LaiC (2021) Epistemic Gradualism Versus Epistemic Absolutism

14680114, 2022, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/papq.12356 by Vanderbilt University, Wiley Online Library on [30/12/2022].
See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
EPISTEMIC
GRADUALISM VERSUS
EPISTEMIC
ABSOLUTISM
BY
CHANGSHENG LAI
Abstract: Epistemic absolutism holds that knowledge-that is ungradable, while

epistemic gradualism argues the opposite. This paper purports to remodel the
gradualism/absolutism debate. The current model initiated by Stephen
Hetherington fails to capture the genuine divergence between the two views,
which makes the debate equivocal, and the gradualist side lacks appeal. I propose
that the remodeled debate should focus on whether knowledge-that is a ‘threshold
concept’ or a ‘spectrum concept’. That is, whether there is a threshold
distinguishing knowledge from non-knowledge. The reconstructed model enjoys
significant advantages as it enables the debate to be more balanced and philo-
sophically interesting.
1. Gradualism and absolutism
Knowledge-how is ordinarily taken to be gradable1 – one can know how to

play the piano better than his or her friend does. Similarly, it is widely
accepted that knowledge of a person or a place can admit of different
degrees – Jack knows Julia better than Jim does, but Jim knows London
better than Jack does. That is to say, knowledge-how and many types of
knowledge-wh are conventionally taken to be gradable. So do belief and
justification – two essential concepts that are taken to be necessary for
constituting knowledge-that by many epistemologists. But when it comes
to the gradability of knowledge-that itself, the most discussed sort of
knowledge in contemporary epistemology, things become complicated.
Pacific Philosophical Quarterly 103 (2022) 186–207 DOI: 10.1111/papq.12356
© 2021 University of Southern California and John Wiley & Sons Ltd
186
14680114, 2022, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/papq.12356 by Vanderbilt University, Wiley Online Library on [30/12/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
EPISTEMIC GRADUALISM VERSUS EPISTEMIC ABSOLUTISM 187
Is one’s knowing that p an absolute yes-or-no affair, or a matter of degree?
The orthodox view holds that propositional knowledge is ungradable (refer
to Ryle 1949; Dretske 1981; Kvanvig 2003; Stanley 2005; Pavese 2017; etc.).
That is, knowledge-that is an absolute concept that does not admit of
degrees. A proposition can either be known or not known – it can neither
be known less or more, nor better or worse. Knowing a proposition,
according to Dretske’s often-quoted analogy, ‘is like being pregnant: an all
or nothing affair’ (1981, p. 363). This orthodoxy is named epistemic
absolutism by Hetherington (2001, 2011) who urges us to reject this
orthodoxy and embrace its opposite, namely, epistemic gradualism – a view
that knowledge-that is gradable.
One key task of epistemology is to investigate various conceptual
properties of knowledge-that. The gradability of knowledge-that is an
important and interesting one, especially when many closely related
concepts, such as knowledge-how, knowledge-wh, belief and justification,
are all seen as gradable by the traditional view. Hence, the debate
between gradualism and absolutism is supposed to be significant and
valuable. However, despite Hetherington’s unremitting and ingenious
defence, gradualism does not seem to have adequate appeal to make
epistemologists pay more attention to the debate. Three mutually related
questions are essential for gradualists to answer in this debate: (1) Why
should we call the absolutist orthodox into question? (2) How does
propositional knowledge come in different degrees if gradualism is true?
(3) And most fundamentally, what does it mean to say that propositional
knowledge is gradable or absolute? This paper has two main goals. First,
I will argue that Hetherington’s answers to the three questions are
unsatisfactory. The upshot is Hetherington fails to construct gradualism
appealingly and thus is unable to provide a solid ground for rejecting
absolutism. Second, I will remodel the debate between absolutism and
gradualism. It will be argued that the key divergence between
absolutism and gradualism should be whether knowledge-that is a thresh-
old concept or a spectrum concept, to be more specific, whether there is
a cut-off point (or ‘threshold’) between knowledge and lack thereof.
This paper will proceed as follows. Section 2 will introduce how the
current framework of the debate proposed by Hetherington answers the
three essential questions aforementioned. Section 3 will reveal why those
answers are defective and how the current debate is misled by a false focus.
Section 4 will elaborate on two crucial notions of my proposal – ‘threshold
concept’ and ‘spectrum concept’. I will show how gradualism can be
constructed based on the idea that propositional knowledge is a spectrum
concept rather than a threshold concept, and how this can remodel the
debate between gradualism and absolutism. Section 5 will show advantages
that the new framework of debate enjoys over the original one.
188 PACIFIC PHILOSOPHICAL QUARTERLY
2. The current debate
This section will give a brief sketch of the current debate against absolutism
by introducing how Hetherington’s gradualist proposal answer the three
essential questions mentioned in Section 1.
Why should we doubt absolutism? Hetherington argues that the absolutist
orthodoxy is dogmatic rather than well-grounded. He notes that the primary
reason for seeing knowledge-that as ungradable is a linguistic one: that we
do not use ‘knows’ as a gradable predicate in our ordinary language (refer
to Hetherington 2001, 2011, ch. 5). For example, expressions like ‘A knows
that p better than B does’ are ordinarily deemed infelicitous. In response,
Hetherington provides counterevidence that we do sometimes felicitously
use ‘knows-that’ constructions in a gradable manner. For example, ‘I know
that I feel pain. I know that you do, too. But I know better that I do than
that you do – whereas you know better that you do than that I do!’
(Hetherington 2001, p. 1).2
After undermining the linguistic ground3 for endorsing absolutism,
Hetherington reminds us that knowledge-that is homologous with some
paradigmatic gradable notions. For instance, he argues that knowing a fact
is sometimes equivalent with understanding that fact, which amounts to
understanding that (1) the fact obtains; and (2) how that fact obtains.
Hetherington also cites Edward Craig’s (1990) view that knowing-how is
sometimes indistinguishable from knowing-that. In light of this alleged
homology, knowing-that is concluded to be as gradational as understanding
and knowing-how. However, at that stage, Hetherington did not provide a
strong argument for the analogousness between knowledge and those
gradable notions. This line of argument was further developed in
Hetherington’s (2005, 2011) later works, where Hetherington systematically
argued for an equivalence between knowledge that p and how-knowledge
that p – the latter refers to knowledge of how it is that p. It is argued that,
on the one hand, knowing that ‘p obtains’ is the minimal aspect of knowing
how it is that p; on the other hand, knowing how it is that p requires at least
knowing that p obtains. Hetherington thus advocates a biconditional that he
calls ‘↔H’, to wit, knowing that p both entails and is entailed from knowing
how it is that p. Therefore, ‘knowledge that p is how-knowledge that p’
(Hetherington 2011, p. 177). Given that how-knowledge that p is gradable
in the sense that one can know one or more aspects of how it is that p, so
is knowledge that p. Now we have seen that there are two versions of
gradualism proposed by Hetherington. The difference between the two
versions is also reflected in Hetherington’s answers to the second essential
question that we care: How does propositional knowledge come in
different degrees?
In his early works, Hetherington argues that the quality of knowledge can
be graded in terms of the justificatory strength4 (refer to Hetherington 2001).
The weaker one’s justification5 for p is the worse one knows that p. But how
bad is a piece of knowledge allowed to be? Hetherington’s answer is as
follows: mere true belief is sufficient for constituting the minimal
knowledge (2001, ch. 4). This is unorthodox again,6 as it rejects the standard
justificationist view that justification is necessary for knowledge. Although
he also admits that perhaps it is practically impossible for one to possess this
sort of minimal knowledge, Hetherington still insists that an unjustified true
belief suffices to be recognised as, albeit very poor, knowledge. One alleged
theoretical advantage of this anti-justificationist view is that it can solve the
so-called boundary problem of knowledge, which asks where the boundary
between ignorance and the worst justified knowledge is. Hetherington argues
that it is almost impossible for fallibilists7 to non-arbitrarily locate a cut-off
point between sufficient justification and insufficient justification, insofar as
we hold that knowledge requires justification. However, if justificationism is
discarded, then the problem can be easily resolved: the cut-off point falls at
mere true belief, namely, zero justification.
Hetherington (2011) takes a different approach to answer how
knowledge-that can be graded in his second version of gradualist proposal.
As we have mentioned before, he equates knowing that p with knowing
how it is that p – and the latter, according to Hetherington, is a kind of
ability. In fact, a core thesis for Hetherington’s second version of gradualism
is what he calls knowledge-as-ability hypothesis, which interprets knowledge
as an ability to manifest various accurate representations of p. This ability
can be graded in accordance with how detailed one knows how it is that p.
The more aspects of how it is that p one knows, the better one’s knowledge
that p is.
It is noteworthy that Hetherington’s early gradualism only rejects the idea
that knowledge-that is absolute in the sense that knowledge cannot be better
or worse despite stronger or weaker justification. He still concedes that there
is an absolute threshold (i.e. ‘cut-off point’) for knowledge, which is mere
true belief. In summary, Hetherington admits that he is committed to a type
of local gradualism that consists of two views:
(i) Knowledge is to be absolutely distinguished from whatever is not knowledge (whatever is

external to knowledge). This is because there is an absolute cut-off point between knowing
and not knowing,
(ii) But within the category of knowing, non-absolutism is true. That is because it is possible that
some cases of knowledge that p are better as knowledge that p, or more clearly knowledge that p,
than other cases of knowledge that p. (2001, p. 7)
Hetherington calls view (1) ‘external absolutism’ and view (2) ‘internal
non-absolutism’. Given Hetherington’s anti-justificationist position, it is
not hard to understand why he endorses external absolutism. For

Hetherington, the central divergence between gradualism and absolutism
is not whether there is an absolute threshold for knowledge, but whether
knowledge can be improved beyond the threshold – by stronger justification.
This commitment to external absolutism is also inherited by
Hetherington’s later account of gradualism. As we have noted, this version
of gradualism sees knowledge as an ability. Notwithstanding, Hetherington
still emphasises that this account is compatible with there being a threshold
for having the ability:
In general, abilities are gradational … … This is consistent with there being an absolute cut-off
point, either precise or not, between having the ability and not having it. (So that is not the sense
of knowledge-absolutism I am denying.) (2011, p. 49)
This allows us to outline Hetherington’s answer to the third essential

question: What does it mean to say that propositional knowledge is gradable
or absolute? To say that knowledge-that is gradable is to say that one can
know that p is better or worse in terms of the strength of justification, or
how detailed one knows how it is that p, after one reaches the threshold of
knowing that p. To say that knowledge-that is absolute is to say that one’s
knowledge that p cannot be improved, despite one’s better justification or
more detailed knowledge of how it is that p, after one reaches the threshold
of knowing that p. The key divergence between gradualism and absolutism
only occurs in the ‘internal’ aspect. He, as well as the majority of philoso-
phers, never doubted the existence of knowledge’s threshold due to the
commitment to external absolutism. In what follows, I will argue that this
commitment paints Hetherington’s gradualism into a corner.
3. Defects of the current framework
So far, I have outlined Hetherington’s (two versions of) epistemic gradual-

ism. In summary, his gradualist theories have two significant traits: (1) being
associated with, if not based on, either ‘anti-justificationism’ or ‘↔H’; (2)
being committed to external absolutism. These two features, as I will show
in this section, render Hetherington’s gradualism both too controversial
and too uncontroversial, such that his answers to the three essential
questions mentioned above are defective.
On the one hand, Hetherington’s gradualism is too controversial to be
attractive due to the controversy of anti-justificationism and ↔H.
Hetherington’s commitment to anti-justificationism severely undermines
the appeal of his gradualism, as numerous epistemologists still see justifica-
tion as necessary for knowledge, especially when ‘justification’ is defined
broadly. The idea that true belief suffices to constitute knowledge is accused
of violating the widely shared intuition that knowledge is incompatible with
mere epistemic luck (refer to Greco 2003; Pritchard 2007; Sosa 2007) and the
intuition that knowledge should be a kind of cognitive achievement (refer to
Greco 2003; Pritchard 2007; Sosa 2007; Pritchard 2012). One can luckily
form a true belief via mere guessing or wishful thinking. But a true belief
formed in these ways can hardly be admitted as knowledge because it is just
luckily true without manifesting one’s relevant cognitive competence. A
theory’s being too controversial of course does not imply its being false,
but the unpopularity of anti-justificationism does constitute a dialectical
defect of Hetherington’s debate framework: gradualism itself is already an
unpopular view and thus needs more explanations than absolutism.
Combining gradualism with another (even more) controversial view is thus
unwise for gradualists – it makes gradualism more dialectically disadvan-
taged and the debate more unbalanced. Admittedly, Hetherington’s
gradualism is not logically based on his anti-justificationism; however, the
two views are closely connected to each other. Anti-justificationism
completes Hetherington’s first version of gradualist picture as an important
piece of jigsaw. Recall that Hetherington’s local gradualism insists that there
is a cut-off boundary distinguishing knowledge from non-knowledge.
Consequently, a natural question to be asked is where the boundary of
knowledge is. Given the difficulty in locating a non-arbitrary boundary,
Hetherington appeals to a simple and straightforward solution, namely,
anti-justificationism. Anti-justificationism is motivated by Hetherington’s
commitment to external absolutism and is used to, in turn, answer an
important problem invited by this commitment. Anti-justificationism is of
course not the only (nor the optimal) option for Hetherington, but it does
constitute an unignorable part of his gradualist storytelling as a matter of
fact and thus affect the convincingness of his gradualist proposal as a whole.
Also, Hetherington’s later account of gradualism based on ‘↔H’ is subject
to a vital logical lacuna. Hetherington’s proof of ‘↔H’ relies on an inference
as follows:
1 A minimal aspect of φ entails that φ, and vice versa;

2 Hence, a minimal aspect of φ is equivalent to φ;
3 Thus, if φ is gradable, then the minimal aspect of φ is gradable.
In Hetherington’s original argument, φ refers to ‘how-knowledge that p’, of

which the minimal aspect (and thus equivalence) is ‘knowledge that p’. Given
that how-knowledge that p is gradable, so is knowledge that p.
However, this inference is a non sequitur. ‘Being dual-purpose’
entails ‘being versatile’, as the former is the minimal aspect of the latter.
And hence ‘being versatile’ also entails ‘being (at least) dual-purpose’.
However, it is untenable to thereby conclude that ‘dual-purpose’
is as gradable as ‘versatile’. ‘Versatile’ is a gradable concept, but
‘dual-purpose’ is not. Similarly, ‘having two colours’ is the minimal aspect

of ‘being colourful’, while the latter is gradable and the former is not.
‘Multicultural’, ‘multiracial’, and ‘multilingual’, are all gradable, never-
theless, their minimal aspects ‘bicultural’, ‘biracial’, and ‘bilingual’ are
not. Therefore, the gradability of a concept is not as smoothly transmitta-
ble to its minimal aspect as Hetherington has it. In summary, both
versions of Hetherington’s gradualism are based on problematic grounds
and hence lack sufficient appeal. Call this the dubious ground problem.
This problem renders Hetherington’s answers to the first two essential
questions that his gradualism is supposed to accommodate unconvincing
(recall that the two questions ask why the absolutist orthodoxy should be
doubted and how knowledge-that is graded).
On the other hand, Hetherington’s local gradualist stance is not
controversial enough to make the alleged debate between absolutism and
his gradualism a genuine debate. Many commentators point out that
Hetherington’s commitment to external absolutism obscures the distinction
between gradualism and absolutism. Contra what Hetherington alleges,
many epistemologists who do not take themselves to be gradualists should
already be glad to consent that, at least in a loose sense, there can be better
knowledge in terms of better justification. Hence, Hetherington’s local
gradualism is not as unorthodox as he presents it to be. For example, when
commenting on Hetherington’s gradualist thesis, Carl Ginet writes
There is at least one way in which I can understand this thesis and in which it strikes me as
correct. One’s knowledge that p might be more or less good according as one’s justification
for believing that p is more or less vulnerable to being wrong. (Ginet 2010, pp. 20–21)
Richard Feldman, who is ‘not persuaded by the central arguments’ of

Hetherington (Hetherington 2001), points out that
It is therefore rather uncontroversial that some items of knowledge can be better justified than
others. And if one wants to say that knowledge that involves better justification is better knowl-
edge, then that strikes me as rather benign. I do not see what exactly denying this would imply.
(Feldman 2002)
This remark is echoed by Romy Jaster, who shares the impression that
knowledge can be graded in a loose sense (e.g. in terms of the justificatory
strength), but strictly speaking, ‘knows’ is still not a gradable predicate:
We can say that the property of being tall enough is satisfied to a higher degree the taller the
subject is. And we can say that the property of being sufficiently flat is satisfied to a higher degree
the flatter the subject is. In the same derived sense we can grade knowledge relations. We can say
that they are realized to a higher degree the stronger the required epistemic position is. But we
should keep in mind that this is just a very loose formulation of what is really claimed.
(Jaster 2013, pp. 320–321)
Hetherington also holds that non-minimalists deny that better
justification yields better knowledge, and thus, they would be unable to ac-
commodate the phenomenon of epistemic improvement. This point is called
into question by Brower (2004):
In any case, the core problem at the intuitive level is that epistemologists who reject minimalism
may grant that knowledge can be better or worse from an epistemic perspective … in general a
non-minimalist who accepts degrees of justification, yet sets justificatory standards short of infal-
libility, can easily accept gradualism while rejecting absolutism. (Brower 2004, p. 107)
Therefore, it seems that most of the absolutists grant that the quality of
one’s knowledge can be improved by virtue of better justification – nonethe-
less, this does not imply that the concept of knowledge is ultimately
gradable. A direct upshot of Hetherington’s local gradualism is an equivocal
attitude towards the gradualism/absolutism debate. The idea is knowledge
is gradable in one sense (viz., internal gradualism), a sense in which there
is no genuine disagreement between Hetherington and absolutists. Mean-
while, knowledge is ungradable in another sense (viz. external absolutism),
a sense in which Hetherington and absolutists reach a consensus. The result
is, we can see no genuine debate between absolutism and Hetherington’s
gradualism, as nothing is actually in conflict given the two distinct senses.
Absolutists can agree with Hetherington in both senses, but still refuse to
accept that knowledge-that is an ultimately gradable concept, as it is still
ungradable in the external sense.
Call this the two-sense problem. This problem makes Hetherington’s
answer to the third essential question aforementioned unsatisfactory. The
question, recall, asks what it means to say that propositional knowledge is
gradable or absolute. Hetherington distinguishes the external sense from
the internal sense of answering this question. However, in neither sense,
there is genuine disagreement between absolutists and him.
Hetherington endorses the existence of knowledge’s threshold as an un-
questionable fact – the problem is not whether this threshold exists, but
how we can (even if only approximately) locate that threshold.
Hetherington (2006) admits that it is extremely difficult to locate it precisely
and that our ignorance of the exact location of knowledge’s threshold can
significantly undermine our knowledge that knowledge has a threshold.
Nevertheless, he insists that this does not affect the fact that the threshold
exists. At least, an achievable goal is to ‘narrow the area or span’ within
which the threshold lies at. This attitude is also shared by Michael Hannon,
who argues that we just need to provide ‘a reasonable degree of approxima-
tion’ (2017, p. 608) of the threshold’s location.8 Call this the epistemicist
attitude, which will be highlighted later on.
However, as we have seen before, this attitude would invite the two-sense
problem and trivialise the debate between gradualism and absolutism.
Furthermore, it seems that external gradualism is what genuinely worthy of

discussing in some interesting but underexplored aspects of the debate. Con-
sider Dretske’s representative analogy of pregnancy. Knowing a proposition
is like being pregnant – either ‘yes’ or ‘no’ – the threshold is just there. Cer-
tainly, one can be more heavily pregnant than another, but this is not what
Dretske considers when he asserts that being pregnant is absolute.
Similarly, the idea that ‘knowing’ has an absolute threshold would be the
more decisive reason underpinning absolutism insofar as we take Dretske’s
analogy seriously. In what follows, I will develop a new account of gradual-
ism, which interprets knowledge as a spectrum concept instead of a threshold
concept.
4. Threshold versus spectrum
4.1. THRESHOLD CONCEPTS
What do I mean when I say that a concept has a threshold? Ordinarily,

‘threshold’ is understood as a point on a scale at which something starts to
happen or change. The threshold of a concept is a cut-off point where any-
thing starts counting as an instance of that concept. If something is taken
to be of a threshold, then usually this concept can be evaluated or measured
along a certain scale. For example, when we say that someone has a high
threshold for pain, what is evaluated is the strength of pain that one can en-
dure – when the strength reaches the threshold, the pain starts being unbear-
able. Accordingly, let me define:
[Threshold Concept]
A concept C is a threshold concept, iff, at the relevant scale S along which C is evaluated, there is
a cut-off point T distinguishing instances of C from everything that falls short of C.
A threshold concept might be evaluated along multiple scales. For exam-

ple, if knowledge is taken to be a threshold concept, then relevant scales can
include the scale of belief, the scale of justification and the like (e.g. for
Hetherington, how detailed one knows how it is that p). The list of relevant
scales depends on which account of knowledge that one endorses and which
factors are relevant for analysing knowledge in accordance with that
account. The trade-offs between these different scales might ultimately settle
an overall threshold, which typically comes in the form of necessary and suf-
ficient conditions. Threshold concepts are often taken to be ungradable, for
example, ‘extinct’, ‘empty’ and ‘completed’. For these concepts, the thresh-
old is located at the endpoint of the scale and thereby renders these concepts
absolute – either reaches the threshold or not (e.g. either extinct or not
extinct), it is a yes-or-no affair. That is also why the idea that knowledge
has a threshold matches external absolutism. Before we move forward to
applying the notion of ‘threshold concept’ to knowledge, some caveats are
in order.
First, when saying that there is a ‘cut-off point’, I do not deny the fact that
a threshold concept can still have its borderline cases. In other words,
‘threshold’ is compatible with ‘vagueness’ insofar as we understand vague
concepts as concepts having borderline cases. Thus, it is not the case that
gradualism is right and absolutism is wrong just because knowledge, as well
as every philosophically interesting concept, of course, is somewhat vague.
There are many ways to understand how a threshold concept can be
vague. For example, consider epistemicism (refer to Sorensen 1988;
Williamson 1994; Horwich 2000; Hu 2015). Epistemicism of vagueness
argues that there is a sharp boundary between the extension and the
anti-extension of a vague predicate. For example, there is a cut-off point
for one’s counting as ‘bald’. It is just because of our cognitive inability that
we cannot know where the threshold is – but this does not affect the fact that
the threshold exists. Accordingly, vagueness is just the ignorance of the loca-
tion of the threshold. As aforementioned, Hetherington’s attitude (as well as
Hannon’s) towards the boundary problem of knowledge can be seen an
epistemicist one: there is a sharp boundary for knowledge, and our
ignorance of where the boundary is located just undermines our knowledge
that there is a threshold, but it does not undermine the fact that there is a
threshold. Another way is to appeal to fuzzy logic that admits degrees of
truth. One can endorse a form of infinite-valued logic in which the truth
value of a proposition Fφ (φ stands for any potential instance of F) varies
from 0 (definite falsehood) to 1 (definite truth) inclusive. The predicate F
is vague, as any Fφ whose truth value is a real number n (0 < n < 1) that
is very close to 1 can be seen as a borderline case of F (refer to Shapiro 2006;
Smith 2008; Hyde 2010). Correspondingly, F’s threshold can be interpreted
as the point where Fφ starts being designated to the truth value of 1, namely,
the right endpoint of the closed interval [0, 1].
The second caveat: a threshold concept can be context sensitive in the
sense that its threshold might vary from context to context. There is a thresh-
old for a male athlete’s being ‘heavyweight’, but the threshold varies in
different sports – for boxing, the minimum is 200 pounds; while for
wrestling, it is 184 pounds. So when a ring announcer describes an athlete
as ‘heavyweight’, the threshold depends on the context of utterance.
I hope my clarifications are clear enough. So what does it mean to see
knowledge as a threshold concept? Remember that a threshold concept
should be able to be evaluated along a corresponding scale. When it comes
to the threshold for knowledge, what is(are) the scale(s) that we should take
into consideration? There can be plural answers. Most current discussions
around knowledge’s threshold focus on the scale of justificatory strength.
For example, Hetherington (2001) holds that knowledge has a threshold
that lies at 0 justification. Besides, insofar as we take belief to be necessary

for knowledge, the scale of credence for belief is also relevant. In this aspect,
there are disputes about where the threshold for sufficient credence for
yielding knowledge-apt beliefs is. The credence-one view argues that beliefs
require the maximal confidence or credence 1 (refer to Clarke 2013;
Greco 2015), while the threshold view holds that beliefs only require credence
above a threshold that is below degree 1 (refer to Kaplan 1996; Foley 2009).
It is noteworthy that although the credence-one view is seen as an opposite
to the threshold view, it does not deny the existence of a belief’s threshold
(in terms of the degrees of credence). The disagreement between the two
views is only about where the threshold is. Apart from these, various other
interpretations of knowledge’s threshold can be provided in accordance with
different accounts of knowledge. For instance, anti-luck virtue epistemology
advocates an analysis of knowledge as follows:
‘S knows that p if and only if S’s safe true belief that p is the product of her relevant cognitive
abilities (such that her safe cognitive success is to a significant degree creditable to her cognitive
agency)’ (Pritchard 2012, p. 20; emphasis mine).
Accordingly, the threshold for knowledge can be understood as a cut-off

point indicating how creditable is creditable enough for yielding knowledge.
This paper will discuss knowledge’s threshold in a general sense without
being limited to any specific account of knowledge. In summary, external
absolutism is committed to this thesis:
[K-THRESHOLD]
Knowledge is a threshold concept in the sense that there is a threshold Tk, which distinguishes
knowledge from everything that falls short of knowledge.
Recall that many epistemologists who endorse K-THRESHOLD also

admit that it is very hard (if not impossible) to know where Tk is (e.g.
BonJour 2003; Hetherington 2006; Foley 2009; Bovell 2012; Hannon 2017).
Call this attitude moderate K-THRESHOLD, which might be a position
that most absolutists adopt. In contrast, there are also voices claiming that
we know exactly where the threshold is (e.g. Goldman 1999;
Hetherington 2001; Clarke 2013; Greco 2015). Call this attitude radical
K-THRESHOLD. In addition, we have noted that a threshold concept
can also be context sensitive, so the third attitude defending
K-THRESHOLD argues that knowledge has its threshold, while Tk can
vary from context to context in accordance with practical factors, such as
stakes (e.g. Hawthorne 2003; Stanley 2005; Fantl and McGrath 2009;
Hannon 2017). Call this attitude contextualist K-THRESHOLD. Both
moderate and radical K-THRESHOLD are compatible with contextualist
K-THRESHOLD. Endorsing any version of K-THRESHOLD means
being committed to external absolutism.
K-THRESHOLD is prima facie plausible and is deeply entrenched in
standard epistemological thinking. For Hetherington, its apparent
plausibility comes from two aspects: first, linguistic data (refer also to
Stanley 2005; Dutant 2007) suggest that we use ‘knows’ as a ‘yes-or-no’
predicate and the distinction between ‘yes’ and ‘no’ ought to be clear.
Second, most epistemologists are fallibilists who hold that knowledge is
allowed to be fallible to a certain degree. While if there is no such a threshold
for knowledge, then we cannot find the certain degree to which knowledge is
allowed to be fallible, which would significantly undermine the persuasive
power of fallibilism (refer to Hetherington 2006; Hannon 2017).
Now we have outlined the absolutist side of the remodelled debate by
employing K-THRESHOLD. So what about the gradualist side? How can
we conceive of a concept of knowledge with no threshold? I suggest that
gradualists can resort to the idea that knowledge is a spectrum concept.
4.2. SPECTRUM CONCEPTS
What is the threshold for a colour’s counting as red? Where is the boundary
distinguishing ‘red’ from ‘yellow’, ‘blue’ or any other colour? It is not only
practically difficult to draw this boundary but also seemingly impossible in
principle to locate such a threshold. That is because ‘red’, as well as other
colours, refers to a range of colour across a spectrum, of which the
‘boundary’ is too gradient to be regarded as a threshold. Admittedly, if
one is taken through colour samples that start with blue and become gradu-
ally more reddish, there would be some point at which one starts describing
them as red. Nevertheless, this only proves that we are able to tell a spectrum
concept A from another spectrum concept B, but not that there is a cut-off
point distinguishing A from B. Consider eye tests. In an eye test, as the
doctor is adjusting the phoropter, the patient who stares at a picture through
the lens would be constantly asked whether the vision is clear now, until the
patient replies ‘yes’. However, this does not imply that there is a threshold
for a picture’s being ‘clear’ or ‘vague’. When asking for a threshold for ‘clear’
or ‘red’, one is conducting a category mistake. A concept’s threshold is ordi-
narily utilised to identify whether a given instance belongs to the extension of
that concept. However, we do not identify a shade of colour as red by
resorting to the so-called ‘red’s threshold’. Instead, we usually determine
whether a colour is an instance of red by (roughly) comparing it with
paradigmatic cases of red. The more that colour resembles those paradig-
matic cases of ‘red’, the more likely the colour would be recognised as red.
Also, there is no clear threshold for a case’s being paradigmatic. Identifica-
tion of a concept like ‘red’ is not a matter of ‘yes’ or ‘no’, but one of degrees.
So is ‘cold’/‘hot’. A cup of water’s being ‘ice-cold’, ‘cold’, ‘warm’ or ‘hot’ is
not a matter that can be determined by a given threshold. These grades of

‘warmness’ refer to different ranges of temperature on a spectrum of which
there are no cut-off beginning points or endpoints. The spectrum of temper-
ature is continuous, gradient and intrinsically gradable.
Now we have roughly classified a cluster of concepts such as ‘red’, ‘cold’
and ‘clear’ (similarly, ‘hate’, ‘enjoy’ and ‘understand’), which should be
better understood as a matter of spectrum rather than something with a
threshold. They can be categorised as a type of concepts that are opposite
to threshold concepts:
[Spectrum Concepts]
A concept C is a spectrum concept, iff, at the relevant scale S on which C is evaluated, instances
of C fall into a range of more or less paradigmatic cases that come in different degrees, while
there are no cut-off points distinguishing instances of C from anything that falls short of C.
Again, some caveats would be helpful. First is the relationship between

spectrum concepts and ‘vague concepts’. It is an attractive view for a lot of
philosophers that every philosophically interesting concept is somewhat
vague. So are all ‘vague’ concepts spectrum concepts? No. As noted before,
there are some vague threshold concepts. Therefore, again, I am not
delivering a trivial point that knowledge is a spectrum concept just because
it is vague. Also, being a spectrum concept is not the only way for a concept
to be ‘vague’. For example, some functionally defined concepts that are
neither threshold nor spectrum concepts can also be vague. For instance,
‘pen’, ‘desk’ and ‘fork’. Where is the boundary between a tree twig and a
pen or a fork? Can a large flat rock be counted as a desk? The boundaries
are fuzzy. However, they are not spectrum concepts as there are no intrinsic
relevant scales on which those concepts can be evaluated and graded. A
spectrum concept requires a gradable scale intrinsic to the concept’s nature.9
Second, not all context-sensitive concepts are spectral. We have noted that
there are context-sensitive threshold concepts. Also, some indexical terms
such as ‘I’ and ‘this’ are context sensitive, but they are neither threshold
nor spectrum concepts due to the lack of relevant scales.
Third, although spectrum concepts are typically gradable concepts, not
every gradable concept is spectral. A concept might be graded not on a
continuous scale but only in a discrete sense. In other words, the concept’s
different grades do not constitute a spectrum, but a ‘staircase function’.10
For example, academic degrees. One might see ‘academic degree’ as a
gradable notion in that it comes in different levels – bachelor, master and
PhD. However, for each particular level, there is a threshold for being
awarded such a degree to the effect that nothing in between counts as a
higher degree (‘PhD candidate’ is not a degree). Therefore, I am not saying
that being gradable entails having no absolute threshold. Instead, what I
suggest is just that if a concept is spectral, then it is gradable. For a concept’s
being spectral, two conditions have to be met: first, the concept can be
evaluated along a certain scale; second, it does not have a sharp threshold.
It is relatively uncontroversial that knowledge-that can be evaluated along
some relevant scales (e.g. justificatory strength), so the key to interpreting
knowledge as a spectrum concept is rejecting the existence of Tk. This is
why the debate over the existence of Tk is essential for the reconstructed
debate between gradualism and absolutism.
So what does it mean to interpret knowledge as a spectrum concept?
Contrar to K-THRESHOLD, external gradualism holds that in a given
relevant scale, there is no such thing serving as knowledge’s threshold.
Instead, there are better or worse (and of course, more or less paradigmatic)
cases of knowledge that come into different degrees constituting a spectrum.
Again, various relevant scales that can be considered when evaluating the
quality of a piece of knowledge, for example, justificatory strength,
credence, reliability and the creditability.11 To put it more clearly, epistemic
gradualism should strive to defend the following thesis:
[K-SPECTRUM]
Knowledge is a spectrum concept in the sense that there is not a threshold Tk distinguishing
knowledge from everything that falls short of knowledge, but only better or worse instances of
knowledge that can be graded in different degrees along the relevant scale S.
K-SPECTRUM is opposed to all three versions of K-THRESHOLD.

Because K-SPECTRUM denies the existence of Tk, it is incompatible
with both moderate and radical K-THRESHOLD. As for impurism,
K-SPECTRUM can remain neutral regarding whether ‘knows’ is context
sensitive, but insist that in no context, there exists such thing as Tk.
Why should gradualists find K-SPECTRUM promising? First of all, qua
the core thesis of external gradualism, K-SPECTRUM perfectly captures
those motivations for embracing gradualism in the original framework. If
knowledge is a spectrum concept, then it is no wonder that there are
linguistic data using ‘knows-that’ as a gradable phrase. Likewise, it can also
explain why knowledge-that notably resembles knowledge-how and under-
standing – they are homologous in the sense that they are all spectrum
concepts. Finally, K-SPECTRUM can also accommodate the gist of
internal gradualism: a spectrum reading of knowledge can naturally
allow an instance of knowledge to be improved by stronger
justification (this should be clearer when the prototype-theoretic picture of
K-SPECTRUM is introduced later). Hence, reasons supporting gradualism
in the original debate could be retained to support K-SPECTRUM. Apart
from these, K-SPECTRUM also lends new support to gradualism. New
motivations for embracing gradualism can be brought by K-SPECTRUM
in at least two ways:
First, K-SPECTRUM is opposed to K-THRESHOLD, which is

the gist underlying external absolutism. However, K-THRESHOLD is
ill-motivated. Recall that K-THRESHOLD is primarily motivated by two
reasons: that linguistic evidence supports the ungradability of knowledge
and that fallibilism would be undermined if knowledge has no threshold.
Both reasons are untenable. On the one hand, as we have noted, linguistic
evidence does not favour absolutism over gradualism.12 On the other hand,
it is unclear why fallibilism has to presume the existence of Tk. An archer’s
being adroit does not require her to hit the bullseye every time – adroitness
is thus ‘fallible’ in the sense that it is compatible with failures. However, this
does not imply that there has to be a precise hit rate distinguishing adroit
from not-adroit. The lack of such a threshold does not undermine the fact
that adroitness is fallible. Analogously, the lack of Tk will not affect the
plausibility fallibilism. Hence, K-THRESHOLD is questionable. Grounds
for doubting K-THRESHOULD constitute grounds for endorsing
K-SPECTRUM, and thus, gradualism.
Second, insofar as K-SPECTRUM is accepted, the boundary problem
could be thereby resolved because it makes no sense to search for a spectrum
concept’s threshold – spectrum concepts are intrinsically threshold-less. We
have seen that it is hard for epistemologists to locate a non-arbitrary
threshold for knowledge, which in turn inspires us to abandon the
ill-grounded presumption that knowledge must have a threshold.
So how can we distinguish knowledge from what falls short of
knowledge if no threshold is available for us to refer to? By endorsing
K-SPECTRUM, gradualism can reply that it is just like how we tell
whether an object is red.13 Admittedly, there can be no easy and precise
criteria when making judgements like these. In our real-life epistemic
practice, judgements about ‘whether one knows’ are not always easy and
precise. Instead, when encountering some borderline cases, our judgements
are inclined to be full of uncertainty and hesitation – just like when we see
a half-ripe apple. The spectrum view of knowledge can mirror and explain
this hesitation. Nevertheless, it is undeniable that we all have various types
of representative instances of knowledge in mind, for example, logical
knowledge, mathematical knowledge and perceptual knowledge. Those
most representative cases of a concept are referred to as ‘prototypes’ of
that concept per prototype theory (refer to Osherson and Smith 1981;
Kamp and Partee 1995; Hampton 2007; Decock and Douven 2014). As
opposed to the traditional necessary-and-sufficient-conditions approach
to accounting for concepts, prototype theory is a rising alternative, which
argues that whether or not something falls under a given concept depends
on how similar14 the entity is to that concept’s prototype(s). Graded
membership is thus accounted for in terms of similarity/distance to
prototype(s). The closer/more similar to prototype(s) an item is, to the
larger degree, the item falls under the concept. For example, if a shade
of colour S1 is closer to the prototype of red than another shade S2, then
S1 is more reddish than S2. In Hetherington’s case of pain, your knowl-
edge of my pain can be a paradigmatic case of knowledge. Nonetheless,
due to the first-person privilege of self-knowledge, it is also legitimate to
hold that my knowledge of my pain is a more prototypical case of
knowledge and hence a better case of knowledge as knowledge.
With prototype theory in play, it should be easier for us to see how internal
gradability and external gradability can constitute a continuum of spectrum.
A natural implication of prototype theory is to deny the existence of a sharp
boundary between knowledge and non-knowledge (refer to Decock and
Douven, 2014, p. 654) and to embrace a gradational boundary region
consisting of borderline cases. Accordingly, the whole ‘conceptual space’
(a la Kamp and Partee; Hampton; and Decock and Douven) of
‘knowledge-that’ consists of prototype(s) (the central region), more or less
paradigmatic cases (the circum-centre region), and borderline cases
(the boundary region). Internal gradability mainly relates to the circum-
centre region, and external gradability mainly relates to the boundary
region. So does the central region admit of internal gradability? It depends
on, first, whether a concept is taken to be with multiple prototypes; second,
if the answer to the first question is yes, then whether some prototypes are
more prototypical than the others. For example, one might hold that a cer-
tain shade of red is the most prototypical case of red (e.g. RGB [255, 0, 0]),
but there are other prototypes (say, crimson, vermilion and scarlet) that are
less prototypical. In this case, internal gradability can also be found in the
central region of red’s conceptual space. Of course, one can also argue that
all prototypes are equally prototypical or there is only one prototype of red.
In these cases, internal gradability does not apply to the central region. But
no matter what one’s answers to the two questions are, it does not prevent
red from being gradable and spectral.
It is noteworthy that this also indicates that K-SPECTRUM does not
imply that knowledge has no necessary conditions. Just because red is a
spectrum concept, it does not mean that every colour can be recognised as
red. Analogously, the spectrum reading of knowledge does not mean that
‘anything goes’. K-SPECTRUM is compatible with the existence of
necessary conditions of knowledge especially when those conditions
constitute indicators of paradigmatic non-knowledge. In summary, the
more fine-grained and detailed the conceptual space of knowledge is
constructed, the better we can accordingly distinguish knowledge from
non-knowledge – although the distinction can never, and should not be
expected to, be clear-cut and easy.
Although it is beyond the scope of this paper to provide a full argument
for K-SPECTRUM, I hope I have said enough to enable the reader to see
how K-SPECTRUM sketches a promising gradualist picture. That being
said, now we can remodel the absolutism/gradualism debate by
reinterpreting the central divergence as the competition between

K-THRESHOLD and K-SPECTRUM. What are the benefits of doing so?
5. Remodelling the debate

The extant model of debate between gradualism and absolutism focuses on
the competition between internal gradualism and internal absolutism. We
have noted that this model of the debate has two defects, that is, it gives rise
to the dubious ground problem and the two-sense problem. Also, the compe-
tition between external absolutism and external gradualism is neglected in the
current model of debate. That is because Hetherington is also an external ab-
solutist and he does not spell out what an attractive external gradualist
reading of knowledge can be like. In contrast, by reconstructing the debate
on the basis of two irreconcilable theses, to wit, K-THRESHOLD and
K-SPECTRUM, my new model of debate enjoys advantages over the
current one in at least three ways:
First, the distinction between gradualism and absolutism is characterised
more explicitly. The central disagreement between K-THRESHOLD and
K-SPECTRUM is whether there is a threshold for knowledge. Absolutism
says ‘yes’, gradualism argues ‘no’ – the divergence is clear and certain. It
avoids the two-sense problem that the extant debate succumbs to, as there
is no space for an equivocal attitude – a concept cannot both be with and
be without a threshold.
Second, the divergence between gradualism and absolutism is captured
more accurately. Drawing on Hetherington’s own definition of ‘external
absolutism’, this new model bases absolutism on its commitment to the
existence of a threshold for knowledge. Moreover, there is a notable amount
of philosophers advocating the existence of Tk on different relevant scales,
which prevents gradualists from tilting at windmills. By way of illustration,
for the scale of belief, some philosophers endorse credence-one or
threshold-view of belief. For the scale of truth, there are numerous
bivalentists and epistemicists. For the scale of justification, we have
infallibilists (refer to Williamson 2000; Littlejohn 2008; Dodd 2011;
Dutant 2016; Davies 2018; etc.), anti-justificationists (e.g. Hetherington
and Goldman) and a large number of aforementioned proponents of
moderate K-THRESHOLD who can be seen as epistemicists on the
boundary problem of knowledge. Compared with Hetherington’s model
of debate, of which the ‘internal absolutist’ side has no genuine proponents,
the remodelled absolutism can avoid becoming a straw man.
Third, the remodelled debate is more balanced as both sides of the debate
become more appealing. On the one hand, the reconstructed gradualism has
more appeal as it does not need to be combined with the counter-intuitive
‘anti-justificationism’ or the problematic ‘↔H’. Rather, it can be applied
to different accounts of knowledge. Hence, the remodelled debate also
avoids the dubious ground problem. On the other hand, this new model of
debate is not unfairly unfavourable for absolutism. Instead, it also boosts
motivations for advocating absolutism by shifting the focus of the debate
from internal absolutism to external absolutism. My new model preserves
all motivations for being an absolutist that one can hold in the original
model (e.g. linguistic evidence). Besides, it further absorbs some motivations
for absolutism that are compatible with internal gradualism, such as
epistemicism, moderate K-THRESHOLD and the credence-one/threshold
view of belief – it is unclear how these sources can motivate (internal)
absolutism within the original framework of the debate.
The last point that I shall mention without expanding is that this new
model can also lead us to two distinct directions of epistemological
investigation. By endorsing K-THRESHOLD, epistemology should strive
for locating the threshold for knowledge, or at least, narrowing down the
span within which Tk is located. Call this paradigm of epistemological
investigation the threshold paradigm. In contrast, if K-SPECTRUM is ac-
cepted, then we should not even attempt to narrow down the span, as there
is no threshold for knowledge at all. Rather, it is more meaningful to figure
out indicators of prototypical cases of knowledge and how to measure the
similarity/distance between a given item and those prototypes. In doing
so, we can depict the conceptual space, to wit, the spectrum of knowledge
in a more fine-grained manner. Call this the spectrum paradigm.
Under the threshold paradigm, many advocates of K-THRESHOLD see
it as a primary task for epistemology to give a reductive analysis of
knowledge,15 or at least an approximation. That is because a threshold for
knowledge will usually serve as an analysis of knowledge or be used to indi-
cate the necessary and sufficient conditions for an epistemic standing’s being
qualified as ‘knowing’. On the contrary, under the spectrum paradigm, the
anti-reductive-analysis line (refer to, e.g. Zagzebski 1994; Williamson 2000)
is what we should take. It is in principle impossible to define a spectrum con-
cept ‘red’ in terms of necessary and sufficient conditions whereby we can dis-
tinguish red from any other colour. What is important for comprehending a
spectrum concept is not analysis, but characterisation – characterising what a
typical instance of that concept is like, what a less typical instance would be
like and so forth. It is not totally crazy to grasp a concept without resorting
to its necessary and sufficient conditions. Consider family resemblance con-
cepts, such as ‘games’. It is impossible to draw a sharp boundary dividing all
members of ‘games’ from all members of ‘not-games’. Admittedly, it is
worthy of debate whether knowledge is a family resemblance concept16 –
what I try to illustrate here is just that a spectrum paradigm of conceptual
investigation is not only conceivable but maybe also more promising than
it appears to be.
The fruitfulness of the remodelled gradualism/absolutism debate also

embodies in both views’ applications in solving epistemological issues. For
example, which view can better answer the value problems of knowledge?
Can the Gettier problem gain better resolutions within a gradualist frame-
work? Is it more reasonable to argue that knowledge-how is reducible to
knowledge-that, if they are both gradable? How should we rethink the
relation between knowing and understanding?
Important and interesting as they are, this paper is unable to address these
problems due to the limited length. Nevertheless, at this stage, I hope that I
have accomplished the second goal of this paper – to reconstruct a better
framework for the gradualism/absolutism debate.
6. Concluding remarks
The gradualism/absolutism debate is an epistemologically valuable issue
that is worthy of further discussions. However, the extant model of debate
interprets the core divergence of gradualism and absolutism in a problematic
way, and thus succumbs to the two-sense problem and the dubious ground
problem. I suggest that we should remodel the debate by focusing on the
competition between the threshold reading and the spectrum reading of
knowledge-that, given that it can help us to construct a more philosophically
meaningful and fruitful debate. I am far from believing that the consider-
ations developed in this paper make a conclusive case for the debate between
gradualism and absolutism. Instead, I hope that this paper would be seen as
a restart rather than a conclusion of the relevant discussions.17
Department of Philosophy, School of Humanities

Shanghai Jiao Tong University, Shanghai, China
NOTES
1
For an exception, refer to Pavese (2017), where Pavese argues that knowledge-how, as
well as knowledge-that, does not admit of degrees.
2
For more linguistic data, refer to Hetherington (2001). In addition, Dutant (2007) also
lists a considerable amount of English sentences involving gradable uses of ‘knows-that’.
3
Apart from English linguistic counterevidence aforementioned, it is noteworthy that in
Chinese language, ‘知道’ (meaning ‘knows that’) is naturally used as a gradable term.
Hazlett (2010) and Hetherington (2011) also cast doubt upon the linguistic methodology that at-
tempts to reveal the conceptual nature of ‘knows’ by how it is used in the ordinary language.
4
More precisely, Hetherington proposes that knowledge can be graded in terms of its
‘failability’. In brief, ‘knowing failably’ is a broader concept than ‘knowing fallibly’ – the latter
means that S knows that p but S’s belief that p could have been false, while the former means that
S knows that p but S could have failed to do so. For a full discussion about the distinction be-
tween failability and fallibility, refer to Hetherington (1999).
5
Hetherington discusses justification in both internalist and externalist senses. Unless spe-
cially specified, ‘justification’ appearing in this paper can also be understood in either sense.
6
But of course, it is not to say that Hetherington is the only one or the first one who holds
this view. For example, the anti-justificationist position of Goldman (1999) is cited by
Hetherington.
7
Infallibilism can argue that the cut-off point lies at 0 fallibility; however, this answer is
unpalatable for fallibilists.
8
Similar ideas are echoed by Bovell (2012).
9
One might also find my definition of ‘spectrum concepts’ somewhat vague. However, this
does not mean that the concept ‘spectrum concepts’ is a spectrum concept – it lacks relevant
gradable scales.
10
Sosa (2001, 2015) interprets knowledge in a highly similar way. He holds that knowledge
can be graded into three levels: animal knowledge, reflective knowledge and knowledge full well.
However, one might also argue that, within each level, there is a corresponding threshold for the
very level of knowledge.
11
Besides, for the AAA model of virtue epistemology, the scales can be about accuracy,
aptness and adroitness.
12
Elsewhere (refer to Lai 2021), I have provided a throughout objection to the linguistic
argument for absolutism. I argued that, first, linguisitc evidence does not favour absolutism over
gradualism. Second, there is an explanatory gap between how ‘knows’ is used in the ordinary
language and how knowledge-that is understood epistemologically. It is methodologically
mistaken to assume that the ordinary uses of ‘knows that’ can reveal the conceptual nature of
knowledge-that.
13
One may argue that we can describe an object as being ‘kind of red’, ‘pretty red’ or ‘red-
ish’, but it is odd to say something like ‘kind of knowledge’ or ‘knowledge-ish’ (thanks to an
anonymous reviewer reminding me of this). However, there are many other less odd gradable
expressions for knowledge claims. For example, in contexts where the idea of epistemic
gradualism is explained, it should be acceptable to say something like ‘kind of know’ (in fact,
1475 sentences involving this expression can be found in the Google Books Corpus) or ‘knows
pretty well’.
14
For a set-theoretic account of ‘similarity’, refer to Tversky (1977) and Cazzanti and
Gupta (2006); for a geometric account, refer to Kamp and Partee (1995) and Decock and
Douven (2014).
15
One can deny that knowledge is reductively analysable but still endorse
K-THRESHOLD. The point here is just that a typical characteristic of the threshold paradigm
is the pursuit of reductive analysis of knowledge. At least, the threshold paradigm has more
motivations to pursue such an analysis compared with the spectrum paradigm.
16
Kusch (2011) and Kusch and McKenna (2020) argue that knowledge is a family
resemblance concept.
17
I am grateful to Duncan Pritchard, Martin Smith, Stephen Hetherington, Aidan
McGlynn, J. Adam Carter and two anonymous reviewers for valuable comments on earlier
drafts of this paper. I also owe sincere thanks to many people for helping me to improve this
paper, including but not limited to Jiaming Chen, Sanford Goldberg, Donghui Han, Chang
Liu and Yijie Shen.
REFERENCES
BonJour, L. (2003). ‘A Version of Internalist Foundationalism,’ in L. BonJour and E. Sosa (eds)

Epistemic Justification. Maiden, Mass: Blackwell, pp. 21–23.
Bovell, C. R. (2012). Ideas at the Intersection of Mathematics, Philosophy, and Theology.
Eugene: Wipf and Stock.
Brower, B. (2004). ‘Review of Stephen Hetherington, Good Knowledge, Bad Knowledge,’
International Philosophical Quarterly 44(1), pp. 107–108.
Cazzanti, L. and Gupta, M. R. (2006). ‘Information-theoretic and Set-theoretic Similarity,’ 2006
IEEE International Symposium on Information Theory, pp. 1836–1840.
Clarke, R. (2013). ‘Belief Is Credence One (in Context),’ Philosophers’ Imprint 13, pp. 1–18.
Craig, E. (1990). Knowledge and the State of Nature: An Essay in Conceptual Synthesis. Oxford:
Oxford University Press.
Davies, A. (2018). ‘Infallibilism and Easy Counter-Examples,’ Grazer Philosophische Studien
95(4), pp. 475–499.
Decock, L. and Douven, I. (2014). ‘What Is Graded Membership?’ Noûs 48(4), pp. 653–682.
Dodd, D. (2011). ‘Against Fallibilism,’ Australasian Journal of Philosophy 89(4), pp. 665–685.
Dretske, F. (1981). ‘The Pragmatic Dimension of Knowledge,’ Philosophical Studies 40(3),
pp. 363–378.
Dutant, J. (2007). ‘The Limited Gradability of “Know”,’ Available at: http://julien.dutant.free.
fr/papers/JDutant_TheLimitedGradabilityOfKnow.pdf
Dutant, J. (2016). ‘How to Be an Infallibilist,’ Philosophical Issues 26(1), pp. 148–171.
Fantl, J. and McGrath, M. (2009). Knowledge in an Uncertain World. New York: Oxford
University Press.
Feldman, R. (2002). ‘Review of Stephen Hetherington, Good Knowledge, Bad Knowledge: On
Two Dogmas of Epistemology,’ Notre Dame Philosophical Reviews 2002(8).
Foley, R. (2009). ‘Beliefs, Degrees of Belief, and the Lockean Thesis,’ in F. Huber and C.
Schmidt-Petri (eds) Degrees of Belief. Netherlands: Springer, pp. 37–47.
Ginet, C. (2010). ‘Reply to Hetherington,’ Veritas (Porto Alegre) 55(2), pp. 18–23.
Goldman, A. (1999). Knowledge in a Social World. Oxford: Oxford University Press.
Greco, D. (2015). ‘How I Learned to Stop Worrying and Love Probability 1,’ Philosophical
Perspectives 29(1), pp. 179–201.
Greco, J. (2003). ‘Virtue and Luck, Epistemic and Otherwise,’ Metaphilosophy 34(3),
pp. 353–366.
Hampton, J. A. (2007). ‘Typicality, Graded Membership, and Vagueness,’ Cognitive Science 31,
pp. 355–384.
Hannon, M. (2017). ‘A Solution to Knowledge’s Threshold Problem,’ Philosophical Studies
174(3), pp. 607–629.
Hawthorne, J. (2003). Knowledge and Lotteries. Oxford: Oxford University Press.
Hazlett, A. (2010). ‘The Myth of Factive Verbs,’ Philosophy and Phenomenological Research
80(3), pp. 497–522.
Hetherington, S. (1999). ‘Knowing Failably,’ Journal of Philosophy 96(11), pp. 565–587.
Hetherington, S. (2001). Good Knowledge, Bad Knowledge: On Two Dogmas of Epistemology.
New York: Oxford University Press.
Hetherington, S. (2005). ‘Knowing (How It Is) That P: Degrees and Qualities of Knowledge,’
Veritas 50(4), pp. 129–152.
Hetherington, S. (2006). ‘Knowledge’s Boundary Problem,’ Synthese 150(1), pp. 41–56.
Hetherington, S. (2011). How to Know: A Practicalist Conception of Knowledge.
Wiley-Blackwell.
Horwich, P. (2000). ‘The Sharpness of Vague Terms,’ Philosophical Topics 28(1), pp. 83–92.
Hu, I. (2015). ‘Epistemicism, Paradox, and Conditional Obligation,’ Philosophical Studies
172(8), pp. 2123–2,139.
Hyde, D. (2010). ‘Vagueness, Logic and Ontology,’ The Bulletin of Symbolic Logic 16(4),
pp. 531–533.
Jaster, R. (2013). ‘Contextualism and Gradability,’ GAP.8 Proceedings, pp. 319–323.
Kamp, H. and Partee, B. (1995). ‘Prototype Theory and Compositionality,’ Cognition 57,
pp. 129–191.
Kaplan, M. (1996). Decision Theory as Philosophy. Cambridge: Cambridge University Press.
Kusch, M. (2011). ‘Knowledge and Certainties in the Epistemic State of Nature,’ Episteme 8(1),
pp. 6–23.
Kusch, M. and McKenna, R. (2020). ‘The Genealogical Method in Epistemology,’ Synthese
197, pp. 1057–1076.
Kvanvig, J. L. (2003). The Value of Knowledge and the Pursuit of Understanding. Cambridge:
Cambridge University Press.
Lai, C. (2021). ‘Against epistemic absolutism,’ Synthese. https://doi.org/10.1007/s11229-020-
02964-x
Littlejohn, C. (2008). ‘From E = K to Scepticism?’ The Philosophical Quarterly 58(233),
pp. 679–684.
Osherson, D. N. and Smith, E. E. (1981). ‘On the Adequacy of Prototype Theory as a Theory of
Concepts,’ Cognition 9, pp. 35–58.
Pavese, C. (2017). ‘Know-How and Gradability,’ Philosophical Review 126(3), pp. 345–383.
Pritchard, D. (2007). ‘Anti-Luck Epistemology,’ Synthese 158(3), pp. 277–297.
Pritchard, D. (2012). ‘Anti-Luck Virtue Epistemology,’ Journal of Philosophy 109(3),
pp. 247–279.
Ryle, G. (1949). The Concept of Mind. Hutchinson & Co.
Shapiro, S. (2006). Vagueness in Context. Oxford: Oxford University Press.
Smith, N. J. J. (2008). Vagueness and Degrees of Truth. Oxford: Oxford University Press.
Sorensen, R. A. (1988). Blindspots. Oxford: Clarendon Press.
Sosa, E. (2001). ‘Human Knowledge, Animal and Reflective,’ Philosophical Studies 106(3),
pp. 193–196.
Sosa, E. (2007). A Virtue Epistemology. New York: Oxford University Press.
Sosa, E. (2015). Judgment & Agency. Oxford: Oxford University Press UK.
Stanley, J. (2005). Knowledge and Practical Interests. New York: Oxford University Press.
Tversky, A. (1977). ‘Features of Similarity,’ Psychological Review 84, pp. 327–354.
Williamson, T. (1994). Vagueness. London: Routledge.
Williamson, T. (2000). Knowledge and Its Limits. Oxford: Oxford University Press.
Zagzebski, L. (1994). ‘The Inescapability of Gettier Problems,’ The Philosophical Quarterly
44(174), pp. 65–73.

LaiC (2021) Epistemic Gradualism Versus Epistemic Absolutism

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

LaiC (2021) Epistemic Gradualism Versus Epistemic Absolutism

Uploaded by

Copyright:

Available Formats

14680114, 2022, 1, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/papq.12356 by Vanderbilt University, Wiley Online Library on [30/12/2022].

Abstract: Epistemic absolutism holds that knowledge-that is ungradable, while

1. Gradualism and absolutism

Knowledge-how is ordinarily taken to be gradable1 – one can know how to

2. The current debate

(i) Knowledge is to be absolutely distinguished from whatever is not knowledge (whatever is

not hard to understand why he endorses external absolutism. For

This allows us to outline Hetherington’s answer to the third essential

3. Defects of the current framework

So far, I have outlined Hetherington’s (two versions of) epistemic gradual-

1 A minimal aspect of φ entails that φ, and vice versa;

In Hetherington’s original argument, φ refers to ‘how-knowledge that p’, of

‘dual-purpose’ is not. Similarly, ‘having two colours’ is the minimal aspect

Richard Feldman, who is ‘not persuaded by the central arguments’ of

Furthermore, it seems that external gradualism is what genuinely worthy of

4. Threshold versus spectrum

4.1. THRESHOLD CONCEPTS

What do I mean when I say that a concept has a threshold? Ordinarily,

A threshold concept might be evaluated along multiple scales. For exam-

that lies at 0 justification. Besides, insofar as we take belief to be necessary

Accordingly, the threshold for knowledge can be understood as a cut-off

Recall that many epistemologists who endorse K-THRESHOLD also

4.2. SPECTRUM CONCEPTS

not a matter that can be determined by a given threshold. These grades of

Again, some caveats would be helpful. First is the relationship between

K-SPECTRUM is opposed to all three versions of K-THRESHOLD.

First, K-SPECTRUM is opposed to K-THRESHOLD, which is

reinterpreting the central divergence as the competition between

5. Remodelling the debate

The fruitfulness of the remodelled gradualism/absolutism debate also

Department of Philosophy, School of Humanities

BonJour, L. (2003). ‘A Version of Internalist Foundationalism,’ in L. BonJour and E. Sosa (eds)

You might also like