Open Journal of Philosophy
2012. Vol.2, No.4, 260-267
Published Online November 2012 in SciRes (http://www.SciRP.org/journal/ojpp) http://dx.doi.org/10.4236/ojpp.2012.24038
Copyright © 2012 SciRes.
Situating Cornerstone Propositions
Department of Philosophy, Universit y of Ottawa, Ottawa, Canada
Email: pphilie@uot tawa.ca
Received July 31st, 2012; revised September 2nd, 2012; accepted September 14th, 2012
Ostensibly, Wittgenstein’s last remarks published in 1969 under the title On Certainty are about episte-
mology, more precisely about the problem of scepticism. This is the standard interpretation of On Cer-
tainty. But I contend, in this paper, that we will get closer to Wittgenstein’s intentions and perhaps find
new and illuminating ways to interpret his late contribution if we keep in mind that his primary goal was
not to provide an answer to scepticism. In fact, I think that the standard reading (independently of its
fruitfulness with dealing with scepticism) leads to a distorted view of Wittgenstein’s contribution in On
Certainty. In order to see that, scepticism will first be briefly characterised, and then I will attempt to ci r-
cumscribe more precisely the standard reading of On Certainty. In Criticism of the Standard Interpreta-
tion, three exegetical arguments against the standard reading are offered—the hope being that the weight
of these three arguments, taken together, instils doubt in the reader’s mind about the correctness of the
standard reading. The paper concludes with an attempt to gesture at the philosophical significance of On
Certainty once we set aside the standard reading.
Keywords: Wittgenstein; Skepticism; On Certainty; Cornerstone Propositions; Hinge Propositions
Scepticism is not irrefutable, but obvious nonsense if it tries
to doubt where no question can be asked.
For doubt can only exist where a question exists; a question
can only exist where an answer exists, and this can only exist
where something can be said.
Ostensibly, Wittgenstein’s last remarks published in 1969
under the title On Certainty are about epistemology, more pre-
cisely about the problem of scepticism. After all, Wittgenstein
is here reacting to Moore’s “common sense” answer to the scep -
tic. Moreover, the central focus of these remarks is on Moore’s
claim to know some propositions (the so-called cornerstones),
against the sceptic’s claim that these propositions are not
known. Wittgenstein’s celebrated suggestion is that we do not
know these propositions (contrary to what Moore claimed and in
agreement with the sceptic), but that nevertheless these proposi-
tions play a special part in our conceptual scheme. It is usually
agreed that this privileged status is what insulates these propo-
sitions from the sceptical threat according to Wittgenstein.
Hence, on this “standard” reading of On Certainty, the task of
circumscribing a privileged class of propositions is a means
towards the end of proving the sceptic wrong—and the accom-
plishment of that task is the driving theme of the work.
This interpretation is widespread in the literature surrounding
On Certainty. For example, Michael Williams, in his penetrate-
ing study of scepticism, talks of an “anti-sceptical strategy” in
relation to On Certainty and also, perhaps more cautiously, of a
“Wittgensteinian” response to scepticism (Williams, 1991: p.
28). Marie McGinn also appears to accept the standard inter-
pretation when she writes that the “primary aim” of On Cer-
tainty is to “steer a course between the sceptic’s doubt and
Moore’s dogmatism” (McGinn, 1989: p. 104).1 Neil Gascoigne
explicitly asserts that “it is clear that through an engagement
with Moore’s essays, On Certainty constitutes an attempt to
respond both to the argument from ignorance (Moore’s ‘Proof’)
and the Agrippan argument (Moore’s ‘Defence’)” (Gascoigne,
2002: p. 143). P. M. S. Hacker is closer to the alternative inter-
pretation to be suggested here—but nevertheless accepts the
standard view—when he claims, talking about Wittgenstein’s
attitude towards scepticism in On Certainty, that the sceptical
point of view is “not to be answered by proving that we do
know what the sceptic doubts, but rather by showing that the
sceptical doubts make no sense” (Hacker, 1986: p. 208). Stroll,
in his study of the book (1994: p. 81), claims that what Witt-
genstein is “getting at” in On Certainty “leads to a way of
thinking about scepticism that is partly directed to those at-
tracted to it, partly to those, like Moore, who wish to reject it”.
He also refers to Wittgenstein’s “treatment” of scepticism
(1994: pp. 8, 166).
What all these interpretations have in common is their attri-
bution of the motivation behind On Certainty, namely that it
should be read as a response to scepticism. This is the claim I
want to put under pressure here. It should be stressed that it
must not be confused with a rebuttal of the claim that Wittgen-
stein’s remarks in that book can be “borrowed” or made to fit
an anti-sceptical strategy in a fruitful and interesting way—the
pregnant works by the authors just mentioned prove it elo-
quently. There is nothing wrong with such attempts—they even
showcase the fertility of the work. But I contend, in this paper,
that we will get closer to Wittgenstein’s intentions and perhaps
find new and illuminating ways to interpret his late contribution
1Also: “Wittgenstein, like Moore, Austin and Cavell, accepts that any phi-
losophically satisfactory response to the sceptic must yield an unqualified
form of common sense”
. 101: e
if we keep in mind that his primary goal was not to provide an
answer to scepticism.2 In fact, I think that the standard reading
(independently of its fruitfulness with dealing with scepticism)
leads to a distorted view of Wittgenstein’s contribution in On
Certainty. In order to see that, scepticism will first be briefly
characterised, and then I will circumscribe more precisely in
The Standard Interpretation of on Certainty, the standard read-
ing of On Certainty. In Criticism of the Standard Interpretation,
three exegetical arguments against the standard reading are
offered—the hope being that the weight of these three argu-
ments, taken together, instils doubt in the reader’s mind about
the correctness of the standard reading. The paper concludes in
The Purpose of On Certainty, with an attempt to gesture at the
philosophical significance of On Certainty once we set aside
the standard reading.
Moore and Scepticism
What is the nature of scepticism in the present context? In
the broadest possible terms, the sceptic is this unreal but very
nagging philosopher who defends the view according to which
we cannot claim to know any empirical proposition.3 Let us
compare this with Moore’s proof in his “Proof of an External
World”. It is important to bear in mind that Moore’s proof is
designed primarily to prove the existence of external things, not
to prove that we can know empirical propositions. Indeed,
Moore stands up, shows his hands, and claims to have proven
the existence of external things.4 Hence, his proof is directed at
those who doubt the existence of the external world, and such a
characterisation seems to be more appropriate to describe ide-
alists rather than sceptics. However, Moore’s proof bears di-
rectly on the issue at hand, thanks to his argument to the effect
that his “proof” displays all the features of a real proof. Part of
this argument consists in claiming that the premises of the
proof are known, hence that Moore knows that he has a hand
(contrary to what t he sceptic claims).5 It is therefore undeniable
that Moore’s proof is relevant to scepticism as understood
here—offering what is in fact a “straight” response to scepti-
cism, that is, one denying what the sceptic is claiming. A useful
way to look at the matter is to formulate both Moore’s proof
and the sceptic’s claim in terms of Crispin Wright’s I-II-III
template. Here is Moore’s proof as understood by Wright6:
1) It looks to me as if I have a hand.
2) I have a hand.
3) There is a material world.
In this “proof”, 1) justifies 2) and 2) justifies 3). The sceptic,
however, is likely to point out that both inferential moves are
not warranted. Let us start with the move from 1) and 2) to 3).
Moore, according to the sceptic, got things the wrong way
round: a correct demonstration of “I have a hand” should start
with, or assume, the claim that there is a material world. We
cannot start with how things appear to us and then infer from
these appearances that things are as a matter of fact thus-and-so
in order to prove that there is a material world. On the contrary:
it is only by presupposing or having already established that
there is a material world that a knowledge-claim such as “I hav e
a hand” is possible at all. The move from 1) to 2) is likewise
unwarranted according to the sceptic: she will argue, contra
Moore, that appearances can always be misleading and that we
cannot trust our senses—this is in fact the main argumentative
weapon in the sceptic’s arsenal. Since the claim that there is a
material world can only be proved with claims such as 2), it
follows that establishing what Moore wants to establish is im-
possible. Hence the sceptic’s reaction to Moore’s proof is
two-pronged: she holds that Moore’s proof does not succeed in
establishing that he knows he has a hand, and also, independ-
ently of the success of this demonstration, that it does not any-
way succeed in establishing that there are external things.
Hence we can identify two kinds of scepticism, and it will be
useful to distinguish them by name: I will call the kind of scep-
ticism that takes issue with the move from type-I to type-II
propositions knowledge scepticism, and the kind of scepticism
that takes issue with the move from type-I and type-II to
type-III propositions external world scepticism.
Wright’s characterisation of Moore’s proof and its correlated
sceptical reaction is a template and can be applied to any em-
pirical knowledge-claim, in which case 1) and 2) will differ in
content. It will be useful to assign types to 1), 2) and 3) above.
A type-I proposition, in this context, will be any proposition
pertaining to how things appear to a subject. A type-II proposi-
tion will be a proposition that states that things in the empirical
world are as they seem to be in its associated type-I proposition.
A type-III proposition will be a proposition whose status as a
justified belief, according to the sceptic, is presupposed for the
establishment of any type-II proposition.7
2One must not forget that a lot of interesting and highly relevant work per-
taining to On Certainty but, crucially, not exclusively centered around the
issue of s cept ici sm is cur ren tly b ein g don e (s ee fo r i nst ance Mo yal- Sh arro ck
monograph (2005) and the essays she edited with Brenner (2007). See also
Coliva (2010)). The point I wish to make is not affected
y the exis tence o
such work . All I wish to argu e for is a thes is about the dr iving theme of On
Certainty. It is important to note also that I am not implying that On Cer-
tainty is a mere offshoot of the Philosophical Investigations or that it does
not say any t h ing over and above PI, etc.
3I go along with Stroud’s characterisation of the sceptic’s main claim: “we
can know no th ing about the world around us” ( Stroud, 1984: p. 39).
4Here is Moore’s proof: “I can prove now, for instance, that two human
hands exist. How? By holding up my two hands, and saying, as I make a
certain gesture with my right hand, ‘Here is one hand’, and adding, as I
make a certain gesture with the left, ‘and here is another’. And if, by doing
this, I have proved ipso facto the existence of external things, you will all
see that I can also do it now in numbers of other ways: there is no need to
multiply examples.” (Moore, 1939: pp. 165-166).
5These explanations about Moore’s target and its relation to scepticism are
worth mentioning. Indeed, Moore’s proof is often understood and assumed
as being directed towards scepticism as defined here, but this does not seem
quite right. The sceptic (as understood here) does not need to deny the exis-
tence of external things. What he denies is that we can know things about
the world, not that the world exists.
The Standard Interpretation of on Certainty
Wittgenstein writes, in the opening lines of On Certainty:
“1) If you do know that here is one hand, we’ll gr ant you all the
In terms of the I-II-III template, Wittgenstein starts by sug-
6See Wright (2002, 2004a). Moore’s proof as stated in “Proof of an External
World” starts with 2), but Wright points out that the sceptic is likely to point
out that 1) is a necessary premise that should be added : ‘The sceptic will insis t
that Moore did not formulate his ‘proof’ properly – that he begins in the
wrong place, since his premise is something which rests on more basic
evidence and is thus more properly viewed as a lemma” (Wright, 2004b: p. 26).
7It should be noted that this I-II-III schema can also be adapted to other
forms of scepticism, namely scepticism about other minds and scepticism
about induction. This generalisation is what prompts Wright to suggest a
unified strategy against scepticism. However, we need not, for the pur
at hand, go into these details here.
Copyright © 2012 SciRes. 261
gesting (“all the rest”) the consequences of establishing that one
can know a type-II proposition. But what are these conse-
quences? There is an ambiguity here. Firstly, Wittgenstein
could be referring to Moore’s objective in his “Proof of an Ex-
ternal World”, namely the establishing of the existence of t hing s
external to the mind. Secondly, he could also be referring to
what Moore is trying to show in “A defense of Common Sense”,
namely that he knows some propositions with certainty. Indeed,
both objectives would be achieved if Moore really knew that he
has a hand. Either way—Wittgenstein was perhaps implying
both consequences—the conditional is an indication that this
kind of knowledge-claim will be the focus of the discussion.
Indeed: as is well-known, it will be the main point of Wittgen-
stein’s remarks in On Certainty taken as a whole to study what
it means to say that in normal cases, we do not know such
things.8 The second remark sets the stage in that regard:
“2) From its seeming to me—or to everyone—to be so, it
doesn’t follow that it is so.
What we can ask is whether it can make sense to doubt it”.
This remark is seminal for the rest of the book: indeed, it
raises the fundamental issue of the meaningfulness of doubting
some empirical propositions, an issue on which Wittgenstein
returns again and again and on which the plausibility of every-
thing else he says depends. But §2 is also ambiguous. If we
consider what he said in §1, Wittgenstein is talking about a
type-II proposition (“here is a hand”) and its relation to a type-I
proposition (“it seems to me as if I have a hand”). On this
reading, the proposition whose status is under study is a type-II
proposition. But if we consider §4, in which Wittgenstein dis-
cusses “I know that I am a human being” and puts under pres-
sure the meaningfulness of that claim, then he could as well be
discussing type-III propositions.9 I mention this because the
following point is not often remarked: in Moore’s proof, his
‘here is one hand’ is one for which, according to Wittgenstein,
doubt is not meaningful—namely a “cornerstone”—and this
fact is worth noticing since the “proof” would not have the
strength it is intended to have if the type-II proposition was
rather “there are 5 pennies in my pocket”.10 Hence, when the
sceptic insists—as Wright contends she should—to add the
type-I proposition in Moore’s proof and then exploits the gap
between the type-I and the type-II proposition, she is doing so
in relation to a special type, so to speak, of type-II proposition,
namely one that is a cornerstone proposition according to Witt-
genstein. This clearly shows that a Wittgensteinian cornerstone
is not necessarily a type-III proposition: some type-II proposi-
tions are also to be excluded from doubt, and this must be kept
in mind when evaluating the convincingness of Moore’s
“proof” and Wittgenstein’s reaction to it. When he raises the
question of the intelligibility of doubting a proposition on the
basis of things seeming thus and so in §2, Wittgenstein is pre-
sumably starting to draw a cleavage between “ordinary” type-II
propositions and cornerstone propositions such as “I have a
hand”. This dialectical issue between Wittgenstein and the
sceptic is important, since if Wittgenstein can convincingly
show that some type-II propositions cannot be doubted, he will
have eliminated one of the sceptic’s main weapons, namely her
claim that since every and any empirical proposition can be
doubted (because no type-I proposition can ever ground ade-
quately a type-II proposition), we do not know anything. In any
case, Wittgenstein is in the business of examining the status of
cornerstone propositions. On this particular and crucial point in
relation to Wittgenstein’s intentions in On Certainty, every-
one—including the present writer—is in agreement.
From what has been said so far, it is very tantalising to inter-
pret On Certainty as a work of traditional epistemology, con-
cerned mainly with replying to the sceptical threat. Indeed,
Wittgenstein had started to think about the issues making up the
bulk of On Certainty upon his reading of Moore’s papers,
which are directly addressing the sceptic. Moreover, it has just
been pointed out that one of the main topics of the book is
about the possibility of doubting some propositions which are
central to our conceptual scheme. If some propositions cannot
be meaningfully doubted, then the sceptic’s dialectic cannot get
off the ground, since her main argumentative strategy is to say
that any and every proposition can be doubted, and hence can-
not be established as known (I shall have more to say on the
relation between knowledge and doubt below).
An instance of the standard reading of On Certainty sees it as
a sceptical response11 to scepticism (whereas Moore’s is a
straight one). According to that reading, Wittgenstein accepts
the sceptical conclusion according to which cornerstone propo-
sitions can never be justified. However, even if we’re not justi-
fied, we nevertheless possess another, weaker kind of warrant
for them: we are entitled to them. We can “accept” them with-
out doing any of the usual evidential work that is required for
normal, non-cornerstone empirical propositions, and this “free”
acceptance12 is warranted since these cornerstone propositions
are necessary for thought and cognitive endeavours. So on this
interpretation, the sceptic is right when she concludes that we
are not justified in claiming to know cornerstone proposi-
tions—it’s just that her conclusion is not as dramatic as she
takes it to be. To put it in relation to Moore’s proof, Wittgen-
stein is arguing, according to this reading, against Moore that
type-III propositions can neither be proved nor known, and in
that sense the sceptic is right. But the sceptic’s dramatic con-
clusion—that nothing can be known since all knowledge de-
pend on type-III propositions and the latter are not knowable –
is hasty since we possess a warrant for cornerstones and this
warrant is enough to give us a good conscience about our cog-
nitive practice and our claims to know type-II, “normal” em-
8One way to characterise the distance between my reading of On Certainty
with the standard one would be the following: I agree that Wittgenstein tries
to establish that we do not know a certain class of propositions and that this
has important consequences for scepticism. However, I think that what he is
mainly striving at is to illuminate the status of these propositions and their
situation in our conceptual scheme. It does not take Wittgenstein—or so I
want to suggest here—676 remarks to (merely) establish that we cannot be
said to know a certain kind of proposition. The bulk of his effort consists in
trying to understand, to describe their role and their pec uliar status.
9It is sometimes difficult to classify propositions in the I-II-III schema. “I
know that I am a human being” belongs to that recalcitrant category. Indeed, it
cannot be, in a natural way, said to be a proposition supported by a type-I
proposition. But it does not have the generic character of a typical type-III
propositi on eith er. Wright charact erises such proposit ion as “ suppor ted by—
by normal standards—an overwhelming body of evidence, whoses i gn if i ca nc e
would have to be overridden if they w ere doubted” (Wright, 2004b).
10Wright is a notable exception. He carefully (Wright, 2004b: pp. 38-47)
distinguishes between the different kinds of cornerstone propositionsdiscussed
11In his recent writings on entitlement, Wright champions that interpretation
(2002, 2004a, 2004b).
12This is what prompts Wright (2007) to talk of a “Welfare State Episte-
mology” with regard to these propositions.
Copyright © 2012 SciRes.
What is common to most readings13 of On Certainty (includ-
ing the one just mentioned) is their understanding of the dialec-
tical place of cornerstone propositions.14 They are seen as
mainly motivated by a desire to provide an answer to the scep-
tical threat, in such a way as to make it almost natural and in-
evitable to read On Certainty as a work whose primary target is
scepticism. Interpreted along these lines, Wittgenstein is here
offering an argument against scepticism: while the sceptic con-
tends that everything can be doubted or is not justified, Witt-
genstein enters the debate and offers an articulated and argu-
mentative reply to the effect that contrary to what the sceptic
holds, not everything can be meaningfully doubted. The reason
why not everything can be doubted—always assuming the
standard reading—is that some propositions form the backbone
of our conceptual scheme in the sense that they make the game
of giving and asking for reasons possible. It is this feature that
is responsible for their privileged status.
Criticism of the Standard Interpretation
There are many reasons to resist that reading of On Certainty.
In what follows, I will identify and discuss three of them:
In his Philosophical Investigations and other writings lead-
ing to it, Wittgenstein had already “dealt” with scepticism.
It would thus be incongruous, at best, if he devoted his very
last remarks to pursue (or at best to deepen) an already vis-
The standard reading naturally leads to interpret Wittgen-
stein as advancing a kind of transcendental argument, and
this is in tension with his philosophical outlook.
More generally, the claim that Wittgenstein is mainly con-
cerned, in On Certainty, with elaborating a response to the
sceptical threat implies that he is in the traditional philoso-
phical arena, tackling a traditional problem. But this is in-
consistent with a crucial aspect of his so-called “later” phi-
losophy, namely his professed quietism.
Let us begin with the first point. I will start with the first
form of scepticism, namely knowledge scepticism. According to
a venerable tradition15 of Wittgensteinian interpretation, the
celebrated “private-language argument” of the Philosophical
Investigations and what comes before it in the book has impor-
tant ramifications with regard to the “problem” of knowledge-
scepticism. This tradition of Wittgensteinian interpretation de-
serves to be taken seriously. Indeed, in the course of showing
the impossibility of a private language, Wittgenstein is led to
argue against a pervading view of the mind, that of an ‘internal’
theatre in which private sensations are accessible through some
kind of inner perception (the “mind’s eye”, to use an evocative
and established expression). This conception of the mind finds
its source in Descartes, the father of modern epistemology. That
the birth of modern epistemology—with its obsession with
knowledge-scepticism—and this conception of the mind come
simultaneously in the history of ideas is not a coincidence. In-
deed, Cartesian doubt finds its source in the idea that we have
no guarantee that the information about the world obtained
through our senses is trustworthy. Its lack of trustworthiness
stems from the fact that all we can be certain of are our sensa-
tions. To move from what the world appears to be thanks to our
private sensations to the claim that the world really is thus-and-
so is an inferential step that can never be warranted due to the
possibility of error, malicious daemon, evil genius—whatever
standard sceptical scenario you might imagine. A crucial prem-
ise of knowledge-scepticism thus conceived is that our access
to the world is necessarily mediated by the possession of these
private “representational” items (whether they are labelled
“sense data”, “sensations”, or “ideas”). It is from the problem-
atic inferential step from “things appear to me as being thus-
and-so” (thanks to the private, representational items) to “things
are thus-and-so” that knowledge-scepticism finds its source. To
put it in terms of the I-II-III template, knowledge-scepticism is
possible because we can never be justified in inferring type-II
propositions from type-I propositions.
The exegetical work has already extensively been done to
show that Wittgenstein is indeed arguing against that Cartesian
picture of the mind.16 Nevertheless, it would be helpful to study
a few remarks of the Investigations that are particularly relevant
in relation to scepticism and certainty. Let us have a look at an
important passage (for the present purpose) of the Investiga-
tions, one related to our understanding of colour-words:
“273. What am I to say about the word ‘red’?—that it means
something “confronting us all” and that everyone should really
have another word, besides this one, to mean his own sensation
of red? Or is it like this: the word ‘red’ means something
known to everyone; and in addition, for each person, it means
something known only to him? (…)
13Rush Rhees i s a notable ex ception. See in p articular th e notes of his semi-
nars as compiled by D. Z. Phillips in Rhees (2003). The present paper is
written while having in mind the spirit,if not the letter, of Rhees’ lifelong
investigations pertaining on Wittgenstein’s so-called “third” masterpiece. It
could be said that Rhees did not make a case to the effect that Wittgenstein’s
thoughts in this se t of remarks are not about scepticism because it was obvious
for him that Wittgenstein was primarily interested in something el se— na mel y
our huma n cognitive condition, of which more in the last section.
14A note on the expression “cornerstone proposition” is in order here. Ac-
cording to some interpreters—notably Stroll (1994)—cornerstones are pre-
cisely sp ecial i n th at th ey are not propositions. However, since I wish in this
paper to rem ain neut ral on this question, I will continue to use the consecr ated
expression “cornerst one pr oposition ” without there by comm itting my self to a
propositional understanding of cornestones.
15Stanley Cavell’s The Claim of Reason is the locus classicus of this strand.
More recent writings influenced by Cavell’s sophisticated reading include
Andrea Kern’s and Marie McGinn’s contributions in the collection of essays
on Wittgenstein and Scepticism edited by D. McManus (2004). In what
follows, I do not claim to follow Cavell’s interpretation—perhaps my inter-
pretation diverges from his, but the spirit is preserved in that I agree with
Cavell about the link between scepticism and what Wittgenstein is aiming at
in his Philosophical Investigations. Entering into an exegesis of Cavell’s
views would require a separate article.
275. Look at the blue of the sky and say to yourself ‘How
blue the sky is!’—When you do it spontaneously—without
philosophical intentions—the idea never crosses your mind that
this impression of colour belongs only to you. And you have no
hesitation at exclaiming that to someone else. And if you point
at anything as you say the words you point at the sky. (…)”
Here Wittgenstein is saying, in essence, that it is a mistake to
say, when we use colour words to represent the world, that their
16Rorty’s interpretation of Wittgenstein as criticising the Cartesian picture o
the mind in his (1979) has been highly influential. While I agree with Rorty
on this particular point, I wish to distance myself from his anti-foundationa-
list interpretation of Wittgenstein. Incidently, I also wish to distance mysel
from foundationalist interpretations—such as Stroll (1994)—since the at-
tempt to categorise Wittgenstein as either a foundationalist or an anti-foun-
dationalist seems to imply ascribing to him a predominant concern with
scepticism. In any case, I would like to remain neutral on this issue for the
time bein g, if only for the sake of trying to gain a fr esh persp ective.
Copyright © 2012 SciRes. 263
use is ultimately justified through the mediation of private
items of consciousness, of “sense data”. To paraphrase one of
Wittgenstein’s well known remarks, when we say that some-
thing is red, “we do not stop anywhere short of the fact”, but we
mean this-is-red (§95). We are tempted, when we do philoso-
phy, to “assume a pure intermediary between the propositional
signs and the fact” (§96).17 But this is, Wittgenstein says, a
“chimera” (ibid.)—in other words, there is no such thing as a
“representationalist” interface between us and the world.
The foregoing is a picture painted in broad strokes, no doubt
about it. But it could be argued that some of the most important
and lasting movements of 20th-Century philosophy have been
concerned with criticising that picture. Indeed, if its contours
are adequate, that picture has been the target of philosophers
such as Wittgenstein, Heidegger, Sellars, and, closer to us,
Davidson, Rorty, and McDowell. All these philosophers jetti-
son the Cartesian conception of the mind and try to show that
the result of doing without that conception reveals either that
knowledge-scepticism is a pseudo-problem or that scepticism is
futile and uninteresting.
In any case, it is difficult to dispute that Wittgenstein is criti-
cising the Cartesian picture of the mind in the Investigations. If
this is right, and if that Cartesian picture includes the “repre-
sentational” role of intermediary private entities such as sense
data (or any other conception of the “Given”), then Wittgen-
stein is targeting a conception of the mind that has been under-
lying modern philosophy for four centuries. Since that concept-
tion of the mind made a kind of scepticism possible—notably,
the knowledge-scepticism addressed by Moore and by the
I-II-III template—then it seems fair to say that Wittgenstein, in
the Philosophical Investigations, had “dealt” with knowledge-
scepticism and revealed it for what it was—a pseudo-problem
that should never have surfaced in the first place.
External World Scepticism
What I have been arguing so far is that Wittgenstein, by the
time of the Philosophical Investigations, had the complete
framework to address the issue of knowledge-scepticism. It
remains, however, to get to grips with the other kind of scepti-
cism, namely what I have called external world-scepticism.
This kind of scepticism takes issue with the step from type-II to
type-III propositions and claims that we cannot be said to know
that there is an external world since in order to establish that,
we have to rely on type-II propositions such as “I have a hand”
or “this chair is orange” (that is, both types, so to speak, of
type-II propositions—the ones respectively identified by Moore
as ‘certain’ and the ordinary ones). But such a reliance leads to
a circularity, since in order to be said to know a type-II proposi-
tion, one must already have established that there is an external
world. But one can only do that on the basis of a type-II propo-
sition—we’re caught in a vicious circle. Does the Investigations
have the resources to deal with that second kind of scepticism?
I wish to argue that it does.
Let us reflect here that the key idea in relation to external
world-scepticism, in On Certainty, is to stress the peculiar na-
ture and role of statements such as “I have a hand” and “there is
an external world”. There is no consensus as to what, exactly, is
the status assigned to these propositions in On Certainty. Are
they rules of grammar?18 Are they meaningless?19 This is, ob-
viously a key issue—not only for the debate around scepticism
of any kind, but also with regards to the favoured reading here,
namely with respect to an investigation about the human cogni-
tive condition. What I would like to point out is that it is agreed
on all sides that Wittgenstein’s fundamental idea here is that
such propositions cannot be meaningfully be doubted, and for
that reason we cannot be said to know them. It also leads to the
ideas that these propositions have a special role for us and also
that since they cannot be doubted, they do not stand in need of
justification. They are outside the game of “giving and asking
for reasons”. How this fundamental idea is worked out in detail
in On Certainty is, as mentioned, a matter of dispute.20 How-
ever, the fundamental idea, independently of the precise pedi-
gree it takes at the time of On Certainty, is in itself enough to
discard the meaninfulness of external world-scepticism. I will
now try to show that this connection is already made in PI.
In the Philosophical Investigations, Wittgenstein discusses
the notions of doubt and knowledge in relation to pain:
“246. In what sense are my sensations private?—Well, only I
can know whether I am really in pain; another person can only
surmise it.—In one way this is wrong, and in another non-
sense. If we are using the word ‘to know’ as it is normally used
(and how else are we to use it?), then other people very often
know when I am in pain.—Yes, but all the same not with the
same certainty with which I know it myself!—It can’t be said
of me at all (except perhaps as a joke) that I know I am in pain.
What is it supposed to mean—except perhaps that I am in pain?
Other people cannot be said to learn of my sensations only
from my behaviour—for I cannot be said to learn of them. I
The truth is: it makes sense to say about other people that
they doubt whether I am in pain; but not to say it about myself”.
Let us examine this key passage carefully. The first thing to
point out is that Wittgenstein is not denying that sensations are
private, but rather the sense in which they can be said to be
private. They are not private in the sense that only the one who
feels the pain can “really” know that he’s in pain (the “really”
should be a red herring here—Wittgenstein seems to take it as a
pointer that we’re talking metaphysically, that language is on
vacation). In fact, it’s the other way round: other people can
know that someone else is in pain, and the one who is in pain
cannot, usually, be meaningfully said to know it. Wittgenstein
makes it clear, in the last sentence of the remark, that the reason
why the one who is in pain cannot be said to know it is because
he cannot doubt it. In these remarks of the Investigations the
crucial connection between knowledge and the possibility of
doubt is made. It is then natural to apply the fundamental idea
to “I have a hand” and “there is an external world”. Indeed, that
connection is repeatedly reminded to us in On Certainty. In the
first two remarks of that book discussed in the first section of
this paper, Wittgenstein makes it in relation to Moore’s
“knowledge” that he has a hand: it is said there, in essence, that
one cannot know that one has a hand because it does not make
sense for one to doubt it. In On Certainty, Wittgenstein does
not primarily try to convince us that we cannot doubt such
propositions on the basis that we cannot be said to know th em—
that is, he does not try to convince us of their special status: he
18See Coliva (2010) and Wr i ght (1985) f or this kind of in t erpretation.
19See Williams (1991) for this kind of interpretation.
20If my readin g of OC is corr ect, then the elu cidat ion of the stat us of corner -
stones is what the book is all about .
17McDowell stresses these passages in his own perspective on scepticism,
resting on his version of a direct realism (see in particular Mind & World,
page 27, where he invokes these remarks).
Copyright © 2012 SciRes.
is rather interested in uncovering the nature of that special
status and the role it plays in our form of discourse (more on
this in the last section of this paper).
In terms of the I-II-III template, once propositions such as “I
have a hand” and, even more obviously, “there is an external
world”, are shown to be outside the game of “giving and asking
for reasons” because they cannot be doubted, external world-
scepticism cannot get off the ground. Indeed, this kind of scep-
ticism requires that such propositions be just like any other
empirical propositions since it rests on the possibility of asking
meaningfully: “on what ground can one be said to know that
there is an external world or that one has a hand?”. Showing
that type-III propositions and some type-II propositions cannot
neither be doubted nor known implies that there is no sceptical
problem of trying to establish proper justification for type-III
propositions such as “there is an external world”. The key is to
see that the connection between knowledge and the possibility
of doubt is already there in the Investigations—in fact, it is
already there in the Blue Book.21
Let us briefly recapitulate: in the Philosophical Investiga-
tions, Wittgenstein has all the elements he will ever need to
show that scepticism is a pseudo-problem. The driving theme
of this masterpiece—the deconstruction, so to speak, of the
Cartesian picture of the mind—leads Wittgenstein to discuss a
cluster of issues that are directly related to scepticism as tradi-
tionally conceived, namely the issues of privacy, of under-
standing, and of the role of mental images. These considera-
tions lead to a recognition that knowledge-scepticism is not a
philosophical problem. In that book, Wittgenstein also makes
the crucial connection between knowledge and the possibility
of doubt which is fundamental for dealing with external world-
scepticism. It would be, in my opinion, irresponsible to ignore
how these themes of the Philosophical Investigations resonate
with scepticism. But if they do resonate in this manner, then is
it plausible at all to hold that the main thrust of On Certainty is
to revisit scepticism, the difference merely being, presumably,
that he does so through a different set of considerations?
The Transcendental Reading
The second reason why On Certainty should not be read as a
work whose primary target is scepticism in the aforementioned
sense is the following. Interpreting On Certainty along the lines
of the standard reading makes it very tempting to see it as put-
ting forward a transcendental argument against scepticism. The
reason is this: if On Certainty is taken to put forward the view
that there are some propositions that cannot be doubted in vir-
tue of their special role in our conceptual scheme, and if this
claim is taken to be made against the sceptic’s contention that
any proposition can be doubted, then Wittgenstein is here de-
fending a view about the nature of language, about its essence
and its conditions of possibility. This is the case if we reflect on
the dialectic underlying the work according to the standard
interpretation: the sceptic argues that any empirical proposi-
tion—type-II or type-III—can be doubted, hence that knowl-
edge is impossible. Wittgenstein, on this interpretation, set
down to work and tried to prove that this is not the case by
writing the remarks contained in On Certainty. The objective
was thus to show that contrary to what the sceptic claims, there
are some propositions that are immune from doubt. If Wittgen-
stein can make a case that such propositions exist, he will have
revealed scepticism as false (or nonsensical, depending on the
version of the favoured reading). In On Certainty (always ac-
cording to the interpretation under study), he shows the exis-
tence of such propositions by pointing out that our cognitive
endeavours—more broadly the language game of giving and
asking for reasons—depend for their very possibility on there
being such privileged propositions, immune from doubt and
outside the arena of giving and asking for reasons. Thus it is a
condition of possibility of language, or thought, that there are
such propositions. Since thought exists (since the language
game in question exists), and since its existence requires that
some propositions be exempt from doubt, then scepticism is
false (or nonsensical).
The problem is that this line of thought attributed to Witt-
genstein has the form of a transcendental argument.22 Its aim is
to identify essential features of language and thought by start-
ing off with general, accepted facts about language and th-
ought—this being followed by an examination of what makes
these facts possible, that is, an inquiry into their conditions of
possibility. However, it is not a very controversial exegetical
claim to affirm that one of the main feature of Wittgenstein’s
philosophy of the ‘second’ period is that it is resolutely anti-
transcendental and anti-essentialist. In the Tractatus, he was
under the spell of essentialism and he was trying to find the
essence of logic (of language and thought). But in his later
work—notably the Blue Book (passim) and the Philosophical
Investigations (§§ 89-133)—he is pointing the way towards a
way of philosophising that does without such pretensions. The
goal of philosophy, for Wittgenstein, is to describe, not to say
anything “surprising”. A transcendental argument is not de-
signed to merely describe a phenomenon, but typically reaches
a substantial conclusion, and has a normative aspect—the result
being that we are compelled to accept the conclusion, however
surprising it can be or even counter-intuitive.
The third exegetical problem related to the standard reading I
would like to examine is the following. If we assume that
Wittgenstein is trying in On Certainty to solve a traditional
philosophical problem, then we are forced to dismiss a central
claim of his later philosophy—his quietism. It is uncontrover-
sial that Wittgenstein would conceive of scepticism as being a
traditional philosophical problem generated by a misuse of
grammar—generated, that is, by an inappropriate use of ex-
pressions containing the concepts of knowledge, doubt, cer-
tainty and others. So there is no doubt that his position on the
nature of philosophy and of philosophical problems—as ex-
posed in the Philosophical Investigations §§ 89-133—pre-
cludes him from any attempt to address scepticism in the way
discussed in the first section of this paper.
21See the Blue Book, page 30, where Witgenstein discusses the use of an
expression such as “I know what I wish”: “Of course I know” could here be
replaced by “Of course there is no doubt” and this interpreted to mean “It
makes, in this case, no sense to talk of a doubt”. In this way the answer “O
course I know what I wish” can be interpreted to be a grammatical state-
ment.” Note here that Wittgenstein seems to be gesturing towards the
rule-like nature of such statements (but just gesturing—a fuller description
of how it works precisely is the topic of On Certainty).
22See Stern (2000), chapter 3, and especially pages 87-89. Stern is prudent in
that he does not seem to endorse the view that Wittgenstein is a “transcen-
dentalist”but he does acknowledge and discuss there the extent to which
Wittgenstein can be interpreted as putting forward a transcendental argu-
ment in On Certainty.
Copyright © 2012 SciRes. 265
The key concern, of course, is whether this is merely Witt-
genstein’s “official” position, to be contrasted with what he
actually does in his later work (including in the Investigations
and in On Certainty). Indeed, his celebrated “rule-following
considerations” and his claims about the nature of privacy and
self-knowledge have often been interpreted as revealing the
dysfunctional aspect of his later contributions: on the one hand,
we have Wittgenstein’s dramatic position on the nature of phi-
losophy—his therapeutic stance and his claim that traditional
philosophy is created by a misuse of grammar—which should
forbid him to engage in theoretical philosophising, and on the
other hand we have immensely influential and original “theo-
ries” about, e.g., the nature of the mind and about the most
pressing epistemological issue of modern times (scepticism).
This way of looking at the matter, however, is completely mis-
guided. This issue has been discussed quite extensively in the
contemporary literature. I agree with Crispin Wright’s analysis,
in which he resolves the tension by pointing out, with respect to
the Cartesian picture of the mind, that Wittgenstein’s criticism
of this picture is not to be seen as a bit of philosophical theo-
rising, but rather as a necessary step towards recognising it for
what it is, i.e., a “bad” picture resulting from misinterpretations
of the grammar of words and concepts belonging to a certain
family.23 It is crucial to remember that he does not replace the
Cartesian picture with another (better, more accurate) one.
Rather, Wittgenstein lets us see that there is nothing special
about the nature of the mind as far as philosophy is concerned,
there is no mystery and nothing to be explained, since every-
thing lays open to view: it’s a matter of looking at things in the
right way, not a matter of finding the ‘right’ theory about the
mind and intentionality.
The same can be said about his attitude towards scepticism.
That issue is not to be “solved” (because there is no meaningful
‘problem’ awaiting a solution). The mystery is not that we seem
unable to answer the sceptic—it’s rather that there is an ap-
pearance of mystery (the mystery of scepticism). The role of
the philosopher is to show that this mystery is a chimera. If this
is right—if Wittgenstein’s essential attitude towards scepticism
is to treat it as senseless (unsinnig)—then one upholding the
standard interpretation criticised here would thus be committed
to the claim that the main thrust of On Certainty is to reveal
scepticism as a pseudo-problem. But this seems, at best, an odd
thing to say—the remarks seem to do much more than that, and
if they do reveal scepticism as senseless, it’s rather as a by-
product of the work, not its essential thrust. We are indeed led
to see that scepticism is senseless, but this is something that the
reader of the Investigations already knows,24 thanks to the criti-
cism of the Cartesian picture of the mind.
To sum up, I have pointed out three broad and related exe-
getical reasons to the effect that Wittgenstein had already “dealt
with” scepticism prior to writing the bulk of On Certainty.
Maybe none of them is totally conclusive—I think they are, but
ultimately it’s for the reader to judge. However, the sheer
weight of these considerations, taken together, should at least
ignite a sparkle of doubt, so to speak, towards the standard
interpretation. If the latter is misguided, it should provoke the
possibility of an alternative reading of On Certainty, one that
sees it as essentially being about something else.
The Purpose of On Certainty
If On Certainty is not a work primarily directed at providing
a response to scepticism, then what is it? Considerations of
space will force me to be brief here but there are some impor-
tant points that can be addressed. There can be no doubt that
Wittgenstein is impressed by the special status of the proposi-
tions “known” with certainty by Moore. There can also be no
doubt that Wittgenstein is trying to defend the view according
to which we can’t be said to know these propositions, and that
this defence takes up a large portion of his resources in that
book. Finally, there is also no doubt that if we don’t entertain
such a cognitive relation to these propositions, scepticism finds
itself thoroughly shaken. But the primary purpose of doing so is
not to reveal scepticism as a pseudo-problem. That would be
(or so it seems to me) the equivalent of smashing a mosquito
with a brick—and a dead mosquito at that.
It would be more fruitful, and more plausible, to read On
Certainty as being a work pertaining to the human cognitive
condition.25 Cornerstone propositions play a special role in the
way we deal with the world and with each other. They form the
backbone of reasoning, of thought, of language, of argumenta-
tion. But what exactly is their status? Where do they belong in
our conceptual scheme? Where are they situated? These are the
questions—or so I contend—that had a strong hold on Wittgen-
stein when he worked on these remarks. He is trying to make
sense of these propositions, and by doing so he is trying to
make sense of our form of life, since the latter is, in good
measure, determined by the cornerstone propositions that form
the backbone of thought.
The recurrent themes of On Certainty fit well with the char-
acterisation of Wittgenstein’s objectives as suggested here. It is
not easy to determine a definitive list of the central themes of
the book, but the following three certainly play a crucial role:
The relation we entertain with cornerstone propositions—is
it epistemic or not? If not, what is that relation? Do we sim-
ply assume them? What does it mean to say that we “take
them on trust”, that we don’t question them but somehow
Cornerstone propositions have the form of empirical propo-
sitions, but they differ from “regular” empirical proposition s.
How to account for this contrast, how should we explain it?
There is a “feeling” of certainty attached to cornerstone
propositions. What is the source of this certainty, and how
is it manifested?
It seems uncontroversial that these questions are central ones
in On Certainty. They were prompted upon reading Moore’s “A
Defence of Common Sense”. These propositions struck Witt-
genstein as peculiar. He wanted to identify their peculiarity, and
by doing so he was exploring our human cognitive condition—
he was not merely, or primarily, trying to provide an answer to
a sceptical threat (one that he had already dealt with anyway).
His goal had a dignity (describing the human condition) that is
lacking when we characterise his writing as directed towards a
position (scepticism) that no one really takes seriously outside
the philosophical arena (in the sense of believing it to be true,
not in the sense that it’s not a “serious” and “pregnant” philoso-
25See for in stance Rh ees when h e holds that Wittg enstein thinks that corn er-
stone propositions “play a curious role in our speaking and thinking. An
investigation of this role (and that is what the remarks from the beginning to
the end of his book are) leads to a better understanding of human language,
thought, a n d l anguage-games” ( Rhees, 2003: p. 3).
23See the “Study Note on Wittgenstein” in his 2001.
24Something that even the reader of the Notebooks knows (se e epigraph )!
Copyright © 2012 SciRes.
Copyright © 2012 SciRes. 267
Wittgenstein’s exploration of our form of life did not take the
form of traditional philosophical theorising, but rather took, in
conformity with his claims about the nature of philosophy in
the Investigations and elsewhere, the descriptive form. He de-
scribes how, as a matter of fact, we use the concepts of knowl-
edge, certainty, mistake, and the like. It is a study of the gram-
mar of key cognitive concept s. There is nothing pr esc rip tive 27—
or normative—about his inquiry into the nature of these con-
cepts. It is more akin to a “grammatical anthropology”, a study
of how human beings use certain key words. The hope is that
such a study will provide an insight into our human cognitive
condition—or, to put it more cautiously, will command a “clear
view” of who we are qua cognitive beings. It is not a bit of
philosophical theorising, but it is nevertheless difficult and deep
because it is about what is somehow presupposed, unquestioned,
taken for granted. To borrow Robert Brandom’s expression, On
Certainty is an attempt to make explicit what is implicit in our
cognitive endeavours. This explicitness does not consist in
constructive theoretical philosophy, but rather in a thorough
description of crucial aspects of our form of life. It consists in
an investigation of the nature and place of cornerstone proposi-
tions. Rush Rhees, in his rich study of On Certainty, put it well
when he wrote (2003: p. 70) that cornerstones “underlie the
possibility of speaking. Underlie—how? This is the main theme
[of On Certainty]”.
From 1915 to his death in 1951, Wittgenstein held on to two
key theses in epistemology: firstly, that the problem of scepti-
cism is a pseudo-problem, and secondly that this is because
doubt does not always make sense. In his mature philosophy—
predominantly in the Philosophical Investigations—Wittgen-
stein gives grounds for both these claims. These grounds ad-
mittedly differ from those expounded in the Tractatus and in
the Notebooks. However, a close study of the exposition of the
nature of philosophy and the discussion of the issue of privacy
in the Investigations show that these standpoints towards the
traditional problem of scepticism were sufficiently established
by then to invalidate a reading of On Certainty that would see it
as merely another attempt to do just that. In fact, it does much
more, and it would be better read as a treatise about the explo-
ration of our human cognitive condition. That interpretation of
On Certainty is consistent with the claims of the previous sec-
tion, namely: it does not concern itself primarily with the prob-
lem of scepticism, it is not transcendental, and it does not con-
sist in theoretical, constructive philosophy. It is a reading of On
Certainty that makes it continuous with Wittgenstein’s mature
Cavell, S. (1979). The claim of reason. Oxford: Oxford University
Coliva, A. (2010). Moore and Wittgenstein. Basingstoke: Palgrave
Gascoigne, N. (2002). Scepticism. London: Acumen.
Hacker, P. M. S. (1986). Insight and illusion: Themes in the philosophy
of Wittgenstein (2nd ed.). Oxford: Clarendon Press.
Kerns, A. (2004). Understanding scepticism: Wittgenstein’s paradoxi-
cal reinterpretation of sceptical doubt. In D. McManus (Ed.), Wit-
tgenstein and Scepticism (pp. 200-217). London: Routle dge.
McGinn, M. (1989). Sense and certainty: A Dissolution of scepticism.
McGinn, M. (2004). The everyday alternative to scepticism: Cavell and
Wittgenstein on other minds. In D. McManus (Ed.), Wittgenstein and
Scepticism (pp. 240-259). London: Routledge.
Moore, G. E. (1925) A defense of common sense. In T. Baldwin (Ed.),
G. E. Moore: Selected writings (pp. 106-133). London: Routledge.
Moore, G. E. (1939) Proof of an external world. In T. Baldwin (Ed.), G.
E. Moore: Selected writings (pp. 147-170). London: Routledge.
Moyal-Sharrock, D. (2007). Unders t an d i ng Wittgenstein’s On Certainty.
Basingstoke: Palgrave Macmillan.
Moyal-Sharrock, D., & Brenner, W. H. (Eds.) (2007). Readings of
Wittgenstein’s On Certainty. Basingstoke: Palgrave Macmillan.
Rhees, R. (2003). Wittgenstein’s On Certainty: There—Like our life.
Oxford: Blackwell. doi:10.1002/9780470776247
Rorty, R. (1979). Philosophy and the mirror of nature. Princeton:
Princeton University Press.
Stern, R. (2000). Transcendental arguments and scepticism. Oxford:
Stroll, A. (1994). Moore and Wittgenstein On Certainty. Oxford: Ox-
ford University Press.
Stroud, B. (1984). The Significance of philosophical scepticism. Oxford:
Clarendon Press. doi:10.1093/0198247613.001.0001
Williams, M. (1991). Unnatural doubts. Oxford: Blackwell.
Wittgenstein, L. (1953). Philosophical investigations. Oxford: Black-
Wittgenstein, L. (1958). The blue and brown books. Oxford: Blackwell.
Wittgenstein, L. (1961). Notebooks 1914-1916. Oxford: Blackwell.
Wittgenstein, L. (1969). On Certainty. Oxford: Blackwell.
Wright, C. (1985). Facts and certainty. Proceedings of the British
Academy, 71, 429-472.
Wright, C. (2001). Study note on Wittgenstein. Rails to Infinity: Essays
on themes from Wittgenstein’s philosophical investigations (pp.
433-443). Har vard: Harvard University Press .
26It is surprising to notice that upon consulting the index of the book, one
will find mention of scepticism only the once (§37). This may seem to be a
superficial and certainly not sufficient reason to reject the targeted reading,
but nevertheless it is a surprising fact to realise that a work apparently de-
voted to criticise a position for which we have no other useful word than
“skepticism” should barely contain it or its cognates (“skeptical”, “sceptic”).
While I agree that Wittgenstein often fails to mention theses that he’s dis-
cussing , it seems that thi s cas e is un li ke, say , “b eh avi our ism” or “ rep res ent a-
tionalism” or “psychologism”. “Scepticism” is not a term of art like the
aforementioned. Discussing scepticism is different from discussing an
“-ism” in a typical philosophical conversation. Scepticism is not a “theory”
like behaviourism. It could be argued that to discuss scepticism isin fact to
engage into epistemology, into a whole field of philosophical study, one in
which many of these “-isms” can figure—reliabilism, internalism, fo und ati on -
alism and what not. In that sense, Wittgenstein would presumably have no
difficulty in using that term and its cognates if he felt that he had to. Seen in
that light, the fact that he barely used that terminology might be telling.
27I mean that Wittgenstein is not putting forward a thesis that will have
revisionary consequences on how we think, act, use words, etc. Not to be
confused with the idea that he is studying rules and that they have a prescrip-
Wright, C. (2002). (Anti-)sceptics simple and subtle: G. E. Moore and
John McDowell. Philosophy and Phenomenological Research, 65,
Wright, C. (2004a). Warrant for nothing (and foundations for free?).
Proceedings of theAristotelian Society Supplementary Volume, 78,
Wright, C. (2004b). Wittgensteinian certainties. In D. McManus (Ed.),
Wittgenstein and Scepticism (pp. 22-55). London: Routledge.
Wright, C. (2007). The perils of dogmatism. In S. Nuccetelli, & G.
Seay (Eds.), Themes from G. E. Moore: New Essays in Epistemology
and Ethics (pp. 25- 48). Oxford: Oxford University Press.