Content uploaded by Kit Fine
All content in this area was uploaded by Kit Fine on Jun 01, 2019
Content may be subject to copyright.
Chisholm’s Puzzle and Unconditional Obligation
In this paper, I would like to suggest a new angle on Chisholm’s puzzle concerning
contrary-to-duty obligations. It differs from previous approaches in its conception of what the
problem is and how it is to be solved. I shall argue that the problem is, in no small part, about
unconditional, rather than conditional, obligation and that, once it is viewed in this light, we
obtain a somewhat different perspective on how the puzzle might be solved.1
We may state the original puzzle in the following way (Chisholm ):
(1) You ought to go to the meeting (premiss).
(2) If you go you ought to say you are going (premiss)
(3) If you do not go you ought not to say you are going (premiss)
(4) You do not go (premiss)
(5) You ought to say you are going (from (1) and (2))
(6) You ought not to say you are going (from (3) and (4))
(7) You ought to say you are going and not say you are going (from (5) and (6)).
It looks as if circumstances might be such that the premisses (1) - (4) of the argument are true
while the conclusion (7) is false. So something must have gone wrong with the reasoning of the
argument. But what?
Barring some failure in the ‘structural’ rules of inference or a failure in the inference from
the intermediate steps, at (5) and (6), to the conclusion, at (7), the puzzle seems to present us
with a choice between two principles of Detachment. There is, on the one hand, the principle of
Deontic Detachment, which entitles us to infer ‘Ought Q’ from ‘Ought P’ and ‘If P then Ought Q’
and which is exemplified by the inference from (1) and (2) to (5). There is, on the other hand,
the principle of Factual Detachment, which entitles us to infer ‘Ought Q’ from ‘P’ and ‘If P then
Ought Q’ and which is exemplified by the inference from (3) and (4) to (6). Viewed in this light,
the puzzle is a puzzle about how best to reconcile the two principles of Detachment, both of
which seem to be involved in ordinary deontic reasoning; and this, indeed, is how the import of
the puzzle is commonly construed.2
A similar puzzle puts pressure on Factual Detachment, quite apart from its conflict with
Deontic Detachment. For consider the following argument:
(1) You ought to go to the meeting (premiss). ʹ
(2) If you do not go you ought to apologize for not going (premiss)ʹ
(3) You do not go (premiss)ʹ
(4) You ought to apologize for not going (from (2) and (3) )ʹ ʹ ʹ
(5) You ought to go to the meeting and to apologize for not going.ʹ
1I originally gave this paper as a talk at the Ranch Metaphysics Meeting in Tucson,
Arizona, January 2018, and subsequently at the Jowett Society in Oxford, February 2019. I
should like to thank Jason Turner, Boris Klemt, Kris McDaniel, Sarah Moss, Jack Spencer and
other participants, and also Ruth Chang and Martin Pleitz, who read earlier drafts of the paper,
for many helpful comments.
2 As suggested, for example, in McNamara , ‘A Bit More of Chisholm’s Paradox’.
The conclusion (5) to the argument is still problematic in this case (though perhaps more on ʹ
normative than logical grounds)3; and so we can see the puzzle as arising without any appeal to
I believe we can go even further in “stripping down” the puzzle and can see it as arising
independently of either principle of Detachment. For let us suppose that Sheila previously
incurred an obligation to go to the meeting at noon, that she does not in fact go to the meeting,
and that nothing happens - unless it be the emergence of an obligation to apologize - that might
release her from the obligation. What then are her unconditional obligations at noon?
Not her obligations to do such and such if such and such conditions obtain, but her obligations
In setting up the case, it will be helpful to avoid certain irrelevant issues over the timing
of her actions. So let us also suppose that just before noon Sheila has her arms out-stretched,
with the index fingers of each hand lightly touching each of two buttons, G (for go) and A (for
apologize). Precisely at noon, when the clock strikes twelve, she can either veer to the left and
press button G and thereby be immediately whisked off to the meeting or else she can veer to the
right and press button A and thereby immediately send an apology - or she can sit still. We may
suppose, furthermore, that this is her only opportunity for doing these things and that, in fact, she
fails to press button G. Our question then is whether at noon she has an obligation to go to the
meeting (press button G) or an obligation to apologize (press button A) or an obligation to do
This, then, is the puzzle; for there appear to be great difficulties in providing a
satisfactory answer. There is, in the first place, a strong intuitive argument for supposing Sheila
to be under both obligations at the time in question. In the original case, it was supposed that she
did not go but it was left open whether or not she apologized. Suppose that in fact she
apologized. Then surely she has failed at noon to fulfil one of her obligations (to go at noon) and
yet succeeded in fulfilling another (to apologize at noon). She deserves blame for the one (not
going) and praise for the other (apologizing). Suppose, on the other hand, that she does not
apologize. Then surely, in this case, she has failed at noon to fulfil two of her obligations and
deserves blame on both counts. It therefore seems, in either case, that she falls under two
obligations at noon, an obligation to go and an obligation to apologize.
These intuitive considerations can be reinforced by further argument. For suppose she
does in fact go at noon and is therefore under no obligation to apologize and compare this to the
original case in which she does not go at noon. We would then have no hesitation in saying that
in the new case she is (at noon) under an obligation to go. But if she has the obligation in the
new case then why not in the original case? How could the failure to do what the obligation
would have required make the obligation disappear (if only it could!)?
Similarly in regard to the obligation to apologize. Suppose there is merely an
expectation, not an obligation, that she go to the meeting. Surely she might then still be under an
obligation (at noon) to apologize given that she does not go. But if in this case then why not in
the original case? Indeed, the presence of an obligation, rather than an expectation, to go should
strengthen the obligation to apologize.
In attempting to avoid these conclusions, one might take the actions (of going or of
apologizing) to relate to different times. But recall the set up is such that she can only perform
these actions at noon; and so we cannot sensibly take the obligations to relate to the performance
3I here use ‘apologize’ in a non-factive sense - so that one can apologize, or say ‘sorry’,
for what one has not done. The factive sense of ‘apologize’ raises issues which are irrelevant to
my present concerns.
of these actions at some other time. One might, alternatively, take the obligations themselves to
relate to different times. Thus it might be thought that she has an obligation up to, but not
including, noon to go at noon.
But surely we want to be able to say that she failed at noon to fulfil an obligation she was
under, viz. the obligation to go at noon. And how can this be so if she was no longer under the
obligation at noon? The chairperson might have released Sheila from the obligation to go to the
meeting at noon. But even if he releases her from the obligation exactly at noon then surely, by
failing to go to the meeting at noon, she has not in any blameworthy way failed to fulfil an
obligation. And so how can she, in the original case, have failed in some blameworthy way to
have fulfilled the obligation, given that she was no longer under the obligation?
The point becomes even clearer when put in terms of success in fulfilling an obligation.
For suppose that Sheila does in fact go to the meeting. Then if she was under no obligation, at
noon, to go to the meeting in case she did not in fact go, it is hard to see how she could come
under the obligation to go to the meeting in case she did in fact go. This, again, would mean that
she could remove an obligation she would otherwise have by failing to fulfil it.
But if in the case in which she does go to the meeting she is not under an obligation to go
at noon, but only just before, then how can she be said to have fulfilled the obligation to go? To
take an analogous case, suppose that Sheila is under an obligation to feed the dog and that this
obligation ceases exactly at noon (perhaps it then becomes someone else’s obligation). Then
feeding the dog at noon is not a way for her to fulfil her obligation. And no more, it would seem,
is going to the meeting should the obligation have lapsed.
Similarly in regard to the obligation to apologize. One might think that Sheila falls under
the obligation to apologize at noon, not at noon itself, but immediately after noon or perhaps some
time before. But both options are quite bizarre. Surely the obligation to apologize arise from the
fact that she did not go to the meeting at noon; and so how can she be under the obligation to
apologize even before she failed to go? And how can she now be responsible for fulfilling an
obligation which will only arise in the future? Indeed, once one grants that she is at noon under
an obligation to apologize, it is hard to see on what basis one might deny that she is at noon also
under an obligation to go - or vice versa. The point, quite simply, is that in either case one can
only succeed or fail in fulfilling an obligation at a given time if one is under the obligation at that
time; one cannot presently fulfil or violate an obligation that lies only in the past or in the future.
It therefore looks as if Sheila is both under an obligation to go and under an obligation to
apologize. But this would then seem to imply that she is under an obligation both to go to the
meeting and to apologize for not going; and yet surely, in the given case, this is not so. Now there
may indeed be cases of moral dilemmas - as perhaps with Sartre’s example of the son who is torn
between leaving home to avenge his brother and staying at home to look after his mother - in
which Agglomeration (the inference from OP and OQ to O(P & Q)) does not go through. Thus
even though the son is obliged to leave home to avenge his brother and also obliged to stay at
home to look after his mother, he is not, one might think, obliged both to leave home and to stay
But the present case does not appear to be a case of a moral dilemma. For one thing,
moral dilemmas arise from the impossibility of fulfilling two obligations. But in the present case
we may suppose that, by stretching her arms in both directions, Sheila is able both to go to the
meeting and to apologize.4 But nor is the present case even a case in which the obligations
derive from two independent systems of norms (such as duty to one’s country and duty to one’s
family) which cannot properly or readily be combined. If anything, it is the opposite, and it is the
same coherent system of norms which requires, in the given case, both that one go and that one
apologize for not going. There would therefore appear to be no good reason to reject
Agglomeration in the present case.
We might call the current version of Chisholm’s puzzle the objectual puzzle and the
previous two versions the inferential puzzle. Although they have a common basis in the existence
of ‘secondary’ obligations, which arise from the failure to fulfil certain ‘primary’ obligations, they
differ in a number of important respects.
In the first place, the inferential puzzle, as the name suggests, is concerned with the
validity of certain inferences, and with the validity, in particular, of Factual Detachment (and of
Deontic Detachment as well under the original version of the puzzle). The objectual puzzle, by
contrast, is concerned with how things stand and is not directly concerned with the validity of
certain inferences (except perhaps Agglomeration). In setting up the puzzle, we have appealed to
no detachment principles, we have simply asked for our judgements concerning a particular case.
Indeed, even if we questioned both detachment principles we might still be strongly inclined to
make the very same judgements about the case.
This is not, of course, to deny that an answer to the one puzzle might constrain, or make
more plausible, an answer to the other. Thus someone who rejects Factual Detachment as a
solution to the inferential puzzle might think, in response to the objectual puzzle, that Sheila is not
under an obligation to apologize, since one of the principal reasons for thinking this has been
removed. However, there is nothing in principle to prevent him from granting that Sheila is under
an obligation to apologize even though this does not follow from the application of Factual
Detachment. Similarly, someone who accepts Factual Detachment might think that Sheila is not
under an obligation to go to the meeting since this seems to be in normative conflict with her
being under an obligation to apologize. But, again, there is nothing in principle to prevent him
from accepting that Sheila is under both obligations.
In the second place, it is taken to be given, in the context of the inferential puzzle, that
Sheila ought to go to the meeting; this is not something that is meant to be questioned. But in the
context of the objectual puzzle, it need not taken for granted. We have supposed that Sheila
previously incurred an obligation to go the meeting and that there are no circumstances of a
normal sort that might release her from the obligation. But it then remains open whether she is
still under an obligation to go to the meeting, granted that she did not in fact go and is thereby
under an obligation to apologize. Thus one possible solution to the puzzle, though not one that I
4The original formulation of Chisholm’s puzzle is therefore less robust, in this one respect,
than the present formulation. For one might argue against the agglomerative inference from ‘you
ought to say you are going’ and ‘you ought not to say you are going’ on the grounds that it is
impossible to fulfil both obligations, whereas there need be no such reason for disputing the
agglomerative inference in the present case.
myself would recommend, is that the obligation to apologize somehow defeats the obligation to
In the third place, the focus of interest of the inferential puzzle is the conditional ought.
We wish to understand its inferential behavior and whether, in particular, it conforms to the
Detachment Principles. To this end, it is helpful to provide a semantics for conditional oughts,
through which our stand on the various inferential principles involved in the puzzle might then be
justified. The focus of interest of the objectual puzzle, by contrast, is with unconditional
obligation.5 We wish to know which unconditional obligations the agent is under in the given
case and whether, in particular, she is under both of the specific obligations, just one, or neither.
To this end, as we shall see, it is helpful to provide an account of what it is to have an
unconditional obligation, since it is through such an account that our stand on which
unconditional obligations she has might then be justified. Thus consideration of the two puzzles
leads in different directions, with the one puzzle leading to the development of a semantics for the
conditional ought and with the other leading to a metaphysical account of the nature of
More generally still, the inferential puzzle is concerned with the use of language, whether
this be the natural language in which the argument was originally formulated or a more formal
language in which the argument might be formalized. Our own puzzle, by contrast, is not
concerned with the language of obligation but with obligation itself.6 We have envisaged a
certain scenario and ask how things stand in that scenario. Of course, we use language in
providing an answer. But our answer is principally concerned with how things are in the given
scenario and not with the language by which we say how things are. We do not - or certainly are
not required to - take a linguistic turn.
I do not in any way wish to underestimate the interest or importance of the inferential
puzzle. The problems to which it gives rise are very real and very difficult; and no account of
5§2 of Thomason and Saint-Croix  present a version of the puzzle without
conditionals, but still involving inference rather than straight description.
6We might compare it, in this regard, with Donnellan’s take on Kripke’s puzzle concerning
belief. He writes:
‘Kripke makes use of a nest of principles having to do with our treatment of the believer’s
linguistic avowals of belief. I am inclined to protest that the puzzle really is a puzzle about belief,
about the psychological state itself, which at best manifests itself in one guise as a derivable
consequence of such principles.’(Donnellan , p. 275).
I am similarly inclined to hold that Chisholm’s puzzle is really a puzzle about obligation and not
about the ‘nest of principles’ by which the linguistic avowals of obligation might be derived.
The parallel with Kripke’s puzzle runs quite deep. In each puzzle, there is clear evidence
in favor of one belief (or obligation) and clear evidence in favor of another, even though it would
appear to be impossible to have both beliefs (or both obligations). Moreover, the considerations
in favor of the different premisses are somewhat analogous. Thus it is a principle of
‘agglomeration’ that prevents one from having both beliefs (or both obligations) and there are
arguments from the irrelevance of additional factors that work in both cases in favor of the subject
having each of the individual beliefs (or obligations).
deontic reasoning can be considered satisfactory until they are resolved. But despite the great
sophistication of the work in this area, I feel that the focus on the inferential puzzle has to a large
extent deflected attention away from the equally important and, in some ways, more fundamental
issues connected with the objectual puzzle. For, as I hope will become clear, it is only by settling
on what unconditional obligations the agent has in the puzzle case that we will be in a good
position to say whether or in what way the detachment principles might hold.
So how is the objectual puzzle to be solved? What I would like to suggest is that the
formulation of the puzzle rests on an improper presupposition and this is that we may adequately
state what her obligations are in terms of what she is obliged to do at a given time. We should,
instead, be describing her obligations in terms of what she is obliged to do in a given situation.
However, this line of response will call for considerable elaboration.
There is, in the first place, a significant ambiguity in what we understand by the ‘situation’
in which an agent acts or has an obligation. For whenever someone is called upon to act, we may
distinguish between the situation the agent is in when she acts and the situation she is in once she
has acted. The first, which we may call the choice or pre-choice situation, is one in which certain
things are taken to be fixed and others are taken to be open in the sense that nothing is specified in
regard to the question of whether they obtain. The second, which we may call the post-choice
situation, includes, in addition to the things which are taken to be fixed in the pre-choice situation,
those actions left open in that situation which she happens to have performed.
Although I have talked of a pre- and post-choice situation, I am not thinking of the pre-
choice situation as one which obtains before she acts and of the post-choice situation as one which
obtains after she acts but intend that each situation should obtain at the very same time. Thus
suppose Sheila presses the A button at noon. Then the pre-choice situation would consist of the
apparatus being as it is at noon etc., and the post-choice situation would consist of the
combination of the pre-choice situation and her act of pressing the A button at noon. The post-
choice situation is, in this way, a contemporaneous extension of the pre-choice situation.
An obligation relates most directly to one’s (pre)-choice situation. Given that the agent is
in a certain situation, we may ask, prospectively, how she ought to act in that situation. But once
an agent has acted in a certain way, we may retrospectively consider whether she ought to have
acted in this way - meaning by this, not the somewhat nonsensical thought of whether she ought
to have acted in this way when how she acted is already taken to be given, but the more sensible
thought of whether she ought to have acted in this way when how she in fact acted is taken to be
open. Thus the obligations one retrospectively has in the post-choice situation are simply those
one prospectively has in the corresponding pre-choice situation though this does not rule out, of
course, that one may regard the post-choice situation as itself a pre-choice situation, giving rise to
its own set of obligations.
We can now explain our differing intuitions about the puzzle case and why the conflict
between them is only apparent. For we are inclined to say that Sheila is, at noon, under an
obligation to go to the meeting since she is under that obligation in the choice situation prevailing
at noon in which it is left open whether she goes; and we are inclined to say that she is, at noon,
under an obligation to apologize since she is under that obligation in the choice situation
prevailing at noon in which it is given that she does not go. Thus there is no basis for supposing
there to be a single choice situation in which she falls under both obligations.
What may cause confusion on this point is that there is a sense in which in the situation in
which she does not go she is both under an obligation to go and an obligation to apologize. But it
is not an unequivocal sense. For she is under a prospective obligation to apologize in that
situation (treating it as pre-choice situation) while only under a retrospective obligation to go in
that situation (treating it as post-choice). There is therefore no real conflict between the two
obligations, which ultimately relate to different situations. But if one simply asks what she is
obliged to do in the situation in which she does not go, there is the danger that one answers the
question equivocally, treating it as pre-choice in regard to her apology and as post-choice in
regard to her going to the meeting.
There is also a sense - indeed, an unequivocal sense - in which, at the given time, Sheila
has an obligation to go and an obligation to apologize. For one may be said to have an obligation
at a given time in so far as one has an obligation in some situation at that time; and in this
unequivocal sense, it will be true that, at noon, Sheila has each of the obligations. But there is
then no good reason to suppose that Agglomeration will hold or that there will be a genuine
conflict between the two obligations since, as before, they will ultimately relate to different
situations. In a way, then, our treatment of the case is analogous to an approach to moral
dilemmas in which obligations are relativized to an imperative or norm (as in van Fraassen (,
§VI), with situations playing the same formal role as the norms. However, for us, the situations,
unlike the norms, will not be in competition for the same portion of normative space and hence
will not, in themselves, give rise to any form of moral conflict.
Although the present line of solution might appear to be relatively straightforward, I
believe that it runs counter to some deeply held principles and that it is only once these are
brought to light that the true import of the puzzle becomes clear.
For one might well think that, at any given time at which the agent is called upon to act,
the choice situation is given by the things that are fixed, or not open to the agent, at that time.
Call this ‘Fixity’. But this means that the situation in which Sheila does not go is not in fact a
choice situation. Since it is still open to her at that time to go, it is merely a post-choice situation;
and we are not therefore justified in taking there to be a choice situation in which she is under an
obligation to apologize.
Fixity has two consequences, which must also be given up. The first is that, at each time
at which the agent is called upon to act, the agent faces only one choice situation. It is this which
might have been taken to justify us in talking of what she ought to do at a given time, since what
the agent ought to do at a time is what she ought to do in the choice situation she faces at that
time.7 The other consequence is that, any given time at which the agent is called upon to act, the
choice situation consists only of things that are fixed for the agent at that time.
We must give up the first of these consequences since we have supposed that Sheila faces
two choice situations at noon, the one in which it is left open whether she goes, and the other in
which it is taken to be fixed that she goes. The second consequence must also be given up since
the second choice situation is one in which she does not go and, even though this has been taken
to be fixed at the time in question, it is not something that is fixed at the time in question.
7Chang (, 215, 223) can also be seen to allow there to be different choice situations
at a given time on account of a difference in the relevant circumstances, although the cases she
has in mind seem to be somewhat different from my own.
Thus the simple picture of choice in which the agent steers a path through the options
made available to her from one moment to the next must be replaced by one in which the options
made available to her at any given time may themselves fractured or layered, so that the choice of
one option at a given time can be seen to create another situation which at that very time is one in
which she is called upon to act. Under the simple picture, the choice of an option at one time
may, of course, create a new choice situation for the agent at a later time, since what was once
open to the agent has now become fixed. But this model of something becoming fixed through
time must now be taken to apply within a single moment of time. There is, so to speak, a nominal
passage of time that runs through the possible actions of the agent at a given moment, so that
within this nominal passage of time an action that was once open for her can, in this purely
nominal sense, subsequently become fixed.8 Moreover, the ‘layering’ of the agent’s actions at a
given time can itself submit to an iterative and branch-like structure. For perhaps there are
several primary obligations at a given time and different secondary obligations depending upon
which of them are fulfilled; and perhaps the failure to fulfil some of the secondary obligations can
lead to tertiary obligations; and so on through ever higher levels of derivative obligation.
Giving up Fixity also has significant consequences for the formulation of obligation
claims. For in stating what one is obliged to do, it is no longer sufficient in general to specify a
time, one must also specify the choice situation which is taken to prevail at the given time. We
should not simply say that Sheila ought to apologize at noon, but only that she ought to apologize
at noon in the situation in which she does not go the meeting.9
Even granted the relativity to a situation, there arises the question as to whether the
situation should be taken to qualify the obligation itself or the content of the obligation. Is
Sheila’s-apologizing-in-the-situation obligatory or is Sheila’s apologizing obligatory-in-the-
It seems clear in the case of time (and this is something we presupposed in formulating the
objectual puzzle) that we should admit both forms of relativity. Consider the case in which Sheila
promises at an earlier time t1 to go to the meeting at t2. She then incurs an obligation at t1 to go
the meeting at t2. The first of these temporal indices is external to the obligation, it qualifies the
obligation itself, while the second of them is internal to the obligation, it qualifies the content of
the obligation; and it seems that both kinds of index are required.
8This provides a mild form of vindication for those, such as Åqvist & Hoepelman ,
van Eck  and Loewer & Belzer , who have wanted to solve the puzzle by
distinguishing the different times at which the actions are performed. For this is indeed correct if
the times in question are taken to be nominal rather than real.
9There is a resemblance between my account and Humberstone’s (, p. 22), since he
also appeals to actual circumstances in accounting for an agent’s unconditional obligations.
However, there are some critical points of difference: (i) he explains unconditional obligation in
terms of conditional obligation (‘what a person ought to do, tout court, is what his conditional
obligations dictate, when they are taken as conditional upon the circumstances in which he finds
himself’ (p. 20)); (ii) he thinks of the circumstances for an agent as fixed once given a time (and a
world); and (iii) he takes the circumstances to be ‘those features of the situation which lie ...
outside of the agent’s control’ (p. 20).
It might be thought that we could gain the same effect in this case by using a double index.
Thus it might be said that she has an obligation at t1 by way of t2 to go to the meeting. However,
this strategy will not work for more complicated cases. Perhaps she has, at t1, an obligation either
to apologize at t2 or to go to another meeting at t3. In cases such as these, there appears to be no
reasonable way to externalize all the indices.
What goes for times, it seems to me, should also go for the situations. The situational
index should be permitted to qualify both the obligation and the content of the obligation. Thus it
may be that Sheila previously failed to apologize for not going to the meeting; and it might then
be thought that in that situation she then incurs the obligation to apologize especially profusely in
a situation in which she again fails to go to the meeting. Thus the one situation, in which she fails
to apologize, relates to the obligation, while the other situation, in which she later fails to go,
relates to the content of the obligation.
A further problem concerns the relationship between the temporal and situational indices.
Do we want to work with two separate indices or should we somehow incorporate them into a
single index, so that it is built into the identity of the situation, broadly construed, that it prevails
at a given time? My inclination on this matter is to let one index incorporate both the time and
the situation. For the situation itself, apart from the time, is of no interest; and should we only be
interested in the time, we can identify it with the minimal situation at a given time, in which
nothing in addition to what is actually fixed is taken to be fixed.10
A further problem still concerns the ‘depth’ of the choice situation. Consider again the
choice situation which consists of the circumstances which are fixed at the time Sheila makes her
choice plus the circumstance of her not going to the meeting. Do we also want to allow the
situations which consists, in addition, of further facts in the past? I think there is no harm in
allowing such situations but they serve no purpose. For the obligations Sheila has in the original
situation are the same as those that she has in any of its past-oriented extensions. Thus for the
purposes of determining her obligations, at least in so far as they pertain to a particular time, we
can simply suppose that the choice situation is merely present and has no tail that extends into the
It might be asked: given that Sheila may be in different choice situations at the given time
then how should she deliberate about what to do? I believe the answer to this question (once we
set aside the external costs of deliberation) is that the normative structure of her deliberations
should mirror the normative structure of her behavior. In the situation in which it is left open
whether she goes, she ought to go; likewise, in the situation in which it is left open whether she
decides to go, she should decide to go. In the situation in which she does not go, she ought to
apologize; likewise, in the situation in which she decides not to go, she should decide to
But still, does she not face a choice as to which choice situation she should take herself to
be in? Indeed, why should she not ‘cheat’ by taking herself to be in the situation in which it is
10I here ignore the complications which arise from taking future facts to be fixed. I should
also mention that I am inclined to treat acting in a given situation as itself an act (XXX). This
means that when one is singing in the rain, for example, one can distinguish the mere act of
singing, which happens to take place in the rain, from the act of singing in the rain. But this is a
separate matter and raises issues far removed from our present concerns.
taken to be fixed that she does not go to meeting and thereby avoid the burden of going to the
meeting? 11 The answer, I believe, is that this is not a case in which she can cheat. For she will
find herself in the minimal situation, whether she likes it not, and hence should act and deliberate
accordingly. It is therefore a mistake to think that she has complete choice as to which choice
situation to be in. She does have a choice as to the post-choice situations, for she may or may not
choose to fulfill her obligations, but she will have no choice over the initial pre-choice situation
and hence there will be no way for her to escape from her primary obligations.
There is a great temptation to think that in such cases there should be a notion of
obligation that simply applies at a time rather than to a situation. For when the moment to act
arises, do I not ask myself ‘what ought I to do now?’, not ‘what ought I to do in such and such a
situation?’ We may agree that this is the question I ask myself since it is taken to relate to the
minimal situation, i.e. to the situation one is in at the beginning of the deliberation. But should I
decide not to do the things - X, Y and Z - that I ought to do, the question can be raised anew
‘given that I have not done X, Y and Z, then what ought I now to do?’; And the ‘ought’ will now
relate to the updated situation. In each case, therefore, there is an implicit reference to an
Still, it has to be admitted that there is something odd about the logical or normative
structure of deliberation in such cases. Normally, when we construct models of deliberation or
reasoning, we presuppose that the agents reason as they should. But in the present cases we are
allowing for the possibility that they do not. In the case of reasoning about matters of fact this
would normally be regarded as disastrous; unless the mistake is corrected, there would be no
reason to have faith in the conclusion. But in the case of deliberation, we might well think that
there are better or worse ways to proceed in the light of a previous bad decision. Imagine that a
robot has been programmed to deliberate about what to do but that it is subject to malfunction.
We might then set up a secondary program that kicks in should the robot act on the basis of a
malfunction in the primary program or even a tertiary program should the secondary program
malfunction; and so on. It seems to me that it is something like this model of deliberation, in
which we explicitly allow for errors in the deliberative process, that must be seen to operate in the
It will be helpful to compare the proposed solution to Chisholm’s puzzle with some of the
other solutions to be found in the literature. The comparison is somewhat tricky since the
literature deals, for the most part, with the inferential puzzle, whereas my interest is in the
objectual puzzle. Still, a great deal that has been said about the one version of the puzzle can be
seen to have application to the other version.
11 Powers (, 396-7) has a related example. In discussing this example, Powers
writes, “An adequate solution of Chisholm's puzzle must contain an explanation precisely of the
fact that there appears to be more than one total relevant circumstance”, though I did not find it
clear whether he thinks that there merely appears to be more than one total relevant
circumstances, in which case the explanation should be of there appearing to be more than one
such circumstance rather than of there actually being more than one such circumstance..
12 We should imagine the deliberation taking place within the specious present. Or,
alternatively, we can suppose that it is rehearsed some time prior to the moment to act.
One point of difference is that much of the literature on the inferential puzzle proceeds by
providing an account of statements of obligation in terms of what is best.13 Thus on a crude
account of this sort, a proposition will be obligatory if it is true in the best possible worlds and a
proposition will be obligatory conditional upon an antecedent proposition if it is true in the best of
the possible worlds in which the antecedent is true. This may lead one to think that the objectual
puzzle can be solved in similar fashion by providing a description of the puzzle situation in terms
of what is best. Thus the best possibility is one in which Sheila goes to the meeting, the second
best is one in which she does not go but apologizes for not going, the third best is one in which
she does not go and does not apologize, and the fourth best, perhaps, is one in which she goes but
apologizes for not going.
But even if this provides a correct and even a full description of the puzzle situation, it
does not answer our original question, which is what, in the given situation, is Sheila
unconditionally obliged to do. Of course, one might try to relate the description in terms of what
is best to the desired description in terms of what ought to be done. But I doubt that much is to be
gained by following such a circuitous route - in regard either to the objectual or to the inferential
puzzle - and something is certainly lost, since the correctness of the account will now be hostage
to the principles connecting what ought to be done to what is best, principles which are
contentious at best and which would be regarded by those of an anti-consequentialist cast of mind
as thoroughly misguided.14
One apparent point of contact with the existing literature is that a number of philosophers,
in dealing with the inferential puzzle, have appealed to the idea of what is fixed and have thought,
in particular, that Factual Detachment will only hold when the antecedent of the ought conditional
is taken not merely to be the case but to be fixedly the case. A typical example is Carmo & Jones
.15 However, it seems to me that their notion of what it is to be fixed is very different from
my own. In a revealing passage, they write:
What do we mean when we say that for some reason or another a fact of the situation ... may be
deemed a fixed ... feature of that situation? Well, there are various ways in which this "fixity" might arise;
[there are temporal reasons] .... But temporal reasons, although very common, are not the only reasons
why things become fixed ... ; for instance, it is not for temporal reasons that the deed of killing, once done,
cannot be undone. What explains fixity in this case is not temporal necessity, but rather causal necessity.
Nor need temporal considerations have any role to play in explaining why the presence of a dog may be, to
all intents and purposes, an unalterable feature of the situation ... . From the practical point of view - from
the point of view of deciding which obligation actually applies to the situation- the key feature is that the
possibility of satisfaction of the requirement that there be no dog is effectively eliminated.’ (Carmo and
Jones , 283-4).
I believe that this passage embodies a fundamental error (relating to the second
consequence of Fixity). They seem to think that if it is not in virtue of a temporal feature of a
situation that a fact is fixed then it must be in virtue of some other feature of the situation - there
13As in Danielsson , Hansson , Lewis [1973, 1974], and Feldman , for
14This is another point of contact with Kripke’s puzzle. As he points out, ‘It is no solution
in itself to observe that some other terminology, which evades the question ... , may be sufficient
to state the relevant facts.’ (Kripke , p. 445).
15 See also Hansson (, §13) and Greenspan (1975], p. 265)
must be some kind of real practical impediment to the fact’s not obtaining. But this need not be
so. We may legitimately consider the question of whether Sheila ought to apologize in the
situation in which she does not go to the meeting without there being any further feature of the
situation which somehow renders it impossible that she go; she simply does not go.
True, it is very tempting to think that something intervenes between her not going and her
apologizing (or failing to apologize) that somehow renders it fixed that she does not go when she
apologizes (or fails to apologize). Perhaps she decides not to go or perhaps something in her
brain prevents her from going. But there need be no such thing. And there had better not be if
we are to be warranted in saying that she ought to go at the time in question, since this requires
that it not be then fixed that she does not go.
It is also true that the situation may be one in which her not going is ‘deemed’ to be fixed.
But it is not deemed to be fixed because it is somehow fixed. It is deemed to be fixed merely in
the sense that we evaluate what she is obliged to do in the situation in the same way as we would
if it were in fact fixed.
Of course, not all situations can properly be regarded as choice situations; they must be
appropriate extensions of the minimal situations in which the facts are genuinely fixed. But the
role of taking a fact to be fixed is not to specify some further feature of the situation but merely to
pick out that choice situation from among the others. There may, of course, be reasons why we
would want to pick out a particular situation rather than another; but these reasons relate to our
interest in the situation - to the further obligations to which it gives rise, for example - and not to
some further feature of the situation.16
The authors introduce a new modal operator □ to indicate that a fact is (actually) fixedly
the case. But if I am right, such an operator, at least in its intended application, is meaningless;
there is in general no fact which a statement of the form □A can properly be taken to record. This
is not to say that there is no use for such an operator in developing a system of deontic logic. But
its role will not be to record a further fact but to specify the choice situation of interest to us. It
does not describe what is fixed but fixes on the situation that is described.
There is another important point of contact with the existing literature. For many
philosophers have wanted to take the obligations in play in Chisholm’s puzzle to be relative to
different contexts or choice situations. For example: Jackson (, p. 180) takes the statement
that it is obligatory that P to be relative to some set of alternatives; Prakken & Sergot (, §5)
introduce a dyadic deontic operator O, where OAB has the meaning that B is obligatory given the
(possibly non-ideal) circumstances A; Saint-Croix & Thomason  evaluate a statement of
obligation OB relative to set of alternatives; and, in a somewhat different vein, McDaniel &
Brady  make desires, promises, intentions and the like conditional on a given proposition.
But, again, there are significant differences between my account of the relativity and
accounts such as theirs (quite apart from the fact that they work in a proposition- rather than an
action-oriented framework and identify the propositions in question with sets of possible worlds).
In the first place, most philosophers have taken the relativity to be external rather than internal to
the obligation; it is the obligation itself that is taken to be relative to the situation or context rather
16Ironically, the authors come close to admitting as much when they talk of the reasons for
deeming a fact to be fixed. But these reasons turn out in the end for them to be reasons for why
the fact itself is fixed.
than the content of the obligation.17 We, on the other hand, have also allowed the relativity to be
internal to the obligation. Indeed, as we have seen, this would appear to be necessary if we are to
deal with obligations whose content is disjunctive or logically complex; and it is a serious
shortcoming of the usual externalist approaches that they are not able to deal with cases of this
In the second place, the relativity of these other approaches is to something proposition-
rather than fact-like - to a possible rather than to an actual situation or to a set of possible rather
actual alternatives. For us, by contrast, the choice situation is always actual, one that actually
obtains. Thus in the puzzle case in which Sheila does not go to the meeting, we can say that she
is obliged to apologize in the ensuing situation but we say nothing of her obligations under the
hypothetical scenario in which she does go to the meeting. We therefore work with a factive
rather than a hypothetical conception of the situation in which an obligation is taken to obtain.18
This difference may appear slight, but it has a number of significant consequences. First,
working exclusively with the hypothetical conception of a situation or context does not enable us
to solve the objectual puzzle. For even if I know that Sheila is under an obligation to go in the
hypothetical choice situation in which it remains open whether she goes and that she is under an
obligation to apologize in the hypothetical choice situation in which she does not go, this still
does not tell us what her actual obligations are.
Now it might be thought that this further question could be easily answered by saying
which of the hypothetical choice situations actually prevail at the given time. But this further
question is not straightforward and raises the very issues which we have already had to confront
in dealing with the objectual puzzle. For how can two choice situations prevail at the very same
time? And how can a choice situation prevail at a given time when the facts within it are not all
fixed? This is not, of course, an objection to the very idea of a hypothetical choice situation. It is
merely to say that appeal to such an idea leaves unanswered questions that are crucial to the
solution of the puzzle .
Second, appeal to a hypothetical choice situation is problematic in a way in which appeal
to an actual choice situation is not. If we specify an obligation as one that holds in an actual
situation then it is relatively clear what is involved, since the world itself will tell us what other
circumstances to take into account - that Sheila, for example, has no aversion to pressing buttons,
is of sound frame of mind, has not forgotten about the meeting. But if we specify an obligation as
one that holds in a merely hypothetical situation then it is not so clear what these other
circumstances might be, since it is no longer so clear which are the relevant worlds in which the
hypothetical situation is meant to obtain.
Again, this is not an objection to the very idea of a hypothetical choice situation. The
point rather is that the appeal to such an idea raises considerations irrelevant to the issue at hand.
17The underlying semantical framework may have encouraged this point of view since if
the choice situation is identified with a set of worlds, i.e. with the alternatives that the situation
leaves open, then this is naturally taken to qualify the obligation rather than its content.
18Again, the underlying semantical framework may have encouraged the alternative point
of view, since if we think of the proposition expressed by the antecedent of a conditional
obligation as helping to fix that to which the obligation in the consequent is relative then it will do
so regardless of whether it is true or false.
For even though we might have an interest in what ought to be done under certain hypothetical
circumstances, such considerations are not required to make sense of secondary obligations,
which can be seen to arise in actual circumstances without any regard to what they might be in
any given counterfactual circumstance.
But most important of all, the hypothetical account fails to address the question of how
conditional obligations are capable of making a claim on our behavior. Take an ordinary
conditional obligation, such as ‘if it rains then you ought to take an umbrella’. If it is not raining,
the conditional obligation is ‘inoperative’ and makes no claim on our behavior while, if it is
raining, the conditional obligation becomes ‘operative’ and does make a claim on our behavior. It
is plausible to suppose that a conditional obligation becomes operative by virtue of giving rise to
an unconditional obligation. If it were conditional obligations “all the way down” then we would
have no idea of how such an obligation could ever make a claim on our behavior (just as we
would have no idea how dispositions could impact on the world if it were dispositions “all the
Now the obligation to do something in a hypothetical situation is, in effect, a conditional
obligation. It is operative only when we are actually in the hypothetical situation. The question
therefore arises as to how the conditional obligation becomes operative. To what unconditional
obligation does it give rise? To this further question, the hypothetical account appears to provide
One might think, again, that the answer is obvious. The conditional obligation becomes
operative through the hypothetical situation becoming actual. But this is not to answer the
question. For the thought is that the hypothetical situation becoming actual serves merely as a
trigger. It is through the operation of this trigger on a conditional obligation that some other,
unconditional, obligation comes into existence - or, at least, if not an unconditional obligation,
then some other conditional obligation, which might in its turn be triggered, until eventually we
arrive at unconditional obligation. And yet we have not been told what this unconditional
obligation might be.
Or again, it might be thought that the unconditional obligation is simply the obligation to
do something in the ‘null’ situation, in which the antecedent condition is automatically satisfied.19
But this still leaves us with the problem that there appear to be two unconditional obligations in
the puzzle situation, an obligation to go and an obligation to apologize. They cannot both very
well be an obligation to do something in the null situation.
This problem is avoided on our own account since an obligation to do something in an
actual situation is no more a conditional obligation than is the obligation to do something at a
given time. In neither case is there any further condition that must be satisfied in order for the
obligation to become operative. Thus rather than understanding an obligation to do something in
an actual situation as a conditional obligation whose antecedent condition is satisfied or as a
degenerate form of a conditional obligation, we understand it as a form of unconditional
obligation in its own right; and the problem of explaining how a conditional obligation can
become operative is removed.
19This would correspond to the standard view in which the unconditional obligation of B is
‘reduced’ to the conditional obligation of B on the trivial proposition ⊤.
It is essential, if the present view is properly to be understood, to distinguish between the
categories of conditional, contingent and relative obligation. A conditional obligation is one
which is conditional upon the circumstances; it is only becomes operative in those circumstances.
Thus Sheila’s obligation to apologize if she fails to go to the meeting only becomes operative if
she fails to go to the meeting. A contingent obligation is contingent upon the circumstances; it
only exists in those circumstances. Thus Sheila’s obligation to go to the meeting is perhaps
contingent upon her having made some kind of commitment to go to the meeting. A relative
obligation is one which is relative to the circumstances or situation; the identity of the obligation,
what it is, depends upon those circumstances. Thus Sheila’s obligation to apologize in the
envisaged scenario just is the obligation to apologize in the situation in which she has failed to go
to the meeting.
Once these distinctions are made, it becomes clear that they are largely independent of one
another. Thus Sheila’s obligation to apologize if she fails to go to the meeting only becomes
operative if she fails to go to the meeting but it exists regardless of whether or not she goes to the
meeting and so is not contingent upon her failing to go to the meeting. Thus one can have
conditionality without contingency (or relativity for that matter). Again, Sheila’s obligation to go
to the meeting is perhaps contingent upon her having made some kind of commitment to go to the
meeting but its identity does not depend upon her having made such a commitment. The
obligation simply is the obligation to go to the meeting, not the obligation to go to the meeting in
a situation in which she has made some kind of commitment to go to the meeting. Nor is her
obligation to go to the meeting conditional upon her having made such a commitment; the
obligation is already operative, so to speak, and nothing more is required for the obligation to
become operative. Thus one can have contingency without conditionality or relativity. Or again,
Sheila’s obligation to apologize in the situation in which she does not go to the meeting is not
conditional upon her not going since, as with Sheila’s obligation to go to the meeting, the
obligation is one that is already operative. In this case, however, there is kind of contingency on
the situation, since the obligation could not even be the obligation that it is without the situation.
Thus one can have relativity without conditionality and without, at least, the normal non-identity-
based forms of contingency
I suspect that the failure to recognize some of these distinctions may have been partly
responsible for what, from my point of view, are certain errors or oversights in the existing
literature. For if we assimilate the relativity of an obligation to a situation to its contingency upon
that situation, we lose the distinctive way in which an unconditional obligation may be relative to
a situation and therefore the whole foundation upon which a proper view of unconditional
obligation should rest. If we suppose, moreover, that an obligation can only be contingent upon
the prevailing situation through the situation being objectively fixed, then we shall find ourselves
in the position of Carmo and Jones, looking in vain for some way in which situation pertaining to
an obligation might actually be fixed. And if we assimilate the contingency of an obligation on
the circumstances to its conditionality upon those circumstances, then we lose the very distinction
between conditional and unconditional obligation.
I turn finally to the question of how, in the light of the previous solution to the objectual
puzzle, we should solve the two inferential puzzles.
A straightforward answer to the second of them suggests itself. To say, in the context of
that puzzle, that Sheila has an obligation to apologize conditional upon her not going is to say:
(*) If she does not go then the (minimal) situation in which she does not go will be one in
which she has an obligation to apologize.
This is a familiar line of thought: the situation described in the antecedent is carried over
to the consequent. However, our account of what is carried over and how is very different from
the usual accounts. For us, the situation in which she does not go is an actual not a hypothetical
situation. We are not using the antecedent of the conditional to pick out a hypothetical situation
and then using the consequent to say that she ought to go in that hypothetical situation. Rather the
description ‘the situation in which she does not go’ is being used to pick out an actual situation. If
the antecedent is true then the description will indeed pick out that situation. But if the antecedent
is false then the description will fail to pick out any situation and nothing, in contrast to the
standard accounts, will be carried over to the consequent. This becomes clearer if we eliminate
the description à la Russell:
(**) If Sheila does not go then there is a single (minimal) situation s in which she does not
go and which is such that she is obliged to apologize in s.
When the antecedent is false, there will then be no value of the variable ‘s’ for which the
consequent is true. The embedded statement of obligation ‘she is obliged to apologize in s’ in the
consequent is indeed an unconditional statement of obligation, but indexed to the variable ‘s’
rather than to a term for a particular situation.20
The conditional itself could, of course, be taken in a number of ways. But for the
purposes of the puzzle, it would seem that nothing essential is lost by taking it to be the material
conditional. Thus the conditional premiss now reads:
(***) Sheila does not go (there is a single situation s in which she does not go and ⊃
which is such that it is obligatory for Sheila to apologize in s).
The philosopher who adopts a hypothetical conception of the situation that is carried over cannot
provide a reductive account of conditional obligation in this way since, for him, the consequent
will already have the form of a conditional obligation.
However, in providing a solution to the inferential puzzle, philosophers have usually
wanted to regiment the argument within something like the standard symbolism of deontic logic
or within some standardized version of ordinary language. This requirement is not satisfied by
our proposed solution, since explicit reference is made to situations in the consequents of the
conditionals in (*), (**) and (***).
I would like to suggest that we meet this requirement by adopting the framework of truth-
maker semantics. Within such a framework, the explicit reference to situations in the conditionals
can be replaced by implicit reference to them within the semantics. Thus, in evaluating the
conditional, any situation which verifies the antecedent can be ‘silently’ carried over to the
consequent and treated as a situation relative to which any obligation stated in the consequent is to
However, this brings another problem in its train. Factual Detachment will no longer
hold! For upon detaching the consequent ‘Sheila ought to apologize’, it will no longer be moored
20There remains the question as to whether the situation s qualifies the obligation or the
content of the obligation. Presumably, in this case, it should be taken to qualify both.
to the antecedent. Thus what we will in effect be concluding, in making an application of Factual
Detachment, is that she ought to apologize in the minimal situation, not in the situation in which
she does not go.
I believe we can rectify this problem by taking the minor premiss ‘Sheila does not go’ in
the inference not only to state that Sheila does not go but to fix the situation as one in which she
does not go. Her subsequent obligation to apologize will then indeed be relative to this situation.
But it is important to bear in mind that in taking the situation to be fixed, we are not saying it is
fixed, but merely treating it as fixed.
What of the original inferential puzzle, which requires us also to accommodate Deontic
Detachment? Unfortunately, the present solution in terms of the material conditional will no
longer do. For suppose Sheila does not go and consider the conditional ‘if she goes then she
ought to destroy the world’. This conditional is true under the material construal of the
conditional since the antecedent is false. It is also true, we have supposed, that Sheila ought to
go; and so it follows by Deontic Detachment that she ought to destroy the world!
I believe we can rectify this problem by taking the conditional to concern hypothetical
situations. Thus in saying that Sheila ought to destroy the world if she goes we are saying that she
ought to destroy the world in any hypothetical situation in which she goes, which presumably is
false. This might appear to make obligations relative to hypothetical situations, which is
something we have wanted to avoid. But we can interpret the new conditional as a counterfactual
conditional in which reference to actual situations is made within the scope of the counterfactual
operator. Thus what we are in effect saying is: if it were to be the case that there is an actual
situation in which she goes then it would be the case that she ought to destroy the world in that
Deontic Detachment can then be seen to hold. For it is plausible, given that someone
ought to do X in a given situation and that they ought to do Y in the extension of the given
situation in which they do X, that they ought to do Y in the given situation. There is therefore a
critical difference between the present case and the case of contrary-to-duty obligations since, in
the latter case, no retraction of the situation in which the obligation is taken to hold is generally
permissible. It is for this reason that it is secondary obligations conditional upon one’s failing to
meet one’s primary obligations, rather than secondary obligations conditional upon one’s
succeeding in meeting one’s primary obligations, that require one to differentiate the choice
situations that may hold within a given time.
Although we can validate the two forms of Detachment in this way, each assigns a very
different role to the minor premiss. With Factual Detachment, the role of the minor factual
premiss, as we have seen, is not simply to state how things are but to fix how they are; and, for
this reason, the extended situation referenced in the conclusion of the inference will be the same
as the situation referenced in the consequent of the conditional premiss. With Deontic
Detachment, by contrast, the role of the minor deontic premiss is merely to state how things are;
and for this reason, the unextended situation referenced in the conclusion will be the same as the
situation referenced in the minor premiss. Chisholm’s puzzle is avoided, since once the premiss
‘Sheila does not go’ has been used to fix the situation to one in which Sheila does not go, we are
no longer justified in re-affirming that she ought to go.
To some extent, the present line of solution could be accommodated within other
frameworks.21 But I believe the focus on situations provides a number of additional benefits. On
the philosophical front, it makes clear the foundation for all talk of obligation in terms of
unconditional obligation, where this is now taken to be indexed to actual rather than to
hypothetical situations; and, on the more technical front, it enables us to employ the resources of
truth-maker semantics in giving an account of the semantics and logic of conditional and
unconditional obligations that relates them directly to the situations in which they are properly
said to obtain.22
Åqvist, L. & Hoepelman J.  ‘Some Theorems About a ‘Tree’ System of Deontic Tense
Logic’, in Hilpinen , pp. 187–221.
Carmo, J. & Jones A.,  ‘Deontic Logic and Contrary-to-Duties’, in D. M Gabbay and F.
Guenthner , 265–343.
Chang R.,  ‘Comparativism: the Grounds of Rational Choice’, in ‘Weighing Reasons’ (eds.
E. Cord and B. McGuire), Oxford: Oxford University Press.
Chisholm, R. M.  ‘Contrary-to-Duty Imperatives and Deontic Logic’, Analysis 24, 33–36.
Danielsson, S.  ‘Preference and Obligation’, Studies in the Logic of Ethics. Uppsala:
Donnellan K.,  ‘Belief and the Identity of Reference’, Midwest Studies in Philosophy 14,
Feldman, F.  ‘Doing the Best We Can : An Essay in Informal Deontic Logic’, Dordrecht,
Boston: D. Reidel Publishing Company.
Fine K.,  ‘Command and Compliance: I & II’, Review of Symbolic Logic, vol 11.4, 609-33
Fine K.,  ‘Act and Embodiment’, to appear in ‘Beings and Doings’ (eds. Alec
Hinshelwood), Oxford University Press.
Gabbay, D.M. & Guenthner, F. (eds.)  ‘Handbook of Philosophical Logic, Volume VIII’,
Kluwer Academic Publishers, Amsterdam.
Greenspan, P. S.  ‘Conditional Oughts and Hypothetical Imperatives’, Journal of
Philosophy, 72: 259–276.
Hansson, Bengt  ‘An Analysis of Some Deontic Logics’, Noûs 3, pp. 373–398.
Hilpinen R., (ed.)  ‘New Studies in Deontic Logic’, Dordrecht: D. Reidel.
Humberstone I. L.,  'The Background of Circumstances', Pacific Philosophical Quarterly,
Jackson F.  ‘On the Semantics and Logic of Obligation’, Mind, 94: 177–195.
Kripke S.A.  ‘A Puzzle about Belief’, in Margalit A. (eds) ‘Meaning and Use’, Synthese
21 The definition of contextualized satisfaction in Saint Croix & Thomason (, §4.2),
for example, enables a factual claim to restrict the subsequent evaluation of deontic claims.
22An application of truthmaker semantics to unconditional commands and obligation can
be found in the two parts of the author’s . The application to conditional commands and
obligation will be given in the third part of the paper.
Language Library (Texts and Studies in Linguistics and Philosophy), vol 3. Springer,
Lewis, D. K.  ‘Counterfactuals’, Oxford: Blackwell.
Lewis, D. K.  Semantic Analyses for Dyadic Deontic Logic’, in Stenlund  1–14.
Loewer B. & Belzer M.  ‘Dyadic Deontic Detachment’, Synthese 54, pp. 295–318.
McDaniel K., & Bradley B.,  ‘Desires’, Mind, v. 177, 267-302.
McNamara, P.  ‘Deontic Logic’, Entry on ‘Deontic Logic’ in Stanford Encyclopedia of
Powers L.  ‘Some Deontic Logicians’, Noûs 1(4): 381-400.
Prakken, H. and Sergot M.  ‘Contrary-to-Duty Obligations’, Studia Logica, 57(1): 91–115.
Saint Croix C., Thomason R.H.  ‘Chisholm’s Paradox and Conditional Oughts’, in Cariani
F., Grossi D., Meheus J., Parent X. (eds) ‘Deontic Logic and Normative Systems’, DEON
2014, Lecture Notes in Computer Science, vol 8554, Springer, Cham
Saint-Croix C. & Thomason R. H.  ‘Chisholm's Paradox and Conditional Oughts’,
International Conference on Deontic Logic in Computer ScienceDEON 2014: Deontic
Logic and Normative Systems pp 192-207 Journal of Logic and Computation, exw003,
Stenlund, S. (ed.)  ‘ Logical Theory and Semantic Analysis: Essays Dedicated to Stig
Kanger on His Fiftieth Birthday’, Dordrecht: D. Reidel.
van Eck, J.  ‘A System of Temporally Relative Modal and Deontic Predicate Logic and Its
Philosophical Applications’, Logique et Analyse, 25: 339–381.
van Fraassen, B. C.  ‘Values and the Heart’s Command’, Journal of Philosophy 70(1): 5-19