Self-Knowledge, Transparency', and The Forms of Activity
Self-Knowledge, Transparency', and The Forms of Activity
Self-Knowledge, Transparency', and The Forms of Activity
Terms of Use This article was downloaded from Harvard University’s DASH
repository, and is made available under the terms and conditions
applicable to Open Access Policy Articles, as set forth at http://
nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of-
use#OAP
1
DRAFT
Introduction
philosophy has given it a place central to the understanding of rationality and agency
themselves, and in doing so it treats the ability to know one's own mind as something
understandings of the nature of mentality and agency themselves. This paper begins
by setting out in very general terms some considerations that would link self-knowledge
to a certain form of agency, and then considers two recent studies which seek in their
different ways to show that a certain form of account of self-knowledge (involving appeal
2
to the "transparency" of belief) must be divorced from any appeal to rational agency. I
will be arguing that in both cases the account that emerges from this divorce ends up
with a kind of agency in the picture after all, only of the wrong kind, so that the ordinary
exercise of coming to know what one believes requires the person to exert a kind of
external control over her own attitudes. In seeking to develop the relevant notion of
rational agency in contrast to this, I cannot claim to be doing more than pointing to the
place where I think we need some such notion. In the course of doing so, I try to
characterize the sense of 'activity' or 'agency' that is relevant to a central class of cases
of self-knowledge, and distinguish this sense of activity from the sense of activity
In thinking about self-knowledge and rational agency there are two broad
directions from which we may begin to ask how they may be related to each other. We
may ask, first of all, how self-knowledge matters to agency itself, that is, whether the
specifically human forms of rational agency can be understood apart from the capacity
for self-knowledge of the mental life that is expressed in that agency. Is our capacity to
act for reasons, to be self-guided in that sense, dependent on our ability to know our
mental life 'immediately'? Can the ordinary ability to respond to reasons in one's
thinking, to consider reasons for and against some belief and respond accordingly, be
understood apart from our capacity for immediate self-knowledge? And would the
absence of the ordinary capacity for self-knowledge make no essential difference to our
rational agency?1 And from the other direction we can ask how rational agency itself
1
Recent philosophical work has drawn connections between self-knowledge and agency in a variety of
different ways. Here I will just mention Burge (1998), Bilgrami (2006), and O'Brien (2007). Particularly
helpful to me in thinking about the issues of this paper has been Matthew Boyle's paper 'On 'Making up
your Mind and the Activity of Reason' (forthcoming).
3
may matter to the understanding of self-knowledge; that is, whether the ordinary
capacity to know what one thinks about something is part of the same capacity to
determine one's thought about that thing. Is our ability to know what we believe
'immediately', and with a kind of authority not shared by what we say about the beliefs
of others, tied to the fact that our beliefs and other attitudes are expressions of our
her attitudes that is related to the capacity for first-personal knowledge of them?
One recent way of relating both sets of questions begins with the example of
belief, and appeals to a notion of 'transparency' between a question about one's belief
and a corresponding question about the object of one's belief. Thus, it has seemed to
can answer a question about her own belief by addressing herself to the corresponding
question about the topic of that very belief.2 Hence, if asked do I, RM, think it will rain
today, I can answer this question by giving my answer to the corresponding question
about the rain, and not by inquiring in to the state of mind of a particular person. The
fact that the answer is given by a particular person, the very person whose state of mind
the first question directed itself to, must surely be part of the answer to how it is possible
or legitimate to answer the question about one's belief by reference to the question
about rain. For 'transparency' of this sort surely doesn't apply to a question I may ask
about the beliefs of another person. If the topic of my question is the beliefs of some
other person, then my efforts to answer that question must address themselves to the
facts concerning the state of mind of that particular person. That's how it is, after all,
2
See Evans (1982), Edgley (1969), and Moran (2001). More recently Byrne (2005) and Shah and
Velleman (2005) appeal to a notion of 'transparency' for belief, but offer very different accounts of it.
4
with my efforts to answer questions about other topics. In seeking to answer a question
about the inflation rate in China, I must direct my attention to China. Why, then, if the
topic is some fact (attitudes of a certain kind) about oneself is it legitimate to answer the
question in a way that seems to neglect the fact that it is about a particular person, and
instead treat it as a question concerning the topic of the attitude itself (e.g., the
weather)? Again, the fact that in such a case the answer is delivered by the very
person whose state of mind is in question must be central to accounting for this, for if
relations to oneself and relations to others. The identity of the person whose state of
mind is inquired into and the person answering the question must matter here.
vindicate the applicability of transparency to the first-person case, and with it the claim
that in delivering an answer in this way, the person is indeed speaking from knowledge.3
The form of account I give appeals to a form of agency that is part of a person's being a
creature with beliefs, and I claim that the transparency which various philosophers have
found attractive cannot be accounted for without appeal to this agency. It would not, in
general, make sense to answer a question about my state of mind (e.g., my belief about
the weather) by attending to a logically independent matter (the weather itself) unless it
were legitimate for me to see myself as playing a role in the determination of what I
believe generally. However, while this form of account takes the topics of self-
knowledge and agency to be closely related, the agency in question does not involve
any kind of voluntarism about belief, and indeed the form of rational agency I have in
3
Moran (2001).
5
mind has as a consequence that such voluntarism is false. The sense in which I see
belief and other attitudes as forms of activity is deeply related to the fact that they are
not matters of choice for the person, and hence the agency involved here is not that
which is exercised when, say, a person chooses to raise her arm and then does so.
though it will be important to recognize that there will be differences between, for
instance, believing, wanting, fearing, intending, hoping, intending, and caring about.
an activity's being worth caring about. In this they differ from brute sensations, which I
take to be aspects of our passive or receptive nature. Part of what is meant by this is
that different forms of the question 'Why?' will apply to items in these two categories,
and the relevance of asking the question of the person herself will also be different.
There may be no special reason to ask me why I'm experiencing pain in my lower back,
although I might know something about it. And the answer to that question 'why?' will
refer to such things as a previous injury, bad work habits, or compression on a disk.
That is, the answers will contribute to the explanation of the coming to be of the
sensation I'm experiencing, but do not seek to say anything about the apparent point, or
the good, or the intelligibility of the state I am in. Naturally, such causal versions of the
6
question 'why?' will also apply to items in the other category as well (e.g., one's beliefs,
hopes, desires and intentions), but it is internal to them that a different form of the
question 'why?' also applies to them. Hence we can ask why someone believes it will
rain when we are asking for that person's reasons for the belief, and we can ask why
someone wants a saucer of mud when we want to know what could seem good or worth
having in such a thing. This form of the question 'why?' thus seeks a certain normative
However, just as with the question seeking the origins of some condition, the answer to
the question 'why?' is meant to tell us something relevant not just about the character
of the attitude itself, but something relevant to the fact of the person's having that
attitude. That is, the way in which the attitude is found reasonable or intelligible by the
person is assumed to be relevant to the question why she has that attitude, how it came
to be part of her mental life, or what maintains it there. So this is something that this
question shares with the kind of question 'why?' that applies to the person's sensations,
in that in both cases we are inquiring into how something comes to be or maintains
itself. The difference is that in the case of the question 'why?' as applied to the person's
beliefs and intentions, the question of how it comes to be is tied to this normative
question involving notions like the reasonable and the worthwhile. The second broad
type of 'why?' question assumes a kind of dependence between these two sets of
what could be seen to be reasonable, intelligible or worthwhile in it. And finally, there
is a further difference in how these two types of question 'why?' are treated, and that is
that for the broadly normative type of 'why?' question as applied to someone's attitudes,
7
we typically do ask that question of the person herself. Unlike the question 'why?'
some attitude she holds. 'Uniquely relevant' or 'indispensible' here does not mean
infallible or incorrigible, but rather that if she cannot tell us why, then we may begin to
doubt whether there is a good answer to that why-question. This is not an assumption
we make about the question 'why?' concerning sensations or other bodily conditions,
When philosophers claim that the notions of reason and justification are internal
to the notion of belief, this is not meant to deny that children and animals can have
beliefs, even though they do not themselves have these notions of reason and
justification. To say this much is so far just to say that certain norms of rational
assessment apply to beliefs. This is not yet different from the idea that, e.g., it is
internal to something's being a heart that it pumps blood and can be assessed as
healthy or malfunctioning, even when the creature with the heart has no conception of
such things. Controversy begins with the thought that, in mature humans, it belongs to
the notion of belief that reason and justification not only apply to it, as a form of
normative assessment, but that the believer play some role with respect to justification,
a role that a creature does not typically play in the good functioning of its heart. The
believer can be asked for her reasons for a certain belief, and the believer typically
recognizes the applicability of that question, even when unable to give any convincing
belief. In these ways we hold the believer responsible for her beliefs in ways that we
do not typically hold the person responsible for the condition of her heart.
We do not ask very young children and non-human animals for their reasons for
believing something, and yet it does seem to many philosophers to be central to the
very notion of belief in mature humans that believing something opens one to the norms
of justification, and the responsibility for conforming to those norms. We ask the
believer herself for her reasons, and the believer recognizes that she is indeed the
person to ask, that the request for reasons is properly addressed to her. We do not
address similar questions to the person about the condition of her heart. We do not
have the same expectation that she will be the person in a special position to know
about its condition and address the question of its good functioning. If this much is
true, it raises several questions. One, what are we presupposing about the believer's
knowledge of her belief and its justification when we address such questions to her
specifically? Two, what are we presupposing about her agency with respect to her
belief when we hold her responsible in these ways, different from any responsibility she
may have for the condition of her heart? And three, how does this system of reason-
asking and reason-giving among mature believers relate to the capacities for belief
among young children and non-human animals? That is, if we grant that such
creatures do have beliefs, and yet do not have the concepts and capacities that would
mature believers, then what is left to the claim that it is somehow internal to belief
among mature believers that the system of reason-asking and reason-giving belongs to
it, and that the role of justification is recognized as such by the believers themselves?
9
In favor of retaining some version of the idea that the asking and giving of
reasons belongs to the nature of belief and other attitudes themselves, I'll just say the
following. While some philosophers have gone so far as to doubt that the concept of
belief can apply to animals and to children before they are language-users, no one
would want to deny that they are both capable of action in a perfectly ordinary sense.
But at the same time, it seems we also don't want to say that having reasons for what
one does is something only added on at a later stage and does not belong to the idea of
action itself. Actions are purposive and goal-oriented. Both the child and the adult
may have reasons for reaching across the table toward a glass of milk. The ability of
the person to tell us what she is doing and why she is doing it is something that
develops later, as part of the growth of various capacities and the initiation into various
forms of responsibility. But the fact that rational assessment applies to the action, and
that later these forms of assessment can be posed as questions that we direct to the
person herself, asking her just what she is up to and why, is all a development of the
same idea of action. These first-person capacities and responsibilities are no more
extraneous to the idea of action than is the related capacity of a speaker to tell us what
she means by something she said. At an early stage of the ability to talk, the child will
not be expected to tell us what she means by her words. That is also something which
develops later. But it is surely internal to the development of the child's very capacity
position to tell us what she means, what she's talking about, and will not count as a
speaker absent any such capacity. Here again, we need not suppose that a speaker
has unbounded authority over the question of what she is saying, or that allowance
10
cannot be made for one form of semantic externalism or another, but the fact that we
can and do ask the speaker to clarify what she is saying is surely part of the very notion
of saying something. In all these cases then, it will only in very special circumstances
makes sense to say "Why ask me?" in response to a question about what one is doing,
If the second, normative version of the 'why?' question is also, like the first one,
meant to shed light on how something comes to be, then it may be asked why it is that
we take the person herself to be particularly relevant as the person to ask here. Why
information on this topic? And this question suggests the possibility that the reason we
ask the person herself is closely related to the fact that the normative 'why?' question
contains within it both a question about a form of normative assessment and a question
about how something comes to be or maintains itself (the belief, the action). The
condition is what it is quite independently of the person's cognitive relation to it, and it is
not the person herself who is the locus of that assessment . By contrast, a person who
believes it will rain, or who hopes it will not, is not just in a condition that can be
assessment, and the believing or the hoping themselves are what they are in virtue of
the person's overall sense of what would support or undermine them as attitudes.
When the belief or the action is judged or found wanting, this is an estimation of the
person, and not simply of some condition she is in. If attitudes such as these can
themselves be seen as forms of normative engagement on the part of the person, then
11
the question 'why?' is applied to them in a particular form. Part of what's meant in
calling the believing itself a form of normative assessment is simply that to believe P is
justifiable. And although we do not choose our beliefs and do not perform them like
oneself to the question "why believe that?", and to do something is to take the action to
be worth doing in some way, and thus to open oneself to the question "why are you
doing that?". And in both cases the person takes the answer to the normative 'why?'
question. With this much in place, we can see it as making a certain sense to see the
person herself as the one to ask when we want to know why she believes this or hopes
for that, and why she will indeed recognize herself as the right person to ask, and might
indeed insist that treating her relation to this question as something in principle
We have special reason to ask the person herself both what she is doing and
why, or what she believes about something and why, because it is possible to see both
or as ways of resolving oneself. A belief is the answer to the question about what the
evidence points to, or what best explains some happening; and an action expresses
one's resolution with respect to the question what is to be done in a certain situation, or
4
This formulation has been stressed by Pamela Hieronymi in recent work, although I suspect my
understanding of the connection differs from hers.
12
invite the normative question 'why?', and assume the responsibility of the person to be
able to speak to that question. By contrast, another internal condition of mine, like the
(practical or theoretical). I may be responsible in one way or another for either the
the heart or the sensation itself, whereas on this view actions and attitudes, are modes
person's ability to say what she is doing or what she thinks about something, and to
know this without having to make the kind of observations of herself that she would if
the question were about someone else's belief or action, is related to the fact that she
bears a certain responsibility for her belief and action, fundamentally different from the
responsibility a person may have for the sensations she finds herself with. In the case
of action, we ask the person herself what she is doing because we ordinarily take her to
know, and we take her to know what she is doing because we take her to know why she
is doing it, and we take her to know what she is doing and why because we expect this
of her, it is her business to know. This brings a certain notion of agency into the picture
knowledge with respect to actions and attitudes is tied to their being expressions of the
rational, active side of one's nature. Hence the relevant notion of 'activity' belongs to
(sensation or bodily condition), rather than to one's relation to a particular item in that
category. Both a sensation of pain or a belief about my chances of winning the next
various degrees of success. This notion of 'control' applies just as well to my relation to
the perfectly inert objects in my immediate environment and is not relevant to the notion
of agency being appealed to with respect to my doing and believing. Rather, if there is
anything to this difference in category, then believing, intending, and hoping are
beliefs are not chosen by her, nor are they typically "controlled" by her. Rather, what we
call a person's beliefs are the precipitate of her ongoing rational activity. It is only
derivatively that a person is 'active' with respect to a particular attitude itself. In the
normal case, I find myself wanting to learn Russian, or suspecting that my relatives
won't be visiting for the summer after all. Such attitudes do not emerge out of nowhere,
of course, but rather become mine in the course of my ongoing thinking and acting, and
are not aimed at as states to put myself in. I do not aim at acquiring some particular
attitude, and it's rationality is not expressed by my singling it out for control or
manipulation. But for all that, my wanting, suspecting and caring about something are
expressions of my active nature, to which some form of the normative 'why?' question
naturally applies, along with my taking myself to be the person who is answerable for
5
For a related argument, see Christine Korsgaard's Presidential Address 'The Activity of Reason', in
particular p. 37.
6
For more on the distinction between exerting external control over one's attitudes and assuming
rational responsibility for them, see Moran (2002).
14
Various recent writers have taken the 'transparency' of belief to be part of the
explanation for the person's ordinary ability to say what she believes about something
without having to base what she says on empirical observation of herself, but have
found the appeal to rational agency to be misplaced. For some, this appeal has been
thought to falter on an ambiguity between how I know what I already believe about
something and how I know what I believe about it now, upon considering the question.
Thus in a recent paper, Nishi Shah and David Velleman say the following:
"The question "Do I believe that P?" can mean either "Do I
already believe that P (i.e., antecedently to considering this
question)?" or "Do I now believe that P (i.e., now that I am
answering the question)?" [...] Now, either of these questions can
give way to the question whether P. If the question is whether I
already believe that P, one can assay the relevant state of mind by
posing the question whether P and seeing what one is
spontaneously inclined to answer. In this procedure, the question
whether P serves as a stimulus applied to oneself for the empirical
purpose of eliciting a response. One comes to know what one
already thinks by seeing what one says -- that is, what one says in
response to the question whether P. But the procedure requires
one to refrain from any reasoning as to whether P, since that
reasoning might alter the state of mind that one is trying to assay.
Hence asking oneself whether P must be a brute stimulus in this
case rather than an invitation to reasoning." 7
This is said in the context of seeking to vindicate a notion of the transparency of belief,
but it is not immediately clear just what sort of problem they see here, or how the kind of
7
'Doxastic Deliberation', Philosophical Review, vol. 114, no. 4, p. 16. A related objection is made in the
paper by Alex Byrne discussed in the next section, as well as in Gertler (forthcoming).
15
insulation from rational agency they have in mind could still deliver an answer about
what I already believe about something. Does engaging my rational capacities corrupt
approach, we could adapt an example from Sydney Shoemaker, and suppose that I am
asked who I believe was the President of the Confederacy during the American Civil
War.8 It may happen that in responding to the question I start to say "Robert E. Lee"
and then correct myself and say "Jefferson Davis". In this way, I failed to treat the
capacities, but none of this provides a reason for thinking that I have thereby produced
a new belief rather than reported what I believed all along. Surely it may be the case in
an example like this that I did in fact believe all along that it was Jefferson Davis, but
blurted out the wrong answer and corrected myself, all the while being faithful to what I
need not be seen as substituting a new belief for what I already believed.
On the other hand, there are difficulties in seeing how my response to such a
here, and still be seen as reporting on a belief of any kind. To begin with, let's recall
that with respect to any belief of mine, it counts as a belief insofar as I take it to be true
(this, of course, is what makes some sort of appeal to 'transparency' seem attractive). If
I relate to my "stored" belief as something I take to be true, it will be hard to see how I
can see my relation to it, however spontaneous, as insulated from the engagement of
my rational capacities for determining what is true or false. It cannot, for instance, be
8
'Moran on Self-Knowledge', European Journal of Philosophy 3, (3):391-401
16
seen by me (or my auditors, if they are the ones applying the stimulus) as simply some
name that is produced upon receiving the stimulus, for it has to represent what I take to
background involving my rational agency that has to be assumed for my response to the
stimulus to count as my spontaneous answer to the question. Consider the fact that
there are countless words and names that may be floating around in one's mind, and
any one of them might be what comes out when the stimulus is applied. None of them
will count as indicating one's belief about the matter unless, minimally, one can
recognize the word as a name, and not something else, and recognize it as the name of
a person, and recognized the name of this person to be relevant somehow to the
stimulus such that it can serve as indicating what one's belief is about this matter.
Simply hearing oneself coming out with something in response to a brute stimulus will
provide no more reason for thinking this represents one's belief about something than if
one were to sneeze in response to the stimulus. Rather, for my response to the
stimulus to be seen as telling us what I already believe about the question, I have to
relate myself in various ways to the name I come out with, and not just hear myself say
it. I must, at a minimum, understand the words I am saying, and understand them as
mean something like 'replying to the question', for in a broader, more neutral sense of
'responding' we would need an additional reason to think that this 'response' bears any
relation to my beliefs about anything (which is what the stimulus is supposed to deliver).
For any mental content, word association, or exclamation that may be produced by the
stimulus, it will count as relevant to the question of my belief only if I am relating to that
17
content or word as representing my belief about the matter. That however, will mean
engaging with my rational capacities in a way that involves my reflecting on the facts of
the matter, which was supposed to be excluded from the process because it was
thought to "contaminate" it.9 But if that were so, then it is hard to see how a person
could ever perform the ordinary task of telling us what they already believe about
something.
What's meant by the "activity of reason" here needn't be explicit, nor need it
involve the production of a new belief. I may hear the question "Where was Balzac
married?' and come out spontaneously with the name "Berditchev", and when asked
why, perhaps I can't say much more than "I must have read this somewhere." 10 But to
know that much about my response is already to know a great deal, is already for me to
person, and in connection with an event which left some written record I could have
encountered. The confidence with which I spontaneously come out with the answer is
surely dependent on such things as that I understand the statement proposed to me,
that the proposition in question makes sense to me, and that the possibility it presents
seems perfectly plausible to me, even if I can presently see no special reason to think it
true. (Compare: the answer I come out with is "On the moon".) All of this and more is
isolating my response to the question from all of this, and still have me responding to a
question. It is only when we take for granted that this background is in place that we
9
"As we pointed out, one cannot engage in reasoning aimed at answering the question whether P
if one wants to find out what one already believes, because such reasoning would contaminate the result
by possibly altering the state that one is trying to assay." Shah and Velleman (2005), p. 17
10
See Daniel Dennett, 'How to Change Your Mind', in Brainstorms (MIT, 1981)
18
then confront the scenario of my replying to the question, with perhaps nothing more to
go on than a feeling of familiarity, or the sense that I must have heard this before, but by
then the work of reason has already prepared the place for my answer. We can, of
course, isolate my response from reason if I do not understand the language in which
the question is posed, or if the response I come out with makes no sense to me,
something I just find myself saying, or if I come out with it as a sheer guess, but in none
of those cases will we think that this response expresses my belief about anything.
And with respect to other things I already believe, which are more integrated in the rest
of my life and thought, things are more complex, and it is even harder to imagine what
could be meant by isolating my response from the influence of reason. That is, if the
question about what I already believe is something like: 'Do you believe there are
people living in Phoenix?' or 'Do you believe that you can buy food with money?', then
my spontaneous answer will be 'yes', not because I have reasoned my way to this
conclusion, but because so much else of what I believe would have to be up-ended if
this were not true, and I would have no idea how to begin such a revision of my beliefs.
I don't have to think about it, but not because my answer is insulated from the influence
of reason, but precisely because my answer is so fully integrated with the rest of my
situation of wanting to know what someone thinks about something, but prior to having
that person consider the question itself. If I'm playing poker with someone, I may want
11
See Wittgenstein, Philosophical Investigations § 478: "What kind of reason have I to assume that my
finger will feel a resistance when it touches the table? What kind of reason to believe that it will hurt if this
pencil pierces my hand? -- When I ask this, a hundred reasons present themselves, each drowning the
voice of the others. "But I have experienced it myself innumerable times, and as often heard of similar
experiences; if it were not so, it would .....; etc."
19
to know whether he thinks I'm bluffing, but without actually raising that question with
him, for that would risk alerting him to a possibility that I don't want him to consider if he
isn't doing so already. In this sort of case, I might do various indirect things to elicit
information on this point, without raising the question itself. This might be thought of as
a "brute stimulus" in the relevant sense, but it will importantly not involve my poker
partner responding to a question as a "brute stimulus". Now, in my own case, it's much
harder to imagine what would count as my "wanting to know what I think about some
possibility, but without my considering the possibility itself". In the two-person case, I
can raise the question of what this other person believes about P, without that person
considering the question of the truth of P itself. In my own reflections about him, I can
speculate about what may or may not be going on there in this other mind, knowing that
my reflections are strictly mine, and not part of what constitutes the state of mind I am
speculating about. But with only one person on the scene, this is not a real possibility.
I cannot pose the question to myself of whether I believe that P without raising the
question of the truth of P, for there is only one mind under consideration here, inquiring
about itself. Naturally, this does not mean that I cannot, in certain circumstances, seek
to "assay" what I really believe about something, in a way that brackets the question of
the truth of what I seem to believe. In various circumstances, a person can indeed take
such an 'outsider's' perspective on her own belief, even though the result may be an
inherently unstable one ("Well, I know that the plane is safe, but clearly I'm also in the
grip of a fear that it is not safe." ) But even this possibility is quite different from the
idea that I could raise the question of whether I really think the plane is safe without
considering the question of whether the plane is safe. I have to understand what
20
question I am asking myself (the way my poker partner does not have to understand my
understand the relation between the stimulus I am applying and the question I am
seeking to answer. And I cannot understand either of these things without considering
the content of the state of mind I am inquiring into, which orients me with respect to the
question of its truth, reasonability, or comprehensibility. Inquiring into one's own mind
about what one already believes about something cannot be insulated from one's
rational agency in the way that it can be with respect to the mind of another person, but
nor does this involvement mean that what I already believe is elusively out of my reach,
engaged. Unlike my relation to my partner in poker, in the first-person case, inquiry into
the mind has to be intelligible to the very mind being inquired into, and hence involves
the engagement of reason on the part of both the inquirer and the mind inquired into, for
explanation of what makes appeal to Transparency possible. The paper presents a re-
thinking of several issues concerning self-knowledge, and its general aim is to vindicate
what Shoemaker criticizes under the name of the Broad Perceptual Model of self-
Byrne has his sympathies here12), and comes down to the two claims that (1) our
detection of our mental states is based on some sort of causal mechanism, and (2) that
our mental states themselves obtain independently of our access to them (Byrne, p. 86).
Shoemaker's case against the "broad perceptual model" depends on challenging (2)
rather than (1), since Shoemaker defends the idea of a "constitutive relation between
believing something and believing that one believes it" (2009, p. 35). The constitutive
relation between one's mental life and one's first-person access to it is defended by
rationality.13 In this sense, then, although we do not infer or reason our way to
belief in the first place and having one's beliefs available to one, without the need for the
kind of evidence one would need for knowledge of the beliefs of others.
understood as the result of our following a rule for belief-formation, which he calls BEL.
The phenomena to be accounted for include both privileged access, described as the
idea that our "beliefs about our mental states acquired through the usual route are more
likely to amount to knowledge than beliefs about others' mental states" (p. 80), and what
he calls "peculiar access"; that is, the idea that one has a special method for learning of
12
To be sure, much of the paper is devoted to showing how the case against "inner sense" turns out to
be unconvincing, and hence counts as a defense of that view. But at the same time "inner sense" is
described as an "extravagant" rather than "economical" account of self-knowledge, in that it requires
appeal to an additional capacity or mechanism, beyond what is already needed for our general capacity
for rationality (p. 92). The account Byrne later goes on to defend, in terms of the BEL rule, is
recommended for being economical rather than extravagant (p. 99), hence I take it that Byrne should be
understood as defending a version of the "broad perceptual model", but not "inner sense". ("The account
is not a version of the inner-sense theory.", p. 99)
13
"[B]elieving that one believes that P can be just believing that P plus having a certain level of
rationality, intelligence, and so on" (Shoemaker, 1994, p. 244, quoted by Byrne on p. 89).
22
one's own beliefs that cannot be applied to the beliefs of others (p. 81). The BEL rule is
it is legitimate for a believer to answer a question about her belief about something, by
reflection not on herself, but on the topic of the belief in question. Hence he presents
This rule for belief is introduced by comparison with another rule, called DOORBELL,
which states
door.
DOORBELL is what Byrne calls a "good rule" insofar as it "tends to produce knowledge
about one's visitors" (p. 94), and this goodness will depend, naturally, on various
empirical conditions obtaining in the environment of the person following this rule. It
can lead one astray, and hence is not failsafe, but may be reliable enough in practice to
count as a rule the following of which results in knowledge. BEL, however, has
resulting second-order belief will be true (p. 96). This has the nice feature of also
capturing "peculiar access", since BEL will only be self-verifying when applied to one's
own beliefs. The variant of BEL "If P, believe that Fred believes that P", is not a good
rule. Finally, BEL is an especially "safe" rule in that the resulting second-order belief will
be true, even if one tries but does not succeed in following it. As the rule is formulated,
actually applying it requires the obtaining of some fact 'P', hence in the absence of that
23
fact the BEL rule is not in fact being followed. However, BEL has the special virtue that
even trying to follow it will result in a true second-order belief, since if I mistakenly take
'P' to obtain, and then seek to follow BEL on the basis of this mistake, I still end up with
BEL is described as a rule which we follow, and the following of this rule is
described as form of reasoning (p. 94). And indeed, the following of this rule is
described by Byrne as no more problematic than the following of less exotic epistemic
rules, such as DOORBELL.14 At the same time, Byrne does acknowledge BEL as an
unusual epistemic rule since, unlike DOORBELL, it recommends a transition from the
first fact being evidence for what is believed in the resulting belief. One way to display
the difference between the two rules is by noting that, despite the imperative form in
And indeed, upon hearing the recommendation of the original DOORBELL rule, it would
be natural to assume that the only reason for following it was that there was just such a
relation between the two contents, such that the fact of the ringing supported or was
evidence for the fact of someone's being at the door. Absent such a connection
between the contents in the two parts of the rule, it would be difficult to embark upon
complying with it, unless one could somehow install beliefs in oneself, exerting a kind of
14
"Given that we follow rules like DOORBELL, it should not be in dispute that we can follow BEL." p. 96
24
pragmatic agency with respect to one's beliefs. As mentioned, Byrne is well aware of
this difference between BEL and DOORBELL, since in a sense this difference simply
question about a particular person's beliefs, by appeal not to facts about that person,
but to facts relating to the content of the belief? The parallel with the revised
As Byrne points out, however, "this is a bad rule: that P is the case does not even make
it likely that one believes that it is the case". (p. 95). How then are we to understand
following the original BEL rule as a form of reasoning when it does not present a rational
relation of support between the two contents? If BEL is a rule we can be said to follow,
and the result of following it is the formation of a particular belief, then there must be
some answer to the question of the person's entitlement to make the transition in
thought that is being recommended by the rule.15 Whatever "following the rule" comes
demands of belief (as it would be for instance with the recommendation, "Believe that P,
because then you'll stop worry so much that not-P"), for the account is meant to be an
account of self-knowledge. The fact that BEL is "safe" in the sense of being unlikely to
mechanism of belief-formation, but the person seeking to follow the rule, the way she
15
I speak of "transitions" of thought here so as not to pre-judge the question whether we should
understand BEL as an inference, as a rational connection of another sort, or as a kind of "blind" rule
which nonetheless produces "good" results ("safe" beliefs). Shoemaker (2009) questions whether
following the BEL rule can be understood as a form of reasoning (p. 36).
25
follows DOORBELL, will still be in need of some reason relating the two contents.
Otherwise the rule is reduced to saying, as it were, "Act upon yourself, do whatever it
having reasoned one's way to it, and without now being able to provide justification for
it, so long as there is in fact a reliable connection between one's belief and the fact in
question. A reliable mechanism for belief-production need not represent itself in terms
of a rational connection between belief-contents, but may instead operate without any
understanding of it on the part of the person herself. BEL, on the other hand,
recommendation. As such, BEL is not just the description of a good disposition for
someone to have, but rather requires the believer to do something in response to it, and
in accordance with it. This much is needed by the parallel with DOORBELL, and the
understanding of BEL as itself a form of reasoning. Following a rule for belief, however,
but requires from the rule-follower some understanding of, and an endorsement of, the
rational connection between the contents mentioned in the rule. DOORBELL does
provide this, given the relation between the contents as represented in DOORBELL*,
original BEL rule, an imperative or recommendation addressed to some "you", and the
revised rule BEL* relating two contents, especially when the corresponding two versions
of DOORBELL don't display this difference in goodness, and indeed seem to support
26
each other? And can the imperative form retain any force when divorced from the
problem of "two topics" (e.g., the weather, my belief), we might look at the issue in terms
(2) Considering the candidate "Jones believes P", I likewise require evidence or
other truth-centered reasons concerning Jones, since he is the subject of the content in
question.
(3) However, considering "I believe that P" as a candidate for my own belief; that
is, the question of attributing the belief P to myself, I don't appeal to evidence for the
content "I believe that P", where that is taken to refer to the beliefs of a particular
person, as in (2). Rather, I appeal to the sorts reasons mentioned in (1), reasons in
favor that content, the one that is embedded in (3), and which doesn't mention any
person.
it raises the question of what could make legitimate the appeal to reasons relating to the
embedded content (P), rather than to reasons relating to the ostensible content of the
attribution itself ('I believe that P'). On this way of looking at it, the beliefs governed by
Transparency are not altogether independent of the appeal to evidence or other truth-
27
centered reasons, but the question is what makes legitimate the exclusive appeal to
reasons relating to the embedded content (1) rather than reasons concerning the
apparent content of a psychological attribution (2). What breaks the apparent parallel
between (2) and (3), such that (3) is answered as though it were a version of (1)?
As with the original Transparency condition, the appeal to the BEL rule also
relates the question of the content P (1) to the question of the self-attribution (3), and
hence incurs a similar burden to explain how it can be legitimate to make the transition
from 'P' to 'I believe that P'. The explanation adverts to the self-verifying character of
following BEL, but that will only be a good explanation if we have a better understanding
of what "following" it comes to, and the conditions under which doing so is possible. In
particular it bears explaining how the imperative form of BEL could make sense to
corresponding version of the rule (BEL*) which describes relations among the contents
of beliefs ("If P, then X believes that P"). If we distance ourselves from the imperative
form of the rule for a moment, we can describe the believer who is guided by the BEL
rule as someone who makes a transition from the apprehension of some fact P ("It's
raining out") to a belief about herself. This "belief about herself" is the product of
following the BEL rule, which the imperative form presents as addressed to "you"
("Believe that you believe that P"). In taking it out of the imperative form and
representing the transition in thought that the rule recommends, we need some
substitute for "you", and it will be crucial that the substitution retains the status of BEL as
a "good rule". If, for instance, the description of the transition takes us from "P" to
"Jones believes that P", this will be a bad rule, even if the person making this transition
28
is indeed Jones. For he may not recognize himself as "Jones", in which case the BEL
rule would be taking him from the apprehension of some fact "P" to a belief about the
beliefs of some person named Jones. The problem of "two topics" would re-emerge for
legitimate in the transition from a thought about, e.g,, the weather, to a thought about
some person's beliefs. Hence the transition described in the BEL rule is subject to
certain conditions, familiar from the philosophical discussion of the first-person. The
transition described in BEL will only be legitimate when the person following it is
identical to the person whose beliefs are mentioned in the resulting second-order belief.
And this identity must be recognized by the person making this transition, and hence the
name we substitute for "you" when we describe this transition outside the imperative
form of the rule, must be one under which the person in question could not fail to
oneself with the person whose beliefs are the topic of the resultant second-order belief,
the application of the BEL rule would be as illegitimate as the transition from some
The 'I', then, is what lies behind the 'you' who is the implicit addressee of BEL in
its imperative form. And in particular the "subject use" of 'I' is crucial to understanding
how it is that, while the attribution of belief to a person under some name, description or
demonstrative must appeal to evidence concerning that person (as in (2)), the first-
person discourse of belief appeals to reasons concerning the content of the belief rather
than evidence or identifying information about the believer.16 When the BEL rule
16
For the initiating discussions of the idea of this use of 'I' proceeding independently of "identifying
information" about the person, see Wittgenstein (1958), Shoemaker (1968), and Evans (1982), Chapter 7.
29
instructs someone to "believe that you believe that P", this "you" must not be in need of
identifying information anymore that the subject-use of "I" is, if the following of the rule is
to be safe or self-verifying. If identifying information were required to pick out the right
person named in the instruction "believe that you believe that P", or if evidence about
that person were needed to determine what that person's beliefs were, then forming the
belief about the beliefs of this person "you" would be in need of the same kind of
support, and subject to the same semantic and epistemic risks, as would the
corresponding belief about the beliefs of Jones. Under these conditions BEL would be
"bad" in the same way noted by Byrne in connection with the "neutral schematic"
version of BEL: the obtaining of some fact P would be treated as a reason for
So a fuller picture of the conditions under which BEL would be a good rule are
that (a) it is only knowledge of one's own beliefs, and not those of another person, that
following BEL will make possible, and (b) that in this knowledge I recognize myself as
the person whose beliefs are the topic of this knowledge, and (c) that my recognition of
myself as the person whose belief is known in this way not be based on identification
information of any sort. One situation in which a person may indeed need to avail
herself of identification information regarding some attitude is when she takes a belief or
a fear of hers to be evident in her behavior but otherwise inaccessible to her because it
is not responsive to her sense of the reasons that would support the belief or fear. If I
have reason to believe that what sustains my actual attitude regarding someone's
actual competence or trustworthiness, then I will not take myself to be entitled to make
30
the transition described in the Transparency condition or the BEL rule. If I think that my
actual belief about him (or people "like him") is controlled by prejudices whose
the reasons for or against P, then I will not take myself to be entitled to answer the
question 'Do you believe that P?' by reflection on the question 'P?'. Likewise, I will not
be entitled to follow the BEL rule which tells me that if "He is just as competent as the
others", I may help myself to the belief that I believe he is just as competent. The
difference between the topics of the two questions will re-emerge, I will recognize that I
may need to give different answers to the two questions. I may know about this
tendency of mine, and know that the beliefs that are its product are resistant to change,
and may not be reflected in my overt, sincere declarations when considering the content
itself.17
Thus, the appeal to either Transparency or the BEL rule has various conditions,
which matter to the understanding of the kind of rationality that is assumed in making
the transition from 'P' to 'I believe that P' . For a person to be entitled to make this
transition, he must recognize the identity between himself and the person whose belief
is the topic of 'I believe that P', and he must take himself to be able to speak for the
beliefs of that person, in the sense that he presents himself as the very person whose
sense of the reasons in favor of 'P' are expressed in the statement 'I believe that P'.
Transparency fails if the statement I arrive at, 'I believe that P', is delivered in a purely
Attributive mode, for that would require evidence about me, which Transparency does
17
See Moran (2003), pp. 407 - 408 for more on different ways of imagining such failures of transparency.
18
On "attributive" see Moran (2001), Chapter 3.3.
31
reflection on a particular content 'P' and the reasons in favor of it, and the answer to the
question of what my actual belief is. That connection presumes that my belief about
some matter 'P' is determined by my sense of the reasons in favor of 'P', and not by
forces independent of those reasons. This is not a matter of constructing a new belief,
This requirement does not mean that the declaration 'I believe that P' cannot be
the expression of what one already believes,19 but that the statement of belief must be
delivered in the mode of endorsing the content of the belief, and not as the attribution of
belief (to oneself, to another person) in a way that leaves open the question of
endorsement. The idea of endorsement points to a form of rational agency that both
Byrne, and Shah and Velleman, find misplaced in the account of Transparency, but in
distancing themselves from it, they each in their different ways picture the self as
exercising the "wrong kind of agency" with regard to its own beliefs, either by seeking to
elicit one's own beliefs through a kind of stimulus insulated from the exercise of reason,
"externally" since we lack a rational connection between the two belief-contents. The
centrality of the first-person to the understanding of transparency points to the fact that
19
Byrne, p. 85. Brie Gertler's paper, 'Self-Knowledge and the Transparency of Belief' makes a similar
criticism, but unlike Byrne, she makes it in the course of denying the appeal to transparency, rather than
as a reconstruction of it.
20
Boyle (forthcoming), argues for seeing the notion of agency relevant to self-knowledge as a form of
"self-activity", and against what he calls the "process conception" of deliberation.
32
not an inference from evidence about a particular person, but rather something more
like a general presupposition of rational thought, to the effect that, from the first-person
point of view, I must take what I believe about something to be the expression of my
sense of the reasons relating to the content of that belief. As we have seen, it is a
presupposition that may lapse in cases of compromised rationality, but at the same
time, it is hard to see what could be more basic to rationality than the idea that I take the
means for determining one's attitudes, for if I could not learn what I (really) think about X
by reflection on X itself, that would mean that my belief about X was out of reach of my
reflection on the facts concerning X itself. If we assume a subject with mastery of the
conditions of compromised rationality that a person could believe that it's raining,
possess the concept of belief, and yet be unable to know her belief through reflection on
the weather. 21
21
Compare Byrne, in summarizing his account as a version of the Broad Perceptual Model: "Thus there
is an appropriate causal mechanism. The state detected is independent of its detection. The subject
might not have followed BEL, in which case the first-order belief would have been present without the
second-order belief. What's more (we may fairly supposed) someone might believe that it's raining,
possess the concept of belief, and yet not even have the capacity to follow BEL." (p. 98)
33
Part of the sense of 'rational agency' in this domain lies in the fact that, with
committed to, then at a minimum I should be able to see my having this belief as a
potential reason for someone else to believe the same thing. My confidence in some
belief of mine may vary, but this will vary in accordance with whether what I take myself
to discover is a belief rather than a hunch or a mere guess. Hence I may not be
prepared to assert it, but if it is a belief at all I must take the fact that I think so to be
something that could be part of someone else's reason for believing the same thing.
Hence, while I am may not be 'active' with respect to it in the sense that it is a
conclusion I have arrived at just now, seeing it as a belief of mine means such things as
knowing what would counter it, knowing what sorts of reasons are relevant to it, my
learn.
More importantly, even for perceptual beliefs that are in some sense forced on
me, these still differ from mere feelings (sensations) in that a perceptual belief is still
others, and not just myself. Here we already see something of my agency involved in a
passive with respect to both of these, but in the case of my perceptual belief the sense
favor of the belief to be unassailable. This is not a matter of something merely befalling
me, like a headache. Rather, my 'passivity' here stems from the fact that, for me, to
34
change my belief would require a reason, and a reason of a special sort (viz., a reason
connected with truth), and hence I do not have what Pamela Hieronymi calls 'discretion'
with respect to this change, as I do with respect to the question of changing my shoes. 22
So, in that sense, my 'agency' is quite restricted here. But that restriction is just the
epistemic reasons for my beliefs, taking them to be the basis for what I can assert. And
in the case as we are imagining it here, within this restricted range of admissible
reasons there is dominance with respect to a single conclusion: I have every reason to
trust the visual evidence here and no reason to doubt it, hence I cannot help but believe
there is, e.g., a tomato in front of me. This is not like being assailed by a stabbing pain
(something not subject to justification in the first place), nor is it like being carried away
by a compulsion to believe that I will win the next round of black jack because the
stakes are so high. Rather, it is an instance of what is called being compelled by the
belief requires reasons, because beliefs are things I am prepared to defend, and in this
case the reasons are overwhelming in a particular direction. This 'passivity', then, is an
expression of my rational agency, and is utterly unlike my passivity or lack of choice with
In this way, then, the 'passivity of belief' is the reverse side of a person's rational
agency as a believer, for it is because one's beliefs are the expression of one's rational
relation to the world that they cannot be simply "chosen". If what I believe were not
answerable to the ways the world is, then I could indeed treat my beliefs as states which
22
'Controlling Attitudes', Pacific Philosophical Quarterly, vol. 87, no. 1, pp. 45 - 74.
35
I could seek to produce in myself for reasons unrelated to their truth. I could "control"
them in a way that enjoys "discretion" over the kinds of considerations that will count for
of another person, I may also see them as states to be produced or removed for purely
practical reasons of my own. In this sense I can be 'active' with respect to them the
way I am when I seek to manipulate the objects in my environment. The fact that I am
not active in this sense with respect to my own beliefs does not, however, mean that I
am passive with respect to them the way I am passive in relation to a sudden pain or
relation to my own attitudes is an expression of the fact that they are my cognitive and
affective relation to the world, and hence demand from me reasons connected with the
kinds of attitudes they are. A belief or a hope represents the world in a certain way, is
something to which justification is internal, and which stands in logical relations to other
attitudes. While there may be reasons for having a sensation, the sensation is not itself
relevant to believing are not reasons for having the belief, but rather reasons in favor of
the content of the thought, that is, reasons relating to its truth. The fact that in the case
of attitudes there is a contrast between reasons in favor of having the attitude and
reasons in favor of the content of the attitude itself is something that distinguishes them
When it is said that we cannot adopt a belief "at will", part of what this means is
that one cannot adopt a belief on a whim, in response to a simple request from another
36
person, or otherwise for reasons unrelated to its truth. Discretion of this sort represents
a kind of freedom or liberty we are familiar with in other contexts, a form of freedom that
appears to characterize our relation to our ordinary actions (e.g., changing one's shoes),
but does not appear to apply to our relation to our beliefs and other attitudes. Hence it
may be natural to conclude from this that the notions of agency and activity have no
bearing on the nature of belief and other attitudes, that they are "forced upon us", in
much the same way as are the facts of gravity or the inert objects in one's environment.
Or it may suggest that if a form of agency did characterize a person's relation to her
beliefs, that could only be as a form of managing or manipulating a set of items within
one's purview. This is, of course, not only unrealistic as a picture of psychological life,
but neglects the fact that the great majority of one's beliefs and other attitudes manage
to maintain themselves in a rough rational systematicity without any need for monitoring
or intervention from the person as such. And hence it may seem natural to conclude
from this that self-knowledge of attitudes like belief has no particular bearing on their
I've tried to suggest room for a different picture of the relation between self-
knowledge and agency, which shifts the picture of self-knowledge from that of
monitoring an internal condition to the ordinary ability to say what I'm doing and why, or
what I think about something and why. This also involves a difference in the kind of
active nature, as well as how I take myself to be answerable for what I believe. With
37
respect to the beliefs and attitudes of another person, I may exercise a kind of
productive agency, aiming at specific results. Here my reasons for favoring the
production of the belief as state may be fully separate from reasons in favor of what is
the fact of having some belief but rather the commitment to its truth and what that
requires of me. Detaching my relation to a state of belief (mine or another's) from the
commitment to its truth is precisely what would allow for discretionary reasons in
relation to its production. The first-person point of view presumes the absence of such
separation, presumes the identity of the considerations in favor of thing believed with
the fact of one's believing it. The absence of such separation characterizes the kind of
rational agency involved, specifically that it is not a matter of acting upon oneself or
taking oneself as an object, but rather of being resolved about some matter. It is this
namely that it makes possible a form of 'transparency', the fact that a person can
normally tell us what she thinks about some possibility by reflecting on that possibility
itself.23
Richard Moran
23
In writing this paper I have benefitted profoundly from conversations with Matthew Boyle, and from his
seminar on 'Self-Consciousness and Self-Knowledge' in the Fall of 2009. I had very helpful comments
from and conversation with Endre Begby as well as other participants at the conference 'Self-Knowledge
and Rational Agency', sponsored by the Center for the Study of Mind in Nature, in Oslo, June 2010. I am
grateful as well to Jonathan Vogel, Pamela Hieronymi, and Doug Lavin for conversations about these
issues, and to the editors of this volume for very helpful comments on the final version.
38
Department of Philosophy
Harvard University
39
BIBLIOGRAPHY
------, (forthcoming) 'On 'Making up your Mind and the Activity of Reason'.
Burge, Tyler, (1998), 'Reason and the First-Person', in Knowing Our Own Minds, edited
by Crispin Wright, Barry C. Smith, and Cynthia Macdonald: Oxford
Hieronymi, Pamela, 'Controlling Attitudes', Pacific Philosophical Quarterly, vol. 87, no. 1,
pp. 45 - 74
Shah, Nishi, and Velleman, David, (2005), 'Doxastic Deliberation', Philosophical Review,
vol. 114, no. 4.
(1988), 'On Knowing One's Own Mind', in J. Tomberlin, ed., Philosophical Perspectives:
Epistemology, Ascadero, CA: Ridgeview Publishing Co.
------, (2009), 'Self-Intimation and Second-Order Belief', Erkenntnis vol. 71, no. 1
-----, (1958), The Blue and the Brown Books, (Blackwell, Oxford).