, and that is a general version of
intuition 1) from Nagel’s Bank Case. To compare Spicer’s second intuition
with intuition 2) from Nagel’s Bank Case, we need to understand that, if we
can’t ascribe to S that he knows that not-H, neither can we ascribe to S that
he knows that O, because H is an alternative to O. To put it differently:
Spicer’s description of the third intuition entails that we should claim that
S does not know that he has hands (O), because we can’t exclude the
possibility that he is a handless brain in a vat (H). Since this alternative has
been mentioned, we’re in a high context. So the intuition means that in a
high context a denial of knowledge to S is correct, like Nagel’s second
intuition. Spicer’s descriptions of the third and fourth intuitions are
simply generalized versions of the last two intuitions from Nagel’s Bank
Case, so we can conclude that the four intuitions do not just apply to
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 24
particular cases, but are general intuitions about knowledge ascriptions
in situations with different stakes.
2.2.2 A contextualist treatment of intuitions
Before we move to an invariantist treatment of the stake intuitions, we
will develop a rough understanding of how contextualism accommodates
and explains them based on the theory as explained in chapter 1.
Contextualism accommodates the first three intuitions:
1) An ascription of knowledge to Lo is correct.
2) A denial of knowledge to Hi is correct.
3) If Lo knows that p, then Hi also knows that p.
It can accommodate them, because of the core principle that the
truth-value of a knowledge ascription varies with the ascriber’s context.
Both subjects Lo and Hi are in the same epistemic position to know, but
the context is different. In Lo’s case, there’s a low stake context which
makes us intuitively ascribe knowledge and in Hi’s case there is a high
stake context, which makes us intuitively deny knowledge. The
contextualist’s explanation for this difference in knowledge ascription is
that the ascriber means something different with ‘know’ in each case. The
contextualist cannot accommodate the fourth intuition:
4) Since Lo and Hi possess the same initial information, it’s a
contradiction to claim that Lo knows and Hi doesn’t know.
We see that the same explanation applies. Contextualism entails that it’s
not a contradiction to claim that Lo knows and Hi doesn’t, because we don’t
mean the same thing with ‘know’ in these two cases. Let me clarify the
difference in meaning with a notation that I borrow from Spicer (2006). We
say that Lo knowslow that the bank will be open and that Hi does not
knowhigh that the bank will be open. Although the concepts knowlow and
knowhigh are similar in many respect, they also differ in the respect that
we put the bar for what counts as knowledge higher when we use
knowhigh than in cases where we use knowslow. This is in broad terms the
main explanation for the stake intuitions that contextualism provides.
Since this is the central claim of contextualism, it looks like the theory is
developed with the purpose to accommodate the two first intuitions that
show how we intuitively ascribe knowledge in high and low stake
situations.
The invariantist does not accept the explanation and the contextualist’s
claim that ‘know’ can mean different things in different contexts.
Therefore invariantism needs to present another explanation. The last
chapter will be devoted to an extensive description of how invariantism
accommodates and explains the stake intuitions.
Chapter 3: An invariantist’s treatment of the stake intuitions
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 25
In my invariantist’s treatment of the stake intuitions, we focus on the
process of belief formation. The stake intuitions involve beliefs in two
ways. First of all, there’s the subject forming a belief about p. The subject
needs to believe that p to know that p, so this is one of the
truth-conducive factors of a knowledge ascription. It would be false to
ascribe knowledge to a subject that doesn’t believe that p. Since Nagel’s
Bank Case doesn’t say anything about the beliefs of Lo and Hi, we need to
analyse the belief formation of Lo and Hi, to understand the influence on
the truth-value of the knowledge ascriptions. I will analyse belief
formation according to the concept of ‘need for closure’, which I will
explain in the next part of this chapter. Secondly, the stake intuitions
involve the ascriber’s belief that S knows that p or that S doesn't know
that p. A particular knowledge ascription can be correct, but if the
ascriber doesn’t believe this to be the case, it would not be
conversationally appropriate to claim so. I will explain more extensively
what it means for a knowledge ascription to be conversationally
appropriate later in this chapter, when I analyse the belief formation for
the ascriber. The concept of ‘need for closure’ applies to the belief
formation for both the subject and the ascriber, so I will start with that.
3.1 Need for closure in belief formation
Nagel (2008) adopts a psychological concept called ‘need for closure’. It
proves to be a highly relevant factor when it comes to knowledge
ascriptions. The concept has been introduced by psychologist Arie
Kruglanski. ‘Closure’ refers to the moment where a subject settles on a
belief and when the formation of a belief turns into the possession of a
belief. Kruglanski and Webster (1996): “The need for cognitive closure
refers to individuals’ desire for a firm answer to a question and an
aversion toward ambiguity” (Kruglanski and Webster 1996, p. 264).
We can imagine many reasons why reaching closure is something
desirable, like the benefit of being able to act or decide in time. Imagine
that you need to catch a train and you see one on a track. You don’t know if
that’s is the train you need to catch though. If you don’t settle on your
belief fast, the train may leave before you made a decision to take it or
leave it. This example is a case of a high need for closure. A subject can
have different levels of need for closure, where a high need for closure
means that a subject wants to settle on a belief fast and a low need for
closure means that a subject does not experience pressure in forming his
belief and is willing to spend more time and effort on forming his belief.
Low need for closure arises typically in situations where it’s very
important to form a true belief. So when the stakes are high, a subject is
usually willing to spend more time and cognitive effort on forming a
belief.
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 26
There’s a general correlation between need for closure and the accuracy
of a judgement. When there’s a high need for closure, the formed belief is
normally less accurate and when there’s a low need for closure we can
expect it to be more accurate. As Nagel (2008) describes the difference:
“Kruglanski dubs the compromised thinking under elevated
need-for-closure ‘seizing and freezing’: in this condition we pounce on
information to form a belief and then become resistant to evidence
contrary to the conclusion first attained. By contrast, in low
need-for-closure we delay judgement and explore evidence more
thoroughly (and typically with greater accuracy)” (Nagel 2008, p. 289). A
greater accuracy follows from the amount of time and cognitive effort
spent on forming the belief when the need for closure is low. We can be
more sure that we collect all important evidence and value it properly.
Consequently, the accuracy drops when the need for closure is raised. It
means that we don’t take the time and cognitive effort that is necessary to
reach a correct belief. In that case, we settle on a belief before we
collected all the evidence or we process the information poorly.
When a subject initially experiences a low need for closure, there can be
several factors raising his need for closure. Some factors that Kruglanski
and Webster (1996) mention are: time pressure, peer pressure, difficulty
of forming a belief and being bored or distracted. These factors can raise
the need for closure so that eventually there’s a much lower accuracy
than the subject initially desired. Often the accuracy is even lower when
there’s a raised need for closure than when a subject experiences a
neutral or high need for closure. Imagine a subject in a high stake
situation. He will initially experience a low need for closure and spend
time and effort on collecting alternatives to p. If he wants to claim that he
knows p, he has to rule out these alternatives. If he is in a position to know
that p, he will be able to do this, but only if he takes the time and effort for
it. When his need for closure is raised, he needs to settle on a belief,
before ruling out the alternative with the consequence that he will say
that he doesn’t know that p. If he would have experienced a high need for
closure initially, he would not have been aware of the alternative and
simply settled on the belief that he does know that p.
Interestingly, when there’s a high need for closure (and therefore a lower
accuracy than when there’s a low need for closure), subjects nonetheless
tend to be more convinced that they made the correct judgement. They
will seize on early information and freeze, not taking any new information
into account. Therefore they are not aware of any possible alternatives,
which makes their judgement the only available plausible answer, making
them feel very assured that it must be correct.
I will apply the concept of need for closure to the belief formation that is
involved in a knowledge ascription. This way, I will be able to provide an
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 27
invariantist explanation for the stake intuitions and why invariantism can
accommodate them or not. I will divide my explanation in an analysis of the
belief formation for the subject and the belief formation for the ascriber,
because a knowledge ascription involves both the subject’s belief that p
and the ascriber’s belief that S knows that p. How they form their beliefs
plays a crucial role in the invariantist’s explanation of the stake intuitions.
3.2 Belief formation for the subject of a knowledge ascription
The belief formation for the subject of a knowledge ascription is highly
influenced by the experienced need for closure. This has important
consequences for the accommodation and explanation of the stake
intuitions. We can only ascribe knowledge to a subject that believes that
p, because the subject’s belief is one of the conditions for knowledge. It is
therefore important to define how the subject will form his belief, when
he settles on a belief and what the chance is that he forms a correct belief.
The first stake intuition concerns a low stake subject; an ascription of
knowledge to Lo is correct. We can expect a low stake subject to be in a
neutral or high need for closure. It’s not particularly important for her to
be correct, so she will not spend more time or cognitive effort on
evaluating alternatives. Since Lo has good evidence that the bank will be
open, in a normal situation like this, she will easily form the belief that p.
When we read Nagel’s Bank Cases carefully, we find no information on
whether Lo actually believes that p. When you’re familiar with a theory of
knowledge (like JTB), you might notice that this information is left out, but
the average reader who trusts his intuitions is not aware of this and has no
reason to doubt that S believes and knows that p. So far we can
accommodate and explain the intuition that an ascription of knowledge to
Lo is correct.
The second stake intuition entails that we would intuitively say that a
denial of knowledge to Hi is correct. This stake intuition concerns a high
stake subject. Initially a high stake subject will experience a low need for
closure. This means that he should be willing to spend more time and
cognitive effort on forming his belief than a subject in a low context. It is -
for whatever practical reason - more important for the subject to be
correct in a high stake context. In relation to this, Nagel (2008) mentions
the results from two research papers in psychology; one by Ziva Kunda,
the other by Jennifer Lerner and Philip Tetlock: “In general, high-stakes
subjects think more systematically and less heuristically, relying more on
deliberate and controlled cognition and less on first impressions and
automatic responses” (Nagel 2008, p. 281). A low need for closure results
in more accurate beliefs, but we can expect Hi to be less confident. To
form his belief, he went through ‘a state in which the assessment of
evidence has not yet solidified into belief’ (Nagel 2008, p. 286). The
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 28
assessment of evidence entails coming up with possible alternatives and,
as already mentioned, since Hi is aware of alternatives, this lowers his
confidence in the correctness of his belief.
Hi will refrain from settling on a belief for as long as possible or
necessary, which makes the timeframe in which we cannot yet ascribe
knowledge larger. So one possibility is that, as long as he doesn’t settle on
a belief, he doesn’t have knowledge, and during this process of belief
formation a denial of knowledge to Hi would be correct. In that case, the
second intuition is accommodated. Since in Nagel’s Bank Case, there is no
information about the belief formation for either Lo or Hi, we don’t know if
Hi comes to believe that p.
Another possibility is that, when Hi (with a low need for closure),
eventually settles on a belief and correctly believes that p, he also knows
that p. If that is the case, the invariantist cannot accommodate the second
intuition, because ascribing knowledge to Hi would be correct. Although,
in this case, we cannot accommodate the second intuition, we will be able
to provide an explanation why our intuitive knowledge ascription would
be incorrect, when we analyse the belief formation for the ascriber later.
High stakes only entail a higher accuracy in forming a correct belief when
the subject experiences a low need for closure. When the subject
experiences a raised need for closure - for example because of time or
peer pressure - the accuracy is lower and the chance bigger that Hi
settles on a false belief. Raised need for closure means that he desires to
settle on a belief before assessing all the evidence properly. He is already
aware of alternatives, because he started with a low need for closure in
which he used more cognitive effort to assess alternatives. Due to the
raised need for closure, he might make a wrong assessment and
overestimate the relevance of the alternative. This leads to settling on the
false belief that p is not correct. In this case the second intuition is
accommodated. When the subject doesn’t believe that p, he doesn’t have
knowledge and a denial of knowledge would be correct.
The third stake intuition entails that if Lo knows, then Hi knows as well.
This can be accommodated and explained. The invariantist would say that
this intuition is correct, because Lo and Hi are both in the same epistemic
position. Only the context is different, but that doesn’t influence their
knowledge. ‘Know’ always has a fixed meaning, so two subjects in the same
epistemic position must both have the same knowledge.
This does imply that Hi went through the process of belief formation with
a low need for closure and correctly believes that p. If Hi experiences a
raised need for closure he might settle on a false belief or, with a low
need for closure, he might not yet settle on any belief, while Lo does
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 29
believe that p. This would not be in line with the third stake intuition. If
this is the case, the invariantist will of course explain that the belief
condition for knowledge is not satisfied for Hi, whereas it is for Lo. Then,
Lo and Hi are not in the same epistemic position and in that case, Hi does
noy know that p, whereas Lo knows. This is an explanation for not
accommodating the intuition, without resorting to contextual elements.
The fourth stake intuition also entails that both Lo and Hi are in the same
epistemic position to know, but it focuses more on the ascriber’s role.
When it comes to the influence of the belief formation for the subject, it is
quite like the third intuition. If Hi (like Lo) does know that p, we should
ascribe knowledge to both subjects and the intuition is accommodated. If
something goes wrong in Hi’s belief formation and he falsely doesn’t know
that p, it would be correct to deny knowledge to Hi. We can again explain
why the intuition is not accommodated in this case, because the belief
condition for knowledge is not satisfied for Hi.
3.3 Belief formation for the ascriber of a knowledge ascription
The belief formation for the ascriber gives us insight in the knowledge
ascription that the ascriber is prepared to make. Most importantly, it
might be that a particular knowledge ascription would be correct to
make, but that the ascriber’s belief formation withholds him from doing
so. In such a case it would be incorrect, but conversationally appropriate
for the ascriber to make a false knowledge ascription. I will first explain
more extensively what it means for a knowledge ascription to be
incorrect, but conversationally appropriate. Afterwards, I will apply the
concept in the treatment of the stake intuitions.
Remember Jacobson’s (2010) comparison of ‘know’ with the term ‘water’:
depending on the context, we can speak of water in more or less loose
ways. This doesn’t mean that the meaning of the word ‘water’ changes, but
just that - for the sake of the situation and conversation - we use the word
incorrectly to convey our message. The invariantist claims that the same
applies to our conversational use of ‘know’. Brown (2005) calls this use
the term ‘know’ incorrect, but conversationally appropriate.
She explains that we can only use ‘know’ conversationally appropriate, if
we actually believe that our knowledge ascription is correct. Brown
(2005): “Of course, the conditions for an utterance to be conversationally
appropriate may include epistemic conditions. [...] DeRose claims that S is
warranted in asserting p only if S knows that p” (Brown 2005b, p. 266).
DeRose’s claim is called the ‘knowledge rule for assertion’, which is
commonly endorsed in philosophy. The belief condition is one condition
for knowledge, so the knowledge rule for assertion entails that S is only
warranted in asserting p, if S believes that p. Applied to the ascriber, this
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 30
means that the ascriber is only warranted in claiming that S knows that p,
if he believes that S knows that p. He cannot believe that S knows that p, if
he doesn’t believe that p is true, because that’s a condition for S to know
that p. We can conclude therefore that the ascriber is only warranted in
claiming that S knows that p, if he believes that p himself. Otherwise, he
would violate the knowledge rule of assertion in claiming something he
doesn’t believe to be true.
We can imagine cases in which S knows that p, but the ascriber doesn’t
believe that S does. An ascription of knowledge would be correct, but it
would not be conversationally appropriate, because he would claim
something he believes to be false. An example would be when you say that
your friend knows that there still is one beer in the fridge, whereas you
don’t believe that there’s any beers left. This in not conversationally
appropriate. At best you can say that your friend believes that there’s a
beer in the fridge, because his belief can be false. You would be meaning
to say that he has a false belief. Knowledge can’t be false, so you can only
say that he knows, if you actually believe that he knows or you’re
undermining your own beliefs. In case that there actually is a beer left in
the fridge, your belief was false and an ascription of knowledge would
have been correct. Your denial of knowledge was incorrect, but
conversationally appropriate, because you didn’t believe that p.
The ascriber and the subject may have different beliefs that p, because
they have different evidence. It is also possible that they have the same
evidence, but still have different beliefs. This might be, because S is in a
high stake situation. S will experience a low need for closure, so it’s more
difficult (cognitively more demanding) for him to form the belief that p
and therefore to know that p. The ascriber is aware that it’s more difficult
to know that p. He does not experience a low need for closure himself
though, because he is not in a high stake situation. The difference in need
for closure will result in the subject correctly believing that p, because
the evidence is sufficient, but the ascriber will not believe that p, because
he finds it too difficult to know (that p or that S knows that p). Either the
ascriber doesn’t believe that p or he doesn’t believe that the evidence is
sufficient for S to know that p.
Remember that the high stakes doesn’t mean that S needs more evidence
to know that p. The epistemic standard is fixed. But a high stake situation
can influence the ascriber in such a way that he applies an incorrect,
skeptically high epistemic standard instead of a correct meetably low
epistemic standard. Bach (2005): “What is decisive here is not the
attributor’s lack of belief but her raised threshold for (confidently)
believing. By this I mean that before believing the proposition in question,
at least with the confidence and freedom from doubt necessary for
knowing [...], the attributor demands more evidence than knowledge
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 31
requires” (Back 2005, p. 27).
I will now look again at the stake intuitions in the light of the belief
formation for the ascriber. In Nagel’s Bank Case the ascriber is always in a
low stake situation. It is not important if we are right in claiming that Lo
and Hi have knowledge or not. This entails that we are in a high or neutral
need for closure. We have seen in the Airport Case that the ascriber can
also be in a high stake situation. The consequence is a low or raised need
for closure. My main thesis is about our general, everyday intuitions
concerning subjects with different stakes, so I will focus on an ascriber in
a low stake situation. To emphasize the influence that a difference in
stakes can have, in some cases I will also mention the belief formation for a
high stake ascriber.
The first intuition concerns a low stake subject. When the ascriber is also
in a low stake situation, invariantism simply accommodates the intuition.
Lo knows that p, a low stake ascriber would believe so and therefore
claim that S knows that p. We can explain the process of belief formation
for the ascriber again with the concept of need for closure. A low stake
ascriber experiences a neutral or high need for closure in forming the
belief that Lo knows that p. It means that he will be fast in forming his
belief with a bigger chance of being wrong, but with a high confidence. The
evidence is sufficient for knowledge and for the ascriber to believe that
Lo knows that p, so the first intuition is easily accommodated.
When the ascriber experiences high stakes, the situation is different. The
Airport Case shows that Smith knows that p, but the high stakes ascribers
Mary and John are not in a position to ascribe knowledge to him. In this
case, their knowledge ascription is incorrect, but conversationally
appropriate. They initially experience a low need for closure and will
therefore spend more time and cognitive effort on forming the belief that
Smith knows that p. Mary and John become aware of possible alternatives
(the itinerary may be wrong), which lowers their confidence in believing
that p and therefore claiming that Smith knows that p. Although the
alternative is not relevant, Mary and John demand more evidence than
knowledge requires. They do not believe that p, so they cannot ascribe
knowledge to Smith. The invariantist can explain that, in this exceptional
case, the first stake intuition would be wrong. A high stake ascriber can
make an incorrect, but conversationally appropriate knowledge
ascription to a low stake subject.
The second stake intuition concerns Hi, a high stake subject that does
know that p. Intuitively we would say that a denial of knowledge to Hi is
correct. According to invariantism, Hi does know that p, so a denial is
actually false. We need to explain why our intuition is wrong. In Nagel’s
Bank case, the ascriber is in a low stake situation, so he experiences a
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 32
neutral or high need for closure. This means that he will take less time and
cognitive effort to form a belief with the result that his belief is less
accurate. Special circumstances can more easily lead to a false belief. The
high stakes for the subject have an impact on the ascriber’s belief
formation. As Bach (2005) describes: “In high-standards Airport and Bank
cases, a special interest gives the attributor reservations about the truth
of the proposition in question and, accordingly, raises her bar for
attributing knowledge to someone else. Even so, the subject knows” (Bach
2005, p. 28). So the ascriber’s confidence regarding p changes with the
information about the subject’s high stakes. The ascriber interprets the
high stakes as entailing that it’s more difficult to form the belief that p.
This is indeed the case, because high stakes lead to a low need for closure,
which raises the cognitive effort to form a correct belief. But the ascriber
does not experience a low need for closure, so instead of using more
cognitive effort to reach a correct belief, he misjudges the amount of
evidence than knowledge requires and reaches a false belief. The ascriber
is not confident in his belief that S knows that p and not in a position to
value the evidence correctly.
The situation would be worse if there also was mentioned an alternative.
In DeRose’s Bank Case High the alternative is mentioned that the bank
could have changed its hours. The low stake ascriber is then aware of the
alternative, but doesn’t take the cognitive effort to properly evaluate it.
The chance that the bank changed its hours in the last two weeks is so
small, that we can reasonably ignore it. It is likely that the ascriber
overrates the alternative though, because it seems more likely since it’s
mentioned and in the focus of attention. Mikkel Gerken coined this
phenomenon an epistemic focal bias . A low stake ascriber is in a high or9
neutral need for closure, so any special circumstances, whether practical
(high stakes) or skeptical (a mentioned alternative) will bias his
judgement. In the case of the second stake intuition, it means that the
ascriber loses confidence that Hi knows. Although this is incorrect, a
denial of knowledge to Hi is conversationally appropriate. In any situation
that matches the second stake intuition, the invariantist can claim that
either ascribing knowledge to Hi would be correct or a denial would be
false, but conversationally appropriate.
The third stake intuition is about the subject’s epistemic position. The
ascriber’s situation is therefore irrelevant. Two subjects are in as good of
an epistemic position to know when they have the same initial information
and the belief formation for the ascriber doesn’t affect their epistemic
position. There’s nothing to say about the belief formation for the ascriber
in relation to the third stake intuition.
9 See Gerken, M. (2011), ‘Epistemic Focal Bias’. In Australasian Journal of Philosophy,
Volume 91, Issue 1 (2013), p. 41 - 61.
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 33
Finally we need to define what the belief formation for the ascriber can
mean for the fourth stake intuition. We’ve seen that the intuition is
accommodated by invariantism when both Hi and Lo correctly form the
belief that p and therefore both know that p, because it would be a
contradiction to claim that Lo knows, but Hi doesn’t. Nagel remarks
however, that from the ascriber’s point of view, in general we do find it
easier to ascribe knowledge to Lo than to Hi. Again, we can explain when
the ascriber incorrectly judges that Lo knows, but Hi doesn’t, that he
makes this mistake because it is conversationally appropriate. As a reader
of Nagel’s Bank Case, we are in a low stake situation and therefore
experience a high need for closure. After all, Nagel presents her Bank
Case in the beginning of her article, so as a reader we don’t want to spend
more time than necessary on forming the belief that Lo and Hi know that p
or not. We see that the only difference between Lo and Hi is the level of
stakes. It makes sense to claim that it is more difficult for Hi to know than
for Lo, because in high stake situations, people demand more evidence
than necessary for knowledge to have enough confidence. Hi has a low
need for closure and will therefore spend more time and cognitive effort
on forming his belief. We should though not mistake this claim (that it’s
more difficult for Hi to know) for the claim that Hi doesn’t know. This
mistake is more likely to happen when we, as ascribers, are in a low stake
situation, because it takes some extra consideration to separate these
claims. The high need for closure makes us settle to easily on the false
belief that Hi doesn’t know, instead of claiming that it is only more difficult
for him to know. With a low need for closure, we should be able to see this
difference and settle on the belief that both Lo and Hi know and that,
indeed, it would be a contradiction to claim that one knows but not the
other.
Altogether, the belief formation for the ascriber means a lot for the
invariantist’s treatment of the stake intuitions. In cases when a subject has
knowledge, but the ascriber is not willing to ascribe knowledge to the
subject, we can assume that the ascriber simply doesn’t belief that S
knows that p. It’s a general rule of conversation that you can’t claim that
something is the case, if you don’t believe it to be the case. The ascriber
may form a false belief about the subject’s knowledge, because he
experiences a high/neutral or raised need for closure. It is not likely that
he forms a false belief when he experiences a low need for closure,
because it means that he makes a good assessment before he makes his
judgement. A high/neutral or raised need for closure has the consequence
that the ascriber settles on a belief too fast without a proper assessment
of all the relevant information. This could mean that he gives too much
weight to an alternative that has been mentioned or that he misjudges
what the subject believes. Anyway, when the ascriber doesn’t believe that
S knows that p, he cannot properly claim that S knows that p. It is not
conversationally appropriate to do so. This explains why, according to
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 34
invariantism, an intuition may turn out to be false in situations with special
circumstances like high stakes.
Conclusion
All together, contextualism does not have an advantage as a theory of
knowledge ascription on the grounds of its treatment of the stake
intuitions. A proper treatment of intuitions explains why the theory can or
cannot accommodate the intuitions. An accommodation is not per
definition favourable, because our intuitions can be wrong. If this is the
case, a theory needs to include in its explanation why an intuition is not
correct.
The stake intuitions concern a case in which the evidence is good enough
to meet a low epistemic standard, but not a high epistemic standard. In
such a case, intuitively we would ascribe knowledge in a low stake
situation, but deny knowledge in a high stake situation. Contextualism can
accommodate these pretheoretic stake intuitions. It explains why, by
claiming that the meaning of ‘know’ is variable according to the context. In
a high stake situation we apply a higher epistemic standard than in a low
stake situation. Consequently, the same evidence can lead to an ascription
of knowledge in a low stake situation and a denial of knowledge in a high
stake situation. According to the contextualist, both knowledge
ascriptions are correct.
Invariantism cannot always accommodate both pretheoretic stake
intuitions, because it entails that the epistemic standard is fixed and
‘know’ has the same meaning in every context. The moderate version of
invariantism claims that we have to apply a low epistemic standard, which
means that we should ascribe knowledge in both the low and the high
stakes situation, given that the subject has a true belief that p. My
invariantist treatment of the stake intuitions first notes, based on Nagel’s
(2008) argument, that the subject’s belief that p is not a given and if S
doesn’t believe that p, neither does he know that p and should we claim
that he knows. A high stake subject experiences a low need for closure in
his belief formation, which means that he takes more time and cognitive
effort to form a belief. As long as he doesn’t form a belief, he doesn’t know
yet. If he, during the process, experiences a raised need for closure, most
likely, he will form a false belief. Overall, the chance that the subject
forms a true belief - and therefore that he knows - is smaller when the
stakes are high. When he doesn’t, a denial of knowledge would be correct
and the intuition is accommodated.
When the subject forms a true belief, a denial of knowledge would be
incorrect. We can explain, mainly based on Bach (2005) and Brown (2005b)
why the ascriber would make an incorrect knowledge ascription, because
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 35
this might be conversationally appropriate. The ascriber cannot
coherently claim that S knows that p, if he doesn’t believe this to be the
case. Either, the ascriber doesn’t believe that p or he doesn’t believe that
S knows that p. When S is in a high stake situation, he will settle on a
correct belief, if he experiences a low need for closure during the whole
process of belief formation. The ascriber mistakenly will settle on a false
belief, when he experiences a high need for closure and doesn’t assess the
evidence correctly. He will demand more evidence than knowledge
requires, which will lead to an incorrect, but conversationally
appropriate denial of knowledge.
In this paper, I tried to make clear that invariantism has a proper
treatment of the stake intuitions. It can coherently explain why our
intuitions are correct in some cases, but false in other cases by referring
to process of belief formation for the subject and the ascriber.
References
Bach, K. (2005), ‘The Emperor's New “Knows”’. In Preyer G. & Peter, G.
(eds.), Contextualism in Philosophy: Knowledge, Meaning, and Truth. Oxford
University Press.
Brown, J. (2005a), ‘Adapt Or Die: The Death of Invariantism?’. In The
Philosophical Quarterly, Vol. 55, No. 219 (April 2005), p. 263 - 285.
Brown, J. (2005b), ‘Comparing contextualism and Invariantism on the
Correctness of Contextualist Intuitions’. In Grazer Philosophische Studien
69 (2005), p. 136 - 164.
Brown, J. and Gerken, M. (2012), ‘Knowledge Ascriptions; Their Semantics,
Cognitive Bases, and Social Functions’. In Brown, J. and Gerken, M. (eds.)
(2012), Knowledge Ascriptions. Oxford University Press, p. 1 - 30.
Cohen, S. (1999), ‘Contextualism, Skepticism, and the Structure of
Reasons’. In Philosophical Perspectives 13, Epistemology (1999), p. 57 - 89.
DeRose, K. (1992), ‘Contextualism and Knowledge Attributions’. In
Philosophy and Phenomenological Research Vol. LII, No. 4 (December
1992), p. 913 - 929.
DeRose, K. (2002), ‘Assertion, Knowledge, and Context’. In The
Philosophical Review, Vol 111, No. 2 (April 2002), p. 178.
Fantl, J. (2012) ‘Knowledge How’. In Zalta, E.N. (ed.) The Stanford
Encyclopedia of Philosophy (Winter 2012 Edition), URL =
<http://plato.stanford.edu/archives/win2012/entries/knowledge-how/
>.
Gettier, E. (1963), ‘Is Justified True Belief Knowledge’. In Sosa E. et al.
(eds.) (2008), Epistemology: An Anthology. Blackwell Publishing, p. 192 -
193.
Glanzberg, M. (2006), "Truth". In Zalta, E.N. (ed.) The Stanford
Encyclopedia of Philosophy (Spring 2013 Edition), URL =
<http://plato.stanford.edu/archives/spr2013/entries/truth/>.
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 36
Gopnik, A. and Schwitzgebel E. (1998), ‘Whose Concepts Are They,
Anyway? The Role of Philosophical Intuition in Empirical Psychology’. In
DePaul, M.R. and Ramsey W. (eds.) (1998), Rethinking Intuitions. Rowman
and Littlefield.
Hawthorne, J. (2003), Knowledge and Lotteries. Oxford University Press.
Ichikawa, J.J. and Steup, M. (2001), ‘The Analysis of Knowledge’. In Zalta,
E.N. (ed.) The Stanford Encyclopedia of Philosophy (Fall 2013 Edition), URL
=
<http://plato.stanford.edu/archives/fall2013/entries/knowledge-analy
sis/>.
Jacobson, S. (2010), ‘Contextualism, Skepticism, and Invariantism’. In
Principia 14 (3) (2010), p. 375 - 391.
Lewis, D. (1996), ‘Elusive Knowledge’. In Sosa E. et al. (eds.) (2008),
Epistemology: An Anthology. Blackwell Publishing, p. 691 - 705.
Nagel, J. (2007), ‘Epistemic Intuitions’. In Philosophy Compass 2/6 (2007), p. 792
- 819.
Nagel, J. (2008), ‘Knowledge ascriptions and the psychological consequences of
changing stakes’. In Australasian Journal of Philosophy, 86:2 (2008), p. 279 - 294.
Spicer, F. (2006), ‘Epistemic Intuitions and Epistemic Contextualism’. In
Philosophy and Phenomenological Research Vol LXXII No. 2, (March 2006), p. 366
- 385.
Jasper Witjes - An Invariantist’s Treatment of Stake Intuitions | p. 37