ChapterPDF Available

Success and Knowledge in Action: Saving Anscombe’s Account of Intentionality


Abstract and Figures

According to Anscombe, acting intentionally entails knowledge in ac- tion. This thesis has been near-universally rejected due to a well-known counter- example by Davidson: a man intending to make ten legible carbon copies might not believe with confidence, and hence not know, that he will succeed. If he does, however, his action surely counts as intentional. Damaging as it seems, an even more powerful objection can be levelled against Anscombe: while act- ing, there is as yet no fact of the matter as to whether the agent will succeed. Since his belief that he will is not yet true while his action is in progress, he can- not possibly know that he is indeed bringing about the intended goal. Knowl- edge in action is not only unnecessary for intentional action, it seems, but–at least as regards success-bound types of action–impossible to attain in the first place. In this paper I argue that traditional strategies to counter these objections are unsatisfactory and propose a new account of knowledge in action which has two core features: (i) It invokes an externalist conception of justification which not only meets Davidson’s challenge, but also casts doubts on the tacit internalist premise on which his example relies. (ii) Drawing on recent work about by John MacFarlane, the proposed account conceives of claims to in action as assessment-sensitive so as to overcome the factivity objection. From a retrospective point of evaluation, previous claims about future events and actions can not only be deemed as having been true, but also as having been known.
Content may be subject to copyright.
Markus Kneer
Success and Knowledge in action: Saving
Anscombes Account of Intentionality
Abstract: According to Anscombe, acting intentionally entails knowledge in ac-
tion. This thesis has been near-universally rejected due to a well-known counter-
example by Davidson: a man intending to make ten legible carbon copies might
not believe with confidence, and hence not know, that he will succeed. If he
does, however, his action surely counts as intentional. Damaging as it seems,
an even more powerful objection can be levelled against Anscombe: while act-
ing, there is as yet no fact of the matter as to whether the agent will succeed.
Since his belief that he will is not yet true while his action is in progress, he can-
not possibly know that he is indeed bringing about the intended goal. Knowl-
edge in action is not only unnecessary for intentional action, it seems, butat
least as regards success-bound types of actionimpossible to attain in the first
In this paper I argue that traditional strategies to counter these objections
are unsatisfactory and propose a new account of knowledge in action which
has two core features: (i) It invokes an externalist conception of justification
which not only meets Davidsons challenge, but also casts doubts on the tacit
internalist premise on which his example relies. (ii) Drawing on recent work
about future contingents by John MacFarlane, the proposed account conceives
of claims to knowledge in action as assessment-sensitive so as to overcome
the factivity objection. From a retrospective point of evaluation, previous claims
about future events and actions can not only be deemed as having been true, but
also as having been known. This research was supported by SNSF grant
Note: This research was supported by the SNSF Ambizione grant PZ00P1_179912. Thanks to
Barry Maguire and the participants of Prof. Peter SchabersresearchcolloquiumattheUniver-
sity of Zurich for very helpful comments.
Markus Kneer,UniversityofZurich(
This is a proof. Please cite the final version of
the paper.
1 Knowledge in Action
In order to count as acting intentionally, Anscombe (1963) claims, an agent must
know what he is doing. More precisely, my φ-ing intentionally entails knowing
that Imφ-ing; it is a necessary (though not sufficient) condition for intentional
action. If, for instance, I am asked why Im tapping my foot and respond that I
wasnt aware of it, I do not count as having done so intentionally.
Despite its intuitive plausibility, the entailment thesis is near-universally re-
jectedprimarily due to Davidsons well-known carbon copier counterexample.
There is, however, an even more damaging objection that can be levelled against
Anscombe, an objection which does not only question whether knowledge is a
necessary condition for intentional action, but whether it is so much as possible.
In the following, I will briefly outline Anscombes theory of action in very broad
strokes, introduce the objections and assess traditional strategies of defending
the entailment thesis. Since the prospects of the latter are dim, I will sketch a
new account of knowledge in action, which, I hope, can put these worries to
rest. The resulting epistemology of action is less the fruit of Anscombe exegesis
than an independent attempt to make sense of the entailment thesis. It is consis-
tent with many of the rather idiosyncratic features of Anscombes philosophy of
action, yet relies on them as little as possible. As such, it is intended to appeal to
both Anscombians and her critics alike.
1.1 Textbook Anscombe
The knowledge a man has when acting intentionally, Anscombe argues, is spe-
cial in various respects. It is knowledge without observation, i.e. not based on
perception or inference, and as such groundless.¹In opening the window,An-
scombe writes, I do not pause and think to myself: Let me see, what are my
movements bringing about? The opening of a window.’” (1963: 51). To help intu-
ition along, knowledge without observation is frequently compared to bodily
awareness. Just as I know the position of my limbs without having to look, I
do not have to observe my bodily actions in order to come to know what it is
that Im doing. The similarities between the two types of knowledge are limited
though. As regards action, observational evidence cannot reveal what I am
For recent discussion of Anscombes theory of action, see Moran (2004) and Setiya (2007). For
knowledge without observation in particular, see Pickard (2004) as well as the survey of the lit-
erature by Schwenkler (2012).
132 Markus Kneer
doing, since a particular episode of bodily behavior could count as any number
of actions. When Im tapping my foot, what is given through observation from a
third-person perspective cannot settle whether I am communicating in codes
with the tenant below, following the rhythm of the music or am engaged in
yet another action (cf. Anscombe 1963: 11). In order to be φ-ing intentionally, I
have to conceive of my behavior under a particular description, that is, as
φing.²Importantly, what is known without observation is not merely what I
am taking myself to do, or what I am trying to do, butas Anscombe insists
what happens, namely, the event I am bringing about in the external world.
The epistemic attitude entailed by intentional action is thus distinctive in
three respects: (i) its content, as the agent has to know what she is doing, and
what is happening, under a particular description definitive of the action; (ii)
its source, as the knowledge one has in acting intentionally is not derived
from observation and finally (iii) the character of the epistemic attitude, which
is rather demanding. It does not suffice to believe, or to believe justifiedly,
that I am engaged in a particular action. According to Anscombe, nothing
short of knowledge in the full-blown sense will do.³
Features (i) and (ii)content and sourcehave given rise to considerable
controversy. Regarding the third featurethe knowledge criteriongeneral opin-
ion is more uniform: Although it is commonly accepted that intentional action is
accompanied by some sort of psychological attitude, the claim that the latter
amounts to an agentsknowledge, i.e. (at the very least) a true, justified, belief
of what she is doing, has triggered widespread criticism. Such an account
faces two principal problems. One concerns true belief, the other concerns belief
as such (or, as I will argue, justified belief).
1.2 The Factivity Problem
Lets begin with truth. In φ-ing, I have no guarantee whatsoever that I will reach
my goal. When attempting to swim across the Channel, for instance, I cannot be
sure of success. I might well believe that I am swimming across the Channel, or I
might know that I am attempting to swim across the Channel. But since there is,
so far, no fact of the matter about whether or not I will succeed, the belief that I
[T]osaythata man knows he is doing X is to give a description of what he is doing under
which he knows it(1963: 12).
As is well known, theresa fourth feature: the practical,ratherthanspeculativenature of
knowledge in action, in virtue of which what is known in action is the cause of what it under-
Success and Knowledge in action: Saving AnscombesAccountofIntentionality 133
am indeed swimming across the Channelthat I am doing what happens’—can-
not yet be true. Hence, there is no possible way for such a belief to amount to
knowledge. In stubbornly calling this attitude knowledge, non-observational or
not, we would have to jump to conclusions(Paul 2009) or take an epistemic
leap of faith(Langton 2004).
In the literature, the fundamental nature of the factivity problem is not suf-
ficiently appreciated and frequently glossed over.For instance, those arguing
against non-observational knowledge frequently end up with a two-factor ap-
proach(amad accountaccording to Anscombe): the agent knows what he in-
tends non- observationally, yet his knowledge that he is in fact bringing about the
intended event is based on perception.Since there is not yet a fact as to whether
the agent will indeed succeed in bringing about his goal while the action is still
in progress, however, perception is an inadequate source of knowledge to solve
the problem. The core problem concerns the absence of an object of knowledge
rather than the appropriate mode to apprehend it.
1.3 The Doxastic Problem
The second problem is that φ-ing intentionally might not even entail believing
that one is φ-ing, as Davidsons famous carbon copier example (1978: 9192)
demonstrates. A man is attempting to make ten legible copies by pressing his
LangtonsprimarytargetisnotAnscombesaccount,butVell emans, which is even more de-
manding epistemically. Whereas Anscombe makes knowledge in action a necessary requirement
for intentional action, Vellem an (198 9) iden tif i es int ent i on as suc h with a particular type of
knowledge (wishful, self-fulfilling true belief).
A notable exception is Grice, who gives an early statement of the problem. His focus lies with
future-directed intentions (IintendtoA), but translates to as yet uncompleted actions (Iwill
indeed succeed in doing A): AmanwhoexpressesanintentiontodoA(whosaysIintendto
do A)isinvolvedina factual commitment; he is logically committed to subscribing, with this or
that degree of firmness, to a factual statement to the effect that he will do A. [. . . ] The standard
source of entitlement to make such a factual statement is not available for this case, since the
ordinary concept of intention is such that if one intends to do A, one is logically debarred from
relying on evidence that one will in fact do A. No alternative source, however, of a different, non-
evidential kind, for the entitlement to say IshallinfactdoAseems to be forthcoming(1972:
Cf. Adams & Mele (1989) and Falvey (2000) for discussion.
This is not to say that the epistemic reach of non-observational knowledge should not be scru-
tinized. It might prove inadequate for reasons independent of factivity. We should be careful,
however, not to level the factivity problem against non-observational knowledge qua non-obser-
vational knowledge, as is not uncommon.
134 Markus Kneer
pen hard on a stack of blank sheets interspersed with carbon paper. All the
while, he is deeply sceptical of his success. As such, he does not believe that
he will indeed produce all the ten copies he needs. If he succeeds, however, it
would be astonishing to deem his action non-intentional. But if φ-ing intention-
ally does not entail that the agent believes that he is φ-ing, it will certainly not
entail that he knows he is φ-ing. Lets call this the doxastic problemfor knowl-
edge in action. Given that knowledge in action seems neither necessary (the dox-
astic problem) nor even so much as possible (the factivity problem), Anscombes
account might sound completely implausible.
In the next section, I will briefly survey three attempts to defend Anscombes
proposal, all deemed insufficient for one reason or another. Thereafter, Ill argue
against the two objections just raised. The goal is not to defend Anscombes com-
prehensive action theory in word and letter, but rather to show that the plausi-
bility of the entailment thesis is not as easily dispelled as is frequently assumed.
I will thus largely refrain from a cumbersome exegesis of Anscombe, yet occa-
sionally point out how the suggested account squares with the more general pic-
ture proposed in Intention (1963).
2 Defending the Knowledge Criterion
2.1 Reduction in Scope
There are different types of strategies to defend some variation of the knowledge
criterion. One might, for instance, attempt to reduce its scope by imposing re-
strictions on what it is that needs to be known by the agent. Even in the carbon
copier example, there is something the agent knows about his actions and, it
might be argued, it is in virtue of that knowledge that the action counts as inten-
tional. As Davidson himself acknowledges, when acting intentionally, what the
agent does is known to him under some description(1971: 50). The carbon cop-
ier, we might hold, certainly knows that he intends to make ten copies (Donnel-
lan 1963), or that he is trying to make ten copies (OShaughnessy 2003).
Some authors deem such a change in the object of knowledge inappropriate.
They propose to stick to the object of knowledge envisioned by Anscombe, which
does not consist in the agents intending or his trying to do something, but his
very doing and that which happens. The complications are to be accommodated
by opting for a less demanding epistemic attitude, that is to say, belief rather
than knowledge. Yet others hold that the only way to defend some form of the
knowledge criterion is to opt both for a different epistemic attitude and a
more modest object of knowledge. Setiya (2007), for instance, does precisely
Success and Knowledge in action: Saving AnscombesAccountofIntentionality 135
that. He proposes that the agent must act in the belief of doing particular things
such as, in Davidsons example, pressing hard on the paper, with the end of mak-
ing ten copies. On this account, it suffices that the agent believes, rather than
knows, that he is engaging in some of the actions constitutive of bringing
about his more general goal. Such strategies are, as I attempt to show below, un-
satisfactory because they concede too much.
2.2 Action Descriptions in Progressive Aspect
A more promising path to pursue is the appeal to the broadnessof the progres-
sive of action verbs.Action descriptions involving atelic verbs in progressive
form (is swimming,is walkingetc.) do not have a success condition built in
and thus impose limited epistemic demands. While swimming, I can know at
any point in the course of doing so that what I am taking myself to do ishard-
core scepticism asidein fact happening. No epistemic leaps of faith required. In
the case of atelic action verbs, the truth of a description in the progressive tense
licenses the truth of its description in the past tense: if I am swimming, it will be
the case that I swam or that I have been swimming. For telic action verbsverbs
with a success condition built into their semanticsthat is not necessarily the
case: if I am trying to swim across the Channel, and turn around after a few
strokes, it will not be true that I have crossed the Channel or that I was crossing
the Channel. While trying to cross the Channel, it is simply impossible for me to
know that I am indeed getting to the other side since any such fact has not yet
been established.
The flexibility introduced by the progressive tense, however, is not limited to
atelic action descriptions. It extends to a considerable class of telic actionswe
might call them weakly telic’—even if the stipulated goal rests unfulfilled. For
instance, one might count as crossing the street though one never makes it to
the other side.Actual completion is not essentialnothing stands in the way
of granting the agent knowledge of her actions, just as in the case of atelic action
descriptions. However, in many cases (call them strongly telic) the completion
Cf. Anscombe (1963: 39), developed in Falvey (2000). Further discussion in Thompson (2008,
Ch. 8), Paul (2009), Haddock (2011) and Schwenkler (2012).
Hence Anscombes insistence that a man can be doing something which he nevertheless
does not do(1963: 39). The interesting though somewhat murky distinction between weakly
and strongly telic actions extends to the past progressive tense.Whereas it sounds infelicitous
to say Mary was crossing the Channelif she returned ashore after a few strokes, theres nothing
wrong with saying that Frank was crossing the streetif he turns back or gets run over.
136 Markus Kneer
of the goal stipulated by the action verb is essential for the action to be consid-
ered as taking place or for the event to happen. I dont count as crossing the
Channel if I give up after a few strokes, as finishing my soup or killing a fly if
I dont, or as walking towards the library if the distance between it and me is
not decreasing.¹But if strongly telic actions are success-bound, how can I
know that I am indeed doing what I intend to do? The appeal to the broadness
of the progressive tense is not effective here to overcome the factivity problem. As
such, it can only serve as a partial defence of the entailment thesis.
2.3 Practical Knowledge
The third strategy to defend the knowledge criterion takes its cue from yet anoth-
er peculiarity Anscombe has in stock when it comes to knowledge in action.
Such knowledge, she proposes, is not like ordinary, speculativeknowledge,
i.e. passive or receptive,in so far as it aims to fit the facts or is derived
from the objects known.Rather, it is practicalin nature and, as such, the
cause of what it understands(Anscombe 1963: 87). If I happen to be mistaken
as to what Im doing, the mistake here is one of performance, not of judgement
(5, 56, 57, 8789). The precise nature of practical knowledge remains one of the
more elusive chapters of Anscombes theory of action. Strategically, this way to
save the entailment thesis should nonetheless already be obvious: it consists in
cashing in maximally on the resources provided by practical knowledge so as to
help overcome the factivity problem. However, the more such practical knowl-
edge is moulded into a type of epistemic state that does not constitutively aim
at representing the facts, the less it deserves its name and the more bewildering
the resulting picture of intentional action.¹¹
This point can be made in somewhat more detail. Lets have a look at An-
scombes discussion of the difference between speculative (or contemplative)
and practical knowledge:
 Last example by Haddock (2011). Here the strongly telic aspect seems to be imposed not by
the verb but by the preposition. Interestingly, there is not only a general success condition at
work in this type of action description, but a stipulated manner of how the result is to be brought
about. Though rather different from standard strongly telic actions, they pose asimilarproblem
for the entailment thesis.
 For adiscussionofAnscombe on practical knowledge, cf. Velle man (1 989) , Moran (2004),
Setiya (2007, 2008, 2009), McDowell (2011), Haddock (2011), Thompson (2011) and Schwenkler
Success and Knowledge in action: Saving AnscombesAccountofIntentionality 137
Can it be that there is something that modern philosophy has blankly misunderstood:
namely what ancient and medieval philosophers meant by practical knowledge? Certainly
in modern philosophy we have an incorrigibly contemplative conception of knowledge.
Knowledge must be something that is judged as such by being in accordance with the
facts. The facts, reality, are prior, and dictate what is to be said, if it is knowledge. And
this is the explanation of the utter darkness in which we found ourselves. For if there
are two knowledgesone by observation, the other in intentionthen it looks as if there
must be two objects of knowledge; but if one says the objects are the same, one looks hope-
lessly for the different mode of contemplative knowledge in acting, as if there were avery
queer and special sort of seeing eye in the middle of the acting (1963, 57).¹²
Heres one way to interpret this passage: the contemplative conception of knowl-
edge seems unsuited for knowing what one is doing, since facts, reality, are
prior, and dictate what is to be said”—and presumably also what can be
known. The factivity of contemplative knowledge would require a very queer
and special sort of seeing eye in the middle of the acting”—an eye that antici-
pates what will come to pass, an eye that jumps to conclusions. Introducing a
second type of knowledge of a practical nature seems to bring an extra compli-
cation into the picture: if there are two types of knowledge, should there not be
two different objects of knowledge as well? Practical knowledge,we might think,
captures what I am (or take myself to be) doing, contemplative knowledge what
happens. But multiplying the objects of knowledge would be a mistake, as An-
scombe vigorously argues in various places. If things work out, the action and
the event are one and the same thing—‘I do what happens. In virtue of being
defined by the same description, they must constitute a single object of knowl-
edge.¹³ Sticking with a single object of knowledge, however, does not prove the
futility of a second kind of knowledge: a single object of knowledge might be ap-
 As so often happens in Intention,various strandsofthediscussionare runtogether. From
this passage it seems that the dichotomy between observational and non-observational knowl-
edge maps onto the one between speculative and practical knowledge. This is of course not the
case. A priori knowledge, such as mathematical knowledge, is clearly responsive to facts in or-
dinary ways without drawing on observational input.
 Heresa characteristic passage: The difficulty however is this: What can opening the win-
dow be except making such-and-such movements with such-and-such a result? And in that case
what can knowing one is opening the window be except knowing that that is taking place? Now
if there are two ways of knowing here, one of which I call knowledge of onesintentionalaction
and the other of which I call knowledge by observation of what takes place, then must there not
be two objects of knowledge? How can one speak of two different knowledges of exactly the
same thing? It is not that there are two descriptions of the same thing, both of which are
known, as when one knows that something is red and that it is coloured; no, here the descrip-
tion, opening the window, is identical, whether it is known by observation or by its being ones
intentional action(1963: 51).
138 Markus Kneer
proached via different modes of knowledge, a ball can be known to be spherical
through touch or vision. Hence the question whether modern philosophy, with
its preference for a contemplative conception of knowledge is not up a blind
alley and should rather approach the same object of knowledge via a different
epistemic mode. What epistemic mode? Practical knowledge. And practical
knowledge, to repeat, is not derived from the objects known, but the cause
of what it understands.
One might reject Anscombes account of practical knowledge outright as
causally perverse and epistemically mysterious(Velleman 1989: 103). As re-
gards the alleged causality, however, we could follow Morans suggestion and
treat practical knowledge as the formal rather than the efficient cause of what
it understands. In virtue of the operative description which defines my doings
as a particular action, practical knowledge specifies intensionally, rather than
causes extensionally what it understands. Anscombes slogan concerns the for-
mal or constitutive role of the description embedded in ones practical knowl-
edge making it the case that this description counts as a description of the per-
sons intentional action. If the agent didntknow this happening under this
description, then as so specified, it would not be what he is intentionally
doing’” (Moran 2004: 54, italics in the original).
Though the charge of causal perversity might thus be averted, its potentially
non-factive nature remains a mysterya mystery that is captured well, yet not
resolved, by Anscombe herself:
If then my knowledge is independent of what actually happens, how can it be knowledge of
what does happen? Someone might say that it was a funny sort of knowledge that was still
knowledge even though what it was knowledge of was not the case! On the other hand The-
ophrastusremark holds good: the mistake is in the performance, not in the judgment
(1963: 82).
Since a non-factive conception of practical knowledge would be a funny sort of
knowledgeindeed, this strategy to save the entailment thesis might be deemed
heroic, but remains implausible from the outset. Theophrastusdictum does little
to dispel the worries.
Lets briefly take stock. On the one hand, not knowing I was φ-ing is an ex-
cellent explanation of my not φ-ing intentionally (subconscious conundrum
aside). On the other hand, it is difficult to explain how it is even so much as pos-
sible to know that I am actually φ-ing while φ-ing, at least as long as we refuse
to make do with atelic actions only, water down knowledge to knowledge of try-
ing or mere belief and shy away from an utterly implausible, non-factive concep-
tion of practicalknowledge. But the knowledge criterion can be defended, I
Success and Knowledge in action: Saving AnscombesAccountofIntentionality 139
think, without making any concessions. In the following, Ill attempt to show
3 Knowledge Proper and the Factivity Problem
Intentional action, conceived as entailing knowledge of what one is doing, we
have said, gives rise to two fundamental problems. Firstly, as Davidson has ar-
gued, in acting intentionally, one might not need to hold a corresponding belief,
let alone know what one is doing. Secondly, it would be reassuring if we could
leave epistemology proper in place, i.e. agree that [k]nowledge must be some-
thing that is judged as such by being in accordance with the facts”—but it is not
obvious how this could be so much as possible. Both problems are strongly re-
lated to the success of ones action: in the former case, what undermines belief is
a low perceived probability as to ones abilities to succeed. In the latter case,
prior to having successfully completed ones action, there simply is no fact
that can be known in acting intentionally.
I will argue that neither problem is specific to action theory. The very possi-
bility of knowledge regarding ones action is a variation of the semantic puzzle of
future contingents, a puzzle that can be solved. Davidsons counterargument, I
will suggest thereafter, is less a matter of belief than of justification. As such
it raises questions regarding internalist and externalist accounts of epistemic jus-
tification, and thus concerns a complication of epistemology proper rather than
one of action theory narrowly conceived. Whereas the general debate between
internalists and externalists has not given rise to a consensus after decades of
argument, externalism, Ill suggest, could well carry the day as regards knowl-
edge in action.
3.1 Assessment-sensitivity of Future Contingents
Lets begin with the factivity problem. On an indeterminist view, the future
course of affairs is a contingent, rather than a settled, or necessary, matter. At
every moment in time, there is a variety of genuine possibilities as to what
world will be actual in the future. Along which branch the world will develop
as the future unfolds is more than just epistemically indeterminate, it is meta-
physically indeterminate as there is as yet no fact of the matter.
Indeterminism gives rise to a semantic paradox regarding utterances of fu-
ture contingents. Take the following example:
140 Markus Kneer
(1) Frank: It will be sunny tomorrow.
At Franks context of utterance, whether it will be sunny the next day is still in-
determinate. Accordingly, the proposition¹expressed by (1) must be judged as
neither true nor false. Call this, following MacFarlane, the indeterminacy intu-
ition. The next day, lets assume, the sun is shining. In retrospect, we are inclined
to hold that what Frank said was true. That is, we deem the proposition ex-
pressed by Franks utterance as true at his context of utterance. Call this the de-
terminacy intuition. Problematically, orthodox semantics cannot accommodate
these conflicting intuitions.
In Kaplanian semantics¹acontext is a potential occasion of utterance for a
sentence S which plays a twofold role.On the one hand, it determines the seman-
tic values for the indexical expressions of the sentence. In Kaplans terms, the
context and the character of the sentence (or sentence type) jointly deliver the
content expressed. Though your utterance I am coldand mine have the same
character, they express different contents because the context provides different
individuals as the semantic value of the indexical I. On the other hand, the con-
text determines the circumstances of evaluation. That is, it specifies under which
set of parameters (a world, a time, potentially a location, a standard of precision
etc.) the truth or falsity of the content should be evaluated. If on Monday,
drenched to the bones, I utter Its rainingand claim the same the next day in
bright sunshine, the difference in truth value is accounted for by the fact that
my different contexts of utterance specify different time parameters and hence
different circumstances of evaluation. In the first case, the circumstance of eval-
uation is the actual world on Monday, in the second its the actual world on
Tuesday. Though identical in content, my utterance on Monday is true, but my
utterance on Tuesday is false. Importantly, whether or not my utterance was
true yesterday depends on the context of utterance, not the context of assess-
mentthere is no such thing on this view. On Tuesday, yesterdays claim that
it was raining remains true, because the time parameter in the circumstances
of evaluation provided by the circumstance of utterance is Monday. The only
context that affects the circumstances of evaluation is the context of utterance,
 This section draws heavily on the work of John MacFarlane. His original statement of the par-
adox was phrased in terms of utterances (MacFarlane 2003), but neither the puzzle nor its sol-
ution resist formulation in terms of uttered propositions (MacFarlane 2008, 2014). For acritical
discussion cf., for instance, Cappelen & Hawthorne (2009) and the contributions to Around the
Tree (Correia & Iacona 2012).
 Cf. Kaplan (1979, 1989).
Success and Knowledge in action: Saving AnscombesAccountofIntentionality 141
so there cannot be any truth value switching which might explain our contradic-
tory intuitions regarding future contingents.
A standard way to make sense of the indeterminacy intuition is to seek re-
course to supervaluationism, a semantics that besides trueand falseintrodu-
ces a third value, indeterminate(or neither true nor false). According to Tho-
mason (1970), for instance, an utterance is true or false simpliciter if true or false
in all possible worlds and otherwise indeterminate. Franks claim It will be
sunny tomorrowis evaluated as neither true nor false, and even though the
next day the sun happens to be shining, it is still not true in all possible worlds,
therefore this evaluation wont change. Though indeterminacy is accounted for,
our retrospective determinacy intuition is not.
MacFarlanes solution¹to the puzzle is to expand the roles contexts can
play: besides the context of utterance, we also have to take the context of assess-
ment into account. Presume that on Monday (t1) there is a genuine possibility for
it to rain the next day (t2). This means that the world actual at c1branches into
sunny worlds (w1) and rainy worlds (w2, w3), as illustrated in Figure 1. Which of
these worlds will turn out actual is, on Monday, still not settled. The context of
Franks utterance, c1, specifies a circumstance of evaluation comprising of the
world of utterance (up to then w@=w1=w2=w3) and the current time t1. At that cir-
cumstance, Franks claim It will be sunny tomorrowis indeterminate. Once the
future has unfolded and w1, a sunny world, has turned out actual, (1)as uttered
at c1(w1=w2=w3, t1) and assessed from c2(w1, t2)is true.
According to the supervaluationist picture, retrospective evaluation of (1)
takes into account all worlds overlapping at the context of utterance c1, that
is, w1, w2and w3. In virtue of not being true at all these worlds, (1) is, and re-
mains, indeterminate. On MacFarlanes view, the retrospective evaluation of (1)
focuses on only some of the worlds overlapping at the context of utterance,
namely those which also branch through the context of assessment. If it turns
out that the latter is c2, (1) was true as uttered at c1and assessed from c2; if it
turns out that the context of assessment is c3(w3, t2), (1) was false as uttered
at c1and assessed from c3. Both the intuition of prospective indeterminacy and
retrospective determinacy are borne out.
 I am simplifying considerably. What matter for our purposes is the basic idea. For details re-
garding the semantic framework, see MacFarlane (2014).
142 Markus Kneer
3.2 Assessment-sensitivity of Knowledge Attributions
This curiosity regarding the truth of a proposition uttered or entertained extends,
I think, to the attribution of knowledge of propositions entertained. Conceive of
knowledge as a justified, true belief, and assume that Franks belief that it will
be sunny tomorrow is well justified. In an indeterministic universe, Frank cannot
know today that it will be sunny tomorrow, because his belief is not true yet
rain tomorrow is a genuine metaphysical possibility. In retrospect, however, if
Frank was right and had good reasons for his belief, it is perfectly felicitous to
say that Frank knew it would be sunny today. With hindsight, not only the alethic
assessment becomes more determinate (a truth-indeterminate proposition gets a
determinate truth value) but also the assessment of the subjects epistemic state
(a truth-indeterminate justified belief becomes a true justified belief, i. e. knowl-
edge, or a justified false belief). At a context of assessment which equals the con-
text at which Frank entertains his belief, his epistemic state is one of justified
belief. Once the future has unfolded and w1, a sunny world, has turned out ac-
tual, claim (1), as entertained at c1(w1=w2=w3, t1) and assessed from c2(w1, t2), will
Figure 1: Context-sensitivity in an indeterministic universe.
Success and Knowledge in action: Saving AnscombesAccountofIntentionality 143
be true and so Frank will count has having held a true, justified beliefthat is,
he will count as having known that it would be sunny the next day.¹
The fact that knowledge attributions are context-sensitive is not news.¹For
instance, in standard scenarios, John might count as knowing that his car is in
his driveway, if thats where he left it. If someone voices the possibility that it
might have been stolen, however, John will likely concede that he doesnt
know. The revision is less a matter of correcting a mistaken claim to knowledge,
than one of adapting to a different, more demanding standard of knowledge.
John will not only acknowledge that he now doesntknow, but also that he
didntpreviously know the whereabouts of his car before the standards were
raised. To make sense of this retraction, we have to evaluate Johns claim not ac-
cording to the (low) epistemic standards of the context of utterance but accord-
ing to the (higher) standards of the context of assessment . The context-sensitivity
at play in such cases, however, is importantly different from the one developed
above. What explains the variation in knowledge attributions when it comes to
different epistemic standards are variations in justification. According to ordinary
standards, Johns belief that his car is in the driveway is justified. Judged by a
differenthigherset of standards, it might not be, and the belief will thus
not count as knowledge. What explains the variation in knowledge attribution
in the previous paragraph, by contrast, regards not justification but the truth
of the subjects belief. Though the context-sensitivity thus affects different con-
stituents of knowledge (truth v. justification), they manifest the same basic rela-
tivist logic: knowledge ascriptions, it seems, are sensitive not only to contexts of
entertainment/utterance but to the context of assessment in several ways.¹
 Attacking the premise of an open future does not weaken the argument, but makes it, if any-
thing, stronger. In that case knowledge ascriptions at the context of utterance are no longer
problematic due to metaphysical and epistemic indeterminism, but only due to the latter: In a
deterministic universe, the future course of events is settled and beliefs about it are either
true or false. Naturally, on certain accounts of justification we might still be loath to call such
beliefs knowledge before the future has materialized. From asuitablecontext of assessment,
however, future contingents can not only be understood as having been true (obviously so,
due to determinism), butif well justifiedas having been known.
 See, for instance, MacFarlane (2005), who covers context-sensitivity regarding standards of
precision. The example discussed in the main text is his.
 The application of assessment-sensitivity to epistemic expressions has become afertilefield
of research recently. Besides the mentioned standards of precision, epistemic modals can also be
seen as sensitive to contexts of assessment (cf. Egan, Hawthorne & Weatherson 2005; Egan 2007;
Stephenson 2007, MacFarlane 2011). The linguistic intuitions inspiring such accounts, however,
are not uncontroversial (cf. Hawthorne 2007, Vo n Finte l & Gillies 2007, 2008; Yalcin 2011; Braun
2012; Kneer 2015, 2020). Still, one can safely accept the assessment-sensitivity of future contin-
gentswhere intuitions are considerably more uniformwithout buying into arampantrelativ-
144 Markus Kneer
3.3 Assessment-sensitivity of Knowledge in Action
Utterances and beliefs regarding actions underway are utterances and beliefs re-
garding events in progress. If the latter can be true, and known, at a context of
being in progress when assessed from a context of completion, so can the for-
mer. There is thus nothing mysterious about the knowledge a man has when
acting intentionally. From a retrospective context of successful completion,
the agent, if holding a suitably justified belief, can be understood as having
known that she was acting as intended at the context of action in progress.
There is no reason to presume that she is making an epistemic leap of faith,
or jumping to conclusions. This is of course perfectly in line with our ordinary
ways of speaking and acting. Having completed an action successfully while
having had good justification that I would, its as natural for me to say that I
knew I was baking a cake or that I knew I was making the boss uncomfortable
as it is to say that I knew thered be a department meeting today. The opposite
would be deeply counterintuitive: if I can have knowledge regarding future
events, it would be astonishing if I could not have knowledge regarding future
events which I can directly influence and whose progress I can monitor.
Just as the attributions of intentionality and knowledge in action go hand in
hand on this account if the action is successful, the absence of one feature will
coincide with the absence of the other.²When my action is unsuccessful, my
claim to practical knowledge fails. My belief about my doings, despite being
maybe well justified, turns out false. Since knowledge is factive, I simply cannot
have known. Relatedly, if I unintentionally bring about B, while trying to do A, I
also lack knowledge in action. I thought of myself as bringing about A, not B,
and as long as B is not constitutive of A, I will cite precisely this factthat I
didnt know I was doing Bas evidence in favor of the assessment that I did
not do B intentionally.
The account of knowledge in action developed here leaves the fundamental
traits of Anscombes picture in place. Th e relevant type of knowledge is still prac-
tical in Morans qualified sense and differs from purely speculativeknowledge.
My conception or descriptionof what I am doing is an essential constituent of
the knowledge I have when acting intentionally. It defines my behavior as a par-
ticular action, and it is in this sense that practical knowledge should be con-
ism regarding epistemic notions in general. Besides, while the solution sketched here is devel-
oped in terms of MacFarlanes relativist semantics, a similarly strong case could be made by em-
ploying more conventional frameworks such as, for instance, the one developed in Brogaard
 I am ignoring counterexamples a
`la Davidson for the moment, but will turn to them shortly.
Success and Knowledge in action: Saving AnscombesAccountofIntentionality 145
ceived as the [non-efficient] cause of what it understands. Knowledge in action
so understood is also in some fundamental way non-observational: I do not
pause and look to see what Im bringing about. Given the underdetermination
of observational evidence as regards the definition of an action, its simply im-
possible to find out. It is in virtue of the very description I have of my doings that
they count as the action in progress, and that description I surely know without
The question whether I can know what I am doing, and what happens, ex-
clusively in a non-observational fashion remains, of course, highly controversial.
Moran, for instance, argues that practical knowledge in action, despite being in
an important sense non-observational, is nonetheless aided by observation. I do
not want to get entangled in this debate. One brief point, however: the question
of non-observational purity can be rephrased as the question whether we can
only attribute knowledge in action post-hoc, if the agent has made sure by per-
ceptual means that he indeed fulfilled his aim. Prima facie, such a condition
does not seem necessary, in which case the chances for exclusively non-observa-
tional practical knowledge might stand better than frequently assumed.
4 The Doxastic Problem
4.1 Belief and Justification
Davidsons famous counterexample is frequently reported thus: the carbon cop-
ier, skeptical as to his success of producing ten legible copies, cannot be said to
know that he will make ten copies, because he does in fact not even believe that
he will. This argument, according to which knowledge is a fortiori out of the
question, is rather unconvincing and it is not what Davidson had in mind.²¹ In
clear-cut cases in which an agent lacks the belief that he is φ-ing, his φ-ing
will not be deemed intentional. For instance, if a man scribbles absentmindedly
on a stack of carbon papers (hence lacking the belief that he is) and produces ten
legible copies, there is little doubt that he didnt make them intentionally. Or, to
take another type of case, presume someone attributes zero probability to his
succeeding in φ-ing, in which case he cannot be said to believe that he is φ-
ing. Pursuing his action nonetheless is much rather a manifestation of irrational
 Things are of course different if we follow Williamsons(2000)suggestionaccordingto
which the verb to knowpicks out asui generis mental state that cannot be factored out into
more basic constituents.
146 Markus Kneer
behavior than of intentional action, quite independently of whether the goal is
attained. The entailment thesis is not under pressure from such cases.
Now, beliefs do not have to be held with certainty. If an agent attributes a
low probability to the truth of his belief, it is no less a beliefeven though, on
some accounts, it might be deemed unjustified. I may, for instance, cling to
the belief that it will be sunny in London all summer despite being aware that
the probability of that is extremely low. Of course, if it turns out to be sunny
in London all summer, my belief, in virtue of being poorly justified, will not
amount to knowledge. The carbon copiercall him Donaldwho considers his
chances of success relatively low, but not inexistent, can thus be conceived as
believing to some degree, but arguably not as knowing, that he is making ten
copies. The doxastic problem, as we called it, does not derive from an absence
of belief, but rather from a lack of epistemic justification for the belief regarding
ones action.
4.2 Internalist and Externalist Justifications
Lets take a look at how Davidson phrases his example:
[I]n writing heavily on this page I may be intending to produce ten legible carbon copies. I
do not know, or believe with any confidence, that Iamsucceeding.ButifIamproducing
ten legible carbon copies, Iamcertainlydoingitintentionally(Davidson 1978: 2001: 80).²²
The argument thus comes to this:
1. I can φintentionally without believing with confidence that Im indeed φ-
2. Knowing that Imφ-ing entails believing with confidence that Imφ-ing.
3. Therefore, I can φintentionally without knowing that Imφ-ing.
The crucial premise is of course the tacit one, i. e. (2). What is the philosophical
motivation for such an assumption? The only explanation that comes to mind is
that Davidson implicitly pledges allegiance to (a strong form of) epistemic inter-
nalism. Internalism is the view that one must be aware of, or be able to become
aware of, the grounds that justify ones belief that p. Differently put, justifiers
must be internalto and directly accessible from the subjects cognitive perspec-
tive. On another account, the subject has to fulfill certain epistemic duties in
order to count as holding a justified belief. Attributing a low subjective probabil-
 For a slightly different formulation, cf. also Davidson (1971, in 2001: 50).
Success and Knowledge in action: Saving AnscombesAccountofIntentionality 147
ity to the belief that one will actually φcan be seen as a defeater to the internal-
ist belief justification. Lacking justification, the belief is no longer a candidate
for knowledge. The man who is skeptical about whether he will succeed in mak-
ing ten legible copies does not hold the belief that he will indeed succeed with
justification and can thus not be considered knowing that he will.
The alternative to epistemic internalism is externalism.²³ Weak externalism
denies that justification should exclusively be conceived as an internal matter,
a strong form denies that justification is ever internalist. According to this
view, the justifiers of a belief need not be accessible to the believer, as they
can be external to her cognitive perspective. What matters instead is (according
to one approach) that the subjects belief reliably tracks the truth, whether or not
she thinks it does. An unconfident examinee reliably giving the right answer to a
certain question can be seen as knowing, despite the fact that she is unsure and
has no inkling of how she came about the information.
Adapted to knowledge in action, we might similarly consider self-confidence
or subjective probability less important than objective probability, that is, the
agents reliable disposition to successfully carry out the intended action. In
this case, the justification of my belief that Imφ-ing will not be defeated by
my lack of self-confidence; what justifies my belief can be external to (and
hence unimpinged by) my cognitive perspective. According to such a view,
there is no prima facie problem in attributing knowledge to Davidsons carbon
copier. The mans reliable disposition to make ten copies (whether or not he
has confidence in himself) is sufficient to justify his belief and thus to warrant
the ascription of knowledge. If, on the other hand, the mans objective probabil-
ity of success is extremely low, his belief that he will make ten copies should not
count as knowledge.²
4.3 Objective and subjective probability
Lets bring the proposed view into somewhat sharper focus by aid of an example.
Sitting in front of a stack of 100 sheets interspersed with carbon papers, Mary is
 Chisholm (1966), Bonjour (1985) and Lehrer (2000) are well-known advocates of internalism;
Goldman (1967), Armstrong (1973), Dretske (1971, 1981) and Nozick (1981) are externalists.
 The variations of internalist and externalist approaches to justification are plentiful and the
line between the two types of account can be drawn in very different ways (cf. Kim 1993). The
internalist/externalist dichotomy, developed here in terms of subjective and objective probabil-
ity, is orthogonal to some of the possible distinctions, yet fits the general thrust of several ac-
counts naturally enough.
148 Markus Kneer
trying to make exactly 57 copiesnot one more, and not one less. Like Donald,
she deems the probability of success low. If she, to her surprise, succeeds, we
would call this a lucky accident rather than an intentional action. Superficially,
the two scenarios are very similar: In both cases the agents attribute a low prob-
ability of truth to the belief that they are indeed doing what they intend to do.
Due to the perceived marginal chances of success, the beliefs are poorly justi-
fiedneither agent has good grounds to believe they will fulfil their goal. Both
beliefs turn out true. We cannot attribute knowledge to either of the agents,
yet Donalds action counts as intentional whereas Marys as a fluke. Why is tha t?
The difference between the two cases is, of course, one of objective probabil-
ity of success.²In Marys case, the probability is extremely low, in Donalds it is
considerably higher. Whether or not an action counts as intentional, it thus
seems, depends less on the subjectively perceived probability of success, and
more on objective probability, i. e. on the actual difficulty of the task for the
agent. This gives us a matrix of four basic cases (Table 1): Lets start with Donald
and Mary, who are both skeptical about their success (bottom row of the table),
but Donalds task is conceived as comparatively easy, whereas Marys as hard.
Intuitively, if successful, Marys action counts as a lucky accident, whereas Do-
nalds counts as intentional.²Presume that John also wants to make ten carbon
copies and that Sally also wants to make 57 out of 100 possible ones, but in com-
parison to Donald and Mary they are both confident as to their success (upper
row of the table), i. e. they attribute a high subjective probability to their belief s
being true. Table 1 summarizes the different cases for the success condition:
Table 1: Dependence of intentionality and knowledge on objective and subjective probability.
High objective probability Low objective probability
High subjective probability John
Intentional Action
Lucky Accident
Low subjective probability Donald
Intentional Action
Lucky Accident
No Knowledge
 More precisely, what matters is the estimated objective probability of success as perceived
from aperspective external to the subjectthe same perspective from which the intentionality
of the action is assessed. It is not the (frequently unknown) actual objective probability of suc-
cess that matters, because the latter is not generally what informs our ascriptions of intention-
ality or our assessment of an agents epistemic situation.
 In fact, empirical evidence suggests that intentionality ascriptions might depend only on ob-
jective probability and not on subjective probability. See, for instance, Mele & Cushman (2007).
Success and Knowledge in action: Saving AnscombesAccountofIntentionality 149
Where subjective and objective probability coincide, the assessment is un-
controversial: Johns action is intentional, his belief is both true and justified
(no matter whether were internalists or externalists) and hence an instance of
knowledge. Mary, if successful, would be deemed lucky. Due to the low probabil-
ity of success, of which she is well aware, she cannot justifiably claim that she
knew she was indeed making 57 out of 100 copies. Such cases of homogenous
probabilities are consistent with both Anscombes and Davidsons views.
Things get complicated where the two types of probability diverge. To attack
knowledge as a necessary condition for intentional action, Davidsonians argue
that Donald doesnt have knowledge in action, whereas his action, if successful,
is nonetheless judged intentional. Considering his chances of success low, Don-
ald does not have reasons to take his beliefi.e. that he will indeed make ten
copiesto be well justified. If said belief, unjustified from his perspective,
turns out true, it would, according to this picture, come somewhat as a surprise
to hear him insist that he knew all along that he was making ten copies.
From an external perspective, however, the belief held by Donald with a low
degree of confidence is not unjustified: there is a high enough objective proba-
bility of succeeding in simple tasks such as this one. From such an external per-
spective, Sallys belief that she will make 57 out of 100 copies, despite being held
with great confidence, on the other hand, does not seem well justified. Hence, on
such an externalist account of justification, Donald can be said to know what
hes doing, whereas Sally cannot. Differently put, according to such a view, An-
scombes entailment thesis faces no obstacle.²
 The general recipe for construing counterexamples to the entailment thesis consists in max-
imizing objective probability by invoking an extremely simple everyday action, while minimizing
subjective probability through imbuing the agent with legitimate doubts as to the success of his
actions. Heresa well-known scenario, standardly attributed to Bratman:
Im recovering from paralysis of my right hand. I try to clench my fist, though am sceptical
whether Ill succeed. If I succeed, I cannot be said to have known Iwould.However,Ihave
surely clenched my fist intentionally.
The simplicity of the action guarantees that it is assessed as intentional. The legitimacy of my
doubts prevent us from ascribing knowledge. The trick lies not only in stipulating astrongmis-
match between the two types of probabilities, but in obfuscating the fact that their contrasting
levels should affect each other and hence the attribution of knowledge and intentionality: if Ido
have legitimate doubts as to whether Icanindeedclenchmyfistsomething that generally
takes months of practice after paralysisthen it is by no means obvious why my unlikely success
should count as intentional. An objective probability of five percent does not make clenching my
fist more intentional than hitting the bullseyewithbowandarrow,justbecausetheformeris
standardly conceived as an easy task and the latter is not. If, on the other hand, I am indeed
150 Markus Kneer
Objective probability captures whether a certain action intended by an agent
reliably comes to pass. As such, it is objective probability, not the agents confi-
dence, that determines whether we deem an action intentional or not. It is sug-
gested that the same type of probability also determines whether we are willing
to consider an agents belief about his action in progress justified.²The strong
relation between intentionality and knowledge in action postulated by An-
scombe can thus at least partially be accounted for by the fact that both inten-
tionality and belief justification derive from a common sourceobjective proba-
bilityand thus give rise to the same necessary condition. If I am able to φ
reliably, and my φ-ing counts as intentional in virtue of that disposition, then
what matters for justifying my belief that Imφ-ing similarly seems to be whether
I am able to φreliably, not whether I so consider myself. Correspondingly, my
inability to φreliably not only undermines my claim that I was φ-ing intention-
ally, but also the justification of my belief that I did.
Davidsons insistence on confidence, on the other hand, is not easy to make
sense of. What subjective probability tracks, if things go well, is precisely ones
ability to bring about the intended action, that is, the objective likelihood of suc-
cess. In the overwhelming majority of cases where the tracking works well, it is a
very useful device. When it doesnt, it seems more appropriate to look to objec-
tive probability itself for justification rather than to its poorly calibrated proxy.
Knowledge in action, misconceived in a way in which said proxy does play
the central justificatory role, isin line with what Davidson holdsmost certain-
ly not a necessary condition for intentional action.
In sum, Davidsons scenario, despite its clever trading on disjoint subjective
and objective probabilities, does little to cast any doubt on Anscombes episte-
mology of action. An externalist account of justification makes the entailment
thesis not only possible, but is also considerably more plausible than an inter-
nalist approach to knowledge in action.
healed and have the usual (or similar) objective chances of clenching my fist, its not at all clear
why my pessimist attitude should be of any major epistemic consequence. The action is so sim-
ple that it is hard to believe how anyone with standard chances could not know that theyre
clenching their fist when they are, whether they are confident or not. Needless to say, if we
buy into the logic of these examples, well have to deny all people with anaturallypessimist
or skeptical disposition knowledge of much of what they are doing.
 Preliminary experimental results confirm that folkascriptions of justification and knowl-
edge do in fact correlate with objective probability and do not correlate with subjective proba-
bility (cf. Kneer, in prep.).
Success and Knowledge in action: Saving AnscombesAccountofIntentionality 151
Adams, Frederick and Mele Alfred. 1989. The role of intention in intentional action.
Canadian Journal of Philosophy 19(4): 511531.
Anscombe, G. E. M. 1963. Intention. Oxford: Blackwell Press, 2nd edition edition.
Armstrong, D. M. 1973. Belief, Truth and Knowledge. Cambridge: Cambridge University Press.
BonJour, Laurence. 1985. The Structure of Empirical Knowledge. Cambridge: Cambridge
University Press.
Braun, David. 2012. An Invariantist Theory of MightMight Be Right.Linguistics and
philosophy 35(6): 461 489.
Brogaard, Berit. 2008. Sea Battle Semantics. The Philosophical Quarterly 58(231): 326335.
Cappelen, Herman and Hawthorne John. 2009. Relativism and Monadic Truth. Oxford: Oxford
University Press.
Chisholm, Roderick M. 1966. Theory of Knowledge. Englewood Cliffs, (N.J): Prentice Hall.
Correia, Fabrice and Iacona Andrea. 2012. Around the Tree: Semantic and Metaphysical
Issues Concerning Branching and the Open Future,volume361.Springer.
Davidson, Donald. 1971. Agency.InRobertBinkley,RichardBronaughandAusonioMarras
(eds.), Agent, Action, and Reason. Toronto: University of Toronto Press.
Davidson, Donald. 1978. Intending. In Y. Yovel (ed.), Philosophy of History and Action.
Dordrecht: D. Reidel. 4160.
Davidson, Donald. 2001. Essays on Actions and Events. Oxford: Clarendon Press, 2nd edition.
Donnellan, Keith Sedgwick. 1963. Knowing What I Am Doing.The Journal of Philosophy 60:
Dretske, Fred. 1971. Conclusive reasons.Australasian Journal of Philosophy,49(1):122.
Dretske, Fred. 1981. Knowledge and the Flow of Information. Cambridge: MIT Press.
Egan, Andy. 2007. Epistemic Modals, Relativism and Assertion. Philosophical Studies,133(1):
Egan, Andy, Hawthorne John and Weatherson Brian. 2005. Epistemic Modals in Context.In
Gerhard Preyer and Georg Peter (eds.), Contextualism in Philosophy. Oxford: Oxford
University Press. 131 168.
Falvey, Kevin. 2000. Knowledge in Intention.Philosophical Studies 99(1): 2144.
Goldman, Alvin I. 1967. ACausalTheoryofKnowing.The Journal of Philosophy,64(12):
Grice, Herbert Paul. 1972. Intention and Uncertainty. Oxford: Oxford University Press.
Haddock, Adrian. 2011. The Knowledge That a Man Has of His Intentional Actions.InAnton
Ford, Jennifer Hornsby and Frederic Stoutland (eds.), Essays on AnscombesIntention.
Cambridge (MA): Harvard University Press.
Hawthorne, John. 2007. Eavesdroppers and Epistemic Modals.Philosophical Issues,17(1):
Kaplan, David. 1979. On the Logic of Demonstratives.Journal of Philosophical Logic,8(1):
81 98.
Kaplan, David. 1989. Afterthoughts.InJosephAlmog,JohnPerryandHowardWettstein
(eds.), Themes from Kaplan (565 614). Oxford: Oxford University Press.
Kim, Kihyeon. 1993. Internalism and Externalism in Epistemology.American Philosophical
Quarterly. 303316.
Kneer, Markus. 2015. Perspective in language. Paris: PhD Dissertation, EHESS ENS Paris.
152 Markus Kneer
Kneer, Markus. 2020. Predicates of personal taste, semantic incompleteness, and necessita-
rianism. Linguistics and Philosophy (online first).
Kneer, Markus. (in prep). Knowledge in action: Anscombe v. Davidson.
Langton, Rae. 2004. Intention as Faith.InJohnHymanandHelenSteward(eds.),Agency
and Action. Cambri dge: C ambridge Uni versity Pres s. 243 258.
Lehrer, Keith. 2000. Theory of knowledge. Boulder: Westview Press.
MacFarlane, John. 2003. Future Contingents and Relative Truth.The Philosophical Quarterly,
53(212): 321336.
MacFarlane, John. 2005. The Assessment Sensitivity of Knowledge Attributions.Oxford
Studies in Epistemology 1: 197233.
MacFarlane, John. 2008. Truth in the Garden of Forking Paths.InM.García-Carpinteroand
Max Kölbe (eds.), Relative Truth. Oxford: Oxford University Press. 81 102.
MacFarlane, John. 2011. Epistemic Modals are Assessment-sensitive.InAndyEganand
Brian Weatherson (Eds.), Epistemic Modality. Oxford: Oxford University Press. 1 44.
MacFarlane, John. 2014. Assessment Sensitivity: Relative Truth and its Applications. Oxford:
Oxford University Press.
McDowell, John. 2011. Anscombe on Bodily Self-knowledge.InAntonFord,Jennifer.
Hornsby and Frederick Stoutland (Eds.), Essays on AnscombesIntention.Princeton:
Princeton University Press. 128 146.
Mele, Alfred R. and Cushman Fiery. 2007. Intentional Action, Folk Judgments, and Stories:
Sorting Things Out.Midwest Studies in Philosophy 31(1): 184201.
Moran, Richard. 2004. Anscombe on Practical Knowledge’”.InJohnHymanandHelen
Steward (eds.), Agency and Action. New York: Cambridge University Press. 43 68.
Nozick, Robert. 1981. Philosophical Explanations. Cambridge: Harvard University Press.
OShaughnessy, Brian. 2003. The Epistemology of Physical Action.InJohannesRoessler
and Naomi Eilan (eds.), Agency and Self-Awareness. New York: Oxford University Press.
345 57.
Paul, Sarah, K. 2009. How We Know What Were Doing.PhilosophersImprint,9(11):124.
Pickard, Hanna. 2004. Knowledge of Action Without Observation.Proceedings of the
Aristotelian Society 104(3): 205230.
Schwenkler, John. 2011. Perception and Practical Knowledge.Philosophical Explorations
14(2): 137 152.
Schwenkler, John. 2012. Non-observational Knowledge of Action.Philosophy Compass
7(10): 731740 .
Setiya, Kieran. 2007. Reasons Without Rationalism. Princeton: Princeton University Press.
Setiya, Kieran. 2008. Practical Knowledge.Ethics,118(3):388409.
Setiya, Kieran. 2009. Practical Knowledge Revisited.Ethics 120(1): 128137.
Stephenson, Tamina. 2007. Judge Dependence, Epistemic Modals, and Predicates of
Personal Taste.Linguistics and Philosophy 30(4): 487525.
Thomason, Richmond H. 1970. Indeterminist Time and Truth-value Gaps.Theoria,36(3):
Thompson, Michael. 2008. Life and action. Cambridge (MA): Harvard University Press.
Thompson, Michael. 2011. AnscombesIntentionandPracticalKnowledge.InAntonFord,
Jennifer Hornsby and Frederick Stoutland (Eds.), Essays on AnscombesIntention.
Cambrid ge: Har vard Uni versity Press. 198 210.
Success and Knowledge in action: Saving AnscombesAccountofIntentionality 153
Velleman, David, J. 1989. Practical Reflection. Princeton: Princeton University Press.
Von Fintel, Kai and Gillies Anthony S. 2007. An Opinionated Guide to Epistemic Modality.
Oxford studies in epistemology,2:3262.
Von Fintel, Kai and Gillies Anthony S. 2008. Cia Leaks.Philosophical Review,117(1):
Williamson, Timothy. 2000. Knowledge and its Limits. Oxford: Oxford University Press.
Yalcin, Seth. 2011. Nonfactualism About Epistemic Modality.InAndyEganandBrian
Weatherson (Eds.), Epistemic Modality. Oxford: Oxford University Press. 295 332.
154 Markus Kneer
This paper reexamines the case for mentality — the thesis that knowledge is a mental state in its own right, and not only derivatively, simply by virtue of being composed out of mental states or by virtue of being a property of mental states — and explores a novel argument for it. I argue that a certain property singled out by psychologists and philosophers of cognitive science as distinctive of skillful behavior (agentive control) is best understood in terms of knowledge. While psychological theories of agentive control that appeal to monitoring mechanisms, such as attention, have been proposed, these theories cannot account for the full scope of controlled action. By contrast, I argue that an epistemic theory of agentive control that invokes knowledge is extensionally adequate. It is when it comes to understanding the hallmarks of skillful performance that the theoretical benefits of thinking of knowledge as mental can be fully appreciated.
Full-text available
The potential capacity for robots to deceive has received considerable attention recently. Many papers explore the technical possibility for a robot to engage in deception for beneficial purposes (e.g., in education or health). In this short experimental paper, I focus on a more paradigmatic case: robot lying (lying being the textbook example of deception) for nonbeneficial purposes as judged from the human point of view. More precisely, I present an empirical experiment that investigates the following three questions: (a) Are ordinary people willing to ascribe deceptive intentions to artificial agents? (b) Are they as willing to judge a robot lie as a lie as they would be when human agents engage in verbal deception? (c) Do people blame a lying artificial agent to the same extent as a lying human agent? The response to all three questions is a resounding yes. This, I argue, implies that robot deception and its normative consequences deserve considerably more attention than they presently receive.
Full-text available
The potential capacity for robots to deceive has received considerable attention recently. Many papers focus on the technical possibility for a robot to engage in deception for beneficial purposes (e.g. in education or health). In this short experimental paper, I focus on a more paradigmatic case: Robot lying (lying being the textbook example of deception) for nonbeneficial purposes as judged from the human point of view. More precisely, I present an empirical experiment with 399 participants which explores the following three questions: (i) Are ordinary people willing to ascribe intentions to deceive to artificial agents? (ii) Are they as willing to judge a robot lie as a lie as they would be when human agents engage in verbal deception? (iii) Do they blame a lying artificial agent to the same extent as a lying human agent? The response to all three questions is a resounding yes. This, I argue, implies that robot deception and its normative consequences deserve considerably more attention than it presently attracts.
Full-text available
According to indexical contextualism, the perspectival element of taste predicates and epistemic modals is part of the content expressed. According to nonindexicalism, the perspectival element (a standard of taste, an epistemic situation) must be conceived as a parameter in the circumstance of evaluation, which engenders "thin" or perspective-neutral semantic contents. Echoing Evans (1985), thin contents have frequently been criticized. It is doubtful whether such coarse-grained quasi-propositions can do any meaningful work as objects of propositional attitudes. In this paper, I assess recent responses by Recanati, Kölbel, Lasersohn and MacFarlane to the "incompleteness worry". None of them manages to convince. Particular attention is devoted to MacFarlane's (2014) argument, which states that if perspectives must be part of the content, so must worlds, which would make intuitively contingent propositions necessary. I demonstrate that this attempt to defend thin content views such as nonindexical contextualism and relativism conflates two distinct notions of necessity, and that radical indexicalist accounts of semantics, such as Schaffer's necessitarianism, are in fact quite plausible.
The book develops a conception of epistemology in which the notion of knowledge is explanatorily fundamental. It reverses the traditional programme of trying to analyse knowledge as a combination of truth, belief, and other factors, such as justification. Rather, belief is a state whose successful form is knowledge, and justification is on the basis of knowledge, which is acquainted with evidence. Knowing is as much a mental state as believing, but it is world‐involving because one can know only what is true; the book extends the externalist conception of mind from the contents of mental states to the attitudes to those contents. As with other mental states, one cannot always know whether one is in the state of knowing. It is argued that this is a special case of a much more general phenomenon; no non‐trivial conditions are such that one is always in a position to know that they obtain whenever they in fact do so. This result has disturbing implications for the nature of rationality, because one is not always in a position to know what it is rational to do. Traditional arguments for scepticism fail because they assume that one is always in a position to know what one's evidence is. The speech act of assertion is also governed by a norm of knowledge. A final chapter explores the limits on what can be known that are revealed by the so‐called paradox of knowability.
Over the past few years, the tree model of time has been widely employed to deal with issues concerning the semantics of tensed discourse. The thought that has motivated its adoption is that the most plausible way to make sense of indeterminism is to conceive of future possibilities as branches that depart from a common trunk, constituted by the past and the present. However, the thought still needs to be further articulated and defended, and several important questions remain open, such as the question of how actuality can be understood and formally represented in a branching framework. The present volume is intended to be a 360 degree reflection on the tree model. The contributions is gathers concern the model and its alternatives, both from a semantic and from a metaphysical point of view.
G.E.M. Anscombe famously claimed that acting intentionally entails knowing "without observation" what one is doing. Among those that have taken her claim seriously, an influential response has been to suppose that in order to explain this fact, we should conclude that intentions are a species of belief. This paper argues that there are good reasons to reject this "cognitivist" view of intention in favor of the view that intentions are distinctively practical attitudes that are not beliefs and do not constitutively involve the belief that one will do what one intends. A theory is then proposed on behalf of Distinctive Practical Attitude views of intention to explain Anscombe's non-observational knowledge phenomenon. It is argued that intentions do not embody non-observational knowledge, but they do provide the evidential basis for it: we know without observation what we are doing by inferring from our intentions.
A great deal of attention has been paid in recent years to the functional roles of intentions in intentional action. In this paper we sketch and defend a position on this issue while attacking a provocative alternative. Our position has its roots in a cybernetic theory of purposive behavior and is only part of the larger task of understanding all goal-directed behavior. Indeed, a unified model of goal-directed behavior, with appropriate modifications for different types of systems, is a long-range ambition.
Relativism has dominated many intellectual circles, past and present, but the 20th century saw it banished to the fringes of mainstream analytic philosophy. Of late, however, it is making something of a comeback within that loosely configured tradition, a comeback that attempts to capitalize on some important ideas in foundational semantics. This book aims not merely to combat analytic relativism but also to combat the foundational ideas in semantics that led to its revival. Doing so requires a proper understanding of the significance of possible worlds semantics, an examination of the relation between truth and the flow of time, an account of putatively relevant data from attitude and speech act reporting, and a careful treatment of various operators. This book contrasts relativism with a view according to which the contents of thought and talk are propositions that instantiate the fundamental monadic properties of truth simpliciter and falsity simpliciter. Such propositions, it argues, are the semantic values of sentences (relative to context), the objects of illocutionary acts, and, unsurprisingly, the objects of propositional attitudes. © Herman Cappelen and John Hawthorne 2009. All rights reserved.
Intuitively, the knowledge of one’s own intentional actions is different from the knowledge of actions of other sorts, including those of other people and unintentional actions of one’s own. But how are we to understand this phenomenon? Does it pertain to all actions, under every description under which they are known? If so, then how is this possible? If not, then how should we think about cases that are exceptions to this principle? This paper is a critical survey of recent attempts to answer these questions. I consider views under three headings: “special source” views, which hold that the knowledge of one’s intentional actions has a non‐perceptual source; “special domain” views, which hold that some but not all aspects of one’s intentional actions are known in this way; and “special character” views, which hold that the knowledge of intentional actions is special not because of where it comes from, but because of some other respect in which it is different in kind from the knowledge of other things.
L'A. defend la these des deux facteurs selon laquelle une personne sait ce qu'elle a l'intention de faire sans recours a l'observation, alors que la connaissance de ce qu'elle est en train de faire repose sur l'observation. Distinguant les notions d'expression et de description, de justification et de capacite, l'A. definit le role de l'auto-attribution dans le processus de la connaissance de l'intention.