Impossible Ideals and the `Ought Implies Can`

Impossible Ideals and the ‘Ought Implies Can’ Principle
Lucas Thorpe – Bogazici University
Kant believes that ‘ought implies can’. And in this paper I will examine how this
principle applies both to what we should do and to what we should be. I will be particularly
concerned with how this principle fits together with Kant’s belief that we have a duty to
instantiate certain ideals, which are in some sense impossible for us to instantiate.
Individually we have a duty to be morally perfect (which Kant often refers to as the duty to
“be holy”). Collectively we have a duty to instantiate the ideal of a world of republics living
together in perpetual peace. Now, given human nature, Kant believes that both of these ideals
are in actual fact impossible for being like us to achieve. So how, if at all, are we to apply the
‘ought implies can’ principles to ideals? One argument would be that such a principle should
only be applied to actions and not to ideals, and I believe that this partially captures Kant’s
attitude, for the ought implies can principle does seem to be a principle that is primarily meant
to be applied to individual actions rather than to, say, one’s whole character. For example, in
the case of particular actions: although is conceivable that someone could teleport themselves
to Tehran the fact that it is physically impossible for someone to do so means that they cannot
have a duty to do so. However, even in the case of actions I will argue that there are good
reasons to limit the application of the ought implies can principle. And I will argue that a
Kantian should distinguish between what we actually ought to do, which is governed by the
ought implies can principle, and what we ideally ought to do, which is governed by a weaker
principle: what I ideally ought to do is limited not by can do, but what I could have been able
to do. When it comes to ideals, however, things are not so obvious. Kant, for example,
believes that we have a duty to be the sort of person who always does the right thing and does
it gladly. This is an ideal. Now, given human nature, for example the fact that we are beings
with needs that sometimes conflict with what we recognize to be the right thing to do, it is
impossible for us to fully realize this ideal, this does not, however imply that we cannot have
a duty to instantiate it.
So how are we to apply the ‘ought implies can’ principle to ideals? Merely logical
possibility is not enough for something to serve as an ideal, the idea of an intuitive intellect
plays an important role in Kant’s theoretical philosophy. This is the idea of an intellect that
knows the world as it is in itself, that has a type of knowledge that is not perspective or
limited. It is the idea of a God’s eye perspective on the universe. Kant believes that there is no
contradiction involved in this idea. We do not, and cannot, however, have a duty to instantiate
this ideal. The reason for this is that although the idea is not self contradictory we have no
positive conception of what it would be like for beings like us to instantiate such an ideal. As
such it cannot function as a yardstick, for we can have no way of knowing whether we are
approaching this idea or not. Conceivability alone, however, is not enough. There are ideas
that are conceivable that cannot function as ideals for beings like us. For example, we can
conceive of a bat or an electron, but it is difficult to make sense of someone who tells us that
their ideal is to be a bat or an electron; I really don’t know what it is like to want to be a bat or
an electron. Kant thinks, however, that we ought to be morally perfect (what Kant calls
having a holy will) and to institute perpetual peace. What is the difference between the ideas
of an intuitive intellect and the ideas of a bat or an electron, which cannot function as practical
ideals, and the ideas of moral perfection and perpetual peace which can?1 One claim is that for
an idea to function as an ideal we must have a conception of what it would be like for flesh
and blood human beings to instantiate the ideal. Although it need not be actually possible to
achieve this. The reason for this is that we can only use an idea as a yardstick to measure our
actions if we have a conception of what is would be like for beings like us to instantiate the
Before examining four specific ideals, let me say a few words about ideals in general.
In Kantian terminology an ‘ideal’ is the idea of an individual, and the moral ideal is a
pure ideal. In his ethics lectures Kant explains that,
Here I disagree with Robert Hanna who argues that, “the theoretical reason of a divine cognizer or “intellectual
intuition” (CPR B72), is (barely) conceivable by us; and such a being would know noumenal objects or thingsin-themselves directly and infallibly by thinking alone. Similarly, the practical reason of a divine agent or “holy
will” (GMM 4: 439) – which is a noumenal subject or person, not a noumenal object or thing – is also (again,
barely) conceivable by us” Kant, Science, and Nature, Oxford Uni9versity Press, 2006, p.19
[T]o expound morality in its full purity is to set forth an Ideal of practical
reason. Such ideas are not chimeras, for they constitute the guideline to which
we must constantly approach. . . We have to possess a yardstick by which to
estimate our moral worth, and know the degree to which we are faulty and
deficient. . . An ideal is the representation of a single thing, in which we depict
such an idea to ourself in concreto. All ideals are fictions. We attempt, in
concreto, to envisage a being that is congruent with the idea. In the ideal we
turn the ideas into a model, and may go astray in clinging to an ideal, since it
can often be defective. . . The ideal is a prototypon of morality.” (29:604-5)2
Now Kant believes that our moral and political ideals must not accommodate
themselves to human weakness. Thus he argues in his ethics lectures that “[t]he moral law . . .
must not be lenient and accommodate itself to human weakness; for it contains the norm of
moral perfection. . . [S]ince ethics also propounds rules, which are meant to be the guidelines
for our actions, they must not be adjusted to human capacity, but have to show what is
morally necessary.” (Ethik Collins, 27:301)3
Now, it may look like Kant is setting himself up for Hegel’s criticism of those whose
will is indeterminate and who will absolute abstraction or universality. Hegel argues that such
people ultimately will nothing (determinate) and so their actions can only be destructive. Thus
Hegel explains,
This is the freedom of the void . . it becomes in the realm of both politics
and religion the fanaticism of destruction, demolishing the whole existing
social order. . It may well believe that it wills some positive condition, for
instance the condition of universal equality . . . but it does not in fact will
the positive actuality of this condition, for this at once gives rise to some
sort of order, a particularization both of institutions and of individuals; but
He makes a similar claim in is ethics lectures where he argues that, “The principle we draw from the weakness
of human nature is this: moral laws must never be laid down in accordance with human weakness, but are to be
presented as holy, pure and morally perfect, be the nature of man what it may. . . the moral law is the archetype,
the yardstick and the pattern of our actions. But the pattern must be exact and precise. . . The highest duty is
therefore to present the moral law in all its purity and holiness, just as the greatest crime is to subtract anything
from its purity” (Ethik Collins, 27:294).
Following Rousseau, Kant criticizes the notion of imitating other particular human beings, however moral, and
constantly stressed the fact that our moral ideal must be pure. “The one and only comparison allowable here is
the relation of his conduct to the moral law, which in respect of its definition is identical with humanity and the
Idea thereof: it is rendered practical, if we conceive thereunder a person adequate to the idea, or an ideal, just as
Christ, for example, is presented to us as an ideal” (27:610).
it is precisely through the annihilation of particularity and of objective
determination that the self consciousness of this negative freedom arises.4
Hegel’s criticism here is that to will a completely abstract ideal is equivalent to willing
nothing. I think that Hegel’s point is valid, for if we have no conception of how our ideal
relates to the here and now, what Kant would call the phenomenal world, to will it would
amount to willing nothing. Hegel’s argument, however, does not apply to Kant. For Kant
insists that our ideals, even if they are in a sense impossible to realize, must be concrete.
Thus in his essay Theory and Practice, he explains that “[I]t would not be a duty to
pursue a certain effect of our will, if this effect were not also possible in our experience
(whether it be thought of as completed or as always approaching completion), and this is the
only kind of theory that is at issue in the present essay” (8:277) I think that this captures
Kant’s conception of how the ought implies can principle applies to ideals. We have to have
some conception of how our ideals relate to our here and now existence, and so they must be
capable of having some sort of empirical content. But in order to function as ideals they do
not need to be actually realizable, they merely need to be approachable. We need to be able to
judge whether we are approaching them or becoming further away.
Intuitively, willing something impossible does not seem to be necessarily destructive,
futile or irrational. For, example, as Nicholas Rescher has argued, “it can make good rational
sense for someone to adopt an unattainable goal, pursuing an objective whose non-realization
is a foregone conclusion”5 For example, “perhaps only by striving for a perfect performance
is the performer (a violin soloist, say, or a figure skater) able to do as well as he can (flawed
though that performance will inevitably be)”.6 There is a rational for this, based upon
considerations to do with the nature of human motivation, nicely illustrated by Machiavelli in
G.W.F Hegel, Element of the Philosophy of Right, edited by Allen W. Wood, Cambridge University Press,
1991, p.38
Nicholas Rescher Ethical Idealism: An Inquiry into the Nature and Function of Ideals, University of California
Press, 1987, p.5.
Ibid. p.12
The Prince. Machiavelli explains that, “the prudent man . . . should proceed like those prudent
archers who aware of the strength of their bow when the target they are aiming at seems too
distant, set their sights much higher than the designated target.” (The Prince, trans. Peter
Bondanella and Mark Musa, Oxford University Press, 1984, p.20). This motivational claim
can be used to support a counterargument to the Hegelian though that willing moral or
political perfection can only be destructive. The violinist aiming at the perfect performance
wills something determinate even if impossible.
Kant himself is also motivated by such beliefs about human motivation believing that
there is a clear motivational difference between choosing to be perfect and choosing to be as
perfect as one can be. Of course, the most we can hope for is to be as perfect as we can be, but
Kant, Rescher, and surprisingly Machiavelli, would seem to agree that we can only become as
perfect as we can be by willing our own perfection. The thought is that given human
motivation if we will to be as perfect as we can we will end up being less perfect than we
could. If we merely aim to play as well as we can, it is likely that we will play less well than
we could have. It seems plausible, then, to think that living up to our potentials involves what
I will call “overshooting”; we need to aim higher than we are able to reach in order to reach as
high as we can.
I will now discuss and compare four Kantian ideals:
(1) The ideals of being a holy will
(2) The ideal Perpetual Peace
(3) The Ideal Solution to Moral Dilemmas.
(4) The ideal of a final finished science.
(1) Being a ‘holy will’
Kant is an ethical idealist in the sense that he believes that to be moral involves
striving to instantiate a moral ideal. This ideal is an ideal of moral perfection. Sometimes he
names this ideal the ‘idea of humanity’ at other times ‘holiness’. Thus Kant claims in the
Critique of Practical Reason that, “It now follows of itself that in the order of ends human
beings (and with him every rational being) is an end in itself, that is, can never be used merely
as a means by anyone (not even by God) without being at the same time himself and end, and
that humanity in our person must, accordingly, be holy to ourselves: for he is the subject of
the moral law and so of that which is holy in itself, on account of which and in agreement
with which alone can anything be called holy” (5:132) and it is quite clear that he identifies
this ‘idea of humanity’ with some sort of ‘moral perfection’. In his lectures from the early
1790s he argues that, “humanity itself, if we wished to personify it, actually lacks any
inclination to evil, but the more a man compares himself therewith, the more he finds out how
far away he is from it” (Ethik Vigilantius, 27:609). Now the idea of a being that lacks any
inclination to evil is what Kant, elsewhere calls a holy being. And one feature of a ‘holy’
human being is that they would always do their duty gladly7 – but this is not possible for
beings like us who are subject to needs.
Now Kant clearly and consistently distinguishes between the ideas of holiness and
virtue. To be virtuous is to strive for perfection, to gradually improve oneself so that one
gradually approaches the ideal. Morally, it is the most that a flesh an blood human being can
hope to be. But this does not mean that virtue itself is Kant’s moral ideal, for one who takes
virtue as their ideal will not be virtuous.
Indeed, the moral law cannot command us to: be virtuous! For, as Kant argues in the
Metaphysics of Morals, “virtue itself, or possession of it, is not a duty (for then we would
have to be put under obligation to duties)” (6:405). Instead, Kant believes that to be virtuous
“[W]hat one does not do with pleasure [mit Lust] but merely as compulsory service has no inner worth for one
who attends to his duty” (Metaphysics of Morals 6:484)]
is to strive towards holiness and that another formulation of the categorical imperative is: be
holy! I believe that this is Kant's considered position. He makes it clear that this is his position
in the ethics lectures he delivered at the time he was working on the Groundwork. In these
lectures he proclaims that, “The ideal of the gospels has the greatest moral purity. The
ancients had no greater moral perfection than that which could come from the nature of man,
but since this was very defective, their moral laws were also defective. . . The principles of
morality are [in Christianity] presented in their holiness, and now the command [i.e.
Imperative] is: You are to be holy!” (Ethik Collins, 27:252 – my emphasis) Here Kant makes
it clear that it is his belief that it is our duty to be holy. For example, in the Metaphysics of
Morals he claims that,
Virtue so shines as an ideal that it seems, by human standards, to eclipse
holiness itself, which is never tempted to break the law. Nevertheless, this is an
illusion arising from the fact that, having no way to measure the degree of a
strength except by the magnitude of the obstacles it could overcome. . . we are
lead to mistake the subjective conditions by which we assess the magnitude for
the objective conditions of the magnitude itself. (6:397)
Similar passages are not hard to find.
However, he also believes that the most that any (biological) human being can do is to
strive towards perfection but he is also committed to the principle that ought implies can, and
this leaves him with a problem. Morality demands that we should be holy/perfect, but it is
impossible for us to ever be perfect. So it might seem that we should not strive for perfection
but merely strive to strive for perfection, that is, that our moral ideal should be virtue not
holiness. Kant struggled with this problem thought his mature period. He found a solution in
the late 1780s or early 1790s with the idea that we can think about a person’s (intelligible)
moral disposition and their (phenomenal) actions as analogous to the relation between a
mathematical function and a series. If we do this, we can think of holiness is the limit of
virtue. So just as the series ½ + ¼ + 1/8 converges on 1 and so the series as a whole is in some
sense equal to one, so the series of acts that constitute the life of a virtuous being who is
striving to be perfect/holy is converging on perfection and so that in some sense virtue is
equivalent to holiness, although there will be no moment in the life of a virtuous individual
when they actually are perfect. Thus Kant explains that, “because of the disposition from
which it derives and which transcends the senses, we can think of the infinite progression of
the good towards conformity to the law as being judged by him who scrutinizes the heart
(through his pure intellectual intuition) to be a perfected whole even with respect to the deed
(the life conduct).” And he explains in a footnote that the disposition “takes the place of the
totality of the series of approximation carried on in infinitum.” (Religion 6:67)8
(2) Perpetual Peace
Kant believes that perpetual peace is, in a sense, Impossible to achieve. Thus in the
Metaphysics of Morals he argues that,
perpetual peace, the ultimate goal of the whole right of nations, is indeed an
unachievable idea. Still the practical principles directed towards perpetual
peace, of entering into such alliances of states, which serve for continual
approximation to it are not unachievable. Instead, since continual
approximation to it is a task based on duty and therefore on the right of human
beings and of states, this can certainly be achieved.” (Metaphysics of Morals
In the footnote he writes: “It must not be overlooked that we do not thereby mean to say that the disposition
should serve to compensate for any lack of conformity to duty, hence for the actual evil, in this infinite series
(the presupposition is rather that the human moral constitution pleasing to God is actually to be found in the
series), but rather that the disposition, which takes the place of the totality of the series of approximation
carried on in infinitum, makes up only for the deficiency which is in principle inseparable from the existence of
a temporal being, [namely] never to be able to become quite fully what he has in mind.” (6:67) My emphasis.
Elsewhere in the religion he explains that, “Even the purest moral disposition elicits in the human being,
regarded as a worldly creature, nothing more than the continuous becoming of a subject well pleasing to God in
actions (such as can be met with in the world of the senses). In quality (as it must be thought of as supersensibly
grounded) this disposition can indeed be, and ought to be, holy and conformable to the archetype’s disposition.
In degree, however, (in terms of its manifestations in actions) it always remains deficient and infinitely removed
from that of the archetype. Nevertheless, as an intellectual unity of the whole, the disposition takes the place of
perfected action, since it contains the ground of its own steady progress in remedying its deficiency.” (Religion
Not only are there reasons to believe that perpetual peace is not achievable in practice
but there are also reasons to believe that perpetual peace is inconceivable. The reason for this
is Kant’s understanding of how ‘always’ judgments work. Thus in an unpublished fragment
Kant argues that:
"It is possible in each throw of the dice that I roll a six, and just as possible as
every other result; but it is not possible for me always roll a six because that
would require a ground of necessity" (# 7170 19:263, p.461).
Now, in his footnote to the translation Guyer writes that "Kant's present argument is
fallacious: that it is not necessary to roll a six, does not mean that it is impossible to do so"
(p.609). I think that Guyer is missing Kant’s point here. Kant is worried about with what is
involved in something always being the case, and his worry is analogous about what involved
in something, such as peace, being perpetual. In both cases he thinks that this cannot be
understood in terms of some sort of ground mechanism that guarantees the outcome. Kant’s
distinction between something being the case each time and it always being the case, could
be defended on various grounds.
(1) Standard probability theory tells us that the probability of always rolling a six is
zero and perhaps Kant has something like this in mind and assumes that for an event (or
sequence of events) to have probability zero implies that it is impossible.9 This might be one
reason to claim that it is possible to throw a 6 on each throw, but not possible to always throw
a 6. There are a number of problems with such an interpretation. One is that it is unclear if
Kant had a fully worked our theory of probability. Secondly, it is standard today to
distinguish a probability of zero from impossibility. Something can have a probability of zero,
but be possible. For example, and I owe this example to Berna Kilinc, if we have to randomly
choose a point on the continuum between 0 and 1, according to standard probability theory
There has been an interesting discussion of this question recently in Analysis. See, Timothy Williamson, ‘How
Probably is an Infinite Sequence of Heads’, Analysis 67.3 July 2007, pp.173-80 and Ruth Weintraub, ‘How
Probable is an Infinite Sequence of Heads: A Reply to Williamson’, Analysis 68.3, July 2008, pp.247-250.
the probably of choosing the point we choose will be zero, but this does not entail that
choosing such a point is impossible.10
(2) Another, and more plausible way, of interpreting Kant’s claims is by appealing to
intuitionistic logic. For an intuitionist it makes sense to distinguish between each role not
being non 6 and always rolling 6. The following conjunction is consistent intuitionistically:
(1) ¬∃x ¬A(x) & ∃x A(x) & ¬∀xA(x)11
If we take the domain to be an infinite set of times at which dice are thrown, and A to
be interpreted as 6 being thrown at a particular moment, then I take the first two conjuncts to
express that claim that each throw is a six, and this is consistent, intuitionistically with it not
always the case that a six is thrown. And so I take this formula to be a way of expressing the
fact that it is (logically) possible for each throw to be a six without this entailing that all
throws are sixes. What is important is that when it comes to an infinite domain, the possibility
of ∀xA(x), interpreted to mean it always the case that A (for all of the infinite times a dice is
rolled is come up 6), cannot be justified merely by appeal to existential judgments. To do so is
to commit something like the naturalistic fallacy. For ‘Every’ judgments, are judgments of
universality, and Kant thinks that a universal judgment involves some necessity, and deriving
an ‘every’ judgment’ from a set of ‘each’ judgments is deriving a universal judgment from a
set of contingent facts. To establish the truth, or even possibility, of such a statement one must
appeal to some ground of necessity that guarantees that at all times A will be the case, or, in
other word that A will be the case perpetually. One way of putting this, suggested by Peter
Although I think there are some problems with such a story. I’m not sure it is actually possible to randomly
choose a point on the continuum, and if the choice is not random, then the probability of choosing the point that
we do might be more than zero. I think there are some quite difficult issues involving the possibility of truly
random choice here. It is also possible that we may only be able to arbitrarily pick a finite subset of points on
the continuum. But this stuff is probably beyond my pay grade.
I suspect, that perhaps Kant might want to claim something stronger than this, namely that the following
conjunction is consistent: (2) ¬∃x ¬A(x) & ∃x A(x) & ¬◊∀xA(x).
Milne in conversation, is that Kant believes that you just can't have what one might call "an
infinite coincidence". This view is not standard; but it is not obviously crazy.
(3) I suspect, however, that Kant’s motivation here might have to do with a
commitment to a certain from of finitism. In claiming that it is possible for each roll to be a 6,
I think kant may mean that any arbitrary long finite sequence of rolls, it is possible that all the
rolls might be 6, and I think that Kant’s point is that it does not follow that this that is possible
that all rolls, in an infinite sequence, are 6. I think Kant would agree that it is not impossible
that for any arbitrarily long finite sequence of rolls it is possible that all of them are sixes, but
that this is not the same as "always rolling a six". It is not the series of number 1/2, 1/4, 1/8 . .
. that converges on 1, but the function. We can only say that this series converges on 1
because there is a function 'behind' the series that makes the series 'necessary'. The function is
the “ground of necessity” that “guarantees” the convergence. The three dots only make sense
because we understand the function (which is that for each member of the series, after the
first, the denominator is twice that of the previous member). To say that we will always roll a
'6' is to think of the series 6, 6, 6 . . . It is to think that not only that (a) every roll up to a
certain point been a 6 (which for any moment is possible) but also that (b) every future roll
will be a six. And that the thought 'every future x will be y' involves the idea of necessity.
Therefore, Kant is committed to the position that any ‘always’ or ‘perpetual’ judgment
involves some appeal to a mechanism, function or ground that will guarantee the necessity of
the outcome.
I suggest that this sort of worry is the main point being addressed in the ‘First
Suplement’ to Perpetual Peace which is entitled “On the guarantee of Perputal Peace”. Kant’s
aim in perpetual peace is to show what mechanisms could function as the “ground of
necessity” of perpetual peace and to show that such mechanisms could in principle be created
by beings like us, indeed even by a “nation of devils” (8.366). His suggestion that lasting
peace could be guaranteed by a lasting peace is would be possible between a federation of
republics enmeshed in a web of trading relations, where the federation had the authority to
settle disputes between the sates concerning their rights.
If he show that and ideal set of such mechanisms could guarantee peaceful co-relations
between states he has shown that Perpetual Peace is both conceivable and that we can
conceive of what it would be like for beings like us to live together in perpetual peace. If he
is successful in showing both of these things then he has successfully shown that the idea of
perpetual peace can function as a practical ideal. He does not need to show that it is possible
for us to actually succeed in fully bring such mechanisms into existence. All we need is a
concrete yardstick that allows us to determine whether particular policies are likely to bring us
closer or further away from the ideal.
(3) Ideal Solutions to Moral Dilemmas.
Kant famously argues in the Metaphysic of Morals that “Because however duty and
obligation are in general concepts that express the objective practical necessity of certain
action… it follows … that a conflict of duties and obligations is inconceivable (obligationes
non collidunto)” (6:224).12 These are strong fighting words – but why does Kant make such a
commitment? As W. A hart Points out, “[T]he idea of the moral law as a harmonious and
consistent set of principles, in as much as it is a regulative idea does not by itself rule out the
The full quote is “A conflict of duties (collisio officiourum s. obligationum)' would be a relation between them
in which one of them would cancel the other (wholly or in part). - But since duty and obligation are concepts that
express the objective practical necessity of certain actions and two rules opposed to each other cannot be
necessary at the same time, if it is a duty to act in accordance with one rule, to act in accordance with the
opposite rule is not a duty but even contrary to duty; so a collision of duties and obligations is inconceivable
(obligationes non colliduntur)/ However, a subject may have, in a rule he prescribes to himself, two grounds of
obligation (rationes obligandi), one or the other of which is not sufficient to put him under obligation' (rationes
obligandi non obligantes), so that one of them is not a duty. - When two such grounds conflict with each other,
practical philosophy says, not that the stronger obligation takes precedence (fortior obligatio vincit) but that the
stronger ground of obligation prevails (fortior obligandi ratio vindt).” (6:224)
possibility of conflicts between duties which are irresolvable”13 instead, he argues, the most
plausible account of Kant’s motivation here is based upon the ought implies can principle.
In this section I do not want to engage in Kant interpretation, but to defend the
position that I think is most plausible, and one that I believe is compatible with a generally
Kantian outlook. The position I wish to defend is close to that of Ruth Barcan Marcus, who
argues, in an influential paper, that,
[T]he existence of moral dilemmas, even where the dilemmas arise from a
categorical principle or principles, need not and usually does not signify that
there is some inconsistency (in a sense to be explained) in the set of principes,
duties, and other moral directives under which we define our obligations either
individually or socially. I wasn’t to argue that… consistency of moral principles
or rules does not entail that moral dilemmas are resovable in the sense that acting
with good reasons in accordance with one horn of the dilemma erases the
original obligation with respect to the other…[Although dilemmas are not settled
without residue, the recognition of their reality has a dynamic force. It motivates
us to arrange our lives and institutions with a view to avoiding such conflicts. It
is the underpinning for a second-order regulative principle: that as rational
agents with some control of our lives and institutions, we ought to conduct our
lives and arrange our institutions so as to minimize predicaments of moral
I can see no good reason why a Kantian should disagree with this.15 And I will suggest
that the best way of cashing this out within a Kantian paradigm is to make a distinction
between what we actually have a duty to do (which is constrained by our actual capacities)
and what ideally we ought to do. I am suggesting that we can think of two senses of ‘can’
here. Our actual duties, at a time, are constrained by what we can actually do at a given
moment. What we should ideally do is constrained by what our capacities could have been.
W.A.Hart “Nussbaum, Kant and Conflicts between Duties.” Philosophy 73, 1998, p.611.
Ruth Barcan Marcus, “Moral Dilemmas and Consistency”, The Journal of Philosophy, 77:3, 1980, p.121.
One possible motivation here, noted by Marcus in her paper, is provided by John Lemmon, who argues that,
“It may be argued that our being faced with this moral situation [a moral Dilemma] merely reflects an implicit
inconsistency in our existing moral code; we are forced, if we are to remain both moral and logical, by the
situation to restore consistency to our code by adding exception clauses to our present principles or by giving
priority to one principle over another, or by some such device. The situation is as it is in mathematics: there, if an
inconsistency is revealed by derivation, we are compelled to modify our axioms; here, if an inconsistency is
revealed in application, we are forced to revise our principles.” John Lemmon, “Deontic Logic and the Logic of
Imperatives, Logique et Analyse, 8:29, 1965, pp. 39-61.
As we shall see, such a distinction might be needed to explain our imperfect duty to develop
particular capacities. Also, given the fact that that the extent of our capacities is not
immediately apparent to us, there are good reasons to think that reflection on what we ideally
ought to do, is something that should often proceed our decision of what we actually have a
duty to do. I think one can cash out the notion of what we ideally ought to do by appealing to
possible worlds. Suppose I cannot save x and y in the actual world, but there is a possible
world in which I exist in which I could save them, even thought I can not actually save both
because there is a possible world in which I could have saved both, it could be the case that
ideally I should save them. This approach respects the ought implies can principle, but does
not limit the principle to what I can do in the actually world but depends upon my capacities
in other possible worlds in which I exist. Such an approach is motivated by the fact that our
capacities are not fixed and we have to make choices about which of our capacities to
develop, and can be blameworthy for not developing certain capacities.
Imagine the following scenario, which I will call scenario one.
(a) Two identical twins, Burak and Burkay, have fallen overboard and
cannot swim and will drown if not saved.
(b) You are the only person on the boat who knows how to swim and you
have some sort of responsibility for the welfare of both of them. (You might
be their mother, or you may have been responsible for their falling in, or
you may be employed as the boat’s lifeguard, professionally and
contractually responsible for saving anyone who has fallen overboard. Or
perhaps you have promised each passenger on the boat, individually, that
you would save them if they fell in.)
(c) You know that you can only save one of them.
What should you do?
One obvious answer here is that you ought to save Burak and you ought to save
Burkay, and so you ought to save both of them. However, the ‘ought implies can’ principle
seems to tell against such answer. And so some commentators, influenced by Kant, suggest
that in such a scenario one has a disjunctive rather than a conjunctive obligation. One ought to
save either Burak or Burkay. But one does not, and cannot, have a duty to save both; if you
save Burak and Burkay drowns, you have not violated any duty you have towards Burkay.
Although it has been argued, for example by Bernard Williams, that in such a scenario
even though you do not have an obligation to save both Burak and Bukay, one does have an
obligation to save Burak and an obligation to save Burkay.16 What fails here is what Williams
calls the principle of agglomeration: the principle that states that if ‘I ought to to A’ and ‘I
ought to do B’ then ‘I ought to do A and B’. And this principle clearly does not hold for other
some domains concerning actions. For example: It might be true that ‘I would be happy if I
married A’ and that ‘I would be happy if I married B’, but this does not imply that it would be
true that ‘I would be happy if I married both A and B’. And perhaps ‘ought to’ behaves in a
similar way. So it is not obvious that the ought implies can principle entails that you don’t
have an obligation to save Burak and an obligation to save Burkay, (and that if you fail to
save Burak in such a situation you have not violated a duty towards him).
Personally, I’m sympathetic to the claim that one does not have a (strict) duty to save
both twins here, and that one has not violated a duty to the twin who you did not save.
However, at the same time I think that it makes sense to say that ideally one ought to save
both. And I think there are a number of reasons for this. Here are four.
First: Imagine that if you had been a better swimmer you would have been able to
save both twins. Now imagine a second scenario. Suppose you were offered a job on a boat
Bernard Williams, ‘Ethical Consistency’ in Problems of the Self, Cambridge University Press, 1973, p.181.
and were knew that scenarios like scenario one happened on a regular basis. I think that
given this knowledge you would recognize that you ought to improve your strength as a
swimmer. However, it is difficult to see how this obligation to improve your swimming skills
would be grounded if in scenarios like scenario one, you only had an obligation to save either
of the twins. It seems clear that in a scenario like scenario one, one might recognize that
ideally one ought to be able to save both twins, and this is what ground the recognition that in
taking up such a job one ought to improve ones swimming skills so that if such a situation
arose one would be able to save both. But the recognition that in scenario one, you should
ideally be able to save both seems to be grounded in the recognition that, ideally, one ought to
save both.
Second (and this point is related to the previous observation): Although in scenario
one, you might not have a duty to save both twins and so have not directly violated a duty to
the unsaved twin, one might be culpable for not having the ability to save both. Imagine a
similar scenario in which you are the owner of the ship and only have one lifebuoy. Even
though you might not have a duty (at the moment) to save both, it seems that you are blame
worthy, and probably also criminally liable, for lacking the ability to save both. And it seems,
to me that the best explanation of this culpability is the fact that ideally you should save both.
In such a situation one ought to have been able to save both; having promised to save each,
you should have ensured that there were enough lifebuoys to have saved all (assuming that
this was something you could have done). And I think what best explains this ought is the fact
that ideally one should save both.
Third: Overshooting is important in the moral sphere. Imagine a scenario where
instead of twins there are three triplets in the water. It might be the case that I know that it is
impossible to save all three, but that I am only able to save two of them if I try and save all
three. So, for motivational reasons, it is important in the moral sphere to recognize what we
should ideally do in order to be able to do the best that we can do.
Finally: In these scenarios I have stipulated that you know what you are able to do.
But this is clearly an unrealistic assumption. [In this I agree with Thomas Reid. 17] We do not
normally know our capabilities until we have tested them. And we are often prey to selfdeception in such matters. This is often the case in the political sphere – where we tell
ourselves that any real meaningful change is impossible, as an excuse to not do what we
recognize to be right.
And this leads to a further question about the ought implies can principle. How are we
to interpret the ‘can’ here? Suppose I believe that doing x is possible but, it is in fact,
impossible? Do I have a duty to try and do x? What if I believe that I know that doing x is
impossible for me to do but it is in fact possible? Have I violated my duty do x? Given our
lack of knowledge of our own capacities, we should first think about what should ideally be
done and then see what we can do to do it.
(4) The Ideal of a Final Science.
The ideal of a finished science might seem closer to the ideal of perpetual peace than
the ideal solution to a moral dilemma. But there are reasons why it might seem that we can
assimilate Kant’s regulative ideal of a final science to the ideal solution to a moral dilemma.
So let me first explain why one might think that the two cases are similar before examining
why and how they might be different.
Reid argues, for example, that “I shall have occasion to shew, that we have very early, from our constitution,
a conviction or belief of some degree of active power in ourselves. This belief, however, is not consciousness:
For we may be deceived in it; but the testimony of consciousness can never deceive. Thus, a man who is struck
with a palsy in the night commonly knows not that he has lost the power of speech till he attempts to speak; he
knows not whether he can move his hands and arms till he makes the trial; and if, without making trial, he
consults his consciousness ever so attentively, it will give him no information whether he has lost these powers,
or still retains them.” Essays on the Active Powers of Man, edited by Knud Haakonssen and James A. Harris,
Edinburgh University Press, 2010, p.9.
(a) Science and Moral Dilemmas. Let me tell a story, that makes it look as if the
regulative ideal of a completed science is analogous to the ideal solution to a moral dilemma.
For Kant the idea of a complete science involves the idea of a fully systematic conceptual
scheme. Kant is working in an Aristotelian paradigm in which systematicity requires the
organization of our concepts according to their genus and species. Such systematicity requires
both complete unity (which involves a single highest genus) and complete specification. It is
tempting to regard unity and specification as analogous to the two twins in scenario one
above. Ideally we want both, but we can only aim at one at a time.
In the Appendix to the Transcendental Dialectic of the First Critique (in a section
called ‘On the Regulative Use of the Ideas of Pure Reason’) Kant argues that,
If we survey the cognitions of our understanding in their entire range, then we
find that what reason quite uniquely prescribes and seeks to bring about is the
systematic in cognition, i.e., its interconnection based on one principle. This
unity of reason always presupposes an ideal, namely the form of a whole of
cognition, which precedes the determinate cognition of the parts and contains the
conditions for determining a priori the place of each part and its relation to the
others. Accordingly, this idea postulates complete unity of the understanding’s
cognition, through which this cognition comes to be not merely a contingent
aggregate but a system interconnected in accordance with necessary laws. One
cannot properly say that this idea is the concept of an object, but only that of the
thoroughgoing unity of these concepts, insofar as the idea serves the
understanding as a rule. (A646/B674)18
Now as this section progresses Kant makes it clear that he is thinking of this systematic
unity of cognition in terms of a set of concepts systematically related in terms of genus-species
relations. Thus he claims that:
We also find this transcendental presupposition hidden in an admirable way in the
principles of the philosophers. . . That all the manifoldness of individual things
does not include the identity of species; that the several species must be treated
only as various determinations of fewer genera, and the later of still higher
And a little later he adds that this “systematic unity (as mere idea) is only a projected unity, which one must
regard not as given in itself, but only as a problem; this unity however, helps us to find a principle for the
manifold and particular uses of the understanding, thereby guiding it even in those cases that are not given and
making it coherently connected.” (A647/B675)
families, etc; that therefore a certain systematic unity of all possible empirical
concepts must be sought insofar as they cab be derived from higher and more
general ones. A651-2/B679-80
The theoretical ideal of systematicity, then demands that we unify our conceptual scheme, and
he calls this demand for unity the principle of genera (which he elsewhere calls the principle
of homogeneity, and assigns to the faculty of “wit”). But the ideal of systematicity also
involves another principle, which, at the very least creates tension with the first principle,
which Kant calls the principle of species (which he elsewhere calls the principle of
specification, and assigns to the faculty of “discrimination”). Thus he claims that:
To the logical principle of genera which postulates identity there is opposed another,
namely that of species, which needs manifoldness and variety in things despite their
agreement under the same genus, and prescribes to the understanding that it be no less
attentive to variety than to agreement. This principle (of discrimination, or of the
faculty of distinguishing) severely limits the rashness of the first principle (of wit);and
here reason shows two interests that conflict with each other: one the one side, an
interest in the domain (universality) in regard to genera, on the other an interest in
content (determinacy) in respect to the manifoldness of species; for in the first case the
understanding thinks much under its concepts, while in the second it thinks all the
more in them. (A654/B682)
Thinking of the ideal of systematicity as involving something analogous to a moral
dilemma, we might say that ideally we ought to be both witty (unifying) and discriminating
(seeking diversity) but we cannot do both – and so, individually, we have a disjunctive duty to
be either witty or discriminating. Collectively, however, as a community we may succeed over
time in being both witty and discriminating. And indeed in the continuation of the quote given
above, Kant seems to suggest such a scientific division of labor between the discriminating
and the witty, claiming that:
This expresses itself in the very different ways of thinking among students of
nature; some of whom (who are chiefly speculative) are hostile to differences in
kind, while others (chiefly empirical minds) constantly seek to split nature into
so much manifoldness that one would almost have to give up the hope of judging
its appearances according to general principles. (A655/B683)
Kant concludes the section by claiming that:
There is nothing here but the twofold interest of reason, where each party takes
to heart one interest or the other, or affects to do so, hence either the maxim of
the manifoldness of nature or that of the unity of nature; these maxims can, of
course be united, but as long as they are held to be objective insights, they
occasion not only conflict but also hindrances that delay the discovery of the
truth, until a means is found of uniting the disputed interests and satisfying
reason about them. (A667-8/B695-6)
Now, it is not clear here why Kant believes that it is obvious that these two maxims can be
combined. But it is clear that the solution involves recognizing that these maxims are not
“objective insights”. And thinking about why Kant claims this will help us understand the way
in which the ideal of systematicity – which involves seeking both unity and diversity is not
really similar to the moral dilemma discussed earlier. Unity and diversity are not really like
twins who both need to be saved, and the “ideal”of systematicity does not really involve
complete unity and complete specification of our conceptual scheme – because complete
specification is an incoherent notion. The idea of a fully systematized conceptual scheme is
not merely unattainable – but it is incoherent and logically impossible.
(2) Unity and Diversity are not like two Drowning Twins, only one of whom we can save.
For Kant there is something fishy about the idea of complete specification, for the
notion of complete specification presupposes the coherence of the notion of lowest species,
and the notion of a lowest species is, Kant believes, incoherent. For any species can be thought
of a a genus that could, in principle, be further specified. This commitment, as we shall see,
lies at the heart of his disagreement with Leibniz. Thus, in the section we have been
discussion, Kant claims that,
[E]very genus requires different species, and these subspecies, and since non of
the latter once again is ever without a sphere (a domain as a conceptus
communis), reason demands in its entire extension that no species be regarded as
in itself the lowest; for since each species is always a concept that contains
within itself only what is common to different things, this concept cannot be
thoroughly determined, hence it cannot be related to an individual, consequently,
it must at every time contain other concepts, i.e., subspecies, under itself. (A6556/B683-4)
In claiming that there is, in principle, no lowest species, Kant is implicitly committing
himself to the position that complete specification is impossible, which suggests that the ideal
of complete systematicity (which involves the idea of complete specificity) is incoherent. An
this is not a passing thought, bus is at the heart of his disagreement with Leibniz (who thought
of Monads as lowest species). It would seem, then that the complete systematicity of our
conceptual scheme cannot function as an ideal.
It is for this reason that the ideal really at work here is not that of complete
systematicity but that of empirically adequate systematicity. Our conceptual scheme does not
need to be completely specifies, but the specificity merely needs to be empirically adequate.
But this notion of empirically adequate specificity presupposes making a certain assumption
about the natural world. Thus Kant claims that,
[I]t is easy to see that even this logical law [of complete specification] would be
without sense or application if it were not grounded on a transcendental law of
specification which plainly does not demand an actual infinity in regard to the
varieties of things that can become our objects – for the logical principle of
asserting the indeterminacy of the logical sphere in regard to possible division
would give no occasion for that; but it [what?] does impose on the understanding
the demand to seek under every species that comes before us for subspecies, and
for every variety smaller varieties. (A656/B684)
As far as I understand this claim, the thought seems to be that the transcendental law
demands merely that within our conceptual scheme we have a concept for every species that
actually exists – and this law is transcendental because it rests upon the assumption that the
number of actual species if finite – and this assumption is a presupposition for the possibility
of (ideal?) knowledge of the phenomenal world. Reason, however demands that however
adequate our conceptual scheme seems to be that we constantly search for possible further
specification. The problem here seems to be that although it might be possible to arrive at a
fully empirically adequate conceptual scheme reason (which obeys the logical law of
specification, not the transcendental law) can never be satisfied with it. So perhaps the
conclusion to draw here is that even if a final finished science is possible, we could never be
in a position to know that we have achieved it.
So – to come back to my original question – is the ideal of sytematicity more like the ideal
of perpetual peace or the ideal solution to a moral paradox? Is it an ideal way of being or an
ideal way of doing? I’m not really sure.
Perhaps I’ll work out what I think while I’m reading the paper.
[In the first part one of this paper I want to think through various ways in which
ideals can be impossible. This section of the paper has little to do with Kant. Although in
thinking thought these questions Kant is always in the back of my mind, and I will discuss
his regulative ideal of a completed finished science. In the second part of this paper I want
to turn back to Kant. I have four main things to say about Kant: (1) That the theoretical
ideal of a completed science is impossible in a different way than Kant’s moral ideal(s) –
and understanding this may help us understand the distinction between regulative and
constitutive ideals. (2) The ‘ought implies’ can principle is too strong to apply to ideals;
the fact that we cannot achieve perpetually peace or actually have a good will at any
moment in time does not imply that we should not strive to realize such ideals. (3)
Recognizing this fact suggests that we need to make some sort of distinction between
different types of oughts. I think that such a distinction is implicit in Kant but he does not
employ a consistent vocabulary to mark this distinction. I suggest stipulating a distinction
between ‘duties’ and what we ‘should’ do on the one hand and what we ‘ought’ to do on
the other hand. I suggest that a strict version of the ‘ought implies can’ principle should be
applied to ‘duties’ and what we ‘should’ do, but a weaker version should be applied to
what we ‘ought’ to do or be. (4) ]
(1) Perfect Performance and “overshooting”. A violinist or footballer, say, may have an
ideal of a perfect performance. A footballer may have an idea of what it is to play perfectly.
Ideally every pass and shot she makes will be on target, and she will be at the right place on
the pitch at every moment of the game, and to play at this level for every game of the season.
At least part of this such ideals is the idea of never making a mistake. I think that many
performers are guided by such ideals. Now, such a performer may recognize that as in
imperfect human being that it is impossible to perform ideally over any sustained period of
time., and that over a seasons she will inevitably make some mistakes. She will miss some
shots and sometimes be badly positioned or fail to pass accurately or when she should. Now,
such ideals do make some concessions to human nature – and to some version of the ought
implies can principle. Such ideals of performance do not, at least in some sense, demand the
impossible. So, for example, the idea of a perfect game does not require that the footballer be
at two places at once, or be able to shoot the ball over 300kph. In addition such ideals are
generally sensitive to the skills and talents of the individual involved. As we get older and less
fit, our ideal of a perfect game normally change. On the other hand, although such ideals or
performance to show a sensitivity to our capacities and abilities, such ideals of performance
are probably humanly impossible – they require a degree of concentration that is probability
impossible for human beings, and we might think that over any extended period of time it is
inevitable that there will be lapses of concentration and attention, at that we will make
Now we might think (perhaps guided by some version of the ‘ought implies can’
principle, that such ideals of performance are unhealthy, and instead of aiming to play a
perfect game, we should lower our standards and just aim to play as well as we can. However,
as Nicholas Rescher has argued, it seems a plausible claim about human nature that in order
to play as well as we can we need to aim at playing better than we can. [I wonder if anyone
knows of any empirical research on this phenomenon.]. If we merely aim to play as well as
we can, it is likely that we will play less well than we could have. It seems plausible, then, to
think that living up to our potentials involves what I will call “overshooting”; we need to aim
higher than we are able to reach in order to reach as far as we can.