‘to serve and protect’: the ends of harm by victor tadros
TRANSCRIPT
ORI GIN AL PA PER
‘To Serve and Protect’: The Ends of Harm by VictorTadros
Kasper Lippert-Rasmussen
� Springer Science+Business Media Dordrecht 2013
Abstract In The Ends of Harm Victor Tadros develops an alternative to consequentialist,
and non-consequentialist retributivist, accounts of the justifiability of punishment: the duty
view. Crucial to this view is the claim that wrongdoers incur an enforceable duty to remedy
their wrongs. They cannot undo them, but they can do something that is almost as good—
namely, by submitting to appropriate punishment, which will deter potential wrongdoers in
the future, reduce their victim’s risk of suffering similar wrongs again. Admittedly, this
involves harming wrongdoers as a means to an end, but according to Tadros the ‘means
principle’ that we should not harm others as a means, properly construed, does not apply to
cases where the victim has an enforceable duty to bear the kind of harm that he or she is
being made to suffer. In this article, I shall express reservations about Tadros’ defense and
interpretation of the means principle. In presenting his position, Tadros also sets out some
interesting anti-retributivist considerations casting doubt on the idea that wrongdoers’
suffering is non-instrumentally good. I shall challenge these. Finally, I shall suggest that
the duty view may have counterintuitive implications in relation to wrongs where the
offender helps to lower the risk that victims will be subjected to similar wrongs in ways
other than by being punished.
Keywords Victor Tadros � Punishment � Deterrence � Retributivism � The means
principle � Selfdefense
Introduction
Justifying state punishment may seem difficult. Even if we set aside the more easily
correctable bad, contingent features of contemporary systems of punishment such as over-
criminalization and discrimination, it is hard to imagine a penal system that is not 1) costly,
2) harmful to innocents, including the children of offenders, or 3) apt to inflict harm on
K. Lippert-Rasmussen (&)University of Roskilde, CUID, Section for Philosophy Universitetsvej 1, Building 3.2.1, Postbox 260,4000 Roskilde, Denmarke-mail: [email protected]
123
Crim Law and PhilosDOI 10.1007/s11572-013-9216-y
some, i.e., convicted offenders, as a means of benefitting others, principally those who
stand to gain from the deterrence payoff.
At least two justifications of punishment are widely accepted. Retributivists believe that
punishment is justified in principle because, in principle, it ensures that wrongdoers suffer
to a degree that fits their crimes. The retributivist takes suffering of this precise kind—
proportionate punitive suffering—to be a good. Consequentialists, by contrast, believe that
punishment is justified in principle, because, in principle, when it takes the form of
imprisonment it prevents wrongdoers from offending, at least in certain ways while
imprisoned, and because a policy of punishing offenders, whether it takes the form of
imprisonment or some other form, deters others from wrongdoing. In standard forms of
consequentialism suffering, even when it is putatively proportionately punitive, is bad in
itself, but punitive measures can be justified by their favorable net effect.1
In The Ends of Harm Victor Tadros proposes a third justification of punishment: the
duty view. He thinks retributivist justifications fail, because the claim that the suffering of
an offender is valuable in itself is implausible. Hence, on this crucial matter, Tadros sides
with consequentialists. However, he also believes that consequentialism is implausible
because, among other things, it fails to recognize that there are moral side-constraints on
our pursuit of the good. In particular, he alleges that consequentialism permits the use of
persons as a means in a wide range of cases where, intuitively, this is impermissible.
Interestingly, however, he stops short of wholesale rejection of the ‘instrumentalist’ jus-
tification of punishment. Indeed, he submits that punishment is justified, in principle at
least, because, and only because (Tadros 2011: 23, 40), of its general deterrence effects. As
he reminds us, like consequentialists, deontologists, too, think that the consequences of
acts (and policies, and so forth) bear on their permissibility—it is just that they are not the
only consideration. In a nutshell, the duty view offers a deontological justification of
punishment that is based on the deterrence effects of punishment, but avoids any com-
mitment to the idea that proportionate punitive suffering is non-instrumentally valuable.
This theoretical positioning is fascinating. It alone ensures that the duty view merits,
and will receive, close scrutiny by theorists of punishment. Add to this the fact that Tadros’
defense of his position is ingenious, draws on a wide range of philosophical resources, and
manifests great sensitivity to moral nuance, and it becomes clear that the duty view is a
powerful arrival among the justifications of punishment available today.
Absolutely crucial to the duty view is the idea that a wrongdoer has a ‘duty to provide a
remedy to the victim for the harm that she has suffered’ (Tadros 2011: 2). If wrongdoers
are unable to fulfill this duty themselves, or are able to, but refrain from doing so, they can
permissibly be made to bear certain costs associated with the provision of the remedy
(Tadros 2011: 277). As Tadros puts it: ‘The appropriate way to remedy [a] wrong is by
providing protection to victims … against future harm. Because offenders are obliged to
provide protection to victims against future attacks … it is permissible for us to harm them
as a means to protect ourselves against future offending. For this reason, it is justified to
punish them for reasons of what we call ‘general deterrence’ … The right to punish
offenders, like the right to force wrongdoers to compensate their victims, is grounded in the
duties that they incur as a result of their wrongdoing’ (Tadros 2011: 2–3). In Tadros’ view
these duties go beyond mere compensation; offenders might have a duty to suffer harms to
1 I say ‘standard forms’, because, among other things, consequentialists can be retributivists, i.e., you canconsistently think that one ought to maximize moral goodness, that ‘the suffering of wrongdoers isintrinsically valuable’ (Tadros 2011: 35), and that, for this reason among others, state punishment ofwrongdoers maximizes moral goodness.
Crim Law and Philos
123
such a degree that victims are more than compensated through increased general deterrence
(Tadros 2011: 288). Also, the harms offenders can be required to suffer may be greater
than those that they have imposed on their victims.
Two apparent weaknesses can be discerned in this cluster of ideas. First, punishing a
particular offender is intended not just to improve the situation of his or her victims, but all
other victims of crime. Second, punishment is even intended to provide more security for
those who are not victims of crime at all. To deal with these Tadros draws on the idea,
which he introduces in the context of self-defense, that offenders have a duty to pool their
responsibilities to protect their victims (Tadros 2011: 193–194, 275). If A has created a
threat to B, and C likewise to D, and if the threat to B can be averted only by harming C
and similarly the threat to D can be averted only by harming A, A and C have a duty to
suffer the harms in question even though, in each offender’s case, his or her being harmed
will not serve to avert the particular threat he or she poses. Extrapolating this idea to the
case of punishment, one can explain why offenders have a duty to provide security for
victims in general and not just their own victims.
In response to the second weakness, Tadros suggests that victims may pass on their
entitlements to others, such as people they love (Tadros 2011: 280). As he notes, the
conclusion warranted by an argument appealing to this consideration seems too weak in the
light of cases where no victim of crime cares about anyone who is not a victim of crime. He
suggests that a stronger conclusion regarding how states may punish wrongdoers ‘in order
to protect all citizens of the state’ by considering the ‘role of the state’ and the ‘permissions
and obligations that citizens have to transfer their right to protection to other citizens’
(Tadros 2011: 281).2
As already indicated Tadros’ book draws in interesting ways on recent discussions of
self-defense (Tadros 2011: 266). To support his core idea about the duties of wrongdoers,
he argues that in cases of self-defense wrongdoers have an enforceable duty to bear the
costs of ensuring that the threat under which they have put their potential victims does not
materialize (Tadros 2011: 188).3 For instance, if I have set a dangerous dog on you, and if
the only way to prevent it from harming you is by my coming between you and the dog,
then it seems very plausible that I have an enforceable duty to put myself between you and
the dog (Tadros 2011: 53). And if I cannot interpose myself, or if I can but will not, others
may have a right to forcibly interpose me. In Tadros’ view: ‘[t]he permissibility of pun-
ishing wrongdoers on grounds of general deterrence … can be understood as a legitimate
extension of the permissibility of defending oneself against wrongful attackers’ (Tadros
2011: 16). Translated to the case of punishment, this view implies that ‘offenders are
obliged to provide protection to victims against future attacks’ (Tadros 2011: 2–3), and this
protection is permissibly provided by the general deterrence effects that result from pun-
ishing offenders.
This article raises some questions for advocates and developers of the duty view.
Section ‘‘Dramatis Personae’’ pins down the duty view a bit more precisely than The Endsof Harm does. Section ‘‘Do Retributivists Get the Criticisms They Deserve?’’ explores
2 It is unclear to me exactly how Tadros develops the additional considerations intended to justify pun-ishment benefiting non-victims. But howsoever that may be, it seems these considerations will still generateconclusions that are too weak from a cosmopolitan perspective. From this perspective, presumably, it ispermissible to punish wrongdoers to protect non-citizens of the state also.3 While one can extrapolate from claims about self-defense to claims about punishment, self-defense differsfrom punishment in that it primarily involves the duty people have not to offend against others, whereaspunishment concerns obligations offenders owe to victims (and others) as a result of having offended againstthem (Tadros 2011: 266).
Crim Law and Philos
123
some of Tadros’ arguments against retributivism. Sections ‘‘The Means Principle and
Punishment’’ to ‘‘The Scope of the Means Principle’’ focus on the means principle. Section
‘‘The Means Principle and Punishment’’ challenges Tadros’ defense of it; Sect. ‘‘The
Means Principle and Moral Status’’ criticizes Tadros’ attempt to link it with ideas about
moral status; and Sect. ‘‘The Scope of the Means Principle’’ raises some doubts about the
scope of the principle. Section ‘‘Implications of the Duty View’’ presents three challenges
to the duty view, including the allegation that it cannot justify punishment of wrongdoers
who have already done more than enough to remedy their wrongs.
Dramatis Personae
Three theories of punishment have significant roles in Tadros’ book: the communicative
view, the retributivist view, and his own duty view.4 I shall set aside the communicative
view. While Tadros thinks that some form of it might be true, he also thinks that, at best, it
plays a minor role, relative to the two other theories, in justifying the institutions of
punishment (Tadros 2011: 110). Clearly, the big fish that Tadros wants to fry is retribu-
tivism, which he identifies with the following two claims:
1) It is intrinsically valuable that offenders suffer in proportion to the gravity of the
offence they have committed because that is what they deserve; and 2) it is per-
missible for the state to ensure that offenders get what they deserve by punishing
them (Tadros 2011: 26).
While he offers other formulations that are not exactly equivalent, he treats the quoted
statement as a canonical formulation of the theory he wants the duty view to displace.5
Unfortunately, The Ends of Harm contains no similarly brief and explicit statement of
the duty view. We can, however, clarify Tadros’ conception of the view by attending to
what he says in the light of the assumption that the duty view is incompatible with
retributivism.
First, then, I take it that friends of the duty view reject retributivism. Hence, we might
suggest that part of the duty view is:
1*) It is not the case that ([a] it is intrinsically valuable that offenders suffer in
proportion to the gravity of the offence that they have committed [b] because that is
what they deserve).
Note that one can affirm 1*) for two different reasons, which is why I distinguish between
denial of [a] and denial of [b]. It is one thing to be hostile to the very idea that it is
intrinsically valuable for offenders to suffer in proportion to the gravity of the offences
4 Consequentialism also plays a role, but Tadros says little about it.5 For instance, elsewhere his rough and ready statement of 1) omits the proportionality element and hisstatement of 2) concerns not the permissibility of state punishment, but whether the state has a ‘right’ topunish (Tadros 2011: 9). In a different passage, he says that retributivism ‘holds only’ that ‘the intrinsicgoodness of the suffering of wrongdoers is a good reason to punish them’ (Tadros 2011: 35–36). However,his official statement is silent on reasons for punishing offenders and on what justifies the state in punishingthem (cf. Tadros 2011: 41). Again, in places Tadros ascribes to retributivists the view that the well-being ofoffenders should be proportionate given their offences (e.g. Tadros 2011: 70, 74). Strictly speaking, how-ever, this does not follow from the definition in 1) and 2). If, in the course of punishment, a wrongdoer issubjected to physical pain and suffers as a result, this level of pain may be proportionate in relation to theoffense, even if, ironically, as a result of his painful experiences his well-being is boosted in dimensionsother than the pleasantness of his mental states (e.g., during the time of suffering he might acquire greaterself-knowledge).
Crim Law and Philos
123
they commit, quite another to reject the notion that the offender’s getting what he or she
deserves is the reason why proportionate punitive suffering is a good. This helps to brings
out that 1) actually contains two claims: an axiological one and a further claim clarifying
the basis on which this claim is made.
I take it that, since Tadros denies the first claim, he also denies the second. In fact, he
might deny 1) because he affirms a position more specific than *1), namely:
1**) [a] It is intrinsically disvaluable that offenders suffer and this disvalue is pro-
portional to their degree of suffering [b] because suffering is intrinsically bad, and
more suffering is worse than less, whatever is the case about the sufferer, provided
the sufferer lacks free will.6
At any rate, Tadros’ remarks about the ‘moral alchemy’ involved in suffering becoming
valuable when the sufferer is a wrongdoer, and his dismissal of the sort of free will he
thinks desert requires, suggest something like 1**). I return to these issues below.
Whichever is the right statement of Tadros’ anti-retributivist view there is a serious
question as to whether it should be included in the definition of the duty view. First, the
retributivist view, as stated by Tadros, does not rule out the possibility that other, non-
retributivist justifications of punishment are correct. Hence, if the duty view includes a
claim to the effect that one, non-duty view based justification is incorrect, the distinction
between retributivism and the duty view becomes heterogeneous, i.e., the two views are
specified along different dimensions. Second, the incorporation of 1*), or 1**), in the duty
view rules out retributivism by definition. However, it is possible to regard punishment as
an activity that can be justified by several reasons, just one being the duty of offenders to
remedy their wrongs. This view is easier to justify than one which asserts this and rules out
retributivism. For these two reasons, I want to state the duty view in a way that parallels the
way Tadros states retributivism.
Given this stricture, the next problem is that, unlike retributivism, the duty view is not
about value. Hence, it contains no statement different from, but analogous to, 1). However,
we can overcome this difficulty by stating the duty view thus:
10) [a] It is permissible to make offenders suffer in proportion to the harm inflicted on
their victims whenever that will suitably remedy the wrong they have done,
[b] because offenders have an enforceable duty to remedy their wrongdoing; and 20)it is permissible for the state to ensure that offenders suffer as specified in 10).7
On this definition, 20) is not at the core of the disagreement between retributivism and the
duty view.8 As Tadros rightly points out, it does not follow from its being permissible to
punish a wrongdoer that it is permissible for the state to do so. For instance, the state may
be illegitimate, or, even if it is legitimate, it may be in no position to punish others for the
relevant kind of wrongdoing because it is involved in the very same kind of wrongdoing on
a much grander scale. However, the duty view and retributivism are united in claiming that
the state can be in a position to punish. They just differ on what makes punishment
permissible as such—a disagreement pertaining to 1) and 10).
6 For textual evidence of Tadros’ endorsement of the first part of 1**) [a], see (Tadros 2011: 12).7 Tadros states that the duty view shares with (non-consequentialist types of) retributivism the idea thatrespect for individuals means that one should not punish the innocent or punish offenders disproportionately(Tadros 2011: 42). Presumably, this is the motivation for the duty view as I have stated it here, not the viewitself.8 Cf. Tadros 2011: 39.
Crim Law and Philos
123
When we look at matters in this way there seem to be two crucial differences between
the duty view and retributivism. First, both views endorse some kind of proportionality
condition, and on both views one of the relevant relata of the proportionality is the degree
to which the wrongdoer is made to suffer. However, while retributivists think that the other
relata is the gravity of the wrong perpetrated (Tadros 2011: 25), the duty view theorist says
it is the harm involved in the wrongdoing. These can diverge. An attempt to kill someone
that merely results in an almost unnoticeable bruise might be a very serious offence even if
the harm involved is very small. Tadros also mentions a separate proportionality consid-
eration relating to the duty view, i.e., the proportionality between the level of harm
imposed on offenders and the goods that punishment might secure (Tadros 2011: 333,
337). Accordingly, Tadrosian proportionality, involves a triadic relation involving harm
suffered by the victim in addition to the two relata just mentioned.
The second difference between the duty view and retributivism concerns the way they
explain the involvement of the proportionality requirement. Here retributivists appeal to
the value of the deserved sufferings of wrongdoers, while the duty view appeals to the
duties of wrongdoers. This means that, in comparing them, we need to focus on which of
1a) and 10a), and which of 1b) and 10b), is the most plausible claim.
I recognize that this definition of the duty view reflects my interpretation of Tadros’
duty view and a reconstructive twist motivated by what I see as the desirability of
maintaining a homogenous distinction between the two views.
Do Retributivists Get the Criticisms They Deserve?
In general Tadros’ discussion does justice to retributivism as well as to the duty view, and
his arguments are admirably balanced. However, in a number of cases I wonder if retri-
butivists might complain that their theory gets a less charitable hearing than the duty view.
First, consider Tadros’ dismissal of the core retributivist claim that proportionate suffering
in wrongdoers is valuable. Suffering is normally morally bad in itself. However, in the case
of wrongdoers, when the suffering is suitably related to their wrongdoing, episodes of
suffering suddenly become valuable, according to retributivists. This claim, however,
seems to involve—to put it as Tadros does quoting H. L. A. Hart with approval—‘a
mysterious piece of moral alchemy in which the combination of the two evils of moral
wickedness and suffering are transmuted into good’ (Tadros 2011: 73).
Retributivists might respond, first, that, whether it is ultimately defensible, this kind of
alchemy is not unusual.9 All it involves is the notion that the moral value of a certain
feature depends on the presence of other features, and this is a theoretical possibility
retributivists are hardly alone in countenancing. Prioritarians assert that the moral value of
a given prudential benefit depends on how well off the beneficiary is, for example.10 More
importantly, Tadros himself makes similar claims of context-dependence. In his discussion
of the communicative view of punishment he notes that: ‘[a]lthough it is sometimes true
9 Retributivists who subscribe to Moore’s principle of organic unities might also deny that ‘the moralvalence of suffering is reversed by wrongdoing’ (Tadros 2011: 73). Suffering is bad whoever the sufferer is,but the organic unity of someone’s suffering as a result of punishment and that person’s being a wrongdoeris good, or at least less bad than the organic unity of someone’s suffering as a result of punishment and thatperson’s being innocent. With regard to parts of valuable wholes Moore puts is as follows: ‘The part of avaluable whole retains exactly the same value when it is, as when it is not, a part of that whole’ (Moore1989: 30). I am grateful to Raffaele Rodogno for this observation.10 See also (Dancy 1993; Dancy 2004; Kagan 1988; Kamm 1996: 67–68).
Crim Law and Philos
123
that it is wrong to make a person feel guilty for their wrongdoing, few people will think
that the fact that feeling guilty is unpleasant provides a very strong reason against con-
demning offenders. We typically need not take the intrinsic badness of the affective
dimension of guilt into consideration in deciding whether to condemn the offender’
(Tadros 2011: 94). What Tadros contends here is that while, as believed by many, the fact
that guilt makes one feel bad is a fact that normally counts against performing an action
resulting in this feeling, in the case of condemning offenders for their wrongdoing this fact
does not count against condemning them.11 Similarly, he argues that the ‘sexual pleasure
rapists get from raping … does not count as a good’ (Tadros 2011: 124), so he certainly
appears to believe that the moral valence of pleasure depends on context. In my view, then,
retributivists might challenge Tadros to explain why—given that the pleasure that rapists
get from raping is not intrinsically good—the suffering of rapists incurred as a result of
raping, through punishment among other things, may not at least be not intrinsically bad,
and perhaps even intrinsically good.12
Secondly, one might worry that Tadros’ charge of moral alchemy has no bite against more
moderate retributivist views than the one he criticizes—e.g., against the non-comparative
view that the punitive sufferings of wrongdoers are less bad, non-instrumentally speaking,
than the sufferings of others, but not good in themselves, and against the comparative view
that it is in itself bad if wrongdoers are as well off as non-wrongdoers. On the comparative
view, benefits to wrongdoers may have as much moral value as benefits to non-wrongdoers. It
is just that the state of affairs in which if they both enjoy the same level of benefits is, in one
respect, non-instrumentally bad. Analogously, telic egalitarians may think that benefits to the
well-off and the badly-off have the same non-instrumental, moral value, it is just that benefits
to the well-off increase unjust inequality and, thus, in one respect increase the non-instru-
mental badness of the distribution (Parfit 1998: 3). Indeed, on this comparative, retributivist
view, punishment might not be justified if we could better restore the appropriate inequality
between wrongdoers and non-wrongdoers by benefiting non-wrongdoers, rather than by
harming wrongdoers punitively.13
Tadros does not base his rejection of retributivism simply on the presumed mystery of
suffering being bad in some contexts and good in others. Drawing on work by Scanlon, he
suggests, further, that one can provide an account of why it may seem good in itself that
wrongdoers suffer even if, as a matter of fact, it is not—he explains away an attraction of
retributivism, in other words. The account says (1) that victims have a right that wrong-
doers recognize their wrongdoing as such and (2) that it is in itself good that wrongdoers
gain the self-knowledge involved in knowing that they have acted wrongly. Given facts
about the human psyche, recognizing that one has done wrong almost inevitably involves
frustration at not being able to undo one’s past wrongdoing, and anxiety as well (Tadros
11 Tadros’ context-sensitivity here concerns reasons, not values. However, it is hard to see why reason–context–sensitivity should be any less of a mystery than value–context–sensitivity.12 Admittedly, Tadros does not make the stronger claim that a feature which is good (a reason for per-forming a certain action) in one context can become bad (a reason against performing it) in another, but weare dealing with the claim that what is a large good in one context can become a small good in another, andonce this concession is made it is hard to see why, in principle, it could not also be the case that a featurewhich is good in one context could be, not just less good in another context, but bad.13 One might think that for this reason the comparative view would not really be retributivist, becauseretributivists want to show more than that punishment is justified in principle. However, to insist on thiscommitment is to go beyond what Tadros’ definition of retributivism involves. Also, the commitment wouldmake the distinction between retributivism and the duty view heterogeneous if the latter is only committedto punishment being justified in principle while the former is committed to something stronger than that.
Crim Law and Philos
123
2011: 45). Hence, our sense that something has gone wrong when a wrongdoer does not
suffer does not reflect a belief that it is in itself good that the wrongdoer suffers. It rather
reflects the fact that a wrongdoer’s not suffering is a reliable sign that she has failed to
fulfill her duty to her victims to recognize her wrongdoing, and an indication of her
deficient participation in the good of self-knowledge (Tadros 2011: 50).14 As Tadros puts
it: it is the judgments [that it is appropriate to make in the light of one’s wrongdoing] alone
that are valuable. Craving and anxiety, in many people, are an inevitable side effect of
making the appropriate judgments. But they are intrinsically disvaluable effects of the
appropriate judgments rather than a good to be sought for their own sake’ (Tadros 2011:
49).
This account might well explain in part why we find it disconcerting when a wrongdoer
does not suffer. However, I find the error theory on offer unconvincing insofar as it is
employed, as it is by Tadros, to support something more ambitious than this rather modest
claim. For one thing, many wrongdoers see, and so do not fail to recognize, that what they
have done is wrong. Often enough they wrong others for self-interested reasons, knowing
quite well that they are acting in a way that is morally impermissible.15 Surely the retri-
butivist will want to say that it is bad if such offenders are not made to suffer through
punishment.
Second, if Tadros’ error theory were correct, it would presumably follow that if we
could invent a pill that would slightly enhance people’s cognitive capacity to recognize
their own wrongdoing, but as a side-effect cause them not to suffer any anxiety and
frustration at all as a result of this realization, we should be in favor of giving wrongdoers
the pill. I suspect, however, that people with retributivist inclinations would not favor this.
They would not be enthusiastic about the suggestion that Hitler should have been given
such a pill if he had been caught alive.
Third, against retributivism, Tadros points to the implausibility of the idea that it is good
in itself when a wrongdoer, as a result of some punishment-unrelated event, comes to
suffer. He claims that this idea is something retributivism, at least in one version worth
discussing, is committed to (Tadros 2011: 68–72). However, it seems to me that Tadros
cannot make this criticism while offering the error theory presently under scrutiny. Clearly,
the error theory does not explain why we might find it bad if a wrongdoer suffers from
anxiety and craves to undo past wrongdoing yet, at the same time, comes to enjoy other
benefits that more than offset these harms, because in such a case the offender’s being
better off is no indication of his or her failure to recognize his or her wrongdoing as such.
Hence, if retributivists are inclined to think this is bad, Tadros’ error theory is flawed if it is
being offered as a general account of retributivist intuitions as such. He must choose
between his error theory and the present criticism.
Tadros also suggests—and, among the arguments he marshals against retributivism, this
is the third one I find unsatisfactory—that we can justify the view that, when harm will
have to fall on someone, it is better that it falls on wrongdoers than on others. Obviously,
this view seems to cohere well with retributivism if we assume that wrongdoers deserve to
suffer while those who are not wrongdoers do not. Tadros, however, offers a Scanlonian
explanation of the view in question which does not appeal to desert, but rather to the value
of choice: ‘where a person has culpably imposed a threat on another person the person who
14 Note that this story cannot explain why retributivists find it bad if punishment fails to inflict suffering onthe punishee, since this failure may not indicate the punishee’s lack of recognition of wrongdoing.15 The internalist position on moral judgment and motivation that if one recognizes that it is wrong to act ina certain way, then one does not so act is an implausibly strong position.
Crim Law and Philos
123
has created the threat could have avoided being liable to be harmed simply by conforming
to what morality already required of them’ (Tadros 2011: 56). It is not similarly the case
that the person threatened by harm could easily have avoided the harm in question; and
because the significance of choice account explains the relevant view about the right
distribution of harm, and because ‘judgements about choice have broader explanatory force
than desert … we should find the significance of choice [account] more appealing than the
desert [account]’ (Tadros 2011: 57).
I do not find this argument decisive in the form in which it is presented. First, for some
situations it may be difficult—perhaps difficult and costly—not to impose a threat on
another person. Thus it may be that the people in question hate the other person’s guts and
find it difficult and exhausting to control themselves. Equally, some people may have an
easy, cost-free way of avoiding having a threat imposed on them. We can imagine that
each month they can swallow a threat-preventing pill as a result of which they will be
immune to threats for the next month. On the significance of choice account it is not clear
that harm, in this kind of case, when it has to fall on someone, should fall on wrongdoers.
Yet, presumably, retributivists think that the harm should fall on the wrongdoers, in which
case the choice account does not really account for the indicated view—at least, not in the
circumstances just imagined.16
Second, the value of choice account may suffer from other problems which Tadros and
others see as reason to reject the account. Plainly problems like these will make it hard for
Tadros to invoke the account to unseat retributivism. Consider a case of self-defense of the
following kind. Faced with an aggressor, I have two options: I can either shoot him with a
lethal weapon; or I can use a stun grenade that will save my life, harming the aggressor
somewhat, at significant cost to myself—we can suppose that I would lose my hearing in
the left ear for 1 month.17 Most of us would agree that in cases of self-defense one is
morally required to defend oneself, if one can, without killing the aggressor. Yet the value
of choice account is unable to justify this view. All of the considerations we have seen
Tadros invoke in an effort to show that harm should fall on the aggressor, not the victim,
speak in favor of the idea that the latter should not bear any costs at all. It might be
suggested that the value of choice account can also accommodate our interest in protecting
ourselves against our own faulty choices. However, given the ambition to provide an
account of intuitions about the morally desirable distribution of harm that would seem best
accounted for otherwise through the consideration of desert, this suggestion appears to
make it difficult to say which principles apply to people who are very likely to make faulty
choices and, thus, have a great interest in protecting themselves against such choices.
Alternatively it might be retorted that while significance of choice is one factor that bears
on permissibility, it is not the only one. Specifically, in the case at hand, welfare con-
siderations might be invoked alongside the significance of choice to explain the relevant
intuition.18 While this reply may be promising, it creates a new problem of explaining how
16 It might be responded that the degree to which the value of all options count depends on the moralqualities of the actions involved in the relevant options (cf. Tadros 2011: 56). I am not able to respondadequately to this suggestion beyond reporting that it seems to undermine the ambition to provide a value-of-choice-based account of the moral qualities of actions (if such an account rests on assumptions about themoral qualities of actions not explained through the value of choice account).17 I accept that, as Tadros shows, retributivists will (also) find it hard to explain why one may not use, say,lethal harm to defend oneself against an offender whom one knows to believe falsely, of himself, that he isfiring at one with live ammunition when in fact he is only firing rubber bullets that will cause at most a slightbruise (Tadros 2011: 176–177).18 I am grateful to Victor Tadros at this point.
Crim Law and Philos
123
to weigh the different factors being invoked against one another. If no such account is
provided, an appeal to the significance of choice offers an incomplete explanation.
Third, I think we need to be cautious about the claim that the value of choice account
explains more—or, in Tadros’ phrase, has ‘broader explanatory force’—than the desert
account. There are cases in which the desert account seems to apply yet the value of choice
account does not. Consider, for example, situations in which people are more or less
deserving in virtue of the moral qualities of the deliberations they have undertaken prior to
choosing. But even if this were not so, and there was a clear sense in which the value of
choice account explains more than the desert account, it would not obviously follow that
we should accept the former and reject the latter. Could a moral claim not be explained by
two logically independent accounts, or principles, one of which is capable of explaining a
wider range of phenomena than the other?
In making his case against retributivism, Tadros also presents a more direct argument.
Although this is more direct, the conclusion is not that retributivism is false, but rather that we
have no reason to believe it. (This is the fourth and final of Tadros’ arguments against retri-
butivism about which I will express skepticism in this paper.) The argument runs as follows:
1. ‘Retributivist arguments suppose a connection between free will and responsibility and
desert.
2. Retributivists believe that desert claims are evidenced by our intuitions.
3. There is good reason to believe that our intuitions about desert are caused by the folk-
psychology conception of free will and responsibility.
4. The folk-psychology conception of free and responsibility is false.
5. There is good reason to doubt the trustworthiness of our intuitions about desert claims.
6. Retributivists lack a good reason to believe desert claims’ (Tadros 2011: 63).
While Tadros makes clear that this is not intended as a knockdown argument (Tadros 2011:
63), it is fair to say that he could have explained more fully how the argument works. First,
I am not sure what 1) adds to the argument. Second, structurally, I think 5) is best regarded
as a conclusion entailed by 3) and 4) together with a suppressed premise:
4a. If a certain set of intuitions is caused by a conception that is false, then there is good
reason to doubt the trustworthiness of these intuitions.
I also think 2) can be advantageously modified in the following way:
2*. Retributivists believe that we have good reason to believe desert claims because of,
and only because of, our intuitions about desert.
With this modification, 6) becomes a conclusion possibly entailed by 5) and 2*). We can
then isolate the following, possibly deductively valid, argument19:
3. There is good reason to believe that our intuitions about desert are caused by the folk-
psychology conception of free will and responsibility.
4. The folk-psychology conception of free will and responsibility is false.
4a. If a certain set of intuitions are caused by a conception that is false, then there is good
reason to doubt the trustworthiness of these intuitions.
5. Hence, there is good reason to doubt the trustworthiness of our intuitions about desert
claims.
19 I say ‘possibly’, because I am uncertain that 5) and 2*) entail 6). One reason is that in an internalist senseof ‘good reason,’ 5) does not say that retributivists have good reasons to doubt the trustworthiness of ourintuitions about desert claims.
Crim Law and Philos
123
2*. Retributivists believe that we have good reason to believe desert claims because of,
and only because of, our intuitions about desert claims.
6. Retributivists lack a good reason to believe desert claims.
Even so revised, the argument is not compelling. To begin with, compatibilists will simply
deny that our intuitions about desert are caused by the folk-psychology conception of free
will and responsibility, where these are assumed to involve some kind of contra-causal
freedom. After all, most compatibilist accounts of free will and moral responsibility
involve denial of the claim that our folk-psychological conception of free will and
responsibility involves any assumption of contra-causal freedom.20 So the compatibilist
will almost certainly reject 4). Nor does Tadros offer any arguments for an incompatibilist
interpretation of our folk-psychological conception of free will and responsibility.21 And if
such an argument could be given, desert theorists would presumably deny 3), anyway, on
the standard compatibilist ground that we often distinguish between the deserving and the
undeserving on bases that do not involve any mention of contra-causal freedom.22 Tadros
allows that we can justify blaming even if determinism is true, but he insists that no such
justification can be given for desert.23, 24 However, this claim is a claim about justification,
not causation, and therefore it lends no support to 3).
A further point can be made ad hominem. As noted above, Tadros holds that the sexual
pleasure of rapists has no moral value. So one might ask why this view does not rest on a
contra-causal conception of free will and responsibility, if retributivist intuitions do as
Tadros also holds. It would seem surprising if these intuitions differed in their relationship
with our putatively contra-causal folk-psychological conception of free will and respon-
sibility. On the face of things, Tadros must either retract his view that the rapist’s sexual
20 Take Hume, for instance (1999: 148–164), on freedom, and Harry Frankfurt (1988:1–10) on moralresponsibility. Presumably, their claims are claims about our folk-psychological conceptions of freedom andmoral responsibility, respectively. If (something like) their analyses are correct, it would be surprising if ouridea of desert presupposes a contra-causal notion of freedom and responsibility rather than our folk-psychological one.21 Tadros refers to the views of Galen Strawson, Ingmar Persson, and Derek Parfit that the ‘folk-psychologyconception of free will grounds what is called either contra-causal or ultimate responsibility’ (Tadros 2011:62), but he does not offer a defense of this claim. For a recent staunch denial that the folk-psychologyconception of responsibility is contra-causal or involves ultimate responsibility, see (Hurley 2003: 80–105).22 See, for instance, P. F. Strawson’s discussion of the ‘self-reactive attitudes’ of offenders who acquiesce intheir punishment without developing any sense of resentment (Strawson 1982: 77–78).23 In wrapping up the section on free will and desert Tadros reports that he knows ‘of no similar argument[to that of why choice is significant even given the falsity of contra-causal conceptions of free will] that canbe mounted in favour of the idea that we should continue to accept that it is intrinsically good that offenderssuffer [even given the falsity of contra-causal conceptions of free will]’ (Tadros 2011: 66). However, it issomewhat surprising to find his argument ending on this note, since the argument against desert he offersseems not to rely on this claim. In the same paragraph he also suggests that ‘[i]t would surely be better if wecould justify punishment on grounds that are less controversial than desert’ (Tadros 2011: 66). Again, thismight be so given that we have as an aim the defense of punishment. But this point is compatible with theview that retributivism is true and that we have good reasons to accept it, even if these reasons that are morecontroversial, i.e., fewer people endorse these reasons, than the reasons we might have for accepting otherjustifications for punishment.24 It is not entirely clear whether he wants to argue that [i] blame and the significance of choice does notdepend on ‘the truth of compatibilism’ (Tadros 2011: 64) or that [ii] ‘there is reason to be optimistic that arobust compatibilist account of blame [and the significance of choice] can be developed’ (Tadros 2011: 64).The latter is consistent with denying the former, because one might think that blame requires the falsity ofincompatibilism (i.e., deny [i]), but also think that compatibilism is actually the true theory (i.e., affirm [ii]).
Crim Law and Philos
123
pleasure has no moral value or concede that retributivist intuitions need not rest on a
contra-causal, folk-psychological conception of free will and responsibility.
In any case, Tadros’ conclusion does not quite deliver what he seems to want it to
deliver. He notes that Parfit, among others, accepts: ‘P: If our acts are merely events in
time, we cannot have chosen our own character, or be responsible for our acts in any way
that could make us deserve to suffer’ (Tadros 2011: 62). He seems to accept P (and I am
drawn to it myself). But he also concedes that Parfit does not defend it, and that ‘[m]any
retributivists, and other defenders of desert, simply deny P’ (Tadros 2011: 63). Given that
latter concession, it is clear that Tadros needs to make the case needs for P.25 However,
strictly speaking, the argument he provides does not establish that P is true; at best it shows
that retributivists have no reason to believe that P is not true.
To sum up, Tadros’ arguments against retributivism are inconclusive. Nor does he
always succeed in putting the duty view to a test that is as stringent as the one to which he
subjects retributivism. I have not addressed all of his arguments against retributivism (cf.
Tadros 2011: 60–61). In particular, I have not addressed his forceful contention that,
whatever appeal the idea that it is good in itself that wrongdoers suffer may have, it loses
this appeal once we specify concretely what kind of suffering—the suffering that offenders
endure because they are separated from their loved ones when imprisoned, or some other
kind—we have in mind (Tadros 2011: 75–78). Also, it should be noted that not all of
Tadros’ criticisms of retributivism attacks its core claim that it is non-instrumentally good
that wrongdoers suffer, e.g., other arguments criticize the views that the state is permitted
to make or capable of making citizens’ level of well-being match their level of desert.
However, my aim in this section has not been to defend retributivism as such, but, less
ambitiously, to challenge some of the reasons Tadros gives for rejecting the view.
The Means Principle and Punishment
In the most succinct statement Tadros offers, the means principle says: ‘it is worse to harm
a person intentionally than it is to harm them as a side effect of one’s actions’ (Tadros
2011: 13; cf. Tadros 2011: 114).26 The principle plays an important negative role in
Tadros’ justification of punishment. It does so because instituting punishment for the
purpose of deterrence seems to violate the means principle: wrongdoers’ harm becomes a
means to greater security. However, in Tadros’ view, the scope of the means principle is
restricted. It is not wrong to inflict harm on a person as a means when that person has an
enforceable duty to suffer that harm and moreover would be morally required to inflict this
harm on him- or herself to avoid a bad outcome.
25 ‘I am not sure what argument we can provide [i] for P that would convince retributivists that [ii] they arewrong to deny that desert rests on a false conception of free will and responsibility’ (Tadros 2011: 63).Tadros then proceeds to present the argument quoted in the main text as an argument of the type to beneeded. However, [i] and [ii] are different claims, and an argument for [ii] is not necessarily an argument for[i]. In particular, the conclusion of the argument offered by Tadros leaves entirely open what the correctaccount of whether desert presupposes contra-causal freedom is.26 Tadros believes ‘it is permissible to harm a person as a means to an end if that person would have had anenforceable duty to avert the threat were she able to do so, even if, in exercising her duty she would beharmed to the same degree’ (Tadros 2011: 129). However, it is unclear if this view means we should revisethe quoted statement of the means principle so that it says it is worse to harm a person intentionally than it isto harm him or her as a side effect of one’s actions unless the person has an enforceable duty to suffer thisdegree of harm to avert some relevant threat, since the permissibility claim is compatible with the view thatit is still harder to justify harming as a means even though the person has an enforceable duty to suffer theharm to remedy the wrong or avert a threat that he or she created.
Crim Law and Philos
123
Tadros seems to adopt the means principle primarily on the grounds that it provides the
best explanation of certain moral intuitions we have (Tadros 2011: 13, 148).27 For instance,
confronted with the famous case devised by Phillipa Foot,28 most of us think it is per-
missible for the driver to divert the trolley off a track with five innocent people on it and on
to another track with just one innocent person on it. Yet, most of us also consider it
impermissible to push a big person off a bridge that crosses a track and into the path of an
oncoming trolley in order to save five innocent people on the track further down. The best
explanation of this difference is that in the bridge case, but not the driver case, the agent
uses the one person as a means of saving the five.
I am not persuaded that this is the best explanation. To explain why, I would ask you to
imagine a variant of the driver case where the driver intends to use, or indeed actually uses,
the one person as a means, and then compare it with a variant of the bridge case in which
the pusher neither intends to use, nor uses, the person on the bridge as a means. How would
these cases look? In the first variant—the driver* case—suppose that the driver is fanatical
about flowers, and that he believes that diverting the trolley onto the track with one person
on it will save some rare flowers. For example, he believes that the track with five persons
rejoins the track with one person at some point after the spot where the one person is
located but before the spot where the rare flowers grow. For some reason, hitting the five
persons would not stop the trolley, but hitting the one person would do so, and stopping the
trolley is a means of saving the rare flowers. Assuming these beliefs are false, the trolley
driver intends to use the single person, whom he kills. Assuming they are true, the trolley
driver actually uses the single person, whom he kills, as a means.
In the second variant we require—the bridge* case—the pusher does not intend to use
one person as a means. We can suppose that he is himself standing on a track and is
threatened by being hit by a trolley. The only way he can avoid being hit by the trolley is
by jumping into the big man on the bridge, effectively pushing him off. If he does so, the
big man will fall and be hit by the trolley, and killed as a result, but the five persons as well
as the pusher will survive.29
I suspect that people who share Tadros’ intuitions about the driver and bridge cases will
draw similar conclusions about the driver* and bridge* cases. That is, they will believe that
the action of the agent in the driver* case is permissible (even though, as they may come to
see, he displays a flaw in reasoning about what to do in the way he does that), while the
action of the agent in the bridge* case is impermissible. This is interesting, because in the
variant cases the means feature is reversed: now, the driver* uses or intends to use the one
person as a means and the pusher in the bridge* case neither intends to use, nor actually
27 In his critique of the retributivist view that the well-being of offenders should be reduced in proportion totheir offenses, Tadros notes that there are other possibilities—e.g., that their rights should be restricted, thatthe well-being independent value of their life should be reduced, or that their non-self-regarding preferencesshould be frustrated in proportion to their offenses. In canvassing their view that it is well-being that counts,retributivists ‘will once again resort to intuition’. Tadros describes this as ‘unsatisfying’ and insists that areason is needed here (Tadros 2011: 74). If so, similar remarks apply to Tadros’ endorsement of the meansprinciple. When we determine which kinds of harm matter for the purpose of applying that principle (e.g.,does frustrating someone’s non-self-regarding preferences as a means violate the means principle?) we runinto the very same question (e.g. Tadros 2011: 130).28 Philippa Foot, ‘The Problem of Abortion and the Doctrine of the Double Effect’, Oxford Review, Number5, 1967; reprinted in Virtues and Vices (Oxford: Basil Blackwell, 1978).29 The fact that six, not five, people will be saved if the big man is pushed off the bridge means that thenumber of lives saved is not equalized across the driver* and bridge* cases. However, given my use of thecases, this is not a problem, because this asymmetry would, if anything, render pushing the big man over thebridge more likely to be permissible.
Crim Law and Philos
123
uses, the one person as a means. As might be guessed at this point, my claim is that if our
intuitions about permissibility remain stable even when we reverse the means feature, the
means principle cannot explain those intuitions.30
Tadros might respond as follows. It would be ‘an option [i.e., a morally permissible
option: KLR] for a person who is properly motivated by the moral reasons that apply to
him’ (Tadros 2011: 156) to divert the trolley in the driver* case. This is what explains our
inclination to say that diverting the trolley is permissible. But the agent’s actual diversion
of the trolley is impermissible because it involves using a person as a means.31 I suspect,
however, that many would say something stronger than this. Many would say that even for
an improperly motivated agent of the sort we imagined diverting the trolley was an option
in the sense that others ought not to try to prevent that agent from diverting the trolley, etc.
However, even if this suspicion is wrong, there is still room for doubt about the idea that an
appeal to the notion of what is an option for a properly motivated agent amounts is
admissible in a defense of the means principle—a principle which concerns not what a
properly motivated agent intends as a means, but what the agent actually intends as a
means, however improperly motivated he or she may be.
Two things should be noted about this criticism of Tadros’ defense of the means
principle. First, I am simply raising doubts about the main argument he offers in its
defense. I am not denying the principle as such, nor that there might be other (pairs of)
cases where we will find it hard to account for differential permissibility without invoking
it.32 Second, while Tadros is committed to the means principle independently of his views
about the justifiability of punishment, and while the principle’s grounding is worth
exploring independently of its implications in the punishment debate, the present criticism
does nothing to undermine the achievement of Tadros’ main argumentative aim—his aim,
that is, to show why punishment is morally justified in principle. The means principle
functions as a constraint on punishment, rendering it harder, not easier, to justify. Indeed
Tadros’ purpose is to explain why ‘self-defense is an exception to the means principle’
(Tadros 2011: 14) and then use this exception to explain why punishment theory can
exploit a similar exception. That said, the means principle is crucial to Tadros’ rejection of
pure instrumentalist, i.e., consequentialist, justifications of punishment; so while he may
not need the principle to show why punishment is justified in principle, it does play a key
role in unseating consequentialist alternatives to his own view.
The Means Principle and Moral Status
Tadros does not rely solely on intuitions favoring the means principle. Drawing primarily
on the work of Frances Kamm, he constructs a deeper rationale for the principle appealing
to the moral status of human beings. He asks: ‘For many non-consequentialists, the meansprinciple provides a constraint on acting. But why is there such a constraint?’ (Tadros
30 It might be said in response that my argument fails to show that harming people as a means has nobearing on moral permissibility; it only shows that such harming is not the only thing with that bearing. Iaccept this point, which is compatible with the conclusion I want to establish in this section. Moreover, Itake it that harming as a means makes a significant difference to permissibility on Tadros’ view. If this iscorrect, there are limits to how much he can rely on the present point.31 Similarly, he might say that in the bridge* case the agent’s action is impermissible because it is not ‘anoption for a person who is properly motivated by the moral reasons that apply to him’ to push the big manover the bridge.32 See, for instance, my remarks in the next section on inflicting pains on animals as a means.
Crim Law and Philos
123
2011: 122) He answers: ‘If it were permissible for me to harm you as a means in order to
prevent five others from being harmed as a means, each of us would be available to be
treated as a means—we would each be available to be harmed for the sake of the greater
good. So each of us would have the same status as an object—available for use by others to
promote the good’ (Tadros 2011: 124).33 We have a more elevated status than objects
have. It is grounded in ‘capacities that are distinctive of human beings’, notably that we are
‘capable of determining what ends to set for ourselves’ (Tadros 2011: 127).34 In fact,
‘[e]ndorsing the means principle is essential to maintaining a certain kind of moral status’
(Tadros 2011: 13).
I do not find this persuasive. First, Kamm herself does not tie status to a justification of
the means principle specifically, and rightly so in my view. Rather she links status and the
justification of rights in general, and the agent-relative constraints that these involve
(Kamm 1996: 271–282; Kamm 2007: 28).35 So, for instance, if, as some people think,
there is an agent-relative principle against doing, as opposed to allowing, harm, Kamm
thinks that what explains why we are not allowed to harm one person (whether as a means
or not), even when that will reduce the overall number of cases where an agent harms
someone, is our moral status. It can be seen, then, that Kamm relies on considerations
relating to moral status to explain a formal feature of the agent-relative constraints: that it
is impermissible for one to violate an agent-relative constraint even when that will prevent
a greater number of violations from taking place.36 This explanation does not pertain to the
content of the relevant agent-relative constraints. As far as considerations relating to moral
status go, the content of those agent-relative constraints that pertain to persons might
concern harming, violating rights, lying, coercing or harming as a means. The concern
about moral status is neutral on this matter and cannot explain why the relevant agent-
relative principle bears upon harming as a means.
One can bring this out is by thinking about the infliction of pain on animals as a means.
Intuitively, the means principle seems to apply to non-human animals as well as human
beings: inflicting pain on animals as a means to an end is harder to justify than inflicting
pain on them as a by-product.37 However, one might suspect that in the case of animals the
means principle does not take an agent-relative form on the grounds that, in the animal case
33 Some readers might be puzzled as to why Tadros suggests that ‘[a]gent-relative constraints are verydifficult to motivate’ (Tadros 2011: 117; cf. Tadros 2011: 125n19, 126). The considerations that he offersseem intended to motivate agent-relative constraints as they are normally understood and were once set outin Scheffler (1982: 80): ‘An agent-centred restriction is a restriction which it is at least sometimes imper-missible to violate in circumstances where a violation would prevent either more numerous violations, of noless weight from an impersonal point of view, of the very same restriction, or other events at least asobjectionable, and would have no other morally relevant consequences’. Tadros, it seems, is aiming tojustify agent-relative constraints so construed. However, he wishes (and this is what I take to be the thoughtbehind the passage quoted in the first sentence in this footnote) to ground the source of agent-relativeconstraints in non-agent-relative moral considerations (cf. Kamm 2007: 29).34 Tadros recognizes that human beings with such distinctive capacities pose a problem for this view, butcursorily suggests that ‘[i]t may be that their relationship with those who have the relevant capacitiesgrounds the application of the means principle to them’ (Tadros 2011: 127n).35 Admittedly, she also discusses moral status in relation to violations of the principle of permissible harm-rights (Kamm 1996: 264) and suggest that, like an agent-relative prerogative, agent-relative constraintsderive from ‘a person’s entitlement to control something, because he is not merely a means to a greatergood’ (Kamm 1996: 262). Of course, these observations also show that Kamm does not tie moral statusspecifically to the means principle.36 The feature is brought out in Samuel Scheffler’s definition of agent-relative constraints: see footnote 33.37 Cf. (Tadros 2011: 127) where he denies that there is an agent-relative constraint of harming animals as ameans. It is not clear from this discussion, however, whether he shares the intuition to which I appeal.
Crim Law and Philos
123
one, is permitted to inflict pain on one animal as a means if that will prevent more cases of
infliction of pain on animals as a means. If this suspicion is correct it shows that, as Tadros
believes, animals do not have the elevated moral status that human beings have, yet the
means principle still applies (in a non-agent-relative form) to animals.
Second, even if moral status explains not simply the form of the moral constraint against
using an individual as a means but also the content of it, I do not think the appeal to
considerations relating to moral status can carry much argumentative weight. For one
thing, the claim that we have a moral status ensuring that we cannot permissibly be harmed
as a means even when doing so will reduce the overall number of cases where someone is
harmed as a means seems, certain embellishments aside, simply to amount to saying that it
is not permissible to harm someone as a means even when that will reduce the overall
number of cases where someone is used as a means. But this is exactly the principle that
those who reject the means principle reject. Hence, it seems that the appeal to consider-
ations relating to moral status begs the question against these deniers of the means
principle.
The charge of question-begging might be denied on the grounds that it overlooks the
benefits we derive from having this moral status recognized by others. We benefit when
others recognize us as having an elevated moral status, among other things as a result of the
pleasure and self-respect we derive from that recognition. Presumably, many of those who
reject the moral status account would agree that these features are morally relevant ben-
efits, so an argument appealing to the benefits would not beg the question against them.
While, for the most part, Tadros is clear that he is not prosecuting this line of thought,
he does make statements like the following: ‘If we reject the permissibility of harming one
person as a means even to save five others from being harmed as a means to the greater
good, our status remains intact’ (Tadros 2011: 125).38 This formulation suggests that if we
do not reject, or perhaps even endorse, the permissibility of harming one person as a means
of saving five others from being harmed as a means to the greater good, our status will notbe intact and this again would suggest that the moral status argument appeals to consid-
erations of the sort I mentioned. However, elsewhere Tadros makes clear that our moral
status is unaffected by what we take, or do not take, to be morally permissible. In any case,
if the moral status argument did appeal to the benefits we derive from the recognition of
our inviolability, the basic argument for agent-relative constraints would surely be rooted
in the promotion of status benefits; it would then display a consequentialist character
incompatible with the conclusion it is supposed to establish.
The third point I want to make is this. Tadros does not subscribe to an absolutist version
of the means principle (Tadros 2011: 138, 169).39 So while he denies that it is permissible
38 Compare this statement: ‘[o]ur desire to maintain our moral independence through endorsing a set ofrules that conform to the means principle may be less significant than the value that we put on avoidingsome much greater harm’ (Tadros 2011: 138).39 Given his critique of Jean Hampton, I wonder if he can consistently take the means principle to provideany more than weak moral reasons. Hampton thinks that lack of respect for one’s victim is important in thejustification of punishment. Tadros expresses skepticism about this on the grounds that ‘if lack of respectdoes not erode moral status it is difficult to see why the obligation to protect people against lack of respect isvery significant in itself’ (Tadros 2011: 108). But this question applies not only to protecting victims fromthe disrespect of other agents, but also to one’s own disrespecting of others. If, as Tadros thinks, all retaintheir moral status irrespective of whether I treat them disrespectfully, it would seem, given Tadros’ critiqueof Hampton, that my reason not to disrespect them is not very weighty—in which case it should bepermissible, say, to use someone as a means when that will prevent two or more cases in which someoneuses another as a means. After all, the one person I use will retain his or her status despite being used as ameans.
Crim Law and Philos
123
to harm one person as a means even when that will prevent five similar cases of harming by
others, he believes that it would be permissible to do so to prevent a sufficiently large
number of cases where an agent harms others as a means. But obviously a defender of the
means principle will have to say something about the ‘sufficiency threshold’ here (cf.
Lippert-Rasmussen 1996). If our moral status—our being more elevated than mere
objects—explains why one is not entitled to harm someone as a means when that would
prevent five similar cases, why does it fail to rule out expedient harmings when the number
of people shielded from harm, and hence the net benefit, is very large?
Tadros agrees that the worry that ‘any restrictions on the scope of the [means] principle
are inevitably ad hoc’ (Tadros 2011: 117) is important. At the same time he addresses the
threshold problem here—moreover, in a way which he takes to strengthen the case for the
means principle considerably. He contends that ‘it is permissible to harm a person as a
means to an end if that person would have had an enforceable duty to avert the threat were
she able to do so, even if, in exercising her duty she would be harmed to the same degree’
(Tadros 2011: 129). Presumably, a person who could prevent, say, one million people from
being harmed as a means by making sure that she would suffer the same harm as a means
has an enforceable duty to suffer that harm as a means. Hence, on the account of the limits
of the means principle that Tadros offers we have an explanation of why we are not
maximally inviolable, i.e., why the impermissibility of using one of us as a means gives
out, at some point, as the number of people saved from harm rises.
The explanation is certainly supportive of the means principle, but I suspect that Tadros
has, in effect, simply shifted an explanatory challenge. True, on Tadros’ account it is no longer
necessary to explain why we have the degree of inviolability that we have. However, we must
now explain why we do not have an enforceable duty to be harmed as a means when that will
prevent the similar harming of five others, but do have an enforceable duty to do so when that
will prevent one million relevantly identical cases of harming others as a means. In other
words, why is the level at which the possibility of avoiding more cases of harming as a means
turns what would otherwise be a supererogatory action into an enforceable duty located where
it is? Surely, if we are puzzled by the threshold at which the means principle gives out, we
should be equally puzzled by the threshold at which an enforceable duty of being harmed as a
means when that would prevent more cases of someone’s being used as a means becomes a
unenforceable duty or perhaps not a duty at all. On the face of things, the second of these
puzzles is no less important than the first.
In short, while Tadros’ enforceable duty account helps to resolve one difficulty with the
means principle, it raises another, just as awkward, difficulty. This is not claim that the
Tadros’ observations and submissions are without value. Successful explanations often
throw up new things to be explained, and sometimes this helpfully redirects our attention to
something that can be fruitfully engaged with.
Finally, I want to raise some doubts about Tadros’ attempt to ground moral status in our
end-setting capacity. A problem here is that it is not immediately clear why the fact that a
being has an end-setting capacity generates agent-relative constraints as opposed to the
moral aim of maximizing the degree to which beings with the capacity exercise it, or the
degree to which they are in a position to exercise it without being preventing from doing so
by others (cf. Scheffler 1982: 99–100). In the context of justifying an agent-relative
constraint on harming others as means, Tadros writes: ‘Our right to set ends for ourselves
can be understood as derived from our independence from each other—the autonomous
control we have over our own lives’ (Tadros 2011: 127). By violating someone’s rights we
might actually boost the degree to which others enjoy autonomous control over their lives.
For example, it is possible that I harm you as a means of preventing more cases in which
Crim Law and Philos
123
others are used as means and, thus, increase the total number of people exercising
autonomous control over their lives. If the underlying concern is our autonomous control
over our lives, why is this kind of (as we might call it) ‘strategic’ violation not permitted?
Note also that some violations of others do not impair their capacity to set ends for
themselves. In fact, they may even boost that capacity. For example, a violation could
make one better able to settle on ends in the future (insofar as this capacity comes in
degrees and can be affected by what others do to one)—e.g., one becomes more headstrong
and willful as a result of one’s experience of being violated—even when it places limits on
the ends that one can pursue successfully. Similarly, pace Tadros, our ‘ability to set ends
for’ ourselves is not reduced because we are morally required to bear great costs for the
benefit of others (Tadros 2011: 211). In fact, it is not even clear that a moral requirement of
this sort limits our ability to permissibly set ends for ourselves relative to a situation in
which we are morally required not to bear such costs.
Note,finally,thatiftherationalebehindthemeansprincipleconcernsourcapacitytosetendsfor
ourselves,onemightaskwhythefocusofthemeansprincipleisharmingasameans,andnot,more
generally,affectingasameans—whereaffectingincludesbenefiting,e.g.,Igiveyouacar,butIdoso
onlybecauseIamacardealerandyouareasportsstar,andIbelievethathavingthiscarwillmakemany
otherpeoplebuysuchacarfrommegeneratingrevenuesfarinexcessofthecostsofmygifttoyou.For
benefits,likeharms,canclashwithendsonehassetforoneself.Itmightbetruethat,almostalways,
onecontrolswhetheronereceivesorretainsacertainbenefit,whereasonerarelycontrolswhether
onesuffersaharmothershaveimposed.Forthatreasonbeingbenefitedasameansrarelyinterferes
withone’ssettingofendsforoneself,orwithone’spursuitofthoseends.However,thisisatbesta
contingentdifferencebetweenbenefitingandharming.Itcannotexplainwhythemeansprinciple
doesnotextendtocasesofbenefitingasameans.40
The Scope of the Means Principle
Whatever grounds the means principle, there is a (connected) question about how exactly
the principle should be understood. In particular, we need to know what ‘harming another
as a means’ involves. There are at least two places in Tadros’ book where I have trouble
following his view in the regard. The first is where he criticizes Judith Jarvis Thomson’s
recently presented view (Thomson 2008) that it is impermissible to divert the trolley to
save the many.41 Thomson considers a case where a bystander has three options: ‘A trolley
is hurtling down the track towards five people. A bystander could turn the trolley onto a
40 Tadros might respond to the present objection by conceding that the means principle, properly construed,says: ‘it is worse to affect [persons] intentionally than it is to affect them as a side effect of one’s actions’.While this version of the means principle fits his rationale for it better than the version he discusses, itsintuitive fit seems worse. For instance, a suitably revised pair of the bridge and the driver cases would notseem to suggest that it worse to benefit the one as means of saving the five than to benefit the one as anunintended side effect of saving the five.41 I am not defending Thomson’s view here. In fact I find her claim that a ‘bystander … cannot impose acost on the person on the second track if he would be unwilling to bear that cost himself’ (Tadros 2011: 119;Thomson 2008, 366) perplexing. This would seem to imply that permissibility hangs on the bystander’sattitudes, which seems to conflict with Thomson’s view that mental states are irrelevant to permissibility.However, I do agree with Thomson that since the agent ‘wouldn’t himself pay the cost of his good deed if hecould pay, there is no way in which he can decently regard himself as entitled to make someone else pay it’(Thomson 2008: 366). Still, it does not follow from one’s inability to decently think that one lacks a certainentitlement that one in fact lacks it, and it is surprising, given her commitment to the irrelevance topermissibility of mental states, that Thomson thinks otherwise.
Crim Law and Philos
123
second track, but there is one person on that track who will be killed if he does that. He
could also turn the trolley onto a third track, but he is on that track and he will be killed if
he does that’ (Tadros 2011: 118). If I understand him correctly, Tadros supposes that it is
permissible for the bystander to divert the trolley provided he shares the burden of saving
the five equally between the stranger and himself. For example, the bystander must decide
whether to redirect the trolley towards himself or the other lone person by flipping a coin
(Tadros 2011: 119–120). He then proceeds to submit that from the (presumed) fact that the
bystander would not divert the trolley, which would have involved a 50 % chance of being
killed, it does not follow that it is wrong for a bystander in the simpler version of the trolley
case, where a bystander can divert the trolley away from five on to one, to kill the one and
save the five. The reason is that the bystander who flips a coin to decide whether to divert
the trolley on to the lone stranger imposes a cost on himself ‘in the service of’ the goal of
saving the five, whereas the cost that is imposed on the lone stranger in the simpler version
of the trolley case is simply a ‘side effect’.
If Tadros’ suggestion here is that the bystander imposes a cost on himself, in taking on a
50 % risk of being killed, as a means to his aim, or end, of saving the five, and that the
means principle therefore applies to the case, I simply fail to see this. Flipping a coin, and
thus imposing a cost on oneself, is a constraint determining the means by which the
bystander selects the pursuit of his end, not a means in itself. To see this, assume that in the
simpler trolley case the bystander conducts a weighted lottery to determine whether to save
the one (1/6 chance) or the five (5/6) because, so he thinks, this is only fair. Surely, this
would not be a case where the bystander imposes a cost on anyone as a means of saving
others. What is the relevant difference between this case and Tadros’ example of the
bystander with three options?42
Second, Tadros discusses self-defense against ‘non-responsible threats’—that is, the
kind of threat I present to another person when, for example, out of the blue, a strong man
appears and throws me at the other person, using me as a missile. He suggests: ‘I must bear
a greater cost to avert threats to others that would come about through my body than I must
bear to avert threats to others that are independent of me’ (Tadros 2011: 255). And in
defending this view he appeals to the means principle, noting that ‘a person’s body is so
intimately bound up with the person that using their body amounts to using the person’
(Tadros 2011: 255).
This strikes me as an odd reading of the means principle as Tadros defines it. The
principle does not discriminate against using a person as such, but against harming a person
as a means, and surely one can use someone’s body without harming them. Suppose I can
extract an antidote from your hip bone. Hence, I painlessly remove it putting in its place an
artificial hip which is actually much better and will last much longer that your own, thus
leaving you better off. Surely, I have not violated Tadros’ means principle simply because I
have used your body as a means, whether or not my action is impermissible on other
grounds.43 Second, setting this aside, the interpretation of the means principle Tadros
42 I am probably missing something in Tadros’ argument here, but it seems to me that he equivocatesbetween individuals using harm to others (or others) as a means and norms requiring or permitting harm toothers (or others) to be used as means. This emerges, for instance, in a passage where Tadros complains thatThomson assumes that ‘[a] being required to bear a cost in the service of a goal as equivalent to [b] having acost imposed on one where that is not done in the service of a goal, but simply as a side effect’ (Tadros 2011:120). Here the [a]-section concerns what costs norms require to be imposed, whereas the [b]-bit concerns theharms individuals impose on others.43 Note that on the extension of the means principle suggested in the last paragraph of the previous section,this case clashes with the means principle suitably construed.
Crim Law and Philos
123
appears to be deploying implies that using someone’s hair or nails as a means is using their
persons as means. This strikes me as too wide a conception of when someone can complain
that he or she has been used.
Implications of the Duty View
I return now to Tadros’ duty view and its implications for permissible punishment. One
problem with consequentialist views of punishment is that they justify not punishing
wrongdoers whenever doing so has bad consequences overall. Another problem is that, in
principle, they justify punishing the innocent. The duty view avoids the latter problem,
because it justifies punishment only in cases where someone has acquired an enforceable
duty through wrongdoing.
However, although Tadros sometimes suggests otherwise, such wrongdoing need not
consist in creating a threat. For instance, Tadros submits that ‘[i]t is because the person
bears responsibility for a threat that she can be harmed to avert a threat that another person
has created’ (Tadros 2011: 194). Suppose a group consisting of 1000 people is under a
threat of death from two sources. All of these individuals suffer from a fatal infection, and
a time-bomb has been planted in the room in which they are located, which is set to
explode shortly after everyone will die from the infection if it is left untreated. B, who is
not responsible for either of these threats, could, without cost, give them a medicine which
would fully cure them, but refuses to do so thereby violating his duty to rescue. C planted
the time bomb in the room. Surely, if the only way the trapped people can save themselves
is by killing C to extract medicine from his body to cure their disease, and by tying B to the
bomb so that his body will shield them from the explosion, it would be permissible for
them to do this. But if that is correct, one can harm someone as a means of averting a threat
for which the person is not responsible even though this person is not responsible for anythreats. While this line of argument is at odds with the view taken by Tadros, I do not think
he needs to resist it in order to defend punishment (e.g., for failures to rescue). Endorsing
the relevant amendment would, however, distance his position further from that of
McMahan’s responsibility-focused account of liability to defensive harm (McMahan
2009).
Let us return to the first of the two problems for consequentialism I mentioned above: a
built-in tendency to justify not punishing wrongdoers whenever doing so has bad conse-
quences on balance. It is an interesting question whether Tadros’ duty view has a similar
implication, although in a different range of cases. On the duty view, wrongdoers have an
enforceable duty to compensate, and perhaps do more than simply compensate, their
victims. Generally, this duty is best discharged when wrongdoers are punished with a
deterrent effect, thus reducing the risk that the victims will suffer similar crimes in the
future. However, it is possible to imagine a number of crimes where the duty view does not
seem to justify punishment, because the relevant duty is either irrelevant or fulfilled in the
absence of punishment.44 First, there are cases in which those who have been subjected to
the crimes do not in any way benefit from their offenders by being punished. For instance,
arguably, people who have been killed are in no way compensated by their killers being
44 This problem is similar to the one facing retributivists who think that punishment is justified as a meansof stripping offenders of the benefits they received through wrongdoing. Their view seems not to justifypunishing offenders who obtain no benefits from their wrongdoing, yet intuitively we might judge suchpunishment to be justified (cf. Tadros 2011: 27).
Crim Law and Philos
123
punished. One might be tempted to argue that victims of this sort are compensated by the
increased security that their loved ones enjoy given the deterrent effect of punishment
(Tadros 2011: 280). However, this need not be the case: it may be that the victim’s loved
ones are already more secure as a result of the wrongdoing—e.g., because the victim of the
killing was very likely to engage in terrible acts of wrongdoing against their loved ones.
Second, there are cases in which wrongdoers arguably fulfill their duty without being
punished. Thus imagine a case in which a particularly gruesome crime has the effect of
hugely boosting the security of others, including the victim, because for some reason it
deters others from committing similar crimes, or indeed crimes at all, in the future. Again,
imagine a case in which the victim of the wrongdoing is a known wrongdoer himself, now
rendered incapable of doing wrong in the future.
Third, suppose that at the very time of my act of wrongdoing I generously donate a very
large sum of money to a crime fighting fund. The donation as such is supererogatory. My
combined act of wrongdoing and donating improves the security enjoyed by the victim of
my crime—an increase much greater than any that could be obtained through the pro-
portionate punishment of those committing similar offences who do not make the generous
donation I have made. It can be argued here that I have fulfilled my duty to my victim to
remedy my wrong, and that it would be impermissible to punish me.
Fourth, sometimes wrongdoers can only fulfill their duty to increase the security of their
victims if they do something that benefits wrongdoers. Suppose I have wronged a number
of people by selling drugs to kids. I repent, and I want to atone now by leading a life of
abstinence in which I selflessly seek to promote the good. However, for some strange
reason my victims can enjoy increased security only if I am forced to live a life of personal
bliss instead. On the duty view, I have an obligation to live it up, and indulge a hedonistic
lifestyle. Indeed others are justified in forcing me to do this if I try to live of perpetual self-
sacrifice instead—which will leave my victims with no compensation. This is unlikely to
go down well with those with retributivist inclinations, who would presumably want to say
that justice is better served in a world where increased security is gained by harming
wrongdoers.
Obviously, one can tinker with the duty view in various ways in an effort to avoid some
of these implications. Thus friends of the duty view might say that in determining whether
a wrongdoer has fulfilled her duty to remedy her wrongs we should disregard supererog-
atory acts improving the security of the one’s victims performed prior to, or simultaneously
with, the act of wrongdoing: it is only victim security that is boosted through the offender’s
punishment that counts. However, it strikes me as odd for an anti-retributivist to make thisreply, since this restriction is most likely to be adopted by those taking the view that it is
somehow fitting that that offenders benefit victims through being punished, and not
through, say, having a good time, or by making voluntary donations. The root problem here
seems to be the instrumentalism involved in the duty view: its instrumentalism saddles the
duty view with the problem of explaining why punishment should not be replaced by
something else, including beneficial treatment, when doing so is the best instrument for
achieving the relevant goals—or alternatively of explaining why this implication is not
problematic after all.45
Let me mention another worry I suspect defenders of the duty view need to address. At
the core of Tadros’ justification lies the right of victims of crime to be compensated, and
45 One might take Tadros’ remarks about his theory being a ‘stepping stone’ to an abolitionist view ofpunishment (Tadros 2011: 40) to indicate that he would simply bite the bullet here (if he sees it as such) andagree that in these cases the wrongdoer should not be punished.
Crim Law and Philos
123
indeed benefited beyond compensation, for the wrongs they have suffered. Now suppose
the state can implement either of two policies of punishment. One powerfully reduces the
likelihood that victims will suffer similar wrongdoing in the future—say, by making the
punishment for offending against someone who has already experienced a similar offense
more severe than punishment for offending against someone who has not previously been
subjected to the relevant wrong. But it does little to reduce the likelihood that non-victims
will suffer similar wrongdoings in the future. The other has a simpler effect: it powerfully
reduces the likelihood that anyone will suffer from the relevant kind of wrongdoing in the
future. It seems to me that advocates of the duty view must maintain that the first of these
policies is required by justice. Yet is it really clear—all things considered, and all other
things being equal—that we should prioritize, say, reducing the number of people who are
raped twice over reducing the total number of rapes?46 Of course, Tadros mentions con-
siderations that explain why punishment should also seek to prevent future harm to non-
victims. For example, he talks about victims transferring their rights to non-victims and the
state’s duty to protect all citizens. However, these considerations do not detach the duty
view from a commitment to the idea that there are stronger reasons, all things considered,
to reduce the likelihood of victims being wronged again than there are to reduce the
likelihood that anyone will be wronged.
Consider finally the issue of ‘barbaric punishments’. Imagine a case in which the
offender can best, or perhaps only, remedy his wrong by donating some of his organs to his
victims. Tadros explains at length why his view does not commit him to the removal of
organs from wrongdoers (e.g., Tadros 2011: 308–310) in this kind of case. He appears to
think the ‘barbarism’ in this practice is unconscionable.47 I do not find his reasons for
denying that the duty view requires the organs to be donated convincing. However, nor do I
think we should infer that the duty view is untenable on this account (even if others, Tadros
included perhaps, might see this inference as an unavoidable one). Suppose I steal two
kidneys from someone in order to sell them on the black market. I am caught, and it turns
out that the only way to save my victim who is dying a slow death is to remove my kidneys
and transplant them into my victim’s body. I fail to see how I could not have an
enforceable duty to offer my kidneys to my victim—at least, if I would have a duty to, say,
suffer 20 years of prison for organ theft if this was the one and only way in which I could
remedy my wrong. Tadros notes that ‘the harm that it is permissible to impose on a person
to avert a threat for which she is not responsible is less than the harm that it would be
permissible to impose on her to avert the threat for which she was responsible’ (Tadros
2011: 308). Even if this is so, it is an open question whether one can, permissibly, punish
organs thieves by taking their organs and using them to save victims of organ theft by
others (assuming this to be the one and only way to save them). We should recall here that
Tadros thinks that to avert a threat posed by someone seeking to steal one of my kidneys I
may impose greater costs on him—e.g., steal both of his kidneys—provided that this is the
only way his wrong can be prevented.48 As Tadros shows, there are pragmatic reasons why
the state should not deal with wrongdoers by adopting such practices as punitive, forced
46 The ‘all other things being equal’ clause excludes the possibility that the marginal harm resulting from acertain wrong might increase when one is subjected to the wrong repeatedly.47 However, quoting (Fabre 2006), Tadros reports doubts about which is the right view here (e.g. Tadros2011: 195).48 Tadros provides additional principled reasons against the death penalty and against confiscating theorgans of wrongdoers (Tadros 2011: 308–309). None seems to be tied specifically to the duty view.
Crim Law and Philos
123
organ removal (Tadros 2011: 309–310). However, I still fail to see how friends of the duty
view can say this kind of practice is unjustified in principle.
Conclusion
In this critical review I have by no means addressed fully the wide range of topics The Endsof Harm deals with. In my view, the duty view is an interesting and promising account of
the permissibility of punishment which, alongside more familiar views, deserves the
careful attention of philosophers of law. Naturally, I have concentrated on elements of
Tadros’ position I was less than wholly convinced by. Critical assessment of this kind can
give a rather negative impression, but of course I would like to think that my comments
will encourage others to delve more deeply into Tadros’ book. Many, on doing so, will
surely want to develop and strengthen the duty view he defends.
Acknowledgments I am very grateful to Nicholas Vrousalis, Alex Voorhoeve, and Zofia Stemplowska,for helpful discussions of the topics dealt with in this review paper, and to Massimo Renzo, RaffaeleRodogno and especially to Victor Tadros, for excellent written comments on an earlier draft of this paper.
References
Dancy, J. (1993) Moral Reasons. Oxford: Blackwell.Dancy, J. (2004) Ethics without Principles. Oxford: Clarendon Press.Fabre, C. (2006) Whose Body Is It Anyway? Oxford: Oxford University Press.Frankfurt, H. (1988) The Importance of What We Care About. Cambridge: Cambridge University Press.Hume, D. (1999) An Enquiry Concerning Human Understanding. Oxford: Oxford University Press.Hurley, S. (2003) Justice, Luck, and Knowledge. Cambridge, MA: Harvard University Press.Kagan, S. (1988) The Additive Fallacy. Ethics 99, 5–31.Kamm. F. M. (1996) Morality, Mortality vol. 2. Oxford: Oxford University Press.Kamm, F. M. (2007) Intricate Ethics. Oxford: Oxford University Press.Lippert-Rasmussen (1996) Moral Status and the Impermissibility of Minimizing Violations. Philosophy &
Public Affairs 25.4, 333–49.McMahan J. (2009) Killing in War. Oxford: Clarendon Press.Moore, G. E. (1989 [1903]) Principia Ethica. Cambridge: Cambridge University Press.Parfit, D. (1998) Equality and Priority. In A. Mason (ed.), Ideals of Equality (pp. 1-20). Oxford: Blackwell.Scheffler, S. (1982) The Rejection of Consequentialism: A Philosophical Investigation of the Considerations
Underlying Rival Moral Conceptions. Oxford: Clarendon Press.Strawson, P. (1982) Freedom and Resentment. In G. Watson (ed.), Free Will (pp. 59–80). Oxford: Oxford
University Press.Tadros, V. (2011) Ends of Harm: The Moral Foundations of Criminal Law. Oxford: Oxford University
Press.Thomson, J. J. (2008) Turning the Trolley. Philosophy & Public Affairs 36.4, 359–374.
Crim Law and Philos
123