self- k now le d ge
a nna lisa c o liv a
Palgrave Innovations in Philosophy
Series Editors
Vincent Hendricks
Philosophy/MEF
University of Copenhagen
Copenhagen, Denmark
Duncan Pritchard
Department of Philosophy
University of Edinburgh
Edinburgh, UK
Palgrave Innovations in Philosophy is a new series of monographs. Each
book in the series will constitute the ‘new wave’ of philosophy, both in
terms of its topic and the research profile of the author. The books will be
concerned with exciting new research topics of particular contemporary
interest, and will include topics at the intersection of Philosophy and
other research areas. They will be written by up-and-coming young phi-
losophers who have already established a strong research profile and who
are clearly going to be leading researchers of the future. Each monograph
in this series will provide an overview of the research area in question
while at the same time significantly advancing the debate on this topic
and giving the reader a sense of where this debate might be heading next.
The books in the series would be of interest to researchers and advanced
students within philosophy and its neighboring scientific environments.
The Varieties of
Self-Knowledge
Annalisa Coliva
Department of Philosophy
Irvine, California, USA
v
Credits
vii
Acknowledgments
This book has developed over several years, since I first became interested
in the topic of self-knowledge during my stay at Columbia University
thanks to a Fulbright Research Fellowship and a Fellowship of the Italian
Academy at Columbia University in 2002–2004. I would like to thank
my host there, Akeel Bilgrami, for his encouragement as well as for regu-
lar conversations, at his place, on this and related topics. Thanks are due
also to Andreas Kemmerling, my host in Heidelberg, where I was visit-
ing in 2006–2007 under the auspices of an Alexander von Humboldt
Fellowship and where I further developed my ideas on the topic of self-
knowledge. I would also like to thank the participants at two workshops
on “The Self and Self-Knowledge” that I organised in Bigorio and at the
Institute of Philosophy in London, in 2004 and 2008, respectively, which
proved invaluable in shaping my ideas on this topic: Dorit Bar-On, Akeel
Bilgrami, John Campbell, Jane Heal, Conor McHugh, Lucy O’Brien,
Ted Parent, Christopher Peacocke, Eva Picardi, Jim Pryor, Ursula Renz,
Carol Rovane, Martine Nida-Rümelin, Barry Smith, Gianfranco Soldati,
Paul Snowdon and Crispin Wright. The workshops were made pos-
sible by the generous contributions of the Fonds National Suisse, the
Institute of Philosophy and the British Academy, which I hereby thank
for their support. Thanks are also due especially to Akeel Bilgrami, Alan
Millar and Margaret Gilbert for their comments on my work on Moore’s
ix
x Acknowledgments
xi
Contents
1 Introduction 1
Bibliography 18
xiii
xiv Contents
3 Varieties of Self-knowledge 51
1 First-personal Self-knowledge 52
1.1 Groundlessness 52
1.2 Transparency 58
1.3 Authority 62
2 Counterexamples from Content Externalism and
Cognitive Sciences? 67
3 Third-personal Self-knowledge 69
4 Summary 74
Bibliography 75
Index 279
1
Introduction
The main and novel idea presented in this book is that self-knowledge—
that is, our knowledge of our own mental states—comes in many ways.
We have first-personal knowledge of our own mental states when, for
instance, we are immediately aware of our occurrent sensations. By con-
trast, we have third-personal knowledge when, for example, we realise
that we enjoy a given mental state by reflecting on our behaviour and by
inferring to its likely cause. Even when distinctively first-personal knowl-
edge is at stake, it must be kept in mind that we have a variety of mental
states. For instance, we enjoy sensations, such as pains and tickles, which
have a characteristic phenomenology, but also perceptions that have both
a phenomenal and a representational content; we have propositional atti-
tudes, such as beliefs, desires and intentions, and these come in various
fashions—that is, as dispositions and as commitments—hence, as the
result of one’s own deliberations based on considering evidence for or
against a given proposition or course of action. Finally, we enjoy emo-
tions, whose nature still escapes philosophical consensus. Such a variety
of mental states invites caution in propounding single, all-encompassing
accounts of how we may know each of these types of mental state. In par-
ticular, although it is clear that sensations and at least some emotions have
1
A lot of them start out by presenting themselves as offering accounts of our knowledge of just this
or that kind of mental state.
2
A notable exception is Boyle 2009, 2011a. His kind of pluralism, however, is more limited than
the one defended in this book, for he mainly stresses the difference between first-personal self-
knowledge of propositional attitudes as commitments and of one’s passive mental states, such as
sensations and perceptions. Furthermore, he thinks that knowledge of our own beliefs is more
fundamental than any other kind of first-personal self-knowledge. No such priority claim is
defended in this book.
3
Several authors whose views will be considered in the following chapters are subject to this criti-
cism, as we shall see.
4
Gopnik 1983 and Cassam 2014 are a case in point.
4 A. Coliva
5
Most theorists whose views will be discussed in the following chapters do that.
6
Cassam 2014, who defends inferentialism, is a case in point. For a critical assessment, see
Coliva 2015a.
1 Introduction 5
authority and groundlessness are not contingent but necessary and a priori
aspects of what goes by the name of first-personal self-knowledge. For mas-
sive failures at this kind of self-knowledge would display either the lack of
the relevant psychological concepts or failures at rationality. Rationality, in
this connection, has to be understood in a “thick”, rather than in a “thin”,
sense. The latter amounts to the idea that we are critical reasoners insofar
as we revise our propositional attitudes and goals on the basis of counter-
vailing reasons. However, I agree with several philosophers (Christopher
Peacocke, Dorit Bar-On and Quassim Cassam, just to mention a few)
who, contra Sydney Shoemaker and Tyler Burge, do not think that self-
knowledge is necessary for being critical reasoners. If that is the notion of
rationality one has in mind, then lack of self-knowledge will not make one
necessarily irrational. Yet we also have a thick notion of rationality, accord-
ing to which making certain psychological self-ascriptions and behaving in
ways which run systematically against them would impugn the idea that
we are confronted with a normal subject, up to the point of rendering her
pronouncements onto herself irrelevant, a mere flatus vocis devoid of any
significance, if not of meaning altogether. These characteristic traits of first-
personal self-knowledge will also be defended against possible objections
stemming from recent findings in cognitive sciences and from scepticism
about knowledge of the content of our own propositional attitudes deriv-
ing from the endorsement of semantic externalism. For instance, several
studies in cognitive science tend to show that we do not have knowledge of
our own character traits, that we are bad at affective forecasting—that is, at
figuring out how we would actually feel if some relevant change happened
to our lives—and, finally, that we are really poor at identifying the causes
of our decisions and further behaviour. None of this, however, shows that
we never have essentially first-personal self-knowledge. Rather, it shows
that its scope is limited and does not extend to our deep-seated and future
dispositions, or to the causal relations among our various mental states,
which are known, if and when they are, in a third-personal way. Yet all this
is compatible with the fact that we have essentially first-personal knowl-
edge of a wide range of mental states, such as our sensations, perceptions,
basic emotions and propositional attitudes as commitments.
Furthermore, some theorists take the rise of content externalism to be
incompatible with at least authoritative self-knowledge regarding one’s
1 Introduction 7
of overt behaviour (or other “inner promptings”). Her views have given
rise to a heated debate, at the interface of philosophy of mind, psychol-
ogy and neuroscience, between supporters of the theory-theory approach
and partisans of so-called “simulation” theories, such as Alvin Goldman
and Robert Gordon. According to simulation theorists, who are other-
wise divided on many issues, knowledge of other people’s mental states
is not based on the application of a theory but on the simulation of the
other person’s point of view, which gives rise to a psychological ascription
based on what one oneself would feel and think if one were in the other
person’s shoes. These views are exposed and critically examined. The main
objection against the inferentialist account is that it implausibly assimi-
lates first-personal self-knowledge to knowledge of other people’s mental
states. Furthermore, it runs the risk of providing a circular account of self-
knowledge and it succumbs as soon as one tries, like in Quassim Cassam’s
recent version of it, to make it transcend its proper domain of applica-
tion. The main criticism against simulation theories, in contrast, is that
they are in fact unclear about how we would get knowledge of our own
minds, on the basis of which we should then gain knowledge of other
people’s mental states, and risk falling back onto other, problematical
models of self-knowledge (such as the inner-sense model). Simulation
theorists, in particular Gordon, also have interesting but underdevel-
oped views about the nature and acquisition of psychological concepts,
such as the concept of belief and of other propositional attitudes. Still,
both inferentialism and simulative accounts have important things to say
about some instances of third-personal self-knowledge, such as knowl-
edge of our deep-seated dispositions and the kind of self-knowledge we
can gain through affective forecasting.
In the fifth chapter, called “Epistemically Weak Accounts”, we intro-
duce and assess various models, which are united in claiming that
self-knowledge is indeed a kind of modest, yet genuinely cognitive
achievement, while trying to avoid construing self-knowledge either as
due to the operations of an inner scanning mechanism or as a form of
inferential knowledge. For this reason, they are called “epistemic” mod-
els. They come in various fashions, however. Some of them can be traced
back to some remarks by Gareth Evans in The Varieties of Reference.
According to Evans, in order to know our own beliefs, we need only
1 Introduction 9
7
For a criticism of Byrne’s development of Evans’s ideas, see Boyle 2011a.
10 A. Coliva
8
Heal 2002 too defends this position, but our exposition will focus on Wright’s more thorough
account.
9
That is to say, by being drilled to substitute their immediate avowal, “P”, “P would be good to
have”, “I will φ”, with the corresponding psychological one—that is, “I believe that P”, “I want/
desire that P”, “I intend to φ”.
1 Introduction 13
mind. However, only in one’s own case can the inference be based on rel-
evant inner promptings, such as sensations, emotions and further mental
states. In some other cases, instead, it can depend on deploying simulative
methods, like when we engage in affective forecasting. Sometimes, we gain
knowledge of our minds by relying on other people's judgements about
us. Our self-knowledge is therefore achieved though testimony. Finally,
knowledge of our dispositional mental states can be obtained by means of
the self-conscious deployment of highly dispositional psychological con-
cepts. In this case, there is inferential reasoning going on, but it is not a
kind of inference to the best explanation. Rather, it consists in subsuming
some aspects of one’s overall behaviour and mental states under a concept
by self-consciously exploiting its characteristic notes.
Strong constitutive accounts have limited purchase also because, con-
trary to what some of their supporters hold, they do not extend to past self-
ascriptions of propositional attitudes as commitments, which are known,
when they are, on the basis of mnestic evidence. Still, it is true that being
able to remember one’s past mental actions, or indeed other mental states,
as well as one’s own past actions, is constitutive of being a cognitively well-
functioning human being. Yet that does not mean that we can account for
our knowledge of these past mental states along constitutivist lines.
Moreover, strong constitutive accounts are not apt to explain self-
knowledge of our sensations and of other mental states that have a dis-
tinctive phenomenology and that are clearly independent of our ability to
self-ascribe them,10 such as bodily sensations, basic emotions, perceptions
and perceptual experiences. Here, the most promising account will have
to forsake the second metaphysical claim at the heart of strong constitu-
tive explanations, according to which psychological self-ascriptions can
at least partially constitute the first-order mental states they ascribe to a
subject. What remains are simply the other characteristic claims of con-
stitutive positions, according to which conceptually competent creatures
are authoritative, at least in the normal run of cases, with respect to their
own mental states and are immediately in a position to self-ascribe them
without observing either their own mental states or their overt behav-
iour. These first-order mental states, however, can exist independently
10
Pace McDowell 1994.
1 Introduction 15
this little theory may often take place in rapid and almost unnoticeable
ways but only because we are already proficient in applying it. Indeed,
genealogically or in new, unexpected cases, it will require time and effort
and possibly help from a third party. For we may well be at a loss about
how to interpret the pool of data about ourselves we may have collected.
That is to say, we may need the intervention of another person to be in a
position to infer that our characteristic feelings and behaviour are signs of
love or envy. Moreover, a lot of our third-personal self-knowledge, such
as affective forecasting or knowledge of our deep-seated dispositions, will
depend on simulating relevant aspects of a given situation to see how we
would react to it, thereby acquiring some insight into our own nature
and character. Reading novels and watching movies can achieve similar
results insofar as we may identify with the protagonists or be prompted to
simulate salient aspects of the plot to see how we would react if we found
ourselves in those situations.
Finally, it should be stressed that, contrary to the kind of self-verifying
self-ascriptions that have commitments as contents, in all cases in which
psychological self-ascriptions substitute more instinctive forms of behav-
iour, there is, however, limited room for error. Owing to slips of the
tongue or to somewhat impaired cognitive conditions, a subject could
actually voice sensations, basic emotions or perceptions she is not actually
enjoying. Yet constitutivism can take care of these possibilities by appro-
priately specifying the relevant C-conditions. By contrast, when the self-
ascription of dispositions or of non-basic emotions is at stake, there is no
default presumption that a subject should be authoritative with respect
to them. For she will be as exposed to error as she would be if she were
applying her psychological theory in order to get knowledge of another
person’s mental states.
At least since Shoemaker’s work, an account of self-knowledge has been
taken to have a bearing on the perplexing yet fundamental phenomenon
of Moore’s paradox—the paradox, that is, consisting in judging “P, but
I do not believe it” or “I believe that P, but it is not the case that P”.
Accordingly, in the Appendix, the proposed account of commitments and
their distinctively first-personal self-knowledge is brought to bear on it.
In particular, it is claimed that only by countenancing propositional atti-
tudes as commitments can Moore’s paradox so much as exist. By contrast,
1 Introduction 17
if one took its doxastic conjuncts to express (the lack of ) beliefs as dispo-
sitions, the paradox would, surprisingly, disappear. Indeed, the case of a
self-deceived subject who discovers her self-deception can perfectly well
illustrate the point. For one may find oneself in a position in which one
would coherently assert “I believe that my husband is unfaithful to me,
but he is not”; where the first conjunct expresses a disposition one has
found out by observing one’s own behaviour and by inferring to its likely
cause, and the second conjunct expresses one’s belief as a commitment,
given one’s knowledge of one’s spouse’s loyal behaviour. By contrast, it
would seem that if, by uttering (or judging) that very sentence, one were
trying, through its first conjunct, to express a commitment, its second
conjunct would actually undo it. This, in fact, would generate a Moorean
paradox. The interesting and novel result is that the existence of Moore’s
paradox can be secured only by countenancing essentially normative
mental states such as commitments.
Hence, to conclude: what goes by the name of “self-knowledge” is a
blend of disparate factors. Sometimes psychological self-ascriptions actu-
ally constitute the corresponding first-order mental states and although
one cannot fail to “know” them, it is not because one entertains a par-
ticular epistemic relation to one’s first-order mental states. Rather, it is
because the self-ascription brings them about and therefore is necessar-
ily authoritative. Some other time, our psychological self-ascriptions are
alternative ways of giving expression to mental states, which can exist
independently of them, resulting from being drilled to substitute their
immediate expression with the relevant linguistic behaviour. Still, under
appropriately specified C-conditions, being in a position immediately to
self-ascribe them and being correct in one’s self-ascription are guaran-
teed to hold a priori and as a matter of conceptual necessity. Finally, in
many cases, self-knowledge is actually the result of the application to
one’s own case of a little psychological theory or of simulative strategies
or, indeed, of an inferential deployment of highly dispositional psycho-
logical concepts, or may depend on testimonial evidence. Only in these
latter cases would self-knowledge be the result of some kind of cognitive
achievement and the term “knowledge” would, accordingly, express an
epistemic relation between a subject and her own mental states. In all
other cases, by contrast, the term “knowledge” would signal rather the
18 A. Coliva
fact that there is no room for error, when self-verifying self-ascriptions are
at stake, or at least not in the normal run of cases, when we are dealing
with self-ascriptions of sensations, basic emotions, perceptions and per-
ceptual experiences. Either way, self-knowledge is valuable either because
of its constitutive links with (“thick”) rationality, concepts’ possession,
and, at least in some cases, responsible agency, or because it can help us
have a better, more integrated and unitary life. Small surprise, then, that
Western philosophy since its inception appropriated the dictum of the
oracle of Delphi, “Know thyself ”.11
Bibliography
Boyle, M. (2009). Two kinds of self-knowledge. Philosophy and Phenomenological
Research, 77(1), 133–164.
Boyle, M. (2011a). Transparent self-knowledge. Aristotelian Society Supplementary
Volume, 85(1), 223–241.
Cassam, Q. (2014). Self-knowledge for humans. Oxford: Oxford University Press.
Coliva, A. (2015a). Review of Quassim Cassam Self-knowledge for humans,
Analysis, doi: 10.1093/analys/anv078.
Gopnik, A. (1983). How we know our minds: The illusion of first-person
knowledge of intentionality. Behavioral and Brain Sciences, 16, 1–15.
Reprinted in Goldman, A. (ed) (1993). Readings in philosophy and cognitive
science. Cambridge MA: MIT Press.
Heal, J. (2002). First person authority, Proceedings of the Aristotelian Society,
102(1), 1–19. Reprinted in (2003). Mind, Reason and Imagination (pp. 273–
288). Cambridge: Cambridge University Press.Please provide page range for
Heal (2002). 1–19
McDowell, J. (1994). Mind and world. Cambridge MA: Harvard University
Press.
11
According to the style of Palgrave Innovations in Philosophy series, while retaining the ambition of
presenting a novel account, the volume also contains a detailed discussion of some prominent
contemporary theories of self-knowledge. However, it does not address the issue of the first person,
even though having a concept of oneself is necessary in order to make psychological self-ascriptions.
The topic is extremely complex and will deserve a volume in its own right. I hope to be able to write
it before too long.
2
Varieties of Mental States
is a planet orbiting around the Sun, in between the Earth and Jupiter, or
that there is a black hole in a given portion of the universe. Even if the
objects of such knowledge are truths about physical entities, the ways
in which we do know them are clearly different—ranging from simply
observational to instrumental and robustly inferential. Equally, realising
the variety of mental states we enjoy will familiarise ourselves with the
idea that the ways in which we know them may be substantially different.
1 Perceptions and Sensations
Throughout their lives, human beings usually enjoy sensations such
as pains and tickles, on the one hand, and perceptions like seeing that
there is a book on the desk or hearing that one’s child is crying, on the
other. Is there a substantial difference between sensations and percep-
tions? According to Tyler Burge’s account of perception (2010), which
is the best developed one to date, the crucial difference between them is
that only perceptions are objective representations of the physical world
around a perceiver. That is to say, only perceptions are representations
of external objects, shapes, colours, and so on and as of objects, shapes
and colours out there—that is to say, as not mere variations in proximal
stimuli. It is therefore useful to present his views in some detail.1
According to Burge, in order to answer constitutive questions about
when objectivity starts, two main methodological assumptions should be
made. The first one is a firmly anti-individualist conception of the mind.
The second one consists in deploying an empirically informed methodol-
ogy. Accordingly, Burge rejects reductionism of the mental to the physical
and looks at the best and most mature science concerning, particularly,
perception to see, firstly, what conceptual categories it makes use of and,
secondly, what kind of data it provides us with regarding the issue of when,
at the ontogenetic and phylogenetic levels, objective representation starts.
1
In what follows, I will not dwell on those aspects of Burge’s overall proposal I am not entirely
happy with, such as the notion of a priori knowledge he seems to subscribe to, with the result that
we would know a priori certain natural norms regarding perception, or his account of smell and
taste according to which these senses do not afford perceptions but merely sensations. For a presen-
tation and a criticism of these aspects of his account, see Coliva 2012a.
2 Varieties of Mental States 21
The crucial issue addressed by Burge is what it means to say that per-
ception affords an objective representation. “Objective” as used here
connotes being a product of objectification, which “is formation of a
state with a representational content that is as of a subject matter beyond
2 Varieties of Mental States 23
2
In this respect, Burge’s account of the content of perception is similar to Christopher Peacocke’s
scenario content. Cf. Peacocke 1992, Chap. 3.
2 Varieties of Mental States 25
cation (…) the broader type of perceptual beliefs ultimately relies on con-
ceptualisations of basic perceptual attributives” (ibid.). In propositional
thought about perceived particulars, the singular elements are inherited
from perception and embedded in an inferential structure, which may
involve also quantificational elements, although it need not do so.3
Sensory states are different from perceptual ones in that they are not
objective. That is to say, they do not represent external elements of real-
ity as such. Take smell, for instance. In many cases, it seems to afford a
manifold of stimuli without attributing them to some relevantly stable
distal cause. For such a reason, sensory states do not have veridicality
conditions. They are not correct or incorrect, true or false representations
of something out there. They are merely subjective variations in proximal
stimuli.
Sensory states, however, are not necessarily sensations. That is to say,
there need not be a subjectively conscious phenomenal aspect to them.
A creature may be hard-wired such that her sensory systems may register
variations in temperature or pressure on her skin, so as to give rise to cer-
tain bodily movements, say, without her being conscious or experiencing
a sensation of cold or heat or of increased or decreased bodily pressure.
Similarly, perceptions, according to Burge, need not involve a conscious,
phenomenal element, even if they are objective representations with
veridicality conditions. Therefore, blind sight, for instance, would be a
case of perception. Of course, this is not to say that this conscious, phe-
nomenal aspect is not present in many, or even most, human perceptions.
3
Burge compellingly criticises Elisabeth Spelke’s claim that bodies are not represented as such in
perception. Moreover, he convincingly argues that cohesion, solidity, boundedness and spatio-
temporal continuity are properties, which can be represented as such in perception. According to
Burge, the ability to discriminate three-dimensional figures from a background and to represent
them as cohesive and bounded, together with the ability to track objects perceptually over time
(although not necessarily in motion or behind occlusions), is “constitutively necessary to visually
representing bodies as such” (2010, pp. 456, 458–459). By contrast, he thinks that perceptual
attributions of solidity are not necessary to that end, even if they are sufficient for it. Notice, more-
over, that, according to Burge, the ability to perceive bodies as such is not necessary for objective
perceptual representation, although it is central to the development of our conceptual system.
26 A. Coliva
tomorrow it will rain, unless one has the concepts necessary to entertain
that proposition.4
Subjects also orchestrate thoughts in inferences. Hence, if one believes
that all humans are mortal and that Socrates is human, one will believe
that he is mortal. If one believes that tomorrow it will rain and one
wishes to avoid getting wet, (ceteris paribus) one will go out only with an
umbrella. Practical inferences are characteristically sensitive not only to
the contents of one’s thoughts but also to the attitude one has towards
them. For if, instead of wishing not to get wet, one desired it, one’s actions
ought rationally to be different.
Recourse to subjects’ attitudes is routinely made in order to explain
and predict their behaviour. For instance, I can explain why Anna voted
for a given party by reference to her convictions and desires for herself
and the society she lives in. Alternatively, knowing her convictions and
desires, I can predict how she will vote at the next general elections.
4
It is a further issue, which will not concern us here, whether those concepts are the constituents of
the proposition itself, or whether the latter consists merely of objects, properties and relations and
whether concepts are needed merely to entertain it in thought, or finally whether it is a set of pos-
sible worlds—that is, all and only those worlds in which it will rain tomorrow in a given place—or
are indeed unstructured abstract entities, which can be entertained just in case one has the relevant
concepts.
Similarly, it is a further issue, which we will not address in this volume, which conditions must be
satisfied in order for a subject to possess the relevant concepts. For an overview and critical assess-
ment of contemporary theories of concepts, see Coliva 2006.
5
Bilgrami (2006, Chap. 5) distinguishes between mental states as “dispositions” and as “commit-
ments”. Scanlon 1998 and Moran 2001 between “brute” or “non-judgement sensitive” and “judge-
ment-sensitive” mental states.
28 A. Coliva
(a) these mental states are not the result of a conscious deliberation, like
a judgement, on a subject’s part, based on considering and, in par-
ticular, on assessing (or even being able to assess) evidence in favour of
P (or of P is worth pursuing, it would be good if P happened, etc.);
(b) these mental states are not within one’s direct control, being rather
something one finds oneself saddled with;
(c) hence, these mental states are not something one will be held ratio-
nally responsible for.
Some examples of mental states that will satisfy these conditions are
(i) mental states that are not formed by having assessed evidence in favour
of P (in the case of beliefs) or of P would be good to have (in the case of
desires, intentions and hopes). Brute urges and needs—insofar as some
of them may (debatably) be considered propositional attitudes—would
fall into this category, but also those propositional attitudes that might
be ascribed to a-conceptual creatures to make sense of their intelligent
behaviour and that, while responsive to some evidence, are not dependent
2 Varieties of Mental States 29
on its appraisal. For, ex hypothesi, these creatures do not have the concepts
necessary to grasp those contents, let alone the ones needed to assess the
evidence in their favour.6
(ii) Mental states that are attributed to subjects to make sense of their
behaviour, of which they themselves may be entirely ignorant. The lat-
ter class of mental states may comprise, but is not exhausted by, uncon-
scious mental states of a Freudian kind, which, however, can be operative
in shaping a subject’s behaviour.7 The idea here is that a subject may
form these beliefs and desires as dispositions, as reactions to experiences
undergone in early infancy and be entirely oblivious to them, while they
do shape much of her overt behaviour. This is not to say that she may
not acquire knowledge of them through therapy, say. Clearly, however,
that would be a case of third-personal self-knowledge. Deep-seated biases
would be another case in point, like gender preferences in offering certain
kinds of job preferably to male (or female) candidates. Again, one may
act on the basis of such a bias and could eventually recognise it, but if
one did recognise it, the first-order mental state, if still in place, would
remain dispositional, at least in the normal run of cases. Indeed, one may
continue to act on its basis while sincerely judging the opposite.8
(iii) Also several propositional attitudes that are neither biases nor
Freudian mental states but are self-attributed on the basis of an act of self-
interpretation, by finding them out through the observation of one’s own
behaviour and other immediately self-known mental states, will fall into
this category. For self-interpretation, when successful, makes one aware
of a mental state that is already there yet is not “one’s making” but rather
something one finds oneself saddled with. A nice example, though not
a case of propositional attitude, is provided by Jane Austen in her novel
Emma, when the protagonist finds out about her love for Mr. Knightley,
6
This makes it disputable that they could have the relevant beliefs as well, if those depended on
having the concepts necessary to grasp the propositions which constitute their contents. If one were
in the grip of such preoccupations, then a-conceptual creatures could at least be granted with
proto-beliefs, desires and intentions. See Dummett (1996, Chap. 12).
7
There may also be mental states which are attributed from a third party to make sense of a subject’s
behaviour, which are unconscious yet are not of a Freudian nature. The example discussed in (iii)
would be a case in point if, instead of being self-ascribed, the mental state were ascribed by another
person.
8
We will discuss this possibility in the context of our treatment of self-deception.
30 A. Coliva
Emma’s eyes were instantly withdrawn; and she sat silently meditating in a
fixed attitude, for a few minutes. A few minutes were sufficient for making
her acquainted with her own heart. A mind like hers, once opening to
suspicion, made rapid progress. She touched—she admitted—she acknowl-
edged the whole truth. Why was it so much the worse that Harriet should
be in love with Mr. Knightley than with Mr. Churchill? Why was the evil
so dreadfully increased by Harriet’s having some hope of return? It darted
through her, with the speed of an arrow, that Mr. Knightley must marry
no-one but herself.
9
The example is presented and discussed in Wright 1998, pp. 15–16, borrowed from Tanney 1996.
Analogous examples could easily be construed for the case of propositional attitudes. Giorgio Volpe
has kindly pointed out to me that also Schopenhauer, in On Freedom of the Human Will, holds the
view that a person’s character traits are known to her through reflection and inference on her past
behaviour.
2 Varieties of Mental States 31
(a’) that they are the result of an action—the mental action of judging
that P is the case (or worth pursuing/having)—on the subject’s part,
10
Cf. Bilgrami 2006, p. 213; Scanlon 1998, Chap. 1; Moran 2001, p. 116.
11
Bilgrami makes extensive use of the term; Robert Brandom too, although he is more interested
in stressing the social dimension of commitments than the former (or indeed myself ). Furthermore,
it is not my contention, somehow built in to the very notion of a commitment, that one should
have knowledge of all the logical consequences of one’s own beliefs and further propositional atti-
tudes. As Bilgrami points out (2006, pp. 371–372, fn. 7, but see also pp. 376–377, fn. 20), the
origin of the use of this term to refer to intentional states (or at least to a class of them) goes back
to Isaac Levi.
32 A. Coliva
12
This is the main difference between the present account of commitments and Bilgrami’s. For, in
his view, commitments are not dependent on a subject’s judgement.
13
One may even hold that they are intrinsically normative and not merely—as it were, externally—
constrained by normative principles. This is indeed the view that I favour and that will be put to
use in the diagnosis of Moore’s paradox (see Appendix). There is no need to take a stance on it at
this stage, though, for the less committal view would still do, in order to mark out propositional
attitudes as commitments from propositional attitudes as dispositions.
14
This is the constraint Bilgrami identifies as essential to commitments, from which, on his view,
(b’) and (c’) follow. However, he gives a moral or evaluative twist that it is best to resist. For, in his
view, would one be held not only rationally responsible for one’s commitments but also accountable
at large. For instance, one might be reproached or resented for having certain commitments (Cf.
Bilgrami 2006, p. 226). However, specified in the way Bilgrami characterises it, (d’) is not sufficient
to mark out the contrast between commitments and dispositions, because one can criticise or be
criticised, and accept to be criticised, also (for) one’s own dispositions, such as the disposition to
smoke, or, to take a more loaded example, for wanting to get rid of other male opponents as a result
of an unresolved Oedipus complex. But, surely, neither mental state is the result of a subject’s
action, for which one can be held rationally responsible, although one may be considered “badly”—
in Bilgrami’s extended sense of the term—for having it. It is then not by chance that, as a matter of
fact, Bilgrami ends up endorsing the view that “we do have transparent self-knowledge of mental
dispositions” (Bilgrami 2006, p. 287). I find this conclusion unpalatable, for, surely, when we do
get knowledge of our unconscious mental states we obtain it through a process of self-interpreta-
tion or of analysis (that may or may not be guided by a therapist) relevantly similar to the ways in
which we may come to attribute mental states to others. So, it seems to me that whatever knowl-
edge we may eventually gain of our unconscious mental states, it is not “transparent” and is actually
grounded in observation and inference.
2 Varieties of Mental States 33
15
Moran 2001, p. 120. Cf. Moran 2003 and Shoemaker 2003, Postscript.
2 Varieties of Mental States 35
16
Cf. Moran 2001, p. 116.
36 A. Coliva
17
“Credo quia absurdum” is usually attributed to Tertullian, even if he did not write it, and is often
associated with fideistic positions.
2 Varieties of Mental States 37
such “oughts” hold for drives, like the urge to smoke a cigarette after din-
ner, which will persist no matter what amount of counter-evidence will
be considered against the advisability of such a practice, or for brute dis-
positions, such as the disposition to form a certain thought upon hearing
a given word, tune, and so on. Hence, for instance, one may rationally
desire to provide one’s child with the best possible education. If one does
so, it will be for reasons, which, as such, may be further assessed. Were
it to turn out—quite implausibly—that countervailing considerations
should outweigh that desire, one should withhold from it, if rational.
Finally, it has to be registered that the distinction between proposi-
tional attitudes as commitments and as dispositions may be misleading,
insofar as it might suggest the idea that commitments exclude having
behavioural dispositions (at large). However, in the present understand-
ing of them, propositional attitudes as commitments will, in fact, be
accompanied by the relevant set of first-order dispositions, at least in
the normal run of cases. Hence, to exemplify with the case of beliefs,
the term “belief ” does not have “divided reference”, as Moran alleges
against Georges Rey, like the term “jade”, say.18 That is to say, it does not
pick out two different kinds of mental states, namely dispositions, on the
one hand, and commitments, on the other. Rather, there is considerable
overlap between beliefs as dispositions and as commitments to justify the
idea that they are more like two species of the same genus and therefore it
is no mere accident that the term “belief ” is, in fact, applied to both. For
instance, if I believe that P as a commitment, then I ought to be disposed
(ceteris paribus) to use P as a premise in a piece of practical or theoretical
reasoning, whether that reasoning is conscious or not. Hence, I should
have at least some of the same behavioural dispositions I would have if
I merely believed as a disposition that P. Similarly, in the case of a desire
as a commitment that P, I ought to be disposed (ceteris paribus) to seek
means to make P happen, just as I would do with a desire as a disposition.
Yet the aetiology, the modes of knowledge and some of the possible con-
sequences of having commitments and dispositions would be different.
However, only in the case of beliefs and other propositional attitudes
as commitments, if I did not actualise such first-order dispositions, I
18
Moran 2001, p. 87.
38 A. Coliva
3 Emotions
Besides having sensations such as heat and cold, perceptions like see-
ing a table in front of one, and propositional attitudes as dispositions
and commitments, human beings are capable of experiencing emotions,
such as fear of a barking dog in front of one, or love for one’s beloved,
or hatred towards a harming enemy, and so on. Although common sense
groups them separately from other mental states and has a specific name
for them, philosophical analyses have usually tried to reduce emotions to
19
This distinction will become crucial in our analysis of Moore’s paradox (see Appendix).
20
I may be reproached for having lied to him, of course, but this would be a moral judgement,
which does not change the fact that I did not have any personal commitment in the first place.
2 Varieties of Mental States 39
David Hume (1739–1740) and William James (1890) are usually consid-
ered to have held the view that emotions are sensations. Hence, the fear
of the barking dog in front of me would be a particular sensation. The
love for my beloved, a characteristic feeling and the hatred for a danger-
ous enemy another kind of feeling.
Various objections have been raised against such an account—for
instance, that it is not always the case that an emotion is accompanied
by a sensation. If love is an emotion, then my love for my children is not
a sensation. Rather, it is a complex mental state, which involves particu-
lar attitudes of care and attention. Similarly, if depression is an emotion,
then it is unclear what kind of sensation it involves. Rather, it seems to
display itself in characteristic attitudes towards people and events or even
life itself.
Moreover, jealousy and envy, for instance, may be akin inasmuch
as one’s feelings are involved, but they clearly differ insofar as the for-
mer concerns the attentions that one’s beloved reciprocates or displays
towards another person, whereas the latter concerns whatever another
person, not necessarily a beloved one, possesses and one would like to
have for oneself.
Sensations, moreover, are not representational, for they lack veridical-
ity conditions; nor are they objective, in the sense of being of objects and
properties represented as such, as we saw in §1.1. Even basic emotions,
such as fear, however, seem to have some form of representational con-
tent. For instance, my fear of the barking dog in front of me seems to
present that dog as dangerous to me and to have, on top of such a repre-
sentational content, a particular phenomenal aspect to it.
40 A. Coliva
Stoics are said to have held a completely different account of the emo-
tions, one recently upheld and developed by Martha Nussbaum (2001).
Accordingly, emotions are taken to be value judgements. Hence, for
instance, my fear of the barking dog in front of me would, in fact,
amount to the judgement, however implicit that might be, that the dog
is dangerous to me. My love for my beloved would be identical to the
judgement that he is important to me. Finally, my hatred for a dangerous
enemy would consist in the judgement that he is mischievous. Feelings,
when they are present, are added on top of the judgement but are not
constitutive of the emotion.
Surely, the proposal does well in accounting for the representational
elements of emotions. For they are taken to be judgements, hence propo-
sitional mental states, with genuine truth conditions. Yet it strikes as too
intellectualistic. After all, in order to make judgements, one needs con-
cepts, at the very least, if not also the ability to perform inferences and to
gather and assess evidence for or against their contents. Arguably, how-
ever, basic emotions are present in infants and animals. Yet it is dubious
that they would possess the relevant concepts, let alone make characteris-
tic inferences and weigh reasons for or against their contents.
Moreover, it is often the case that experiencing a certain emotion is
the basis for making a certain judgement. For instance, it is the fear I am
feeling of the barking dog in front of me which makes me judge (rightly
or wrongly) that it is dangerous. Furthermore, despite my fear, I can issue
a contrary judgement, if, for instance, I have independent and stronger
reasons to think that, after all, it is not dangerous. Yet, although normally
a better judgement overrides a worse one, I can still fear the dog, even if
I judge that, after all, it is not dangerous.
Perhaps the most recent account of emotions is the one due to Julien
Deonna and Fabrice Teroni (2012), who develop suggestions that can be
found in Edouard Claparède (1928). Accordingly, my fear of the barking
2 Varieties of Mental States 41
explanations of the sound just heard, are satisfied. Hence, the emotion
is there, yet knowledge of the complexity of the conditions on which
it would accurately track a real danger prompts the subject to consider
whether they are satisfied. Only when one has reasons to think they are
does the emotion give a subject a reason for the corresponding evaluative
judgement. This, however, is structurally identical to what happens in the
case of perception. In perception, however, the background conditions
that allow us to form the relevant justifications, upon having certain per-
ceptual experiences, are less complex and more easily in place. That is why
there seems to be a difference between the two cases. To see how this need
not be so, however, consider a subject who knowingly lives in fake-barn
country. In that case, she could not take her perceptual experiences at face
value to form the corresponding empirical beliefs. In fake-barn country,
moreover, acting just on the basis of one’s visual experience would be
unjustified and therefore irrational.
A further objection raised against perceptual accounts is that they can-
not make sense of so-called “why questions” with respect to one’s emotions
(Brady 2013; Deonna and Teroni 2012). For instance, we may well ask
“Why do you fear that dog?”, thereby looking for an explanation of the
reasons of one’s fear, when, for instance, it is clear that the dog in question
is totally harmless. Yet, the objection goes, we cannot raise a similar kind
of question in the case of a perception. “Why do you see a dog?” could be
taken only as an inquiry into the physical causes of that perception.
Now, supporters of the objection tend to equate the reasons one would
cite in response to the why question with the correctness conditions of
the emotion. Thus, for instance, given my fear of the dog, the question
“Why do you fear that dog?” could be answered by saying “Because it is
dangerous”. Its being dangerous would be, at once, the reason of my fear
and what would make that emotion correct and therefore rational, on
this view. Alternatively, if the dog were not dangerous, that would ipso
facto prove my emotion not only incorrect but also irrational. According
to the supporters of the objection, perceptions are neither rational nor
irrational, but merely correct or incorrect. Hence, they conclude, emo-
tions cannot be perceptions, because the former can, as such, be rational
or irrational whereas the latter are neither justified nor unjustified or,
more generally, assessable along rational dimensions.
2 Varieties of Mental States 45
Notice that if the objection moved from a sound premise, namely that
emotions are rational (or irrational), that would mean that they would be
assessed as rational (or irrational) as such, not just via (or because of) the
judgements based on them. If that were the case, emotions would be very
different from sensations and perceptions, which are not assessable as ratio-
nal or irrational, and would be much more similar to judgements, beliefs
and other propositional attitudes, which are assessable along that dimension.
Yet, in response to the objection, supporters of the perceptual model
would be well advised to notice that if reasons are identical to correctness
conditions, then perceptions too would, in some sense, be justified, when
correct, or unjustified, when incorrect, and therefore would be rational
or irrational. Of course, that strikes us as odd, but the onus is on the
opponents’ shoulders. For they would have to show that, while having
correctness conditions, those conditions cannot be reasons of one’s per-
ceptions, while they can be reasons for one’s emotions. This seems to be
a difficult task. Moreover, it seems misguided and brought about by too
hasty an equation between correctness conditions and reasons or justi-
fications we had better avoid. Think of beliefs, which paradigmatically
have truth conditions. Yet they can be justified, on most accounts of
justification, even when occasionally false, and, conversely, be true while
unjustified. Yet, if one does not equate correctness conditions and reasons
or justifications, the “why questions” objection evaporates. For both per-
ceptions and emotions would have correctness conditions and yet the fact
that one may assess the rationality of only the latter would not prove that
emotions are, as such, rational or irrational.
The “why questions” objection is therefore misguided, but it stresses
something important, namely that we do assess emotions along ratio-
nal dimensions. The interesting question—independently of whether the
issue can be used to attack perceptual theories of the emotions—is how
to explain such an attitude. The first option would be to say that we are
speaking loosely. What we should say, rather, is that a given emotion is
either correct or incorrect vis-à-vis the event we are witnessing or imag-
ining. It is correct when it tracks the corresponding evaluative property,
incorrect otherwise. In this case, the analogy with perceptions would be
re-established and this would show, once more, that the “why questions”
objection brings no grist to the anti-perceptual theorists’ mill.
46 A. Coliva
Alternatively, one may suggest that emotions are assessed along a ratio-
nal dimension not as such, but derivatively. That is to say, they are con-
sidered rational when they are in line with our best judgement, irrational
otherwise. For instance, fearing a small, harmless dog is irrational because
it contrasts with our best judgement, whereas fearing a big, aggressive-
looking dog is rational because it is in keeping with our best judgement.
On such a view, emotions would differ from perceptions, for we do not
say that perceptions that are/are not in keeping with our best judgement
are rational/irrational, but, rather, veridical or delusionary. Hence, quite
ironically, it is only by renouncing the idea that emotions are, as such,
rational/irrational that one can appeal to their rational assessment to
draw a contrast between emotions and perceptions. Personally, I favour
this latter interpretation. Nevertheless, it would merely add to the other
ones reviewed in this section, which decisively speak against the percep-
tual account of the emotions.
The models reviewed so far share a common trait. They all try to reduce
emotions to something else—that is, either to sensations or judgements
or to bodily attitudes or perceptions. Perhaps it is by no chance that they
all fail. The idea I wish to put forward is that given the extant state of the
art, it is tempting to think that emotions should not be reduced to any
other allegedly better-understood mental state. Perhaps we had better start
taking the measure of the fact that our conceptual scheme is indicative of
something right, when it groups various mental states under the category of
emotions, rather than do without it, at least at a reflexive, theoretical level.
In this connection, it is useful to exploit the analogy with border-
line mental illness. Borderline patients are not those who are borderline
between mental illness and sanity. Rather, they manifest symptoms, in
various combinations and proportions, which belong to the characteristic
traits of neurosis, on the one hand, and psychosis on the other, without
having only those that belong to one or the other of these two catego-
ries. Borderline patients, therefore, are mentally disturbed, but they are
neither neurotic nor psychotic. They belong to a third, autonomous cat-
egory while sharing symptoms belonging to the other two.
2 Varieties of Mental States 47
4 Summary
In this chapter, we have seen the variety of mental states we can enjoy.
They range from non-representational ones such as sensations, to repre-
sentational yet non-conceptual ones like perceptions, to representational
48 A. Coliva
Bibliography
Bilgrami, A. (2006). Self-knowledge and resentment. Cambridge, MA: Harvard
University Press.
Brady, M. (2013). Emotional insight. Oxford: Oxford University Press.
Burge, T. (2010). Origins of objectivity. Oxford: Oxford University Press.
Claparède, E. (1928). Emotions and feelings. In M. L. Reynert (Ed.), Feelings
and emotions: The Wittenberg symposium (pp. 124–139). Worchester: Clark
University Press.
Coliva, A. (2006). I Concetti. Roma: Carocci.
Coliva, A. (2012a). Critical notice of Tyler Burge origins of objectivity.
Disputatio, 4(33), 515–530.
Coliva, A. (2012b). Human diagrammatic reasoning and seeing-as. Synthese,
186(1), 121–148.
Coliva, A. (2015). Extended rationality. A hinge epistemology. London: Palgrave
Macmillan.
Davies, M. (1998). Externalism, architecturalism and epistemic warrant, in
Wright C., Smith B. and Macdonald C. (eds), 321–361.
Davies, M. (2004). Epistemic entitlement, warrant transmission and easy
knowledge. The Aristotelian Society Supplementary Volume, 78, 213–245.
Deonna, J., & Teroni, F. (2012). The emotions. A philosophical introduction.
London/New York: Routledge.
Dummett, M. (1996). Origins of analytic philosophy. Cambridge MA: Harvard
University Press.
Moran, R. (2001). Authority and estrangement. Princeton: Princeton University
Press.
2 Varieties of Mental States 49
1 First-personal Self-knowledge
Knowledge is a factive propositional attitude: if S knows that P, it is the
case that P. As is well known, there is no consensus regarding its analysis or
even if such analysis is possible in the first place.1 For our purposes, how-
ever, it will do to take it to consist in having a true and suitably justified
belief. Knowledge therefore requires the possession of the concepts neces-
sary to entertain the proposition that P, which is the content of one’s belief.
In general, empirical knowledge is either observational or inferential or
a mixture of the two.2 Suppose I want to know what colour the curtains
of my lounge are. I look and see they are pale yellow. Suppose I want to
know whether Socrates is mortal. I reason like this: Socrates is a man, all
men are mortal, therefore Socrates is mortal. Finally, suppose I want to
know where the north is. I look at the sky, see the sun right in front of
me, check my watch and find out it is noon and therefore infer that the
north is in the opposite direction.
1.1 Groundlessness
in Jane Austen’s Emma in which the protagonist realises her love for Mr.
Knightley. However, for the inference to get started at all, Emma must
already have knowledge of those mental states of hers that figure as con-
tents of the premises of her reasoning, such as, say, “I feel terrible about
the fact that Mr. Knightley could return Harriet’s feelings for him”. On
pain of an infinite regress, there must be knowledge of our own mental
states that is not inferential. Somewhere down the line, we must be able
to know our own mental states directly.3
Moreover, quite intuitively, self-knowledge cannot be a matter of
observation either: firstly, mental states are just not the kind of thing one
could observe4. Secondly, the Cartesian picture of an inner eye, which is
supposed to observe mental states that are luminously presented in the
mental arena, is more a recipe for trouble than a viable explanation of the
sense in which self-knowledge could be observational. Very briefly: it is a
recipe for trouble because it would lead to solipsism—the idea that each
of us is caught up in her own world insulated from anyone else, being
unable to know whether others have mental states and are, therefore, full-
fledged persons5. After all—the train of thought would go—how would I
know that other people have mental states at all if those mental states are
intrinsically private to them and foreclosed to me? It may really become
a recipe for catastrophe when taken a step further and taken to involve
conceiving of our psychological language as private. Since Wittgenstein,
however, private language has become synonymous with incoherence.
The idea is this: any language is a rule-governed practice and it is essential
to it that there be a distinction between correct and incorrect applications
of the terms. Still, if the meanings of our psychological vocabulary (or
indeed the contents of our psychological concepts) are constituted by
mental states that are private to each individual, then whatever seems to
one to be the right application of the psychological term “S” is right6. This
3
Cf. Wright 1998, p. 16.
4
This point has been convincingly argued for by Ryle 1949and Shoemaker1996b
5
This point has been vigorously argued for by Ryle 1949.
6
Wittgenstein’s remarks in the Philosophical Investigations about the very possibility of a private
language are profitably read as responses to Russell’s observations on introspection in Russell (1912,
Ch. 5) and Russell (1921). Russell’s views have somehow been resurected by Gertler (2011a) and
Chalmers (2004). Assuming, for the sake of argument, that no criterion of correctness were necessary
54 A. Coliva
just means that the distinction between being right/seeming right, which
is much the same as the distinction between correct/incorrect applica-
tions of a word (or a concept), has vanished. Yet, as we have just seen,
that distinction is essential to there being a (psychological) language (and
concepts) at all. Thus, Cartesianism about self-knowledge may take us to
a conception of our psychological language, which turns it into a non-
language. However, we do have a psychological language—after all, we
tell each other about our own thoughts and feelings, sensations and emo-
tions all the time and we seem to understand each other well. Therefore,
it seems safe to conclude that Cartesianism will not do. If Cartesianism
has to go, so has the idea that self-knowledge could be observational.
Hence, the first two constraints on any feasible account of self-
knowledge are:
for concepts’ possession (as well as for linguistic meaning), it would seem possible to form very short-
lived, demonstrative concepts of the form that S, which would exploit the presence of the very
mental state one is having. Some theorists would object to the very idea of such short-lived concepts,
requiring the possibility of redeployment in order to grant the existence of a genuine concept in the
first place. If so, Wittgenstein’s critique would apply.
If one did not require multiple applications for the existence of the relevant demonstrative con-
cepts, one would avoid Wittgenstein’s objection or at least would considerably deflate it. But then
one would have to live with the unpalatable consequence of having only short-lived, in fact singular
psychological concepts. That would make it difficult to understand how we could ever make identity
judgements about different token mental states, or comparative judgements about them over time.
3 Varieties of Self-knowledge 55
In what follows, we will see how these different intuitions can be devel-
oped into different accounts of self-knowledge. At this stage, it is enough
to register them and to point out that, in a minimal sense, groundlessness
amounts to the idea that first-personal self-knowledge is not the result of
any substantial cognitive achievement such as inner perception or infer-
ence to the best explanation.
56 A. Coliva
to her intentions, she would not know why, for instance, she happens to
be pushing a cart around in a supermarket—as she witnesses herself doing.
She would not act but be acted, and at any instance of the question “Why
did you do/are you doing that?”, she would be entitled to respond “I do
not know. Let me find out” and she would then be entitled to preface
whatever answer she might wish to give, based on inference to the best
explanation, with “Probably I did/am doing that because …” or “It is my
hypothesis that I did/am doing this because …”. Now, these responses,
though understandable from the point of view of a subject like the one
just described, would actually strike us as a symptom of mental illness if
they were given by an actual subject. They would strike us as reports on
another subject, who is the subject of one’s actions, which would happen
to live in one’s body.7 On a “thick” notion of being a rational agent, who
can be considered rationally responsible for (at least the vast majority of )
her actions, knowing of them and their motivations only through infer-
ence to the best explanation and in the way of a hypothesis would actu-
ally represent a major departure from rationality.8 Hence, groundlessness
seems to be an element of genuine first-personal self-knowledge, which in
turn is a constitutive element of being a responsible agent.
Now, to claim that groundlessnessweak is constitutive of first-personal
self-knowledge and to redeem that claim by means of a priori consider-
ations having to do, as we have seen, with the notion of sensation (and
possibly other phenomenologically salient mental states), and with that
of being rational agents who can be held responsible for their actions,
does not mean to prejudge the issue of how a satisfactory account of first-
personal self-knowledge can go about trying to meet this desideratum. In
particular, it is still possible to propose a fully epistemic account of first-
personal self-knowledge that aims to respect this requirement. It is only
by the lights of non-epistemic theories of first-personal self-knowledge
that groundlessness, just like the other features of first-personal self-
knowledge we will examine in the following, cannot be accounted for in
epistemic terms. By the lights of those theories, groundlessness has to be
7
Similar considerations are advanced in Shoemaker 1996b, as we shall see in Chap. 7 (§1).
8
As stressed in Chap. 2, in connection with propositional attitudes as commitments, to be rational
agents does not mean acting necessarily for good reasons. It means, however, to have knowledge of
the actions one is performing and of their motivations, at least for the most part, so as to be in a
position to be held responsible for them.
58 A. Coliva
1.2 Transparency
it is, to preface it with “I believe that”. This recipe for accounting for
our knowledge of propositional attitudes is often called the “transparency
method” because it bypasses one’s inner world, as it were, to direct one’s
mind and attention to P itself and to the reasons for or against P that
should lead us either to believe it or not. We will address this model of self-
knowledge in Chap. 5. The reason why we put it aside until then is that,
as remarked, Evans’s method is supposed to give us an account of how we
can have knowledge of our minds and, in particular, of beliefs and possibly
other propositional attitudes. Therefore, it cannot be taken as one of the
neutral conditions that all accounts of self-knowledge should satisfy.
Some theorists are unconvinced that transparency is a characteristic
feature of our own mental states (see Snowdon 2012 for a criticism of
transparency with respect to sensations, emotions and passing thoughts).
They point out, for instance, that higher-order mammals and infants do
not seem to be capable of self-ascribing sensations and basic emotions,
or even perceptions, although they can enjoy them. To such a worry, one
might respond by noticing that we have been talking of transparency
mostly in terms of awareness and it seems safe to hold that, at least in
the case of sensations, occurrent emotions with a characteristic ongoing
phenomenology, and conscious perceptions, their occurrence seems to
coincide with one’s awareness of them, even if one does not possess the
concepts necessary to self-ascribe them. More precisely:
the idea that also propositional attitudes which may not have a distinctive
phenomenology can be transparently self-known. As we saw at the very
beginning, having formed the intention to cook risotto for dinner, one
may be immediately in a position to self-ascribe that thought, even if,
arguably, there is no distinctive phenomenology to forming such inten-
tions. In such a scenario, the likely move would be to conditionalise trans-
parency to the obtaining of the relevant (C-)conditions. Namely, subjects
should be conceptually endowed, besides being cognitively lucid, attentive
and alert. By means of such a conditionalisation, then, usual counterex-
amples to transparency, often presented in the literature on self-knowl-
edge, would easily be dispensed with. The case of animals and infants
would be countered by the requirement that subjects should possess the
relevant concepts. If, in addition, a subject is under the effect of drugs,
has her attention occupied by a pressing task, or is distracted or distressed
in various ways, she may be having sensations, emotions and perceptions
and yet not be aware of them. Given the conditionalisation just proposed,
these cases would not represent counterexamples to transparency.
A separate remark is apposite in relation to the Freudian idea that
we may have many unconscious propositional attitudes, such as beliefs,
desires and intentions, of which, by definition, we are totally unware.
Obviously, there is no denying that this might be the case. However,
Freudian mental states are mental dispositions (see Chap. 2, §2). Hence,
one may add to the C-conditions that when propositional attitudes are
at stake, they should be beliefs, desires and intentions as commitments,
not as dispositions. We will come back to this issue in due course (see
Chap. 7, §§3–4). For now, it suffices to have shown how transparency
can be understood and limited, so has to hold for at least a good range
of mental states.
1.3 Authority
in Italy are really too hot” and you are sincere and know how to use ‘I’,
‘think’, ‘summers’, ‘Italy’, …, then nobody could challenge you by say-
ing “Are you sure that this is what you are thinking?”. Of course, one
can challenge the subject’s grounds for believing that summers in Italy
are too hot, by pointing out that it is August 15, you are in Italy and the
temperature is only 15 °C. Yet one cannot challenge the subject’s own
believing that she believes it. By challenging the grounds in favour of
that content’s belief, one can lead a subject to revise her belief, but this
possibility depends on the (implicit) acknowledgement that the subject
does believe—erroneously, from your point of view—that summers in
Italy are too hot and that she knows that much about herself. Similarly,
if I sincerely avow a painful sensation, it would be pointless for someone
to challenge me by saying “Are you sure that is what you are feeling?”. To
the restatement of my avowal, I could only add that I am being sincere
and that I can speak English. Therefore, subjects are authoritative with
respect to their own mental states: if they say (or judge) that they have a
certain mental state M, then they have it.
Surely, however, if someone asked you “What is the weather like?” or
“How is your mother?”, from your sincere and conceptually competent
answers “It is raining” or “She is very sad since she retired recently and
she feels useless”, it wouldn’t have followed at all that it would be inap-
propriate for someone to challenge you by saying, for instance, “Are you
sure? You haven’t taken a look out of the window and one of those relax-
ation CDs that make all the noises of nature is on”. Or: “Are you sure? I
saw your mum last night at a party, dancing to ‘Staying Alive’ with your
dad, and she did not look sad at all”. Therefore, authority is another
feature that sets self-knowledge apart from all other kinds of knowledge.
Again, some theorists are wary of authority (see again Snowdon 2012).
With respect to sensations, for instance, they point out that one can
envisage a situation where a subject is asked to locate her sensation and is
uncertain about that. This, however, would merely show that we are not
authoritative with respect to the bodily location of our ongoing sensa-
tions. Yet it would not show that we are not authoritative with respect to
the fact that we are undergoing them.
Another case often discussed in the literature is the one of a subject
who, by dreading the pain a dentist will produce by messing about her
64 A. Coliva
mouth, claims to be feeling pain already when the dentist has not even
touched her. This case can be taken care of by specifying the C-conditions
under which authority is supposed to hold. In fact, the subject should
be cognitively lucid. Arguably, however, intense fear can impair one’s
cognitive functions. Once again, the possibility, on specific occasions, of
mistaken self-ascriptions of sensations should not be taken as a counter-
example to authority, but only as imposing the need of properly charac-
terising its extent.
Another prima facie difficult case to handle is the one of confused or
vague sensations, such as itches that border pain. A subject could then
be unclear whether she is undergoing one or the other kind of sensation
and make avowals she would then like to retract. This, however, would
not be a counterexample to authority, because we are, in fact, dealing
with a confused or vague sensation, which escapes definite categorisation.
Hence, we vacillate between two concepts we do possess, when both of
them seem to apply to it to some extent. In this sense, our self-ascriptions
may be imprecise and retractable. Yet this does not show that, at least
when sensations are not of a confused nature, we would not be authori-
tative with respect to them. More generally, the ubiquitous presence of
vagueness in our concepts should not be taken—by itself—as a serious
challenge to the possibility of first-personal self-knowledge any more
than the ubiquitous presence of vagueness in the domain of colour is
normally taken as a challenge to the possibility of correct colour percep-
tions and judgements.
Another possibility is to be wrong about the content of one’s percep-
tual appearances. For instance, one can say that one’s after image contains
five red dots, when in fact it contains only four. However, authority can
be maintained even in this case because determining the number of dots
involves counting and one may go astray in doing it. Still, one would be
authoritative with respect to the fact of having an after image.
Interestingly, Eric Schwitzgebel (2008) has recently argued that we are
not authoritative with respect to our inner feelings and emotions. The
idea is that we may be bad at judging that the feelings we are currently
experiencing are feelings of anger, say, when in fact this would be clear to
a third party. It should be admitted that sometimes we do not realise what
the feelings we may experience are symptoms of. This, however, could be
3 Varieties of Self-knowledge 65
9
Whether, in turn, they are also constitutive elements of what it means to be a self or a subject at
all, insofar as they are constitutive of being critical reasoners and subjects of moral norms as Burge
2011 maintains, is a further issue, which would need a separate treatment.
3 Varieties of Self-knowledge 67
2 Counterexamples from Content
Externalism and Cognitive Sciences?
Some theorists (Boghossian 1989 and McKinsey 1991) take the rise of con-
tent externalism to be incompatible with at least authoritative self-knowledge
regarding one’s current propositional attitudes. For, if externalism is correct,
a subject may think of being thinking a water thought, say, when she is in
fact entertaining a thought about twater because of her actual causal connec-
tions with an environment in which lakes, rivers and seas are in fact filled in
with XYZ rather than H2O. Hence, it would seem to her to be thinking a
water thought, while she would not. Therefore, she would lack knowledge
of her current thought. For, arguably, knowledge of one’s thought would
entail knowledge of its nature—its being a belief rather than a hope or a
desire—and of its content—namely, its being about water and not twater.
Let us grant, for the sake of argument, that content externalism is cor-
rect. Let us further suppose that, unbeknownst to her, our subject is actu-
ally thinking a twater thought. Still, she would seem to have essentially
first-personal access to it even if she may be wrong about its actual con-
tent. Hence, she would still have transparent access to the fact that she is
entertaining a thought rather than a hope or a wish, and her access would
still be groundless. That is, it would be based neither on observation nor
on inference. Finally, she would still be authoritative with respect to its
seeming or apparent content. Some theorists would call it “narrow con-
tent” and would happily acknowledge its existence alongside with “wide
content” (Putnam 1975). Hence, although strong authority may be
impaired by content externalism, a weaker kind of authority, at least with
respect to the kind of mental states one is enjoying and the narrow con-
tent of one’s propositional attitudes, would be preserved. It is that kind of
authority that, being fundamental to psychological explanation in a way
in which, arguably, authority over the wide content of one’s propositional
attitudes is not, needs to be explained by any account of first-personal
self-knowledge and that we will try to account for in what follows.10
Let us now move on to the challenges to first-personal self-knowledge
raised in recent years by studies in cognitive psychology. They tend to chal-
lenge the idea that we are authoritative with respect to our own mental
10
For a review of classical positions on this topic, see Parent 2013.
68 A. Coliva
11
See Gertler 2011a, pp. 70–86.
12
Gilbert 2006.
13
Nisbett and Wilson 1977; Libet 1985; Wegner 2002; Wilson 2002; Wegner and Wheatley 1999.
3 Varieties of Self-knowledge 69
the thoughts that we take to be the causes of our actions. Once more, the
scope of first-personal knowledge of our own mental states appears to be
limited but not to be vanishing.
Similar results have been achieved by studies concerning our moods,
such as pessimism, like the tendency to regard the future as bleak, and
emotions, such as anger. Gertler correctly points out that moods are
dispositional mental states and that emotions involve, besides a certain
feeling, with respect to which we do have first-personal access, certain
dispositional elements. This idea chimes well with the borderline concep-
tion of the emotions we put forward in Chap. 2 (§3), according to which
emotions cannot be reduced to either feelings or dispositions. It may well
be, then, that although we have privileged access to their phenomeno-
logical aspects, we do not have it to their dispositional elements.
Finally, if we do not have first-personal knowledge of dispositional
mental states, that entails that we are not authoritative, nor do we have
immediate and groundless access to a lot of propositional attitudes, such
as beliefs and desires, which are dispositional. Indeed, this is only to be
expected. Yet we do also have the impression of having first-personal access
to our beliefs, desires and intentions. Our view, presented in Chap. 2, that
we should distinguish between different kinds of propositional attitudes—
that is, those as dispositions and those as commitments—comes in handy
at this stage. For we may say that whereas we lack first-personal knowledge
of the former, we have it of the latter. How we do have it will have to be
explained, and we will consider various explanatory accounts in the fol-
lowing chapters. Yet, although this imposes a further limitation to first-
personal self-knowledge, it does not show that we lack it altogether. Rather,
it merely shows that first-personal self-knowledge is limited to occurrent
sensations, including those characteristic of at least basic emotions, percep-
tions, current thoughts and propositional attitudes as commitments.
3 Third-personal Self-knowledge
So far, we have examined the characteristic aspects of first-personal self-
knowledge—that is to say, those aspects, which set our knowledge of our
own mental states apart from other kinds of knowledge, such as knowl-
edge of truths about the external world, usually achieved through percep-
70 A. Coliva
14
I was pleased to find a similar claim in Cassam 2014.
3 Varieties of Self-knowledge 71
may be deeper unconscious reasons for that. For instance, if the person in
question is a colleague and one is generally convinced that intimate rela-
tions with one’s colleagues should be avoided at all costs in order not to
harm one’s career or public image, one may persistently ignore the evidence
that shows that one is indeed in love with a colleague.
We have also seen how alleged counterexamples to first-personal self-
knowledge drawing on recent findings in cognitive studies can in fact
be interpreted as interesting examples of third-personal knowledge of
our own mental states. Therefore, for instance, we think we have cer-
tain character traits, or that we will have certain emotional responses
in the light of alleged life-changing events, which we actually lack. The
interesting aspect of this failure at self-knowledge is that the reason why
we think we would have these mental dispositions is that we presum-
ably engage in imagining how we would behave in a given situation.
For whatever reason, which may or may not be biased, we think we
would behave in ways that would depend on having a certain mental
state. Yet, when faced with a real-life situation, we do not behave as we
imagined we would do. The kind of procedure followed in these cases,
in order to try to gain knowledge of our mental dispositions, is indeed
similar to the one we would apply if we were to figure out someone else’s
mental states by means of simulation and its correct outcome is obvi-
ously not guaranteed. It is therefore another example of third-personal
self-knowledge.
Another intermediate case, which does not immediately coincide with
knowledge of our own mental states either through inference to the best
explanation or through mental simulation, is the one in which we suddenly
notice an aspect of ourselves we had been previously blind or oblivious to or
indeed reconceptualise an aspect of ourselves by suddenly seeing it in a dif-
ferent way. We may all of a sudden realise that we are not nice to other peo-
ple—we do not actually care about them that much—but only act polite, or
we may suddenly see a character trait of ours not as a symptom of strength
but of weakness. Seeing-as, switches of aspect and noticing aspects are com-
plex phenomena even when they clearly concern perceivable objects and
properties and involve the operation of perceptual faculties.15 Obviously, in
15
Cf. Wittgenstein PI, II, xi. In Coliva 2012b, I consider the application of these notions to the use
of diagrams in geometrical proofs and provide the relevant bibliographical references.
3 Varieties of Self-knowledge 73
the case of our own mental states, these notions find only mediated or indi-
rect employment, for no real perceptual faculty is involved and yet it does
not seem that we engage in any complex inferential thinking or in mental
simulation. Furthermore, the relevant psychological concepts are necessar-
ily needed, whereas, arguably, purely perceptual switches of aspects do not
necessarily require concepts (although they may and often do involve them).
We may say that we suddenly take a pattern of overt behaviour, and possibly
other inner aspects of our psychology, as instantiating this or that psycho-
logical property either because that pattern is manifest to us for the first
time or because the relevant concept is available to us at last and a pattern of
behaviour and inner elements we had possibly noticed before seems imme-
diately to fit the newly acquired concept. Sometimes, these new concep-
tualisations or reconceptualisations contradict previous ones and that gives
rise to a switch of aspect regarding ourselves and our personalities. Still, in
these cases, inference is involved at least in the following sense. While there
is no inference to the best explanation, the characteristic notes of the newly
acquired (or applied) concept guide the process of subsuming the observed
pattern of behaviour and inner elements under the concept. That may hap-
pen very rapidly, and so, phenomenologically, we may not even seem to be
engaging in inferences, yet epistemically that is what grounds our judgement
“I am F” (where F is a psychological concept). It is one possible case of self-
interpretation and it can go wrong because, after all, the pattern observed
may not fit the concept or we may have ignored other features which would,
if taken into account, lead to a different psychological self-ascription.
Another interesting case is the one in which we aquire self-knowledge
by means of someone else's testimony. For instance, a mother tells her
son he is pushy, or even that he is being pushy on a specific occasion.
He believes his mother and forms the belief, based on testimony, that he
is (being) pushy. This would count as a clear case of testimonial—and
therefore third-personal—self-knowledge. Testimonial self-knowledge is
also achieved in more institutionalized contexts in which, for instance, a
therapist makes a diagnosis about a subject's psychological situation and
the latter forms the corresponding belief.
What is important to notice at this stage is that third-personal self-
knowledge exhibits all the typical aspects of any substantive cognitive
achievement, which usually results in knowledge. We base our psychologi-
cal self-ascriptions on various kinds of evidence—both observational and
74 A. Coliva
4 Summary
In this chapter, we have seen the characteristic traits of first-personal self-
knowledge (§1)—that is, groundlessness, transparency and authority—
and the different readings which can be offered of them. Assuming as a
methodological rule the least demanding reading of each of them, compat-
ible with the project of accounting for self-knowledge, we have argued that
it is a necessary and a priori aspect of first-personal self-knowledge, whose
failure would impair a subject’s conceptual mastery with respect to the rel-
evant concepts or her “thick” rationality and therefore her being rightly
considered a responsible agent. We have then defended the thesis that there
3 Varieties of Self-knowledge 75
Bibliography
Boghossian, P. (1989). Content and self-knowledge. Philosophical Topics, 17, 5–26.
Burge, T. (2011). Self and self-understanding. The Dewey lectures (2007–2011),
The Journal of Philosophy, CVIII/6–7. Reprinted in Cognition through under-
standing (pp. 140–228). Oxford: Oxford University Press.
Cassam, Q. (2014). Self-knowledge for humans. Oxford: Oxford University Press.
Chalmers, D. (2004). Phenomenal concepts and the knowledge argument, In P.
Ludlow, Y. Nagasawa, & D. Stoljar (Eds.), There’s something about Mary:
Essays on phenomenal consciousness and Frank Jackson’s knowledge argument
(pp. 269–298). Cambridge, MA: MIT Press.
76 A. Coliva
1
Armstrong focuses in particular on self-knowledge of intentional mental states. Although he is not
always clear on this point, his proposal is generally taken to be an explanation of how we can form
reliable (and therefore knowledgeable) beliefs about first-order mental states. Notice that if it were
otherwise—that is, if his theory accounted just for how we can be conscious of our first-order
mental states, in a way which does not involve forming beliefs about them—then it should be
completed by how such consciousness could give rise to self-knowledge properly so regarded.
2
This, I take it, requires concepts’ possession, since one cannot entertain beliefs if one does not have
the concepts necessary to grasp the propositions which figure as their contents. Alas, Armstrong is
not always clear on this score.
80 A. Coliva
3
This is indeed the driving thought behind Sydney Shoemaker’s objections to Armstrong’s model.
According to Shoemaker, who is a functionalist about mental states, each mental state is individu-
ated by its functional role—that is, its characteristic input and output. He then claims that, con-
trary to Armstrong’s causal model, there cannot be a given first-order mental state unless certain
behavioural outputs are in place. They do, however, require awareness of one’s first-order mental
states. For instance, since pain is individuated as the mental state that depends on physical injury
and gives rise to characteristic behaviour, involving the attempt to get rid of it, the belief (or the
ability to have the belief ) that one is in pain is necessary in order to be in pain. Therefore, there
cannot be self-blind creatures—that is to say, creatures who are capable of having first-order mental
states and yet do not have (or are in principle incapable of having) knowledge of them. Similarly,
beliefs as possessed by rational creatures ought to be revisable based on contrary evidence. This,
however, requires the ability to know that one has a given belief, which should be revised in light
of counter-evidence. Neither objection is decisive though. For Armstrong and Lycan could argue
that, on the one hand, pains are possible and can have a distinctive phenomenology even if one is
not aware of them qua pains. Infants and animals would indeed seem to be capable of feeling pain
without having knowledge of it. Similarly, a subject who lacked the relevant conceptual repertoire
could have a first-order belief and even change it on the basis of contrary evidence and yet not be
capable of having beliefs about one’s own beliefs or of answering the question “Do you believe that
P?”. We discuss Shoemaker’s position in Chap. 7 (§1).
4 Epistemically Robust Accounts 81
in. You then see what mental state she is in when she avows her pain and
hence has the belief that she is in pain. Finally, you test the correlation
between these two mental states to see whether it is reliable.
However, you cannot do this for the hope that peace be reached in the
Middle East: there are no merely behavioural symptoms that correspond
to hoping that peace be reached in the Middle East.5 Hence, in the case of
pain, one may individuate the neural configuration that realises it, inde-
pendently of the avowal of pain and then verify whether a subsequent
avowal is reliably caused by the neural realisation of pain. In the case of
the hope that peace be reached in the Middle East, in contrast, this can-
not be done, because, to repeat, we cannot have access to the first-order
mental state independently of the subject’s avowal that she has it. Hence,
Armstrong’s model, which promised to be scientifically amenable, seems
to fail to be empirically testable and to do so because it must presuppose
self-knowledge (and indeed in the form of an avowal) in order to indi-
viduate the relevant first-order mental states in the first place.6
Of course, one may want to protest that I am being unfair to
Armstrong because I am allowing too narrow a behavioural basis: there
is no reason why the relevant behaviour should be merely physical, as it
could comprise also linguistic items, such as the assertion of “Peace in
the Middle East would be good to have”. If this larger behavioural basis
is allowed—the train of thought would be—then one might find out
what neural state it corresponds to and see whether it is followed by the
appropriate psychological avowal—that is, “I hope peace will be reached
in the Middle East”. Conversely, one might test whether the avowal is
accompanied by the brain state one is in when one is enjoying the first-
order mental state. Let us suppose that by so doing we individuate two
different brain states—the brain state which would correspond to uttering
“Peace in the Middle East would be good to have” and the brain state
5
Or, at any rate, behavioural manifestations underdetermine the individuation of precisely that
mental state.
6
So, although there may be causal mechanisms that enable self-knowledge, as I presume there must
be causal mechanisms that enable thought in the first place, self-knowledge cannot be explained by
appealing only to them. Moreover, it may well be the case that the causal mechanisms that enable
self-knowledge are not anything like a dedicated faculty or scanner mechanism but the ones which
underwrite concepts’ possession and reasoning.
4 Epistemically Robust Accounts 83
observation of one’s own behaviour, she would have lost the ability of
having distinctively first-personal self-knowledge—the very ability which
Armstrong’s model was supposed to explain.
Finally, the test might also show that the two mental states have in
fact the same neural realisation. Then, ironically, the test would support
precisely constitutive positions. In particular, it would support those con-
stitutive accounts, such as Shoemaker’s (see Chap. 7, §1) that hold that
there is just one mental state—with its neural realisation—and that all
is needed to have knowledge of it is a normal degree of rationality, intel-
ligence and conceptual mastery.
First, this whole idea of seeing our own mental states is a muddle.
Indeed, Descartes9 got the phenomenology of self-knowledge wrong: we
do not see our own mental states. The use of that verb is highly meta-
phorical. Hence, the analogy between the child and the seasoned scientist
is beside the point—simply, there is nothing like seeing which should be
accounted for in the first place.10
Secondly, even in the little sketch of reasoning I presented as an exem-
plification of what Gopnik has in mind, there is already a piece of self-
knowledge—that is, the subject’s knowledge of her feeling hungry. To be
fair, Gopnik allows for mental states other than propositional attitudes to
enter the inference and she does not regard this as a problem since she claims
to be merely offering an account of our knowledge of intentional mental
states such as beliefs. However, the point is that one needs knowledge of
one’s own intentional mental states too, to perform the relevant inference.
For, surely, one cannot infer that one desires to have an ice cream just from
observing one’s own physical behaviour—one’s moving towards the fridge,
say—since that behaviour would be compatible with thousands of different
explanations. Nor can one infer one’s desire to have an ice cream from that
behaviour together with one’s awareness of feeling hungry, because, again,
that would still be compatible with many different explanations. Therefore,
in addition, one seems to need the belief that there is an ice cream in the
fridge and that the ice cream is what will satisfy one’s appetite, or that one
feels like having the ice cream that one believes to be in the fridge. Still, in
this case, one will have to have knowledge of feeling like having exactly the
ice cream, which comes very close to already knowing one’s desire to have
it. For either the raw feeling is not phenomenologically distinguishable
from other ones and we need to conceptualise it as that very kind of feeling,
9
Or, at any rate, what goes by the name of the Cartesian model, whether or not it was actually held
by Descartes.
10
So I wholly agree with Moran (2001), who refers to the talk in terms of “inner vision” as a “mislead-
ing metaphor” (p. 13) and writes (p. 14): “While ‘representationalism’ is a controversial thesis about
the ordinary perception of objects in the world, on nobody’s view is the awareness of one’s headache
mediated by an appearance of the headache. And in the case of attitudes like belief, there is simply
nothing quasi-experiential in the offing to begin with. There is nothing it is like to have the belief that
Wagner died happy or to be introspectively aware that this is one’s belief, and that difference does not
sit well with the perceptual analogy”. Of course, Gopnik might recast her point by saying that just
like a seasoned scientist knows immediately—that is, without inference—that there in electron mov-
ing about, so does a subject know her own propositional attitudes. Yet this is not her official position,
and, even if it were, the following criticisms developed in the main text would apply.
4 Epistemically Robust Accounts 87
11
This view does not seem to me very plausible, though. For it seems difficult to distinguish feeeling
like having an ice cream from feeling like having a sorbet or an ice lolly merely on the basis of
phenomenological features.
12
See Cassam 2014, p. 169.
88 A. Coliva
13
See Cassam 2014, pp. 130, 138.
14
That is to say of non-occurrent propositional attitudes, a subject may nevertheless be aware of.
4 Epistemically Robust Accounts 89
15
See Gordon 1995, p. 62.
16
Gordon 1995, p. 63.
90 A. Coliva
17
Ibid.
18
Gordon does not expound his criticism of introspection but merely says that he does not find it
very plausible (see Gordon 1995, p. 59). Anyway, in Chap. 3 and in §1 of the present chapter, we
have already considered several objections to it.
4 Epistemically Robust Accounts 91
19
Of course, one may dispute that such a list captures the constitutive inferences of our concept of
belief or even that the right account of it is inferential. These possible objections are not relevant to
present purposes.
20
Gordon 1995, p. 59. We will come back to this procedure in Chap. 6.
21
Gordon 1995, p. 60.
22
We will come back to this procedure in Chap. 5, §3.
92 A. Coliva
such training, although not sufficient for the acquisition of the concept,
may be the most important step in the acquisition of the concept. (To be
specific, it links the name of a propositional attitude with the verbal expres-
sion of that attitude; and the child who can also expressively use ‘I want’
when speaking for another (that is, in the context of a simulation) comes
close, in my view, to have grasped the general concept of wanting.)28
such utterances, made without ascriptive intent, are in fact the way young
children first employ the linguistic forms of propositional attitude ascrip-
23
Gordon 1995, p. 59.
24
Gordon 1995, p. 61.
25
Ibid.
26
Ibid.
27
See Goldman 1993.
28
Gordon 1995, p. 65, fn. 7.
4 Epistemically Robust Accounts 93
tion. For example, children first use “I want” merely as a way of getting
what they want. Yet the resulting utterance, were it construed as an ascrip-
tion, would generally be a true one. Thus, children are linguistically boot-
strapped from expression to reliable self-ascription. The same should hold
whether one is expressing one's own propositional attitude or vicariously
expressing another’s.29
Thus, the idea seems to be that first we learn to use “I want” and
“I believe” merely as alternative expressions of our wants and convic-
tions, that we may have expressed without introducing those expressions.
Then, by simulating others and by using the same routines we use in our
own case, we would get the semantic continuity between first-person and
third-person uses of those expressions.
Yet, if that is what Gordon has in mind, it seems too scant to give
us full-blown possession of the relevant concepts. For, by Gordon’s very
lights, the latter requires grasp of the distinction between facts and their
representation by subjects, when the two can dramatically come apart.
Furthermore, given Gordon’s specific conception of simulation as total
projection, where even the reference of ‘I’ is shifted from the simulator
to the simulated subject, possibly occurring at a stage when the former is
not even supposed to grasp that ‘I’ normally refers to oneself, simulation
would still be an exercise in first-personal, rather than in third-personal,
understanding. For I would be “attributing” the relevant mental states
to myself, even though my own self would have been totally projected
onto someone else. Therefore, Gordon does not seem to offer us a sound
explanation of the possession of propositional attitudes’ concepts, for the
central inferences which individuate the concept of belief in particular—
that is, (1) and (2) —and first-personal/third-personal semantic continu-
ities do not seem to be accounted for. Failing that, he does not propose
a viable account of self-knowledge, for which possession of the relevant
psychological concepts is necessary.30 It remains that simulation can be
an important tool, more effective and cognitively parsimonious than the
application of a theory to oneself, in order to form predictions about
29
Gordon 2009.
30
As we shall see in Chap. 6 (§2), it is possible to offer an essentially expressivist story as a viable
explanation of the possession of at least a rudimentary concept of belief, but further elements will
have to be built into the picture, besides the ones Gordon allows for.
94 A. Coliva
one’s own behaviour and, granted the possession of the relevant psycho-
logical concepts, a source of knowledge of our own minds in hypothetical
or future situations.
As we shall see in further chapters, however, Gordon’s views, as indebted
to Wittgenstein and Evans as they are, can be built upon to give a satisfac-
tory account of the possession of the relevant psychological concepts. So,
let us consider the epistemic characteristics of this model, assuming for
the sake of argument that it could be complemented by a sound account
of psychological concepts’ possession, alternative to the one which would
fit better with theory-theory accounts.
The account would respect groundlessness at least in its weak form,
for it would be anti-introspectionist and anti-inferentialist. It would
also respect transparency, given the ascent routine our self-ascriptions
would be based on and given the expressivist account of self-ascriptions
of non-intentional mental states provided. However, authority would
be impaired. Surely, simulation would give us more reliable predictions
about our future or hypothetical mental states than the application of a
theory. But they would still be predictions, which remain clearly fallible.
Notice, however, that that would not impair one’s rationality or concep-
tual abilities. A subject who got predictions wrong about her future or
hypothetical states of mind would not necessarily be irrational, nor would
she display lack of the relevant psychological concepts. She may just be
poor at performing the task, by being oblivious to some relevant aspects
of the simulated situation. Or else, in real life, she could actually react
differently from how she thought she would. That is enough to show that
the kind of simulation Gordon has in mind is, after all, a source of third-
personal self-knowledge, rather than of first-personal self-knowledge,
despite its inheritance of some features that characterise the latter.
In this connection, it is worth pointing out that also the theory-theory
could offer important elements to understand the way in which, at least
sometimes, we gain knowledge of our past and dispositional mental states,
in particular. For instance, one could reflect on one’s behaviour on a past
occasion and interpret the data by means of certain psychological “laws”
and therefore attribute to oneself this or that propositional attitude, emo-
tion or character trait, as a disposition. If the theory were in fact a piece
of Freudian theory, the very same procedure could yield knowledge of
4 Epistemically Robust Accounts 95
one’s unconscious mental states. Clearly, one can be more or less reliable
in doing so and versed in this practice. Yet, obviously, one would by all
means get third-personal knowledge of one’s own mental states. Hence,
both theory-theory and simulation theories can offer important elements
to understand third-personal self-knowledge while failing, for different
reasons, to account for distinctively first-personal self-knowledge.
4 Summary
In this chapter, we have considered epistemically robust accounts of self-
knowledge. We have started with recent inner sense theories and we have
examined Armstrong’s proposal in particular. In his view, self-knowledge
is the product of a reliable cognitive mechanism that tracks first-order
propositional attitudes and produces the corresponding second-order
ones. The mechanism is physically realised in our brains and is opera-
tive at the subpersonal level. We have claimed that although the model
accounts for groundlessness, transparency and authority, it severs the
connection between self-knowledge and rationality and concepts’ posses-
sion. Hence, it is deemed unsatisfactory.
We have then moved on to Gopnik’s theory-theory. According to this
model, we know our minds in the same way we know others’—that is,
through inference to the best explanation, based on observation of one’s
overt behaviour and non-intentional mental states and through the appli-
cation of a real theory of the mind subjects acquire around the age of four.
We have taken issue with the model from a merely philosophical point
view, even though it has been variously criticised also from a psychologi-
cal point of view. Moreover, we have done so by granting, albeit merely
for the sake of argument, that it could somehow satisfy or at least assuage
the requirements for distinctively first-personal self-knowledge—that is,
groundlessness, transparency and authority. We have therefore claimed
that it is inadequate from a philosophical point of view because it either
collapses into a crude form of behaviourism or presupposes the very self-
knowledge it aims to explain or something too close to it to avoid the
charge of vicious circularity. We have raised similar worries with respect
to Cassam’s recent defence of inferentialism as well as cast doubts on the
96 A. Coliva
Bibliography
Armstrong, D. (1968). A materialist theory of the mind. London/New York:
Routledge.
Cassam, Q. (2014). Self-knowledge for humans. Oxford: Oxford University Press.
Dokic, J., & Proust, J. (Eds.). (2002). Simulation and knowledge of action.
Amsterdam/Philadelphia: John Benjamins Publishing Company.
4 Epistemically Robust Accounts 97
concept, the concept of belief and those concepts that are necessary for
the specification of the content of your belief ). In Peacocke’s view, when
the self-ascription is formed in this way, it amounts to knowledge because
the second-order belief would be true—in virtue of the conceptual truth
that if one judges that p, one believes it—and justified by the correspond-
ing first-order mental state (namely, the judgement that P).3
In Peacocke’s view, this proposal avoids inferentialism because “to say
that (2) is the thinker’s reason for making the judgement in (3) is not to
say that he infers the self-ascription from a premise that he has made such
a first-order judgement”, together with the bridging principle that if one
judges that P, one believes it.4 Rather, the transition is “blind”—that is,
the subject makes a judgement that P and then self-ascribes the belief,
without inferring to it. Yet the transition counts as rational because it
can be rationalised by appealing to two aspects of it: the fact that the
first-order mental state is a conscious one and the fact that it conforms to
an a priori principle, which links judgement and belief. Peacocke’s pro-
posal also avoids observationalism—both in classic Cartesian and more
modern reliabilist versions—because to say that a first-order mental state
is conscious does not mean, according to Peacocke, that it is an object
of (quasi-perceptual) attention, but, rather, that it occupies a subject’s
attention: being in a certain intentional state contributes to a subject’s
phenomenology—to what things are like from his subjective point of
view—despite the fact that the state is not presented to him as an object.5
Furthermore, contrary to Armstrong’s reliabilist account, on Peacocke’s
view, a first-order conscious mental state is taken to be a subject’s own
reason for making the corresponding self-ascription and not merely its
cause.6 Finally, Peacocke’s model is not a version of constitutivism, because
he denies both the constitutivist’s ontological claim that first-order men-
tal states depend for their existence on the corresponding self-ascriptions
and the constitutivist’s anti-epistemic claim that subjects’ psychological
3
According to Peacocke, a judgement that p may not always occur, in fact, but it should always be
available to a subject, in order for his psychological self-ascription to be justified (cf. Peacocke
1999, pp. 222–223; 241–242).
4
Peacocke 1999, p. 214.
5
Peacocke 1999, pp. 205–209.
6
Peacocke 1999, pp. 224–225.
102 A. Coliva
7
Peacocke 1999, pp. 230–231.
8
Peacocke 1999, pp. 232–233.
5 Epistemically Weak Accounts 103
and assessed. On his account, transparency in its strong form would hold
as well (see Chap. 3, §1). For these transitions from the mere awareness of
a given mental state to their self-ascription would be immediately and non-
inferentially made by conceptually endowed, attentive and lucid subjects.
Peacocke’s proposal thus relies crucially on the claim that first-order
mental states may be given to a subject in such a way as to function as rea-
sons, from his own point of view, for the corresponding self-ascriptions.
Hence, in order for the proposal to be implemented satisfactorily, he
needs a notion of a conscious mental state that can support the claim that
the corresponding self-ascription would be rationally justified merely by
its occurrence.
Concerning what it is for a mental state to be conscious, however, he
offers two, not obviously equivalent, proposals. One is that a mental state
is conscious if there is something it is like to have it.9 This suggestion
amounts to the idea that a mental state is conscious if there is something
subjectively distinctive about being in it. The other proposal is that a
mental state is conscious if it occupies our attention without being an
object of attention.10 This suggestion, in contrast, amounts to the idea
that a mental state is conscious if it is such that it can occupy our atten-
tion in the following sense: it is something whose nature, content, pre-
suppositions and theoretical and practical consequences we can attend to
without turning the state itself into an object of attention.11
9
Peacocke 1999, pp. 205–206.
10
Peacocke 1999, p. 207.
11
For instance, Peacocke (1999, pp. 209–210) writes: “I now attempt some further analysis of the
occupation of the attention by conscious thought. When you have a thought, it does not normally
come neat, unconnected with other thoughts and contents. Rather, in having a particular thought,
you often appreciate certain of its relations to other thoughts and contents. You have a thought, and you
may be aware that its content is a consequence, perhaps gratifying, perhaps alarming, of another conclu-
sion you have just reached; or you may be aware that its content is evidence for some hypothesis that you
have formulated; or that it is a counterexample to the hypothesis” (italics mine). “Now when you think
a particular thought, there is of course no intention in advance to think that particular thought.
But there can be an intention to think a thought which stands in a certain relation to other thoughts
or contents” (italics mine). This clearly suggests the idea that if a thought occupies one’s attention,
then, according to Peacocke, its content must be known to the subject who has it, as well as its
practical consequences—like causing one’s feeling alarmed or gratified—and theoretical ones—
such as its relation to other thoughts. As we shall presently see, however, this requires for Peacocke
too knowledge of the nature of one’s own mental states as well—that is, of their being beliefs as
opposed to wishes, imaginings, and so on.
104 A. Coliva
12
Actually, Peacocke talks in terms of “rational sensitivity” to one’s mental states. One way of
understanding this (rather vague) expression is in terms of “awareness”. However, for a different—
still problematical—interpretation, see the following paragraph.
13
Peacocke himself (1999, p. 216) makes the point: “Now the thinker who successfully reaches new
beliefs by inference has to be sensitive not only to the content of his initial beliefs. He has also to
be sensitive to the fact that his initial states are beliefs. He will not be forming beliefs by inference
from the contents of his desires, hopes, or daydreams”.
5 Epistemically Weak Accounts 105
16
I would like to thank Barry Smith for bringing this point to my attention. McHugh dubs it
“practical” in his 2012 paper. See previous footnote.
17
McHugh 2012 in fact seems to take it as a brute datum about certain kinds of creature.
18
Bonjour (2003, p. 62): “[Awareness of the kind and content of one’s own occurrent mental state
is] not in any way apperceptive or reflective in character: [it does] not require or involve a distinct
second-order mental act with the propositional content that I have the belief in question. Instead,
[it is] partly constitutive of the first-level state of occurrent belief ”.
5 Epistemically Weak Accounts 107
then the problems just reviewed will re-emerge. If, in contrast, it is not,
it is difficult to see how it can avoid collapsing into full-blooded knowl-
edge of one’s first-order mental state. For it may be true that when one
is consciously enjoying a first-order intentional state one does not actu-
ally have to judge that one has that mental state with that content.19
However, if that knowledge has to be somehow explicit, it seems that it
should be constitutive of having a conscious first-order mental state that
one should at least have the disposition to make that kind of judgement.
Yet, in this latter case, Bonjour’s proposal would be a mere notational
variant of more traditional constitutive accounts of self-knowledge,20 a
fact that would betray its whole point. Finally, if such constitutive aware-
ness is neither implicit knowledge nor constituted by the disposition to
judge that one has a given first-order mental state and merely consists in
the fact that the kind and content of that state are part of its phenom-
enology, then the proposal will turn into a version of Peacocke’s first,
phenomenological account of what it means for an intentional mental
state to be conscious and will be open to the criticisms we shall presently
level against it.
Let us now turn to Peacocke’s other, phenomenal gloss on what it
means for a mental state to be conscious.21 Here, too, problems arise. The
first one is that the work the proposal assigns to the phenomenology of
first-order intentional states now restricts it to the case of self-ascriptions
of occurrent mental states and rules it out for standing attitudes that lack
any phenomenological impact. Secondly, even in occurrent cases, the
capacity of the phenomenology of first-order mental states to carry the
resulting epistemological load falls into question as soon as one reflects
that, for example, occurrent desires, hopes or wishes with the same con-
tent need hardly be different from a strictly phenomenological point of
view yet will be required somehow to rationalise different self-ascriptions.
19
According to Bonjour (2003, pp. 65–68) that would either involve us in a vicious regress or never
provide an account of how a given mental state (first-order or otherwise) could ever be conscious.
20
Such as Bilgrami 2006, for instance. In his view, it is a necessary condition for having (a certain
class of ) first-order propositional attitudes that one knows them. There will be more on these
accounts in Chap. 7.
21
Further criticism of this aspect of Peacocke’s proposal can be found in Heal 2012; see also
Peacocke’s response in Peacocke 2012.
108 A. Coliva
22
This point is vigorously maintained, for instance, in McDowell 1994 and Brewer 1999. For
opposite views, see, for instance, Jim Pryor 2005.
23
Peacocke (1999, p. 216) writes: “In cases of consciously based self-ascription of attitudes and
experiences, a thinker […] makes a transition not only from the content of some initial state, but
also makes it because the initial state is of a certain kind (…). In the case of consciously based self-
ascription, the distinction between those events which are occurrent attitudes of the right kind to
sustain the resulting judgement and those which are not is a distinction which is conceptualised by
the thinker”. All this clearly implies that the relevant transitions are made on the basis of how the
first-order mental states are represented to the subject and not just on the basis of their phenome-
nology and this seems to ensure their rationality from the subject’s own point of view. Hence,
although, as Pryor has brought to my attention, Peacocke 2001 maintains that sensations—devoid
of any representational content—can immediately justify one’s corresponding self-ascriptions, he
does not seem to be inclined to offer an analogous account of intentional mental states and of their
self-ascriptions.
Furthermore, for reasons of internal coherence with his earlier work, I think Peacocke should
acknowledge that only representational contents—let them be psychological or otherwise—can
serve as rationalisers of judgements and, in particular, of self-ascriptions of intentional mental
states. Indeed, it is only on such an assumption that one can understand why, instead of defending
the so-called “Myth of the Given” against McDowell’s attacks, he elaborated a notion of non-con-
ceptual yet fully representational content for experiences (see Peacocke 1992, Chap. 4).
5 Epistemically Weak Accounts 109
24
Indeed, I think that it is very unclear how proponents of such a view could solve the problem just
mentioned and known as the ‘arbitrariness problem’ (see Pryor 2005, pp. 192–193). For instance,
Pryor’s own attempt to solve it by appealing to a notion of mental events which are themselves logi-
cally structured is dubious both from a metaphysical point of view, as Achille Varzi has remarked
to me, and from an epistemological one. For an internalist needs a justifier that is given to the
subject and that can play a rationalising role for his self-ascription from his own point of view. The
fact that an event might be, unbeknownst to him, logically structured and suited, in principle, to
rationalise the transition from its occurrence to its self-ascription, is of no use to the development
of a sound internalist epistemology of psychological self-ascriptions.
25
I need to emphasise that Peacocke 2003 does not explicitly consider the case of self-knowledge.
Rather, I am freely extending views he develops with respect to the relation between perceptual
experiences and empirical beliefs to the case of transitions from first-order mental states to second-
order ones.
110 A. Coliva
26
Strictly speaking, as Peacocke himself notices (2003, p. 26), the results of these transitions would
be only “relatively a priori” since they would be justified by the occurrence of particular mental
states. Still, they would not be inferred from them.
27
See Peacocke 2003, p. 11.
5 Epistemically Weak Accounts 111
28
See Peacocke 2003, pp. 12, 101, 177–178.
29
Notice that I am not claiming that the rationalising principles should be self-consciously or even
tacitly employed by a subject, even less that their truth should be appreciated by him in order to
have an account of self-knowledge which would be acceptable by internalist lights.
30
On pain of falling back into either horn of the consciousness dilemma.
31
See, in particular, Burge 1996 and 2011.
112 A. Coliva
justification, for instance, when you are able to say that you believe that
there is a red wall in front of you because you see it. Similarly, you have a
justification for believing that the sum of the internal angles of a triangle
is 180° because … where the dots should be replaced with a proof of that
theorem. Entitlements, on the other hand, are warrants for one’s beliefs
that need not be articulable by a subject and yet can have a rationalising
role with respect to the beliefs held on their basis, because they impinge
on their likely truth (at least in normal conditions). Hence, for instance,
a child has a perceptual entitlement for the belief that there is a red wall
in front of her, thanks to the kind of perceptual experience she is hav-
ing. She has it, even if she is unable to give her reasons for her belief,
perhaps because she lacks the ability to conceptualise her experience.32
Entitlements are thus externalist kinds of warrant.33
Moving on to self-knowledge, roughly speaking, one’s first-order men-
tal states, according to Burge, play the role of entitlements with respect
to one’s psychological self-ascriptions. Yet he thinks that entitlements do
not derive from the reliability of an inner scanning mechanism or from
the reliable operation of any other cognitive mechanism or procedure.
Such reliability is necessary in order to possess an entitlement, but not
sufficient. Entitlements derive, for Burge, from certain a priori consid-
erations concerning constitutive aspects of what it means to be a critical
thinker or a reasoner as well as a subject to whom moral norms may
apply34—whence the title of “rational externalism” for his proposal.
According to Burge, critical thinking is deployed when we provide
proofs for theorems, think through a plan, construct a theory or engage
in a debate. What happens in these cases is that, for instance, we start
out with a belief that P. We find out or are presented with conclusive
32
This is a very simplified account of perceptual entitlements. In fact, several a priori conditions
characteristic of perception will have to be met too, for Burge. His proposal, therefore, differs from
crudely reliabilist versions of perceptual justification (or warrant, in Burge’s terminology). See
Burge 1993 and 2010. For an assessment, see Coliva 2012a.
33
In Burge 2011 (p. 189), it is pointed out that although one can have a perceptual entitlement and
yet lack knowledge of the world around one, if, for instance, one is in a sceptical scenario, one can-
not have an entitlement for one’s psychological self-ascriptions and yet be wrong about one’s own
mental states. Burge dubs this second kind of entitlement “immune to brute error”.
34
Being a reasoner and a subject to whom moral norms apply are, according to Burge 2011, con-
stitutive features of what it means to be a subject.
5 Epistemically Weak Accounts 113
35
See Burge 2011, pp. 192–193.
114 A. Coliva
36
In this case, “I am thinking/judging that P” would be similar to the explicit performative “I
promise you to φ”, whereby one is promising to φ as well as saying that one is so doing. Thus, by
judging “I am thinking/judging that P”, one would thereby think or judge that P while also judging
that one is so doing.
37
Burge (2011, pp. 210–211) claims that “I hereby judge that P” is an impure cogito-like case and
that error is possible. Yet, if it occurred, it would signal a malfunctioning in a subject’s cognitive
capacities and that would actually defeat one’s entitlement for the relevant self-ascription.
5 Epistemically Weak Accounts 115
the perceptual case falter and it would then become mysterious what
would ground the relevant psychological self-ascriptions. This problem
may escape notice if one focusses on cogito-like thoughts, where the
second-order mental state brings the first-order one into existence. Yet it
is a problem when one aims to account along these lines for other kinds
of psychological self-ascriptions.
Another problem emerges in connection with Burge’s account of our
knowledge of our own sensations. For he seems to presuppose an account
of phenomenal concepts as tags for phenomental states, which is highly
dubious, as we saw while presenting Wittgenstein’s remarks against the
possibility of a private language or indeed of concepts designed directly to
refer to one’s phenomenologically salient and distinct experiences.
Moving on to more general worries, if we leave aside cogito-like
thoughts and think of first-order propositional attitudes and their self-
ascriptions as separate existences, to be told that thanks to constitutive
principles concerning what it is to be critical reasoners we have an entitle-
ment—that is, a particular kind of externalist epistemic warrant for the
relevant psychological self-ascriptions—does not tell us anything about
how we actually move from the former to the latter. That is to say, it
may still be the case that a subject self-ascribes the relevant propositional
attitude thanks to the subpersonal, therefore unconscious, operation of
a reliable inner mechanism.38 Burge’s proposal would then be entirely
compatible, from a merely epistemic point of view, with crude forms of
reliabilism such as Armstrong’s. That is to say, Burge could insist that the
kind of rational norms involved in accounting for our entitlement to self-
knowledge are a priori. Yet it is not precluded to a reliabilist to endorse
this claim and still insist that it is one thing to explain how knowledge
of our own propositional attitudes comes about and quite another to
insist that it is a priori true that first-personal self-knowledge is neces-
sary in order to engage in critical reasoning and to be subject to moral
appraisal. The latter does not account for the former, and the former is
epistemically significant whereas the latter is not—no matter how illumi-
nating it may be about the importance of self-knowledge with respect to
38
Indeed, Burge 1996, p. 103, fn. 12 allows that his account is compatible with the presence of “a
causal mechanism that relates attitudes to judgments about them”.
5 Epistemically Weak Accounts 117
Suppose you come home, and see that no car is parked in the driveway. You
infer that your spouse is not home yet … Later, you may suddenly remem-
ber that your spouse mentioned in the morning that the breaks of the car
were faulty, and wonder whether she may have taken the car for repair. At
this point, you suspend your original belief that she is not home yet. For
you come to realise that the absence of the car is not necessarily good evi-
dence that she is not home. If the car is being repaired, she would have
returned by public transport. Then finally you may reach the belief that she
is home after all, given your next thought that she would not have taken
any risks with faulty breaks.
39
A similar worry can be found in Peacocke 1996, Moran 2001 (pp. 109-113), Bar-On 2004 and
Cassam 2014.
40
See Burge 2011, p. 192.
41
Furthermore, it would be dubious that only creatures capable of such a robust form of critical
reasoning would count as selves and that, in turn, would make trouble for Burge’s vindication of
our entitlement to self-knowledge, which heavily depends on constitutive claims about what it
means to be a self.
5 Epistemically Weak Accounts 119
42
Although Evans did not develop his proposal, he actually maintained his model for self-knowl-
edge of beliefs could be extended to other cases of first-personal knowledge of our own mental
states. Hence, he embraced monism. (See Evans 1982, p. 225).
120 A. Coliva
only beliefs but also desires. We will consider only the former for the objec-
tions we will raise against it could be pari passu repeated against the latter.
Fernández’s central thesis is that one’s psychological self-ascriptions are
based on whatever grounds support the corresponding first-order beliefs.
Therefore, the former are at least prima facie justified if one’s beliefs are
as well. Usually they are, if they follow the “production-of-belief prin-
ciple” (2013, p. 46, 55): if one apparently perceives/remembers/acquires
testimonial information/has an intellectual seeming that P, one comes to
believe P; and so does one if one believes Q and believes that Q entails P.
Fernández then addresses the objection that the bypass model is a bet-
ter explanation of how one generates or forms one’s beliefs rather than of
how we know pre-existing ones. Fernández concedes that many times that
is the case, but he contends that there is also a genuinely doxastic reading
of the question “Do you believe that P?” whose answer depends on the
application of the bypass model. He considers a case of a client who asks
his lawyer whether she believes him innocent. He writes (2013, p. 51):
It is not easy to see how this could be a response to the original challenge.
For the lawyer in this case seems actually to be making up her mind
regarding the innocence of her client, by considering evidence for and
against it.
Nevertheless, Fernández then explains how the bypass model satisfies
groundlessness and authority. It does satisfy weak groundlessness because
we enjoy certain mental states (like experiences, memories, etc.) which
naturally and usually lead us to form the corresponding beliefs as well as
second-order self-ascriptions, without any need to observe our own overt
behaviour and without having to reason from the basing mental state to
the second-order self-ascription. By contrast, other people will be in a posi-
tion to ascribe a belief to us only by observing our overt behaviour and by
inferring to its likely cause. If, however, our self-ascriptions are not based
on outer observation and inference to the best explanation, they are not
open to the kinds of error that can beset third-personal ascriptions of belief.
5 Epistemically Weak Accounts 121
Hence, our own access to our own beliefs is different and more secure
than whatever access other people might have with respect to them. For
this reason, our self-ascriptions of belief are at least weakly authoritative.
Fernández also claims that the bypass model is not causal, because the self-
ascription is not causally brought about by the first-order belief, but, rather,
it is grounded on the evidence on which that very belief is based. This
guarantees that also groundlessness in the stronger form is respected since
first-order mental states do not play the role of evidence for the relevant
self-ascriptions.43 Finally, transparency is guaranteed in its strong version
because provided one has the relevant concepts and the relevant first-order
mental states, one will immediately be in a position to self-ascribe them.
Two remarks are in order: first, it seems odd to say that one’s self-ascriptions
of belief should be justified by the evidence that prompts the first-order
belief itself. Intuitively, the self-ascription, if justified at all, should be justi-
fied by the corresponding first-order mental state. This would indeed match
the intuition that although we may have unjustified first-order beliefs, if
they were formed on inappropriate, scant or no grounds, we could never-
theless be justified in self-ascribing them, for the simple reason that we do
have them. To rescue this compelling intuition, Fernández has to say that
although the subject would have evidence but no justification for the first-
order belief, that very evidence would count as a justification for her self-
ascription. Now, besides the difficulty of distinguishing between evidence
and grounds in a satisfactory way, it is clear that this could be so only if we
considered the role of that evidence in giving rise to the first-order belief
which then would tend to correlate with one’s self-ascription of it. However,
if that is the case, in the end what “justifies” one in self-ascribing the belief is
precisely the first-order belief itself, not the evidence that prompts it.
Second, it is difficult to see how the bypass model would count as
non-causal. For, after all, it heavily depends on the fact that once we are
exposed to certain kinds of evidence, we tend to form the corresponding
first-order beliefs and, if conceptually equipped, we tend to form the rel-
evant psychological self-ascriptions. True, if Fernández’s story were right,
43
However, it may be objected, as Giorgio Volpe observed, that Fernández’s account violates a
particular form of groundlessness—namely, one which would ban that the relevant psychological
self-ascription be based on any kind of evidence or grounds whatsoever, even those which would
ground the first-order belief in P. I agree with this objection, but at least the restricted versions of
groundlessness we introduced in Chap. 3 would be respected.
122 A. Coliva
44
For a discussion of further aspects of Fernández’s proposal, see Coliva 2014.
5 Epistemically Weak Accounts 123
45
In Moran 2012 (p. 235), he calls this form of agency “agency as responsiveness to reasons” as
opposed to “agency as production”. The latter is “at will”, the former is constrained by epistemic
reasons. It is a form of agency nonetheless.
46
Moran, however, would probably reject this rendition of his ideas because he wants to preserve a
unitary meaning to “belief ” (see Moran 2001, pp. 83–94). For reasons exposed in Chap. 2 (§2), I
think that we can explain why commitments and dispositions are two species of the same genus
“belief ”.
124 A. Coliva
47
Like Fernández’s account, Moran’s would not respect an unrestricted version of groundlessness
(see fn. 42). For our purposes, though, it is sufficient that it respects indeed both the weak and
strong version of it we have been discussing throughout this book.
48
Implausibly, Cassam 2014 raises this objection to theories that hinge on the application of the
transparency method. For a discussion, see Coliva 2015b. See also Boyle 2011a and 2015. However,
his gloss (Boyle 2011a) on Moran’s position as entailing tacit self-knowledge of one’s first-order
mental states, which needs be made explicit only through reflection, seems to betray the gist of
Moran’s non-evidential account.
5 Epistemically Weak Accounts 125
49
It is difficult, though, to see how psychological self-ascriptions reached through the transparency
method could be wrong. For the subject will have the relevant judgement-dependent attitude, if
she has formed it by considering reasons in favor of its content. Still she may also exhibit a behvior
which at least partially goes against her psychological self-ascription. In that case, it would seem
that Moran’s account would be more consistent with treating cases of self-deception as due to the
presence of two conflicting mental states, only one of which is formed through deliberation and is
self-known by application of the transparency method. There will be more about this account of
self-deception in Chapter 7 (§§3-4).
50
Moran acknowledges this charge in his exchange with Sydney Shoemaker and Lucy O’Brien. See
Moran 2003, especially fn. 4 and pp. 409–417.
126 A. Coliva
different from the ones in which we would know of other people’s mental
states or of those aspects of our minds which are not of our making.51
Along these lines, one might even suggest that a better title for his book
would have been Authorship and Estrangement, leaving aside the issue of
how we can have any special knowledge of (some of ) our own mental
states and thus epistemic authority over them.
In response to this kind of worry, Moran has replied that he welcomes
the suggestion that his account should be completed by endorsing the
idea of an entitlement as a non-discursive kind of warrant.52 In his view,
having an entitlement “is not so much a matter of concept possession, as
a matter of possession of the relevant practical and cognitive abilities”.53
I suppose that what this means is that, in his view, entitlements in this
area would not be a priori arguments based on reflection on certain
concepts, such as the notion of being a responsible agent or a reasoner,
which would render one’s psychological self-ascriptions warranted, even
if one were not able to entertain them. Rather, according to Moran, it
seems that a subject will count as being entitled to her psychological
self-ascriptions just as long as she is actually able to make up her mind
on the basis of weighing reasons for or against P (or P would be good to
have), is able (in the case of desires and intentions as commitments) to
pursue certain courses of action for the time necessary to realise them,
and is able to apply the concepts which are necessary in order to make
the self-ascription.
Hence, a fair rendition of Moran’s account, I think, would be the
following one. Suppose someone asks us “Do you believe that P?”. We
can answer that question by deliberating what to believe on the basis of
assessing evidence for or against P directly. Hence, we can answer “yes”
(or “no”) to a question about our beliefs without having to rely on any
introspection and just by considering reasons for or against P itself. If
we then say or judge “I believe that P”, that should be considered an
avowal—that is, a direct expression—of our having deliberated that P is
51
This objection has been raised by several scholars (such as Bar-On 2004 and Shah and Velleman
2005). For a defence of Moran’s position, see Boyle 2009.
52
This is the gist of his response to Lucy O’Brien’s observation. See Moran 2003, p. 412.
53
Ibid.
5 Epistemically Weak Accounts 127
54
Or at the very least he seems to hold that this kind of self-knowledge is more fundamental than
our self-knowledge of sensations, perceptions, and so on. (For a similar objection, see Finkelstein
2003.) Boyle 2009, who is sceptical of the claim that genuinely first-personal self-knowledge
should be equated with deliberative self-knowledge of one’s propositional attitudes, is however in
favour of the claim that it is more fundamental than any other kind of self-knowledge because it is
inherent to our notion of belief. The latter claim is specially defended in Boyle 2011b.
128 A. Coliva
4 Summary
In this chapter, we have looked at weak epistemic accounts of self-
knowledge. We have started with Peacocke’s rational internalism (§1),
whose characteristic claim is that the conscious occurrence of first-order
mental states grounds one’s psychological self-ascriptions and rationalises
them, without thereby requiring a subject to have observational or infer-
ential knowledge of them. We have shown that various possible glosses of
the notion of what it is for a first-order mental state to be conscious create
problems for Peacocke’s account, either because they would make it cir-
cular or because they would be unsuited to play a genuinely rationalising
role with respect to the relevant self-ascriptions.
We have then looked at Burge’s rational externalism (§2). We have
found it wanting because it relies on an idea of entitlement whose epis-
temic credentials are unclear. If it is modelled after the role of perceptual
experiences vis-à-vis judgements about the outer world, it risks reinstating
the idea that one’s first-order mental states should be conscious, to play a
warranting and rationalising role with respect to the relevant psychologi-
cal self-ascriptions. If it is construed as an a priori argument that should
link the ability to be critical reasoners with having knowledge of our own
mental states, it would actually be compatible with a reliabilist account
of how the latter is achieved. Moreover, on reflection, it is doubtful that
we cannot conceive of a critical reasoner who does not have propositional
5 Epistemically Weak Accounts 129
knowledge of her own mental states. Hence, the required a priori connec-
tion between these notions is dubious. Or else, if it is not, it is because an
ad hoc notion of being a critical reasoner is being presupposed.
We have then moved on to a different kind of proposal which gives
centre stage to Evans’s so-called “transparency method” (§3). According
to that method, in order to get knowledge of our own beliefs, we should
not direct our eyes inward, to introspect our own minds to see whether we
can find certain mental states which we would then self-ascribe. Rather,
we have to direct our gaze outward, to determine whether it is the case
that P. If we find (what we consider) sufficient evidence for P, then we
can simply prefix P with “I believe that”, thereby making the relevant
self-ascription. We have considered two prominent ways of develop-
ing Evans’s method: one more epistemically oriented, recently put for-
ward by Jordi Fernández (§3.1), and another more deliberative-oriented
maintained by Richard Moran (§3.2). The former has been criticised for
implausibly claiming that the evidence which justifies one’s belief in P
would also justify one’s self-ascription of that belief and also because it
would have difficulties in accounting for knowledge of beliefs based on
no evidence in favour of their contents and for self-ascriptions of open-
mindedness with respect to P. The latter, in contrast, has been found
wanting because it identifies first-personal self-knowledge with the ability
to deliberate and make up one’s mind, when intuitively we would like
to account also for the special way in which we do know our ongoing
sensations, emotions and perceptions. Moreover, we have argued that,
its merits notwithstanding, it does not have anything to offer in the way
of explaining how self-ascriptions of beliefs and other propositional atti-
tudes, which are the result of rational deliberation, would amount to
knowledge, despite its ambition to present an alterntive to non-epistemic
accounts of self-knowledge.
Given the difficulties which beset weak epistemic accounts of first-
personal self-knowledge, as well as the strong epistemic accounts we
reviewed in the previous chapter, the time is ripe for considering accounts
that somehow dispense with the idea that what goes by the name of first-
personal self-knowledge is actually the result of some peculiar cognitive
achievement and for reckoning with the fact that it is perhaps a misno-
mer to call it “knowledge” after all.
130 A. Coliva
Bibliography
Bar-On, D. (2004). Speaking my mind. Expression and self-knowledge. Oxford:
Oxford University Press.
Bilgrami, A. (2006). Self-knowledge and resentment. Cambridge, MA: Harvard
University Press.
Bonjour, L. (2003). A version of internalist foundationalism. In L. Bonjour &
E. Sosa (Eds.), Epistemic justification. Internalism vs. externalism, foundations
vs. virtues (pp. 61–68). Oxford: Blackwell.
Boyle, M. (2009). Two kinds of self-knowledge. Philosophy and Phenomenological
Research, 77(1), 133–164.
Boyle, M. (2011a). Transparent self-knowledge. Aristotelian Society Supplementary
Volume, 85(1), 223–241.
Boyle, M. (2011b). Making up your mind. Philosophers’ Imprint, 11(17), 1–24.
Boyle, M. (2015). Critical study of Quassim Cassam Self-knowledge for humans.
The European Journal of Philosophy, 23(2), 337–348.
Brewer, B. (1999). Perception and reason. Oxford: Oxford University Press.
Burge, T. (1993). Perceptual entitlement. Philosophy and Phenomenological
Research, 67, 503–548.
Burge, T. (1996). Our entitlement to self-knowledge. Proceedings of the
Aristotelian Society, 96, 91–116. Reprinted in P. Ludlow and N. Martin (eds.)
Externalism and Self-Knowledge, Stanford, CSLI Publications, 1998,
pp. 239–264.
Burge, T. (2010). Origins of objectivity. Oxford: Oxford University Press.
Burge, T. (2011). Self and self-understanding. The Dewey lectures (2007–2011),
The Journal of Philosophy, CVIII/6–7. Reprinted in Cognition through under-
standing (pp. 140–228). Oxford: Oxford University Press.
Cassam, Q. (2014). Self-knowledge for humans. Oxford: Oxford University Press.
Coliva, A. (2012a). Critical notice of Tyler Burge Origins of objectivity. Disputatio,
4(33), 515–530.
Coliva, A. (2014). Review of Jordi Fernández Transparent minds. Theoria, 81,
442–445.
Coliva, A. (2015b). Review of Quassim Cassam Self-knowledge for humans,
Analysis.
Evans, G. (1982). The varieties of reference. Oxford: Oxford University Press.
Fernández, J. (2013). Transparent minds. Oxford: Oxford University Press.
Finkelstein, D. (2003). Expression and the inner. Cambridge MA: Harvard
University Press.
5 Epistemically Weak Accounts 131
1 At the Origins of Expressivism:
Wittgenstein
Wittgenstein is rightly considered at the origins of expressivism in the
philosophy of mind, especially for his annotations in the Philosophical
Investigations and in the Remarks on the Philosophy of Psychology (vol. 2).
As we shall see, he can also be seen at the origins of constitutive models,
at least to some extent. It is therefore appropriate to expound on his
views. His positive proposal, however, is better appreciated if it is seen
in connection with his criticism both of the introspective model and of
behaviourism.
On the one hand, Wittgenstein attacks introspectionism because it
would lead to a language of sensations and, more generally, of mental
states, which would be private. A private language, however, is, for him,
a category mistake. For the rules of this putative language would be just
“impressions of rules” (PI 259), hence no rules at all. Discussing the pos-
sibility of a subject’s keeping a diary of his sensations, named after a pro-
cedure of private ostensive definition, Wittgenstein writes: “whatever is
going to seem right to me is right. And that only means that here we can’t
talk about ‘right’” (PI 258). That is to say, if “S” is the name I give to a
sensation of mine, that no one else can have access to, its use over time
counts as right whenever I would think so, even when I may use it to
refer to a different kind of sensation or to nothing at all. This, however,
simply means that the distinction between correct and incorrect uses of
6 Expressivism About Self-knowledge 135
“Are you not really a behaviourist in disguise? Aren’t you at bottom really
saying that everything except human behaviour is a fiction?”—If I do speak
of a fiction, then it is of a grammatical fiction. (PI 307)
Again:
“So you are saying that the word ‘pain’ really means crying?”—On the
contrary: the verbal expression of pain replaces crying and does not describe
it. (PI 244)
“But doesn’t what you say come to this: that there is no pain, for example,
without pain-behaviour?”—It comes to this: only of a living human being and
what resembles (behaves like) a living human being can one say: it has sensa-
tions; it sees; is blind; hears; is deaf; is conscious or unconscious. (PI 281)
The human body is the best picture of the human soul. (PI II, iv, p. 178e)
The idea is that only of a living human being (and more generally of a
living animal being that behaves similarly to us), contra a human corpse
or a zombie, can we correctly say that it does have sensations and feelings
and can thus teach a subject to express them in linguistic terms.
138 A. Coliva
4
By strong contextualism, I have in mind what is nowadays called “semantic eliminativism”, a view
maintained by Wittgenstein and, more recently, by Charles Travis. I do not mean anything having to
do with forms of assessment sensitivity. Nor is the label “contextualism”, as used here, indebted to
Kaplan’s notion of context and to Grice’s distinction between semantics and pragmatics. Crudely put,
for Wittgenstein, the notion of context roughly coincides with that of language game and certainly is
not exhausted by the triad subject-place-time or any suitable extension of it. Moreover, given his equa-
tion of meaning and use, there is no room for a sharp division between semantics and pragmatics.
140 A. Coliva
wrong, and in another nonsense. 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 cer-
tainty 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 them.
I have them. 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. (PI 246; cf. PI
II, xi, p. 221e)
5
For an extended examination, see Coliva 2010, Chap. 2.
6 Expressivism About Self-knowledge 141
just repeating that one is in pain? (1) would thus be flouted. One might
think that the reason is the sensation itself. Surely, Wittgenstein would
have objected to that, since, in his view, sensations as such cannot be rea-
sons for their ascriptions. As he famously wrote (PI 289):
Clearly, the question must be answered in the negative, for him, and
in fact, he goes on to say:
6
Hanoch Ben-YAmi has kindly pointed out to me that the correct translation would be “incor-
rectly”, rather than “without right”. This would further support my reading of Wittgenstein and
my criticism of epistemic interpretations.
7
See, in particular, Wright 2004 and Williams 2004a, b.
142 A. Coliva
has pain and gives voice to it. Furthermore, a claim like “I know I am
in pain” would—in the normal run of cases—be totally irrelevant (thus
flouting (4)) and indeed odd in the course of a conversation, up to the
point that, if no proper context for its occurrence could be provided,
one would start doubting whether a subject is in her right mind, or
would start thinking that perhaps she is trying to make a joke, and so on.
Finally, in the normal run of cases, it does not make sense to suppose that
one might not know whether one is in pain (contra (5)). Of course, there
can be cases of confused sensations—an itch which borders pain—but
the phenomenon of vagueness cannot support the idea that when we
experience sensations we may not know that we do. As to the possibility
of unconscious sensations, they would be a conceptual solecism in this
connection. Thus, for Wittgenstein, the use of “I know” in connection
with one’s avowals would not make sense.
Still, there is a grammatical sense of “I know” which would be appro-
priate in connection with one’s psychological avowals. Again, the exis-
tence of such a grammatical use of “I know”, for Wittgenstein, is clear
from his remarks in On Certainty. Two aspects of the grammatical use of
“I know” are particularly relevant with respect to avowals—namely, (1′)
the fact that “I know” would actually mean “A doubt is excluded” or “I
cannot be wrong” (OC 59) and (2′) that the “I” is unimportant (OC 58).
Here are the relevant passages in the Philosophical Investigations:
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. (PI 246)
And with respect to (2′), the idea that “I˝ is not a referential expression
when used in psychological avowals, which is present in Wittgenstein
since his remarks in the Blue Book (pp. 66–67), makes its appearance
again in PI 404–411. Here are some passages:
In saying this [“I am in pain”] I don’t name any person. Just as I don’t name
any person when I groan with pain. Though someone else sees who is in
pain from the groaning. (PI 404)
“But you aren’t in doubt whether it is you or someone else who has the
pain!”—The proposition “I don’t know whether I or someone else is in
pain” would be a logical product, and one of its factors would be: “I don’t
know whether I am in pain or not”—and that is not a significant proposi-
tion. (PI 408)
“I” is not the name of a person, nor “here” of a place, and “this is not a
name”. But they are connected with names. (PI 410)
“Only you can know if you had that intention.” One might tell someone
this when one was explaining the meaning of the word ‘intention’ to him.
For then it means: that is how we use it. (And here “know” means that the
expression of uncertainty is senseless). (PI 247)
144 A. Coliva
contrast, are internally related to their contents. They are not the causes of
the actions that fulfil them (PI 632, cf. BlB 23–24). In his view, moreover,
intentional actions and willing do not stop short of the action willed (PI
622), although there may be cases in which trying and making an effort to
bring about the action are relevant. Nor are they individuated by specific
(bodily) feelings (PI 625), although they may be accompanied by them.
When it comes to our knowledge of intentions and actions (as well
as of beliefs and desires), it depends on the circumstances. In the usual
run of cases, we express, avow or voice our own intentions. In that case,
there is no real epistemology of the mental, for the reasons we rehearsed
in connection with our putative knowledge of our own sensations. That
is, the criteria for the empirical and therefore meaningful use of “I know”
are flouted. To go over them once more: we do not have independent
reasons for “I know I ψ that P” (where ψ is a propositional attitude verb
like intending, desiring, believing, hoping, willing, etc.)—that is, reasons
other than the very mental state itself (contra (1)). A fortiori, even if one
conceded that the mental state could be a reason for our knowledgeable
self-ascriptions, it would not be any stronger a reason than what it is sup-
posed to ground (contra (2)). We do not find out our intentions, beliefs,
and so on. We do have them and give immediate expression to them by
saying or thinking “I intend to φ” or “I /believe/desire that P”, or indeed
we form them through those very self-ascriptions. We do not discover
them either through introspection or inference to the best explanation
given our behaviour or feelings, which may, at least on occasion, accom-
pany the occurrence of these mental states (contra (3)). To say “I know I
ψ that P” must be relevant, but this is not the case in ordinary circum-
stances. In particular, asserting it would not add anything to one’s simple
self-ascription of the relevant mental state (contra (4)). Finally, given the
conditions which are supposed to hold, it would not make sense to say
or judge “I do not know whether I ψ that P” (contra (5)). For that kind
of judgement could meaningfully occur only in a context in which I do
not have yet formed the (specific) intention (or the belief or the desire).
Now, if the expressivist story has to get purchase in relation to propo-
sitional attitudes like intending, believing and desiring, it will have to be
the case that the relevant self-ascriptions are learned as ways of substitut-
ing pieces of instinctive behaviour. The question arises, however, whether
6 Expressivism About Self-knowledge 147
The idea then is that also when it comes to wanting, intending and desiring
there are characteristic primitive reactions. For instance, a child wants to
have a glass of water and tries to reach for it and the adult sees her doing so
and says “You want a glass of water. I’ll give it to you”. Again, a child puts
her coat on and stands by the door. The adult says “You want to go out.
We’ll go in a moment”. Little by little, the child is taught to replace her
primitive behaviour with pieces of linguistic vocabulary, up to the point
where she herself is in a position to say “I want a glass of water/to go out”.
The case of belief, however, is certainly more complicated. For there do
not seem to be specific primitive reactions which would signal one’s believing
that today it is sunny or that one’s favourite toy is soft and tender. In particu-
lar, even if the child reaches for her toy and holds it close to her face, we do
not normally teach her to replace that behaviour with a linguistic one that
makes explicit the attitude of belief. That is to say, we do not say “You believe
that your teddy bear (say) is tender”, even if, to us, the child’s behaviour man-
ifests that attitude. We simply voice the embedded content of the attitude. It
is perhaps for this reason that when Wittgenstein explicitly talks about self-
ascriptions of belief in the present tense, he takes them to be equivalent to the
assertion of their embedded content, just in a more tentative way. In PI II,
x, while addressing Moore’s paradox—“I believe it is raining, but it isn’t”—
Wittgenstein observes, first, that “I believe that” is “transparent” (in Evans’s
sense of the term). For “The expression ‘I believe that this is the case’ is used
like the assertion ‘This is the case’” (p. 190e). Perhaps, sometimes, as a “hesi-
tant assertion [of P]” (p. 192e). This leads him to hold that “to believe” has a
different meaning when used in the first-person present, as opposed to in the
first-person past, or within a suppositional context, or in the third person.
Here are some relevant passages, whose significance is better appreciated by
keeping in mind Wittgenstein’s equation of meaning and use (PI 43, 560):
148 A. Coliva
Moore’s paradox can be put like this: the expression “I believe that this is
the case” is used like the assertion “This is the case”; and yet the hypothesis
that I believe this is the case is not used like the hypothesis that this is the
case. (PI, II, x, p. 190e)
Hence, the idea seems to be that since √1 admits of two possible solu-
tions—1 and −1—both of which belong to real numbers, whereas √−1
does not admit of a solution within the domain of reals, “√” means some-
thing different in the two cases. Although the solutions of the application
of the function are different depending on whether it is applied to 1 or
−1, and indeed in order to provide a solution to √−1 we need to extend
numbers to imaginary ones whereas no such extension is needed in order
to provide a solution to √1, that does not mean that the function itself
is ambiguous, even if its application requires something very different of
us in the two cases. Thus, the analogy provided by Wittgenstein to sup-
port a strong form of contextualism about meaning is dubious. That does
not ipso facto prove his semantic contextualism wrong, though. Yet it can
make us suspicious of it. As we shall see, this is a thesis contemporary
expressivists do not generally hold.
Notwithstanding Wittgenstein’s strong form of contextualism about
linguistic meaning, there is something else worth noticing—namely, the
different function the same words can have depending on context, con-
ceding that they may retain the same linguistic meaning. He writes:
This is how I think of it: Believing is a state of mind. It has duration; and
that independently of the duration of its expression in a sentence, for
6 Expressivism About Self-knowledge 149
Does it make sense to ask “How do you know that you believe?”—and is
the answer: “I know it by introspection”? In some cases it will be possible to
say some such thing, in most not. (PI 587)
equation between meaning and use, comes down to the insistence on the
different meaning the same words can have. Here are some relevant passages:
When I sat down on this chair, of course I believed it would bear me. I had
no thought of its possibly collapsing”. But: “In spite of everything that he
did, I held fast to the belief …”. Here there is thought, and perhaps a con-
stant struggle to renew an attitude. (PI 575)
Similarly,
When someone says “I hope he’ll come”—is this a report about his state of
mind, or a manifestation of his hope?—I can, for example, say it to myself.
And surely I am not giving myself a report. It may be a sigh; but it need
not. If I tell someone “I can’t keep my mind on my work today; I keep on
thinking of his coming”—this will be called a description of my state of
mind. (PI 585)
say at the end of a quarrel ‘All right! Then I leave tomorrow!’; I make a
decision”. References could be multiplied ad libitum. They all go in the
direction of distinguishing between different uses or functions the same
words can have and, at least in the case of self-ascriptions of intentions,
between the self-ascription of a disposition and of (something close to)
a commitment, with the attendant diagnosis of their underlying episte-
mology. None in the latter case, and third-personal in the former.
We have now examined at length Wittgenstein’s expressivism. We
can thus sum up how it fares with respect to an adequate theory of
self-knowledge. In the case of so-called first-personal self-knowledge, it
respects groundlessness, indeed in a strong form, since it not only does
steer away from the observational and the inferential account of self-
knowledge but does not consider occurrent mental states as evidence or
reasons for the corresponding self-ascriptions either.
It also respects transparency, for on this account, any mental state
which can be expressed by means of an avowal is a mental state a subject
is aware of, even if it is not named, described or reported by means of
the corresponding self-ascription and even if its verbal expression presup-
poses a blind substitution of a piece of instinctive and primitive behav-
iour with a characteristic linguistic expression, which is considered the
immediate expression of it.
Finally, authority is respected as well, although it is seen as a prod-
uct of our linguistic practice of according it to subjects. That is to say,
once the C-conditions are satisfied, then there is no room left for doubt
regarding the fact that a subject does have the mental state she actually
avows. Whether authority is satisfied in its weak or strong version in
Wittgenstein’s account is not clear, because he gives no explanation of
self-deception. Yet, at the very minimum, it holds in its weak version.
The problems with Wittgenstein’s expressivism, however, are mostly
related to the kind of semantics of avowals it provides for, such as the
endorsement of a strong form of semantic contextualism and the dif-
ficulty in accounting for the fact that, after all, even when uttered in the
avowing mode, these sentences have a syntactical structure that makes
them truth-apt. In connection with avowals of belief and other propo-
sitional attitudes, moreover, it is not clear what kind of pre-linguistic
behaviour they would substitute. Finally, their equation with the asser-
152 A. Coliva
2 Bar-On’s Neo-expressivism
Bar-On, in her Speaking My Mind: Expression and Self-Knowledge (2004),8
has presented the best-developed expressivist account of self-knowledge
to date. Her view inherits the key idea of Wittgenstein’s expressivism—
that is, that psychological self-ascriptions are avowals, which express one’s
occurrent mental states and are pieces of linguistic behaviour that replace
more natural expressions of ongoing mental states. Yet she thinks this cen-
tral contention can be coupled with a semantic and an epistemic account
whereby the relevant self-ascriptions are truth-apt and are expressions of
privileged self-knowledge.
First, it is important to stress that the kind of more instinctive behav-
iour avowals substitute can be both non-verbal, as in the case of pain, and
verbal. Hence, “I want o” can be acquired by learning to use it instead of
one’s pointing towards an object o while pronouncing its name.9 More
importantly, in the case of self-ascriptions of propositional attitudes, like
belief, “I believe that P” will be acquired as a substitution for one’s assert-
ing that P, mostly in circumstances when one gets corrected by an adult
about the fact that P is not actually the case, even if it seems so to one.10
Second, Bar-On’s idea is that “I intend to φ” and “I believe that P”,
as avowals, express the relevant mental states but have a content which is
truth-evaluable. Consider an explicit performative like “I promise to take
you to the zoo”. It both expresses one’s promise and makes explicit the fact
that one has made such a promise. Hence, the resulting self-ascription is
truth-apt. In particular, it is true if the subject actually intends to take her
child, say, to the zoo and false otherwise. Similarly, according to Bar-On,
“I intend to φ” and “I believe that P” express one’s intention and belief,
8
Another prominent version of neo-expressivism can be found in Finkelstein 2003.
9
See Bar-On 2004, p. 288.
10
See Bar-On 2004, p. 294.
6 Expressivism About Self-knowledge 153
respectively, and are true iff one actually intends to φ and believes that
P. We should then distinguish between the act of avowing one’s intention
or belief and the product of that act, which is a self-ascription that can be
either true or false.11 According to Bar-On, these self-ascriptions can also
express one’s judgement of intending to φ or of believing that P. Yet, even
in this case, the self-ascription is not epistemically based on the relevant
first-order mental state. That is, the mental state elicits one’s intentional
self-ascription and even one’s judgement of being in it and therefore is their
rational cause, but it is not a kind of evidence on which the judgement is
based.12 Notice, moreover, that contrary to explicit performatives “I intend
to φ” and “I believe that P” are not ways of bringing about the relevant
first-order mental states, according to Bar-On. Rather, they are expressive
of them. Hence, she refuses the idea that one’s psychological self-ascrip-
tions can play a constitutive role with respect to one’s first-order mental
states (save for cases of self-verifying judgements, like the ones we reviewed
in our presentation of Burge’s views in Chap. 5, §2). In this respect, she is
entirely with common sense in recognising the independent existence of
mental states from their self-ascription (save for self-verifying thoughts).
Still, recognising a subject’s act as an avowal—that is, as a self-expressive
act—entails “presuming her avowal to be true”.13 For, for an intentional
linguistic act to count as an avowal, it must be prompted by the mental
state avowed. In fact, according to Bar-On, the avowal literally shows the
relevant mental state and thus makes it visible or audible to one’s audience.14
Avowals are therefore characterised by a transparency-to-the-subject
and to-the-world condition. The latter is the analogue of Evans’s notion
of transparency. A subject does not have to look within herself to answer a
question about her belief that P (and other propositional attitudes) but at
the reasons for (or against) P (or its desirability, etc.). Similarly, if an audi-
11
I think this point is not sufficiently appreciated by Boyle (2009), who criticises Bar-On’s proposal
for not explaining the difference between merely parroting “I φ that P” (where φ has to be substi-
tuted with a propositional attitude verb) and actually representing oneself as φ-ing. Along Bar-On’s
lines, see also Finkelstein 2003, Chap. 4.
12
There will be more on how the mental state can play a rationalising role with respect to the self-
ascription in the following.
13
See Bar-On 2004, p. 317.
14
See Bar-On 2004, pp. 417, 422–423.
154 A. Coliva
ence wants to challenge a subject’s avowal, they will take issue with the
reasons for (or against) P itself, not with the fact that the subject has self-
ascribed that belief (or some other intentional mental state). The former,
in contrast, is the presumption of truth we have already reviewed together
with the more problematic claim (as we shall see) that avowals wear the
relevant mental states “on their sleeve”.15 Hence, avowals do not play the
role of mere symptoms of those mental states that supposedly cause them,
according to Bar-On. Thus, one’s audience does not have to infer from the
former to the latter through an inference to the best explanation. Rather,
an audience can directly perceive the mental state in a subject’s avowal.16
Therefore, mental states are “conditions the subjects are in, not states that are
in the subjects”, made directly manifest in their avowals.17
It is worth pointing out that, according to Bar-On, the presumption of
truth avowals enjoy can be overridden. Hence, she is in sharp disagreement
with more traditional expressivist views “that maintain that a subject’s avow-
ing a mental condition conceptually guarantees her being in the condition (so
the avowal must be true), as long as the subject is linguistically or conceptu-
ally competent and sincere”.18 Consider someone who is so scared of the
dentist that as soon as he approaches her with his instruments she screams
“I am in pain” or “It hurts”. Let us suppose that the patient is being fully
sincere. Yet she cannot be in pain, for she has not even been touched. Her
avowal is thus false. Still, according to Bar-On, it is not based on a misiden-
tification of one’s ongoing mental state as pain. The subject has not made
a mistake but has committed an “expressive failure”.19 As Bar-On puts it,
though [the subject] has successfully expressed pain, she has not succeeded
in expressing her pain. She could not have expressed her pain, since there
was no pain for her to express. The subject has used an expressive tool that
is referentially associated with one condition (…) to give vent to a different
condition.20
15
See Bar-On 2004, p. 315.
16
See Bar-On 2004, p. 423.
17
Bar-On 2004, p. 424.
18
Bar-On 2004, p. 325.
19
Bar-On 2004, p. 320.
20
Bar-On 2004, p. 323.
6 Expressivism About Self-knowledge 155
21
See Bar-On 2004, p. 335. This position is not very plausible. Suppose someone is being tortured
and screams “I don’t want this”. It would be weird to say that this is not an avowal but a judgement
based on introspection. Yet it would be equally weird to hold that it is based on some other positive
mental state like wanting something other than what one is being inflicted.
22
See Bar-On 2004, p. 301.
156 A. Coliva
self can be in M and therefore have the relevant entitlement, the account
does explain not only how she has knowledge of her mental states but also
why that knowledge is privileged—that is, available only to the subject
herself. To put it differently, the very truth-maker of one’s psychological
self-ascription is also what warrants it. Now, in the case of false but sincere
avowals, since they are not issued from being in M, they are not warranted.
It may seem so to the subject. Yet they are not. In this respect, expressive
entitlements are not like ordinary instances of justification or warrant that
may be present even if the beliefs eventually based on them are false.
In order to review Bar-On’s proposal, it is worth making use of our
usual template. Hence, we have seen that groundlessness is respected in
its weak form, for it is a tenet of the present account that avowals are
not based on observation or inference to the best explanation. It is also
respected in the more demanding sense that the first-order mental state
does not play the role of evidence for the corresponding self-ascription,
not even in the minimal sense invoked by Peacocke (see Chap. 5, §1).
However, as we have seen, that mental state functions as a non-evidential
warrant for the corresponding self-ascription and thus as its rational, as
opposed to merely brute, cause. Bar-On’s position on this score is, how-
ever, extremely problematic. Consider the case of false yet sincere avowals.
Take a subject who is so frightened of the dentist as to avow pain before
he even touches her mouth, and runs away. In such a predicament, we
would like to say that she did so—from her point of view—because she
thought she was in pain, even though—from our own point of view—she
will have done so out of fear rather than pain. Given Bar-On’s critique of
the idea that avowals are based on how things seem to one in one’s mental
arena, as it were, and her thesis that in such an event the self-ascription
of pain would be unwarranted, as well as false, it is left in the dark how
we could recognise some degree of rationality in the subject, at least from
her own point of view. To say that
we can still give sense to the intuition that there is a common element
shared by acts of avowing truly and acts of avowing falsely: both are acts of
speaking one’s mind whose characterisation as reasonable acts may require
mentioning the (possibly absent) self-ascribed state of mind25
25
Bar-On 2004, p. 395.
158 A. Coliva
does not help, since there is nothing, on the present account, in the sub-
ject’s mind which could warrant her false self-ascription of pain. Nor does
it help to say that “a false avowal may still be regarded as immune to errors
of misascription, as well as misidentification”.26 For, after all, it is a core
aspect of Bar-On’s proposal that avowals can be false and hence that the
presumption of truth they normally enjoy can be overridden. Finally, it
does not help to say that “even if we take an avowal to be false, and thus
take the subject to be wrong, we may still not take her to have gone wrong,
epistemically speaking”.27 For if the warrant is identical to the truth-maker
and the latter is absent, so is the former. Hence, Bar-On cannot conclude
that “false and true avowals may equally enjoy entitlement by default”.28
Furthermore, it is dubious that avowals are direct manifestations of
subjects’ mental states, which actually make the latter immediately per-
ceivable by one’s audience. To see how implausible it is, consider the fol-
lowing thought experiment. Suppose an alien arrives in a country where
there are human beings and zombies, who all behave alike both verbally
and non-verbally. First, it is not clear why an alien should see their pain,
as opposed to merely seeing their overt behaviour (let it be linguistic
or non-linguistic) and should learn to interpret it as expressive of the
relevant mental states. Second, how could he be said to see a zombie’s
pain, say, since, ex hypothesi, a zombie does not feel any pain? Yet, since
he would be behaving (either linguistically or non-linguistically) just as a
normal human being and in Bar-On’s account the latter would be directly
showing her pain, he should be showing his pain too.
Moving on to transparency, Bar-On is extremely cautious in her
remarks. She thinks it is entirely possible to be self-blind creatures—that
is to say, creatures who do have mental states, even conscious ones, and
yet cannot have knowledge of them. Surely, she does not wish to make
self-knowledge constitutive of having conscious mental states,29 nor is she
attracted to the view that, if possessed of the relevant concepts and cogni-
tively lucid and alert and if a subject has a given mental state, she will be
in a position to believe or judge that she does. Still, she is hospitable to the
26
Ibid.
27
Bar-On 2004, p. 396.
28
Ibid.
29
See Bar-On 2004, p. 406.
6 Expressivism About Self-knowledge 159
idea that a subject endowed with the relevant concepts, while cognitively
lucid and alert, may direct her attention so as to pass judgement over her
ongoing mental states, like in the case of non-evidential reportive avowals.
Moreover, she thinks that given the presumption of truth our avowals
enjoy, authority is respected. Yet, given that she allows for cases of false
avowals and in particular of self-deception, as we have seen, authority, in
her view, holds only in its weak version. In this connection, however, it
is worth stressing that Bar-On’s defence of the idea that avowals may be
false yet sincere is highly problematic. For, if it is constitutive of avowals
that they are taken to be elicited from the mental state avowed, it should
be said that there was no real avowal in the first place, when there is no
corresponding mental state. Even more so if one holds, as Bar-On does,
that the mental state is not merely the cause—that is, the symptom—of
the mental state but somehow embodies it. For, in the case of false (yet
sincere) avowals, no mental state is present to elicit and be embodied
in one’s words. If there is none, then one’s words may resemble avow-
als, but they are not. They are, at best, an attempt at avowing a mental
state. Saying that the subject nevertheless succeeds in expressing a mental
state, though not her mental state, is very confusing. Surely, she did not
express someone else’s mental state. Moreover, it is true that we may take
the failed avowal of pain as a symptom of the subject’s fear of the dentist.
However, if we construe this as meaning that she has managed to express
a mental state of hers, though not the one actually avowed, this raises the
question of how come the subject has used a self-ascription of pain, say,
to give vent to a mental state of fear of which, clearly, she is not aware
as such. Thus, the idea that there is a mental state M that gives rise to a
self-ascription of a mental state N suggests that the subject has wrongly
recognised her own mental state—that she has after all mis-taken M for
N—contrary to what Bar-On wants to maintain.
3 Summary
In this chapter, we have analysed early and recent expressivist accounts of
self-knowledge. We have seen how they strongly reject the idea that our
knowledge of our own minds is based on a robust cognitive achievement
and how they hold, in contrast, that psychological avowals are acquired
160 A. Coliva
Bibliography
Bar-On, D. (2004). Speaking my mind. Expression and self-knowledge. Oxford:
Oxford University Press.
Boyle, M. (2009). Two kinds of self-knowledge. Philosophy and Phenomenological
Research, 77(1), 133–164.
Burge, T. (1996). Our entitlement to self-knowledge. Proceedings of the
Aristotelian Society, 96, 1–26.
Coliva, A. (2010). Moore and Wittgenstein. Scepticism, certainty and common
sense. London: Palgrave Macmillan.
Finkelstein, D. (2003). Expression and the inner. Cambridge MA: Harvard
University Press.
Jacobsen, R. (1996). Wittgenstein on self-knowledge and self-expression.
Philosophical Quarterly, 46, 12–30.
Wright, C. (2004). Warrant for nothing (and foundations for free)? The
Aristotelian Society Supplementary Volume, 78, 167–212.
7
Constitutive Theories
1
See Wright 2001c, p. 310; Boghossian 1989, p. 5.
2
That is to say, by being drilled to substitute their immediate avowal, “P”, “P would be good to
have”, “I will φ”, with the corresponding psychological one—that is, “I believe that P”, “I want/
desire that P”, “I intend to φ”.
7 Constitutive Theories 167
3
Shoemaker 1996a, p. 236.
4
Shoemaker 1996a, p. 234.
5
Ibid.
168 A. Coliva
agnostic has to infer to the mental states of the agent starting from the
observation of the latter’s (linguistic and non-linguistic) behaviour. If,
then, there are reasons for us to think that the agent and the agnostic are
one single subject because the agnostic’s beliefs and desires rationalise the
agent’s actions, then the agnostic too will be in a position to know that
his beliefs and desires rationalise the agent’s actions.
Notice that this argument rests on the premise that it is only if the
agnostic makes the relevant psychological self-ascriptions that we can
start making sense of the agent’s behaviour as in fact identical to the
agnostic’s. This, however, is clearly not necessary. In fact, the agnostic
could just say things such as “It is raining. It is bad to get wet” while the
subject opens the umbrella. This would give us reasons to think that the
agnostic and the agent coincide because the relevant first-order mental
states ascribed on the basis of what they say and do, without the need of
any psychological self-ascription on their part, rationalise their behaviour
and give it the necessary unity to make us think that it is performed by
a single subject.
Of course, for the agnostic to know that the agent’s mental states coin-
cide with hers, she would have to have access to her own. But we could
imagine, in keeping with the conditions of this thought experiment, that
she could know of them inferentially. That is to say, she would know that
she said or judged “It is raining. It is bad to get wet” while opening the
umbrella, and she could then infer, by reflection on this kind of overt
linguistic behaviour, that she believes that it is raining and desires not to
get wet. Hence, nothing so far in the argument presented by Shoemaker
requires direct or immediate self-knowledge.
6
Shoemaker 1996a, p. 235.
7 Constitutive Theories 169
Suppose you come home, and see that no car is parked in the driveway. You
infer that your spouse is not home yet… Later, you may suddenly remem-
ber that your spouse mentioned in the morning that the breaks of the car
were faulty, and wonder whether she may have taken the car for repair. At
this point, you suspend your original belief that she is not home yet. For
you come to realise that the absence of the car is not necessarily good evi-
dence that she is not home. If the car is being repaired, she would have
returned by public transport. Then finally you may reach the belief that she
is home after all, given your next thought that she would not have taken
any risks with faulty breaks.
protect me from the rain), I open the umbrella as soon as I get out of my
office and I do so without having to self-ascribe these beliefs and desires
to myself, but just by having them and by letting them shape my further
behaviour. Although “thick” critical reasoning is certainly interesting and
characteristic of a lot of adults’ psychological lives, it is quite clear that
it cannot be taken to be the basic notion in terms of which we could
advance theses about the rationality of human beings in general and, in
this context, as the origin of our self-knowledge.
Let us then focus on “thin” critical reasoning, which, in contrast, seems
to be more basic and constitutive of what it means to be a rational sub-
ject. Of course, being able to engage even in “thin” reasoning requires
some kind of sensitivity not just to the contents of one’s beliefs and desires
but also to the kind of mental states they are contents of. For it makes
a difference whether I believe it is raining or whether I desire it, vis-à-
vis the practical reasoning that would eventually lead me to opening the
umbrella. Yet it does not seem incredible to suppose that a physical system
is so engineered that (other things being equal) it opens the umbrella only
when the content “It is raining” appears in her belief-box, after seeing the
rain fall down, and the content “Not get wet” appears in her desire-box.
One may object to this reconstruction by noticing that if this is indeed
a piece of reasoning the subject is doing at the personal, rather than at the
subpersonal, level, she should be aware of the fact that the former is the
content of a belief and the latter of a desire. Hence, she should have tacit
knowledge of the mental states she is enjoying. That poses the problem
of understanding what one means by “tacit” knowledge. According to
Shoemaker, a subject has tacit knowledge of her belief that P “just in case
it is as if A has an explicit belief in P”—that is, “A’s cognitive dispositions
are relevantly as if A has an explicit belief in P”.8
My claim is that to the extent that a subject is rational, and possessed of the
relevant concepts (most importantly, the concept of belief ), believing that
P brings with it the cognitive dispositions that an explicit belief that one
has that belief would bring, and so brings with it the at least tacit belief that
one has it.9
8
Shoemaker 1996a, p. 241.
9
Ibid.
7 Constitutive Theories 171
There are two strands to this claim. The first one is that if one believes
that one believes that P, one will be disposed to assert “P”, to use P as a
premise in one’s reasoning and to hold P true. These are just the same dis-
positions one would have if one had the first-order belief that P. The sec-
ond sense, however, is that just by being a rational agent in Shoemaker’s
sense of the word, one will be disposed to make the relevant psychological
self-ascription. This is not a disposition one would have just by having
the first-order belief, however. In effect, Shoemaker’s gloss on the notion
of tacit knowledge comes down to this: if a subject is a rational agent,
she has tacit knowledge of her first-order belief that P, just in case she
behaves like someone who believes that P and is disposed to self-ascribe
that belief. Yet, to count as having that disposition, one should, at least
on occasion, make that kind of psychological self-ascription.
Hence, the argument used to establish the necessity of self-knowledge
for critical and practical reasoning need not depend on the strong claim
that these forms of inference necessarily involve a reflection on the sound-
ness of one’s propositional attitudes. It could allow, that is, for a “thin”
notion of inferring and reasoning in a critical way, or in accordance with
practical syllogisms, just as long these inferences are supposed to take
place at a personal, rather than at a subpersonal, level and tacit knowledge
of one’s attitudes is explained along the lines just reviewed—that is to say,
as consisting in being capable, at least on occasion, to self-ascribe the rele-
vant mental states. Notice, however, that tacit self-knowledge thus would
presuppose explicit self-knowledge. If this were the correct account of
tacit self-knowledge, it then would turn out that even being critical rea-
soners in a “thin” sense presupposes being capable of self-ascribing one’s
propositional attitudes. Shoemaker’s claim would be (at least condition-
ally) vindicated; a rational agent—that is, someone possessed of normal
intelligence, rationality and the relevant concepts—cannot be self-blind.
However, it is doubtful that this is the correct account of tacit knowl-
edge. In particular, it explains tacit knowledge in terms of explicit
knowledge and thus it seems to deprive the former of any theoretical
significance. Furthermore, it cannot be used to explain (explicit) self-
knowledge, for ultimately it presupposes it. But if having tacit knowl-
edge of one’s first-order beliefs reduces to the first strand to that notion
(namely, to being disposed to using P as a premise in one’s reasoning, to
172 A. Coliva
10
See Shoemaker 1996a, p. 224. I suspect that Boyle’s account of tacit self-knowledge amounts just
to such a view. See, in particular, Boyle 2011a (cf. Boyle 2011b). Yet this account seems to me in
tension with Boyle’s position in Boyle 2009 (§5.2), where tacit self-knowledge seems to depend on
the ability, at least on occasion, to self-ascribe the relevant mental states.
7 Constitutive Theories 173
Instead of the first-order state and the belief about it having different core
realisations but overlapping total realisations, it might be that they have the
same core realisation and that the total realisation of the first-order state is
a proper part of the total realisation of the first-person belief that one has
it. (…) What my earlier discussion suggested is that if one has an available
first-order belief, and has a certain degree of rationality, intelligence, and
conceptual capacity (…) then automatically one has the corresponding
second-order belief. If it is possible to have the available first-order belief
174 A. Coliva
the conceptual and cognitive competence or the sincerity of the subject. As we shall see, in contrast,
attitudinal avowals, for Wright, are only weakly authoritative.
13
See Kripke 1982.
176 A. Coliva
she may have meant “+” thus all along. For criticism of Wright’s so-called “Intention view” of rule-
following, see Boghossian 2012, Peacocke 2012 and Bilgrami 2012. Wright has recently aban-
doned this position. See Wright 2015.
18
Wright 2001d, p. 312.
19
The set of rules and criteria that constitute our concepts and determine the meaning of the words
used to express them.
20
See Wright 2001f, p. 372.
178 A. Coliva
I take this passage to suggest that just as an attitudinal avowal does not
track a pre-existing mental state, it does not give rise to it either. Elsewhere,
21
Wright has exploited this idea in numerous writings and in different areas of discourse such as
taste, morality (although he is sceptical of the prospects of this strategy in that realm) and second-
ary qualities (colour in particular). See Wright 1992. Concerning propositional attitudes, the most
thorough discussion is in Wright 2001b.
22
Wright 2001b, p. 204 et infra.
23
Wright 2001b, p. 204.
7 Constitutive Theories 179
Wright insists that the self-ascription suffices for having the intention, but
it is not necessary for it.24 Moreover, Wright (2001f, pp. 351–352), in the
course of discussing Burge’s account of self-knowledge (cf. Chap. 5, §2),
criticises the self-verifying thesis regarding attitudinal avowals for explain-
ing too much—that is, infallibility—when the relevant self-ascriptions
may be open to error, essentially due to self-deception. But he also criti-
cises the self-verifying view because, in his opinion, it is guilty of explain-
ing too little. For it would hold only for episodic thoughts, originating
in the relevant self-ascriptions. In contrast, Wright thinks that authority
extends to all kinds of attitudinal avowals, also to those which do not
exploit the characteristic feature of some episodic thoughts and indeed
also to avowals of past intentional states (cf. Wright 2001b, p. 206).
This, however, confers a strange metaphysical status to intentions (and
to other propositional attitudes as well). For intentions are not brought
about by the self-ascription itself. But they do not seem to exist already, at
least as such, either. For, otherwise, it would be difficult to maintain that
our “best” judgement about them—that is, a judgement made sincerely
when there is no reason to think a subject might be self-deceived—deter-
mines their extension. Wright’s position then seems to lean towards a
form of irrealism, of an instrumentalist fashion, which is usually associ-
ated with Dan Dennett’s (1987) idea of an intentional stance and with
Donald Davidson’s (1970) notion of anomalous monism. Intentions
(and other attitudinal mental states) are not metaphysically robust enti-
ties. They are just useful devices—possibly just concepts—employed to
interpret and make better sense of subjects’ behaviour. Indeed, accord-
ing to Wright’s view, the authority we grant to subjects’ avowals of their
intentions is a function of the fact that doing so proves useful in the
application of a mutual interpretational scheme. As he writes:
24
Wright 2001d, pp. 313–314. This way, he fends off the objection that one would be caught in a
regress of beliefs whenever there is anything believed, and that un-self-conscious creatures would
not be able to believe anything.
180 A. Coliva
us that our initiation into the language in which these concepts feature
results in the capacity to be moved, who knows exactly how, to self-ascribe
states of the relevant sorts—and to do so in ways which not merely tend to
accord with the appraisals which others, similarly trained, can make of
what we do but which provide in general a far richer and more satisfying
framework for the interpretation and anticipation of our behaviour than
any at which they could arrive if all such self-ascriptions were discounted.
The roots of first-personal authority for the self-ascription of these states
reside not in cognitive achievement, based on cognitive privilege, but in
the success of the practices informed by this cooperative interpretational
scheme.25
26
Wright 2002.
182 A. Coliva
remembering what one intended to do at some earlier time: you form the
intention of going to the kitchen to start preparing a meal. The phone
rings. Once you hang up, you ask yourself “What was I about to do?”
or “What did I want to do?” and then you suddenly remember that you
wanted to go to the kitchen. Hence, it seems that in either case the rele-
vant self-ascriptions will not be groundless. Nor are they transparent: just
by having had them at some earlier time, a subject—even when equipped
with the relevant concepts and being attentive and alert—is not thereby
able immediately to self-ascribe them at a later time. For one can have
difficulties in figuring out what led one to act a certain way, or one may
forget what one intended to do before receiving a phone call. Authority
is not guaranteed either. For, clearly, one can make mistakes in interpret-
ing oneself, and one can misremember what one was about to do, before
receiving the phone call. These occasional failures, however, contrary
to what Burge maintains (see Chap. 5, §2), do not impair our overall
cognitive well-functioning. Surely, the latter situation is often unverifi-
able, since we do not often speak out loud our ongoing intentions. But
suppose we were so trained that any time we formed an intention, we
immediately gave voice to it. Suppose, furthermore, that we lived in a
community where subjects were always heard by some other member of
the community and where their utterances were always recorded. If that
were so, the idea that we are authoritative with respect to our avowals of
past intentions would evaporate. For others would be just as well placed
to contradict us, based on their recollection (or on records stored some-
where) of what we had said.27
Moreover, as we shall see in the following section, there is an open
issue about the correct account of self-deception, such that it may actu-
ally turn out that, contrary to appearances, self-deception is no bar to
the possibility that avowals of intentions and other propositional mental
states be always authoritative (save for cases of insincerity and of concep-
tual misunderstanding).
27
Consider, moreover, the possible tension between an avowal of an ongoing mental state at t1 and
a different past one at t2. It is not clear, in Wright’s framework, to which one we should accord
priority and why.
7 Constitutive Theories 183
28
He also holds that one incurs various commitments related to the inferential role of the concepts
that specify the contents of one’s commitments. For instance, if I believe as a commitment that
there is a table in front of me, I also ought to believe that if I run hard into it, I will be injured.
184 A. Coliva
29
See Strawson 1962.
30
Free actions are those brought about by non-coercive causes (contra those which are brought
about by coercive ones). Strawson’s idea is that non-coercive actions can be individuated only by
bringing in normative considerations and, in particular, our reactive attitudes and the practice of
blaming and resenting they give rise to. To that picture, Bilgrami adds that our reactive attitudes
should be justifiable ones and, being so, they presuppose that the mental states in which the
resented and blamed actions originated be self-known. If they were not self-known, our reactions
would not be justified and the actions we unjustifiably react to should be considered as non-free.
7 Constitutive Theories 185
There is something honest, then, about those who refuse to grant anything
special to self-knowledge and view it as getting a causal account based on a
7 Constitutive Theories 187
measurably more than usual reliable mechanism that will account for our
intuitions misleadingly expressed as ‘privileged access’. They see it as a nar-
row question in epistemology, they find the exceptions to be ubiquitous,
and they draw their conclusion that there is nothing radically set apart
about self-knowledge. Their conclusion is honestly drawn from their
framework. It is their framework that is wrong. Self-knowledge is unique
only if it is embedded in a much wider framework integrating very large
themes in philosophy such as the normative nature of intentionality and
agency that I have been expounding. (Bilgrami 2012, p. 277).
entail that they are transparently known to their subjects. But this is
clearly absurd. What, however, does preclude the possibility of having
justifiable reactive attitudes towards them if it is allowed that also dispo-
sitional mental states can be the object of such attitudes, as long as those
mental states enter rationalising explanations of subjects' behaviour?
As interesting as Bilgrami’s constitutive proposal appears to be, we
have seen that it is also problematic when it comes to the details of the
vindication of the left-to-right side of the constitutive thesis, based on the
“pincer argument”, and of the idea that propositional attitudes as com-
mitments are intrinsically normative mental states.
31
Coliva 2009, 2012d. The account presented here partially differs from the one presented in those
papers.
7 Constitutive Theories 189
(i) they are the result of an action—the mental action of judging that
something is the case—on the subject’s part;
(ii) they are normatively constrained—that is, they must respond to the
principles governing theoretical and practical reasoning;
(iii) they are so constrained (also) from the subject’s point of view;
(iv) they are mental states for which the subject is held responsible.
32
Hence my proposal sharply differs, on this score, from Bilgrami’s, who, as we saw in §3, main-
tains that transparency holds also for propositional attitudes as dispositions, provided they enter
the explanation of actions that are the object of justifiable reactive attitudes.
190 A. Coliva
33
This can sometimes (and perhaps often) be the case, but it cannot always be the case, because it
would presuppose self-knowledge.
34
See Chap. 5, §1 and this chapter, §1.
192 A. Coliva
“I believe that P”. You teach her to substitute one form of behaviour—one
kind of expression of her mind, that is, the outright assertion of “P”
accompanied by the ability to give reasons for it, which manifests her
first-order belief (based on judgement)—with another, that is, the asser-
tion of “I believe that P”. Similarly, take a subject who says “Peace in the
Middle East would be good to have” and is disposed to offer consider-
ations in its favour and withdraw from it if these considerations did seem
no longer compelling, but if asked “Do you hope/desire that peace be
reached in the Middle East?” did not know how to answer. Then, again
you drill her to use “I hope/desire that P” as an alternative expression of
her mind—that is, of her asserting “Peace in the Middle East would be
good to have” for this and that reason. Let me stress that it is absolutely
essential in order for the proposal to steer away from any observational
model that one be adamant that “I believe that P” or “I desire that Q”
is taught neither on the basis of evidence nor on the basis of the rule “If
you are disposed to judge thus-and-so, then you believe/desire that P/Q”.
Rather, “I believe that P” and “I desire that Q” are taught blindly, as alter-
native expressions of one’s mind: they are ingrained as alternative ways of
expressing one’s first-order beliefs and desires (based on judgement) other
than asserting that P or that Q would be good to have.35 Hence, in this
account, there is no inner epistemology, just a substitution of one form
of behaviour with another. But—and this is crucial—the kind of behav-
iour which would get replaced would already be quite rich. For, in order
to have beliefs (and other propositional attitudes) based on judgement,
a subject will already have to have the ability to differentiate between,
for instance, believing P and P’s being the case, by being sensitive to the
fact that her point of view may be challenged—thus responding with
reasons in favour of it—or indeed proved wrong—thus abandoning it. It
is only on the background of this already complex pattern of behaviour,
which, however, does not seem to require the concept of belief, as we saw
35
Indeed, this seems to me to be the right development of Wittgenstein’s idea that avowals substi-
tute behaviour. It is just that when we move from avowals of sensations to avowals of propositional
attitudes the behaviour we must take into account is not merely physical but also linguistic. As the
reader will recall, Dorit Bar-On makes a similar point, even though in the course of developing a
neo-expressivist account of self-knowledge (cf. Chap. 6, §2). Robert Gordon too seems to endorse
it, even though he does not think it suffices for possessing the relevant psychological concepts. We
criticized Gordon’s view in Chap. 4 (§3).
7 Constitutive Theories 193
36
This is indeed something which goes to the heart of the debate anyway, as Davies and Stone
(1995a) remark in their Introduction.
7 Constitutive Theories 195
understand that their own point of view about the world (or that of other
subjects) may be wrong. This, I take it, is at least a necessary condition
for having beliefs as commitments. Furthermore, the ability to pass the
false-belief test in the case of others may be explained differently than by
appeal to the fact that children would possess a theory of other minds.
For it would be enough to explain their correct answers to suppose that
they issue them as if they themselves were in the other person’s shoes. So,
the ability to pass the false-belief test need not show that children pos-
sess a veritable theory of their own minds as well as of others’. In fact,
it may actually be taken to prove just that they are capable of first-order
beliefs based on judgement and to project themselves onto others and
therefore issue the correct answer to the false-belief test, without thereby
having any explicit knowledge of their own and other minds, which, in
my view, crucially depends on the possession of the relevant psychologi-
cal concepts.
The reason why the data at our disposal do not seem to tell us clearly
what is the case is that—quite understandably—the experiments have not
been designed to test the possibility I am advocating here. For, usually,
children will be exposed to talk in terms of belief when they are actually
in the process of acquiring the ability to have beliefs based on judgement.
This may well have confused the issues: we may have mistaken the fact
that it takes time for children to learn to have these first-order mental
states as a sign of the fact that it takes them time to acquire the concept
of belief. Furthermore, we may have imputed that difficulty to the fact
that mastery of that concept would depend on the acquisition of a theory
of one’s own as well as of others’ minds. A more telling test, then, would
be to look at children who have not been exposed to psychological talk
up to the age of 3 or 4; see whether, around that age, they pass the false-
belief test in their own case as well as in others’ (where, crucially, the test
should not be phrased in terms of beliefs); and then introduce them to
talk in terms of belief (and other propositional attitudes).37 If, at that
37
Should this test prove impossible, one might see whether there are languages which do not have
talk in terms of belief and other propositional attitudes. If speakers of those languages were in fact
capable of beliefs as commitments, as I think there is no reason to be sceptical of, we then could
test how long it would take them to acquire the ability to express their minds by self-ascribing
beliefs and other propositional attitudes in a different language which contained these devices.
196 A. Coliva
stage, it actually takes them a short amount of time to learn how to use
“I believe”, then we would have shown that the account of concepts’ pos-
session I have been proposing would, in fact, be compatible with human
psychological development.38
Furthermore, it has to be noted that the conceptual role of the concept
of belief is what theorists of concepts offer as an abstract individuation
of that concept that supervenes on a practice of its use which, however,
may come about in different ways. In particular, since according to the
present proposal the commitments undertaken by asserting “P”, as an
expression of one’s belief based on judgement, and by asserting “I believe
that P” would actually be the same (save for the fact that the latter would
make explicit what the former leaves implicit—namely, that the assertion
of “P” expresses one’s own point of view that need not be correct), it may
well be that the conceptual role of the concept of belief in the first-person
present specifies the rules for the use of that concept which, in practice,
may have been acquired by becoming able to have first-order beliefs based
on judgement first and then by being blindly drilled to express them by
prefacing one’s assertion of “P” with “I believe that”.
Assuming that what I have been proposing is along the right lines, it is
perhaps worth noticing that it is a consequence of the suggested account of
conceptual mastery that while it may be an open issue whether and to what
an extent language is necessary in order to have first-order propositional
attitudes based on judgement, on my view language is indeed necessary for
38
The apparently difficult case, for my own proposal, would in fact be constituted by autistic
patients affected by Asperger syndrome. Although they do not fail the false-belief test, they seem
not to have a theory of their own minds and to lack a theory of other people’s minds (Frith and
Happé 1999). But several things must be noted: 1) when we look at their reports, what they show
is that subjects affected by this syndrome have different kinds of experiences, particularly of speech,
and different sensations, if not an altogether lack of painful sensations (pp. 15–18). None of this
would show anything relevant with respect to their propositional attitudes and their knowledge of
them. 2) The only report which has a bearing on this issue (from Donna Williams (1994)) in fact
seems to imply that she did not have, as an autistic child, desires as commitments (she writes
(p. 15): “Autism had been there before I’d ever known a want of my own, so that my first ‘wants’
were copies of those seen in others (a lot of which came from TV)”). In such a case, it would not
be surprising that they would have to gain knowledge of their own minds in a third-personal way
and that this would require some kind of theory of other minds. Finally, all the data are based on
personal reports and, obviously, this would not have any bearing on the possibility of having, and
of having knowledge of one’s occurrent commitments. For much of what they say could actually be
due to forms of self-interpretation.
7 Constitutive Theories 197
One might, with Wright, add to the C-conditions that the subject
should not be self-deceived (or anyway that it is reasonable to assume
that she is not). But quite apart from sounding an ad hoc move, it seems
that the very possibility of self-deception would show that constitu-
tive accounts do not have much of a point: after all, how could one be
mistaken about one’s own immediately available mental states if not by
somehow going wrong in identifying them? Wouldn’t such room for error
be compatible only with non-constitutive accounts of self-knowledge?40
Hence, it would be good news if we could account for self-deception dif-
ferently, thereby showing that its existence would not constitute a threat
to constitutive accounts.
Here is where I agree with Bilgrami: self-deception is a case where a
subject self-ascribes a mental state and has it as a commitment yet also
has another, opposite mental state as a disposition. The irrationality is
brought about by the clash between one’s commitments and one’s own
unconscious dispositions. Thus, for instance (the example is mine), take
a jealous wife who openly and sincerely asserts with her friends that she
believes that her husband is totally faithful to her—and has all the rea-
sons in the world to do so—but, then, once at home, is often inquisitive,
searches his belongings, and so on. According to Bilgrami,41 what we
should say is that she believes as a commitment that her husband is faith-
ful—after all, she is prepared to assert it with friends and has all the rea-
sons to think so. Yet she also has the unconscious belief, as a disposition,
that he is unfaithful to her, which is operative in her inquisitive behav-
iour. So, she is self-deceived, in the sense that she sincerely avows a belief
and behaves in ways that run contrary to it. Yet it is not the case that she
40
Wright (2001b, p. 324) seems to me to underestimate the implications of allowing for cases of
self-deception, and so I think does Heal (2002, p. 276).
41
See Bilgrami (2006, pp. 140–157; 278–280); cf. also Stoneham 1998. It must be stressed, in
order to avoid confusions, that I am endorsing Bilgrami’s account of self-deception with respect to
those mental states one would not attribute to oneself on the basis of inference and observation of
one’s own behaviour and further available mental states. I think in the latter cases one could make
genuine mistakes and self-attribute mental states one does not really have. If, then, one were to
restrict self-deception, properly so conceived, only to these cases, as Wright suggested to me in
conversation, then the fact that one might go astray in self-interpreting oneself would not represent
a counterexample to the view that non-observational or immediate self-ascriptions of propositional
attitudes are not open to failures of authority. The authority of immediate attitudinal avowals then
would remain unchallenged.
7 Constitutive Theories 199
has a false belief about her own beliefs. Rather, she has two, different
beliefs, both in kind and in content, that give rise to her distinctively
irrational behaviour.42
Here are some considerations in favour of Bilgrami’s account of self-
deception. First, I agree with Bilgrami that his account does well at
explaining cases of motivated self-deception—that is, cases in which self-
deception is the outcome of a conflict in the subject between, say, believ-
ing that P and believing that not-P. In these cases, one of the two mental
states gets suppressed while the other one is endorsed. Yet the former can
remain operative in shaping (at least part of ) the subject’s behaviour and
lead her to various forms of inconsistency. Secondly, and more generally,
what makes us say that a subject is self-deceived is a conflict between her
psychological self-ascriptions and (some other aspect of ) her behaviour.
Now, conflict is usually brought about by the fact that there are two oppo-
site parties (or more) at fight, neither of which need be wrong but each of
which may simply be responding to different motivations and concerns.
In the case at hand, it makes sense to think that whereas one part of the
subject’s personality is entirely confident and mature, the other is full of
insecurities, which lead her to be suspicious of the behaviour of those
around her. Of course, there may be reasons for both attitudes: on the
one hand, the fully open and trustworthy behaviour of the husband and,
on the other, a perhaps (well-)motivated sense of insecurity about one’s
own power to attract a person and to involve him in a stable relation-
ship. Finally, suppose the subject realises, either through self-analysis or
through the aid of a therapist, that she has such an unconscious belief
about her husband’s infidelity. Now, if it were just a matter of realising
42
One may object that there are also cases of “negative” self-deception—cases, that is, in which one
says “I do not believe that P˝ yet behaves in ways that are explainable only by attributing to them
the belief that P. Stretching the example slightly, but just because that would help make the point
more vividly, think of Pascal who would say “I do not believe God exists (nor that he does not)”
and yet would behave—or, at any rate, recommend to behave—as an irreprehensible Christian. In
this case, one would not be self-ascribing any belief. Hence, the only option seems to say that one
falsely believes that one does not believe that God exists (nor that he does not). But we can recast
the example in such a way that it ceases to be a counterexample to authority. For we could say that
the avowal is still the expression of the subject’s mental state—namely, of her commitment to not
using “God exists” (nor its negation) as a premise of her practical and theoretical reasoning, which
runs against the disposition to behave as a kosher Christian and thus use that belief as a premise of
her practical reasoning.
200 A. Coliva
her own mistake in categorising a state of mind of hers, she should simply
correct her psychological self-ascriptions. After all, when I get to know
that the wall I am looking at is not red, but white and lit by a red light, I
would immediately correct my belief—that is, I would substitute it with
the new one. But clearly this is not what would happen in the case we are
considering. For the subject would (or, at any rate, should) try to realign
her behaviour, motivated by the belief as a disposition that her husband
is unfaithful to her, with her commitment to his fidelity. Obviously, this
can take a lot of time and personal effort and indeed may never fully
succeed. For all these reasons, it seems to me that Bilgrami’s account of
self-deception is by far preferable to the traditional explanation of this
phenomenon in terms of simply mistaken psychological self-ascriptions.
As a result, self-deception is entirely compatible with the fact that a sub-
ject is authoritative with respect to her own mental states as long as it is
clear that the mental states she is authoritative about are merely those as
commitments.
Having dispensed with the counterexample to authority—indeed,
with what is usually regarded as the only counterexample to it—let me
turn to the problem of explaining why it holds. Recall that we are look-
ing for an account of authority that does not make it the result of any
epistemic achievement. For any epistemic achievement may, in principle,
go wrong and, in that case, there could be counterexamples to authority.
But we have just seen that there aren’t any.43 Indeed, there cannot be
any if we want to be serious about the fact that the biconditional holds
as a matter of conceptual necessity. So we should not just be happy with
the result that, as a matter of contingency, self-deception is not really a
threat to authority. For one thing, one might dissent with our account
of that phenomenon and still hold that it raises a problem for authority.
For another, there may be other phenomena which might, at least con-
ceivably, impugn it. Hence, a satisfactory account of authority must dis-
pense with the result-of-a-cognitive-achievement picture, tout court. For,
so long as psychological self-ascriptions are seen as reports on one’s own
43
This is not to say that one’s own avowals of one’s mental states as commitments are always correct
but only that they are open to a very limited form of error: they are incorrect either because of
conceptual incompetence or because of slips of the tongue.
7 Constitutive Theories 201
mental states, the question arises of whether they are true or false. Unless
one is prepared to suppose that our cognitive faculties may be infallible,
one could not account for the claim that authority holds as a matter of
conceptual necessity.
One strategy may consist in maintaining that, since there are not
counterexamples to authority, any competent and sincere assertion of “I
believe that P” (or of “I desire/intend/wish/hope that P/to φ”) would
entail that one has the corresponding first-order belief (or other proposi-
tional attitude, as a commitment). Still, this would hardly be an explana-
tion of why authority holds, but, rather, a simple acknowledgement, or
a consequence, of the fact that it does.44 Again, it is only if one is in the
grip of a Wittgensteinian anti-explanatory mantra that one could leave
matters at that.
The explanation of authority takes two different routes, depending on
whether one forms the belief as a commitment that P through its very
self-ascription or on whether one just gives expression to it by means
of its self-ascription. Let us look at them in this order. In the former
case,45 the crucial point is that when “I believe/desire/intend/wish/hope
that P”—that is, the corresponding psychological self-ascriptions are
acquired along the lines developed so far, as ways of making the same
commitments as the ones undertaken by judging and asserting “P” or “P
is worth pursuing/having”, while having in view reasons in favour of P
(is worth pursuing/having)—, it then becomes possible to use “I believe/
desire/intend/wish/hope that P” in lieu of asserting (or judging) that P
(is worth pursuing or having), in order to form one’s first-order belief
or desire that P directly. The difference between forming the first-order
mental state by means of the second-order judgement, instead of forming
it by means of the first-order one, is just the fact that “I believe/desire/
intend/wish/hope that P” makes explicit what the first-order judgement
leaves implicit—namely, that that is just one’s own particular standpoint
44
As we have seen in §3, this would be Bilgrami’s strategy.
45
As we saw in Chap. 5 (§2), Burge accords pride of place to cogito-like performatives and consid-
ers the ones in which one self-ascribes other propositional attitudes (as commitments) impure,
since they may be affected by error if one were cognitively malfunctioning. In my view, cogito-like
cases are by far less interesting than the other ones, as they would merely bring about episodic
thoughts rather than propositional attitudes as commitments.
202 A. Coliva
on P (or its being worth pursuing or having). Hence, we can bring about
the relevant first-order belief and other propositional attitudes (as com-
mitments) either by judging that P is the case (or is worth pursuing or
having) or by judging “I believe/desire/intend/wish/hope that P”, while
having in view reasons in favour of P (is worth pursuing/having). For,
to repeat, given the role of the latter locutions (either in speech or in
thought) and of “P (is worth pursuing or having)”, I can commit myself
to P (is worth pursuing or having), thus bringing about the correspond-
ing first-order beliefs or other propositional attitudes, either by simply
judging the latter or by judging the former, thus simultaneously making
explicit my commitment to P’s being the case (or to its being worth pur-
suing/having).46 According to this model, there would be a sense in which
it is literally true that we make up or create our minds. Moreover, since
one’s self-ascription would bring into existence the relevant first-order
mental states, those judgements would necessarily be true—in fact, self-
verifyingly so.47 Furthermore, there would be no temptation to think that
one should have the first-order mental state in view first, in order to make
one’s judgement, which thus would result in knowledge. For, if there
is no mental state before making the relevant assertions or judgements,
then of course there is nothing to know, or be aware of, in the first place,
which should be tracked in judgement.
But not all self-ascriptions of propositional attitudes as commitments
have this performative nature of bringing into existence the very first-
order mental state they are about. A lot of them simply avow a first-order
mental state which is, nonetheless, our making, yet not through its very
46
Obviously, it would always be available to one to justify one’s judgement—“I believe/desire/
intend/wish/hope that P˝, say—ex post by appealing to the fact that one’s evidence allow(ed) one to
judge that P (is worth pursuing or having), thus giving rise to one’s belief/desire/intention/wish/
hope that P (as a commitment). The possibility of giving such a justification for one’s judgement “I
believe/desire/intend/wish/hope that P˝, however, should not obscure the fact that the commit-
ment to P (is worth pursuing or having) was actually made by judging “I believe/desire/intend/
wish/hope that P˝. My account of attitudinal avowals then explains why philosophers, most nota-
bly Wittgenstein in The Philosophical Investigations II, x, have been tempted to reduce “I believe
that P˝ to “P˝. Their mistake was due to failing to see that the contents of those judgements are
different, whereas their insight was to recognize that the commitments undertaken by making those
judgements (or the corresponding assertions) are virtually the same. See the Appendix for an
account of Moore’s paradox.
47
Also, the corresponding assertions would have the same effect, as long as sincerity is granted. For
an account of what this means, in the present context, there will be more in the following.
7 Constitutive Theories 203
making any participant aware of the fact that her specific point of view—
formed through an act of rational deliberation—is just one among other
possible ones.
When understood in the way proposed, a judgement (or a sincere
assertion49) such as “I believe/desire/intend/wish/hope that P” is like a
performative—namely, like “I promise to buy you an ice cream”, “I hereby
thee wed”, “I hereby name you so-and-so”, and so on: it makes a certain
thing happen, for it does create the first-order propositional attitude as a
commitment. Where this, to repeat, is possible precisely because judging
“I believe/desire/intend/wish/hope that P” becomes just an alternative
way of undertaking the same commitments one would make by judging
that P (is worth pursuing or having, once provided with the relevant psy-
chological concepts), save for the fact that the former kind of judgement
would also make explicit what the latter leaves implicit—namely, that the
judgement reflects just a subject’s own point of view.
Many have argued against such a view of psychological self-ascriptions
by maintaining that it would commit one to the implausible claim that
they would lack content and could not, therefore, be sensibly prefaced
by negations, or be embedded in suppositions, and otherwise wider con-
texts.50 But this objection, if sound at all,51 could be raised only in the case
49
Notice the sincerity condition placed upon second-order assertions, which cannot, however, be
carried over to judgements, since judgements are—when made—necessarily sincere. This seems to
me enough to dispel the worry that one may judge or assert “I believe that P˝ and yet not form the
corresponding first-order mental state as a commitment. Of course, what remains entirely possible
is that I do not act on the basis of such a commitment and thus fail to actualize the connected
disposition. But this is no objection to the view which is being proposed here.
50
The locus classicus is Geach 1965.
51
The counter is usually that they could retain minimal assertoric content as well as minimal truth.
Accordingly, it would suffice for minimal assertoric content that performatives can be embedded in
negations and suppositions and that they can undergo the usual tense transformations. Obviously
they can. For “I am not going to buy you an ice cream” or “Suppose I bought you an ice cream”
and “I did buy you an ice cream” are perfectly sensible things to say. Minimal assertoric content
then pairs with minimal truth—with the idea that it is enough to qualify as a truth predicate that
some platitudes—in particular, negation and the T-schema—are respected. It would be a further
issue whether these statements should be taken as descriptions or as expressions of commitments one is
(or is not) undertaking thereby. This strategy can be found in Hacker (1986, p. 90) as well as in
Wright (1992, p. 28) and in Jacobsen (1996). Hacker (1986, p. 298), however, denies that minimal
assertoric content would be compatible with truth evaluations. Be that as it may, when we consider
judgements (and assertions) of “P˝ (is worth pursuing/having), we could distinguish between the
content of the judgement (or of the assertion), which obviously is truth-assessable, and what is being
206 A. Coliva
done by means of the act of judging it. It is the act of judging that P that brings about the corre-
sponding belief (or further propositional attitude) as a commitment, which can get expressed in
one’s assertion of “P (is worth pursuing/having)”, although the content of one’s assertion would
remain P (is worth pursuing/having). And when the psychological self-ascription functions merely
as an avowal, it expresses one’s commitment while also saying that one has it. Hence, an avowal too
can have minimal assertoric content. A similar distinction can be found in Jacobsen (1996, p. 26)
and in Bar-On (2004, pp. 251–264).
52
A similar point can be found in Heal (2002, pp. 282–288) but also in Jacobsen (1996, pp. 23–28).
Strangely enough, Jacobsen, who, officially, sets out to characterize an expressivist account of avow-
als, in fact ends up defending the claim they are performatives. See Jacobsen (1996, pp. 26–28).
Thus, I agree with much he says, although I would insist on the difference between expressivism
and constructivism: on the former, first-order mental states are already there and get simply
expressed by the relevant utterances; on the latter, in contrast, utterances (if sincere) bring about
first-order mental states, at least in some cases.
53
Similar considerations could be made for the case in which psychological self-ascriptions do not
have a performative function but merely an expressive one. For one indeed may try to fool one’s
interlocutor by means of fake expressions of the relevant first-order mental states.
7 Constitutive Theories 207
54
As well as in the case of avowals. See previous footnote.
55
Notice that here an asymmetry between assertions and judgements may arise. For the performa-
tive judgement cannot be overridden. So, one way of putting the point is that when I am insincere
I utter a performative sentence without making the corresponding judgement.
56
Cf. Geach 1965, p. 260. As we saw in Chap. 6 (§1), this was indeed Wittgenstein’s unpalatable
position.
57
See also Jacobsen (1996, p. 18) for a similar point, which can be traced back to Wittgenstein (Cf.
Philosophical Investigations, §560).
208 A. Coliva
58
Hence, I would not be too keen to endorse the kind of dilemma Wright thinks there is in the fact
that, on the one hand, avowals are authoritative and, on the other, they self-ascribe a disposition
whose obtaining is assessable from a third-personal point of view. See Wright (2001a). Jacobsen
(1996) and Heal (2002) too seem to be highly struck by this dilemma. The dilemma, however,
seems to me very much a function of a simplistic description of the situation. Hence, it calls more
for a dissolution—for an account of why it does not really stand—than for a positive solution.
7 Constitutive Theories 209
59
There will be more about these cases in the next chapter (§1).
60
In fact, they may make action and deliberation even more difficult to attain.
210 A. Coliva
61
Notice, moreover, that they can also be used to report a commitment previously undertaken. In
such a case, they are based on remembering having made such a commitment and certainly not on
inspecting one’s own mind, as it were. Also, it may be possible to find out about one’s own disposi-
tions in a third-personal way and self-ascribe them and subsequently form a corresponding com-
mitment. All this would require two different mental actions, although the final self-ascriptions
may be identical in form. We dwell on both issues in the next chapter.
62
With the usual caveats having to do with difficulties in exegesis, Wittgenstein seems to have had
this view, as Jacobsen (1996, pp. 14–17) reminds us.
7 Constitutive Theories 211
5 Summary
In this chapter, we have seen several constitutive positions. They all sub-
scribe to the view that so-called self-knowledge is not the result of any
epistemic achievement—however minimal that might be. Rather, what
goes by that name is the constitutive thesis, according to which, given cer-
tain conditions C, one has a given propositional attitude iff one believes/
judges that one does. We have seen how Shoemaker (§1) is interested in
redeeming the left-to-right side of the thesis—that is, the one concern-
ing transparency. He does so by advocating that the C-conditions should
make reference to a subject endowed with rationality, intelligence and
the relevant psychological concepts. We have objected to Shoemaker’s
claim that self-knowledge is necessary for being a reasoner—that is, for
being able to engage in theoretical and practical reasoning and in belief
revision. In addition, we have maintained that even if it were necessary to
that end, that would just show that if one is such a reasoner, one would
have knowledge of one’s own mental states, but that would be silent on
the provenance of that knowledge. We have also seen how, ontologically,
for Shoemaker first- and second-order mental states do not have separate
existences.
We have then moved on to Wright’s account (§2), which is a vindica-
tion of the right-to-left direction of the constitutive thesis—that is to say,
the side of authority. We have seen how he embraces the Wittgensteinian
idea that this is just a feature of how our communal language game with
psychological avowals is played, which allows for undermining only
when there are positive reasons to think that a subject is self-deceived.
Thus, in order to secure the conceptual truth of the relevant side of the
constitutive thesis, the C-conditions must make reference to the absence
7 Constitutive Theories 213
from several possible objections but also pointed out that its scope is
limited only to commitments and to subjects endowed with the relevant
psychological concepts. It is therefore necessary to integrate it with dif-
ferent ones in order to account for our first-personal knowledge of other
mental states and, obviously, for third-personal knowledge of other kinds
of propositional attitudes and further mental states.
Bibliography
Bar-On, D. (2004). Speaking my mind. Expression and self-knowledge. Oxford:
Oxford University Press.
Bilgrami, A. (2006). Self-knowledge and resentment. Cambridge, MA: Harvard
University Press.
Bilgrami, A. (2012). The unique status of self-knowledge, in Coliva A. (2012),
263–278.
Boghossian, P. (1989). Content and self-knowledge. Philosophical Topics, 17,
5–26.
Boyle, M. (2009). Two kinds of self-knowledge. Philosophy and Phenomenological
Research, 77(1), 133–164.
Boyle, M. (2011a). Transparent self-knowledge. Aristotelian Society Supplementary
Volume, 85(1), 223–241.
Boyle, M. (2011b). Making up your mind. Philosophers’ Imprint, 11(17), 1–24.
Coliva, A. (2009). Self-knowledge and commitments. Synthese, 171(3),
365–375.
Coliva A. (2012c) (ed) The Self and Self-Knowledge (Oxford: Oxford University
Press).
Coliva, A. (2012d) “One variety of self-knowledge: constitutivism as construc-
tivism”, in A. Coliva (ed.) The Self and Self-Knowledge, Oxford, Oxford,
University Press, pp. 212–242.
Davidson, D. (1970). Mental events. In L. Foster & J. W. Swanson (Eds.),
Experience and theory. London: Duckworth. Reprinted in Essays on actions
and events (pp. 79–102). Oxford: Clarendon Press.
Dennett, D. (1987). The intentional stance. Cambridge MA: MIT Press.
Frith, U., & Happé, F. (1999). Theory of mind and self-consciousness: What is
it like to be autistic? Mind and Language, 14(1), 1–22.
Geach, P. (1965). Assertion. The Philosophical Review, 74, 449–465. Reprinted
in Rosemberg, J. F. & Travis, C. (eds) (1971). Reading in the philosophy of
language (pp. 250–261). Englewood: Prentice-Hall.
216 A. Coliva
Hacker, P. (1986). Insight and illusion: Themes from the philosophy of Wittgenstein.
Oxford: Clarendon Press.
Heal, J. (2002). First person authority, Proceedings of the Aristotelian Society, 102,
1–19. Reprinted in (2003). Mind, Reason and Imagination (pp. 273–288).
Cambridge: Cambridge University Press.
Jacobsen, R. (1996). Wittgenstein on self-knowledge and self-expression.
Philosophical Quarterly, 46, 12–30.
Kripke, S. (1982). Wittgenstein on rules and private language. Oxford: Blackwell.
Peacocke, C. (2012). Conscious events and self-ascriptions: Comments on heal
and O’Brien, in Coliva A. (ed), 180–188.
Povinelli, D. J., & Vonk, J. (2004). We don’t need a microscope to explore the
chimpanzee’s mind. Mind and Language, 19(1), 1–22.
Stoneham, T. (1998). On believing that I am thinking. Proceedings of the
Aristotelian Society, 98, 125–144.
Strawson, P. (1962). Freedom and resentment. Proceedings of the British Academy,
48, 1–25.
Tomasello, M., Call, J., & Hare, B. (2003a). Chimpanzees understand psycho-
logical states—The question is which ones and to what an extent. Trends in
Cognitive Science, 7, 153–156.
Tomasello, M., Call, J., & Hare, B. (2003b). Chimpanzees versus humans: It’s
not that simple. Trends in Cognitive Science, 7, 239–240.
Williams, D. (1994). Somebody somewhere. London: Corgi Books.
Wright, C. (1992). Truth and objectivity. Cambridge MA: Harvard University
Press.
Wright, C. (2002). What could antirealism about ordinary psychology possibly
be? The Philosophical Review, 111, 205–233.
Wright, C. (2015). The reality of privileged access. In S. C. Goldberg (Ed.),
Externalism, self-knowledge and skepticism (pp. 49–74). Cambridge:
Cambridge University Press.
8
Pluralism About Self-knowledge
rule that individuates the relevant concept. For doing so will, in a large
amount of cases, presuppose knowledge of one’s own mental states already.
We have proposed—in line with some recent expressivist pronounce-
ments, which are otherwise silent on the particular kind of propositional
attitudes avowed—that the relevant concepts are acquired “blindly”, by
being drilled to use them to give alternative expression to one’s mind,
whenever one is already in a position to judge, based on reasons, that P
is the case or is worth having or doing, and therefore has the correspond-
ing judgement-dependent belief, desire or intention. The relatively late
appearance of psychological concepts—or the relatively late development
of simple tags into concepts, as Gordon would have it—in human beings’
cognitive repertoire thus is imputed to the fact that it takes time for them
to acquire the ability to have these kinds of propositional attitude, rather
than to the fact that it takes time for them to develop a theory of their own
and other minds. Sensitivity to reasons is—according to the picture we
have sketched—what takes time to develop. Once one has acquired such
an ability, psychological concepts come virtually for free. For, if one has
“blindly” acquired the relevant tags already, one needs only using them to
give expression to one’s judgement-dependent propositional attitudes. If,
in contrast, one does not yet have those tags, one simply needs to acquire
them—and do so “blindly”—by being taught to substitute one’s direct
expression of judgement-dependent belief, desire or intention with “I
believe/desire that P” or “I intend to φ”.
This kind of constitutivism, however, cannot be applied outside its
proper boundaries. For one thing, it cannot be of avail in connection
with avowals of sensations or basic emotions, for it does not seem right
that we do bring them about by self-ascribing them. If there is room for
some, much weaker form of constitutivism in this connection, as we see
in §2, there is none for the metaphysically robust variant I have been
proposing for propositional attitudes as commitments. For another, it
cannot be applied to our knowledge of dispositional mental states, let
them be propositional attitudes or the dispositional elements of complex
emotions, as is shown in §3. As we shall presently see (§1), however,
the strong form of constitutivism proposed at the end of the previous
chapter has further limitations when it comes to self-ascriptions of past
propositional attitudes as commitments too. A proper, necessarily plural-
ist account needs to take care of them too.
8 Pluralism About Self-knowledge 219
actual deliberation one made. In this case, one remembers what went
on in one’s mind at a given moment in the past and makes a past self-
ascription based on that. In such a case, the self-ascription is based on
mnestic evidence and is akin to a judgement about one’s material sur-
roundings based on one’s perceptual evidence. In such a case, the self-
ascriptions is not groundless—not because one observes one’s past mental
state or because one infers to it starting from one’s behaviour and further
inner promptings to infer to its likely cause—but because one’s judge-
ment of having had that mental state is based on (mnestic) evidence that
stems directly from that mental event.1 In such a case, there is a pre-
sumption of truth—we usually remember our past, including our past
deliberations, correctly—but surely no conceptual guarantee that one’s
mnestic evidence is right and one’s past self-ascription correct.
This kind of case, though real, is not the norm, however. For it is often
the case that we do not actually have a recollection of the precise act of
deliberation we performed, in order to form a given propositional attitude
as a commitment. Yet we constantly engage in past self-ascriptions of the
relevant mental states. For instance, I have no recollection whatever of
when I decided to enrol for a first degree in philosophy. But I certainly
deliberated to that effect. Here again, the psychological self-ascription “I
decided to enrol for a first degree in philosophy” is actually based on mnes-
tic evidence, but not of one’s past mental state of deliberating to do such a
thing. The evidence has rather to do with one’s memories of some relevant
biographical element, like remembering having considered some options,
right after finishing high school, and having actually enrolled for a phi-
losophy degree when the time to enrol came, together with a recollection
of not being forced to do anything or of having enrolled going through
the motions. It is important to resist thinking of the self-ascription as a
case of inference to the best explanation. For, by judging that I did decide
to enrol for a first degree in philosophy, I am not just making sense of
who I am now by inferring that it is likely that at some point I deliber-
ated to do such a thing. I am sure I did make that decision, at some point
1
In order to avoid thinking of self-ascriptions of past commitments as based on having in view
one’s past mental states, it is useful to think of a subject who, having lost her mid- and long-term
memory, keeps a very accurate diary of her daily activities. She could still make past self-ascriptions,
yet in no sense would they be based on having in view one’s past mental states.
8 Pluralism About Self-knowledge 221
2.1 Sensations
2
I think similar considerations can be put forward for self-ascriptions of ongoing thoughts, mental
images and imagination.
8 Pluralism About Self-knowledge 223
3
To be clear, I do not wish to deny that sensations such as pain, pleasure, hunger, thirst, coldness
and hotness have a distinctive phenomenology. What I do deny is that our concepts are tags we use
by having learnt to identify the relevant sensations while having in view their distinctive phenom-
enology. Rather, these sensations have distinctive behavioural manifestations and we are taught to
replace, or at least to accompany those behavioural manifestations with the relevant self-ascrip-
tions. In contrast, I do think that the phenomenological account would be a non-starter in the case
of propositional attitudes, for there does not seem to be anything phenomenologically salient in the
mere propositional attitude (let it be of belief, or desire, or intention, and so on). If there is some-
thing phenomenologically relevant is the kind of feeligs which can accompany those propositional
attitudes. Furthermore, it does not seem to me that phenomenology would help us distinguish
among the various fine-grained propositional attitudes we may enjoy. For instance, I simply don’t
see what could distinguish a hope from a wish at a merely phenomenological level.
226 A. Coliva
would not know that she is in pain because she feels it, but because she
would see herself scream and moan and would infer that she must be
in pain. That would cast doubt on the fact that a subject of capable of
having sensations and other phenomenally salient mental states at all.
Presumably, people affected by congenital insensitivity to pain with anhi-
drosis (CIPA) exemplify such a possibility.
Furthermore, as we already remarked in Chap. 3 (§1), a subject who
had to infer to her painful sensation would strike us as someone who,
in some sense, is alienated from herself; that is, she would bear to her
mental states the kind of relation she bears to those of others. However,
what makes a mental state of ours, ours, is, besides its origin in our brain
activity, the fact that it is known to us in a way which differs from how we
know of other people’s mental states and in which no one else can know of
it. The failure of groundlessness would therefore result in a lack of “sense
of ownership” over the mental states one would be inferring to have.
Furthermore, if she had to find out through inference to the best
explanation whether she was in pain, when asked “How are you?”, she
would be entitled to respond “I do not know. Let me find out” and she
then would be entitled to preface whatever answer she might wish to give
with “Probably I am in pain” or “It is my hypothesis that I am in pain”.
These responses would actually strike us as a symptoms of mental illness
unless there were reasons to think that the subject is indeed affected by
CIPA. They would strike us as reports on another subject, who is the
subject of that sensation and who would happen to live in one’s body.
Hence, groundlessness seems to be a necessary and a priori element of
what counts as being a subject of sensations—indeed, the subject of those
sensations and not merely the object whom one may observe as having
them—and of what counts as being a rational subject.
In order to redeem transparency, one might say that sensations are nec-
essarily conscious—that is, such that having them coincides with being
aware of them. That is to say, their transparency, on this reading, would
be a necessary and a priori aspect of what goes by the name of “sensation”.
Similar considerations may perhaps be put forward in the case of at least
those emotions that have a distinctive phenomenology to them, as we
shall presently see. As we saw in Chap. 2 (§1), we may distinguish between
sensory states, which need not be conscious—being merely registrations
of variations in proximal stimuli—and sensations properly so regarded,
228 A. Coliva
that these mental states have relational objects and correctness conditions
(at least in the case of perceptions). For I am scared of the dog which is
approaching or I see the cat in front of me or even that there is a cat in front
of me, and I don't simply have a certain feeling or a visual appearance.
Yet going back to basic emotions, the eventual self-ascriptions of fear (and
of other basic emotions) will be groundless, transparent and authoritative.
These features, in turn, are constitutively tied to what it means to possess the
concept of fear (and of other basic emotions) as well as to what it means for
someone to be a rational subject (in a “thick” sense, as we saw). Moreover,
they are constitutively tied to what it means for a subject to be capable of
enjoying basic emotions and to be the subject of those very emotions.
the object perceived is in clear view. If all these conditions obtain, together
with the usual ones, namely conceptual mastery, lucidity, attentiveness and
cognitive well-functioning, then if one sees a dog, one will immediately be
in a position to judge “I see a dog” and one’s self-ascription will be guar-
anteed of being correct. Similarly, if one has a red after-image, say, one will
be immediately in a position to judge “I am having an experience of red”
(or “I am appeared to redly”) and one’s self-ascription will be authoritative4.
To redeem the constitutive thesis in connection with the kinds of men-
tal state we are considering here, it is important to stress the strong link
between perception and action. Hence, since we have allowed for the
possibility of unconscious perceptions (see Chaps. 2 (§1) and 3 (§1)),
it is crucial to note that the actions we would perform on the basis of
unconscious perceptions would fall out of responsible agency. As we saw,
a subject affected by blind-sight does not know either that she is catch-
ing a ball or why. Hence, she cannot be held responsible for her action.
Groundlessness, transparency and authority with respect to perceptions
are therefore necessary and a priori elements only of conscious perceptions,
which, in turn, are constitutive elements of responsible outer actions.
Hence, to conclude, groundlessness, authority and transparency will
be necessary and a priori features constitutively tied to what it means to
possess the concept of seeing (and of other perceptual states), of having
perceptual experiences and conscious perceptions, and indeed of what
it means to be the subject of those very mental states. Finally, they are
constitutively tied to what it means for someone to be a responsible agent
and a rational subject (in a “thick” sense).
4
This account seems to me largely in keeping with the gist of Evans’s account of our knowledge of
our seemings, put forward in Evans (1982, chapter 7).
8 Pluralism About Self-knowledge 233
(a) They are not the result of a conscious deliberation—that is, a judge-
ment, on a subject’s part, based on considering and, in particular, on
assessing (or even being able to assess) evidence in favour of P (or of P
is worth pursuing, it would be good if P happened, etc.);
(b) they are not within one’s direct control, being rather something one
finds oneself saddled with;
(c) hence, they are not something one will be held rationally responsible
for.
As we saw, some examples of mental states that will satisfy these condi-
tions are unconscious mental states of a Freudian kind, which, however, can
be operative in shaping a subject’s behaviour. As noted, one can acquire
knowledge of them through therapy, which is clearly a third-personal
means to get knowledge of oneself—a means, moreover, of an interpreta-
tive nature. That is to say, patterns of observed behaviour are subsumed
under a theory which explains them as due to certain unconscious mental
states. Two main sorts of error are possible—namely, either focusing on
irrelevant patterns of behaviour or mistakenly identifying their cause. A
distinctive trait of third-personal self-knowledge of these mental states
is that it is not sufficient, just by itself, to bring about a change in one’s
attitudes even if it can prompt one to try to change them in the future.
As already remarked in Chap. 2 (§2), deep-seated biases would be
another case in point, like gender preferences in offering certain kinds
of job preferably to male (or female) candidates. Again, one may act on
234 A. Coliva
and the internal cues we attend to may remain identical but be ranked
or arranged differently, so as to result in a classic case of aspect-switching.
For, as is made familiar by Wittgenstein's discussion of the duck-rabbit
case, in typical aspect-switching the lines remain identical and yet we
see them differently. Still, in these cases, inference is involved at least in
the following sense. While there is no inference to the best explanation,
the characteristic notes of the newly acquired (or applied) concept guide
the process of subsuming the observed pattern of behaviour and inner
elements under the concept. That may happen very rapidly, and so, phe-
nomenologically, we may not even seem to be engaging in inferences, yet
epistemically that is what grounds our judgement “I am F” (where F is a
psychological concept). It is one possible case of self-interpretation and
it can go wrong because, after all, the pattern observed may not fit the
concept or we may have ignored other features which would, if taken into
account, lead to a different psychological self-ascription.
In other cases, our self-knowledge is based on testimonial evidence. If
a trustworthy person (or at least someone we have no reason to think is
conceptually incompetent or willing to deceive us) tells us we are arro-
gant, say, we may form that belief, based on her telling us so. Our third-
personal self-knowledge would thus crucially have a social dimension to
it, and it would not be confined to the highly peculiar case in which
we gain self-knowledge by consulting a professional therapist. Normal
chats with family and friends can actually be an important source of self-
knowledge. Indeed, even gossip, to the extent that one is willing to con-
sider it a way of gaining testimonial knowledge, can make us aware of
dispositional aspects of our character we might otherwise ignore. In par-
ticular, it can make us aware of the kind of reactions we tend to elicit in
other people and would thus make us aware of those dispositional aspects
of our character. Finally, it should be noted that testimony and gossip
really allow us to gain a third-personal perspective onto ourselves, in a
way in which the other forms of third-personal self-knowledge reviewed
so far will always depend on a first-personal application of third-personal
methods.
As already noted (see Chap. 3, §3), sometimes we figure out our com-
plex emotions, such as love towards someone, by reflecting on our behav-
iour and inner phenomenology in their presence or at the prospect of
238 A. Coliva
meeting or losing them. This is the case illustrated by the passage from Jane
Austen’s Emma quoted in the second chapter. Obviously, we can go wrong
in our self-interpretation, based on abductive reasoning. Our behaviour
and feelings may not be symptoms of love but of being possessive and of
dreading the prospect that a good friend will no longer be so close to us as
the result of getting married to another person. Alternatively, we can mis-
take our behaviour and feelings as symptoms of love, when they are in fact
signs of our vanity and complacent attitude towards those people who are
usually kind to us. Sometimes, in contrast, we are self-blind in the sense
that despite showing many of the characteristic symptoms of love, say,
towards someone, we do not realise that we are in love with them, even
if we do possess the concept of love and are generally capable of correctly
applying to it to other people. These forms of self-blindness can have dif-
ferent causes, which may be due to psychological or to cultural factors or
even to deeper unconscious causes.
There are also less dramatic cases, where we simply conceptualise our
behaviour and inner promptings in a mistaken way because we more or
less consciously ignore or downplay some other aspects of our behav-
iour and feelings. Hence, we arrive at a mistaken individuation of some
of our character traits. For instance, there is a difference between being
driven at work and being obsessed or even ruthless. Application of these
concepts to ourselves is not (necessarily) a case of abductive reasoning.
Quite often, it is just inferential conceptual application, where the char-
acteristic notes of the concept guide our psychological self-ascription
but where we can go astray by not paying (enough) attention to some
aspects of our behaviour and feelings which would actually call for a dif-
ferent categorisation. Similarly, we may not believe what a trustworthy
person tells us about our character or our attitude in a given situation. Of
course, there may be reasons why we do go so astray, or indeed why we
may not trust what we are told by another person. Clearly, it is not nice
to realise that one is obsessed or even ruthless and not merely driven.
Similarly, it is not nice to realise that one is an arrogant person. Such
a realisation would presumably call for the endorsement of a different
conception of oneself and for a re-orientation of one’s behaviour. Both
outcomes would presumably pose problems of a psychological and even
of a moral nature. It is sometimes this kind of phenomenon which is
8 Pluralism About Self-knowledge 239
4 Summary
In this chapter, we have seen how the metaphysically robust constitutive
account, which we uphold for self-knowledge of propositional attitudes
as commitments, cannot be extended to our knowledge of other kinds of
mental states. In particular, in §1, we have seen that even when it comes
to past self-ascriptions of commitments, it will have to be integrated with
an evidential account. By that, we mean simply that the relevant self-
ascriptions are based on mnestic evidence and therefore are not groundless,
even though they are not based on the observation of one's past inten-
tion or on inference to the best explanation starting from the observation
of one's own behaviour. That opens up the possibility of forms of self-
blindness and of mistaken self-ascriptions. Yet there is a general presump-
tion that subjects do remember their past (mental) actions correctly and
that their self-ascriptions—arrived at by taking into account the relevant
mnestic evidence—are correct. It is that presumption that in turn is con-
stitutive of what it means to be a rational, well-functioning human being,
who can be held responsible for her (mental) actions through time. Yet
this does not mean that one’s knowledge of those past commitments can
be accounted for by means of a constitutive account of self-knowledge.
We have then moved on to our self-knowledge of sensations, basic
emotions, perceptions and perceptual experiences (§2). Here again, a
8 Pluralism About Self-knowledge 241
The paradox arises from the fact that the conjuncts in (1) and (2) have
independent truth conditions and therefore can be both true (or false) at
once. If so, then, (1) and (2)—that is, these conjunctive propositions—
could be true and yet judging or asserting them would be so deeply absurd
as to result in self-defeat. We sense that a subject cannot really believe
that it is raining, say, while also judging that it is not, or that he cannot
disbelieve that it is raining while also judging that it is. Yet the nature
of Moore’s paradox remains elusive in many ways and several different
explanations of it have been proposed. Despite its elusiveness, study
1
Sometimes (1) is phrased differently—namely, (1*) “P and I believe that not-P”. Notice that (1*)
can be obtained from (1) through uniform substitution of P with not-P and commutation, pro-
vided that double negation elimination held. (1) (or 1*) is called the commissive form of the para-
dox and (2) the omissive one. See Green and Williams 2007, p. 5.
2
Green and Williams 2007, pp. 3–4.
3
Moran 1997, p. 143.
4
Baldwin 2007, pp. 76–89. An analogous suggestion can be found in Millar 2004, p. 125. It will
be assumed that broadly functionalist accounts of beliefs and normative ones are the main options
on the table. Eliminativism, besides being problematical in its own right, would be a non-starter in
connection with Moore’s paradox, for it would actually condone “It is raining but I do not believe
it”, since, by its lights, there would be no beliefs. See Turri 2010, pp. 35–39. As will become appar-
ent, in some cases (1) and (2) can be coherently judged/asserted. This, however, does not mean that
there are never cases in which those judgements/assertions would result paradoxical. The task will
be to spell out the conditions in which judging or asserting either (1) or (2) would result in
self-defeat.
Appendix: Moore’s Paradox 245
5
Moore 1942, pp. 540–543. See also Moore 1944, p. 204. Wittgenstein PI, II, xi pp. 190–192 and
1980, §§90–96. Among Moorean analyses, see Baldwin 1990 and Shoemaker 1995. Among
Wittgensteinian ones, see Collins 1996; Heal 1994; Lee 2001; Linville and Ring, 1991 and
Malcolm 1995.
6
Heal 1994, p. 6, Shoemaker 1995, p. 213, Moran 1997, p. 144 and Baldwin 2007, pp. 78–9
emphasise this point. Green and Williams 2007, p. 6 date it back to Sorensen 1988. Chan 2010
tries to make sense of the idea that there might be pragmatic paradoxes at the level of thought. Yet,
he concludes that, in the case of Moore’s paradox, this would nevertheless lead to an unsatisfactory
account.
7
Linville and Ring 1991 are committed to this claim. Heal 1994, p. 6, places as a requirement upon
any satisfactory analysis of Moore’s paradox that “the solution must identify a contradiction, or
something contradiction-like, in the Moorean claims”. Both requirements will be criticised at
length, for different reasons, in the text.
246 Appendix: Moore’s Paradox
application of whatever principle, the very content of (1) and (2) comes
down to a contradiction of the form “p and not-p” or appeals to some
pragmatic norm governing the speech act of assertion.
At the core of all Moorean analyses lie two main claims:
(a) Neither (1) nor (2) is a contradiction. For they are conjunctions of
propositions with different truth conditions. Hence, they can both
be true at the same time.
(b) Given (a), there is nothing wrong with (1) and (2) as such. Thus, there
is nothing wrong with believing or entertaining their respective con-
tents in thought. For one can perfectly well think that one has a false
belief—(1)—and that is ignorant with respect to P—(2). Rather, the
oddity of Moorean contents—what, in effect, entitles us to talk about
a paradox in their connection—is the fact that when one asserts either
(1) or (2), then one is doing something pragmatically inconsistent.
8
Heal 1994, pp. 11–2.
Appendix: Moore’s Paradox 247
(b’) is the only feature which, at least prima facie, seems to speak in favour
of Wittgensteinian analyses. For (a’) has the well-known and unpalatable
consequence of depriving one of the means of explaining the following
platitudes.
Embedding: (1) and (2) can be embedded in wider contexts, like “Suppose
that I believe that P, but it is not the case that P” and “Suppose that it is the
case that P, but I do not believe it”. The possibility of embedding (1) and
(2) in suppositions, in turn, explains why they can occur as antecedents in
correct conditional statements.10
9
Moran 1997 denies that this is actually Wittgenstein’s position. This, however, is the vulgata and
constitutes the bulk of what I call “Wittgensteinian” analyses of Moore’s paradox.
10
Wittgenstein himself was aware of this. See in the following for a (failed) attempt at reconciling
this platitude with a Wittgensteinian account of Moore’s paradox.
248 Appendix: Moore’s Paradox
Past: (1) and (2) can be turned into the past. “I believed that P but it was
not the case that P” or “I did not believe that P but it was the case that P”
is perfectly fine.
Third-person: (1) and (2) can be put into the third person. “She believes
that P, but it is not the case that P” and “It is the case that P, but she does
not believe it” can both be used to make accurate statements.
11
One might object that one can suppose anything. For instance, opponents of dialetheism suppose
that P and not-P in order to show that any consequence follows. However, exactly this case shows
that the supposition was not coherent in the first place. At any rate, formal contradictions at least
do not cease to be such when changed of tense or person.
12
By contrast, the meaning of “to believe” both in the first-person past and in the third person is
descriptive of a mental state, which one is either ascribing on the basis of behavioural criteria
(sometimes even in one’s own case) or on the basis of memory.
Appendix: Moore’s Paradox 249
13
de Almeida 2001, p. 38.
14
Heal 1994, p. 7, who favours a Wittgensteinian analysis of Moore’s paradox, acknowledges this
point, but, surprisingly, goes on to offer a uniform account of it, where both (1) and (2) come down
to contradictions of the form “P and not-P”. However, such a result can be obtained only by taking
“I do not believe that P” as equivalent to “I believe that not-P” and thus by taking its assertion as
equivalent to the assertion of “not-P”.
15
This point (see de Almeida 2001, p. 39) is not sufficiently appreciated by Heal (1994, pp. 20–24),
who puts forward the view that self-ascriptions of belief have a double role: that of describing one’s
attitude and that of expressing the belief in p. Only the latter aspect of the concept of belief would
lend itself to a Wittgensteinian treatment of Moore’s paradox. Yet, even if Wittgensteinian analyses
worked for certain instances of Moore-paradoxical sentences, they would not work for others (or,
in Heal’s account, even if they worked for the latter reading of Moore-paradoxical sentences, they
would not work for the former). Thus, Wittgensteinian analyses do not explain away Moore’s para-
dox and, when faced with genuine instances of it, are left with no means to address it.
250 Appendix: Moore’s Paradox
16
This would be obvious if one held the principle that “what can be (coherently) believed con-
straints what can be (coherently) asserted” but not vice versa (Shoemaker 1995, p. 227, 1n holds the
quoted principle but does not explicitly commit himself to the negation of its converse. In contrast,
de Almeida 2001, p. 33 holds the whole principle). Wholehearted Wittgensteinians would proba-
bly reject this principle, but more recent proponents of the Wittgensteinian analysis of Moore’s
paradox, such as Jane Heal, do not explicitly reject it. Therefore, if one were to defend the
Wittgensteinian analysis of Moore’s paradox one had better either motivate the rejection of the
principle, or find a way of extending the Wittgensteinian analysis to the case of Moore-paradoxical
judgements. As we shall see, this is no trivial matter.
17
Evans 1982, pp. 225–226. As we saw in Chap. 5, §3, the idea is that self-ascriptions of beliefs are
not based on looking inwards and on finding out that one has a given mental state. Rather, they
draw on the same epistemic resources that would license the mere judgement (or assertion) that
P. While there is much to applaud in the attempt to avoid the Cartesian epistemology of the mind,
Evans’s point does not help the cause of Wittgensteinians, nor is very convincing as an account of
self-knowledge. For a sustained application of Evans’s point to self-knowledge, see Gallois 1996 and
Fernández 2013. For a critique, see Gertler 2011a, pp. 174–177, Coliva 2014 and Chap. 5, §3.
In response to the suggestion that one may use Evans’s point to explain Moore’s paradox indepen-
dently of developing it into a viable account of self-knowledge, it should be noted that it would be
an extremely implausible move to make. For what is under attack is exactly the distinctive claim
that would be used to explain the paradox—namely, that the very same grounds would support
both “P” and “I believe that P”.
Appendix: Moore’s Paradox 251
18
Square brackets will be used for propositional contents.
19
This is a difficult rule to motivate, since it would imply omniscience.
20
Heal 1994, p. 11. Williams 2007 provides a sustained defense of the application of Evans’s point
(see also his 2004). However, he makes use of two principles—that he calls “Evans’s Principle” and
“Analogue of Evans’s Principle” (at p. 95 and 101, respectively)—that are not clearly Evans’s and
relies on the problematical idea that Evans’s point can be developed into an adequate theory of self-
knowledge. For a discussion, see Vahid 2005; Brückner 2006, 2009. For a response, see Williams
2009.
252 Appendix: Moore’s Paradox
Let us then take stock and put down clearly what we are looking for.
We want to be able to explain why holding in thought [I believe that P,
but it is not the case that P] or [I do not believe that P, but it is the case
that P] would be so absurd as to result in self-defeat. Yet we do not want
to do so by saying that those contents are equivalent to [P and not-P].
Jane is married to Jim. They have been married for several years and have a
daughter. Jane is often at home, on her own, attending to domestic chores.
From time to time, she feels lonely and wishes that she had pursued her
own career. More often than not, however, she feels much rewarded by the
fact that her family is so serene. Indeed, when she meets with her friends,
who sometimes complain about their husbands, she cannot help remarking
that her life makes her happy and that her husband is adorable and com-
pletely trustworthy.
Still, it often happens that, while preparing for the laundry, Jane care-
fully searches Jim’s pockets. While tidying up his studio, she opens and
examines all the drawers. While dusting the furniture, she lingers on the
screen of his laptop, left open on the incoming messages.
One day, Ann, a psychoanalyst friend of Jane’s, approaches her and tells her
about Freud’s theories concerning the unconscious. Little by little, the deep
significance of a whole series of previously meaningless actions is disclosed
to Jane. Ashamedly, she realises that all that attention spent over the con-
tent of her husband’s pockets was a sign of her being insecure about him.
All that dusting the screen of his laptop, a symptom of her thinking that he
might have some intimate correspondence with another woman. Still, Jane
knew all too well that Jim had always been the most truthful of men. The
thought popped into her head: “I do believe that Jim is unfaithful to me,
but he is not”.
254 Appendix: Moore’s Paradox
21
Lee 2001, pp. 366–7 has a similar story. Remarks that are congenial to the gist of Jane’s story can
be found in Moran 1997, Martin 1998, Williams 2006, and Gallois 2007. Chan 2008 presents a
case of wishful thinking in order to make a similar point. Gertler contends, contra Gendler, that
assertions such as Jane’s are, in the relevant context, perfectly legitimate and such that the doxastic
conjunct expresses full and real belief. See Gertler 2011b, and Gendler 2008.
22
Other kinds of analysis of Moore’s paradox, which hinge on the idea of unravelling the complex-
ity of the notion of belief, are Heal’s and Baldwin’s. They stand opposed to other strategies, such as
Shoemaker’s, which try to derive an explanation of Moore’s paradox from an account of self-knowl-
edge. Self-knowledge is a necessary condition for giving rise to the paradox, but it is not sufficient
for it.
23
The reader will make the necessary adjustments to apply the same kind of claim to (2).
Appendix: Moore’s Paradox 255
24
Gallois 2007 defends this point, although he prefers to say that the belief must be consciously
held. Williams 2006 talks of consciousness in this connection, but he is happy to endorse
Rosenthal’s higher-order account of consciousness, in which the latter comes down to
self-knowledge.
25
See Chap. 7 (§§3–4).
26
Self-deception would then be a case in which one’s (unconscious) dispositions are not in keeping
with one’s explicitly held commitments and this creates a conflict in one’s outward behaviour (con-
tra what Moran 1997 maintains).
27
As will become evident in the following, Jane’s irrationality is not the irrationality of knowingly
holding a contradiction or of undertaking mutually inconsistent commitments (if this is a genuine
possibility at all). Rather it is the irrationality of not aligning one’s dispositions with one’s
commitments.
28
Functionalism treats beliefs as dispositions, and therefore as lacking intrinsic normativity. Hence,
as noted in Heal 1994, pp. 12–20, it is bound to dissolve Moore’s paradox. A similar point can be
found in Baldwin 2007, p. 84.
29
Bilgrami 2006, chapters 4–5. Millar 2004 considers beliefs as intrinsically normative mental
states. For further discussion, see chapter 1 (§2) and chapter 6 (§§3–4).
256 Appendix: Moore’s Paradox
30
When they are, they are usually known in a third-personal way. That is to say, by observation and
inference to the best explanation.
31
Freudian unconscious mental states do not exhaust the category of dispositional beliefs and
desires. There may be also beliefs and desires that shape one’s character traits. Moreover, insofar as
one is willing to grant beliefs and desires to animals and infants, their intentional mental states
would be dispositions.
Appendix: Moore’s Paradox 257
32
See in particular Moran 2001 and Millar 2004. Cf. Chap. 3 (§2) and Chap. 7 (§§3–4).
33
Bilgrami 2012, p. 213.
258 Appendix: Moore’s Paradox
the kind of actions that one ought to perform given that belief. However,
theorists disagree on whether, in order to count as really having the com-
mitment in question, one also has actually to comply with the course of
action it mandates, at least on occasion, or whether no such behaviour
is required in order to count as having the commitment.34 Those who
favour the former option will appeal to the idea that of a subject who,
for example, persistently professed “I ought to help the poor” and never
did anything to that effect, we would presumably be inclined to say that
she thinks she has the commitment to help the poor but actually she
does not. In contrast, those who favour the latter option would argue
that insofar as such a subject recognised what would be required of her
and were self-critical or accepted criticism for not complying with her
commitment, she would count as having it nonetheless. Clearly, the two
readings impose more or less stringent requirements with respect to the
conditions that have to be met in order to have beliefs as commitments
and propose a different configuration of the interplay between commit-
ments and dispositions. This issue is of great interest and significance; it is
not, however, of immediate relevance for the present discussion and I will
show why in due course.35 What is relevant, at this stage, is simply that
the internal link between the content of one’s belief as a commitment
and the course of action it mandates be recognised (yet not necessarily
implemented, on at least one possible understanding of the notion of a
belief as a commitment) by the subject who could not otherwise have
that belief as a commitment.
Let us now turn to the kind of analysis of Moore’s paradox that can be
offered if one held a “liberal” notion of commitment, which allows for the
possibility that a subject could knowingly and willingly undertake incom-
patible commitments. Accordingly, the analysis of Moore’s paradox would
be as follows.36 Being committed to [P], one ought to use it as a premise
34
Millar 2004 seems to maintain the former option, whereas Bilgrami (2006, 2012) defends the
latter.
35
For a defence of the latter interpretation, see Chap. 2, §2 and Chap. 7, §§3–4.
36
I surmise that Baldwin 2007 (p. 86) puts forward something along these lines, for he allows for
the possibility of finding “ourselves from time to time with inconsistent commitments”. What
would be absurd, on his analysis, would be “to make commitments whose inconsistency is obvious
in the very judgment itself ”.
260 Appendix: Moore’s Paradox
37
Obviously, she would also commit herself to using [I believe that P] as a premise of her
reasoning.
Appendix: Moore’s Paradox 261
mutandis, given the omissive form of Moore’s paradox, one could then be
open-minded with respect to [P], while also endorsing it). In particular,
it does not seem impossible to reason from contradictory premises which
are actually knowingly and willingly upheld and not merely supposed,
like in the case of (1), or to reason from genuine open-mindedness with
respect while also actually holding that P, in the case of (2). It is just that
these doxastic situations would—unluckily—be irrational ones. On such
a view, rational norms would be external to one’s beliefs as commitments
and would be brought to bear on them just as norms of etiquette can be
brought to bear on one’s way of sipping tea or holding a cup.38 That is to
say, as post hoc normative evaluations of doxastic situations (or actions)
which would have been instantiated (or performed), nevertheless. Yet,
if it were possible for a subject to endorse incompatible commitments,
what would then prevent one from judging and even from asserting that
one did? Of course such a situation would be comparatively rare—or so
one would hope. However, if it is a possibility, the corresponding judge-
ment and assertion should be too. The absurdity perceived in hearing
either (1) or (2) would be just due to the fact that we are inductively
unfamiliar with open admissions of irrationality, which, however, on the
proposed conception of commitment, would be entirely possible.
Yet, if this were the explanation of why we find the assertion of (1)
or (2) absurd, then to label Moorean contents “paradoxical” would be
an overstatement. There are plenty of assertions that strike us as weird
because—supposing they are true and sincere—they are at odds with
our experience, but this would not justify calling them “paradoxical” (for
example, “He is not breathing but he is alive”). Furthermore, it was a
datum of the problem that when we hear Moorean sentences we perceive
some kind of real impossibility—a form of self-defeat—despite the lack of
a contradiction. Whereas, so far, our analysis has revealed no impossibil-
ity, just the fact that if those judgements were actually made, they would
correspond to an irrational state of mind.
In order to justify the idea that there is a real impossibility here and
hence a genuine paradox, we should understand the notion of a belief as
38
I would like to thank Margaret Gilbert for raising an objection that made me realize more clearly
what seems to me to be wanting in the first proposed account of Moore’s paradox.
262 Appendix: Moore’s Paradox
39
A similar conclusion can be reached by considering the ‘oughts’ as input that should be appreci-
ated by a subject in order to have the relevant beliefs as commitments—namely, that one ought to
Appendix: Moore’s Paradox 263
about logic and the internal link between a belief as a commitment and the
kind of action it mandates and one ought to see oneself as bound to, in order
to count as having the commitment. Of course, since we are open to the
possibility that a subject may have a commitment while behaving contrary
to what it mandates, we have to be open to the possibility that she might act
contrary to what the belief (as a commitment) that it is raining mandates,
in the kind of scenario just envisaged. However, in such an event, so long as
she did recognise that she has not “lived up” to her commitment, she would
count as having it in the first instance, nonetheless. Notice, once again, that
even if she behaved contrary to her commitment and therefore did not have
the relevant first-order disposition, she should have a related second-order
one, which depends precisely, as required, on seeing herself as bound to act
in certain ways. Hence, if she did not act accordingly, she should be self-
critical or accept criticism from others for not doing so. What could not be
done, therefore, on such a less demanding notion of a commitment, is not to
have a belief as a commitment and act in ways that run contrary to it. Rather,
it is to have the former and think it may be all right to act in ways that run
contrary to it. Yet, to stress, the constitutive connection that is necessary in
order to come to grips with Moore’s paradox—that is, the one concerning
the content of one’s commitments and the kind of actions mandated by
them one should see oneself as bound to—would be respected nonethe-
less. That is to say, even if one might fail to bring these actions about. Since
the courses of action mandated by inconsistent commitments are mutually
exclusive, one can’t have the belief as a commitment that P while also know-
ingly and willingly assenting to its negation (or be open-minded with respect
to [P] while knowingly and willingly assenting to it or to its negation), even
if one actually behaved in ways that run contrary to [P].
By contrast, as we have seen with Jane’s case, one can perfectly well
have the belief as a disposition that p while also knowingly and willingly
assenting to its negation. Hence, one might be saddled with beliefs (as
believe that P only if there is sufficient evidence in its favour. So one could self-ascribe the belief as
a commitment that P only if one had sufficient evidence in favour of [P]. Yet one cannot have suf-
ficient evidence for [P] and also and at the same time for its negation. Hence, one cannot possibly
have the inconsistent commitments entailed by (1). Similarly, one can be committed to open-
mindedness with respect to [P] only if one had neither sufficient evidence in favour of [P] nor for
its negation. Yet one cannot have, at once, no sufficient evidence for [P] and [not-P] and sufficient
evidence for of [P]. Hence, one cannot possibly have the inconsistent commitments entailed by (2).
264 Appendix: Moore’s Paradox
dispositions) that one does not endorse reflectively, act on them and even
overtly recognise the kind of situation one is in. Yet it seems that, on
such a more demanding notion of a belief as a commitment, one cannot
even have the belief as a commitment that P if one also assented—and
thereby committed oneself—to its negation. The assent to the negation
thus would undo one’s previous commitment to [P].
Therefore, given (1), if the doxastic conjunct were the self-ascription
of a commitment, understood in the way proposed, that commitment
could so much as exist only as long as no known and willing assent to
the negation of its content were in place. Hence, if the doxastic conjunct
is a self-ascription of a genuine commitment, no assent to its negation is
possible. Conversely, if such an assent is in place, then there cannot be
any commitment to [P] in the first place.
Take now an instance of (2)—[I do not believe that P, but it is the
case that P]—where the first conjunct is taken as a self-ascription of a
commitment, understood in the way proposed. By judging that P is the
case, one would undo one’s previous commitment to open-mindedness
with respect to [P]—that is, one’s commitment to using neither [P] nor
[not-P] as a premise of one’s reasoning. Hence, again, if the doxastic con-
junct is an ascription of a genuine commitment to open-mindedness, no
known and willing assent to its embedded content is possible. Yet, if such
an assent is in place, there cannot be any genuine commitment to open-
mindedness in the first place.
Therefore, the moral to draw is that any creature capable of beliefs as
commitments could not possibly judge either (1) or (2), despite the fact
that their conjuncts have independent truth conditions, when (1) and
(2) are construed as satisfying conditions (I)–(III). For to do so would
actually be self-defeating.
Finally, the absurdity of asserting either (1) and (2) can be explained
as follows. Usually, though not invariably (as we saw, for instance, with
Jane’s case), we assert the contents of our beliefs as commitments.40 As
we have seen, however, if (1) and (2) are asserted as involving beliefs as
40
This does not require taking issue with any of the various proposals about the norm of the asser-
tion available in the growing literature on that topic. For the claim is that we usually assert the
content of our commitments, not that this is the norm of correct assertions.
Appendix: Moore’s Paradox 265
An Objection
The previous account of Moore’s paradox gives rise to an immediate
objection, for it seems that sometimes we find ourselves with inconsistent
commitments. Consider, for instance, the case of a man with two families
who at lunchtime says to his first family “I will go on holiday with you[family
1]” and at dinnertime says to his second family “I will go on holiday with
you[family 2]” and thereby undertakes inconsistent commitments.41
Let us analyse this case in more detail.42 There seem to be only four
options. First, he means to compatibilise by dividing his holidays
between the two families (or by taking them all to the same place at the
same time). Second, he has changed his mind between lunch and din-
ner and will go on holiday with the second family only. Third, he has
forgotten the promise made over lunch and therefore is not knowingly
and willingly undertaking two incompatible commitments.43 Finally, by
willingly trying to bind himself to knowingly incompatible courses of
41
Another possibility is to have beliefs as commitments that, though not incompatible as such, in
given circumstances, may impose incompatible courses of action. Consider the belief as a commit-
ment that a supervisor should help her students and the other belief as a commitment that one
should give an academic position to the best candidate. Obviously, there is no incompatibility
between the two commitments. Hence, one can knowingly and willingly assent to both. If, how-
ever, the situation arises when there is a conflict, one will have to give priority one commitment
over the other.
42
Even if, as Marcello Fiocco pointed out to me, this case is not one of allegedly inconsistent beliefs
as commitments, I think it is useful for illustrative purposes, since beliefs as commitments are spe-
cies of commitments.
43
Cases like Frege’s puzzle where one can assent to “Hesperus is a planet” while denying or not
assenting to “Phosphorous is a planet” are cases in which the beliefs are incompatible but this is not
known to the subject, who precisely ignores that Hesperus is identical to Phosphorus. Hence, it is
important to qualify the claim here: it is only for knowingly incompatible beliefs as commitments
that one cannot (knowingly and willingly) have both of them at once. Another way to see the same
point is that to judge “I believe that Hesperus is a planet, but Phosphorus is not” (for a subject who
did not know that Hesperus is Phosphorus) would not be Moorean-paradoxical. I would like to
thank John Hawthorne for raising an objection that made me think more about these cases.
266 Appendix: Moore’s Paradox
Gallois, A. (1996). The world without, the mind within. Cambridge: Cambridge
University Press.
Gallois, A. (2007). Consciousness, reasons and Moore’s paradox. In M. Green
& J. N. Williams (Eds.), Moore’s paradox. New essays on belief, rationality and
the first person (pp. 165–188). Oxford: Clarendon Press.
Geach, P. (1965). Assertion. The Philosophical Review, 74, 449–465. Reprinted
in Rosemberg, J. F. & Travis, C. (Eds.) (1971). Reading in the philosophy of
language (pp. 250–261). Englewood: Prentice-Hall.
Gendler, T. (2007). Self-deception as pretense. Philosophical Perspectives, 21,
231–258.
Gendler, T. (2008). Alief and belief. Journal of Philosophy, 105(10), 634–663.
Gertler, B. (2011a). Self-knowledge. London/New York: Routledge.
Gertler, B. (2011b). Self-knowledge and the transparency of belief, in
Hatzimoysis A. (ed), 125–145.
Gilbert, D. (2006). Stumbling on happiness. New York: Knopf.
Goldman, A. (1993). The psychology of folk psychology. Behavioral and Brain
Sciences, 16, 15–28.
Gopnik, A. (1983). How we know our minds: The illusion of first-person
knowledge of intentionality. Behavioral and Brain Sciences, 16, 1–15.
Reprinted in Goldman, A. (ed) (1993). Readings in philosophy and cognitive
science. Cambridge, MA: MIT Press.
Gordon, R. (1995). Simulation without introspection or inference from me to
you, in Davies M. and Stone T. (1995b) (Eds.), 53–67.
Gordon, R. (2007). Ascent routines for propositional attitudes. Synthese, 159,
151–165.
Gordon, R. (2009). Folk psychology and mental simulation, The Stanford ency-
clopedia of philosophy (Fall 2009 Edition), Edward N. Zalta (Ed.), URL =
http://plato.stanford.edu/archives/fall2009/entries/folkpsych-simulation/
Green, M., & Williams, J. N. (Eds.). (2007). Moore’s paradox. New essays on
belief, rationality and the first person. Oxford: Clarendon Press.
Hacker, P. (1986). Insight and illusion: Themes from the philosophy of Wittgenstein.
Oxford: Clarendon Press.
Hatzimoysis, A. (Ed.). (2011). Self-knowledge. Oxford: Oxford University Press.
Heal, J. (1994). Moore’s paradox: A Wittgensteinian approach. Mind, 103(409),
5–24.
Heal, J. (2002). First person authority, Proceedings of the Aristotelian Society,
102(1), 1–19. Reprinted in (2003) Mind, reason and imagination (pp. 273–
288). Cambridge: Cambridge University Press.
Bibliography 273
Wegner, D., & Wheatley, T. (1999). Apparent mental causation: Sources of the
experience of will. American Psychologist, 54, 480–492.
Williams, D. (1994). Somebody somewhere. London: Corgi Books.
Williams, J. N. (2004a). Moore’s paradoxes, Evans’s principle and self-knowl-
edge. Analysis, 64(4), 348–353.
Williams, M. (2004b). Wittgenstein, truth and certainty. In M. Kölbel & B.
Weiss (Eds.), Wittgenstein’s lasting significance (pp. 249–284). London/New
York: Routledge.
Williams, J. N. (2006). Moore’s paradoxes and conscious belief. Philosophical
Studies, 127(3), 383–414.
Williams, J. N. (2007). Moore’s paradox, Evans’ principle, and iterated beliefs.
In M. Green & J. N. Williams (Eds.), Moore’s paradox. New essays on belief,
rationality and the first person (pp. 90–113). Oxford: Clarendon Press.
Williams, J. N. (2009). Justifying circumstances and Moore-paradoxical beliefs:
A response to Brückner. Analysis, 69(3), 490–496.
Williamson, T. (2000). Knowledge and its limits. Oxford: Oxford University
Press.
Wilson, T. (2002). Strangers to ourselves. Discovering the adaptive unconscious.
Cambridge, MA: MIT Press.
Wimmer, H., & Perner, J. (1983). Beliefs about beliefs. Representation and
constraining function of wrong beliefs in young children’s understanding of
deception. Cognition, 13, 103–128.
Wittgenstein, L. (1980). Remarks on the philosophy of psychology (Vol. 1). Oxford:
Blackwell.
Wright, C. (1985). Facts and certainty. Proceedings of the British Academy, 71,
429–472.
Wright, C. (1986). Does philosophical investigations §§258–60 suggest a cogent
argument against private language? In P. Pettit and J. McDowell (Eds.),
Subject, thought and context. Oxford: Oxford University Press. Reprinted in
Wright C. (2001), 223–290. (Referred to as Wright 2001c).
Wright, C. (1987). On making up one’s mind: Wittgenstein on intention. In P.
Weingartner & G. Shurz (Eds.), Logic, philosophy of science and epistemology:
Proceedings of the XIth international Wittgenstein symposium. Vienna: Holder-
Pickler-Tempsky. Reprinted in Wright C. (2001), 116–142. (Referred to as
Wright 2001a).
Wright, C. (1989a). Wittgenstein’s rule-following considerations and the central
project of theoretical linguistics. In A. George (Ed.), Reflections on Chomsky.
Oxford: Blackwell. Reprinted in Wright C. (2001), 170–213. (Referred to as
Wright 2001b).
Bibliography 277
A argument
affective forecasting, 6, 8, 14, 16, 68 Fregean, 185
agency, 18, 62, 123, 123n45, 183, Moorean, 185
186, 187, 232, 267 open-question, 185
mental, 211, 256, 257 pincer, 185, 188
agent, 57, 66, 167, 168, 171, 179, why questions, 44, 45
183, 209, 257 Armstrong, David, 7, 77–84, 90, 95,
responsible, 57, 74, 81, 122, 125, 101, 116, 167
126, 232, 241, 242 assertion, 82, 92, 127, 128, 147,
alienation, 56 148, 151–2, 174, 192,
anti-individualism, 21–2 196, 201, 202, 202n46–7,
a posteriori, 80, 185 204, 205, 205n49, 205n51,
a priori,, 6, 10, 12, 13, 17, 19n, 51, 206, 206n51, 207, 207n55,
52n2, 56–8, 60–2, 66, 74, 209, 211n63, 214,
80, 83, 101, 109, 110, 244n4, 245–9, 249n14,
110n26, 112, 112n32, 116, 250, 250n17, 252, 254,
125–9, 163, 164, 167, 186, 254n21, 260, 261,
223, 227, 232, 241 264n40, 267
attention, 9, 39, 43, 59, 60, 71, 77, 217, 244, 257–62, 262n,
79, 87, 101, 103, 103n11, 264, 265n41–3
104–6, 106n16, 108n23, as dispositions, 5, 17, 19, 29, 31,
109, 119, 145, 155, 159, 37, 69, 149, 186, 198, 200,
238, 253, 254 208, 210, 234, 235, 255,
Audi, Robert, 42 255n28, 256, 256n31, 258,
Austen, Jane, 29, 53, 71, 238 263, 264
authority judgement-dependent, 194, 218
strong, 67 perceptual, 24, 25
weak, 125 Bilgrami, Akeel, 12, 13, 27n5,
avowals 31n11, 32n12, 32n14,
extension determining, 178 107n20, 165, 166, 183–8,
negative, 155 189n32, 198, 198n41, 199,
reportive, 155, 159 200, 201n44, 213, 214,
avowing 255n29, 257, 258
descriptive function of, 138–9 blind-sight, 61, 62, 232
expressive function of, 138 Boghossian, Paul, 67, 163
awareness BonJour Laurence, 106, 106n18,
conscious, 105 107, 107n19
phenomenal, 115 Boyle, Mathew, 3n2, 124n48,
127n54, 153n11, 172n10
Brady, Michael, 43, 44
B Brandom, Robert, 31
background assumptions, 43 Brewer, Bill, 108
Baldwin, Tom, 244, 254n22, Brüeckner, Anthony, 251n20
255n28, 259n36 Burge, Tyler, 6, 9–11, 20–5, 20n1,
Bar-on, Dorit, 6, 10, 134, 192n35 25n3, 26, 66n9, 100,
behavior 111–18, 128, 153, 156n24,
linguistic, 17, 66, 71, 91, 92, 152, 169, 179, 182, 201n45
158, 168, 228 Byrne, Alex, 9, 9n7
non linguistic, 66, 71, 158, 168,
210, 228
behaviourism, 78, 87, 95, 134, 135, C
180 Cartesian picture, 53
belief(s) Cassam, Quassim, 4n6, 8, 78, 84–8,
as commitments, 36, 37, 60, 125, 95, 124n48
183–4, 186, 189, 190, 195, cause(s)
195n, 202, 204, 208–10, of actions, 146
Index 281
dispositions externalism
first-order, 31, 37, 185, 263, 267 about content, 6, 67–9
second-order, 38, 262 epistemological, 116
distinct existences, 173 rational, 100, 111–18, 128
Dokic, Jérôme, 85n8
Dummett, Michael, 29n6
F
family resemblance, 47, 139
E Fernández, Jordi, 89, 100, 119–22,
eliminativism, 244n4 124, 124n47, 129
embedding, 247 Finkelstein, David, 127n, 152n,
emotions 153n11
basic, 6, 11, 14–16, 18, 39, 40, Frege, Gottlob, 185, 265n43
52, 59, 61, 69, 164, 218, Frith, Uta, 196n
222–32, 240 functionalism, 255n28
borderline view of, 19, 46–7
as evaluative judgments, 5, 43, 44
as felt bodily attitudes, 5, 40–1 G
as perceptions of evaluative Gallois, André, 89, 250n17, 254n21,
properties, 41–6 255n24
as sensations, 39 Geach, Peter, 205n50, 207n56,
entitlement(s), 11, 100, 110–13, 211n63
115–18, 126–8, 131, 141, Gendler, Tamar, 254n21
156, 158 Gertler, Brie, 53n6, 68, 69, 254n21
expressive, 11, 157 Gilbert, Daniel, 68n12
epistemic achievement, 9, 58, 62, 66, Goldman, Alvin, 8, 78, 88–96
133, 163, 200, 212, 225 Gopnik, Alison, 3n4, 7, 77, 84–8, 95
Euthyphro contrast, 178 Gordon, Robert, 8, 78, 88–96,
Evans, Gareth, 8, 58, 94, 100, 119, 192n, 218
119n, 250n17 grammatical, 15, 133, 142–5, 204
Evans’s point, 8152, 250, 250n17, fiction, 136
251, 251n20 Green, Mitchell, 243n, 244n2,
evidence 245n6
behavioural, 28, 123 Grice, Paul, 139n
mnestic, 14, 220–1, 240 groundlessness
observational, 73 strong, 113, 121, 225
expressivism, 11, 15, 133–60, 222 weak, 56, 79, 94, 113, 120, 124,
neo-, 152–9 157, 225
Index 283
I
imagination, 43, 74, 144, 222n, 240 J
individuation, 75, 82n5, 196, 229, Jacobsen, Rockney, 205n51,
235, 238 206n51–2, 207n57,
infallibility, 114, 145, 179 208n58, 210n62
inference James, William, 39
to the best explanation, 5, 13, 14, judgement, 5, 9, 10, 13, 14, 28,
55, 57, 70, 72, 73, 75, 77, 32n12, 34–6, 38n20,
79, 83, 95, 99, 120, 139, 40–3, 43n21, 44–6, 64,
141, 146, 154, 157, 176, 73, 83, 88–91, 101n3,
220, 225, 227, 104, 105n14, 107, 108,
236, 237, 240, 241, 108n23, 110, 111, 113,
256n30 124, 127, 128, 146, 153,
deductive, 257 155n21, 156, 159, 178,
inductive, 233, 236, 257 179, 190–2, 195–6, 201,
inferential theories, 84–8 202, 202n46, 203, 203n,
inner observation, 79 204, 205, 205n49, 206,
inner promptings, 8, 14, 88, 220, 207, 207n55, 209, 214,
235, 238 220, 224–6, 233, 237,
inner sense theories, 11, 77–84, 95 244, 245, 247, 249,
insincerity, 182, 214 250n17, 251, 254, 260,
instrumentalism, 180, 183 261, 267
intention, 1, 2, 5, 19, 28, 29n6, justifications, 43–5, 55, 110–12,
31–4, 56, 57, 60, 62, 68, 112n32, 121, 127, 129,
69, 72, 89, 103n11, 114, 141, 157, 202n46, 251.
118, 126, 143, 145, 146, See also reasons
284 Index
reasons, 6, 9, 28, 30, 34–7, 40, 41, 105, 109n24, 110, 112–17,
43–5, 57n8, 59, 61, 62, 72, 120–8, 133, 134, 138, 152,
95, 102, 103, 105, 108n23, 153, 155, 157, 160, 164–6,
111–13, 118, 119, 123–5, 168, 171, 181, 193, 197,
123n45–6, 125n49, 126, 199–205, 206n53, 207,
140, 141, 143, 145–6, 151, 209, 210, 213, 214,
153, 156n24, 160, 168, 219–21, 234, 237, 238,
176–8, 181, 192, 198–202, 240, 242
212, 217–19, 227, 228, self-blind person, 167
235, 238, 239, 258. See also self-deception, 41, 200, 255
justifications motivated, 199
reliabilism, 7, 116, 156 self-interpretation, 29, 30, 32n14,
representationalism, 86n10 65, 71, 182, 196n, 209,
Rey, Georges, 37 219, 234, 235, 237, 238
Ring, Merrill, 245n5, 245n7 self-knowledge
rule-following, 175, 176n17 constitutive account(s) of
rules, 107, 133, 134, 144, 160, 164, robust, 219, 222, 240, 241
196 weak, 241
Ryle, Gilbert, 53n4–5, 58 epistemically weak account(s) of,
8, 99, 100n1, 102, 105–7,
109, 109n25, 112, 114,
S 116, 117, 118n41, 123,
Scanlon, Thomas, 27n5, 31n10 125, 127n, 129
scanning mechanism, 8, 11, 79, 81, expressivist account of, 152
112 first personal, 52–66
Schopenhauer, Arthur, 30n inferential model of, 125, 217
Schwitzgebel, Eric, 64 neo-expressivist account of, 192n
seeing-as, 42, 72, 236 observational model of, 192, 193,
Self-ascriptions 223, 230
of propositional attitudes third personal, 2, 4, 8, 16, 29, 51,
as commitments, 31–8, 221 69–75, 95, 96, 149, 190,
as dispositions, 13, 28, 31, 219, 232–42
32n13, 149, 183, 187, sensations, 1, 3n2, 4–6, 9–11,
189n32, 208, 218, 219, 14–16, 18–26, 38, 39,
232–40 45–7, 54–61, 63, 64, 70,
psychological, 2–6, 11–14, 16, 74, 83, 108n23, 115, 116,
17, 18n11, 54, 55, 59, 123, 129, 134–7, 139–45,
65–8, 73, 89, 90, 99–102, 164, 167, 174n12, 192n,
Index 287
V
T Varzi, Achille, 109n24
Tanney, Julia, 30n Velleman, J. David, 126n51
Tappolet, Christine, 42 veridical, 21–3, 46
Teroni, Fabrice, 40, 44 Volpe, Giorgio, 30n, 121n
testimony, 14, 35, 73, 75, 233, 237, voluntarism, 33, 34
241 Vonk, Jennifer, 197n
288 Index